• Author / Uploaded
• enrique cruz salomon

Citation preview

ARISTOTLE UNIVERSITY OF THESSALONIKI LABORATORY OF SOIL MECHANICS, FOUNDATION & GEOTECHNICAL EARTHQUAKE ENGINEEERING

4th International Conference on Earthquake Geotechnical Engineering 25-28 June Thessaloniki Greece

Proceedings of: Workshops 1, 2, 4 Invited Lectures Delayed Papers editor: Kyriazis Pitilakis

Table of Contents Invited Lectures Presentations TITLE

Authors

STABILIZATION OF THE LEANING TOWER OF PISA. BEHAVIOUR OF THE TOWER AFTER STABILIZATION WORKS 2001-2004

M. JAMIOLKOWSKY, C. VIGGIANI

NEW ORLEANS LEVEE PERFORMANCE IN HURRICANE KATRINA: LESSONS FOR CALIFORNIA’S LEVEE SITUATION

R. B. SEED

Page 1

29

Keynote Lecture Presentation TITLE PILE RESPONSE TO LATERAL SPREADING: FIELD OBSERVATIONS AND CURRENT RESEARCH

Authors

Page

R. DOBRY, C. MEDINA, T. ABDOUN, S. THEVANAYAGAM

61

Delayed papers ID 1700

TITLE SEISMIC DESIGN LOADS FOR METROPOLITAN SUBWAY TUNNELS: THE CASE OF THESSALONIKI METRO

Authors

Page

Kiriazis PITILAKIS, Anastasios ANASTASIADIS, Dimitrios RAPTAKIS, Nikolaos BOUSOULAS, Elena PAPAGEORGIOU

131

WORKSHOP 1: LARGE SCALE FACILITIES, GEOTECHNICAL STRONG MOTION ARRAYS AND EXPERIMENTAL SITES ID

TITLE

Authors

Page

W1-1005

REFLECTIONS ON THE IMPORTANCE OF THE QUALITY OF THE INPUT MOTION IN SEISMIC CENTRIFUGE TESTS

Jean-Louis CHAZELAS, Gopal SP MADABHUSHI

146

W1-1004

STUDY OF PILE RESPONSE TO LATERAL SPREADING USING PHYSICAL TESTING AND COMPUTATIONAL MODELING

Ricardo DOBRY, Sabanayagam THEVANAYAGAM, Tarek ABDOUN, Ahmed ELGAMAL, Usama EL SHAMY, Mourad ZEGHAL, and Claudia MEDINA

161

W1-1010

TRENDS AND OPPORTUNITIES IN THE FURTHER USE AND DEVELOPMENT OF THE EU LARGE RESEARCH INFRASTRUCTURES FOR EARTHQUAKE ENGINEERING

Michel GERADIN and Fabio TAUCER

174

W1-1006

VISUALIZATION OF LARGE-SCALE SEISMIC DATA RECORDS

Falko KUESTER , Tara C. HUTCHINSON, Tung-Ju HSIEH

175

W1-1007

NONLINEAR WAVE PROPAGATION AND TRENDS AT A LARGESCALE CENTRIFUGE FACILITY

Bruce KUTTER and Dan WILSON

187

W1-1011

A 3-D VISUALIZATION SYSTEM FOR LARGE-SCALE EXPERIMENTAL GEOTECHNICAL EARTHQUAKE DATABASES

Jorge MENESES, Masayoshi SATO and Akio ABE

199

W1-1012

15 YEARS OF EUROSEISTEST

Kyriazis PITILAKIS, Dimitris RAPTAKIS, Konstantia MAKRA, Francisco CHAVEZ GARCIA, Maria MANAKOU, Pashalis APOSTOLIDIS, George MANOS

211

W1-1008

LARGE-SCALE GEOTECHNICAL SIMULATIONS TO ADVANCE SEISMIC RISK MANAGEMENT FOR PORTS

Glenn J. RIX, Ellen M. RATHJE, Patricia M. GALLAGHER, and Ross W. BOULANGER

225

W1-1009

INSTRUMENTED GEOTECHNICAL SITES: CURRENT AND FUTURE TRENDS

Jamison H. STEIDL

234

W1-1001

DENSE SEISMIC INSTRUMENTATION OF SMALL SOFT BASINS

Bill (W.R.) STEPHENSON

246

W1-1003

RECENT DEVELOPMENTS OF GEOSYNTHETIC-REINFORCED SOIL STRUCTURES TO SURVIVE STRONG EARTHQUAKES

Fumio TATSUOKA, Junichi KOSEKI, Masaru TATEYAMA, Daiki HIRAKAWA

256

W1-1002

LARGE-SCALE SHAKE TABLE TESTS FOR EARTHQUAKE GEOTECHNICAL ENGINEERING AT NCREE, TAIWAN

Tzou-Shin UENG, Meei-Ling LIN, Wen-Jong CHANG, Chia-Han CHEN, and Kuo-Lung WANG

274

WORKSHOP 2: GEOTECHNICAL EARTHQUAKE ENGINEERING RELATED TO MONUMENTS AND HISTORICAL CENTRES ID

TITLE

Authors

Page

W2-1012

SEISMIC RESPONSE ANALYSIS OF ANCIENT COLUMNS

Nikolaos ARGYRIOU, Olga-Joan KTENIDOU, Maria MANAKOU, Pashalis APOSTOLIDIS, Francisco CHAVEZ GARCIA, Kyriazis PITILAKIS

284

W2-1015

DESIGN AND IMPLEMENTATION OF ENGINEERING MEASURES FOR THE PROTECTION OF A HISTORICAL MONUMENT AT THE SEISMIC AREA OF MOUNT ATHOS PENINSULA GREECE

Stavros BANDIS, Christos SCHINAS, Elias BAKASIS

302

W2-1007

SEISMIC RESPONSE OF HISTORICAL CENTERS IN ITALY: SELECTED CASE STUDIES

Antonio COSTANZO, Anna D’ ONOFRIO, Giuseppe LANZO, Alessandro PAGLIAROLI, Augusto PENNA, Rodolfo PUGLIA, Filippo SANTUCCI DE MAGISTRIS, Stefania SICA, Francesco SILVESTRI, Paolo TOMMASI

319

W2-1001

A RESEARCH ON THE PERFORMANCE OF THE CONCRETE STRUCTURES AND THE REASONS OF THEIR FAILURE IN BAM EARTHQUAKE AND DESIGN SUGGESTIONS

Roozbeh ETTEHAD, Hamed JAHANGIRI

343

W2-1013

UNDERGROUND MONUMENTS (CATACOMBS) IN ALEXANDRIA, EGYPT

Sayed HEMEDA, Kyriazis PITILAKIS, Ioanna PAPAYIANNI, Stavros BANDIS, Mohamed GAMAL

348

W2-1016

EFFECT OF STRONG WIND TO THE CENTRAL TOWER, BAYON, ANGKOR THOM, CAMBODIA

Yoshimori IWASAKI

370

W2-1011

DAMAGES OF EARTHEN STRUCTURES AT ARG-E-BAM CAUSED BY THE EARTHQUAKE OF DEC. 26, 2003, THE CITADEL AT BAM, IRAN

Yoshinori IWASAKI, Mahmoud NEJATI

378

W2-1002

THE RECONSTRUCTION OF THE TEMPLE OF TEMPLE OF ZEUS AT NEMEA: RECENT PROGRESS AND FUTURE PERSPECTIVES

Nicos MAKRIS, Theodoros PSYCHOGIOS

386

W2-1003

SEISMIC PERFORMANCE OF ROCK BLOCK STRUCTURES WITH OBSERVATIONS FROM THE OCTOBER 2006 HAWAII EARTHQUAKE

Edmund MEDLEY, Dimitrios ZEKKOS

398

W2-1005

REGIONAL SUBSIDENCE AND EARTHQUAKES AS THREATS TO ARCHITECTURAL MONUMENTS IN MEXICO CITY

Efrain OVANDO-SHELLEY, Marcia PINTO DE OLIVEIRA, Enrique SANTOYO

410

W2-1004

USING CLASSICAL MONUMENTS FOR THE ASSESSMENT OF PAST EARTHQUAKE SCENARIOS

Ioannis PSYCHARIS

427

W2-1008

SEISMIC PERFORMANCE OF THE 4TH CENTURY A.D., BYZANTINE LAND WALLS OF THE CITY OF THESSALONIKI, GREECE

Anastasios G. SEXTOS, Kosmas C.STYLIANIDIS

439

W2-1010

INFLUENCE OF ENGINEERING AND GEOLOGICAL ENVIRONMENT ON Erbol SHAIMERDENOV, Askar ZHUSUPBEKOV, ARCHITECTURE’S MONUMENTS Tursun ZHUNISOV

452

W2-1009

THE INVERSE PROBLEM: MODELLING PAST EARTHQUAKES FROM Stathis C. STIROS, Villy A. KONTOGIANNI THEIR EFFECTS ON ANCIENT CONSTRUCTIONS – THE CASE OF THE AD365 EAST MEDITERRANEAN EARTHQUAKE

457

W2-1014

EARTHQUAKE RESPONSE AND VULNERABILITY ASSESSMENT OF MASONRY STRUCTURES

Costas SYRMAKEZIS, Athanasios ANTONOPOULOS, Olga MAVROULI

465

W2-1006

GROUTING OF THREE-LEAF MASONRY: EXPERIMENTAL EVIDENCE ON COMPRESSIVE AND SHEAR STRENGTH ENHANCEMENT

Elizabeth VINTZILEOU

477

WORKSHOP 4: HOW CAN EARTHQUAKE GEOTECHNICAL ENGINEERING CONTRIBUTE TO SAFER DESIGN OF STRUCTURES TO RESIST EARTHQUAKES? ID

TITLE

Authors

Page

W4-1001

KEY ISSUES IN THE ANALYSIS OF PILES IN LIQUEFYING SOILS

Misko CUBRINOVSKI, Hayden BOWEN

489

W4-1004

STATE OF ART KNOWLEDGE VS. STATE OF PRACTICE IN SEISMIC RISK MITIGATION – THE ITALIAN EXPERIENCE AFTER THE 2002 S. GIULIANO EARTHQUAKE

Mauro DOLCE , Giacomo DI PASQUALE , Agostino GORETTI

499

W4-1003

THE CONTRIBUTION OF GEOTECHNICAL ENGINEERING TO SAFER DESIGN OF EARTHQUAKE RESISTANT BUILDING FOUNDATION

Michele MAUGERI, Francesco CASTELLI and Maria Rossella MASSIMINO

511

W4-1002

INFLUENCE OF DYNAMIC LOADS (EARTHQUAKE LOADING AND Jost A. STUDER, Hansjürg GYSI AIRPLANE CRASHES) ON THE BEARING CAPACITY OF PILES, A CASE STUDY

541

4th International Conference on Earthquake Geotechnical Engineering

Invited Lectures Presentations

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

 

29

30

                                                  

        

        

 

31





32



33

34



  



   

        





  



         

          

        

35



  



          

        





  



 











36

 



 

 



37



38

39

40

41



42

    

    

43

44

 

 

 

45

46

 

 

47



48

 

49

  

     

 

 

 



     

























 

50

 

 

51

52



53

  

  

54

       

       

  

 

 

       

       

 

 























 

 



 

 

 



 



















55







56

57





 

58



 

59

             

60

4th International Conference on Earthquake Geotechnical Engineering

Keynote Lecture Presentation

PILE RESPONSE TO LATERAL SPREADING: FIELD OBSERVATIONS AND CURRENT RESEARCH

61

Acknowledgments

62

Kobe Earthquake 1995

63

Outline

64

Miwa et al. (2000)

Fukaehama Island, Kobe 1995

65

Usuoka et al. (2007)

Fukaehama Island, Kobe 1995

66

Miwa et al. (2006)

Foundation System

67

Miwa et al. (2006)

Soil Profile

68

Miwa et al. (2006)

Ground Acceleration and Pore Pressure Buildup in Liquefied Layer

69

Miwa et al. (2006)

Analytical Model for Dynamic Soil-Pile-Structure Interaction

70

Miwa et al. (2006)

Pile Failures at Depth just before Liquefaction (Kinematic Effect)

71

Miwa et al. (2000, 2006)

Response Computed by Dynamic SoilPile-Structures Analysis

72

Inertial damage to superstructure and pile at shallow depths may occur much before liquefaction Kinematic damage due to large cyclic ground deformations associated with liquefaction tends to concentrate at the two interfaces between liquefied and nonliquefied layers; it occurs just before or after soil liquefies Kinematic damage due to large permanent ground deformation associated with lateral spreading also tends to concentrate at the top and bottom of liquefied layer; it develops after soil liquefies

Lessons from this Case History

73

NCEER Case Studies of Japanese and U.S. Earthquakes (Hamada and O’Rourke; O’Rourke and Hamada, 1992) Two Special Issues on Kobe Earthquake of ‘Soils and Foundations’ Journal, 1996 and 1998 Tokimatsu (1999) Dobry and Abdoun (2001) Oregon State University Report (Dickenson et al., 2002) Ishihara (2003) U. California Davis Report (Boulanger et al., 2003) Bhattacharya et al. (2004) Proc. Workshop U. California Davis (Boulanger and Tokimatsu, 2005)

Summaries of Case Histories

74

Free field ground deformation D very important parameter Spatial variation of D under structure may contribute to damage Top of piles may move about same as D , or much less if very stiff foundation (high EI, pile groups, batter piles, superstructural constraints) If shallow nonliquefied crust above liquefied layer, passive thrust of that crust is key factor Damaging maximum bending moments occur at top / bottom boundaries of liquefied layer Is pile buckling in liquefied layer a problem? 75

76

Some Unsolved Engineering Questions

77

In-depth studies of case histories (mostly in Japan) using advanced technologies Field tests with blasting (Rollins et al., 2005; Ashford et al., 2006) Large-scale 1 g shaking tests in Japan and U.S. (6m tall inclined laminar boxes), use of advanced sensors Small-scale centrifuge testing (Japan, U.S. at UC Davis and RPI, UK at Cambridge U.) Use of advanced IT tools for data integration, system identification and visualizations Numerical simulations and analyses (DEM, FEM, dynamic and static p-y, limit equilibrium / pushover analyses)

Ongoing Research, Research Tools

78

Outline

79

80

81

82

Outline

83

RPI 150g-ton Centrifuge

84

85

Slightly cemented sand

Nevada sand (Dr=40%)

10 -300 -200 -100

8

6

4

2

0

0

100

200 300 400

T=15sec T=20sec T=30sec

cemented sand

Slightly

Pile Bending Moment Profiles During Shaking

Soil depth (m)

86

87

88

89

90

Outline

91

92

Moment (kN-m)

Disp. (cm)

Disp. (cm)

Ground and Pile Response

93

94

95

96

Maximum bending moment profiles in single pile tests: 1g test (water) and centrifuge tests (water and viscous fluid)

Height [m

97

Distance [m]

Distance [m] 98

Comparison between centrifuge and full-scale shaking table test results

99

100

101

102

Maximum bending moment profiles in single pile tests: 1g test (water) and centrifuge tests (water and viscous fluid)

Height [m

103

104

Outline

105

Recent Results from 1g Test on Lateral Spreading

106

107

108

109

Depth (m)

110

111

Acceleration (g)

112

SAA vs Ring Accelerometer 3m Depth

113

SAA vs Potentiometer Displacement at Surface

114

Deflection (cm)

Base Excitation

Input Base Excitation (displacement of base)

115

Pore Pressure Development

Depth (m)

116

Deflection (cm)

Lateral Spreading

117

Depth (m)

Lateral Spread Initiation - 1

Deflection (cm)

118

Lateral Spread Initiation - 2

Deflection (cm)

119

We are starting to understand better detailed mechanics of lateral spreading and interaction with pile foundations Some significant parameters: – Free field permanent ground deformation – Shallow nonliquefiable layer – Lateral stiffness (and strength) of pile foundation system including superstructural constraints – Areas of pile foundation exposed to soil lateral pressures – Density, permeability of liquefiable layer Active research in several countries, combining field observations and testing, 1g and centrifuge tests, numerical simulations, advanced sensors

Conclusions

120

Thank you!

121

4 th International Conference on Earthquake Geotechnical Engineering June 25-28, 2007 Keynote Lecture No.4

PILE RESPONSE TO LATERAL SPREADING: FIELD OBSERVATIONS AND CURRENT RESEARCH RICARDO DOBRY 1 CLAUDIA MEDINA TAREK ABDOUN SABANAYAGAM THEVANAYAGAM

LIST OF REFERENCES Abdoun, T. (1997). “Modeling of Seismically Induced Lateral Spreading of Multi-layered Soil and its Effect on Pile Foundations,” Ph.D. Thesis, Rensselaer Polytechnic Institute, Troy, NY. Abdoun, T., and Dobry, R. (2002). “Evaluation of Pile Foundation Response to Lateral Spreading,” Soil Dyn Earthq Eng, Vol. 22, No. 9, pp. 1051-1058. Abdoun, T., Dobry, R., O’Rourke, T.D., and Goh, S.H. (2003). “Pile Response to Lateral Spreads: Centrifuge Modeling,” J Geotech Geoenviron Eng, Vol. 129, No. 10, pp. 869-878. Abdoun, T., Dobry, R., O’Rourke, T.D., and Goh, S.H. (2005). Closure to “Pile Response to Lateral Spreads: Centrifuge Modeling,” J Geotech Geoenviron Eng, Vol. 131, No. 4, pp. 532-534. Abdoun, T., Dobry, R., Zimmie, T., Zeghal, M., and Gallagher, P. (2005). “Centrifuge Research of Countermeasures to Protect Pile Foundations Against Liquefaction-induced Lateral Spreading,” Journal of Earthquake Engineering, Vol. 9, Special Issue 1, pp. 105-125. Abdoun, T., Danisch, L., and Ha, D. (2005). “Advanced Sensing for Real-Time Monitoring of Geotechnical Systems,” ASCE Geotech Special Public No. 138 Site Characterization and Modeling (Ellen M. Rathje, ed.), Proc Geo-Frontiers 2005 Conference, January 24-26, 2005, Austin, TX. Abdoun, T., Danisch, L., Ha, D., and Bennett, V. (2006). “Advanced Sensing for Real-Time Monitoring of Geotechnical Systems,” TRB 85th Annual Meeting, January 22-26, 2006, Washington, DC. Abdoun, T., Abe, A., Bennett, V., Danisch, L., Sato, M., Tokimatsu, K., and Ubilla, J., (2007). “Wireless Real Time Monitoring of Soil and Soil-Structure Systems,” Geotech Spec Publ No. 161 Embankments, Dams, and Slopes: Lessons From the New Orleans Levee Failures and Other Current Issues (F. Silva-Tulla and P. Nicholson, eds.), Proc GeoDenver 2007 Conference: New Peaks in Geotechnics, February 18-21, 2007, Denver, CO. Adalier, K. (1996) “Mitigation of Earthquake Induced Liquefaction Hazards,” Ph.D. Thesis, Rensselaer Polytechnic Institute, Troy, NY.

1

Professor, Department of Civil and Environmental Engineering, Rensselaer Polytechnic Institute, Troy, NY, E-mail: [email protected]

122

Ashford, S.A., Juirnarongrit, T., Sugano, T., and Hamada, M. (2006) “Soil-Pile Response to BlastInduced Lateral Spreading. I: Field Test,” J Geotech Geoenviron Eng, Vol. 132, No. 2, pp. 152162. Bartlett, S.F., and Youd, T.L. (1992). “Case Histories of Lateral Spreads Caused by the 1964 Alaska Earthquake,” Ch. 2 of Case Studies of Liquefaction and Lifeline Performance During Past Earthquakes, Vol. 2: United States Case Studies (O’Rourke and Hamada, eds.), NCEER, SUNYBuffalo, Buffalo, NY, Tech. Rept. NCEER-92-0002, pp. 2.1-2.127 Beavers, J.E., (ed.) (1991). “Costa Rica Earthquake Reconnaissance Report,” Earthquake Spectra, EERI, Supplement B, Vol. 7, 127 pages. Benuzca, L. (ed.) (1990). “Loma Prieta Earthquake Reconnaissance Report,” Earthquake Spectra, EERI and NRC, Supplement to Vol. 6, 448 pages. Berrill, J.B., Christensen, S.A., Keenan, R.J., Okada, W., and Pettinga, J.K. (1997). “LateralSpreading Loads on a Piled Bridge Foundation,” Seismic Behavior of Ground and Geotechnical Structures, (Seco E. Pinto, ed.), Balkema, Rotterdam, pp. 176-183. Bhattacharya, S., Madabhushi, S.P.G., and Bolton, M.D. (2003). “Pile Instability During Earthquake Liquefaction,” ASCE Engineering Mechanics Conference EM2003, July 16-18, 2003, Seattle, WA. Bhattacharya, S., and Bolton, M.D. (2004). “Errors in Design Leading to Pile Failures During Seismic Liquefaction,” Proc 5th International Conference on Case Histories in Geotechnical Engineering, April 13-17, 2004, New York, NY. Bhattacharya, S., Madabhushi, S.P.G., and Bolton, M.D. (2004). “An Alternative Mechanism of Pile Failure Deposits During Earthquakes,” Geotechnique, Vol. 54, No. 3, pp. 203-213. Bhattacharya, S., Madabhushi, S.P.G., and Bolton, M.D. (2005). Closure to “An Alternative Mechanism of Pile Failure Deposits During Earthquakes,” Geotechnique, Vol. 55, No. 3, p. 263. Boulanger, R.W., Mejia, L.H., and Idriss, I.M. (1997). “Liquefaction at Moss Landing During Loma Prieta Earthquake,” Journal of Geotechnical and Geoenvironmental Engineering, Vol. 123, No.5, pp.453-467. Boulanger, R.W., Meyers, M.W., Mejia, L.H., and Idriss, I.M. (1998). “Behavior of a Fine-grained Soil During the Loma Prieta Earthquake,” Canadian Geotechnical Journal, Vol. 35, pp.146-158. Boulanger, R.W., Kutter, B.L, Brandenberg, S.J., Singh, P., and Chang, D. (2003). “Pile Foundations in Liquefied and Laterally Spreading Ground During Earthquakes: Centrifuge Experiments and Analyses,” Report No. UCD/CGM-03/01, Center for Geotechnical Modeling, University of California, Davis, CA, 205 pages. Boulanger, R.W. and Tokimatsu, K., eds. (2006). Proceedings Workshop on Seismic Performance and Simulation of Pile Foundations in Liquefiable and Laterally Spreading Ground, March 16-18, 2005, University of California, Davis, CA, ASCE Geotech Special Public No. 145, pp. 50-60. Brandenberg, S.J., Boulanger, R.W., and Kutter, B.L. (2005). Discussion of “Pile Response to Lateral Spreads: Centrifuge Modeling” by Dobry et al., J Geotech Geoenviron Eng, Vol. 131, No. 4, pp. 529-531. Brandenberg, S.J., Boulanger, R.W., Kutter, B.L., and Chang, D. (2005). “Behavior of Pile Foundations in Laterally Spreading Ground During Centrifuge Tests,” J Geotech Geoenviron Eng, Vol. 131, No. 11, pp. 1378-1391.

123

Brandenberg, S.J., Boulanger, R.W., Kutter, B.L., and Chang, D. (2007). “Liquefaction-Induced Softening of Load Transfer Between Pile Groups and Laterally Spreading Crusts,” J Geotech Geoenviron Eng, Vol. 133, No. 1, pp. 91-103. Cubrinovski, M., Kokusho, T., and Ishihara, K. (2004). “Interpretation From Large-Scale Shake Table Tests on Piles Subjected to Spreading of Liquefied Soils,” Proc 11th Intl Conf on Soil Dyn and Earthq Eng, January 7-9, University of California, Berkeley, CA, Vol. 2, pp. 463-470. Cubrinovski, M., Kokusho, T., and Ishihara, K. (2006). “Interpretation From Large-Scale Shake Table Tests on Piles Undergoing Lateral Spreading in Liquefied Soils,” J of Soil Dyn and Earthq Eng, Vol. 26, pp. 275-286. Dale, G. (2002). “Prediction of Pile Bending Response to Lateral Spreading by Elastoplastic Analysis Validated by Centrifuge Tests,” M.S. Thesis, Rensselaer Polytechnic Institute, Troy, NY. Dickenson, S.E., McCullough, N.J., Barkau, M.G., and Wavra, B.J. (2002) “Assessment and Mitigation of Liquefaction Hazards to Bridge Approach Embankments in Oregon,” Final Report SPR 361, Oregon Department of Transportation and Federal Highway Administration, 272 pages. Dobry, R., and Abdoun, T. (1998). “Post-triggering Response of Liquefied Sand in the Free Field and Near Foundations,” Proc ASCE Geotechnical Earthquake Engineering and Soil Dynamics III, Seattle, WA, August 3-6 1998, Vol. 1, pp. 270-300. Dobry, R., and Abdoun, T. (2001). “Recent Studies on Seismic Centrifuge Modeling of Liquefaction and its Effect on Deep Foundation,” State-of-the-Art Paper, Proc 4th Int Conf on Recent Advances in Geotech Earthq Eng and Soil Dyn (S. Prakash, ed.), Paper SOAP 3, Vol. 2, March 26-31, 2001, San Diego, CA, 30 pages. Dobry, R., Taboada, V., and Liu, L. (1995). “Centrifuge Modeling of Liquefaction Effects During Earthquakes,” Proc 1st Int Conf on Earthq Geotech Eng (IS-Tokyo), Tokyo, Japan, Keynote Lecture, pp. 1291-1324. Dobry, R., Abdoun, T., O’Rourke, T.D., and Goh, S.H. (2003). “Single Piles in Lateral Spreads: Field Bending Moment Evaluation,” J Geotech Geoenviron Eng, Vol. 129, No. 10, pp. 879-889. Dobry, R., Abdoun, T., O’Rourke, T.D., and Goh, S.H. (2005). Closure to “Single Piles in Lateral Spreads: Field Bending Moment Evaluation,”. J Geotech Geoenviron Eng, Vol. 131, No. 4, pp. 532-534. Dobry, R., Thevanayagam, S., Abdoun, T., Elgamal, A., El Shamy, U., Zeghal, M., and Medina, C. (2007). “Study of Pile Response to Lateral Spreading Using Physical Testing and Computational Modeling,” Proc. 4th Intl Conf Earthquake Geotechnical Engineering, Thessaloniki, Greece, June 25-28. Dungca, J.R., Kuwano, J., Saruwatari, T., Izawa, J., Suzuki, H., and Tokimatsu, K. (2004). “Shaking Table Tests on the Lateral Response of a Pile Buried in Liquefied Sand,” Proc 11th Intl Conf on Soil Dyn and Earthq Eng, January 7-9, University of California, Berkeley, CA, Vol. 2, pp. 471477. Dungca, J.R., Kuwano, J., Takahashi, A., Saruwatari, T., Izawa, J., Suzuki, H., and Tokimatsu, K. (2006). “Shaking Table Tests on the Lateral Response of a Pile Buried in Liquefiable Sand,” J of Soil Dyn and Earthq Eng, Vol. 26, pp. 287-295.

124

Elgamal, A., Zeghal, M., Taboada, V., and Dobry, R. (1996). “Analysis of Site Liquefaction and Lateral Spreading Using Centrifuge Testing Records,” Soils and Foundations, JGS, Vol. 36, No. 2, pp. 111-121. Elgamal, A., Parra, E., Yang, Z. and Adalier, K. (2002). “Numerical Analysis of Embankment Foundation Liquefaction Countermeasures,” J Earthq Eng, Vol. 6, No. 4, pp. 447-471. Elgamal, A., Yang, Z., and Parra, E. (2002). “Computational modeling of Cyclic Mobility and PostLiquefaction Site Response,” J of Soil Dyn and Earthq Eng, Vol. 22, Issue 4, pp. 259-271. Elgamal, A., Yang, Z., Parra, E., and Ragheb, A. (2003). “Modeling of Cyclic Mobility in Saturated Cohesionless Soils,” Int J of Plast, Vol. 19, Issue 6, pp. 883-905. Elmekati, A., and Zeghal, M. (2007). “Systen Identification Analysis of Test LG-0,” Internal Progress Report, Rensselaer Polytechnic Institute, Troy, NY. Finn, W.D.L. (2004). “An Overview of the Behavior of Pile Foundations in Liquefiable and NonLiquefiable Soils during Earthquake Excitation,” Ishihara Lecture, Proc 11th Intl Conf on Soil Dyn and Earthq Eng, January 7-9, University of California, Berkeley, CA, Vol. 1, pp. 57-67. Gonzalez, L., Abdoun, T., and Dobry, R. (2006). “Effect of Soil Permeability on Centrifuge Modeling of Pile Response to Lateral Spreading,” ASCE Geotech Special Public No. 145 (Boulanger and Tokimatsu, eds.), Proc Workshop on Seismic Performance and Simulation of Pile Foundations in Liquefiable and Laterally Spreading Ground, March 16-18, 2005, University of California, Davis, CA, pp. 50-60. Gonzalez, L., Abdoun, T., and Dobry, R. (2007). “Effect of Soil Permeability on Centrifuge Modeling of Pile Response to Lateral Spreading,” J Geotech Geoenviron Eng (accepted). Guo, W.D., and Ghee, E.H. (2006). “Behavior of Axially Loaded Pile Groups Subjected to Lateral Soil Movement,” ASCE Geotech Special Public No. 153 Foundation Analysis and Design: Innovative Methods (Parsons, R.L., Zhang, L., Guo, W.D., Phoon, K.K., and Yang, M., eds.), Proc Geoshanghai 2006 Conference, June 6-8, 2006, Shanghai, China. Haigh, S.K., and Madabhushi, S.P.G. (2006). “The Effect of Pile Flexibility on Pile-Loading in Lateral Spreading Slopes,” ASCE Geotech Special Public No. 145 (Boulanger and Tokimatsu, eds.), Proc Workshop on Seismic Performance and Simulation of Pile Foundations in Liquefiable and Laterally Spreading Ground, March 16-18, 2005, University of California, Davis, CA, pp. 24-37. Hamada, M. (1992). “Large Ground Deformations and Their Effects on Lifelines: 1964 Niigata Earthquake,” Ch. 3 of Case Studies of Liquefaction and Lifeline Performance During Past Earthquake, Vol. 1: Japanese Case Studies (Hamada and O’Rourke, eds.), NCEER, SUNY-Buffalo, Buffalo, NY, Tech. Rept. NCEER-92-0001, pp. 3.1-3.123. Hamada, M. (1992). “Large Ground Deformations and Their Effects on Lifelines: 1983 NihonkaiChubu Earthquake,” Ch. 4 of Case Studies of Liquefaction and Lifeline Performance During Past Earthquake, Vol. 1: Japanese Case Studies (Hamada and O’Rourke, eds.), NCEER, SUNY-Buffalo, Buffalo, NY, Tech. Rept. NCEER-92-0001, pp. 4.1-4.85. Hamada, M., and O’Rourke, T.D. (eds.), (1992). “Case Studies of Liquefaction and Lifeline Performance During Past Earthquakes. Vol. 1: Japanese Case Studies,” NCEER, SUNY-Buffalo, Buffalo, NY, Tech. Rept. NCEER-92-0001.

125

Hamada, M., Yasuda, S., Isoyama, R., and Emoto, K. (1986). “Study on Liquefaction Induced Permanent Ground Displacement,” Research Committee Report, Association for the Development of Earthquake Prediction, Tokyo, Japan, 87 pages. Hamada, M., R. Isoyama, K. Wakamatsu, (1995). “The 1995 Hyogoken-Nanbu (Kobe) earthquake: Liquefaction, Ground Displacement and Soil Condition in Hanshin Area,” Assoc. for Development of Earthquake Prediction, School of Science and Engineering, Waseda University, and the Japan Engineering Consultants, Publ. by the Assoc. for Development of Earthquake Prediction, 194 pages. Hamada, M., Wakamatsu, K., and Ando, T. (1996). “Liquefaction-Induced Ground Deformation and its Caused Damage During the 1995 Hyogeken-Nanbu Earthquake,” Proc. 6th Japan-U.S. Workshop on Earthquake Resistant Design of Lifeline Facilities and Countermeasures Against Soil Liquefaction (Hamada and O’Rourke, eds.), NCEER, SUNY-Buffalo, Buffalo, NY, Tech. Rept. NCEER-96-0012, Vol. 1, pp. 137-152. Harada, N., Towhata, I., Takatsu, T., Tsunoda, S., and Sesov, V. (2004). “Development of New Drain Method for Protection of Existing Pile Foundations From Liquefaction Effects,” Proc 11th Intl Conf on Soil Dyn and Earthq Eng, January 7-9, University of California, Berkeley, CA, Vol. 2, pp. 498505. Harada, N., Towhata, I., Takatsu, T., Tsunoda, S., and Sesov, V. (2006). “Development of New Drain Method for Protection of Existing Pile Foundations From Liquefaction Effects,” J of Soil Dyn and Earthq Eng, Vol. 26, pp. 297-312. He, L., Elgamal, A., Abdoun, T., Abe, A., Dobry, Meneses, J., Sato, M., and Tokimatsu, K. (2006). “Lateral Load on Piles Due to Liquefaction-Induced Lateral Spreading During 1G Shake Table Experiments,” Proc. 8th U.S. National Conf on Earthq Eng, April 18-22, 2006, San Francisco, CA, Paper No. 881. Hwang, J.I., Kim, C.Y., Chung, C.K., and Kim, M.M. (2004). “Behavior of a Single Pile Subjected to Flow of Liquefied Soil in an Infinite Slope,” Proc 11th Int Conf on Soil Dyn and Earthq Eng, Vol. 2, pp. 573-580. Idriss, I.M., and Boulanger, R.W. (2004). “Semi-empirical Procedures for Evaluating Liquefaction Potential During Earthquakes,” Proc 11th Intl Conf on Soil Dyn and Earthq Eng, January 7-9, University of California, Berkeley, CA, Vol. 1, pp. 32-56. Ishihara, K. (2003). “Liquefaction-induced Lateral Flow and its Effects on Foundation Piles,” Proc. Fifth Natl Conf on Earthquake Engineering, Istanbul, Turkey, May 26-30, Invited Lecture, 28 pages. Ishihara, K., Yasuda, S., and Nagase, H. (1996). “Soil Characteristics and Ground Damage,” Soils and Foundations – Special Issue on Geotechnical Aspects of the January 17 1995 Hyogoken-Nambu Earthquake, JGS, Special Vol. January, pp. 109-118. Iwasaki, T. (1986). “Soil Liquefaction Studies in Japan,” Technical Memorandum of PWRI No. 2239, Public Works Research Institute, Ministry of Construction, Japan, 143 pages. Juirnarongrit, T., and Ashford, S.A. (2006). “Soil-Pile Response to Blast-Induced Lateral Spreading. II: Analysis and Assessment of the p-y Method,” J Geotech Geoenviron Eng, Vol. 132, No. 2, pp. 163-172. Japanese Geotechnical Society (1998). Special Issue of ‘Soils and Foundations’ on the Effects of the 1995 Hyogoke-Nambu Earthquake.

126

Kachadoorian, R. (1968). “Effects of the Earthquake of March 27, 1964, on the Alaska Highway System,” U.S. Geological Survey Professional Paper 545-C, U.S. Geological Survey. Kagawa, T., Sato, M., Minowa, C., Abe, A., and Tazoh, T. (2003) “Centrifuge Simulations of LargeScale Shaking Table Tests: Case Studies,” J Geotech Geoenviron Eng, Vol. 130, No. 7, pp. 663672. Knappett, J.A., and Madabhushi, S.P.G. (2006). “Modelling Effects of Axial Load on Behaviour of Pile Groups in Laterally-Spreading Soil,”. Proc 6th International Conference on Physical Modelling in Geotechnics, August 4-6, 2006, Hong Kong, pp. 1007-1012. Kawashima, K., Shimizu, K., Mori, S., Takagi, M., Susuki, N, and Nakamura, S. (1988). “Analytical Studies on Damage to Bridges and Foundation Piles Caused by Liquefaction-Induced Permanent Ground Displacement,” Proc. Japan-U.S. Workshop on Liquefaction, Large Ground Deformation and Their Effect on Lifeline Facilities, Nov. 16-19, Tokyo, Japan, Association for the Development of Earthquake Prediction, (Japan) and NCEER (U.S.), pp. 99-117. Koyamada, K., Miyamato, Y., and Tokimatsu, K. (2006). “Field Investigation and Analysis Study of Damage Pile Foundation During the 2003 Tokachi-Oki Earthquake,” ASCE Geotech Special Publ No. 145 (Boulanger and Tokimatsu, eds.), pp. 97-108. Lin S., Tseng, Y., Chiang, C., and Hung, C. (2006). “Damage of Piles Caused by Lateral SpreadingBack Study of Three Cases,” ASCE Geotech Special Public No. 145 (Boulanger and Tokimatsu, eds.), Proc Workshop on Seismic Performance and Simulation of Pile Foundations in Liquefiable and Laterally Spreading Ground, March 16-18, 2005, University of California, Davis, CA, pp. 121133. Matsui, T. and K. Oda. (1996). “Foundation Damage to Structures,” Soils and Foundations – Special Issue on Geotechnical Aspects of the January 17 1995 Hyogoken-Nambu Earthquake, JGS, Special Vol., January, pp. 189-200. McCulloch, D.S., and Bonilla, M.G. (1970). “Effects of the Earthquake of March 27, 1964 on the Alaska Railroad,” U.S. Geological Survey Professional Paper 545-D, U.S. Geological Survey. Miwa, S., Ikeda, T., and Sato, T. (2006). “Damage Process of Pile Foundation in Liquefied Ground During Strong Ground Motion,” J of Soil Dyn and Earthq Eng, Vol. 26, pp. 325-336. Mizuno, H. (1987). “Pile Damage During Earthquakes in Japan (1923-1983),” Proc ASCE Session on Dynamic Response of Pile Foundations (T. Nogami ed.), April 27, 1987, Atlantic City, pp. 53-77. O’Rourke, T.D., and Hamada, M. (eds.), (1992). “Case Studies of Liquefaction and Lifeline Performance During Past Earthquakes. Vol. 2: United States Case Studies,” NCEER, SUNYBuffalo, Buffalo, NY, Tech. Rept. NCEER-92-0002. O'Rourke, T.D., and Pease, J.W. (1992). “Large Ground Deformations and Their Effects on Lifeline Facilities: 1989 Loma Prieta Earthquake,” Ch. 5 of Case Studies of Liquefaction and Lifeline Performance During Past Earthquake, Vol. 2: United States Case Studies (O’Rourke and Hamada, eds.), NCEER, SUNY-Buffalo, Buffalo, NY, Tech. Rept. NCEER-92-0002, pp. 5.1-5.85. Parra, E. (1996). “Numerical Modeling of Liquefaction and Lateral Ground Deformation Including Cyclic Mobility and Dilation Response in Soil Systems,” Ph.D. Thesis, Rensselaer Polytechnic Institute, Troy, NY.

127

Pyke, R. (2003). Discussion of “Liquefaction Resistance of Soils: Summary Report from the 1996 NCEER and 1998 NCEER/NSF Workshops on Evaluation of Liquefaction Resistance of Soils” by Youd et al., J Geotech Geoenviron Eng, Vol. 129, No. 3, pp. 283-284. Radwan, H., and Abdoun, T. (2006). “3D Data Viewer of Centrifuge Physical Models,” Geotech Special Publ Proc GeoCongress 2006 Conference: Geotechnical Engineering in the Information Technology Age (D.J. DeGroot, J.T. DeJong, J.D. Frost, and L.G. Baise, eds.), February 26-March 1, 2006, Atlanta, GA. Ramos, R. (1999). “Centrifuge Study of Bending Response of Pile Foundation to a Lateral Spread Including Restraining Effect of Superstructure,” Ph.D. Thesis, Rensselaer Polytechnic Institute, Troy, NY. Ross, G.A., Seed, H.B., and Migliaccio, R.R. (1973). “Performance of Highway Bridge Foundations”, The Great Alaska Earthquake of 1964, Engineering Volume, National Academy of Sciences, Washington, D.C., pp. 190-242. Seed, R.B., Dickenson, S.E., Riemer, M.F., Bray, J.D., Sitar, N., Mitchell, J.K., Idriss, I.M., Kayen, R.E., Kropp, A., Harder Jr., L.F., and Power, M.S. (1990). “Preliminary Report on the Principal Geotechnical Aspects of the October 17, 1989 Loma Prieta Earthquake,” Report No. UCB/EERC90/05, Earthquake Engineering Research Center, University of California at Berkeley, Berkeley, CA., 139 pages. Shea, G.H., (ed.) (1991). “Costa Rica Earthquake Reconnaissance Report,” Earthquake Spectra, EERI, Supplement B, Vol. 7, Chapter 6 – Bridges, 127 pages. Shengcong, F., and Tatsuoka, F. (1984). “Soil Liquefaction During Haicheng and Tangshan Earthquakes in China; A Review,” Soils and Foundations, Vol. 24, No. 4, pp 11-29. Shenthan T. (2001). “Factors Affecting Liquefaction Miitigation in Silty Soils Using Stone Columns,” MS Thesis, University at Buffalo, SUNY, NY, p. 129. Shibata, T., F. Oka, and Y. Ozawa. (1996). “Characteristics of Ground Deformation due to Liquefaction,” Soils and Foundations – Special Issue on Geotechnical Aspects of the January 17 1995 Hyogoken-Nambu Earthquake, JGS, Special Vol. January, pp. 65-80. Stewart, H.E., Miura, F., and O’Rourke, T.D. (1988). “Pile Damage Due to Large Ground Displacement,” Proc. Japan-U.S. Workshop on Liquefaction, Large Ground Deformation and Their Effect on Lifeline Facilities, Nov. 16-19, Tokyo, Japan, Association for the Development of Earthquake Prediction, (Japan) and NCEER (U.S.), pp. 173-182. Susuki H., and Tokimatsu, K. (2004). “Effect of Pore Water Pressure Response Around Pile on p-y Relation During Liquefaction,” Proc 11th Intl Conf on Soil Dyn and Earthq Eng, January 7-9, University of California, Berkeley, CA, Vol. 2, pp. 567-572. Susuki H., Tokimatsu, K., Sato, M., and Abe, A. (2006) “Factor Affecting Horizontal Subgrade Reaction of Piles During Soil Liquefaction and Lateral Spreading,” ASCE Geotech Special Public No. 145 (Boulanger and Tokimatsu, eds.), Proc Workshop on Seismic Performance and Simulation of Pile Foundations in Liquefiable and Laterally Spreading Ground, March 16-18, 2005, University of California, Davis, CA, pp. 1-10. Swan ,S.W., Flores, P.J., and Hooper, J.D. (1996). “The Manzanillo Mexico Earthquake of October 9, 1995,” NCEES Bulletin, The Quarterly Publication of NCEER, Vol. 10, No. 1, January.

128

Taboada, V.M. (1995). “Centrifuge Modeling of Earthquake-Induced Lateral Spreading in Sand Using a Laminar Box,” J Geotech Geoenviron Eng, Vol. 129, No. 10, pp. 879-889. Tamura, S., and Tokimatsu, K. (2006) “Seismic Earth Pressure Acting on Embedded Footing Based on Large-Scale Shaking Table Tests,” ASCE Geotech Special Public No. 145 (Boulanger and Tokimatsu, eds.), Proc Workshop on Seismic Performance and Simulation of Pile Foundations in Liquefiable and Laterally Spreading Ground, March 16-18, 2005, University of California, Davis, CA, pp. 83-96. TGC (1995). “Geotechnical Response of Quay and Loading Areas in Puerto Manzanillo During October 1995 Earthquake,” (in Spanish) Report TGC Geotecnia 95-1763-C, October, Mexico. Tokimatsu, K. (1999). “Performance of Pile Foundations in Laterally Spreading Soils,” Proc 2nd Int Conf on Earthq Geotech Eng (P. Seco e Pinto, ed.), June 21-25, 1999, Lisbon, Portugal, Vol. 3, pp. 957-964. Tokimatsu K., Mizuno, H., and Kakurai, M. (1996). “Building Damage Associated with Geotechnical Problems,” Soils and Foundations– Special Issue on Geotechnical Aspects of the January 17 1995 Hyogoken-Nambu Earthquake, JGS, Special Vol. January, pp. 219-234. Tokimatsu, K., H. Oh-Oka, Y. Shamoto, A. Nakazawa, and Y. Asaka. (1997). “Failure and Deformation Modes of Piles Caused by Liquefaction-Induced Lateral Spreading in 1995 Hyogoken-Nambu Earthquake,” Proc 3rd Kansai International Geotechnical Forum on Comparative Geotechnical Engineering (KIG-Forum '97), Kobe, January, Publ. by the Kansai Branch of the Japanese Geotechnical Society, pp. 239-248. Tokimatsu, K., H. Oh-Oka, K. Satake, Y. Shamoto, and Y. Asaka, Y. (1998). “Effects of Lateral Ground Movements on Failure Patterns of Piles in the 1995 Hyogoken-Nambu Earthquake,” Proc ASCE Geotechnical Earthquake Engineering and Soil Dynamics III, Seattle, WA, August 3-6, 1998, Vol. 2, pp. 1175-1186 Tokimatsu, K., Suzuki, H., and Sato, M. (2004). “Influence of Inertial and Kinematic Components on Pile Response during Earthquakes,” Proc 11th Intl Conf on Soil Dyn and Earthq Eng, January 7-9, University of California, Berkeley, CA, Vol. 1, pp. 768-775. Tokimatsu, K., Suzuki, H., and Sato, M. (2005). “Effects of Inertial and Kinematic Interaction on Seismic Behavior of Pile with Embedded Foundation,” J of Soil Dyn and Earthq Eng, Vol. 25, pp. 753-762. Towhata, I., Sesov, V., Motamed, R., and Gonzalez, M. (2006). “Model Tests on Lateral Earth Pressure on Large Group Pile Exerted by Horizontal Displacement of Liquefied Sandy Ground,” Proc. 8th U.S. National Conf on Earthq Eng, April 18-22, 2006, San Francisco, CA, Paper No. 1227. Ubilla, J. (2007). “Physical Modeling of the Effects of Natural Hazards on Soil-Structure Interaction,” Ph.D. Thesis, Rensselaer Polytechnic Institute, Troy, NY. Uzuoka, R., Sento, N., Kazama, M., Zhang, F., Yashima, A., and Oka, F. (2007). “Three-dimensional Numerical Simulation of Earthquake Damage to Group-Piles in a Liquefied Ground,” J of Soil Dyn and Earthq Eng, Vol. 27, pp. 395-413. Yao, S., Kobayashi, K., Yoshida, N., and Matsuo, H. (2004). “Interactive Behavior of Soil-PileSuperstructure System in Transient State to Liquefaction by Means of Large Shake Table Test,” J of Soil Dyn and Earthq Eng, Vol. 24, pp. 397-409.

129

Yasuda, F., A. Nanjoh, and K. Kosa. (1997). “Investigation and Checking of Load Bearing Capacity of Damaged Piles,” Proc. of the 3rd Kansai International Geotechnical Forum on Comparative Geotechnical Engineering (KIG-Forum '97), Kobe, January, Publ. by the Kansai Branch of the Japanese Geotechnical Society, pp. 249-257. Yokoyama K., Tamura, K., and Matsuo, O. (1997). “Design Methods of Bridge Foundations Against Soil Liquefaction and Liquefaction-Induced Ground Flow,” Proc 2nd Italy-Japan Workshop on Seismic Design and Retrofit of Bridges, February 27-28, 1997, Rome, Italy, pp. 109-131. Yoshida, N., and Hamada, M. (1991). “Damage to Foundation Piles and Deformation Pattern of Ground Due Liquefaction-Induced Permanent Ground Deformations,” Proc. 3rd Japan-U.S. Workshop on Earthquake Resistant Design of Lifeline Facilities and Countermeasures for Soil Liquefaction, NCEER, SUNY-Buffalo, Buffalo, NY, Tech. Rept. NCEER-91-0001 , pp. 147-161. Yoshida, N., Watanabe, H., Yasuda, S., and Mora, S. (1992). “Liquefaction-Induced Ground Failure and Related Damage to Structures During 1991 Telire-Limon, Costa Rica, Earthquake,” Proc. 4th Japan-U.S. Workshop on Earthquake Resistant Design of Lifeline Facilities and Countermeasures for Soil Liquefaction (Hamada and O’Rourke, eds.), NCEER, SUNY-Buffalo, Buffalo, NY, Tech. Rept. NCEER-92-0019, Vol. 1, pp. 37-52. Yoshida, N., Towhata, I., Yasuda, S., and Kanatani, M. (2005). Discussion to “An Alternative Mechanism of Pile Failure Deposits During Earthquakes” by Bhattacharya et al., Geotechnique, Vol. 55, No. 3, pp. 259-263. Youd, T.L. (1993), “Liquefaction-Induced Damage to Bridges,” Transportation Research Record, published by the TRB and the National Research Council, Washington, DC., No. 1411, pp. 35-41. Youd, T.L., Rollins, K.M., Salazar, A.F., and Wallace, R.M. (1992). “Bridge Damage Caused by Liquefaction During the 22 April 1991 Costa Rica Earthquake,” Proc. 10th World Conference on Earthquake Engineering, July 19-24, Madrid, Spain, Vol. 1, pp 153-158. Youd, T.L., Idriss, I.M., Andrus, R.D., Arango, I., Castro, G., Christian, J.T., Dobry, R., Finn, W.D.L., Harder Jr., L.F., Hynes, M.E., Ishihara, K., Koester, J.P., Liao, S.S.C., Marcuson III, W.F., Martin, G.R., Mitchell, J.K., Moriwaki, Y., Power, M.S., Robertson, P.K., Seed, R.B., and Stokoe II, K.H. (2001). “Liquefaction Resistance of Soils: Summary Report from the 1996 NCEER and 1998 NCEER/NSF Workshops on Evaluation of Liquefaction Resistance of Soils,” J Geotech Geoenviron Eng, Vol. 127, No. 10, pp. 817-833. Youd, T.L., Idriss, I.M., Andrus, R.D., Arango, I., Castro, G., Christian, J.T., Dobry, R., Finn, W.D.L., Harder Jr., L.F., Hynes, M.E., Ishihara, K., Koester, J.P., Liao, S.S.C., Marcuson III, W.F., Martin, G.R., Mitchell, J.K., Moriwaki, Y., Power, M.S., Robertson, P.K., Seed, R.B., and Stokoe II, K.H. (2003). Closure to “Liquefaction Resistance of Soils: Summary Report from the 1996 NCEER and 1998 NCEER/NSF Workshops on Evaluation of Liquefaction Resistance of Soils,” J Geotech Geoenviron Eng, Vol. 129, No. 3, pp. 284-286. Zeghal, M., and El Shamy, U. (2004). “A Continuum-Discrete Hydromechanical Analysis of Granular Deposit Liquefaction,” Int J Numer Anal Met Geomech, Vol. 28, No. 14, pp. 1361-1383. Zeghal, M., and Oskay, C. (2003). “A Local Identification Technique for Geotechnical and Geophysical Systems,” Int J Numer Anal Meth Geomech, Vol. 27, No. 11, pp. 967-987. Zeghal, M., Elgamal, A.W., Tang, H.T., and Stepp, J.C. (1995). “Lotung Downhole Array. II: Evaluation of Soil Nonlinear Properties.” J of Geotech Eng, Vol. 121, No. 4, pp. 363-378.

130

4th International Conference on Earthquake Geotechnical Engineering

Delayed Papers



     

 

                                                                                                                                                                                         

  

                                                                 

131

                                                                                                                                                                        







132



                                                               

   

133

                                                                                                                          



                                   

134







 

              

              

               

              

 



 

              

              

              

                 

                

                                                                                  







  

135

                          



 













    



























  

        















 

 



 

 

 





































 

    

  









   

    

 

















  

















  

                



 











    











                                                                    

136

                                    

         





                                               

137



     





                

138

             



             



 

 

 

  

 





                      

 

                   

            



                  

      

    

  

 

    

        

   

     

     

             

 





139

    





 















 

                   

  

  

   

















































 









 

                                                                                                                                       

140

                                              

                               







 









































 



       







 











































                          







 













































         

141

              

 





 











































 

                 

 





 













































                

 





 













































           

142

          

 





 

 

 





 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 





                

 





 















































                                                                               

143

                                                                                                                                                                                                                                                      

144

                                                               

145

4th International Conference on Earthquake Geotechnical Engineering

WORKSHOP 1 Large scale facilities, geotechnical strong motion arrays and experimental sites

4 th International Conference on Earthquake Geotechnical Engineering June 25-28, 2007 Paper No. W1-1005

REFLECTIONS ON THE IMPORTANCE OF THE QUALITY OF THE INPUT MOTION IN SEISMIC CENTRIFUGE TESTS Jean-Louis CHAZELAS 1, Gopal SP MADABHUSHI2 ABSTRACT Dynamic centrifuge modelling is widely accepted as the experimental technique that can be used to understand complex behaviour of soil-structure systems subjected to earthquake loading. In Europe, the centrifuge facility at Schofield Centre in Cambridge has been active for more than 30 years in the area of dynamic centrifuge modelling. The earthquake loading is simulated on the Cambridge centrifuge using simple mechanical actuators that produce sinusoidal shaking. While such input motions are simple and helpful in deciphering certain aspects of soil behaviour particularly while assessing damaging effects of the earthquakes, the complex non-linear behaviour of soils requires more sophisticated earthquake actuators that can simulate multi-frequency nature of real earthquakes. Such a servo-hydraulic shaker has been established on the LCPC centrifuge in Nantes, similar to the shaker at C-Core centrifuge facility in St Johns, Canada. In this keynote paper, the importance of the quality of input motion is investigated. The difficulties in generating complex motion aboard centrifuges are discussed. Another aim of this paper is to discuss some of the exciting developments that are occurring in the modelling of earthquakes on centrifuges. The outline design of a 2-D (horizontal and vertical) shaker being developed at Cambridge is presented. Similarly creation of distributed testing facilities that are networked within UK under the UK-NEES project that is linked to US-NEES and other similar networks opens up a new era of collaborative testing in earthquake engineering.

Keywords: actuators, earthquakes, input motions, geotechnical engineering, centrifuge modelling

INTRODUCTION Physical modeling in earthquake engineering with reduced scale experiments in the centrifuge is now widely considered as the established experimental technique of obtaining data in controlled conditions to help engineers and researchers to understand the mechanism involved in the response of soil – structure systems to seismic loading. This experimental approach recreates the stress state in soils which is a fundamental condition to observe realistic soil behaviour. Of course, as any other experimental method it has its limitations, among which the most evident is the boundary effect due to the fact that the soil model mounted in the centrifuge is necessarily of limited dimensions. This is classically treated by using laminar or shear stack box which allow the natural deformation of a soil column. Another important limit is the ability to impose a realistic input at the base of the soil model. Earthquakes generate a complex sequence of vibrations that can lead to 3D displacements with a broad frequency content of very variable duration. Firstly, it is difficult to build a 3D shaking table in the 1

Senior Researcher, Laboratoire Central des Ponts et Chaussées, Nantes, France, Email: [email protected] 2 Reader in Geotechnical Engineering, Department of Engineering, University of Cambridge, Cambridge, UK, Email: [email protected]

146

limited volume of the basket of a centrifuge. Secondly, it is difficult to generate independent time histories for each axis of shaking: it is of course simpler to generate harmonic inputs than to reproduce broad band records of real earthquakes. Note that these two problems apply as well to full scale shaking tables. The purpose of this paper is to present recent and upcoming experimental facilities developed in Europe and discuss the effort that led to an improvement in the quality of the simulations. Establishing the quality of simulations essentially means to define and justify the choice of 1, 2 or 3D movements, the ability to simulate single or multi frequency inputs, and the quality of the resulting inputs in comparison with the target input. There is an on-going discussion amongst centrifuge modellers on the best type of input motions that may be used in dynamic centrifuge modelling. The input motion that has simple, sinusoidal tone bursts at different frequency will lend itself to easy analysis of the response the soil and the superstructure. This is used extensively at Cambridge on a wide range of boundary value problems in which the key mechanisms of failure are eloquently deciphered. This choice to some extent is independent of the actuators available with RPI centrifuge facility using sinusoidal inputs even though their servo-hydraulic shaker is capable of simulating realistic earthquakes. Similarly, use of a more realistic input motion from a previous earthquake such as Kobe motion or Northridge motion would be considered useful from the design of future structures point of view. Further, the role of multi-frequency input motion on the dynamic behaviour of soils is not fully understood. It is generally argued that for soil liquefaction problems use of simple input motions is sufficient. Recently, finite element analyses were carried out by Ghosh and Madabhushi (2003) and dynamic centrifuge modelling was carried out by Madabhushi, Ghosh and Kutter (2006) to investigate the role of type of input motions in the generation of excess pore pressures. These investigations revealed that the amount of excess pore pressure generated in loose, saturated sands may be not be effected by sinusoidal input motions or more realistic input motions. However, the amount of lateral spreading of sloping ground that can occur may be quite different if sinusoidal motions are used as the dilation spikes that occur during strong shaking cycles are more pronounced compared to a more realistic input motion with only a few strong cycles of shaking. First attempts of centrifuge shaking tables were mechanical 1D harmonic devices based on leaf spring device (Morris, 1979), bumpy road tracks (Kutter, 1982) or cams systems (Suzuki et al., 1991, Kimura et al, 1998, ). Other technologies have been tested – such as explosives (Zelikson et al., 1981), piezo-electric jacks (Arulanandan, 1982), electromagnetic motors (Fujii, 1994) - but the majority of the existing devices are now electro-hydraulic (Ketcham et al., 1988, Van Laak et al. 1994) because of the ability of electric servo-valves to accept complex driving functions and the command hydraulic jacks with a rich frequency content. Few 2 D devices have been developed either with two horizontal shaking directions or one horizontal and one vertical. The difficulty of avoiding uncontrolled frequency contents – especially harmonics due to mechanical guidance and clearance in sine inputs – and spurious movements – especially yawing and rocking – is largely increased from 1D to 2D. Note that these considerations apply as well to full scale shaking tables but with specific aspects due to the fact that the device is embarked in the basket of a rotating machine. Among main issues in a shaker design are how to shake the biggest mass – i.e. the largest model – and how to control that the acceleration at the basis of the soil box with reduced spurious movements and reduced transmissions of vibration to the centrifuge. Till recently, the basic option was to use the basket of the centrifuge as the reaction mass for the shaking of the table. The heaviest payload was then dependent on the basket weight and the level of vibration tolerated by the centrifuge. The question of the spurious movements arises first from problems of symmetry in the case of a simple actuator, and from the coordination in the case of multiple actuator. The type of bearings is also of main importance in this design – with two opposite schools; rigid guidance on rails and no guidance at all on oil bearings. C-Core Laboratory in Newfoundland, Canada, and Laboratoire Central des Ponts et Chaussées, France, have purchased a new concept 1D earthquake simulator (EQS) designed by ACTIDYN that overcome these two basic difficulties with a highly sophisticated set of technological solutions

147

(Chazelas et al, 2006). On the other hand, Cambridge University has long operated 1D mechanical shakers (ref of the bumpy road, Madabhushi et al, 1998), and is now developing a new 2D simulator. We propose here to present the different aspects of the performances achieved at LCPC's facility, to present the priorities assigned by Cambridge to its new design and then to open a debate on the interest of achieving such quality of input and control.

LABORATORY SPECIFICATIONS OF A HIGH QUALITY 1D SHAKING TABLE AT C-CORE AND LCPC The specifications imposed to providers by C-Core and LCPC were very similar; they result from a bibliographical review of existing device giving an insight on what was possible and what have been the recent evolutions. Both Laboratories chose a 1D shaker, as very few 2D devices are now operational as there is still much work to realize in 1D modelling. The direction of the shaking, conventionally noted Y, was naturally fixed horizontal regarding the model in flight and parallel to the axis of rotation of the centrifuge regarding the fix natural repair in order to limit the Coriolis forces. The maximum g level of operation was fixed between 80 and 100 g as many models are achieved at 40, 50 or 80 g in the domain of foundations. Over 100 g, the models become very small and difficult to be instrumented. The most important specifications were the level of the horizontal acceleration – a minimum of 0,4 g prototype scale – this figure commonly admitted since Kobe 1995 earthquake. The control on the acceleration – an not on the displacement – was considered as compulsory in relation to the fact that the design of a reduced scale experiments in the centrifuge is controlled by the level of the centrifuge acceleration. C-Core called for proposal in 2001 and finally negotiated with Actidyn the performances recalled in table I. LCPC followed two year later and just increased certain values of the specifications (see table I). It must be emphasized that these specifications correspond to sine inputs; this inputs imply the highest power on the jacks shaking the table and the largest volume of hydraulic storage. These specifications included upper bounds for the spurious moments, expressed as a maximum 10% ratio between the X and Z acceleration on the table and the Y acceleration in the direction of the shaking. Note that these accelerations were to be recorded at the extremity of the table, the most severe position to evaluate these movements. Table I : Specifications of C-Core and LCPC Earthquake simulators Specifications

C- Core

LCPC

Maximum centrifuge acceleration of operation (MAO) 80 g

80 g

Maximum payload

400 kg

400 kg

Maximum horizontal acceleration (Y direction)

0.5 MAO

0.5 MAO

Duration of full power sine shake

1s

1s

Maximum velocity

1m/s

1m/s

Maximum displacement

2.5 mm

5 mm

Bandwidth of operation in sine

20 – 200 Hz

20 – 250 Hz

Maximum spurious accelerations in X and Z direction RMS X and RMS Z < recorded at the Y end of the shaking table 10% RMS Y

RMS X and RMS Z < 10% RMS Y

LCPC added a last category of specifications: the ability of the shaker to reproduce earthquakes with very specific signatures supposed to solicit differently the machine. Four records from reference earthquake were specified. They are recalled in table II with their respective selection criteria.

148

TECHNOLOGICAL RESPONSE AND PERFORMANCES In addition to the fulfilment of the above mentioned specifications, the main Actidyn's ambitions were to demonstrate its ability to isolate the centrifuge from the vibrations of the payload and to limit the spurious movements applied to the model. The technological choices have been detailed in [Perdriat et al, 2002] and [Hutin et al, 2004], and will only be described briefly here: - the main innovation was the concept of a permanent dynamic equilibrium between the payload on its shaking table and a counterweight embarked together in the basket of the centrifuge. Actidyn was then able to increase largely the mass of the payload, provided the centrifuge was able to support the total mass. This last condition was easily fulfilled in both laboratories, - the second technological bet was to opt for oil bearings – two superimposed, one for the counterweight on the basis of the shaker and one for the payload table on the counterweight (see fig.1). There is not any longitudinal guidance so as to avoid high frequencies due to micro shocks in mechanical clearance, Counterweight

Payload

Oil bearings Basket Platform

Figure 1. Actityn’s EQS - Dynamic equilibrium payload – counterweights - another essential design option was to install one jack on each side of the shaking table between the payload table and the counterweight table. These two jacks and their servo-valves are operated through a multi-axis control system developed by Data Physics, Hutin et al., 2002. In the simulation of the real earthquakes, specifications and technological limits of the machine constraint the results: the control system computes the drive function sent to the servo-valves in a bandwidth of 20 to 400 Hz. Actidyn considered that the jacks would probably have a flat response till 200 or 250 Hz and then fixed the bandwidth controlled by the Data Physic software controller to 20 – 400 Hz. In the domain of broad band signals, the records were first filtered in the 20 - 350 Hz bandwidth because the power needed is much reduced as regard to sine tests. This means first that the prototype record has to be filtered in 0,5 - 8.75 Hz for 1/40th tests or 0.25 - 4.75 Hz for 1/80th test, for example before time scaling. It must then be pointed that the record from Kobe cannot be correctly reproduced lower than 1/55 th because its Fourier spectrum still contain much energy at 0,4 Hz. At higher reduction scales, the bandwidth of the system will accept such low frequency earthquakes. The physical limits of the machines in terms of maximum accelerations, velocities and displacements also impose constraints : the maximum horizontal acceleration is limited to 0.5 the centrifuge acceleration, the velocity is limited to 1 m/s and the maximum displacement is limited to 5 mm. It is necessary to control, by a double integration process, that none of these limits will be overcome by the reference input. This leads to eventually apply a reduction factor on the filtered record as shown in the last column of table II.

149

With the combination of the specifications and the physical limits of the machine, any earthquake cannot be simulated without care but table II shows that a wide rage of characteristic earthquakes was theoretically possible to run. Figures 2 and 3 show the quality of the fitting of the model earthquake to the reference record for two of these references. All these figures are expressed in the prototype 1g scale. The payload was in a first time a rigid concrete block of mass 400 kg. In the case of Mexico, the high frequency content corresponds to non controlled high frequencies (around 360 Hz) at the time of these tests. Since then, with a fine calibration of the accelerometers in the control loop, these spurious frequencies have been largely reduced. This is proved by the comparison of the Landers reference to records in the tree directions at the extremity of the shaking table supporting the payload (figure 5). This test was run at 40 g and the payload was a shear stack box with sand, for a total mass of 350 kg. Table II: Real earthquake records selected as references in order to evaluate the performance of LCPC’s earthquake simulator Selection criteria

Span

Landers- Lucerne Valley Station Component N09E 28/06/1992 Kobe DAI8-G - Component N43W - 17/01/95 Mexico - Sec. Com. Y Transport Station Component 090 – 19/09/85 Northridge Tarzana Station Component 90 – 17/01/94

Short – Strong amplitude in low frequencies and important velocity spike. Long span and high amplitudes of acceleration Long span – Rich spectrum in low frequencies

48 s

0.3 0.2 0. 0 -0.1 -0.2

Very impulsive. Very high acceleration peaks

Acceleration - g

Acceleration - g

Site

32

33

34

35

time - s

36

Acceleration - g

Acceleration - g

0.2 0 -0.2 -0.4

39.5

40 time - s

0 dB 120 s

40.5

0.41

- 3 dB at 50 g - 4 dB at 80 g

180 s 0 dB 60 s

0.4 0.3 0 -0.2 -0.4 -0.6

36

36.5

0.2 0.1 0 -0.1 -0.2 -0.3

44

- 8,5 dB at 50 g - 5.6 dB at 80 g

37

time - s

Y table reference

0.4

Reduction ratio after filtering

45

46

37.5

47

38

48

time - s

Figure 2. Sequences of the Landers earthquake tested at 50 g centrifuge - 400 kg rigid payload

150

0.16 Mexico 50 g

Y table refererence

0.08

10-2 Y table reference

2. Acceleration -

Acceleration - g

0.12

2.8

0.04 0

- 0.04

2 1. 1. 0.

- 0.08

0.

- 0.12

0

- 0.16 2

3

3

4

4 5 time - s

5

6

6

7

7

0

2 4 6 Frequency - hz

8

Figure 3. Principal sequence of Mexico earthquake(left) and Fourier spectrum (right) tested at 50 g centrifuge with a 400 kg rigid payload 10-2

reference Landers 40 g

1

X table Y table Z table

0.875 0.75 0.625 0.5 0.375 0.25 0 125 0

0

1.25

2

3.75

5

frequency - Hz

6.25

7.5

8.75

10

Figure 4. Fourier Spectrum of Landers earthquake simulation with spurious accelerations This last figure shows that the system is able to face a certain lack of equilibrium of the dynamic balance as the counterweights are not of tunable mass. The above-mentioned tests verified the ability of the device to simulate true 1D broadband earthquakes with different frequency content. Of course in the domain of sine inputs, known to be much more demanding for the machine, the main problem was to control the ability to carry out full power sine inputs under any centrifuge accelerations from 40 to 80 g and during 1 second (model scale, which correspond respectively to 40 to 80 s at model scale). In brief, during the acceptance tests, we could carry out very pure sine inputs from 40 Hz up to 100 Hz in this range of centrifuge accelerations and with the rigid 400 kg payload. Pure, here, meant with no spurious movements – acceleration in the X (yawing) and Z (rocking) direction measured at the extremity of the shaking table – greater than 15 % (we had specified 10%). As indicated earlier, last tests with a fine calibration of the accelerometers involved in the control loops largely widened this frequency range as we achieved tests at 40 g centrifuge, 18 g horizontal acceleration (or 0.45 g prototype scale) from 32 to 200 Hz with spurious moments in the X and Z direction limited to 10%. Two other types of controls have been achieved during acceptance test: an analysis of the harmonic content of the sine inputs and a verification of the vibration of the basket and the arms of the centrifuge. The harmonic content had not been specified. The correction of the harmonic content is realized by the Data Physic control software through an iterative fitting process applied to a dummy payload. The most troublesome harmonics to be corrected were generally the third and the fifth and, at the end of the fitting process, appeared to be limited to less than 10 % of the main frequency

151

component in the band 40 – 100 Hz. This should be improved with the recent corrections in the control loops. Of course, when the sine input will be over 130 Hz, correcting the third harmonic will be impossible due to the 400 Hz limit of the controller. The control of the vibration of the basket is not yet completely checked: it has been only conducted at 40g’s due to aging of the centrifuge itself. The isolation was determined at different frequencies with sinusoidal motion. It was calculated as the ratio of the acceleration measured on the shaking table to the one measured at the basket floor or on the arms. It is to be compared to the “inertial isolation ratio”, the ratio of the basket and simulator mass to the payload mass. The inertial isolation ratio is 6.5 while from 40 to 100 Hz the real isolation is from 30 to 60, that is to say an improvement of a factor 5 to 9. At lower frequencies the isolation improvement is reduced to 3 and at higher frequencies, it vanished progressively but these tests should be renewed next with the corrections introduced in the control loops. Globally speaking, the earthquake simulator at LCPC produces earthquakes with a very good fitting to the reference signal, as well sinusoidal motion of up to 0.5 g (prototype scale) as a wide range of broad band real records at 40 and 50 g centrifuge in the bandwidth 40 – 200 Hz for sinusoidal motion and 0 – 350 Hz for broad band record. At higher level of centrifuge acceleration – up to 80 gc - the bandwidth of acceptable response has been tentatively controlled to be narrower – up to 100 Hz - but should be once more controlled with recent improvements. Good fitting means limited spurious movements – yawing and rocking accelerations at the extremity of the table less that 10 to 15% - and superior harmonics for sine inputs limited to 10% in amplitude. EARTHQUAKE GEOTECHNICAL ENGINEERING RESEARCH AT CAMBRIDGE UNIVERSITY Current Facilities: The success of earthquake geotechnical engineering at Cambridge depended to a large extent on the simple mechanical actuators that have been used for more than 30 years. The current earthquake actuator that relies on Stored Angular Momentum (SAM) to deliver powerful earthquakes at high gravities was developed and is in operation for 12 years, Madabhushi et al (1998). In Fig.5 the front view of the SAM actuator while in Fig.6 a view of the SAM actuator loaded onto the end of the 10m diameter Turner Beam centrifuge is presented. The model seen in Fig.6 was from an investigation carried out by Haigh and Madabhushi (2002), on lateral spreading of liquefied ground past square and circular piles.

Figure 5. A view of SAM earthquake actuator at Cambridge

152

The SAM earthquake actuator is a mechanical device which stores the large amount of energy required for the model earthquake event in a set of flywheels. At the desired moment this energy is transferred to the soil model via a reciprocating rod and a fast acting clutch. When the clutch is closed through a high pressure system to start the earthquake, the clutch grabs the reciprocating rod and shakes with an amplitude of 2.5 mm. This is transferred to the soil model via a bell crank mechanism. The levering distance can be adjusted to vary the strength of the earthquake. The duration of the earthquake can be changed by determining the duration for which the clutch stays on. Earthquakes at different frequency tone bursts can be obtained by selecting the angular frequency of the flywheels. Recent modifications to the SAM actuator were carried out to further enhance its capabilities and to improve the performance envelope. Early earthquakes using this device were non-symmetric as the clutch migrated downwards to an end stop once the centrifugal acceleration was applied. This meant that at the start of the earthquake the clutch body was hitting the end stop if it grabbed the reciprocating rod during its downward motion. This problem has been rectified by incorporating a pneumatic actuator that centralises the clutch prior to every earthquake. Logic controls automatically turn the air to the pneumatic actuator off once the earthquake is fired and the clutch starts to move with the reciprocating rod. In its original conception the SAM actuator was mounted onto the end of the beam centrifuge and shook a package on the special swing, reminiscent of the Bumpy Road actuator, Kutter (1982). However this arrangement was modified and a self-contained swing platform was developed that could house the SAM actuator as shown in Fig. 5 following a research grant (No:GR/L90415/01) from EPSRC, UK. This has transformed the usage of the SAM actuator and since 1994 nearly PhD students utilised this facility and several industrial, EPSRC and EU projects were successfully completed using this actuator.

Figure 6. A view of SAM earthquake actuator loaded onto the end of the beam centrifuge (Model seen is that of lateral spreading of soil past square and circular piles, Haigh and Madabhushi (2002)) The technical specifications of the SAM actuator are listed in Table III. In its normal operational mode the SAM actuator is used to deliver strong shaking to model packages either at particular but different tone bursts that could recreate damaging cycles felt by the structure during an earthquake loading or to apply a swept sine wave motion to detect the frequency response of the soil-structure system.

153

Table III: Specifications of the SAM actuator Parameter Maximum g-level of operation Dimension of the soil models Earthquake strength of choice Earthquake duration of choice Earthquake frequency of choice

Value 100 g 56 m (L) 25 m (B) 22 m (H) 80 m (L) 25 m (B) 40 m (H) Upto 0.4g of bed rock acceleration From 0 s to 150 s From 0.5 Hz to 5 Hz Swept sine wave capability

Note: All parameters above are in prototype scale As mentioned earlier the SAM actuator was used in the investigation of several boundary value problems. As an example of the input motions generated by the SAM actuator the following investigation of liquefaction induced lateral spreading problem is presented. The dynamic behaviour of the slope was studied using miniature instrumentation for the measurement of pore-pressures and accelerations throughout the slope. Analysis of these signals has revealed interesting details about the response of these slopes to earthquake loading. The accelerometer time-histories in Fig.7 show the measured base (ACC 9082), mid layer acceleration (ACC 8076) and surface accelerations (ACC 8025) in one of the models. It can be seen that whilst the base motion is approximately constant from cycle to cycle, the surface response late in the earthquake shows alternate cycles having profoundly different behaviour.

Figure 7. Acceleration time-histories This shows itself as an amplified frequency component at half of the fundamental earthquake frequency upon study of FFT’s. Measurement of the phase lag of acceleration between base and surface of the models, as could be achieved from the time-histories shown in Fig.8, allows estimates of the shear wave velocity to be made at different times during the earthquake. From this data it can be shown that as the soil liquefies and softens, the shear wave velocity falls to such a point that the natural frequency of the soil column becomes approximately 25 Hz, half that of the earthquake excitation. It is thus postulated that the soil column is resonating at this natural frequency, hence giving the behaviour described above.

154

Figure 8. Upward propagation of S wave It is also interesting to note the dilative response of the soil slope. All of the PPT’s present in the model show significant dilative behaviour occurring with the generation of a “suction-spike” once per cycle. Examining the timing of these spikes with respect to depth illustrates that this suction pulse propagates vertically from the base of the model to the surface at the shear wave velocity. This behaviour is illustrated in Fig.9. It is postulated that this is due to the dynamic shear stress applied by the wave, superimposed on the initial static shear stress causing the soil stress path to cross the characteristic state threshold and hence the soil to dilate. This pore-pressure behaviour will cause a slip-stick motion of the soil down the slope, with velocity and displacement being accumulated while the base is accelerating upslope and then locking up on the other half-cycle when dilation occurs. As described earlier using a sinusoidal input motion may lead to an under-estimate of the amount of lateral spreading of sloping ground that can occur as strong cycles of shaking are applied throughout the model earthquake, making the liquefied soil to dilate in each half cycle and hence stopping the lateral displacement of the ground. If a more realistic earthquake motion is applied then the ground liquefies during one or two strong cycles applied and stays liquefied during the smaller cycles that follow allowing the ground to suffer much larger lateral displacement. This is one example where type of input motion can have a bearing on the output from the centrifuge test.

Figure 9. Upward propagation of the suction spike

155

Future Developments: At Cambridge University, preliminary design of a new 2-D earthquake actuator has been recently completed. This actuator, when fabricated will be able to shake the soil models both horizontally and vertically akin to the 2-D actuator recently established at UC Davis in the USA. The vertical ground accelerations can play an important role in the ultimate performance of a structure. Recent earthquakes have yielded many recordings of vertical accelerations which are quite large (in some cases up to 0.8g to 1.0g). Current design codes only allow for a fraction of these as vertical accelerations. Also the combination of vertical shaking followed by strong horizontal shaking can lead to unexpected and interesting failure mechanisms in a wide range of civil engineering structures. With this in view the Cambridge 2-D earthquake actuator project has been initiated and is currently at an early stage. A schematic diagram of the 2-D earthquake actuator assembly is presented in Fig.10 below. A ProEngineer CAD drawing is presented in Fig.11. The design of this 2-D actuator for the Cambridge centrifuge is quite demanding as the payload capacity of the Turner beam centrifuge is limited to 1 tonne. In addition there are severe space constraints. Further, the Turner beam centrifuge is used extensively for non-earthquake testing which means that the 2-D earthquake actuator needs to be loaded and unloaded on and off the centrifuge quite frequently. These bring in additional complexities such as breaks in high pressure hydraulic lines, contamination of the hydraulic fluid etc. Despite these difficulties the design of this 2-D actuator is progressing well and with suitable funding should be available for use in a few years time. This would become the only 2-D earthquake actuator to serve the European Community. The specifications of the 2-D shaker were drawn taking to consideration the special requirements of the Turner beam centrifuge. Unlike the LCPC shaker the entire centrifuge is used as the reaction mass as the swing platform on which the 2-D shaker is mounted is locked onto the centrifuge when the centrifuge is speeded up beyond 10g’s. Also the use of the 2-D shaker is expected to complement the SAM earthquake actuator that can operate at high gravities and deliver powerful sinusoidal earthquakes. The 2-D shaker will be used at relatively lower g levels but with more realistic earthquake input motions in horizontal and vertical directions.

Figure 10. Schematic view of the 2-D actuator assembly

156

Table IV: Preliminary design specifications of the 2-D earthquake actuator Parameter Maximum g-level of operation Dimension of the soil models Earthquake strength of choice – Horizontal direction – Vertical direction Earthquake duration of choice Earthquake frequency of choice

Value 50 g ~ 80 g 56 m (L) 25 m (B) 22 m (H) 80 m (L) 25 m (B) 40 m (H) Up to 0.8g Up to 0.6g From 0 s to 150 s From 0.5 Hz to 5 Hz

Note: All parameters above are in prototype scale There is a good collaboration between Cambridge group and the geotechnical centrifuge modellers at UC Davis through the EPSRC funded UK-NEES project and at LCPC, Nantes through the funding provided by British Council in France. The lessons learnt on the quality of input motions from servohydraulic shakers at Davis and Nantes will be extremely valuable in the development of the 2-D shaker at Cambridge.

UK NEES Project: Another exciting development in the field of earthquake engineering research is the NEES project in the USA that established the concept of distributed testing at geographically distributed sites. This concept is extremely useful for Europe given the expertise in earthquake engineering in Europe and the geographical distances between the centres of excellence. Having distributed experimental facilities that are linked to a dedicated network will enable research workers in the whole of Europe to not only access the experimental data but to actually have tele-observation and tele-participation capabilities. The USA-NEES project has been well set up and a similar network in Europe benefits from the technological advances already achieved in the USA. For example, network protocols for data sharing and data archiving are already available.

Figure 11. Schematic view of the 2-D actuator To complement the US-NEES, EPSRC funded a research project to develop a UK-NEES network among Cambridge, Oxford and Bristol universities. This project is at an early stage and a further opportunity arose to collaborate with NZ-NEES program in New Zealand. In Fig.12 a snapshot of one of the meetings is presented which shows the teams from Cambridge, Oxford, Bristol and Auckland are taking part with the Auckland team making a Power point presentation.

157

DISCUSSION The importance of quality of input motion cannot be over-emphasised. The discussion on quality of input motions can focus on three main aspects: uncontrolled movements limit the capabilities of the facility, the experimental work can be of limited quality, given the non linearity of the soil behaviour, the realistic simulation of earthquake input motions is a requirement. Introducing vibrations in a rotating machine such a centrifuge is not very healthy for it. This a strong contraindication in the case of the “free top” design of the Actidyn centrifuges where avoiding unbalance is a priority. Even with more rigid designs of the centrifuge, the protection of the rotational bearings impose to limit drastically the mass ratio between the payload and the basket when it is used as a reaction mass. This induces a reduction of the size of the models, and then concerns arise with respect to boundary effects. Any uncontrolled spurious movements increase this constraint, with the risk of hitting a resonance frequency of the whole device that would bias the experimental observation or, once more, limit the experimental possibilities. In this regard, one of the very positive aspects of the Turner beam centrifuge at Cambridge where the swinging platform locks onto the end of the centrifuge as the centrifuge speed is increased. This leads to the whole of the beam centrifuge acting as a reaction mass.

Bristol

Oxford

New

Oxford

Cambridge

Figure 12. UK-NEES INSORS Facility Second, vibration analysis is complex enough to avoid any questions of the type: where does this frequency comes from? With harmonic inputs, the problem is of limited importance but with broad band tests it can be troublesome. So much effort is done for the consistency of experiments with scale effects and scaling laws that it would not be consistent not to pay attention on the quality of the seismic input especially in terms of harmonics: as pointed earlier, a first condition is that the simulator is control in acceleration. If the acceleration amplitude of the fist harmonic is 10% of the fundamental input of the test, the displacement generated by this harmonic is 40 time lower, which introduces an acceptable error of 2,5%. With higher components, it can no longer be considered as negligible. With the third and fifth harmonics, the problem is reduced due to the reduction ratio of 9 and 25 between

158

acceleration and displacement. Moreover, facing the non linear behaviour of the soil, adding harmonics and spurious movements will skew the observations. The role of experimental works in the scientific process is both to enable the observation of behaviour mechanisms to confirm or to enrich the numerical models and to provide data for the validation and/or for the fitting of parameters of these numerical models for design purposes. Both applications need fully controlled experiments: being sure of the input motion, the fact it is not polluted by spurious movements and unexpected harmonics is an imperative The non linearity of the soil response to vibrations is the main justification of this reflection because it conditions all the scientific method. Arguments are in favour of simplifications of experimental works: it seems simpler to study the soil and soil-structure responses at discrete frequencies and at different levels of amplitude to provide information on the vibration response, on the frequency dependence of parameters and on the non linearity of the response with the amplitude of the input. The centrifuge is adapted for such a parametric approach but so long as the data are not biased. On the other hand, this approach does not give any information on the dependence of the soil-structure response on multi-frequency input motion, the resonance of non-linear systems and the evaluation of damping in such non-linear systems. The example of dependence of the generation of excess pore pressure and the amount of lateral spreading obtained discussed earlier emphasises the importance of input motion. CONCLUSION The European resources in experimental facilities present an important evolution: Schofield Laboratory in Cambridge University has long been the only research centre able to achieve reduced scale earthquake simulations, limited to 1D harmonic inputs. LCPC offers now a new 1D facility with enlarged possibilities in terms on harmonic and broad band inputs. The quality of the input on this new device is particularly well controlled due to technological innovations. Schofield Laboratory is developing now the second generation of a simulator with 2D inputs. Similar efforts on the quality of the input will drive its design. 1D and 2 D shaking is demonstrated here not to be opposite but complementary in the global approach of the non linearity of the soil and soil – structure response to vibrations.

REFERENCES Arulanandan K., Canclini J., Anadarajah A., Similation of earthquake motions in the centrifuge, Journal of the Geotechnical Engineering Division ASCE, Vol 8, n°GT5, mai 1982, pp 730-742 Chazelas J-L, Derkx F., Thorel L., Escoffier S., Rault G., Buttigieg S., Cottineau L-M, Garnier J., Physical modelling of earthquakes in the LCPC centrifuge, First European Conference on Earthquake Engineering and Seismology, Geneva, Switzerland, 3-8 September 2006, Paper Number: 1064 Fujii N, Development of an electromagnetic centrifuge earthquake simulator, Ko H.Y., McLean F.G. (eds), Centrifuge 91, Boulder, CO, 13-14 juin 1991, Balkema, Rotterdam, 1991, pp 351-354. Ghosh, B. and Madabhushi, S.P.G., A numerical investigation into single and multiple frequency earthquake input motions, Journal of Soil Dynamics and Earthquake Engineering, 2003, Vol. 23, pp 691-704. Haigh, S.K. and Madabhushi, S.P.G., Dynamic Earth Pressures on Piles due to Lateral Spreading. Proc. International Conference on Physical Modelling in Geotechnics, St John's, Newfoundland, Canada, July, Balkema, Rotterdam, 2002. Hutin C., Perdriat J., Rames D., Dynamically balanced braod frequency earthquake simulation system, in Phillips R., Guo P.J., Popescu R. (eds), Physical Modelling in Geotechnics, ICPMG'02, St John's , NF - Canada, 10-12 july 2002, pp 175 – 178.

159

Ketcham S.A., Ko H.Y., Sture S., An electrohydraulic earthquake simulator for centrifuge testing, Corté J.F. (ed), Centrifuge 88, Proceedings of the international conference on geotechnical centrifuge modelling, Paris, France, 25-27 avr. 1988, Balkema, Rotterdam, 1988, pp 97-102. Kutter, B. L., Centrifugal modelling of the response of clay embankments to earthquakes, PhD thesis, Cambridge University, UK, 1982. Kimura T, Takemura J., Saitoh K, Developpment of a simple mechanical shaker using cam shaft, Corté J.F. (ed), Centrifuge 88, Proceedings of the international conference on geotechnical centrifuge modelling, Paris, France, 25-27 avr. 1988, Balkema, Rotterdam, 1988, pp 107-110. Madabhushi S.P.G., Schofield A.N., Lesley S., A new strored angular momentum (SAM) based earthquake actuator, Kimura, Kusakabe & Takemura (eds), Centrifuge 98, Tokyo, Japon, 23-25 sept.98, Balkema, Rotterdam, 1998, pp 111-116. Madabhushi S.P.G., Ghosh B., Kutter B. L., Role of input motion in excess pore pressure generation in dynamic centrifuge modelling, Int. J. of Phys. Modelling in Goetech. 3, 2006, pp 25-34. Morris, D.V., The centrifuge modelling of dynamic behaviour, PhD Thesis, Cambridge University, UK, 1979. Perdriat J., Phillips R., J. Nicolas Font J., C. Hutin, Dynamically balanced broad frequency earthquake simulation system, in Phillips R., Guo P.J., Popescu R. (eds), Physical Modelling in Geotechnics, ICPMG'02, St John's , NF - Canada, 10-12 july 2002, pp 169 – 174. Suzuki K., Babasaki R., Suzuki Y., Centrifuge tests on liquefaction-proof foundation, Ko H.Y., McLean F.G. (eds), Centrifuge 91, Boulder, CO, 13-14 juin 1991, Balkema, Rotterdam, 1991, pp 409 – 415. Van Laak P.A. , Elgamal A.W.,Dobry R., Design and performance of an electrohydraulic shaker for the RPI centrifuge, Leung CF, F.H. Lee, Tan T.S. (eds), Centrifuge 94, Singapore, 31-08/2-09-94, Balkema, Rotterdam, 1994, pp 139-144. Zelikson, A., Devaure, B.,Badel, D., Scale Modeling of soils structure interaction during earthquakes using a progammed series of explosions during centrifugation, Proceedings of the international conference on recent advances in geotechnical earthquake engineering and soil dynamics , 1981, pp 361 – 366.

160

4 th International Conference on Earthquake Geotechnical Engineering June 25-28, 2007 Paper No. W1-1004

STUDY OF PILE RESPONSE TO LATERAL SPREADING USING PHYSICAL TESTING AND COMPUTATIONAL MODELING Ricardo DOBRY1, Sabanayagam THEVANAYAGAM2, Tarek ABDOUN3, Ahmed ELGAMAL4, Usama EL SHAMY5, Mourad ZEGHAL6, and Claudia MEDINA 7 ABSTRACT A progress report is presented on a 3-year, NEES research aimed at detailed clarification and quantification of the mechanics of soil-pile interaction during lateral spreading, with emphasis on the action of the liquefied soil on pile foundation. The project involves cooperative research between five US universities. The centerpiece is 1g base shaking of 6-m tall geotechnical models using the laminar box recently developed at the U. at Buffalo. Results are presented of a free field experiment of liquefaction of a level loose sand deposit. Additional 1g lateral spreading experiments obtained by inclining the laminar box are scheduled for 2007. They include one baseline free-field lateral spreading test, and two more tests involving a stiff pile and a flexible pile, respectively. The 1g testing involves use of state-of-the-art instrumentation techniques to measure soil and pile responses, including a new advanced sensor array of MEMS accelerometers. The instrumentation aims at recording the dynamic response as well as permanent changes in the free field, along the piles, and, especially, in the near field soil around the pile where some of the most important and challenging phenomena take place. Three other complementary efforts simulate these 1g full scale experiments using: (i) small scale centrifuge model experiments at Rensselaer Polytechnic Institute (RPI); (ii) Finite Element Modeling (FEM) computer simulations at the U. of California, San Diego; and (iii) Discrete Element Modeling (DEM) computer simulations at RPI and Tulane U. System Identification and Visualization techniques are systematically used to process the data. Compatible data bases from the 1g and centrifuge experiments, as well as from the FEM and DEM simulations are fed to a newly developed 3D Data Viewer for comparison and integration of the various results.

Keywords: Liquefaction, lateral spreading, piles, large scale testing, centrifuge testing, DEM simulations, FEM simulations 1

Professor, Department of Civil and Environmental Engineering, Rensselaer Polytechnic Institute, Troy, NY, E-mail: [email protected] 2 Associate Professor, Department of Civil, Structural and Environmental Engineering, University at Buffalo, State University of New York, NY, E-mail: [email protected] 3 Associate Professor, Department of Civil and Environmental Engineering, Rensselaer Polytechnic Institute, Troy, NY, E-mail: [email protected] 4

Professor and Chair, Department of Structural Engineering, University of California, San Diego, CA, E-mail: [email protected] 5

Assistant Professor, Department of Civil and Environmental Engineering, Tulane University, New Orleans, LA, E-mail: [email protected] 6 Associate Professor, Department of Civil and Environmental Engineering, Rensselaer Polytechnic Institute, Troy, NY, E-mail: [email protected] 7 Postdoctoral Research Associate, Department of Civil and Environmental Engineering, Rensselaer Polytechnic Institute, Troy, NY, E-mail: [email protected]

161

INTRODUCTION Liquefaction-induced lateral spreading of sloping ground and near waterfronts is a major cause of earthquake damage to deep foundations. Earthquake case histories in the US, Japan, Mexico, Costa Rica, Turkey and other countries, have shown damage to buildings, bridges, port facilities and other pile-supported structures. Effects include cracking and rupture of piles at both shallow and deep elevations, rupture of pile connections, and permanent lateral and vertical movements and rotations of pile heads with corresponding effects on the superstructure (McCulloch and Bonilla, 1970; Hamada et al., 1986; Mizuno, 1987; Hamada and O’Rourke, 1992; O’Rourke and Hamada, 1992; Youd, 1993; Swan et al., 1996; Ishihara et al., 1996; Tokimatsu et al., 1996; Yokoyama et al., 1997; Tokimatsu, 1999; Dobry and Abdoun, 2001; Koyomada et al., 2005; Lin et al., 2005). Extensive experimental research on the effects of lateral spreading on deep foundations in the last decade have included: full scale soil-pile field experiments to blast-induced lateral spreading (Ashford et al., 2006); full scale as well as model 1g tests in shaking tables (Suzuki et al., 2005; Tamura and Tokimatsu, 2005; Towhata et al., 2006; He et al., 2006); and centrifuge model testing (Abdoun et al., 2003; Haig and Madabhushi, 2005; Brandenberg et al., 2005, 2007). A number of these centrifuge and 1g shaking tests of lateral spreading are conducted in inclined laminar boxes attempting to approximate the spreading of an infinite slope (Fig. 1). Case histories as well as experiments suggest that the effect of lateral spreading on piles has the character of a pseudostatic, kinematic soil-structure interaction phenomenon, driven by the permanent lateral movement of the ground in the free field. Bhattacharya et al. (2004) have suggested that another mechanism contributing to the observed failures is pile buckling under gravitational loads, due to decreased lateral support provided by the surrounding liquefied soil. Various foundation analysis and design methods have been proposed to account for the lateral loading effect, where the soil applies static lateral forces to the pile foundation, either (i) as a function of the relative displacement between the foundation and the free field (Beam-on-Winkler-Springs or p-y approach); or (ii) taking the maximum possible values of these lateral soil static forces which depend on the soil strength within an overall limit equilibrium method. When the lateral loads and resulting pile bending moments are controlled by a strong shallow nonliquefied soil layer, the ground pressures can be reasonably related to the passive failure of the soil against the foundation, and hence to the shear strength of the soil (Berrill et al., 1997; Dobry et al., 2003). However, a much larger uncertainty appears when the lateral pressures applied by the liquefied soil play an important role. One source for this uncertainty is the vertical area over which these lateral pressures are applied in pile groups (only the areas of the piles, or also the areas of soil between the piles?). Another source of uncertainty in this case is the significance of soil permeability, not included in current design methods but which recent full scale and centrifuge tests suggest may be extremely important, with less pervious soil such as fine or silty sand tending to produce higher pile bending moments due to the lack of enough time for the negative pore pressures near the pile to dissipate in the liquefied but still dilative soil (Dungca et al., 2004; Hwang et al., 2004; Gonzalez et al., 2005; Suzuki et al., 2005). This paper constitutes a progress report on a 3-year study (2005-08), aimed at detailed clarification and quantification of the mechanics of soil-pile interaction during lateral spreading, with emphasis on the action of the liquefied soil on the pile foundation. The effort is supported by the US National Science Foundation through the NEES Consortium, NEES Inc. (NEES: George E. Brown, Jr. Network for Earthquake Engineering Simulation). It involves cooperative research between five US universities (University at Buffalo SUNY (UB), Rensselaer Polytechnic Institute (RPI), University of California at San Diego (UCSD), and Tulane University (TU)), as well as Japanese collaborators at Tokyo Institute of Technology (TIT) and the National Research Institute for Earth Science and Disaster Prevention (NIED). The centerpiece of the effort is 1g full scale testing in the large laminar box recently developed at UB, as sketched in Fig. 1. The figure shows two experiments of a stiff pile and a flexible pile, respectively, subjected to lateral spreading induced by shaking at the base of the inclined laminar box containing loose saturated sand. These two experiments, as well as a third baseline free-field experiment of the inclined laminar box without a pile, are scheduled for mid 2007. Two successful liquefaction free field experiments have already been conducted using this laminar box in a horizontal

162

position, simulating liquefaction of level sand deposits, and results are presented at the end of the paper for one of them (Test LG-0). The video of Test LG-0 can be viewed at https://central.nees.org/?projid=122&loc=Public#. Future experiments will be web cast live at http://www.nees.buffalo.edu/projects/NEESPiles/. This 1g testing involves the use of state-of-the-art instrumentation techniques, including an advanced, dense sensor array of MEMS accelerometers, as well as more traditional piezometers, potentiometers, and digital image analyzers used to monitor the time-history of accelerations, excess water pressures, and vertical and lateral spreading within the ground and along the piles, with additional strain gages on the piles monitoring bending moments. These sensors aim at recording dynamic fluctuations as well as permanent changes in the free field, along the piles, and, especially, in the near field soil around the pile where some of the most important and challenging phenomena take place. A 3D Data Viewer originally developed at RPI for centrifuge testing has been adapted to display these responses on a computer screen, allowing the researchers to visually capture the complex response of the soil and pile foundation before and during liquefaction and lateral spreading. Three other parallel efforts are underway to simulate these 1g full scale experiments at UB using: (i) small scale centrifuge model experiments at RPI; (ii) Finite Element Modeling (FEM) computer simulations at UCSD; and (iii) Discrete Element Modeling (DEM) computer simulations at RPI and TU. Comparison between centrifuge and 1g tests should allow clarification of the applicable scaling law(s); the DEM simulations can be considered to be numerical experiments providing additional physical insights; and the FEM simulations should both provide additional insight and serve as a calibrated platform for analysis and design of actual systems in engineering practice. Compatible data bases from the 1g full scale and centrifuge experiments, as well as from the FEM and DEM simulations are fed to the 3D Data Viewer, which is then used by the research team to compare and integrate the various results.

Figure 1. Full scale testing of lateral spreading and its effect on pile foundations, scheduled in 2007 using the new geotechnical laminar box at the University at Buffalo. The laminar box sits on rubber/sliding bearings on the strong floor and is shaken at the base by two fast actuators connected to a reaction wall

RESEARCH TOOLS The research combines two advanced NEES testing facilities at UB and RPI, and advanced computational simulation facilities at UCSD, RPI and TU. Full details are available at three NEES websites (www.nees.rpi.edu, www.nees.buffalo.edu, http://nees.ucsd.edu). Summaries of the tools used in the research are presented below.

163

1g Full-Scale Prototype Testing at UB 1-g Full-scale prototype testing at the UB-NEES experimental site utilizes the new 2D modular laminar box sketched in Fig. 1 (2.75 5 6.2m), two newly acquired 100 ton fast actuators, reaction wall (9.1 m high, 12.5 m wide), strong-floor (close to 300 m2), image processor, and controllers. This facility was developed in 2000-04 with NEES support. The tests utilize state-of-the-art sensors (pore pressure, acceleration, LVDT, laser, fiber optic shape cables), CPT site characterization, and image analysis tools acquired as part of the NEES development. The 2D laminar box is made of 24 laminates, separated by ball bearings to facilitate 2D motions, and has the ability to simulate sloping ground subjected to large deformations. The laminar box is supported by the strong floor which rests on a steel shaking base frame having high capacity ball bearings and shaken by two 100 ton fast actuators mounted on the new UB-NEES reaction wall (Figs. 2a and 2b). The total weight of the box filled with sand is about 150-170 ton, whereas the maximum horizontal dynamic actuator capacity is 90 ton in each horizontal direction simultaneously, or 180 ton in any one direction. Thus very large shaking g levels are possible subjected to safety limitations. The actuators can be fed with an arbitrary target motion, and the controllers compensate for compliance effects so as to accurately shake the base of the soil to meet the desired target motion. Sand is deposited by hydraulic filling (Fig. 2c). Data acquisition systems are available to monitor up to 256 channels at high frequencies. High resolution imaging tools can be positioned to capture deformation patterns at selected zones in the soil box. Interpretation software is available for accurately capturing deformation/strain characteristics at any location. Figure 3a shows the laminar box filled with sand and fully instrumented, during a cone penetration test (CPT) done before the shaking. Figure 3b shows front, side, and top views of the laminar box during the shaking.

(a) (b) (c) Figure 2. 1g full-scale geotechnical laminar box: (a) Base shaker of the geotechnical laminar box; (b) Laminar ring stack; and (c) Hydraulic filling of soil and instrumentation

(a) (b) Figure 3. 1g full-scale geotechnical laminar box – liquefaction experiment: (a) UB researchers setting up instruments and CPT testing before shaking; and (b) Video of UB liquefaction experiment (https://central.nees.org/?projid=122&loc=Public#)

164

Centrifuge Model Testing at RPI The RPI geotechnical centrifuge (Fig. 4) was commissioned in 1989 and started producing 1D earthquake model simulation tests in 1991. Over the years it has produced about 600 documented model experiments of soil and soil-structure systems, with most but not all being earthquake shaking tests. In 2000-04, the centrifuge was upgraded and integrated into the NEES grid with NSF support. This included an increase in capacity from 100 g-ton to 150 g-ton; a new 2D in-flight earthquake shaker (Fig 5a), and 2D laminar box for shaking in the two prototype horizontal directions (Fig 5b); and a 4-degree-of-freedom robot capable of performing in-flight operations such as construction and excavation, pile driving, ground remediation, cone penetration (CPT), and model foundation loading tests without stopping the centrifuge. Also, a model may now be instrumented with dense arrays of advanced sensors (shape-acceleration-arrays and tactile pressure sensors) connected to RPI’s new web-based data acquisition system. Currently, the 3m-radius, 150 g-ton centrifuge facility has two 1D shakers capable of providing in-flight base excitation to several available rigid and laminar (flexible wall) model box containers; the new 2D shaker and 2D laminar box; split box model containers for simulating the effect of large ground displacement on buried pipes; the new 4-degree-of-freedom robot; LVDT, accelerometer, pore pressure, and earth pressure sensors as well as load cells and strain gages, other advanced sensors and a high speed camera and image processing software to monitor model response; data acquisition system with 128+ channels and Internet capability; client-server data acquisition/control software for web-based observation and operation of centrifuge experiments; local area network connected to the high speed RPI network connection making it possible to share experimental data and other information in real time between the RPI research staff and outside parties; a control/teleparticipation and a teleconference room; and model preparation, robot and electronic rooms and other equipment and peripheral facilities.

(a) (b) Figure 4. (a) RPI 150g-ton geotechnical centrifuge; and (b) Tele-control room

(a) (b) Figure 5. RPI geotechnical centrifuge: (a) 2D in flight servohydraulic shaker (two prototype horizontal directions); and (b) 2D laminar container

165

Advanced Sensors A main objective of this investigation is to monitor as closely as possible the 1g and centrifuge shaking tests conducted at UB and RPI, by producing more and better data using advanced instrumentation, thus providing a higher resolution picture (in space and time) of the model response during and after shaking. Test data is collected using suites of traditional state-of-the-art sensors (pore pressure, accelerometers, strain gages, etc.). In addition, advanced sensing techniques as described below are also used for measuring soil and pile response. These new measurements are especially suited to the needs of visualization, system identification, and comparisons with numerical simulations. The two new advanced techniques used in the project are: Shape-acceleration-arrays sensor (1g testing) A novel field shape-acceleration-array (SAA) sensor has been developed, taking advantage of promising new advances in the fiber optic and micro-machined electromechanical sensor (MEMS) technologies. These sensors are capable of simultaneously measuring acceleration and permanent ground deformation down to 30 m depth, at a cost that is 1/10th or 1/20th of existing inclinometer and borehole technologies. In fact, the significance of the new wireless SAA sensor, reside in the type of ground measurement that is recorded. The array is capable of measuring in situ, 3D ground deformation as well as 2D soil acceleration every 0.25 m interval down to a soil depth of 30-40 m. Each sensor array is connected to a wireless sensor node to enable real time monitoring as well as remote sensor configuration (Abdoun and Danisch, 2005; Abdoun et al. 2006, 2007). Figure 6 shows photos of the sensor array placed in the free field and on the pile foundation before placing the soil in a 1g test in Japan. These new sensor arrays are being utilized in the 1g tests at UB.

(a) (b) Figure 6. Shape-Acceleration-Arrays used in a 1g shaking table test in Japan: (a) Sensor array placed in the free field; and (b) Sensor array attached to a pile foundation. Photos courtesy of NIED (Abdoun et al., 2007) High speed camera (centrifuge testing) RPI’s high speed camera is a Phantom 5 capable of recording full digital images at rate of 30,000 frames/sec. The use of high speed camera allows production of a high resolution picture of centrifuge model response. This includes measurement of the ground deformation (and especially the permanent, DC component of this deformation) at enough locations and times during and after shaking. The high speed camera is used in the tests in conjunction with markers at the surface and sides of the model to provide this information at the boundaries of the soil model. Visualization tools and System Identification The sensors used in this project are producing a growing wealth of data which must be integrated and compared with other experimental and analytical data. System identification (SI), visualization (of complete response of soil-foundation system), and a new 3D Data Viewer, are three main tools being used in the project.

166

System Identification System identification techniques provide tools to deduce a model or model parameters of a real system based on its output (response) and input (excitation). Currently, these techniques are enjoying increasing attention in earthquake engineering, in view of substantial enhancements of computational resources as well as experimental data and earthquake records. For this research, the same simple (Zeghal et al., 1995) as well as sophisticated (Zeghal and Oskay, 2003) procedures are being employed to process data from both 1g and centrifuge experiments, so as to allow comparison and ensure consistency of the results. Visualization of Experiments Visualization of Experiments is especially suited to test data measuring the earthquake response of a system as done in this project. Such visualizations are essentially movies illustrating the actual system’s response to the base earthquake shaking. Visualizations of several centrifuge tests have already been implemented at RPI and are being used systematically in the investigation. Figure 7a is a frame taken from one of these "movies" which can be viewed at the RPI-NEES Web site (www.nees.rpi.edu). 3D Data Viewer The 3D Data Viewer (Fig. 7b) was developed at RPI initially for the processing and study of centrifuge results (Radwan and Abdoun, 2006). It is a software especially designed to display both 3D models as well as 2D plots side by side (http://nees.rpi.edu/3dviewer). This tool can be used for simple 3D visualization, or more specifically to organize data from various sources (sensors) which can be represented in 3D space. To organize all these data, a virtual representation of the physical model is displayed and the sensors became interactive elements so the user can select them and plot their data. This aids greatly in organizing the data and getting a good high-level understanding of the test results. As mentioned before, this same 3D Data Viewer has been adopted as a main integrative tool of the whole research project. Compatible data bases are produced for the same 1g UB experiment at the same sensor locations by the four tools of the project (1g tests, centrifuge tests, DEM simulations, and FEM simulations), which are then uploaded to the 3D Data Viewer for side-by-side comparisons and further integration of the results by the team researchers.

(a) (b) Figure 7. Visualization tools: (a) Visualization of RPI liquefaction Model 3 centrifuge experiment (see complete test animation at www.nees.rpi.edu); and (b) 3-D Data Viewer software for visualization of the response of the ground and foundations during strong shaking (download software from http://nees.rpi.edu/3dviewer)

167

Discrete Element Method (DEM) Micro-Mechanical Numerical Simulations A coupled hydro-mechanical model is being employed to analyze the liquefaction and lateral spreading of a saturated deposit of cohesionless particles (Zeghal and El Shamy, 2004), as well as the soil-pile interaction when subjected to a dynamic base excitation. The fluid motion is idealized with averaged Navier-Stokes equations, and the discrete element method (DEM) is used to model the assemblage of solid particles (Fig. 8a), with appropriate fluid-particle interactions. The model is employed to analyze liquefaction of both level and sloping saturated loose deposits of cohesionless particles subjected to dynamic base excitation. Periodic boundaries and a high gravity field are used to reduce the number of particles to a manageable value. The DEM numerical experiments, validated by the physical tests, are starting to provide for the first time detailed information at the level of individual soil grains and groups of grains, on the response of the liquefied soil. It is expected that the upcoming DEM simulations of pile-soil response will be very valuable in this respect, especially for the soil close to the pile foundation, where complicated phenomena take place, including aspects such as dilative stress-strain response, rapid transient water flows, strain localization, shear banding, and flow of the liquefied soil around the pile. New insights provided by the DEM simulations that are validated by the 1g and/or centrifuge results will be used to refine important unsolved aspects of the constitutive relations used in the FEM analyses. Finite Element (FEM) Numerical Simulations An effective-stress solid-fluid finite element computational framework (Program CYCLIC http://cyclic.ucsd.edu/) together with an appropriate plasticity-based pressure-sensitive soil model (Parra, 1996; Elgamal et al., 2003) is used to simulate pile foundations subjected to liquefactioninduced lateral spreading in sloping ground (Fig. 8b). Centrifuge experimental programs conducted at RPI have been a major source of calibration and validation for CYCLIC (Dobry et al., 1995; Taboada, 1995; Elgamal et al., 1996; Dobry and Abdoun, 1998; Elgamal et al., 2002a,b). The 1g large shaketable models conducted at NIED in Tsukuba City, Japan (He et al., 2006), as part of a previous RPIUCSD-NIED US-Japan project supported by the National Science Foundation (NSF), were used for further calibration of CYCLIC to simulate pile foundations. Recent developments funded by NSF in collaboration with Prof. Kincho Law of Stanford University extend CYCLIC to perform Parallel Computations (ParCYCLIC), allowing for high fidelity analyses of lateral spreading effects on piles. RPI centrifuge tests (Abdoun, 1997; Abdoun and Dobry, 2002) have been used so far as a basis of further calibration for displacement, pore pressure and acceleration.

(a) (b) Figure 8. Numerical simulations: (a) DEM micromechanical computer model of sloping saturated sand layer and pile; and (b) FEM mesh of pile in liquefiable soil model (different colors show mesh partitioning for computation on 16 processors)

168

FULL SCALE LIQUEFACTION TEST AT UB (LG-0) Following a successful shaking 3m demonstration test at UB, one full-scale laminar box shaking test was conducted to evaluate the response of a level deposit of loose saturated sand in the free field when subjected to strong shaking motion (Thevanayagam and Ecemis, 2007). The total depth of soil in the laminar box in Test LG-0 was 4.9 m. The soil was fully instrumented with a MEMS shape accelerometer array (SAA), traditional accelerometers, piezometers, and potentiometers. Potentiometers measure displacement of a contact point, and are used to measure horizontal and vertical displacement of selected locations on the top of the ground and along the laminar box side wall. Saturated Ottawa sand with D10 = 0.161 mm and D50 = 0.230 mm was used. The average permeability for the relevant range of void ratios, e = 0.61-0.66, is k = 1×10-5 m/s (Shenthan, 2001). The base input acceleration was essentially sinusoidal with a frequency of 2 Hz, containing 4 phases of increasing acceleration amplitude: (i) 0.01g (non-destructive phase) with a duration of 5 sec; (ii) 0.05g with a duration of 10 sec; (iii) 0.15g with a duration of 10 sec; and (iv) 0.30g with a duration of 10 sec. Figure 9a shows the accelerations measured during the non-destructive phase and beginning of the second, 0.05g phase (when the soil liquefied in the first few cycles). Figure 9b presents the shear stress-strain hysteretic response in the soil evaluated using system identification (Elmekati and Zeghal, 2007) for the non-destructive phase (i.e., first 5 seconds of shaking during which only a small pore pressure buildup took place). Figure 10 shows the input shaking as well as the excess pore pressures measured in the soil. A video of this experiment can be seen at https://central.nees.org/?projid=122&loc=Public#.

(a) (b) Figure 9. Results from full-scale laminar box 1g shaking Test LG-0: (a) Acceleration time histories recorded at soil and rings during non-destructive phase (0-5 sec); and (b) 1D shear stress-strain hysteretic response in the soil during non-destructive phase, obtained using SI from ring acceleration time histories The accelerations during Test LG-0 were recorded using three accelerometer systems. Two of these systems recorded respectively outside the laminar box and inside the soil, using submersible traditional accelerometers, while the third system consisted of the advanced SAA discussed before in the paper. The comparison between the records from the three systems in Fig. 9a indicates good

169

agreement in the first 5 seconds, that is, during the non-destructive, 0.01g shaking. Shortly after the second phase of 0.05g started at 5 seconds, the deposit liquefied and the acceleration records in Fig. 9a diverged, probably due to rotation of the traditional accelerometers inside the soil following liquefaction. This phenomenon of sensor rotation has been previously reported in the literature for centrifuge experiments (Adalier, 1996). Thus, ring and soil accelerometers show different behavior after liquefaction. The shear stress-strain loops in Fig. 9b, obtained by SI processing of the acceleration records from the rings in the non-destructive phase, are being used to evaluate the dynamic properties of the sand deposit. The loops become steeper at greater depths, indicating (as expected) an increase in shear modulus with depth within the sand deposit. Figure 10 includes comparisons between FEM predictions and measurements for Test LG-0 (Elgamal and Forcellini, 2007), with reasonable agreement. While the comparison of accelerations in Fig. 10a is restricted to the first 7 seconds, the pore pressure comparisons in Fig. 10b cover the whole shaking. Both measurements and FEM predictions indicate that liquefaction was induced by the second, 0.05g phase of the shaking, but with the experimental data showing a higher rate of pore pressure build-up and an earlier onset of liquefaction than predicted by the FEM simulation.

(a) (b) Figure 10. Comparison between experimental recorded data and FEM simulation: (a) Acceleration time histories during non-destructive phase (0-5 sec time window); and (b) Pore pressure time histories during whole shaking Efforts are currently under way in the project to complete the simulation of 1g level test LG-0 using both DEM computer analysis (El Shamy and Zeghal, 2007), and centrifuge model testing (Abdoun and Gonzalez, 2007).

CONCLUSIONS Critical NEES research is under investigation on the problem of pile foundations subjected to liquefaction-induced lateral spreading, with emphasis on the action of the liquefied soil on the piles. The complex nature of the problem requires a fundamental approach using several available experimental and analytical tools, The research includes use of the UB 1g 6m-tall laminar box container NEES facility, and the RPI NEES centrifuge facility; use of novel advanced sensors to

170

measure soil accelerations, deformations and pressures during shaking; and micromechanical discrete element modeling (DEM) and finite element modeling (FEM) analyses at RPI, U. of California, San Diego, and Tulane University; all integrated by an appropriate identification and analysis framework including system identification and visualization capabilities.

AKNOWLEDGEMENTS This material is based upon work supported by the National Science Foundation under NEESR-SG Grant No. 0529995. This support is gratefully acknowledged. The authors wish to thank the staff members of the UB-NEES Site: Thomas Albrechcinski, Andrei Reinhorn, and Mark Pitman; and of the RPI-NEES Site: Inthuorn Sasanakul, and Javier Ubilla, for their valuable support in the 1g and centrifuge testing at these two NEES sites. The authors wish to thank NEES IT staff members Hassan Radwan and Jason Hanley for their contribution to the visualization tools. Special thanks are extended to graduate students Raghudeep Bethapudi, Victoria Bennett, Tewodros Dessalegn, Nurhan Ecemis, Ahmed Elmekati, Davide Forcellini, Marcelo Gonzalez, Peng Hao, and Thangalingam Kanagalingam for their dedication, creativity, and enthusiasm in conducting this research. Finally, previous 1g testing shaking experiences in Japan and ideas for this research provided by Drs. Akio Abe, Masayoshi Sato, and Kohji Tokimatsu, are also most gratefully acknowledged.

REFERENCES Abdoun, T., “Modeling of Seismically Induced Lateral Spreading of Multi-layered Soil and its Effect on Pile Foundations,” Ph.D. Thesis, Rensselaer Polytechnic Institute, Troy, NY, 1997. Abdoun, T., and Danisch, L., “Advanced Sensing for Real-Time Monitoring of Geotechnical Systems,” Proc Geo-Frontiers 2005 Conference, Austin, TX, January 24-26, 2005. Abdoun, T., and Dobry, R., “Evaluation of Pile Foundation Response to Lateral Spreading,” Soil Dyn Earthq Eng, Vol. 22, No. 9, pp. 1051-1058, 2002. Abdoun, T., and Gonzalez, M., “Centrifuge Model Test on Pile Foundation Subjected to LiquefiedInduced Lateral Spreading: Calibration Factors,” Internal Progress Report, Rensselaer Polytechnic Institute, Troy, NY, 2007. Abdoun, T., Dobry, R., O’Rourke, T.D., and Goh, S.H., “Pile Response to Lateral Spreads: Centrifuge Modeling,” J Geotech Geoenviron Eng, Vol. 129, No. 10, pp. 869-878, 2003. Abdoun, T., Danisch, L., Ha, D., and Bennett, V., “Advanced Sensing for Real-Time Monitoring of Geotechnical Systems,” TRB 85th Annual Meeting, Washington, D.C., January 22-26, 2006. Abdoun, T., Abe, A., Bennett, V., Danisch, L., Sato, M., Tokimatsu, K., and Ubilla, J., “Wireless Real Time Monitoring of Soil and Soil-Structure Systems,” GSP Proc GeoDenver 2007 Conference, Denver, CO, February 18-21, 2007. Adalier, K., “Mitigation of Earthquake Induced Liquefaction Hazards,” Ph.D. Thesis, Rensselaer Polytechnic Institute, Troy, NY, 1996. Ashford, S.A., Juirnarongrit, T., Sugano, T., and Hamada, M., “Soil-Pile Response to Blast-Induced Lateral Spreading. I: Field Test,” J Geotech Geoenviron Eng, Vol. 132, No. 2, pp. 152-162, 2006. Berrill, J.B., Christensen, S.A., Keenan, R.J., Okada, W., and Pettinga, J.K., “Lateral-Spreading Loads on a Piled Bridge Foundation,” Seismic Behavior of Ground and Geotechnical Structures, (Seco E. Pinto, ed.), Balkema, Rotterdam, pp. 176-183, 1997. Bhattacharya, S., Madabhushi, S.P.G., and Bolton, M.D., “An Alternative Mechanism of Pile Failure Deposits During Earthquakes,” Geotechnique, Vol. 54, No. 3, pp. 203-213, 2004. Brandenberg, S.J., Boulanger, R.W., Kutter, B.L., and Chang, D., “Behavior of Pile Foundations in Laterally Spreading Ground During Centrifuge Tests,” J Geotech Geoenviron Eng, Vol. 131, No. 11, pp. 1378-1391, 2005. Brandenberg, S.J., Boulanger, R.W., Kutter, B.L., and Chang, D., “Liquefaction-Induced Softening of Load Transfer Between Pile Groups and Laterally Spreading Crusts,” J Geotech Geoenviron Eng, Vol. 133, No. 1, pp. 91-103, 2007.

171

Dobry, R., and Abdoun, T., “Post-triggering Response of Liquefied Sand in the Free Field and Near Foundations,” Proc ASCE GEESD, pp. 270-300, Seattle, WA, August 3-6, 1998. Dobry, R., and Abdoun, T., “Recent Studies on Seismic Centrifuge Modeling of Liquefaction and its Effect on Deep Foundation,” State-of-the-Art Paper, Proc 4th Int Conf on Recent Advances in Geotech Earthq Eng and Soil Dyn (S. Prakash, ed.), Paper SOAP 3, Vol. 2, 30 pages, San Diego, CA, March 26-31, 2001. Dobry, R., Taboada, V., and Liu, L., “Centrifuge Modeling of Liquefaction Effects During Earthquakes,” Proc 1st Int Conf on Earthq Geotech Eng (IS-Tokyo), Keynote Lecture, pp. 12911324, Tokyo, Japan, 1995. Dobry, R., Abdoun, T., O’Rourke, T.D., and Goh, S.H., “Single Piles in Lateral Spreads: Field Bending Moment Evaluation,” J Geotech Geoenviron Eng, Vol. 129, No. 10, pp. 879-889, 2003. Dungca, J.R., Kuwano, J., Saruwatari, T., Izawa, J., Suzuki, H., and Tokimatsu, K., “Shaking Table Tests on the Lateral Response of a Pile Buried in Liquefied Sand,” Proc 11th Intl Conf on Soil Dyn and Earthq Eng, Vol. 2, pp. 471-477, 2004. El Shamy, U., and Zeghal, M., “DEM Class A Prediction of UB Level Ground Shaking Table Tests,” Internal Progress Report, Tulane University, New Orleans, LA; and Rensselaer Polytechnic Institute, Troy, NY, 2007. Elgamal, A., and Forcellini, D., “Experimental and Computational Study of Pile Foundations Subjected to Liquefaction-Induced Lateral Spreading: Simulation of Test LG-0,” Internal Progress Report, University of California, San Diego, CA, 2007. Elgamal, A., Zeghal, M., Taboada, V., and Dobry, R., “Analysis of Site Liquefaction and Lateral Spreading Using Centrifuge Testing Records,” Soils and Foundations, JGS, Vol. 36, No. 2, pp. 111-121, 1996. Elgamal, A., Parra, E., Yang, Z. and Adalier, K., “Numerical Analysis of Embankment Foundation Liquefaction Countermeasures,” J Earthq Eng, Vol. 6, No. 4, pp. 447-471, 2002a. Elgamal, A., Yang, Z., and Parra, E., “Computational modeling of Cyclic Mobility and PostLiquefaction Site Response,” J of Soil Dyn and Earthq Eng, Vol. 22, Issue 4, pp. 259-271, 2002b. Elgamal, A., Yang, Z., Parra, E., and Ragheb, A., “Modeling of Cyclic Mobility in Saturated Cohesionless Soils,” Int J of Plast, Vol. 19, Issue 6, pp. 883-905, 2003. Elmekati, A., and Zeghal, M., “Systen Identification Analysis of Test LG-0,” Internal Progress Report, Rensselaer Polytechnic Institute, Troy, NY, 2007. Gonzalez, L., Abdoun, T., and Dobry, R., “Effect of Soil Permeability on Centrifuge Modeling of Pile Response to Lateral Spreading,” ASCE GSP No. 145 (Boulanger and Tokimatsu, eds.), pp. 50-60, 2005. Haigh, S.K., and Madabhushi, S.P.G., “The Effect of Pile Flexibility on Pile-Loading in Lateral Spreading Slopes,” ASCE GSP No. 145 (Boulanger and Tokimatsu, eds.), pp. 24-37, 2005. Hamada, M., and O’Rourke, T.D. (eds.), “Case Studies of Liquefaction and Lifeline Performance During Past Earthquakes. Vol. 1: Japanese Case Studies,” NCEER, SUNY-Buffalo, Buffalo, NY, Tech. Rept. NCEER-92-0001, February, 1992. Hamada, M., Yasuda, S., Isoyama, R., and Emoto, K., “Study on Liquefaction Induced Permanent Ground Displacements,” Research Rept., Assn. for Development of Earthquake Prediction, Japan, 87 pages, November, 1986. He, L., Elgamal, A., Abdoun, T., Abe, A., Dobry, Meneses, J., Sato, M., and Tokimatsu, K., “Lateral Load on Piles Due to Liquefaction-Induced Lateral Spreading During 1G Shake Table Experiments,” Proc., 8th U.S. National Conf on Earthq Eng, Paper No. 881, San Francisco, CA, April 18-22, 2006. Hwang, J.I., Kim, C.Y., Chung, C.K., and Kim, M.M., “Behavior of a Single Pile Subjected to Flow of Liquefied Soil in an Infinite Slope,” Proc 11th Int Conf on Soil Dyn and Earthq Eng, Vol. 2, pp. 573-580, 2004. Ishihara, K., Yasuda, S., and Nagase, H., “Soil Characteristics and Ground Damage,” Soils and Foundations, JGS, Special Issue: January, pp. 109-118, 1996. Koyamada, K., Miyamato, Y., and Tokimatsu, K., “Field Investigation and Analysis Study of Damage Pile Foundation During the 2003 Tokachi-Oki Earthquake,” ASCE GSP No. 145 (Boulanger and Tokimatsu, eds.), pp. 97-108, 2005.

172

Lin S., Tseng, Y., Chiang, C., and Hung, C., “Damage of Piles Caused by Lateral Spreading-Back Study of Three Cases,” ASCE GSP No. 145 (Boulanger and Tokimatsu, eds.), pp. 121-133, 2005. McCulloch, D.S., and Bonilla, M.G., “Effects of the Earthquake of March 27, 1964 on the Alaska Railroad,” Professional Paper 545-D, U.S. Geological Survey, 1970. Mizuno, H., “Pile Damage During Earthquakes in Japan (1923-1983),” Proc ASCE Session on Dynamic Response of Pile Foundations (T. Nogami ed.), pp. 53-77, Atlantic City, April 27, 1987. O’Rourke, T.D., and Hamada, M. (eds.), “Case Studies of Liquefaction and Lifeline Performance During Past Earthquakes. Vol. 2: United States Case Studies,” NCEER, SUNY-Buffalo, Buffalo, NY, Tech. Rept. NCEER-92-0002, February, 1992. Parra, E., “Numerical Modeling of Liquefaction and Lateral Ground Deformation Including Cyclic Mobility and Dilation Response in Soil Systems,” Ph.D. Thesis, Rensselaer Polytechnic Institute, Troy, NY, 1996. Radwan, H., and Abdoun, T., “3D Data Viewer of Centrifuge Physical Models,” Geotech Special Publ Proc GeoCongress 2006 Conference, Atlanta, GA, February 26-March 1, 2006. Shenthan T., “Factors Affecting Liquefaction Miitigation in Silty Soils Using Stone Columns,” MS Thesis, University at Buffalo, SUNY, NY, p. 129, 2001. Suzuki, H., Tokimatsu, K., Sato, M., and Abe, A., “Factor Affecting Horizontal Subgrade Reaction of Piles During Soil Liquefaction and Lateral Spreading,” ASCE GSP No. 145 (Boulanger and Tokimatsu, eds.), pp. 1-10, 2005. Swan S.W., Flores, P.J., and Hooper, J.D., “The Manzanillo Mexico Earthquake of October 9, 1995,” NCEES Bulletin, The Quarterly Publication of NCEER, Vol. 10, No. 1, January, 1996. Taboada, V.M., “Centrifuge Modeling of Earthquake-Induced Lateral Spreading in Sand Using a Laminar Box,” Ph.D. Thesis, Rensselaer Polytechnic Institute, Troy, NY, 1995. Tamura, S., and Tokimatsu, K., “Seismic Earth Pressure Acting on Embedded Footing Based on Large-Scale Shaking Table Tests,” ASCE GSP No. 145 (Boulanger and Tokimatsu, eds.), pp. 8396, 2005. Thevanayagam, S., and Ecemis, N., “NEESR-SG Project – Experimental and Micromechanical Computational Study of Pile Foundations Subjected to Liquefaction Induced Lateral Spreading,” Internal Progress Report, University at Buffalo, SUNY, Buffalo, NY, 2007. Tokimatsu, K., “Performance of Pile Foundations in Laterally Spreading Soils,” Proc 2 nd Int Conf on Earthq Geotech Eng (Seco E. Pinto, ed.), Vol. 3, pp. 957-964, Lisbon, Portugal, June 21-25, 1999. Tokimatsu K., Mizuno, H., and Kakurai, M., “Building Damage Associated with Geotechnical Problems,” Soils and Foundations, JGS, Special Issue: January, pp. 219-234, 1996. Towhata, I., Sesov, V., Motamed, R., and Gonzalez, M., “Model Tests on Lateral Earth Pressure on Large Group Pile Exerted by Horizontal Displacement of Liquefied Sandy Ground,” Proc 8th U.S. National Conf on Earthq Eng, Paper No. 1227, San Francisco, CA, April 18-22, 2006. Yokoyama K., Tamura, K., and Matsuo, O., “Design Methods of Bridge Foundations Against Soil Liquefaction and Liquefaction-Induced Ground Flow,” Proc 2 nd Italy-Japan Workshop on Seismic Design and Retrofit of Bridges, pp. 109-131, Rome, Italy, February 27-28, 1997. Youd, T.L., “Liquefaction-Induced Damage to Bridges,” Transportation Research Record, published by the TRB and the National Research Council, No. 1411, pp. 35-41, Washington, D.C., 1993. Zeghal, M., and El Shamy, U., “A Continuum-Discrete Hydromechanical Analysis of Granular Deposit Liquefaction,” Int J Numer Anal Meth Geomech, Vol. 28, No. 14, pp. 1361-1383, 2004. Zeghal, M., and Oskay, C., “A Local Identification Technique for Geotechnical and Geophysical Systems,” Int J Numer Anal Meth Geomech, Vol. 27, No. 11, pp. 967-987, 2003. Zeghal, M., Elgamal, A.W., Tang, H.T., and Stepp, J.C., “Lotung Downhole Array. II: Evaluation of Soil Nonlinear Properties.” J of Geotech Eng, Vol. 121, No. 4, pp. 363-378, 1995.

173

4 th International Conference on Earthquake Geotechnical Engineering June 25-28, 2007 Paper No. W1-1010

TRENDS AND OPPORTUNITIES IN THE FURTHER USE AND DEVELOPMENT OF THE EU LARGE RESEARCH INFRASTRUCTURES FOR EARTHQUAKE ENGINEERING Michel Géradin and Fabio Taucer ABSTRACT As stated in the European Strategic Research Agenda presented by the Earthquake Engineering community, the research institutions operating the European large testing facilities for Earthquake Engineering as well as their users community face the need to better integrate their research activities and make a coherent use and development of infrastructures. A coordinated and collaborative research strategy has to be adopted to take full advantage of the continuous advance in the knowledge of the associated methodologies and in the technologies to implement them. In particular, information technologies allow evolution towards global connectivity to support interoperability, hybrid and distributed testing, to ensure remote access of the laboratories and to provide access to a common data base of experiments and results. If properly achieved, such evolution will greatly facilitate international collaboration as developed in the context of the different working groups created under the umbrella of the World Forum on Collaborative Research in Earthquake Engineering. It will also provide industry and SMEs easier access to performing testing infrastructures. Following this strategy will imply making best use of the specific instruments provided by the FP7 Capacities Programme. .

174

4th International Conference on Earthquake Geotechnical Engineering June 25-28, 2007 Paper No.W1-1006

VISUALIZATION OF LARGE-SCALE SEISMIC DATA RECORDS Falko Kuester 1, Tara C. Hutchinson2, Tung-Ju Hsieh 3 ABSTRACT Visualization of data, whether it is measured or from numerical simulations, is key to allow a clear interpretation of physical processes. Learning and understanding is substantiated by orders of magnitude if crisp visualization of an event, data, or other information, is provided to the user. In this paper, measured data from multiple scales (global, local, and the building-level) is used to demonstrate a variety of visualization methods, each of which provide the user with the ability to maximize their understanding of its underlying physical mechanisms. These scales encompass the use of field acquired seismic measurements, digital elevation data, and building-level seismic sensors. The concept of visual analytics is discussed in the context of these datasets, and demonstrated via data presentation on a variety of hardware interfaces, including a massive high resolution display system and an immersive virtual reality system. Keywords: Visualization, display systems, virtual reality, seismic sensors, digital elevation maps

INTRODUCTION Large-scale data acquisition, computing and networking initiatives are providing substantial resources for the modeling, simulation, measurement, and analysis of complex geotechnical and structural systems. The result is an overwhelming amount of scientific data being created at increasingly shorter time intervals and ever-higher resolution. For example, massively networked, sensor grids provide instantaneous access to time varying multivariate records capturing seismicity at the global and local level. These data sources can then be further augmented with data from airborne imaging systems delivering topology and topography data at centimeter resolution, forming geo-spatially referenced records, capturing pre-event, event and post-event information. To cope with data of a complexity and detail that greatly exceeds the capabilities of traditional 2D plots for 3D data expression and analysis, a new generation of visual analytics techniques, tools and infrastructure are required that allow researchers to collaboratively view, interrogate, correlate and manipulate data at a broad range of temporal and spatial scales and resolutions. The objective of visual analytics is to enable data analysis and hypothesis generation by means of interactive data visualization and fusion. The underlying idea is that collaborative visual comparison of theory and practice, by teams of engineers, will allow new insights to be derived more rapidly. In this paper, a set of processing and visualization strategies for seismic records is described, that leverage high-resolution display walls and virtual reality environments specifically designed to support visual analytics. Application examples considering the global, local, and building-scales are provided to demonstrate the potential of these tools. 1

Associate Professor, Departments of Structural Engineering and Computer Science and Engineering, University of California, San Diego, CA, USA. Contact: [email protected] 2 Associate Professor, Department of Structural Engineering, University of California, San Diego, CA, USA. 3 Postdoctoral Researcher, Calit2, University of California, San Diego, CA, USA

175

RELATED WORK A conventional approach for the analysis of field-collected data sets is to study earthquake induced effects as time histories by plotting and comparing two-dimensional (2D) discrete waveforms. The frequency characteristics can then be studied by overlaying different plots and evaluating ratios and relationships. However, when properly geo-referenced and treated, seismic data sets can be presented in a more natural and intuitive form that facilitates the understanding of governing events and mechanisms. In particular, three-dimensional (3D) display can provide additional visual clues via an intuitive representation of the time-varying wave propagation that is difficult to represent with traditional 2D techniques. When combined with an immersive display environment, the spatial relationships may be further strengthened. Nayak et al. (2003) used 3D glyphs, graphics primitives or symbols with various geometric and color attributes, to represent the measured seismic data. These glyphs were rendered in real time and combined with a 3D topography terrain map. An interactive 3D visualization system was developed by (Hsieh et al., 2003b) to facilitate the intuitive analysis of field measured seismic data sets. Field collected seismic data sets were rendered in 3D and geo-referenced with Digital Elevation Models (DEMs), and subsequently displayed in a monoscopic or stereoscopic display environment. In another example (Hutchinson et al., 2005) presented a visualization system used to enhance the understanding of the time-varying movements of a case study building subjected to the 1994 Northridge earthquake. Simulation of seismic wave propagation characteristics of a region can further aid with estimating potential damage caused by hypothetical or actual earthquakes. For example, the ground motion of the 1906 San Francisco earthquake was simulated by (Mavroeidis & Papageorgiou, 2001) and they concluded that intense horizontal ground displacements were developed along the entire length of the Saint Andreas fault plane that ruptured during the earthquake for the San Francisco Bay Area. Their simulation results were plotted on a 2D map using vectors at each grid cell, pointing into the direction of displacement. A series of grayscale images, taken at equal time intervals, was produced indicating the evolution of the ground displacement field over time. These snapshots show the qualitative value, yet reveal no other geological information. Another Simulation of wave propagation for the San Francisco Bay Area using a numerical finite difference method was performed by (Antolik el al. 1997). One result was a 90-second video showing wave propagation overlapped with a 2D contour map. The result of 3D simulation was displayed on a 2D surface plane with red color representing positive amplitudes and blue color representing negative amplitudes. Another larger scale simulation was conducted by (Komatitsch et al., 2003) to simulate wave propagation resulting from large earthquakes. An earth-scale 5.5-billion-grid-point (average grid surface spacing of 2.9 km) crustal model was constructed and simulated on a 1944-processor supercomputer. The obtained results were visualized as seismic waveforms plotted on a 2D world map. In order to fully explore the results of high-resolution simulation runs in combination with field collected data, new rendering algorithms are needed that fuse different data sources to provide a geospatially and temporally anchored, visual representation of the studied physical phenomenon. Akcelik et a. (2003) used a 3000-processor supercomputer to simulate an earthquake in the Los Angeles Basin using one billion grid points. This data was subsequently used to develop parallel rendering algorithm capable of rendering time-varying large-scale earthquake simulation results in real-time. Ma et al. (2003) rendered the earth crust volume data semi-transparent allowing viewers to see through the volume, revealing 3D seismic wave propagation originating from the hypocenter of a simulated earthquake event. Compared to conventional 2D approaches, volume rendering in this case provides a perspective viewpoint, allowing the simulated data to be explored in the spatial and temporal domain. These visual paradigms provide an enhanced understanding of the field measured seismic data sets. Instead of just focusing on the local scale, this paper presents visualization techniques that allow scientists to study seismic data across different spatial (global, local and building-level) and temporal scales.

176

CASE STUDY: SEISMIC RECORDS AT THE GLOBAL, LOCAL, AND BUILDING LEVEL Throughout the past century, millions of earthquakes have been recorded worldwide, processed (e.g. (NEIC, 2007)) and studied to construct increasingly accurate analytical and numerical models. The development of these improved models requires a careful and thorough understanding of historic earthquake events and their effects. High-quality earthquake field data in combination with topological and geological structures and new exploration techniques are of utmost importance for a better understanding of the distribution of earthquakes in space, time, and magnitude. Recent deployment of seismograph networks and the growth in number of installed digital seismic measuring instruments has led to a rapid increase in the size of seismic data sets (Hudnut et al, 2001). These massive field measured data sets, consisting of digitized seismic waveforms, are now readily available to the public for download in near real-time. While visualization strategies for simulated data have been explored in the past, real-time visualization of unstructured field collected earthquake data is a largely unexplored research area. Global-Scale The United States Geological Survey (USGS) provides access to DEMs representing elevation information in a raster grid of regularly spaced intervals. The DEMs are available from the USGS Earth Resources Observation Systems (EROS, 2007) Data Center. The GTOPO30 (GTOPO30, 2007) DEMs, with a horizontal grid spacing of approximately 1-kilometer, are used in this study. In addition, the used historical earthquake records were obtained from NEIC. As part of the USGS Earthquake Hazards Program, NEIC provides an extensive seismic database for scientific research compiled from records of global seismograph networks. The obtained earthquake information includes position, depth, magnitude, and date/time of a seismic event, i.e. the description of its hypocenter. For this study, a total of 500,000 hypocenter records were used. Local-Scale In dense urban regions, depending of course upon the ground motion, site, and structure characteristics, even a magnitude 4.0 earthquake may result in significant losses in widespread areas. Beyond the tragic loss of life, important civil infrastructure such as buildings, dams, and bridges may be damaged or destroyed and critical lifeline systems such as power grids, water and gas lines, interrupted. California in particular is classified as one of the areas with the highest earthquake activity in the United States. For the past three decades, the California Strong Motion Instrumentation Program (CSMIP) has been installing accelerometers at various representative locations throughout California, making it one of the most extensively monitored states in the nation. Currently, more than 900 stations are operational, including 650 ground-response, 170 building, 20 dam and 60 bridge stations. The corresponding data records are maintained by the Consortium of Organizations for Strong-Motion Observation Systems (COSMOS, 2007). The objective of COSMOS is to promote the advancement of strong-motion measurement on the ground and in structures in densely populated urban areas likely to be struck by future earthquakes and to develop national policies for the improvement of strong-motion earthquake measurements and their application. Among the readily available records are those of stations that recorded the characteristics of the 1994 Northridge earthquake. The 1994 Northridge earthquake, with a magnitude 6.7, was the most damaging earthquake in the United States since the San Francisco Earthquake of 1906 and its cumulative damage to public property and California's infrastructure was estimated to be \$20 billion U.S. (Aurelius, 1994; EERI, 1995). However, the earthquake also generated a wealth of sensor records that when combined with other historic events, provide statistical data that contributes towards the improvement and development of seismic design guidelines. Augmenting these data sets with urban planning networks such as telecommunication or power grids may then also assist in pre- and post-event planning and hazard mitigation.

177

Building-Scale Temporal loading plays a large role for the extreme loading conditions civil engineers must design for. The studied building is a seven-story reinforced concrete structure located in central San Fernando Valley, California. Numerous seismic sensors were placed throughout the building and recorded its performance during the 1994 Northridge Earthquake. The current system of accelerometers includes one measuring the vertical component of motion on the first floor, ten measuring the north-south direction on floors one, two, three, six, and the roof, and five measuring the east-west direction on floors one, two, three, six, and the roof level. Figure 1 shows the instrumentation layout throughout the structure. The Northridge earthquake caused significant structural and nonstructural damage to this building, rendering it unsafe for occupancy (Trifunac et al., 1999). All exterior columns in the longitudinal direction of the building on the fourth floor developed large shear cracks. On the South facing wall, there were wide shear cracks in the columns. Analysis of response records concur with visual observations, indicating that the building behaved roughly in the nonlinear range during the earthquake.

Figure 1. Sensor locations throughout the studied building.

VISUALIZATION STRATEGIES Even though these data records provide spatial and temporal details, which make them suitable for fusion with high-resolution digital elevation models, satellite imagery, and inventory data to provide geo-spatially and temporally anchored information, conventional data analysis techniques are commonly still limited to 2D plots and maps. In the past, this limitation was a function of the processing and rendering performance of commodity PCs. However, with the recent performance increases of graphics processing units (GPUs) on today's graphics cards, new hardware accelerated technique for data exploration are viable. Real-time interactive visualization of large-scale terrain models tended to be challenging because of the huge amount of polygons that have to be rendered. Utilizing adaptive out-of-core techniques (Duchaineau et al., 1997), which load data from external storage as needed, in combination with hardware accelerated rendering, complex terrain meshes can now be rendered at interactive frame rates and augmented with detailed texture maps. A variety of 2D images, such as aerial photos, satellite images, and inventory data in image form, can be seamlessly mapped onto the terrain surface and programmable graphics hardware used to move selected geometric computations from the CPU to GPU, increasing the rendering speed (Amor et al., 2005). Methodology Interactive visualization is at the heart of any visual analytics environment, requiring a seamless data processing, fusion, visualization and interrogation framework. For the presented visualization of seismic records, pipeline stages include, (i) terrain mesh creation and adaptive refinement, (ii) data

178

augmentation and (iii) visualization of static and time-varying seismic records. Figure 2 illustrates the building blocks of this visualization pipeline. These components are described in detail in the following sections.

Figure 2. The components of the visualization pipeline. Adaptive 3D Rendering and Terrain Mesh Refinement Extensive high-resolution digital elevation models are available today, in combination with aerial photographs and satellite imagery, providing a means for the creation of an accurate 3D model on a global or local scale. In this study, USGS GTOPO30 DEMs, with a horizontal grid spacing of approximately 1-kilometer and complexity of 43,200x21,600 pixels were used in combination with satellite imagery and color-coded maps indicating terrain elevation. Interactive rendering of these terrain meshes at full resolution is still beyond the capability of most commodity graphics hardware, however, it is possible to create close approximations with good visual quality. One solution is to approximate the original terrain mesh with a coarser mesh adaptively as the viewpoint changes by selecting a subset of the vertices from the full resolution height field. In view-dependent mesh refinement, the construction process is performed at run-time. A modified version of the SOAR terrain rendering engine (Lindstrom & Pascucci, 2002) was used for the construction of the terrain. The mesh is constructed adaptively at each frame such that the flat neighborhoods and distant areas are triangulated more coarsely than close and high-detail neighborhoods. Figure 3 shows the snapshots of an adaptively rendered terrain mesh. In order to articulate terrain variations on a global scale, terrain elevation is exaggerated by a factor of 90.

(a) (b) (c) Figure 3. Level-of-detail terrain model rendering and texture mapping. (a) A coarse terrain mesh with 4,143 triangles and texture mapping enabled, (b) coarse terrain mesh without textures, and (c) a refined terrain approximation using 133,984 triangles.

179

Terrain Texture Mapping Textures can be defined as rectangular arrays of data, such as aerial photos, terrain height fields, and ground motion data. Texture mapping is a computer graphics technique that applies an image to the surface of a polygon. A rectangular texture can be mapped to nonrectangular polygons by indicating how the texture is to be applied to the vertices that define the polygons. Figure 4 shows an example of a 2D texture map using a cylindrical projection of the earth's surface obtained from NASA Goddard Space Flight Center. At one kilometer resolution (43,200x21,600 pixels), it is one of the most detailed, continuous and freely available, true-color image of the Earth to date.

(a) (b) (c) Figure 4. Screenshots of EQVis: 3D rendering of the DEMs and NEIC earthquake data sets at global scale, (a) global 3D seismicity map of 500,000 hypocenters, (b) shaded terrain model, (c) 3D terrain texture mapped with satellite image for North America plate, South American plate, and Caribbean plate. Rendering Static Seismic Data Sets Hypocenters from the NEIC database are rendered as spherical proxies, and color-coded based on depth. Different glyphs can also be used to represent other components of hypocenter records such as event data/time, location, depth and magnitude in a geo-referenced context. In addition, detailed 3D surface and surface level information can be shown, to provide important features to the viewer. Figure 5 illustrates different data representation modes. Figure 5 (a) was provided by NEIC for earthquake records from Alaska (Latitude = 60 to 66 North, longitude = 144 to 156 West). The total number of earthquakes recorded in this area was 14,876. The difference with respect to visual impression is significant when the same data set is studied in 3D as shown in Figure 5 (b). The observed viewing volume in this example is 29.5 million km3. A spherical proxy is rendered at the earthquake hypocenter location, defining the longitude, latitude and depth of the event. The proxy radius represents the magnitude of the event and the color the earthquake year. In addition, detailed 3D surface and aboveground level information is shown to provide context and orientation to the viewer in respect to important surface features. Correlations with epicenters (hypocenters projected to the terrain surface) are optionally encoded into the surface texture map, and geometrically connected. Epicenter information alone would fail in revealing the spatial characteristics of the earthquake event as shown in Figure 5(a). When viewed in an interactive environment, spatial correlation is straight forward as users navigate through the data and view it from arbitrary viewpoints. Figure 6 shows a 3D rendering of hypocenters in the greater Anchorage area. Interactive rendering of the historic hypocenter data provides the user with spatial and temporal relationships and can aid in the visual identification of fault plane locations.

180

(a) (b) Figure 5. (a) 2D earthquake epicenter map for the greater Anchorage area. (Latitude: 60 to 66 North, Latitude: 144-156 West). Source NEIC. (b) Example of USGS NEIC earthquake data sets for the Alaska area, combining terrain and static earthquake data sets. (Latitude: 60 to 66 North, Latitude: 144-156 West, 1973-2006).

(a) (b) (c) (d) Figure 6. 3D renderings of Alaska DEMs and hypocenters viewed from different directions. Imposters vs. 3D Proxies In order to interactively render complex earthquake data sets in combination with high-resolution digital elevation maps and auxiliary reference information, substantial resources are required. Since it is desirable to have this technology available on standard computational platforms, optimized algorithms and rendering techniques have to be employed to achieve the desired results. Since welldefined geometric proxies are used to represent earthquake hypocenters, rendering techniques based on imposters (billboards or depth sprites) can be used to achieve even better performance gains. These techniques are particularly popular in flight simulation, in which static objects such as trees can be approximated through simple images (textures) mapped onto a single viewpoint aligned polygon. The challenge for this particular application is that the system has to handle tens to hundreds of thousands of billboards. For the presented data, billboards and depth sprites are particularly suited to replace the rotational symmetric spherical proxies of hypocenters, which allow rotation over an arbitrary axis through the proxies' origin without the need to change the texture image representing it. Since all hypocenter proxies have the same shape, only a single texture is needed.

181

Rendering Dynamic (Time-Varying) Seismic Data Sets The time-varying nature of earthquake data records obtained from discretely positioned seismographs is suitable for dynamic display, enabling intuitive correlation of the observed ground motion records. Synchronized display of ground motion records can then reveal seismic wave propagation. This knowledge could ultimately aid in developing refined numerical or analytical simulations of these events. The horizontal orthogonal components of one-dimensional acceleration records can be combined and displayed in 3D using billboarding techniques for seismograms and geometric proxies to encode time varying information. When properly time stamped and geo-referenced, data from multiple stations can be studied simultaneously. This study uses CSMIP free-field measured tridirectional acceleration records of the 1994 Northridge earthquake, obtained via the COSMOS database. The data of 72 representative stations was used for the visualization. Figure 7(a) shows the acceleration amplitudes for the 1994 Northridge earthquake rendered as billboards. Since the textures for the ground motion records are generated dynamically, additional information such as time stamps or global reference markers can easily be augmented onto the data as it is being rendered. This enables the correlation of the observed events and more intuitive analysis of interconnected information. Figure 7(b) shows a visual representation of the peak ground accelerations in the form of bounding boxes aligned with the respective data site. Bounding box dimensions represent the peak accelerations observed during the event and are updated based on the seismogram records. An additional proxy is rendered at each station location to visualize the waveform history during playback. From these ground motion records it is easy to identify at which times particular areas were subjected to the most significant shaking.

(a) (b) Figure 7. (a) Rendering of 1994 Northridge seismogram data in east-west direction for the Los Angeles basin. (b) Bounding box for maximum acceleration amplitude in east-west direction. Visualizing the Seismic Response of a Building A 3D model of the building was constructed for subsequent visualization for this case study. Major components of the structure were subdivided into groups of solid objects, such as columns, beams, floors (including the roof level), paneling, and facia. This allows groups of objects to have common texture skins and to be independently manipulated. The field measured data sets are discretely mapped onto the geometric model and used to generate temporal simulations reproducing the deformation patterns observed.Records from the 1994 Northridge earthquake consist of 16 channels of seismograms measuring acceleration in both horizontal directions, and select measurements in the vertical direction. Accelerations can be double-integrated and processed to obtain displacement records. Since sensors are not placed at each floor level, or at each extreme edge of the building, to fully describe the 3D movement of the structure, interpolation was applied. Floor levels are structurally rigid and no local damage to the members was observed, therefore, the assumption of linear interpolation across the plan of each of the floors is reasonable.

182

Each seismogram consists of 3,000 sample points with a sample spacing of t=0.02 second, for a total duration of the time-varying visualization of 60 s. Figure 8 shows a sequence of time steps and snapshots of the fully deformed building model using shade (color) coding applied to the components of the model. In the real simulation for the students, red is used to represent the maximum deformations (over the entire record), while blue is used to represent minimum values. Interpolation is applied across the color spectrum and mapped onto the model. Color provides an intuitive, rapid assessment of the current state of the structure during time. It is based on the assumptions regarding the unknown deformation points, as described. The ground floor level in this case is fixed and all subsequent floors are shown relative to ground surface. It should be noted that deformations of the building are amplified by 20 to enhance their visual perspective. It clearly shows the extreme in-floor level torsion the structure undergoes during 3D oscillation, as well as the large (dominant) longitudinal (long axis of the building) displacement demand. Displacement measurements indicate peak values in the north-south direction of 17 cm on the east end and 23 cm on the west end. This significant difference between the east and west-end displacements indicated that the building suffered large torsional movements during the earthquake, which concurs with the clear depiction illustrated. As observed in the field after this earthquake, the longitudinal load-resisting system was severely damaged, thus significantly softening the structure in this direction, and the 3D interactive visualization clearly shows this. In addition, between time steps shown, a sharp reversal in the direction of movement can be observed (e.g., between parts i, j, and k of Figure 8. The visualization provides a unique opportunity to show an overview of the structural response of the building for a given ground motion history. In addition, the articulation of the spatial transitions assists in locating the building components, which may have been damaged on a local level.

(a) 6396 sec

(b) 7.02

(c) 7.08

(d) 7.14

(e) 7.20

(f) 7.26

(g) 7.32

(h) 7.38

(i) 7.44

(j) 7.50

(k) 7.56

(l) 7.62

(m) 7.68

(n) 7.74

(o) 7.80

Figure 8. Image sequence showing building deformations (color coded) between time t = 6.96 7.80 seconds (sub-sampled to a t = 0.06 seconds) - 1994 Northridge earthquake.

183

VISUAL ANALYTICS ENVIRONMENT The presented visual analytics techniques can be further enhanced with a visualization environment that matches demands imposed by the data to be analyzed. For the presented case studies, concurrent visual data correlation is enabled through the use of HIPerWall, a wall-sized display. HIPerWall offers a large-scale workspace for massive visual data correlation, providing the ability to display and study multiple data sets in great detail. More precisely, HIPerWall supports the integration of multiple data and analysis modalities, across different scales and information domains. The display consists of fifty 30" display tiles arranged in a ten-by-five matrix, each operating at a resolution of 2,560 by 1,600 pixels (4 mega-pixels), for an overall resolution of over 200 mega-pixels (25,600 by 8,000 pixels). A combined display space of 7m x 2.2m is available for data visualization and collaborative analysis, providing a visual, satisfying 20/20 vision across its entire surface at a distance of one meter from the screen and outward. This effectively allows users to digitally and physically zoom into areas of interest by either using an input device to adjust magnification levels or simply walking towards the display. The display array is powered by 25 PowerMac G5s equipped with dual, dual-core processors (quads), NVIDIA Quadro FX 4500 graphics cards, 2GB of on-board memory and gigabit Ethernet. All nodes are interconnected via a dedicated Level 3 gigabit switch and data is served up by a dedicated storage node, while a stand-alone control node functions as the front-end. Figure 9 illustrates powerof-ten capabilities that are used to study data at different levels of resolution, seamlessly transitioning between a global, local and building-level data representation. In many cases, spatial and temporal data characteristics also benefit from a stereoscopic data representation that allows the user to explore a truly three-dimensional representation of the data. A wall-sized passive-stereo display, operating at 8 megapixel resolution (4 megapixel per eye) was used to provide access to this additional dimension. Users access 3D space with a wand, a spatially tracked 3D pointing device, supporting unconstrained interaction. Figure 10 shows a series of models spanning the global to local scale.

(a)

(b)

(c) Figure 9. Visual Analytics on HIPerWall, leveraging an ultra-high resolution display wall to explore data across different time and spatial scales, (a) global, (b) local and (c) building-level visualization with “powers of ten” support, allowing for a seamless transition between levels.

184

(a) (b) (c) Figure 10. Visual Analytics on CaveWall, (a) global, (b) local (c) building-level visualization in a fully semi-immersive virtual reality environment.

CONCLUSIONS An overwhelming amount of scientific and engineering data is being created at increasingly shorter time intervals and at ever-higher resolution. This is particularly true in the seismic, geotechnical, and structural domains, where sensing of the physical environment is essential to advancing understanding. Visualization of this data, whether it is measured or from numerical simulations, is key to allow a clear interpretation of physical processes. Learning and understanding is substantiated by orders of magnitude if crisp visualization of an event, data, or other information, is provided to the user. In this paper, measured data from multiple scales (global, local, and the building-level) is used to demonstrate a variety of visualization methods, each of which provide the user with the ability to maximize their understanding of its underlying physical mechanisms. These scales encompass the use of field acquired seismic measurements, digital elevation data, and building-level seismic sensors. The methods are simple, yet robust, as they rely upon out-of-core processing techniques, and simplistic rendering strategies, to speed up the visualization activities for the user. Rendering methods include spherical proxies for representing earthquake hypocenters, bounding boxes for representing maximum structural deformations, and tracing elements to describe history. The concept of visual analytics is discussed in the context of these datasets and their associated rendering methods, and demonstrated via data presentation on a variety of hardware interfaces, including a massive high-resolution display system and an immersive virtual reality system.

AKNOWLEDGEMENTS This research was supported in part by the National Science Foundation Combined Research Curriculum Development Program under Grant Number EIA-0203528 and the Major Research Instrumentation Program under award number MRI-0421554, as well as the Holmes Fellowship Foundation. Data for this study was provided by USGS, NEIC and COSMOS. Ms. Rebecca Chadwick assisted with preparation of the waveform data sets and the structural model. The above support is greatly appreciated.

REFERENCES Akcelik V., Bielak J., Biros G., Epanomeritakis I., Fernandez A., Ghattas O., Kim EJ., Lopez J., O’Hallaron D., Tu T., and Urbanic J. “High resolution forward and inverse earthquake modeling on terascale computers,” In SC ’03: Proc. of the 2003 ACM/IEEE conference on Supercomputingpage 52, Washington, DC, USA, 2003. IEEE Computer Society.

185

Amor M., Boo M., Strasser W., Hirche J., and Doggett M. “A meshing scheme for efficient hardware implementation of butterfly subdivision using displacement mapping,” IEEE Computer Graphics & Application, 25(2):46–59, 2005. Antolik M., Stidham C., Dreger D., Larsen S., Lomax A., and Romanowicz B. “2-D and 3-D models of broadband wave propagation in the san Francisco bay region and north coast ranges,” Seismological Research Letter, 68:328, 1997. Aurelius E. The January 17, 1994 Northridge, CA earthquake: An EQE summary report. Technical report, EQE International, Houston, TX, 1994. COSMOS. Consortium of Organizations for Strong-Motion Observation Systems (COSMOS), 2007. http://www.cosmos-eq.org/ CSMIP. California Strong Motion Instrumentation Program (CSMIP), 2007. http://www.consrv.ca.gov/cgs/smip/about.htm Duchaineau M., Wolinsky M., Sigeti DE., Miller MC., Aldrich C., and Mineev-Weinstein MB. “Roaming terrain: real-time optimally adapting meshes,” Proc. VIS ’97 the 8th conference on Visualization ’97, pages 81–88, Los Alamitos, CA, USA, 1997. IEEE Computer Society Press. EERI. “Northridge earthquake of January 17, 1994 reconnaissance report vol.1,” Earthquake Spectra, Supplement C to vol.11, 1:377–420, April 1995. EROS. United States Geological Survey, National Center for Earth Resources Observation and Science (EROS), 2007. http://edc.usgs.gov/ GTOPO30. United States Geological Survey, GTOPO30, 2007. http://edcdaac.usgs.gov/gtopo30/gtopo30.asp Hsieh TJ., Kuester F., and Hutchinson T. “Terrain mapping for interactive visualization of earthquake datasets,” Proc. IASTED International Conference on Modeling and Simulation 2003, pages 489– 496, Palm Springs, California, USA, February 2003. Hsieh TJ., Kuester F., and Hutchinson T. “Visualization of large-scale seismic field data,” Proc. ISC High Performance Computing Symposium (HPC 2003), pages 163–170, Orlando, Florida, USA, March 2003. Hudnut KW., Bock Y., Galetzka JE., Webb FH., and Young WH. “The southern California Integrated GPS Network (SCIGN),” Proc. International Symposium on Crustal Deformation Measurement, pages 129–148, 2001. Hutchinson T., Kuester F., Hsieh TJ., and Chadwick R. “A hybrid reality environment and its application to the study of earthquake engineering,” Virtual Reality, 9(1):17–33, 2005. Komatitsch D., Tsuboi S., Ji C., and Tromp J. “A 14.6 billion degrees of freedom, 5 teraflops, 2.5 terabyte earthquake simulation on the earth simulator,” Proc. SC ’03 the 2003 ACM/IEEE conference on Supercomputing, page 4, Washington, DC, USA, 2003. IEEE Computer Society. Lindstrom P. and Pascucci V. “Terrain simplification simplified: A general framework for viewdependent out-of-core visualization,” IEEE Transactions on Visualization and Computer Graphics, 8(3):239–254, 2002. Ma KL., Stompel A., Bielak J., Ghattas O., and Kim EJ. “Visualizing very large-scale earthquake simulations,” Proc SC ’03 the 2003 ACM/IEEE conference on Supercomputing, page 48, Washington, DC, USA, 2003. IEEE Computer Society. Mavroeidis GP. and Papageorgiou AS. “Simulation of long-period near-field ground motion for the great 1906 San Francisco earthquake,” Seismological Research Letters, 72:227, 2001. Nayak AM., Lindquist K., Newman R., Kilb D., Vernon FL., Johnson A., Leigh J., and Renambot L. “Using 3D glyph visualization to explore real-time seismic data on immersive and high-resolution display systems,” Earth-Oceans-Atmosphere (EOS) Transactions of American Geophysical Union (AGU), 84(46): Fall 2003 AGU Meeting. Abstract ED32C–1208, 2003. NEIC. National Earthquake Information Center (NEIC), 2007. http://neic.usgs.gov/ Trifunac MD., Ivanovic SS., and Todorovska MI. “Instrumented 7-storey reinforced concrete building in Van Nuys, California: description of damage from the 1994 Northridge earthquake and strong motion data,” Technical Report CE 99-02, Univ. of Southern California, Los Angeles, California, 1999.

186

4th International Conference on Earthquake Geotechnical Engineering June 25-28, 2007 Paper No.W1-1007

NONLINEAR WAVE PROPAGATION AND TRENDS AT A LARGESCALE CENTRIFUGE FACILITY Bruce KUTTER 1 and Dan WILSON2 ABSTRACT This paper describes the UC Davis centrifuge facilities, with an emphasis on recent developments in data acquisition systems. An example application presents some centrifuge results that illustrate the nonlinear propagation of waves through liquefiable soil. These waves are associated with the strainstiffening behavior of undrained dilatant soil. When dilation triggers negative pore pressures, the soil stiffens and large acceleration spikes can develop. Similar behavior has been observed in the field during real earthquakes. Shock wave theory can be used to describe many aspects of the propagation of these spike waves. The paper closes with a summary of recent trends in large-scale modeling on shared facilities. Trends in experimental hardware, instrumentation, research directions, and networking are outlined, and the importance of collaborative and multidisciplinary research is discussed. Keywords: centrifuge, earthquake, wave propagation, instrumentation, equipment, collaboration

INTRODUCTION The UC Davis centrifuge has the largest radius and largest platform area of any geotechnical centrifuge in the US; it is one of the top few in these categories in the world. The centrifuge can carry five-ton payloads and operate at 75 g (at the effective radius of 8.5 m). This centrifuge and its associated equipment have capabilities for simulating the effects of earthquakes on geotechnical models and models of soil-foundation-structure interaction. Figure 1 shows researchers installing instrumented model piles into sand in one of the flexible shear beam model containers. The model container will be placed on a shake table mounted on the swinging bucket at the end of the centrifuge (shown in the background). The basic configuration of the shake table on the UC Davis centrifuge is illustrated in Figure 2; the cylindrical accumulators shown in the diagram can be seen on the bucket in the background of Figure 1. Typical geotechnical model testing in the past used limited amounts of instrumentation, which yielded observations at selected points in a physical model. Progress in earthquake engineering was hindered by incomplete and ambiguous data sets that allowed for subjective and potentially mistaken interpretation. The philosophy behind the development of the NEES Centrifuge at UC Davis capitalizes on the size of the facility and innovations in instrumentation and information technology to enable researchers to generate much higher resolution information (more control, sensors, and images) and to create more realistic physical models that will provide unambiguous experimental data for assessments of numerical simulations.

1

Professor, Department of Civil Engineering, University of California, Davis, USA, Email: [email protected] 2 Associate Project Scientist, Department of Civil Engineering, University of California, Davis, USA

187

Figure 1. Researchers installing model bridge bents in a flexible shear beam container. The end of the centrifuge arm is in the background. (Photo by Peter Essick) The data acquisition (DAQ) infrastructure of the centrifuge includes about a dozen computers and externally-mounted data acquisition chassis. The major components of the DAQ infrastructure are: RESDAQ-Main: 128 channels of amplifiers connected to a high-speed A/D converter for conventional sensors RESDAQ-ERT: 48-channel electrode switching system useful for electrical resistivity measurements, electro-location techniques, and electrical resistance tomography RESDAQ-Aux: An auxiliary DAQ chassis dedicated to specialized systems such as the one that drives eight channels of bender element sources and monitors 16-channels of bender element receivers. It also contains a digital oscilloscope card useful for high-speed applications. WiDAQ: Fifty-six 8-channel wireless data acquisition systems that can be used for conventional sensors and MEMS accelerometers. VIDDAQ: Six computers, each dedicated to one high speed (up to 200 frames per second) camera.

Figure 2. Basic configuration of the shaking table mounted on the centrifuge bucket (note the corresponding cylindrical accumulators in Fig. 1)

188

Figure 3 shows a model, similar to that being prepared in Figure 1, mounted on the centrifuge with all instrumentation installed just prior to spinning. Visible in the image are cameras, strobe lights and wireless DAQ units, all monitoring the behavior of six individual pile supported bent structures (Ilankatharan et al. 2006). Aluminum blocks representing a bridge deck mass were supported on two piles for each bridge bent model. The blocks were oriented in varying directions to enable a study of the effect of shaking in different directions relative to the alignment of the piles. The size of the centrifuge enabled approximately 200 sensors to be used to monitor a simultaneous test of the six structures under identical soil conditions and identical sequences of ground shaking. At UC Davis we have developed miniature data acquisition systems that can be mounted in a centrifuge soil model to reduce the number and complexity of signal wires emanating from highly instrumented experiments. The WiDAQ Wireless Data Acquisition System was developed as part of the NEES construction award at UC Davis. It is now in continuous use by researchers at UC Davis (Wilson et al. 2007). Figure 4 shows a collage of images associated with the WiDAQ system. In the upper left, 4 two-channel MEMS accelerometers are shown connected to one of the WiDAQ sensor managers. In the upper right, one of our low-cost, tiny packaging systems is shown. An Analog Devices ADXL sensor is attached to a miniature circuit board, the sensor cable is stress relieved to the circuit board, and then the system is dipped in epoxy to seal it. Just below that, the WiDAQ radio board, DAQ board, and a AAA battery are assembled outside of their housing. On the bottom left of Figure 4, individual Silicon Designs, Inc., MEMS accelerometers are shown. The image on the bottom right is a depiction of how the sensor managers can be deployed around a model container to reduce the number of wires emanating from the experiment. The sensor managers can be buried in the soil or mounted at convenient locations on the model container.

Pile strain gauges routed to WCM/SM MEMS accelerometers on superstructure routed to WCM/SM

Pile strain gauges bundled and routed to DAQ conditioning ICP accelerometers on superstructure bundled and routed to DAQ conditioning Figure 3. Piles with bent caps skewed relative to the direction of shaking. Visible instrumentation includes strobe lights, video cameras, displacement potentiometers, accelerometers and strain gauges.

189

Figure 4. Clockwise from top left: MEMS accelerometers in an open sealable plastic housing; a MEMS accelerometer with custom packaging, side view of Radio, DAQ card and battery; model container with embedded sensor managers; and MEMS accelerometers. The capabilities of the equipment to support research at the UC Davis Center for Geotechnical Modeling are illustrated in the example study on nonlinear wave propagation described below. Additional information about the centrifuge is available at http://nees.ucdavis.edu. This paper concludes with a discussion of future directions for shared major equipment and collaborative research.

RESEARCH ON NONLINEAR WAVE PROPAGATION IN LIQUEFIABLE SOIL An example of ongoing research is presented to provide one realization of an experiment on the large centrifuge. The example chosen is on the study of nonlinear wave propagation in liquefiable soil. Observations from ground motion instrumentation indicate that liquefaction generally results in a softening of the soil and an increase in the predominant period of the motion. But at the same time, high frequency spikes or cusps of acceleration have been observed. The magnitude of these spikes can be quite large (see data from Frankel et al. (2002) shown in Figure 5), producing the peak ground acceleration at a site, and hence they should be accounted for in evaluation of shaking hazards. Youd and Holzer (1994), after analyzing the pore pressure and acceleration data from the Wildlife Site, explained that as the ground oscillated back and forth displacement was arrested by dilatancy that caused a sudden drop of pore pressure accompanied by a sudden deceleration. They explained that the concurrent dips in acceleration and pore-pressure records indicate that the liquefied soil was in a state of cyclic mobility. Youd and Holzer (1994) explained that the spikes in surface acceleration occurred later in the record because ground oscillation actually increased in amplitude for about 20 s after the primary train of acceleration pulses had passed through the site. More recently, Holzer and Youd (2007) explained that Love waves were responsible for the long period motions. The spiky accelerations that occurred as a result of long period excitation are clearly a result of highly nonlinear soil behavior.

190

Figure 5. Ground surface accelerations at a site a few kilometers south of downtown Seattle in the M 6.8 Nisqually, Washington earthquake. Note the large one-sided spikes between t = 24 s and 29 s. (Data from Frankel et al. 2002) At this point, it is useful to distinguish two types of nonlinearity: 1) Strain-softening nonlinearity: shear modulus degrades as strain increases (e.g., hyperbolic stressstrain law and conventional modulus degradation relationships that show the shear modulus monotonically decreases as shear strain increases) 2) Strain-stiffening nonlinearity: shear modulus increases as strain increases (e.g., effect of dilatancy on saturated sand which produces stress-strain behavior indicated in Figure 6).

Strain-Stiffening Nonlinearity and Shock Waves Zeghal et al. (1994), Youd and Holzer (1994), Bonilla et al. (2005), and Holzer and Youd (2007) analyzed acceleration and pore pressure data from the Wildlife Site obtained during the Superstition Hills Earthquake and found that pulses of negative pore water pressure coincided with large shear strains. They also observed stiffening of the soil at the site associated with the reduction in pore water pressure and the increase in effective stress. They were able to back-calculate stress-strain and stress path relationships for the soil at the Wildlife Site and observed that the patterns were similar to those observed in laboratory element tests (e.g., Figure 7). Several researchers have demonstrated that acceleration spikes, similar to those observed in the field, can be reproduced in models tested on geotechnical centrifuges with shaking tables (e.g., Fiegel and Kutter 1994, Coelho et al. 2004, Kutter and Wilson 1999, Taboada and Dobry 1998). Some results of centrifuge experiments to study the formation, evolution, and propagation of spiky acceleration pulses is introduced here.

Figure 6. Stress-strain relationship for a strain-stiffening material.

191

Figure 7. (Top) Undrained torsional shear behavior of Fuji River Sand at a relative density of 75% (Ishihara 1985). Note the strain-stiffening behavior in all except the first stress-strain loop. The stress-strain curve at the bottom (from Holzer and Youd 2007) was obtained by processing accelerometer data at the Wildlife Site according to the procedures developed by Zeghal and Elgamal (1994). Labels on the circled points on the curves correspond to the times at which peak pore pressures were observed. Kutter and Wilson (1999) recognized that many aspects of the propagation of acceleration spikes in centrifuge model tests could be explained using shock wave theories. Shock phenomena occur in other media (e.g., breaking water waves and sonic booms) due to nonlinearities that cause the peak of a wave to travel faster than the front. In laboratory tests, the tangent shear modulus of dilatant sand under undrained conditions has been observed to increase as shear stress increases. This is analogous to the increase in the tangent bulk modulus of air as pressure increases – a necessary condition for formation of shock waves in air. The extent to which the acceleration spikes can be described by shock-wave theory and understanding the factors that control the width and amplitude of acceleration spikes is the subject of ongoing research. Nonlinear wave-propagation centrifuge results Figure 8 shows data from Kutter et al. (2004) obtained from a dense vertical array in a layered soil profile. The left side of Fig. 8 shows a schematic of the soil layering, and the accelerometer locations are indicated by arrows. The base layer (with H2 = 9 m) consisted of sand with relative density of 80%, the layer with H1 = 4.5 m consisted of sand with relative density of 50%, and the top layer is an approximately 2-m thick sloping clay layer. The entire box is sloping to the left so that lateral spreading could be studied. From these data it was clear to see that pulses travel at a variety of speeds, some of them disappear, and some of them focus into spikes. The shape of the acceleration spikes is clearly a function of the soil type; the spikes are sharper in dense sand than loose sand, and their

192

Figure 8. Data from Kutter et al. (2004) showing location of sensors in a centrifuge model container, and the acceleration records in a large shaking event. Time and depth have been scaled by a factor of 30 to prototype scale. amplitude decreases and width increases in the clayey surface layer. The spikes are unsymmetrical due to the existence of a static shear stress in the direction of the sloping ground. Some pulses were observed to travel as slow as 11 m/s. Figure 9 presents a closer look at the data from this vertical array, and also presents velocity records, obtained by integration and baseline correction of the acceleration data. In Fig. 9, fine lines have been added to enable visualization of the propagation of various wave features. Wave velocities are obviously not constant, some small waves coalesce into sharper waves, sometimes forming a sharp spike and then the spike waves sometimes grow and sometimes disperse as they propagate through the soil. Interestingly, the inelastic waves seem to die as they reach the surface without significant reflection. The nonlinear wave propagation problem is just one application for which dense arrays of sensors are desirable to capture the phenomenon. The multi-component experiment shown in Figure 3 is another example where the performance of a variety of structures can be compared in a single model test, provided that sufficient instrumentation is available.

TRENDS IN CENTRIFUGE MODELING OF EARTHQUAKE ENGINEERING PROBLEMS Facilities (experimental hardware) Over recent years there has been a significant growth in ability for robotic manipulation of experiments while they spin on the centrifuge. Robotics may be used for inspection with cameras, site investigation with penetrometers or resistivity probes, and for simulation of construction processes such as excavation or pile installation. Another area where progress has been significant is in the development of flexible shear beam model containers that permit waves to propagate through the model without over-constraining the behavior of the experiment. In shake table testing on the centrifuge and elsewhere, there is an interaction between the shaking apparatus and the experiment, the interaction alters resonant frequencies of the system and results in energy transfer between the experiment and the apparatus. In centrifuges the ratio of mass-ofexperiment to the mass of the shaking table or to the reaction mass is typically larger than it is in 1-g shaking tables. This significance of the interaction between the apparatus and the experiment should be accounted for.

193

Figure 9. A closer look (12 to 19 s time window) of the data in Figure 6. The top graph shows acceleration, the bottom shows velocity obtained by integration of acceleration record. Light lines were added by eye to help visualize the evolution of nonlinear waves. Spacing of traces is in proportion to the spacing between sensors in the vertical array. Time is scaled by a factor of 30 to convert model data to prototype scale. Instrumentation and data acquisition The trend in instrumentation can be summarized by: more, faster, and different. Commercial high speed video cameras are becoming more and more capable, reliable, and have better resolution. In selecting video cameras, one is faced by the decision of using one very fast (1000’s of frames per second) and expensive camera or opting for use of a larger number of inexpensive intermediate speed cameras (200 frames/second). Wireless data acquisition systems are now being routinely used at Davis and we invite collaborators to help maintain and develop this platform. The system enables 100’s of sensors to be economically added to a facility with minor increases in wire congestion in the experiments. The WiDAQ system is able to record data from MEMS accelerometers, which promises to reduce the cost associated with increasing instrumentation. Geophysical testing and tomography are really promising tools for better monitoring of experiments. In this category we include arrays of bender elements to enable monitoring of the evolution of shear wave velocity in the experiments and p-wave ultrasound imaging which has been shown to hold promise for detecting interfaces and pockets of unsaturated soils in the centrifuge models. Electrical

194

methods also have been demonstrated to hold promise for measuring volumetric and shear strains in soil, for detection of movements of buried electrodes, and producing tomographic images of resistivity distribution (Li et al. 2006). Fiber-optic sensors, as they become more and more economical, have great potential to revolutionize the ability to measure strain distributions on structural components such as piles or retaining walls. Laser scanners and close range photogrammetry could potentially provide accurate 3-D maps of experiments. There are so many exciting instrumentation technologies, and many of them require different equipment and software to make them work. In dynamic centrifuge testing, the speed of measurements is frequently an issue. The cost of increasing instrumentation can be measured in purchasing cost or in the time required to install the sensors, measure their locations, mount the equipment on the centrifuge and learn or develop the software to process the resulting data. Research topics With the development of shared major experimental facilities in the US, interdisciplinary research is likely to continue to increase. This increases the breadth of knowledge of researchers, and forces them to figure out how to communicate the results of their research to members of another discipline. This must be beneficial for growing the impact of the research. Recent efforts to model more complex systems in the large centrifuge have included the dynamic behavior of a nonlinear building frame supported on shallow foundations and multi-bent sections of a bridge structure. The complex experiments are both stimulated and made possible through collaboration. (e.g. Kutter and Wilson 2006). Networking, data sharing, simulation and visualization tools, and collaboration Due to the increase in complexity of the experiments being performed, the planning of large scale experiments and construction of the models is best done by teams of researchers. With the advent of teleconferencing tools provided by NEESit (the information technology provider for the Network for Earthquake Engineering Simulation), cross country teleconferences have become quite commonplace in the US. One of the common justifications for model tests is to provide data for verification or development of numerical procedures. This is benefited by the provision of open source community simulation tools (e.g., the OpenSees platform http://it.nees.org/software/opensees/). An oft cited benefit of model testing is that a valuable data set will be produced and archived for use by others. Data sharing is a great idea because it has the potential to reduce the need for repeating experiments that have already been done. Sharing data does not mean posting a file to a web site. The file needs to be accompanied by the metadata necessary for others to understand the formats of the data (e.g. Kutter et al. 2002). The complexity of data from large scale experiments is an additional non-technical barrier to data sharing. Visualization tools could play an important role in enabling complex data sets to be understood by others. A visualization tool called N3DV is intended to be useful for presenting the data from a large number of sensors in an intuitive format (Weber et al. 2003). A screenshot from N3DV showing sensor metadata and sensor locations is presented in Figure 10. Dynamic data may be presented by animating the behavior or icons that represent different sensors in an experiment. Visualization tools such as N3DV are hoped to be a “carrot” that encourages people to archive there data in a standard format that can be imported by N3DV and understood by others. The added benefit is that the tool can hopefully be used on-the-fly during an experiment enabling researchers to quickly peruse their data and identify sensors that produce anomalous data.

195

Figure 10. Data and Metadata from centrifuge experiments are shown using a tool called N3DV. Each sensor is represented by an icon for which the color, size, and location may be animated in proportion to the data recorded.

DISCUSSION Research progress in civil engineering has been steady but slow over the last century. Research has historically been accomplished by individuals with small groups of researchers. A small fraction of these individuals have been successful in closing the loop from identifying a problem, researching a solution and being successful in getting the solution applied in engineering practice. The slow but steady progress in civil engineering contrasts with relatively rapid progress that has been made in the computing industry. New computer chips are designed by large teams of engineers, and parallel advances in data storage, displays, networks, batteries, and software are essential for the computing revolution that has occurred. Individuals play a minor role in the computing revolution; the progress is driven by highly coordinated organizations. There is great potential to revolutionize research progress in civil engineering through collaboration, sharing, visualization, and knowledge generation. IT infrastructure will eventually be established to facilitate seamless collaboration and shared access to data. Access should be considered in the broadest sense. It includes ability to obtain and analyze the data as well as visualization tools that enable efficient understanding of the data. Changing the emphasis toward collaborative interdisciplinary research requires a change in research culture. We cannot underestimate the difficulty of changing a culture that is well-established and has a successful history, even when there is broad consensus that the change is for the better. There is a lot of incompletely understood physics associated with dynamics of particulate, multiphase materials. The nonlinear wave propagation issue presented in this paper presents just one example. A goal of NEES is to provide equipment and an environment in which we can make the fundamental advances to make a real impact on understanding of the complex physics.

196

CONCLUSIONS There are many problems involving nonlinear dynamic soil behavior that remain to be resolved. The fact that 1-D nonlinear wave propagation in strain-stiffening soil can produce large acceleration spikes, which can be modeled as shock waves, is an example of a basic phenomenon that has become well understood by studying the results of model tests. The NEES centrifuge at UC Davis, with its large size and versatile data acquisition network, provides a facility for production of comprehensive data sets that enable description of model behavior as opposed to sparse sampling of data at selected points. Hundreds of sensors can be used to enable significantly improved resolution of the behavior and mechanisms that occur during shaking. Geophysical tools are being developed to enable researchers to measure internal deformations and strains. A large centrifuge experiment using all of the resources described herein will be more than simply a “test.” With the advanced instrumentation and large size of the experiments, the experiments should be viewed as “repeatable” case histories. Due to the large quantity of information to be captured from one model test sequence, theories and design procedures can be improved and tested. There is still work to be done to train the profession to fully appreciate and to understand what may and what may not be learned from the large-scale centrifuge model tests. Increased collaboration with practicing engineers and development of software tools for simulation and intuitive visualizations of the data are needed to improve the technology transfer from university to practice. The Center for Geotechnical Modeling facilities are intended for use by researchers from academia and industry, from both the US and abroad, as a shared-use resource. We have an “all hands on deck” philosophy; one person cannot quickly operate a shaking table, all the sensors, tomography tools, as well as the sequence of events being tested. Future experiments will be performed by teams of researchers; with remote researchers able to control their experiments at UC Davis as easily as local researchers.

ACKNOWLEDGEMENTS Support for operations and maintenance of the centrifuge is provided by National Science Foundation Award No. CMS - 0402490 provided via the Network for Earthquake Engineering Simulation (NEES). The experimental work was made possible by funding of the Earthquake Engineering Research Centers Program of the National Science Foundation, under award number EEC-9701568 through the Pacific Earthquake Engineering Research Center (PEER). Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect those of the National Science Foundation. The facility development and work presented in this paper were made possible by the staff and colleagues in the UC Davis Center for Geotechnical Modeling: Ross Boulanger, Lars Pedersen, Chad Justice, Cypress Winters, Ray Gerhard, Peter Rojas, and Nick Sinikas.

REFERENCES Bonilla, L. F., Archuleta, R. J. and Lavallee, D. (2005). “Hysteretic and Dilatant Behavior of Cohesionless Soils and Their Effects on Nonlinear Site Response: Field Data Observations and Modeling,” Bull. Seism. Soc. Am., 95, pp. 2373-2395. Coelho, P. A. L. F., Haigh, S. K. and Gopal Madabhushi, S. P., (2004). “Centrifuge modelling of earthquake effects in uniform deposits of saturated sand,” Fifth International Conference on Case Histories in Geotechnical Engineering, April 13-17, 2004, New York, Vol. 1, pp. 8

197

Fiegel , G. L. and Kutter, B. L. (1994). “Liquefaction-Induced Lateral Spreading of Mildly Sloping Ground,” J. Geotech. Engrg. 120, pp. 2236. Frankel, A. D., D. L. Carver, and R. A. Williams (2002). “Nonlinear and linear site response and basin effects in Seattle for the M 6.8 Nisqually, Washington earthquake,” Bull. Seism. Soc. Am., 92, pp. 2090–2109. Holzer, T,L., and T.L. Youd (2007) “Liquefaction. Ground Oscillation, and Soil Deformation at the Wildlife Array, California,” Bulletin of the Seismological Society of America, In Press. Ilankatharan, M., Kutter, B.L., H. Shin, P. Arduino, S.L. Kramer, N. Johnson, and T. Sasaki. (2006). “Comparison of Centrifuge and 1g Shaking Table Models of Pile Supported Bridge Structure,” International Conference on Physical Modeling in Geotechnics, ISSMGE, Hong Kong, pp. 165 170 Ishihara, K. (1985). “Stability of natural deposits during earth-quakes,” Proc., 11th Int. Conf. on Soil Mechanics and Foundation Engineering, Vol. 2, San Francisco, pp. 321–376. Kutter, B. L. (2006). “Phenomena associated with undrained and partly drained dilatant soil,” International Conference on Physical Modeling in Geotechnics, ISSMGE, Hong Kong, pp. 165170. Kutter B.L., Gajan, S.M., Manda, K.K. and Balakrishnan, A., (2004), “Effects of layer thickness and density on settlement and lateral spreading,” ASCE Journal of Geotechnical and Geoenvironmental Engineering, Vol. 130, No.6, pp. 603-614. Kutter B.L & Wilson D.W. (1999): “De-liquefaction shock waves,” Proceedings 7th U.S Japan Workshop on earthquake resistant design of life line facilities and countermeasures against liquefaction, pp. 295-310. http://cgm.engineering.ucdavis.edu/Publications/reports/shockpaper.pdf Kutter, B.L. & Wilson, D.W. (2006) “Physical Modelling of Dynamic Behavior of Soil-FoundationSuperstructure Systems,” International Journal for Physical Modeling in Geotechnics, Volume 6, No. 1, pp. 1-12, 2006. Kutter, B.L., Wilson, D.W., and Bardet, J.P. (2002). “Metadata Structure for Geotechnical Physical Model Tests, ” Proc. Int. Conf. On Physical Modelling in Geotechnics, St. Johns, Canada (Phillips, Guo, Popescu, eds.), Balkema, Lisse, Netherlands, pp. 137-142. Li, Z., Kutter, B. L., LaBrecque, D. and Versteeg, R., (2006). “A New Electrode Switching System (ESS) and a Scheme for Measurement of the Movement of Buried Objects,” 6th International Conference on Physical Modeling in Geotechnics 2006, Hong Kong, China. Taboada, V.M. and Dobry, R. (1998). “Centrifuge modeling of earthquake-induced lateral spreading,” J. Geotech. and Geoenvir. Engrg., 124, pp. 1195. Weber, G.H., Schneider, M., Wilson, D.W., Hagen, H., Hamann, B., and Kutter, B.L. (2003). “Visualization of Experimental Earthquake Data.” Visualization and Data Analysis 2003, Proc. SPIE, (Boerner, K., Chen, P.C., Erbacher, R.F., Groehn, M. and Roberts, J.C., eds.), August, Vol. 5009, Bellingham, Washington. Wilson, D.W., Boulanger, R.W., Feng, X., Hamann, B., Jeremic, B., Kutter, B.L., Ma, K.-L., Santamarina, C., Sprott, K.S., Velinsky, S.A., Weber, G., and Yoo, S.J.B. (2004). The NEES Geotechnical Centrifuge at UC Davis, 13th World Conference on Earthquake Engineering, Vancouver, B.C., Canada, August 1-6, Paper No. 2497. Wilson, D.W. and Kutter, B.L. (2002). Dense Instrumentation Arrays, Proc. Int. Conf. On Physical Modelling in Geotechnics, St. Johns, Canada (Phillips, Guo, Popescu, eds.), Balkema, Lisse, Netherlands, pp. 131-136. Wilson, D.W., Kutter, B., Ilankatharan, M., Robidart, C. (2007) “The UC Davis High-Speed Wireless Data Acquisition System,” in press, International Symposium on Field Measurements in Geomechanics, ASCE Geoinstitute, Boston, September. Youd, T. L., and T. L. Holzer (1994). “Piezometer performance at the wildlife liquefaction site,” California, J. Geotech. Eng., 120, no. 6, pp. 975–995. Zeghal, M., and A. W. Elgamal (1994). “Analyses of site liquefaction using earthquake records, “ J. Geotech. Eng., 120, no. 6, pp. 996-1017.

198



    W1-1011 

 

                                                                                                                                                                                                                                                                                              

  

    

199

                                                                                                                                                                                                                                                      

200

                                                                                                                           

201

                                                                               

                        

202

  

                       

     

203

                                                                                                                                                                                                                              

204

   

                              

205

                          

       

206

                                                                                                                                                                                                       

207

        

                                    

208

                                                                                                                                              

 

209

                                                                                                                                                                                 

210

4 th INTERNATIONAL CONFERENCE ON EARTHQUAKE GEOTECHNICAL ENGINEERING 4th INTERNATIONAL CONFERENCE Thessaloniki, Greece 25-28 June 2007

ON EARTHQUAKE GEOTECHNICAL ENGINEERING Thessaloniki, Greece 25-28 June 2007

15 YEARS of EUROSEISTEST K. PITILAKIS (Project Coordinator) D. RAPTAKIS, K. MAKRA, F. CHAVEZ-GARCIA (Site Effects) M. MANAKOU, P. APOSTOLIDIS (3D Soil Model) G. MANOS (Structural Studies)

4 th INTERNATIONAL CONFERENCE ON EARTHQUAKE GEOTECHNICAL ENGINEERING Thessaloniki, Greece 25-28 June 2007

GENERAL PRESENTATION

European Projects: EUROSEISTEST

EUROSEIS-TEST (EV5V-CT.93-0281) EUROSEIS-MOD (ENV4-CT.96-0255) EUROSEIS-RISK (EVG1-CT-2001-00040)

Euroseis-Test site Thessaloniki Urban Area

Area: 60 km2

Volvi Lake

N Stivos

Profitis

EuroseisTest site

211

4 th INTERNATIONAL CONFERENCE ON EARTHQUAKE GEOTECHNICAL ENGINEERING Thessaloniki, Greece 25-28 June 2007

SEISMICITY – STRESS FIELD Geodetical surveys

Recording period 1973-2002, M>3 (Geophysical laboratory AUTH)

Permanent Seismological network

Recording period 1981-2003

Focal mechanism and stress field seismotectonic model

(GALANIS et al., 2004)

(VAMVAKARIS et al., 2004)

4 th INTERNATIONAL CONFERENCE ON EARTHQUAKE GEOTECHNICAL ENGINEERING Thessaloniki, Greece 25-28 June 2007

VOLVI EARTHQUAKE 20/6/1978, M w=6.5

The only record until the 1993 Volvi earthquake 20/6/1978, Mw=6.5

200 100 0 -100 -200 200

MN' 78 (L)

AF '78 (L)

A single record for the time period 1978-1993 MN '78 (T)

AF '78 (T)

MN '78 (V)

AF '78 (V)

100 0 -100 -200 200 100

40

20

0 -100 -200

0

-20

-40

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 TIME (sec)

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 TIME (sec)

0

4

8

12

16

20

Time (sec)

212

4 th INTERNATIONAL CONFERENCE ON EARTHQUAKE GEOTECHNICAL ENGINEERING Thessaloniki, Greece 25-28 June 2007

GEOPHYSICAL – GEOTECHNICAL EXPERIMENTS

In situ measurements Sampling boreholes In situ geotechnical surveys (NSPT, CPT) Seismic surveys for P-wave refraction and Seismic Wave Inversion (SWI) Geoelectrical tomography and soundings Cross-hole and Down-hole surveys Logging measurements Microtremor array measurements (SPAC method)

Laboratory measurements Conventional geotechnical tests Resonant column Triaxial Oedometer Cyclic triaxial tests

4 th INTERNATIONAL CONFERENCE ON EARTHQUAKE GEOTECHNICAL ENGINEERING Thessaloniki, Greece 25-28 June 2007

FIRST EXPERIMENTS 1993 - 1995

EUROSEIS-TEST project (1993-1995) 12 km surface measurements -1 km borehole measurements Refraction - Surface wave inversion – microtremors – Cross-hole – Down-hole

before starting project

STI

250 200

PRO

150 100

TST

50 0 -50 -100 -150 -200 -16500

-16000

-15500

-15000

-14500

-14000

-13500

DISTANCE (m)

after the end

250

STI

200

PRO

150 100

TST

-50

A B C D E

-100

F

-150

G* G

50 0

-200 -16500

-16000

-15500

-15000

-14500

-14000

-13500

DISTANCE (m)

213

4 th INTERNATIONAL CONFERENCE ON EARTHQUAKE GEOTECHNICAL ENGINEERING Thessaloniki, Greece 25-28 June 2007

GEOPHYSICAL – GEOTECHNICAL RESULTS 1993-1995 S14

S7

S6

0

2

S5

0 SM-SC 10 20 30 40

CL ML-CL

A

SM SC CL

B

370 m/s Vp

Silty Clay Sand Silty Sand and Sandy Clay Grey - Black

5 10

15

15 15

25

25

S3

5

12

10 10

S2

S1

5

4

4

10

12

12

4. 5 8

S8 4

4. 5

S9 2. 5

4

S10

15

17 25

5

S15

35 40

40 40

S12

5 10

10

10

12

15 20

35

40

30

37 40

40

40 45

45 45

45 45

50

45

45

55

55

75

75 75

75

50 55

55

55

55

65

65 65

65 65

6 5 65

80

80

55 55

6655

65 65

55 5 5

5 5

6655

70 75

75

75 80 85 90

85 85

85 8 5

8 5

135

13 5

1 35

90 95

95

100

Vs (m/ s)

105/25 cm 50/7cm

C

Silt - Silty Sand Sandy Clay Light Brown

115

11 5

115

Soi l Type

70/15cm 70/15cm 70/15cm 106

145

400-550 165

165

160 165 170

185

CH

200

1650 m/s

170

180

CL-M L 190190

D

560-650

16 5

SM -M L

70/ 13cm

CH

145

SM 17 0

Clayey Silt and Marly Clay Grey - Green

150-250 250-400

CL

150

CL

60 ML-CL

95-150

115

125 130

SM . SM -SC, M L

Qs

15 20

20

30

SM -SG

70

S13

2 5

15 15

25

160

50

S11

2. 5 10

15

21

1450 m/s Vp

ML SM CL SM

12

BODY WAVES N30-SPT VELOCITIES (km/sec) 0.0 0.5 1.0 1.5 0 20 40 60 80 100

S4

2 5

SOIL USCS SOIL TYPE DESCRIPTION

190

190

195 20 0

195

200

180

600-900

190

1000-1700

18 5

19 5

200

200

190190 19 5 200

2300-2600

200

CL s tiff

70/13cm 118

SC-CL g ravel s Wheat her Rock Rock

80

250

300

90 SM-SC 100 110

E CL

Vs

Silty Clay with sand and gravels Brown to Grey

FORM ATIONS Vp (m /s) Vpw (m/ s) Vs (m/s) Qs d (t/ m 3)

250

NNW 200

A B C D E F G* 330 450 550 - 1500 1600 2000 2500 2600 3500 300 450 650 800 1250 130 200 20 30 40 15 2.05 2.15 2.0, 2.15 2.10 2.15 2.20 2.5

G 4500 2600 200 2.6

SSE

93/25cm

Vp

120 SM-CL 130 ML-SM

150 94/25cm

Silty-clayey Sand

CL

150

CH

160

C G* G

B

100

140

C

F ML

170 SM-ML

Sandy Clay or Silty Clay with gravels

90/15cm

C E F G* G

50

2700 m/s 0

B C D E F

100/8cm

Clayey-silty Sand

-50

A B C

D E

D

F G* G

E G*

180

CPT

SC Vp

190 SCHIST G* 200 GNEISS G

Vs (RC) Qs (RC)

Weathered schist Gneiss

3D ACCEL EROMETERS

F4

3100 m/s Vp 4900 m/s

0 10 20 30 40 50 60 0 60 120 180 240 300 qc-CPT (kgr/cm 2) QUALITY FACTOR (Qs)

G

-100

-150

G* WEATHERED ROCK : V P BELOW WATER TABLE ( PW

*V

** 2.0

t/m : CENTRAL PART 2.15 t/m3 : EDGES

-200 0.0

F

0.5

1.0

2.0

F2

G

F3 1.5

F1

G*

)

3

2.5

3.0

3.5

4.0

4.5

5.0

5.5

DISTAN CE (Km)

(RAPTAKIS et al., 2000)

4 th INTERNATIONAL CONFERENCE ON EARTHQUAKE GEOTECHNICAL ENGINEERING Thessaloniki, Greece 25-28 June 2007

FIELD EXPERIMENTS 1996 - 2005

EUROSEIS-MOD project (1996-1999)

sensor

EUROSEIS-RISK project (2002-2005)

214

4 th INTERNATIONAL CONFERENCE ON EARTHQUAKE GEOTECHNICAL ENGINEERING Thessaloniki, Greece 25-28 June 2007

GEOPHYSICAL RESULTS 1996-2005 R3 R2

2

R1 1 o 12 0

4

3

1.1

1

Experimental dispersion curves Phase velocity Rayleigh waves

GER1 GER2 GER3 SCH LAG1 LAG2 BRG LIM2

0.9

Inverted Vs profile

1 2 EVA VATH PER2 DA B2 MAN ANA THE FTE TRI FRM TOPO VOL FID VRO KAP MES

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1 1

2

3

4

5

6

7

8

MANAKOU, 2007 (PhD THESIS)

(Hz)

4 th INTERNATIONAL CONFERENCE ON EARTHQUAKE GEOTECHNICAL ENGINEERING Thessaloniki, Greece 25-28 June 2007

GEOPHYSICAL RESULTS 1996-2005

60 m 130 m

I S P2

120 m S P1

BRG

120 m II SP3

s ve wa

wave s

ce rfa Su

Body

4

(RAPTAKIS et al., 2005)

215

4 th INTERNATIONAL CONFERENCE ON EARTHQUAKE GEOTECHNICAL ENGINEERING Thessaloniki, Greece 25-28 June 2007

GEOPHYSICAL – GEOTECHNICAL RESULTS 1996-2005

+ Vs: 130-275m/sec Thickness (m) 0-30

400

350

90-120

350

300

300

250

250

200

200

75 76

150

De pth (m )

30-60 60-90

BA

N 400

S1L TST TSTT OPO TRI

50

VRO

91 90 DD2

100 50

A+B C+D

0

Rock

-5 0

120-150

Vs: 280-450m/sec

180-210 210-240

150-180

150

NIK1 79

100

C+D

0

Rock

E+F

-5 0

-1 0 0

240-270

-1 0 0

-1 5 0

-1 5 0

Rock

-2 0 0

270-300

-2 0 0

-2 5 0

-2 5 0

F- GNSP

F- Sx

-3 0 0

-3 0 0

-2 4 00 0

-2 3 00 0

-2 2 00 0

-2 10 0 0

-20 0 0 0

- 19 0 0 0

-1 8 0 0 0

-1 7 0 00

-1 6 0 00

-1 5 00 0

-1 40 0 0

-13 0 0 0

-12 0 0 0

- 11 0 0 0

-1 0 0 0 0

Dista nce (m )

ABA 2 00

1 00 50 0

15 0

MES L AG

BRG -

geo -2003 BRG S3 a rtBRG

KO K

exp- 2003 ANA THE L EF

ANA

DAB2

geo -2004 FID EXP VOL

TOPO TSE PED ATHATH exp- 2004 ANA L OS THE S1LTST 85 TRI TST 8 4 DA B A+B

10 0

C+D

50 0 -50

E+F

-10 0

-15 0

-15 0

-20 0 -25 0

Vs: 500-800m/sec

G GEO

-5 0 -10 0

E+F

20 0

1 50

-20 0

Rock

-25 0

-30 0

-30 0

-35 0

-35 0

F-Sx -40 0 -2 15 00 -21 00 0 -2 05 00 -20 00 0 -1 95 00 -19 00 0 -1 85 00 -18 00 0 -1 75 00 -17 00 0 -1 65 00 -16 00 0 -1 55 0 0 -15 0 00 -1 4 50 0 -1 40 00 -13 50 0 -1 30 00 -12 50 0 -1 20 00 -11 50 0 -1 10 00 -10 50 0 -1 00 00

-40 0

Distance (m )

MANAKOU, 2007 (PhD THESIS)

4 th INTERNATIONAL CONFERENCE ON EARTHQUAKE GEOTECHNICAL ENGINEERING Thessaloniki, Greece 25-28 June 2007

GEOPHYSICAL – GEOTECHNICAL RESULTS 1996-2005

3D GEOLOGICAL STRUCTURE FOR SITE RESPONSE STUDIES

RELIEF

AB/CD INTERFACE CD/EF INTERFACE

EF/BEDROCK INTERFACE

BEDROCK TOP SURFACE

MANAKOU, 2007 (PhD THESIS)

216

4 th INTERNATIONAL CONFERENCE ON EARTHQUAKE GEOTECHNICAL ENGINEERING Thessaloniki, Greece 25-28 June 2007

SITE RESPONSE – SITE EFFECTS STUDIES

SITE RESPONSE – SITE EFFECTS STUDIES To investigate complex site effects (2D and 3D soil structures, differential motion due to edges effects, lateral variation) To develop models and tools for ground motion estimates combining source, path and site effects. Experimental study on azimuthal dependencies and directivity effect Experimental evaluation of soil non-linearities using records from DH array To study liquefaction phenomena (if any). Evaluation methodologies for liquefaction susceptibility and risk

of

existing

To study the spatial variability of ground motion, to identify the relative importance of the parameters involved and to validate the existing methods for loss of coherency estimation. To evaluate the reliability of the empirical techniques used for site amplification estimation To improve the existing numerical and analytical modeling techniques as well as to develop new simulation methods and software. To contribute to the more explicit definition of site effects for the design earthquakes and spectra in Eurocode 8.

4 th INTERNATIONAL CONFERENCE ON EARTHQUAKE GEOTECHNICAL ENGINEERING Thessaloniki, Greece 25-28 June 2007

INSTRUMENTATION – PERMANENT ARRAY

Guralp

PRO

K2

ROA

Guralp!

K2

WT3

WT2

WT1

Etna

Etna

Etna

GRA TST

Etna

FRM

Guralp

STH

Guralp

STC

Etna

STE

ET1

ET2

ET3

Etna

Etna

Etna

217

4 th INTERNATIONAL CONFERENCE ON EARTHQUAKE GEOTECHNICAL ENGINEERING Thessaloniki, Greece 25-28 June 2007

INSTRUMENTATION – PERMANENT ARRAY

NNW PRO

GRA

PRO-33m BRG

KOK

ANA

TST

W TST-21m

TSE

ATH

DAB E

FRM BUT

TST-40m TST-72m TST-136m

STC STI SSE

TST-197m

Out o f sca le

MORE THAN 300 EVENTS No EventDate EventTime Magnitude EcoordN EcoordE Depth 68 27/12/1998 10:14:25 3.1 40.723 23.277 9.4 69 17/8/1999 0:01:37 7.4 40.76 29.97 18 70 12/12/1999 19:25:58 4.5 40.554 23.646 10 71 31/1/2000 9:56:12 3.9 40.636 23.638 10 72 12/3/2000 8:32:42 3.3 40.625 23.153 2 73 10/6/2000 18:35:56 2.7 40.713 23.349 9 74 16/12/2000 2:49:52 2.8 40.595 23.428 75 13/1/2001 0:42:24 3.9 40.67 23.47 1 76 26/7/2001 0:21:39 6.4 39.05 24.35 19 77 10/8/2001 21:49:39 4.8 40.71 23.45 5 78 8/10/2001 4:50:21 4.4 40.64 23.16 11 79 8/10/2001 5:32:17 4.4 40.45 23.08 3 80 8/10/2001 6:55:30 3.5 40.52 23.08 5 81 8/10/2001 8:00:27 3.7 40.54 23.13 5 82 29/10/2003 21:15:48 4.7 40.71 22.83 25 83 27/3/2004 10:39:26 * * * * 84 8/6/2004 22:38:23 * * * * 85 15/7/2004 0:40:31 * * * * 86 15/7/2004 4:12:15 * * * * * No information yet available

4 th INTERNATIONAL CONFERENCE ON EARTHQUAKE GEOTECHNICAL ENGINEERING Thessaloniki, Greece 25-28 June 2007

SITE RESPONSE – SITE EFFECTS STUDIES STC

STE

10 ARNE AS' EARTHQUAKE 1997, MAY 4, M=5.8

1

SSR HORIZ SSR VERT HVS R

FRM

TST

GRB

GRA

10

1 300

250

NN W

FORMATION A B C D E F G* G Vp (m/s) 330 450 550 V pw (m/s) - 1500 1600 200025002600 35004500 Vs (m/s) 130 200 300 450 650 800 12502600 D ensity (t/m 3) 2.05 2.15 2.0, 2.15 2.10 2.15 2.20 2.5 2.6

200

SS E STE ST2 2

10

S TC 150

PR O

C F RM

100 G RA G RB C E F G

50

B C

1

TST

A TS17 B C

D E F

TS72 F

-50

G

C

D E

0

D

B

G

D E

PRO G*

10

F4 F

-100 *V

-150

3D ACCELEROM ETERS PW : V P BELOW WA TER TABLE (

)

10 F REQ UEN CY (Hz)

Figure 5. Comparison of the results obtained from the HVSR technique (thin line), SSR for horizontal (thick line) and for vertical components (dashed line)

F1

G*

** 2.0 t/m 3 : CENTRAL PA RT F2 G F3 2.15 t/m 3 : EDGES G * WEATHERED RO CK -200 0.00 0 .25 0.5 0 0.75 1.00 1.2 5 1.50 1.75 2.00 2.25 2.50 2.75 3.00 3.25 3 .50 3.7 5 4.00 4 .25 4.5 0 4.75 5.00 5.2 5 5.50 5.75 D ISTA NCE (Km)

RAPTAKIS et al., 1998

1

G

1

1

10 F REQ UEN CY (Hz)

218

4 th INTERNATIONAL CONFERENCE ON EARTHQUAKE GEOTECHNICAL ENGINEERING Thessaloniki, Greece 25-28 June 2007

SITE RESPONSE – SITE EFFECTS STUDIES TRANSVERSAL COMPONENT

RADIAL COMPONENT 40

40

36

36

32

32

28

28

24

24

20

20

16

16

12

12

8

8

4

4

0 3.0

0 3.0

2.5

2.5

2.0

2.0

1.5

1.5

1.0

1.0

0.5

0.5

0.0

0.0

300 250 200 150 100 50 0 -50 -100 -150 -200 -250 -300

300 250 200 150 100 50 0 -50 -100 -150 -200 -250 -300 0

500

1 000

1500 2000 2500 3 00 0 3500 4000 4500 DISTANCE (m)

50 00 5500 6000

0

500

1 000

1500 2000 2500 3 00 0 3500 4000 4500 DISTANCE (m)

50 00 5500 6000

RAPTAKIS et al., 2000

4 th INTERNATIONAL CONFERENCE ON EARTHQUAKE GEOTECHNICAL ENGINEERING Thessaloniki, Greece 25-28 June 2007

SITE RESPONSE – SITE EFFECTS STUDIES

PRO

0.0

GRA GRB

1.0

2 .0

TST

3.0

FRM

4.0

ST C STE

5.0

Distance (km)

CHAVEZ-GARCIA et al., 2000

219

4 th INTERNATIONAL CONFERENCE ON EARTHQUAKE GEOTECHNICAL ENGINEERING Thessaloniki, Greece 25-28 June 2007

SITE RESPONSE – SITE EFFECTS STUDIES

2D PARAMETRIC ANALYSES

Our results suggest that, in order to improve current schemes to take into account site effects in building codes, the more to be gained comes from consideration of lateral heterogeneity. MAKRA et al., 2005

4 th INTERNATIONAL CONFERENCE ON EARTHQUAKE GEOTECHNICAL ENGINEERING Thessaloniki, Greece 25-28 June 2007

SITE RESPONSE – SITE EFFECTS STUDIES

T2

T1 0.40 sec 10.0

T A 1 2 T1

2. 0sec

3

T2 T

A 1 2

2D/1D Aggravation factor

3

3.720 synthetic (2D & 1D) accelerograms and their response spectra

1.0

Sa(2D) A Sa(1D) 0.1 0.1

1.0

T1

T2

period (sec) 10.0

STE

6.0

STC

10.0

5.3 4.5

1.0

1.0

3.8

REC/1D

3.0

2D/1D

0.1 10.0

0.1 FRM

TST

10.0

2.3 1.5 0.8 0.0

1.0

1.0

0.1 10.0

0.1 GRB

GRA

10.0

1.0

0.1

1.0

0.1

1.0 Period (sec)

0.1

1.0 Period (sec)

0.1

MAKRA, 2000 (PhD THESIS)

220

4 th INTERNATIONAL CONFERENCE ON EARTHQUAKE GEOTECHNICAL ENGINEERING Thessaloniki, Greece 25-28 June 2007

SITE RESPONSE – SITE EFFECTS STUDIES

4 th INTERNATIONAL CONFERENCE ON EARTHQUAKE GEOTECHNICAL ENGINEERING Thessaloniki, Greece 25-28 June 2007

STRUCTURAL BEHAVIOUR & SSI EFFECTS

EXPERIMENTAL AND NUMERICAL INVESTIGATION OF SSI EFFECTS Prof. George C. Manos

221

4 th INTERNATIONAL CONFERENCE ON EARTHQUAKE GEOTECHNICAL ENGINEERING Thessaloniki, Greece 25-28 June 2007

FACILITIES FOR TESTING OF STRUCTURAL MODELS

Permanently instrumented field site for SSI effects

React ion Mass

Store House

React ion Mass

7,0

Moving Crane Facility Max. Capacity 20 KNt

2,0

4.2

2.52

3,38

25,0m

1.5 3.5

5,82

2,00

1.52

PIER M ODEL

6,0

A

3.5

2,0

3.5

3.15

3.36

9.2

3.6

2.95

2.46

9,80

A

1.5

3.5 6.46

4,30

5-Story M odel 5,00

5,00m 8,25m 7,0

7,54m 10m

2 00 3 6 EURO SEIS- TEST

4 .0

My

OVERTURNING MOMENT

3 .5

Ut ili ty Room

LAY O UT OF EUROSEIS T EST SITE

3 .0 2 .5 2 .0 1 .5 1 .0

Movin g Crane Facility Max. Capacity 20 KNt Store House Total weight = 153,6t

-0 .5 0 .0

5-Story M ode l 0.4

PIER M ODEL

1 .0

1 .5

2 .0

2 .5

3 .0

3 .5

4 .0

4 .5

5 .0

5 .5 6 .0 t [se c]

-3 .0

6,50m

4,00m

9,00m

0 .5

-1 .0 -1 .5 -2 .0 -2 .5

4,30m

3,70

3,30 0,35

0 .5 0 .0

SECTION A - A

-3 .5 -4 .0

React ion Mass 7,10

3,35 10,0

2,5 8,25

6,75

4 th INTERNATIONAL CONFERENCE ON EARTHQUAKE GEOTECHNICAL ENGINEERING Thessaloniki, Greece 25-28 June 2007

FACILITIES FOR TESTING OF STRUCTURAL MODELS

Study of SSI effects on model structures at Volvi European Test Site Euro-Risk Project Coordinator: Prof. George C. Manos http://euroseis.civil.auth.gr

6th Storey Model

222

4 th INTERNATIONAL CONFERENCE ON EARTHQUAKE GEOTECHNICAL ENGINEERING Thessaloniki, Greece 25-28 June 2007

FACILITIES FOR TESTING OF STRUCTURAL MODELS STRUCTURES OF TEST SITE 5storey building Pier bridge model

me an

9.524E-03

4 th INTERNATIONAL CONFERENCE ON EARTHQUAKE GEOTECHNICAL ENGINEERING Thessaloniki, Greece 25-28 June 2007

FACILITIES FOR TESTING OF STRUCTURAL MODELS

TYPES OF ANALYSES

2. Time domain 3. Various special analysis

ADINA [Hz]

1.67

2.68

3.16

FFTrec [Hz]

1.65

2.77

3.28

Pier material: elastic linear E=32GPa , v=0.3 , =2.5t/m3

Vertical component

Soil material: elastic linear E=97.14MPa , v=0.3 , =2.05t/m3 , Vs=135m/sec

223

4 th INTERNATIONAL CONFERENCE ON EARTHQUAKE GEOTECHNICAL ENGINEERING Thessaloniki, Greece 25-28 June 2007

REFFERENCES

•Chávez-García, F.J., D. Raptakis, K. Makra, and K. Pitilakis (2000). Site effects at Euroseistest II. Results from 2D numerical modeling and comparison with observations, Soil Dyn. Earthq. Engrg. 19, 23-39. •Kudo K., T..Kanno, H. Okada, T. Sasatani, N. Morikawa, P. Apostollidis, K. Pitilakis, D.Raptakis, M. Takahashi, S. Ling, H.Nagumo, K. Irikura, S. Higashi & K. Yoshida, 2002. “S-Wave Velocity Structure at EURO-SEISTEST, Volvi, Greece Determined by the Spatial Auto-Correlation Method applied for Array Records of Microtremors”, Proc. 11th Japan Earthquake Engineering Symposium, Paper No. 62. •Makra, K., D. Raptakis, F.J. Chávez-García, and K. Pitilakis (2001). Site effects and design provisions: the case of Euroseistest, PAGEOPH 158, 2349-2367. •Makra, K., D. Raptakis, F.J. Chavez-Garcia & K. Pitilakis, 2002. “How important is the detailed knowledge of a 2D soil structure for site response evaluation”, Proc. 12th ECEE, Paper No 682. •Makra K., F.J. Chávez-García, D. Raptakis & K. Pitilakis, 2005. “Parametric analysis of the seismic response of a 2D sedimentary valley: Implications for code implementations of complex site effects”. SDEE 25, pp. 303-315. •Manakou (2007). Contribution to the determination of a three dimensional soil model for studying seismic response: application to the Mygdonian sedimentary basin. Ph.D. Thesis (in Greek), Civil Engineering Department, Aristotle University of Thessaloniki. •Manos C. (Presentation by V. Renda on behalf of Prof. G. Manos), 2004. Experimental and Numerical Studies of Model Structures at the Volvi European Test Site. OECD/NEA International Workshop "Seismic Input Motions, Incorporating Recent Geological Studies“, OECD/NEA Workshop proceedings. •Manos, G., Kourtidis, V., Soulis, V., Sextos, A., Renault, P. and Yassin, B., 2005. Numerical and experimental soil-structureinteraction of a bridge pier model at the Volvi-Greece European Test Site. 6th European Conference on Structural Dynamics, Eurodyn 2005, Paris, II, pp. 1335-1340. •Manos, G.C., Renault, P. & Sextos, A.G., 2005. Investigation of the interaction between neighboring model structures at the Euroseistest site. Proceedings: EURODYN, pp. 1297-1302. •Pitilakis, K., D. Raptakis, K. Lontzetidis, Th. Tika-Vassilikou, and D. Jongmans (1999). Geotechnical and geophysical description of EURO-SEISTEST, using field, laboratory tests and moderate strong motion recordings. J. Earthq. Engrg. 3, 381-409. •Raptakis, D., N. Theodulidis, and K. Pitilakis (1998). Data analysis of the EURO-SEISTEST strong motion array in Volvi (Greece): Standard and horizontal to vertical spectral ratio techniques. Earthq. Spectra 14, 203-224. •Raptakis, D., F.J. Chávez-García, K. Makra, and K. Pitilakis (2000). Site effects at Euroseistest I. Determination of the valley structure and confrontation of observations with 1D analysis, Soil Dyn. Earthq. Engrg. 19, 1-22. •Raptakis, D., M. Manakou, F.J.Chávez-García, K.Makra, and K. Pitilakis, 2005. 3D Configuration of Mygdonian basin and preliminary estimate of its site response, SDEE 25, pp. 871-887.

224







    W1-1008 

 

                                                                                                                                                                     

  

                                

225

                                                                                     

 



                          

226

                                                                                                                                                                                                 

  

227

                                                                                        





                                                                                                       

228

                                                                                                                                                                                                                                                                                                               

  

229

                                                                                                                                                                                                                                                                                                                                                                         

230

   

   

 



  

 

 

   

 

 

 

  



                                                                                                                                                                                   

231

                                     

                                                                                              

232

                                                                                                                                                                                                                                                                                                     

233

4th International Conference on Earthquake Geotechnical Engineering June 25-28, 2007 Paper No. W1-1009

INSTRUMENTED GEOTECHNICAL SITES: CURRENT AND FUTURE TRENDS Jamison H. Steidl 1

ABSTRACT A goal of engineering seismology research is to generate analytical and empirical models for accurate prediction of ground shaking, pore water pressure generation, ground deformation and soilfoundation-structure interaction (SFSI), and to help engineers understand how these predictions will affect the built environment. The development of simulation capabilities that can reproduce these effects at various strain levels requires well-instrumented test sites where actual ground response, pore pressure, and deformation can be monitored during earthquake shaking to provide benchmark case histories for verification of the simulation models. In the U.S. alone, there are many “extensive” geotechnical strong motion array facilities available for use in calibration and validation of our modeling techniques. An update on these facilities including recently deployed arrays, and a summary of the current research activities using these facilities will be presented. In particular, the experimental field site facility that is part of the National Science Foundations George E. Brown Jr. Network for Earthquake Engineering Simulation (NEES) will be highlighted. This facility includes two permanently instrumented field sites for the study of ground response, ground failure, soil-foundationstructure interaction, and liquefaction. The current and future trend for instrumented sites seems to be moving to the simultaneous monitoring of both the geotechnical and structural components at a given site. This integration of the two sub-disciplines within earthquake engineering provides opportunities for new collaborations. The performance analysis of instrumented structures incorporating geotechnical and structural aspects should provide advances in our ability to predict the effects of earthquakes on the built environment. Keywords: geotechnical strong motion array, engineering seismology, site response, test sites.

INTRODUCTION Over the last two decades there has been significant activity in terms of the construction and operation of instrumented geotechnical sites. This summary paper is a continuation of the work presented almost decade ago (Archuleta and Steidl, 1998) that outlined the results of borehole array studies in the United States. Two issues come to mind when thinking about the last two decades of activity with instrumented geotechnical sites. First, the need for increased coordination among the agencies and organizations that install, maintain, and disseminate the data form these instrumented sites. This has been repeatedly expressed in the proceedings and reports from various workshops and conferences. Second is the lack of collaboration between the structural and geotechnical engineering communities to improve the design and performance of structures by integrating the components of the problem from both disciplines. Until recently, these components had been treated separately. This workshop contribution will begin with a discussion of the geotechnical strong motion array (GSMA) activities in the United States, including a list of what is thought to be the current operational sites broken into categories. Then some examples of the current experimental activities using these 1

Institute for Crustal Studies, University of California at Santa Barbara, California, USA, Email: [email protected]

234

sites will be provided. Lastly, some discussion of the future opportunities and challenges that the community faces in trying to deal with the two issues of coordination and collaboration mentioned above. EXISTING GEOTECHNICAL STRONG MOTION ARRAYS IN THE UNITED STATES In taking a look at the Geotechnical Strong Motion Array (GSMA) sites in the United States, the first question to ask is how to define a GSMA in the first place. In this case, any strong motion station that contains both a surface and a sub-surface borehole sensor qualifies. However, the GSMA’s are further sub-divided into three categories. Extensive Arrays (4+ borehole sensors): An “extensive” GSMA is a site with four or more borehole accelerometers in addition to the surface sensor. Using this definition, 14 such sites are found. These sites are operated by a mix of agencies and institutions, including; the United States Geological Survey (USGS), the California Geological Survey (CGS/CSMIP), California Department of Transportation (Caltrans), the National Science Foundation (NSF), and the University of California at Santa Barbara (UCSB). The site name, region, and responsible agencies are listed for each of these extensive array sites in Table 1. Note that these sites are primarily in California, with the exception of one site in Anchorage, Alaska. This is primarily reflective of the high cost of installing and operating extensive GSMA’s and the probability of obtaining useful strong motion data over the life span of the array. The high seismicity regions of the United States have historically been the most heavily instrumented, though this is changing somewhat with the new Advanced National Seismic System (ANSS). This is not to say that having strong motion records in the lower seismicity regions like the Central and Eastern United States is not important, however, the potential to “catch” a significant event does play into the funding agencies decisions regarding where to allocate resources for these “extensive” facilities. Table 1. Extensive Array sites in the United States Site Name Bessie Charmichael School Borrego Valley Delaney Park Embarcadero Plaza Eureka Array Garner Valley Hayward, San Mateo Bridge Hollister Observatory Levi Plaza Melloland Array El Centro San Diego Coronado Bridge Treasure Island NGES site Vincent Thomas Array Wildlife Liquefaction Array

Location Bay Area – Northern California Southern California Anchorage, Alaska Bay Area – Northern California Northern California Southern California Northern California Central California Northern California Southern California Southern California Bay Area – Northern California Southern California Southern California

Agency(s) USGS UCSB ANSS/USGS/UAF USGS CGS/CSMIP/Caltrans UCSB/NEES CGS/CSMIP/Caltrans UCSB USGS CGS/CSMIP/Caltrans CGS/CSMIP/Caltrans CGS/CSMIP/NSF CGS/CSMIP/Caltrans UCSB/NEES

As an example of an extensive array we show the recently deployed Anchorage ANSS Delaney Park Array and instrumented structure, the Atwood Building, located 500 meters from the geotechnical array. Figure 1 shows the location of Delaney Park in relation to the Atwood building in the skyline behind the park. Figure 2 shows a schematic of the vertical array with seven 3-component accelerometers located at depths from the surface down to 61 meters. The instrumentation of the Atwood building is also shown in Figure 2. This is an ideal case where the input motion to a wellinstrumented structure is provided through a GSMA located so as to provide the free field input motions to that structure. This instrumented site represents the new trend in collaborative geotechnical and structural engineering monitoring sites.

235

Figure 1. The Delaney Park GSMA with the instrumented Atwood Building in the background

Figure 2. Schematic representation of the GSMA and Structural Array monitoring at Delaney Park and the Atwood Building in Anchorage, Alaska.

236

Moderate Arrays (2-3 borehole sensors): The second category of GSMA facilities in the United States is “moderate” arrays, or sites with two or three borehole accelerometers in addition to a surface sensor. The list of 15 such arrays provided in Table 2 is again heavily weighted towards California. However, looking at these less extensive (and therefore less expensive) arrays, the lower seismicity regions begin to be represented here, with three moderate array sites located in the Central and Eastern United States, in this case operated by the University of Kentucky and Northeastern University, respectively. The same agencies and institutions that operate the previously discussed extensive arrays operate the remaining twelve sites. An example of the moderate class of GSMA is the Southern California Earthquake Centers (SCEC) Long Beach Water Replenishment District (WRD) site in Southern California. This site has a relatively deep borehole accelerometer at 350 meters, an intermediate depth accelerometer at 30 meters, and the surface accelerometer. This is a typical soft deep-basin site in Los Angeles with “engineering rock” just barely being encountered at 350 meters depth. Table 2. Moderate Array sites in the United States Site Name Carquinez Bridge Corona Array Foster City, San Mateo Bridge Half Moon Bay La Cienega Array Long Beach Water District Northeastern University Olmstead Locks and Dam Paducah, KY Parkfield, Turkey Flat Rohnert Park San Francisco Bay Bridge Sassafras Ridge, KY University of California, Riverside Winfield Scott School

Location Northern California Southern California Bay Area – Northern California Northern California Southern California Southern California Eastern US Central US Central US Central California Northern California Bay Area – Northern California Central US Southern California Northern California

Agency(s) CGS/CSMIP/Caltrans CGS/CSMIP/Caltrans CGS/CSMIP/Caltrans CGS/CSMIP/Caltrans CGS/CSMIP/Caltrans UCSB/SCEC Northeastern University University of Kentucky/ACOE University of Kentucky CGS/CSMIP/Industry CGS/CSMIP/Caltrans CGS/CSMIP/Caltrans University of Kentucky UC/SCEC USGS

Surface Borehole Pair (1 borehole sensor): The third category of GSMA sites in the United States is the Surface Borehole Pair. The list below of 17 such sites are again a combination of Central US and California sites, weighted heavily toward Southern California (Table 3). The majority of these sites were targets of opportunity where an existing strong motion station was being drilled and logged to collect geotechnical site characterization data, and another agency (SCEC for example) would leverage the drilling and logging costs to justify funding the installation of well casing in the hole and the installation of a borehole sensor. Since many of these sites exist because site characterization was occurring at the station, information regarding the soil profiles at these sites is available. Having good site characterization data is critical for future analysis of strong motion records obtained at these GSMA sites. Many of these sites are also collaborative with existing monitoring networks where much of the infrastructure (power and communications) is already in place at the site for the existing strong motion instrument. While these sites will not provide the extensive array multi-level detail within the soil column needed to calibrate complex nonlinear simulation models, these sites are still extremely useful. They provide the input motion at a depth where the material behavior during strong shaking in significant earthquakes is expected to remain in the linear stress-strain regime. Thus, they are the control motion that can at a minimum be used to evaluate the degree of nonlinear soil behavior as observations are made at different excitation levels. Assuming that the behavior during small earthquakes is linear throughout the soil column, changes in the site response transfer function from

237

borehole to surface as input motions increase with larger earthquakes can be interpreted as changes in the material behavior (Tsuda and Steidl, 2006; Assimaki and Steidl, 2007). With enough of these observations at various sites, over a range of input motions and site classifications, average site response correction factors can be determined and used in seismic code provisions. Currently, these factors are based primarily on theory, not empirical data. So while these surface-borehole pair sites may not be extensive enough for some research applications, they do make an important contribution. Table 3. Surface Borehole Pair Sites in the United States Site Name Central Fire Station Cerritos College Griffith Park Jensen Filtration Plant Kentucky Bend, KY Mira Catalina School Obregon Park, Los Angeles Pacific Park Plaza, Emeryville Ridgely, TN Rinaldi Substation Stone Canyon Superstition Mountain Tarzana, Ceder Hill University of California, Los Angeles University of California, Santa Barbara University of California, San Diego Wonderland Avenue School

Location Southern California Southern California Southern California Southern California Central US Southern California Southern California Bay Area – Northern California Central US Southern California Southern California Southern California Southern California Southern California Southern California Southern California Southern California

Agency(s) USGS/UCSB USGS UCSB/SCEC UCSB/SCEC/USGS University of Kentucky UCSB/SCEC CGS/CSMIP/UCSB USGS University of Kentucky UCSB/SCEC UCSB/SCEC UCSB/SCEC/USGS CGS/CSMIP USGS/ANSS UC/SCEC UC UCSB/SCEC

HIGHLIGHTS OF ACTIVITIES AT THE NEES SITES The U.S. National Science Foundation George E. Brown Jr. Network for Earthquake Engineering Simulation (NEES) program provides an unprecedented infrastructure for research and education, consisting of networked and geographically distributed resources for experimentation, computation, model-based simulation, data management, and communication. Rather than placing all of these resources at a single location, NSF has distributed its resources among 15 equipment sites throughout the US. To insure that all earthquake engineering researchers can effectively use this equipment, the equipment sites are operated as shared-use facilities, and NEES will be implemented under a new paradigm, as a network-enabled collaboratory. As such, members of the earthquake engineering community are able to interact with one another, access unique, state-of-the-art instruments and equipment, share data and computational resources, and retrieve information from digital libraries without regard to geographical location. One of these 15 NEES equipment sites is the NEES@UCSB facility, which includes two permanently instrumented field sites constructed for the study of ground response, ground failure, soil-foundationstructure interaction, and liquefaction. The two sites are located in Southern California. Both sites are located close to major faults and have previous histories of recording ground motions and pore-water pressures. They also have a history of site characterization studies, and both sites are underlain by soft, liquefiable ground. These field sites are well suited for ambient noise studies, passive earthquake monitoring, and active testing using mobile shakers. Garner Valley Downhole Array The NEES Garner Valley Downhole Array (GVDA) is located in southern California at a latitude of 33° 40.127’ north, and a longitude of 116° 40.427’ west. The instrumented site is located in a narrow valley within the peninsular ranges batholith east of Hemet and southwest of Palm Springs, California. This seismically active location is 7km from the San Jacinto Fault and 40 km from the San Andreas

238

Fault. The valley is 4-5 km wide at its widest and about 10 km long. The valley trends northwestsoutheast parallel to the major faults of southern California. The valley floor is at an elevation of 1310 m and the surrounding mountains reach heights slightly greater than 3,000 m. A panoramic view of the GVDA field site is shown in Figure 4, taken at the completion of the NEES construction in Fall of 2004. The details of the geotechnical site conditions and instrumentation at the GVDA facility can be found at the NEES@UCSB website (http://nees.ucsb.edu/), and in previous studies of the observations from this site (Archuleta et al., 1992; Steidl et al., 1996; Bonilla et al., 2002).

Figure 4. The GVDA site in 2004 after the NEES program upgrade. The NEES GVDA facility exemplifies the trend of instrumented sites moving to multi-disciplinary collaborations between seismologists, geotechnical, and structural engineers. The reconfigurable structure (Figure 4) constructed at the GVDA site is instrumented with pressure cells under the four corners of the foundation, vertical displacement transducers on the four corners, accelerometers on the corners, bottom slab, and top slab, and a rotational sensor on the bottom slab. In addition, a downhole accelerometer and pore pressure transducer are installed below the foundation. The new structure is intended for improving our understanding of soil-foundation-structure interaction (SFSI). Figure 5 is a schematic of the structure and the different input forces that can be used in response testing.

Figure 5. The various input forces used to study soil-foundation-structure interaction at GVDA.

239

In figure 6, the locations of the various sensors installed on and beneath the SFSI test structure are shown. In addition to the instrumented structure, the soil column at GVDA is heavily instrumented with 6 additional downhole accelerometers (6, 15, 22, 50, 150, and 501 meter depths) and 4 additional pore pressure transducers (6.1, 8.8, 10.1, and 12.4 meter depths).

Figure 5. Instrumentation at the GVDA SFSI Facility The SFSI test structure instrumentation is designed to easily capture both rocking and torsional modes of the structure. It was also designed to be re-configurable, so that the stiffness could be modified by adding or removing bracing on any of the 4 sides. The mass of the structure can also be modified through the addition of weight on the roof slab, or even the addition of a second story. A permanent shaker is mounted under the roof slab, and can be operated remotely, providing an excellent tool for teaching SFSI and structural dynamics concepts. The shaker is also used in research by exciting the structure on a regular basis and comparing the response with environmental factors like soil saturation and temperature. A weather station is installed at the GVDA site to provide rainfall data and temperature data, and soil moisture probes are installed below the foundation of the structure. Liquefaction array monitoring at GVDA In addition to the SFSI test facility and ground response sensors at GVDA, pore pressure monitoring in the near surface soil layers is providing new observations to help better understand the liquefaction

240

phenomena and nonlinear material behavior in general. The largest motions recorded so far at the GVDA site are just at the level where the onset of nonlinear soil behavior might be expected, around 0.1 g peak ground acceleration. Observations from the liquefaction array sensors are beginning to show the build up of pore pressure at this level, and then show the slow decay back to the background level. A recent M5.1 earthquake near Anza, CA produced a quality set of observations showing this behavior (Figure 7). Interestingly, the shallow transducers show increases in pore pressure during the strongest shaking, while the deeper transducer seems to show an opposite effect. It is expected that these observations will be used for many years to come as simulation techniques are tested and new physics based models are proposed to model this behavior.

Surface Acceleration (PGA ~

Pore Pressure increase with S-wave and slow recovery back to pre-event level

Figure 7. 150 seconds of surface ground acceleration and sub-surface pore pressure observations from the 2005 M5.1 Anza event recorded at GVDA

Wildlife Liquefaction Array The Wildlife Liquefaction Array (WLA) is located on the west bank of the Alamo River 13 km due north of Brawley, California and 160 km due east of San Diego. The site is located in the Imperial Wildlife Area, a California State game refuge. This region has been frequently shaken by earthquakes with six earthquakes in the past 75 years generating liquefaction effects at or within 10 km of the WLA site. Based on this history, there is high expectation that additional liquefaction-producing earthquakes will shake the WLA site during the 10-year operational phase (2004-2014) of the NEES program. Figure 8 is a view of the WLA site after construction was completed in Fall 2004.

Figure 8. The NEES WLA facility just after construction was completed in 2004.

241

The extensive instrumentation at this site includes 4 surface accelerometers, 6 downhole accelerometers, 11 sub-surface pore pressure transducers, and numerous benchmarks and inclinometer casings for monitoring lateral ground displacements. In this paper, we will focus on some of the experimental activities that have taken place since its inception in 2004, and refer the reader to other publications which contain detailed information on the instrumentation and geotechnical properties of the site (Youd et. al., 2004; Youd et. al., 2007). Both earthquakes and active testing using the NEES@UTA “T-Rex” mobile shaker have been used to examine the response of the WLA site. In the late summer of 2005 the “T-Rex” shaker excited the WLA site and provided a useful test of the system, as well as some provocative observations of pore pressure during local shaking. Figure 9 shows the location of the shaker relative to the accelerometer and pore pressure instruments at the WLA site.

Figure 9. The WLA instrumentation and T-Rex shaker location during testing in 2005. The active shaking from the NEES@UTA shaker only lasts for approximately 10 seconds, however the pore pressure signals that are generated by the active source last for minutes. Pore pressures at the two transducers located closest to the shaker (P7 and P8 in Figure 10) show an immediate increase in pore pressure and slow decay back to the pre-shake level that takes more than 20 minutes. Another interesting observation is that just following the initial pulse, the deeper of the two closest transducers (P8) also shows another slight pressure increase, which can be explained by a pressure wave generated at the source and spreading out and propagating through the sand layer as it dissipates. All of the transducers that are located at or near the top of the sand layer at the site (see Youd et al., 2004, 2007 for details of the soil layers) show an impulsive initial arrival, including the two located further from the shaker (P4 and P1), with the amplitude of the pulse decreasing with both lateral and vertical distance from the source. The time histories shown in Figure 10 are sorted by distance from the shaker source with the top trace being the closest and bottom trace the furthest. Pore pressure continues to rise for as much as three minutes after the shaking has stopped at the deeper and further transducers. Even

242

at 10 minutes after the source has stopped the pore pressures have still not dissipated back to the preshake level.

Figure 10. The WLA instrumentation and T-Rex shaker location during testing in 2005. These active source observations of pore pressure generation and slow dissipation are also seen when earthquakes shake the site. The largest ground motions observed to data at the WLA site are from a local magnitude 5.1 event located approximately 10 km from the site. In addition to the M5.1 event, numerous M3 and M4 level earthquakes also generated very interesting observations at the site. The largest ground accelerations from this swarm of events was only about 10% g, however even these modest levels of shaking, the pore pressure response can clearly be seen in the observations. The observations from the eight pore pressure transducers for a 1-hour period during this swarm of earthquakes is plotted in Figure 11 from shallowest to deepest (top to bottom). Similar to the active source testing, all of the transducers show a clear response to the earthquake activity. In the earthquake swarm observations, the deeper transducers have a larger dynamic response to the passage of the body waves from the events. All transducers show slight pore pressure increase and slow recovery back to the pre-swarm level. The deeper transducers also have this increased pore pressure, but it’s harder to see due to their larger dynamic response. The pore pressure increase during the swarm is approximately 0.2-0.3 kPa, while the dynamic response of the deeper transducers is almost an order of magnitude larger, at approximately 3-5 kPa peak to peak (Figure 11).

243

P4 - LQ1 – 2.7m M4.4

M4.6 P1 - LQ2 – 2.9m P7 - LQ3 – 3.4m Shallow Pore Pressure Increase and Slow Recovery

P3 - LQ4 – 3.7m P4 - LQ5 – 4.4m P8 - LQ6 – 4.4m P2 - LQ7 – 6.4m

P5 - LQ8 – 6.4m

Figure 11. The WLA liquefaction array. Sixty seconds of data showing both dynamic response of the pore pressure sensors and static pressure increase, with slow dissipation when excited by the 2005 Obsidian Buttes earthquake swarm. In addition to the active source and earthquake monitoring at the WLA site, the benchmarks and inclinometer casings are re-surveyed approximately once per year to obtain a baseline for lateral displacements. The free face of the Alamo river bank located just meters from some of the instrumentation, benchmarks, and casings should provide an excellent source for lateral spread activity in the next large earthquake.

CONCLUDING REMARKS In order to maintain a vital monitoring and research program at instrumented geotechnical sites, often many different funding agencies must be involved, and a very broad scope of work must be envisioned. Coordination and collaboration among seismologists, geotechnical and structural engineers to establish instrumented sites that serve the engineering and scientific goals of each of these disciplines represents the current trend for increasing the number of these facilities. It would be useful to also get the practitioners involved, in both defining the scientific and engineering goals for instrumented sites, and also to get instrumentation planning as part of the design phase for new structures. At present there is a lack of coordination among the various responsible organizations, both at the domestic level in the United States as well as internationally, when it comes to the operations and dissemination of data from instrumented geotechnical array sites worldwide. The maintenance and operations of the existing instrumented geotechnical site resources and dissemination of their data should be considered a high priority. There is a need for an umbrella organization or working group

244

with representatives from both the national and international agencies with monitoring programs to facilitate the coordination and collaboration between these agencies. The operations and maintenance of instrumented geotechnical sites, especially in regions of relatively low seismicity, is often a difficult task for the responsible agency or organizations. The lack of recordings of significant events over long time periods can make the funding agencies question the benefit of maintaining these resources. An international organization that regularly re-evaluates the scientific and engineering needs in conjunction with the current inventory of instrumented sites could provide justification needed for agencies to obtain funding for maintaining these facilities.

AKNOWLEDGEMENTS The Author would like to acknowledge the many agencies and individuals who have provided information regarding the operational status of existing GSMA resources, both in the United States and Internationally. The dedication and patience of the organizations that maintain these array sites that serve the geotechnical earthquake engineering and seismological communities is very much appreciated. The author would also like to acknowledge the many funding agencies and collaborators that have been involved with GSMA monitoring at the University of California, Santa Barbara over the last two decades. While many have come and gone, the longevity of these facilities would not have been possible without them. These include the U.S. Nuclear Regulatory Commission, the U.S. Geological Survey, the U.S. National Science Foundation, the French Commissariat à l’Energie Atomic, the Electrical Power Research Institute, Agbabian Associates, Kinemetrics Inc., the California Department of Transportation, Kajima Corporation of Japan, and the Nuclear Power Engineering Corporation of Japan. The GVDA and WLA sites are currently operated under contract to the National Science Foundation as part of the George E. Brown Jr., Network for Earthquake engineering Simulation, award number CMS-0402490. Without the support and cooperation of the Lake Hemet Municipal Water District and the California Department of Fish and Game, the GSMA monitoring at the GVDA and WLA sites would not be possible.

REFERENCES Archuleta, R. J., S. H. Seale, P. V. Sangas, L. M. Baker, and S. T. Swain (1992). Garner Valley downhole array of accelerometers: instrumentation and preliminary data analysis, Bull. Seism. Soc. Am., 82, 1592-1621 (Correction, Bull. Seism. Soc. Am., 83, 2039). Archuleta, R. J. and J. H. Steidl (1998). ESG studies in the United States: Results from borehole arrays, The Effects of Surface Geology on Seismic Motion, Irikura, Kudo, Okada, and Sasatani (eds), Balkema, Rotterdam, v 1, p.3-14. Assimaki, D. and J. H. Steidl (2007). Inverse analysis of weak and strong motion borehole array data from the M w7.0 Sanriku-Minami earthquake, Soil Dynamics and Earthquake Engineering, 27, p. 73-92. Bonilla, L. F., J. H. Steidl, J-C. Gariel, and R. J. Archuleta (2002). Borehole response studies at the Garner Valley downhole array, southern California, Bulletin of the Seismological Society of America, 92, p. 3165-3179. Steidl, J. H., A. G. Tumarkin, and R. J. Archuleta (1996). What is a reference site? Bulletin of the Seismological Society of America, 86, pp.1733-1748 Tsuda, K., and J. H. Steidl (2006). Nonlinear site response from the 2003 and 2005 Miyagi-Oki earthquakes, Earth, Planets, and Space, 58, 1593-1597. Youd, T.L., J. H. Steidl, and R. L. Nigbor (2004), “Lessons learned and need for instrumented liquefaction sites”, Soil Dynamics and Earthquake Engineering, vol. 24, Issues 9-10, p 639-646. Youd, T.L., J. H. Steidl, and R. A. Steller (2007), “Instrumentation of the Wildlife Liquefaction Array”, Proceedings of the 4th International Conference on Earthquake Geotechnical Engineering, June 25-28 th, 2007, Thessaloniki, Greece.

245

4 th International Conference on Earthquake Geotechnical Engineering June 25-28, 2007 Paper No. W1-1001

DENSE SEISMIC INSTRUMENTATION OF SMALL SOFT BASINS.

Bill (W.R.) STEPHENSON 1 ABSTRACT The paper describes the evolution of dense arrays of seismographs on soft sites in New Zealand, detailing the overall seismic response of each of the three most valuable of these arrays. The basins had lateral dimensions of around 400 m and depths of a few tens of metres, and were densely instrumented with three-component velocity seismographs in order to investigate extreme site response to earthquake shaking. At each site a month of recording resulted in several tens of useful earthquake records. These recordings were supplemented with geophysical and geotechnical investigations. Although all three basins were roughly equivalent in their size, depth and flexibility, their responses to earthquakes were strikingly different. The Parkway basin shows several valley-wide frequencies, which correlate with peaks in the Horizontal-to-Vertical Spectral Ratios of microtremors. These valley-wide frequencies are associated with large amplifications. The methods used to elucidate the wave-propagation basis of these amplifications are described, together with comments on international cooperation, data sharing and future experiments. Keywords: Basin Array, Resonance

INTRODUCTION Recent rapid advances in technology have made it economically possible to operate dense seismograph arrays in areas of soft soil where both earthquake-related damage, and the amplitude and duration of earthquake shaking are known to be abnormally great, or are suspected to be abnormally great. Various agencies have taken advantage of these advances, and have installed seismograph arrays of various types in such soft areas. The arrays have varied in the type of instrumentation used, the inter-station distance, and the duration of the field deployment. Both surface arrays and threedimensional arrays have been implemented, some being strictly of a temporary nature, and some being more permanent. A summary of array locations and properties was produced by Zerva et al (2004). Since the time of that summary other arrays have been implemented, notably the Cavola array (Bordoni et al, 2006) in Italy. The majority of the arrays discussed by Zerva et al (2004) do not fall within the compass of the densely instrumented small soft basins that this paper will treat. Examples of the sort of experiment which this paper will not treat are the Roma array in Mexico City (only three locations, and not within a confined basin), the Grenoble array (a small part of a large basin) and the San Jose array in the United States (wide sensor spacing, varying site conditions across the basin). 1

Senior Scientist, Hazards Group, GNS Science, Lower Hutt, New Zealand, Email: [email protected]

246

The use of the term “small soft basins” in the title of this paper is an attempt to restrict the scope of the paper to situations where abrupt discontinuities in material properties in both the lateral and vertical sense define a structure in which energy may be trapped. “Small” is intended to describe areas of less than a few hundred metres diameter, and “soft” is intended to describe shear wave velocities of less than a few hundred metres per second. The prime purpose of this paper is to disseminate information about three seismometer arrays that have been operated on small soft basins in New Zealand, and to encourage further use of the seismic data that they have generated. The paper will also describe the analyses applied to the New Zealand data. The approach to these analyses was conditioned by early observations that suggested that the type behaviour to look for would be the coherent response at a single frequency of a limited area of ground. New Zealand workers have had a long involvement with small arrays, improving their performance as better instrumentation became available and as experience was accumulated. Stephenson (1974) described an early initiative set up in the Hutt Valley in which a trigger system sensing rock motion sent a radio signal to frequency modulated tape recorders, commanding them to record the output of one-second geophones. This system was short-lived because subsequent to preliminary recordings of earthquakes being made, the minicomputer used to digitise the recordings became wholly dedicated to another task and was hence unavailable. At the same time funding to expand the system was diverted to other projects. Nevertheless this first array gave useful insights, stimulating thoughts that a northsouth rock signal could give rise to an east-west resonant soil response. This was formalised by Stephenson (1972) with the suggestion that an appropriate impulse response for the soil would consist of a sequence of two-by-two matrices, thus foreshadowing the current use of matrix transfer functions (Benites, pers. comm.) A second attempted array operation carried out at Pukehou in New Zealand’s North Island (Stephenson & Barker, 1991) was of limited value because strong motion accelerographs were used, with a consequent limitation on the size of useful earthquake. As with the earlier array the system was triggered by a rock-derived radio command signal. Although the Pukehou array was in an area of high seismic activity it was also in a flood-prone region, so that only two stations were operational (the others having been withdrawn in response to rising water) when useable ground motion eventually transpired. The single useable earthquake was recorded at one soil station and one rock station and it provided a clear example of a “directional resonance”. With modern instrumentation and in the absence of flooding the Pukehou site would be an ideal location because it is a small-scale limestone basin filled with peat, thus being bounded on all sides and having a high impedance contrast. The two peat layers had shear-wave velocities of 17 m/s and 34 m/s, while the limestone shear-wave velocity probably was in excess of 1000 m/s and possibly was as much as 2000 m/s. The landowner has expressed reluctance to allow further work because it would interfere with a cash crop. Three subsequent arrays on other basins in New Zealand all yielded useful data. In each of these three cases a basin with a strong valley-wide resonance was sought, but the Alfredton array (Stephenson, 1996) proved non-resonant and as a consequence the Parkway array (Stephenson, 2000) was implemented. A casual inspection of initial seismograms from Parkway and their spectra suggested that valley-wide resonance was absent, and planning therefore started for the Wainuiomata array (Stephenson et al., 2002). The Wainuiomata valley was found to consist of several sub-systems, each showing valley-wide resonance. In the meantime a more sophisticated analysis of Parkway data suggested that valley-wide resonance was present, and that this justified a re-appraisal of the Parkway seismograms. The Parkway array data has proven extraordinarily rich. The locations and responses of these three arrays are discussed more fully in paper 1726 of this conference. The behaviour of each basin was distinct, and was caused by details in the manner in which shear wave velocity varied with depth. If conventional site evaluation criteria had been applied, the three basins would have been deemed equivalent, having similar depths and similar shear wave velocities at the surface.

247

COMPARATIVE RESPONSES OF NEW ZEALAND BASINS Mean Standard Spectral Ratios (SSR) for all well recorded earthquakes at a site near the centre of each of the three basin arrays are shown in Fig. 1. Averaging of spectra over all rock sites, and many earthquakes, ensured that the SSR curves were robust. This evaluation of SSR gives a clear idea of the different responses of the three basins upon which the seismograph arrays were laid out. Standard Spectral Ratios are the ratios of spectra of earthquake motions recorded on soil and on rock. In the case of the basins it was considered best to choose representative stations near the basin centres so as to avoid basin edge effects. 30

10 (c) Wainuiomata W03

(a) Alfredton A02 5

20

0

0

2

4 6 8 Frequency (Hz)

10

15 (b) Parkway P06

10

10

5

0

0

2

4 6 8 Frequency (Hz)

10

0

0

2

4 6 8 Frequency (Hz)

10

Figure 1. Mean values of Standard Spectral Ratio for a site near the centre of each basin array. The dashed lines specify an SSR of 1 and the vertical scales are the same for each plot. Alfredton basin amplifies earthquake shaking to a small extent, over a broad range of frequencies. This is consistent with the shear wave velocities of the soil being low, but becoming progressively higher with depth until an eventual low contrast with bedrock velocity occurs. Parkway basin amplifies earthquake shaking in a complex resonant fashion, with shaking at several frequencies being amplified to various extents. The frequencies being amplified are broadly consistent with what is known about the shear wave velocity profile. Each Wainuiomata sub-basin amplifies earthquake shaking in a resonant fashion, with extreme amplifications at around 1 Hz. At these frequencies the shaking is greatly prolonged. At frequencies above 7 Hz, shaking is attenuated, with the layer of soft soil functioning as a low pass filter. The surface layer of soft soil, having an abrupt impedance contrast with deeper material is responsible for this behaviour.

248

POSSIBLE FUTURE WORK Although recent progress has been very encouraging, the lessons learned from existing arrays could be applied when designing new arrays, and these new arrays could be improved even more by adopting better technology. In addition, the implementation of proven techniques used for the analysis of data from other existing arrays, or from future arrays, could be fruitful. For instance coherency studies were carried out on Parkway array data (Liao et al, 2007) and could be carried out on data from other arrays. The significance of coherency is that it can reveal relative displacements. The motion at Parkway is dominated by fast-travelling waves which should maintain high coherency, whereas the motion between sites at Alfredton is less related, and lower coherency (meaning greater relative displacement) might be expected. Other possible projects might implement the technique of filtering followed by progressive component rotation and the calculation of wavenumber spectra, and the technique of forming real-time movies of particle motion. Both of these techniques have been demonstrated at Parkway and proved most fruitful. The underlying reason for studying the earthquake-induced motion in small basins is to understand why buildings founded on such basins may be subject to much greater damage than buildings on nearby firm ground. It is noteworthy that on the one hand the most comprehensive arrays have been operated in areas where no excess damage has been observed, and on the other hand that arrays in areas associated with excess damage have been limited by small size or lack of dense coverage. Thus to fully understand situations with excess damage it is necessary to properly instrument a restricted location where such damage is known to occur. An opportunity therefore exists to implement a new temporary dense seismograph array in the La Molina alluvial basin near Lima, Peru. Successive damaging earthquakes in 1940, 1966 and 1974 showed that local soil conditions increased intensities from MMVI on nearby stiff soil to as much as MMIX in La Molina, (Kuroiwa, 2002) and a recent as-yet unpublished analysis of a 1984 temporary array deployment in La Molina shows that four of the seven soil stations show a peak in ground motion at 1.2 Hz; the frequency that dominated the velocity response spectrum of an aftershock of the 1974 earthquake. This observation of a single frequency in a restricted area is reminiscent of the initial observations at Parkway (Stephenson, 2000) that eventually led to the conclusion that a superposition of modified Rayleigh waves and modified Love waves travelled down valley, guided by the valley walls. At Parkway, classic Rayleigh waves were modified by a variation in amplitude across the valley. For Love waves, the modifications likewise involved a variation across the valley, but also the adoption of a vertical component near the valley walls in order to accommodate pure shear, while simultaneously allowing the in-plane displacement to fall to zero at the valley walls. A dense array of modern seismographs covering an area of 1 km diameter could resolve the behaviour of the soil within the resonant area of La Molina, perhaps implying behaviour equivalent or similar to that at Parkway. Arrays to date have often used recorders that proved to have inadequate resolution, or to be fixed gain rather than gain-ranging, or to rely on triggered operation. In the case of the 1984 La Molina deployment all three of these deficiencies were present. Any future deployment would best use at least 16-bit gain-ranging recorders and would record ground motion continuously, with useful data being extracted later. This was the procedure followed in the 2004 Cavola array deployment. The Cavola instrumentation would be an excellent bechmark to keep in mind when designing new arrays. A feature of all three New Zealand array deployments was that they were insufficiently supported by geotechnical observations. Because this was not an intended outcome there is a lesson to be learned. In each case it was thought on reasonable grounds that the site consisted of horizontal layers, and that it would be sufficient to characterise each layer using SCPT, and then to determine the extent of each layer with sparse CPT probes. However the Parkway site had an impenetrable thin gravel layer at depth, the Alfredton site was not built up from horizontal layers, and the Wainuiomata site was actually a series of independent sub-basins. It is usual for the costs of an array project to be estimated

249

beforehand, and for the costs of additional work to be difficult to obtain. This was the case for these projects, and only a special grant enabled characterisation at Parkway below the gravel layer. It would be wise when pricing future arrays, to assume an absolute worst-case scenario when planning geotechnical aspects. A strength of all three New Zealand array deployments was that their planning included four rock stations with the intent that the incoming wavefield would be adequately characterised. Yu and Haines (2003) reported that the use of the average motion from four rock sites provided a usefully improved reference signal.

DATA ANALYSIS To date, approaches to the analysis of dense array data recorded by seismographs on soft basins have mainly used techniques adapted from those used for arrays installed at homogeneous sites, where simple phases, well separated in time, are expected. These analyses are often carried out within a wavenumber space and address horizontal propagation. It is usually assumed that Love waves will be associated with solely horizontal motion, Rayleigh waves with a preponderance of vertical motion, and bulk waves with both vertical and horizontal motion. These assumptions may be true for wave motion in a halfspace, but they are not borne out for a layered site in a horizontally-restricted context. The response of a restricted volume of soil, often with its extent being greater than its depth, can be very complex. Monochromatic waves at different frequencies, with different polarisations, and travelling in different directions, will be superposed. The task of unravelling these contributions has been relatively neglected, but the start that has been made at Parkway by filtering followed by a sequence of component rotation and computation of wavenumber spectra, shows what can be done. Summarising the procedures used at Parkway, Fig 2 shows the result of summing Parkway spectra over all soil-based stations and all earthquakes. The peaks at 1.58 Hz and 1.68 Hz show the importance of these frequencies, and subsequent attention was focussed on 1.58 Hz. A long duration earthquake was then chosen so that meaningful narrow band filtering could be used.

Figure 2. Summed normalised spectra of all horizontal seismograms at Parkway, including both components, for all soil sites and all earthquakes. There are site-wide responses at 1.58 Hz, 1.68 Hz and 2.1 Hz. For this long duration earthquake, Figs 3a and 3b show wavenumber spectra for north components and for vertical components, for motion between 1.53 Hz and 1.63 Hz. It is noteworthy that the implied propagation for the north component is neither along nor orthogonal to, the north axis. Furthermore

250

the velocities for the vertical components and for the north components bear no relationship to each other.

(a)

(b)

Figure 3. Wavenumber spectra at Parkway for a long-duration earthquake. (a) This does not correspond to either a transverse or a longitudinal wave. (b) Vertical components of soil stations at 1.58 Hz. This bears no relationship to part (a). However when the coordinate system is rotated to 30º and 120º (the axes natural to the basin), the result is much more reasonable as shown in Fig 4a and Fig 4b. These figures show a transverse wave travelling down valley at 1.3 km/s, and a longitudinal wave travelling down valley at 2.7 km/s. The wavenumber spectrum of the vertical motion, however, does not correspond to the longitudinal wave (imagined to be a Rayleigh wave).

(a)

(b)

Figure 4. Wavenumber spectrum at Parkway for a long-duration earthquake. (a) 30º components of soil stations at 1.58 Hz; 2.7 km/s longitudinal wave. (b) 120º components of soil stations at 1.58 Hz; 1.3 km/s transverse wave. Both waves travel down-valley (210º). When the vertical motion is partitioned between the longitudinal wave and the transverse wave as described by Stephenson (2002) it is clear that the longitudinal wave is a Rayleigh wave guided by the valley and that the transverse wave is similar to a Love wave except that in accommodating the boundary conditions corresponding to a large impedance contrast at the valley wall, some vertical

251

particle motion is involved as indicated by Fig 5. This is further elaborated in paper 1726 of this conference, and in Stephenson (2007).

Figure 5. Freeze-frame from a movie of postulated transverse wave, showing how the constraint of the valley wall enforces some vertical motion. This cross-sectional view illustrates the “sloshing” nature of the wave. The most recent analysis technique applied to Parkway data used real-time movies of particle motion as shown by Stephenson (2007). It clearly shows the nature of the propagating waves and also points to other patterns of ground motion.

MODEL VERIFICATION Workers who devote their efforts to understanding site response tend to fall into one of the two broad camps of data analysis and of modelling, although there may sometimes be overlaps. Both approaches are valid provided that the ultimate goal is attaining a level of understanding which allows future site response to be predicted accurately, independent of the source earthquake. This understanding could be considered complete when (given the incident wavefield) every nuance of every component, at every station, for every earthquake, was reproduced in the modelling. At the moment this is not so. It is common to correctly predict frequencies, but much harder to correctly predict durations. The different analytical techniques described in the previous section all have a part to play in model verification. The situation may be likened to viewing a scene through various colours of glass, with each technique giving emphasis to a different aspect of ground motion. Only when a given model agrees with the data as seen from all these viewpoints should it be considered as being accurate.

INTERNATIONAL COLLABORATION The planning, funding and implementation of a temporary seismograph array of more than 20 recorders is a non-trivial task. Experience in New Zealand suggests that a five year gap between the decision to run a dense seismograph array, and the final availability of both ground motion and geotechnical data, is normal. This time includes initial data reduction plus geotechnical and geophysical studies. The result is typically gigabytes of data which require considerable time and effort to interpret. It is easy to understand the perspective of a researcher who has invested five years of effort in acquiring a great deal of data and who wants to reap some eventual rewards from analysing it, without being forestalled by somebody who has not been involved in the project. At the same time it is easy to understand the perspective of a funder who might take the point of view that public money should result in publicly available data. The funder could correctly assert that the full value of data is only achieved when it is thoroughly analysed from multiple points of view. Perhaps one answer lies in making data freely available on condition that the original team members are sufficiently involved in each use of the data that they become co-authors. This approach has certainly worked in the case of

252

the Parkway array data, and it has the advantage that a people who have a close knowledge of the site can prevent mistakes being made. It is for instance very easy from the other side of the world to insert the wrong valley in a diagram. Funding for travel should be incorporated in collaborative work both to build a true team and to enable non-local workers to get a hands-on feel for the local situation. Another way of ensuring that the people who did the work got some credit would be to establish a foundation paper for each dataset listing all contributors as authors, and to provide copies of the data only on the understanding that the foundation paper be cited in all resultant publications. Finally, the possibility of totally international projects should be considered. This would involve international funding and international operation as well as international data sharing. That would require a considerable organisational effort and would be subject to issues of parochialism.

LESSONS LEARNED Dense temporary arrays of velocity seismographs can provide valuable insights into the responses of small basins filled with soft soil. In the short deployment times of a few weeks the earthquakes recorded by the New Zealand arrays were seldom felt – none for the Alfredton array, one for the Parkway array and one for the Wainuiomata array. Therefore the recordings only apply to low amplitude shaking, where the soil is expected to behave in a linear manner. However an understanding of linear response is a prerequisite to a full understanding. As technology has improved it has become more feasible to record ground motion continuously. At the times that the three New Zealand arrays of this paper were set up however, data storage was at a premium, and it became necessary to allow each recorder to operate autonomously, using preset criteria to determine whether a given data stream was worth recording. We used the STA/LTA method (ratio of short term average to long term average), with an overarching amplitude criterion in case of strong shaking of an emergent nature. This was because during the earlier phases of recording at Alfredton it was found that the signal on soft soil caused by a distant earthquake often rises slowly, and the STA/LTA approach can fail. Our approach today would be to record everything, to examine the recordings on a daily basis, and to reject non-earthquake data quite early on. The establishment and operation of the Alfredton array, together with later recordings of microtremors, confirmed that microtremor recordings can be used to distinguish between resonant and non-resonant basins. Had microtremors been recorded and analysed prior to the Alfredton project, that array would not have been installed because a valley-wide response was being sought. In the case of the Parkway and Wainuiomata arrays, microtremor recording provided insights that both these areas would undergo valley-wide response. Microtremors played an important role in choosing these areas in preference to other areas simultaneously under consideration. In terms of the analytical techniques employed, the varied inter-station spacings employed for the Parkway array were preferable to the regular grid of the Alfredton array or the large inter-station spacings of the Wainuiomata array. When a selection of inter-station spacings is used, spatial aliasing is avoided while at the same time both high and low wave speeds can be accommodated. In the case of studying the behaviour of coherency with distance a spread of inter-station distances enables a better spread of data points, better enabling relevant curves to be plotted. In respect of array layout, Parkway was superior to both Alfredton and Wainuiomata. At Parkway four stations on the rock surrounding the basin were used. The same was not possible in Wainuiomata because one site owner withdrew consent at short notice a few days after recording had started. Four stations allowed better averaging of the rock signal, and potentially can be used to reconstruct the fronts of incident plane waves.

253

Seismic CPT proved an invaluable tool for characterising the basins, especially when the velocity inversion that is present at Parkway is considered. It was our intent for each basin to carry out a very few SCPT probes, and then to correlate any layers identified by CPT with known velocities. At Alfredton this failed because of the inhomogeneity of the sediments; at Parkway it failed because CPT probes were unable to penetrate a hard layer; and at Wainuiomata severe problems were encountered with groundwater because a deep aquifer carried water at high pressure and penetrometry to this depth resulted in flooding in urban parks, with attendant problems of sealing a high pressure leak. The computation of spectra in the wavenumber domain, combined with frequency filtering and component rotation allowed complex propagation phenomena to be untangled, and simplex optimisation allowed vertical motion at Parkway to be partitioned between two waves. When this approach failed at one frequency, displaying the motion as a real-time movie was very instructive, showing the reality (but insignificance) of torsional types of response. Much has been learned from studies of the three basins from the perspectives of both understanding basin response, and instrumenting basins. Nevertheless much still remains to be gained from further studies. At Alfredton the behaviour of coherency as a function of frequency and station separation remains unstudied, and a simple technique to predict the lack of valley-wide response needs to be devised. At Parkway a valley-wide resonant response at 2.05 Hz remains unexplained. For Wainuiomata there still exists a possibility of valley-wide response for the whole valley system including sub-basins. Summing normalised spectra over all soil sites and all earthquakes as accomplished for Parkway (Stephenson & Chávez-García, 1998) could be a useful way of investigating this.

ACKNOWLEDGEMENTS It is difficult to name all the individuals who have influenced the writing of this paper. Many staff at GNS Science, both technicians and scientists, have made the paper possible, and they are hereby thanked. Likewise my overseas colleagues have provided endless ideas and guidance and they are thanked. It is appropriate however to specifically thank Carolyn Hume who drafted the diagrams, and Nick Perrin and Rafael Benites who critically appraised the text.

REFERENCES Bordoni P, Cara F, Cercato M, Di Giulio G, Haines AJ, Milana G, A. Rovelli A, Ruso S. “The use of a very dense seismic array to characterize the Cavola (northern Italy) active landslide body.” Poster 44, Third ESG International Symposium, August 2006. Kuroiwa, J “Reducción de desastres: viviendo en armonía con la naturaleza”. Ediciones del PNUD, Perú, Lima, January 2002. Liao S, Zerva A and Stephenson WR "Seismic spatial coherency at a site with irregular subsurface topography” Proceedings, ASCE Conference “New Peaks in Geotechnics”, Denver, Colorado, February, 2007. Stephenson, WR “An experimental study of normal modes of vibration of saturated alluvium”. In: Proceedings : Symposium on Earthquake Engineering, 5th, Roorkee, India, Nov. 9-11. 119-126, 1974. Stephenson, WR “Microzoning for design” in “Structural design for earthquakes” University of Auckland, New Zealand, August 1972. Stephenson, WR “Basin response to earthquakes : the Alfredton soft soil site”. Institute of Geological & Nuclear Sciences science report 96/03. Institute of Geological & Nuclear Sciences Lower Hutt, New Zealand, 1996. Stephenson, WR “The dominant resonance response of Parkway Basin”. In: 12WCEE 2000 : 12th World Conference on Earthquake Engineering. Upper Hutt, NZ: New Zealand Society for Earthquake Engineering. Proceedings of the World Conference on Earthquake Engineering, 2000.

254

Stephenson, WR “Guided Love- and Rayleigh-waves in Parkway Valley, Wainuiomata, N.Z.” Bulletin of the New Zealand Society for Earthquake Engineering, 35(4), 255-265, 2002. Stephenson, WR "Visualisation of Resonant Basin Response at the Parkway Array, New Zealand" Soil Dynamics and Earthquake Engineering, 27(5), 487-496, 2007. Stephenson, WR and Barker, PR “Results from the Pukehou array”. In: Pacific Conference on Earthquake Engineering, Auckland, New Zealand, 20-23 November 1991, Proceedings, Volume 3. 229-238 Wellington: New Zealand National Society for Earthquake Engineering. 1991. Stephenson, WR, Barker, PR, and Yu, J. “The Wainuiomata "local effects" seismograph network”. Bulletin of the New Zealand Society for Earthquake Engineering, 35(4) 243-254, 2002. Stephenson, WR, and Chávez-García, FJ “Preliminary assessment of resonant phenomena recorded by the Parkway Network”. Institute of Geological & Nuclear Sciences science report 98/18 Institute of Geological & Nuclear Sciences Lower Hutt: New Zealand 1998. Yu, J and Haines, AJ “The choice of reference sites for seismic ground motion amplification analyses: case study at Parkway, New Zealand”. Bulletin of the Seismological Society of America, 93(2): 713-723, 2003. Zerva, A, Piasecki, M, Liao, S and Yargici, V. “Spatial (ground surface) array data: observations and needs”. http://www.cosmos-eq.org/Projects/GSMA/Paper/3.2_Zerva_et_al.pdf. 2004

255

4 th International Conference on Earthquake Geotechnical Engineering June 25-28, 2007 Paper No. W1-1003

RECENT DEVELOPMENTS OF GEOSYNTHETIC-REINFORCED SOIL STRUCTURES TO SURVIVE STRONG EARTHQUAKES

Fumio TATSUOKA1, Junichi KOSEKI2, Masaru TATEYAMA3, Daiki HIRAKAWA4 ABSTRACT A geosynthetic-reinforced soil (GRS) retaining wall (RW) technology has been developed to replace conventional-type soil RWs, aiming at higher cost-effectiveness as well as higher seismic stability. With this technology, full-height rigid (FHR) facing is staged-constructed so that the geosynthetic reinforcement layers are firmly connected to the back of the facing. The following is reported: a) lessons from the failure of conventional RWs on level ground during the 1995 Kobe Earthquake in Japan and those on slopes during the 1999 ChiChi Earthquake in Taiwan and 2004 Niigata-kenChuetsu Earthquake in Japan; b) reconstruction of damaged conventional-type slopes and RWs to GRS RWs with FHR facing; and c) related researches by model shaking table tests. Three new bridge types with geosynthetic-reinforced backfill (i.e., vertically preloading and prestressing the geosynthetic-reinforced backfill, using lightly cemented gravel reinforced with geosynthetic reinforcement and reinforcing the backfill of integral bridge with geosynthetic reinforcement layers connected to the parapet) developed also aiming at higher cost-effectiveness as well as higher seismic stability are described. Keywords: Bridge abutment, full-height rigid facing, geosynthetic reinforcement, soil retaining wall

INTRODUCTION Due to vast and serious damage to a great number of civil engineering structures during the 1995 Hyogo-ken Nambu Earthquake (i.e., the 1995 Kobe Earthquake), the seismic design codes for RC and steel structures in Japan were revised substantially so that the structures can survive such high level seismic load (i.e., the so-called level II design seismic load). A great number of embankments, retaining walls (RWs) and bridge abutments with backfill for railways, highways and residential areas were seriously damaged during not only the 1995 Kobe Earthquake (Fig. 1), but also many other major earthquakes in Japan, including the 2004 Niigata-ken Chuestu and 2007 Noto Hanto Earthquakes. After the 1995 Kobe Earthquake, the seismic design codes for new construction of soil structures have also been revised. However, the revision was generally not as substantial and sufficient as those for RC and steel structures. Correspondingly, soil structures that failed by earthquakes have been usually reconstructed to those of original structural types, although it is possible that the reconstructed soil structures would not survive a similar level of seismic load by which the concerned soil structures collapsed. It seems that this situation reflects such a conventional design concept that collapse of soil structures should be accepted for the following reasons. Firstly, it is most important to as soon as possible recover the original function of failed soil structures even if reconstructed soil structures may not have the same level of structural stability as the original. Secondly, the 1

Professor, Department of Civil Engineering, Tokyo University of Science, Japan, Email: [email protected] 2 Professor, Institute of Industrial Science, University of Tokyo, Japan. 3 Chief Research Engineer, Railway Technical Research Institute, Japan 4 Assistant Professor, Department of Civil Engineering, Tokyo University of Science, Japan

256

reconstruction of failed soil structures to those having a seismic stability higher than the original level, preferably to the level withstanding level II design seismic load could be overly costly. Thirdly, as the number of existing important soil structures is so large, it is financially not feasible to rehabilitate all of them to have sufficiently high seismic stability. It is also the case with important soil structures that are to be newly constructed. Lastly, it is generally much easier and faster to reconstruct failed soil structures when compared with the case of RC and steel structures and, so, soil structures could be allowed to fail or collapse.

a) b) Figure 1. a) Gravity type RW without a pile foundation (Ishiyagawa, Hanshin Railway); and b) leaning type unreinforced concrete RW without a pile foundation (Sumiyoshi, JR West) that collapsed during the 1995 Kobe Earthquake (Tatsuoka et al., 1997). Nowadays, however, these rationales listed above are not relevant with many soil structures that are to be constructed as important infrastructures. Firstly, with line structures, such as highways and railways, collapse of soil structures at a limited number of places results in a long closure of the whole system. Secondly, a great number of huge soil structures, such as massive high embankment, have been constructed for residential and industrial areas, railways, highways and so on by means of modern construction machines, which was not feasible previously. It is usually highly time-consuming and costly to reconstruct these soil structures if they collapse. Thirdly, the reconstruction of damaged and failed conventional type soil structures (i.e., embankments with gentle slopes and gravity- or cantilever-type RC soil RWs) to geosynthetic-reinforced steepened slopes or GRS soil retaining walls could be much more cost-effective with a higher seismic stability than reconstruction to the original structural types. The high cost-effectiveness of GRS soil structures in spite of increased structural stability results from smaller earthwork, a higher constructability under restricted conditions (e.g., in a narrow space, on a steep slope or at a remote place) and a higher construction speed. Moreover, it becomes easier to arrange a drainage system in geosynthetic-reinforced backfill than in conventional type soil structures. For the same reasons, the geosynthetic-reinforced soil technology is also useful to newly construct soil structures in many cases. Therefore, there is no strong reason to reconstruct failed conventional type soil structures to those of the original structure types in usual cases. In this report, firstly the geosynthetic-reinforced soil retaining wall (GRS-RW) technology that is now becoming popular is described. Then, several case histories in which conventional embankments and soil RWs that fully collapsed by seismic loads were reconstructed to geosynthetic-reinforced steepened slopes and GRS-RWs having FHR facing are reported. Finally, several new type bridge abutments taking advantages of the geosynthetic-reinforced soil technology are introduced.

GEOSYNTHETIC-REINFORCED SOIL RETAINING WALL WITH FHR FACING Fig. 2 shows a procedure to construct ‘permanent’ GRS RWs, which is now widely used in Japan (Tatsuoka et al., 1997). This technology has three major features. Firstly, GRS RWs are constructed by the following staged construction procedure: 1) the backfill is constructed to the full wall height with a help of gravel gabions placed at the shoulder of each soil layer; 2) geosynthetic reinforcement layers are arranged with a vertical spacing of 30 cm to ensure a high compaction of the backfill; and 3) after

257

sufficient deformation of the completed geosynthetic-reinforced backfill and the supporting ground has taken place, a lightly steel-reinforced concrete full-height rigid (FHR) facing is constructed by casting-in-place concrete directly on the wrapped-around wall face ensuring a strong connection between the facing and the reinforced backfill. The advantages of the staged construction method are: 1) the damage to the connection between the reinforcement and the facing due to in- and postconstruction deformation of the backfill and supporting ground can be avoided, which makes possible the construction of GRS RWs on compressive subsoil of thick clayey soil on the premise that the stability of wall can be ensured; 2) the backfill immediately back of the wall face can be compacted dense with better mobilization of reinforcement tensile force; and 3) the alignment of completed wall face becomes easy. The second feature is the use of a polymer geogrid with cohesionless soil to ensure a good interlocking with the backfill and a composite of non-woven and woven geotextiles with highwater content cohesive soils to facilitate both drainage and tensile reinforcement of the backfill so that low-quality on-site soil can be used as the backfill, if necessary. The third feature is the use of relatively short reinforcement, which becomes possible by using FHR facing connected to the reinforcement. Drain hole

Gravel gabion Geosynthetic

1) Leveling pad for facing

2) Placing geosynthetic & gravel gabions

3) Backfilling & compaction

4) Second layer

5) Completion of

6) Casting-in-place

RC facing a) wrapped-around wall b) Figure 2. a) Staged construction of GRS RW with FHR facing; and b) a typical wall at Nagoya.

Earth pressure

Earth pressure

Large force in the wall; and large overturning moment & large horizontal load at the bottom of the wall.

Very small force in the facing, resulting to a simple facing structure

Reinforcement

Stress concentration

a)

Needs for a massive or strong wall structure; and a pile foundation

b)

Very small overturning moment & very small lateral force at the bottom of facing, resulting in no need for a pile foundation!

Backfill

Figure 3. Mechanisms of: a) cantilever RW; and b) GRS RW with FHR facing. Roles of full-height rigid facing The facing structure of a conventional type RW is a cantilever structure that resists against the active earth pressure, PA, from the unreinforced backfill activated on the facing by reaction force at the base of the facing (Fig. 3a). Therefore, large internal moment and shear force are activated inside the facing while large overturning moment and sliding force develop at the base of the facing. With reinforced soil RWs, on the other hand, the backfill is retained by the tensile force in the reinforcement. The conventional explanation of the function of the facing, which is actually misleading, is that, because of tensile-reinforcement effects, only very small earth pressure is activated on the back of the facing, accordingly, only a light and flexible facing that can prevent the backfill soil from spilling out is sufficient. However, if the earth pressure activated on the back of the facing is very small, the tensile force at the connection between the reinforcement and the back of the facing becomes very small (Fig.

258

4a) and therefore the soil retaining capability of the reinforcement is largely reduced (Tatsuoka, 1992). That is, the maximum available tensile force, T max, in the reinforcement is obtained by Eq. 1 (Fig. 4c): Tmax= Minimum of [TR, T anchor, Tretain + TWmax]

(1)

where TR= tensile rupture strength of each reinforcement layer; Tanchor= anchorage capacity, which is approximately proportional to the anchorage length, La; Tretain= available retaining strength, which is approximately proportional to the retaining length, L r; and, TWmax= available tensile force at the connection between the reinforcement and the back of the facing, which increases with an increase in the earth pressure activated on the back of the facing and with an increase in the connection strength. When T Wmax decreases to zero, the distribution of the reinforcement tensile force, T, becomes one as those presented in Fig. 4a. In this case, at lower levels in the wall, Tmax cannot become large enough. That is, low connection force results in low tensile force in the reinforcement, which means low lateral confining pressure in the active zone. Then, the active zone becomes more deformable and less stable, particularly when the backfill is a cohesionless soil. On the other hand, large TWmax results in large Tmax (i.e., Fig. 4b). With this distribution pattern, the active zone is effectively confined without exhibiting large strains, and therefore, large bond stresses are not mobilized at the surface of the reinforcement. Consequently, the reinforcement tensile force, T, becomes rather constant in the active zone. When the reinforcement layers are connected to a FHR facing, large earth pressure, which is similar to the active earth pressure in the unreinforced backfill or even higher earth pressure, can be activated on the back of the facing (Fig. 4b), resulting into larger connection force and higher reinforcement force. Then, the confining pressure in the active zone is kept high, which results in higher stiffness and small deformation as well as higher strength and stability of the active zone (i.e., preferred stable conditions). The FHR facing for a GRS-RW behaves like a continuous beam with a large number of supports with a small span (Fig. 4b). Then, even with high earth pressure activated on the back of the facing, only small force is activated inside the facing, resulting in a simple facing structure, and insignificant overturning moment and lateral force activated at the base of the facing, resulting in no need for a pile foundation in usual cases. Active zone

Active zone

Reinforcement Connected

a)

b)

c) Figure 4. Distributions of tensile force in the reinforcement; a) when no facing is used or when the facing and reinforcement are not connected; and b) when a rigid facing and reinforcement are connected; and c) available maximum tensile force in the reinforcement. In summary, the GRS RW system described in Fig. 2 is much more stable while the facing structure and foundation structure are much simpler than the conventional type RWs, which results in lower construction cost, a higher construction speed and the use of lighter construction machines. Despite the above, the wall performance can be equivalent to, or even better than, conventional type RWs. Some case histories The elevated transportation structures in Japan have gradually changed from embankment having a gentle slope towards embankments supported with conventional type RWs (initially masonry RWs, then RC cantilever RWs with a pile foundation) and RC frame structures for higher ones (Fig. 5a). About 35 years ago, the Terre Armee RW technique was introduced to construct railway and highway RWs and had dominated in the permanent reinforced-soil RW market in Japan. More recently, the GRS RW system described in Fig. 2 was introduced and now has become one of the standardized RW

259

technologies for railways in Japan while becoming popular also in other fields. Today, the use of the Terre Armee RWs for railways has declined to nearly zero. Because of a higher cost-effectiveness and a sufficiently high stability and low deformability, the GRS-RW system (Fig. 2) is routinely adopted instead of conventional RC RWs and the Terre Armee RWs. By the end of 2006, GRS RWs (Fig. 2) were constructed at 518 sites with a total wall length more than 80 km (Fig. 6). The stability of GRSRW systems has been validated by excellent post-construction performance of the constructed walls. It should be emphasized that any problematic case, for example with too large wall deformation, has not been reported.

Gentle slope

could be unstable; could be too deformable; and occupies too large space.

(basically with piles)

Higher cost-effectiveness Sufficiently stable and stiff

Some cases

(basically no piles)

a)

b)

Figure 5. a) History of elevated railway and highway structures in Japan; and b) a typical GRS RW in Nagoya, presented in Fig. 2b (Tatsuoka et al., 1997).

Wall length km

80

23

11 53

60

40

20

a)

140

96

82

97

2004 NiigatakenChuetsu Earthquake

Total Annual

0

16

80 km

1995 Kobe Earthquake: a very high-seismic stability of GRS RWs with a FHR facing was proven, which resulted in its more popular use subsequently.

b)

1990

1995

2000

2005

Since 1982: research at the University of Tokyo & the Railway Technical Research Institute

Figure 6. a) Locations of GRS RWs with FHR facing constructed by the staged construction procedure (as of June 2006); and b) annual and total constructed wall length.

a)

b) Figure 7. GRS RW with a FHR facing for railway at Tanata; a) immediately after construction; and b) one week after the earthquake (Tatsuoka et al., 1997).

One of the important case histories with respect to the seismic stability of GRS RW with FHR facing is the one at Tanata site that survived the 1995 Kobe Earthquake (Fig. 7). The damage to the Tanata GRS RW was substantially less serious when compared to many conventional types of soil RWs including those described in Fig. 1. The Tanata wall was completed in February 1992 on the southern

260

slope of an existing embankment for the JR Kobe Line to increase the number of railway tracks from four to five. The total length of the wall is 305 m and the greatest height is 6.2 m. The surface layer in the subsoil supporting the wall consists of relatively stiff terrace soils. The backfill soil is basically a cohesionless soil with a small amount of fines. The reinforcement is a geogrid made of PVA (polyvinyl alcohol) coated with soft PVC (polyvinyl chloride) for protection, having a nearly rectangular cross section of 2 mm by 1 mm and an aperture size of 20 mm with a nominal tensile rupture strength TR = 30.4 kN/m. The wall deformed to a limited extent and moved slightly. The largest outward displacement occurred at the crest of the wall at the location in contact with a RC box culvert structure crossing the railway embankment. The outward lateral displacement was 26 cm and 10 cm at the crest of the wall and at ground level. Despite the above, the performance of the GRS-RW was judged to be very satisfactory because of the following facts. Firstly, the peak ground horizontal acceleration at the site was estimated to be more than 700 gals (i.e., 700 cm/sec2; or 0.7 g, where g is the gravitational acceleration). This is consistent with a high collapse rate of wooden houses at the site (Fig. 7b). Many of the collapsed houses were relatively new, constructed less than approximately ten years before. Secondly, on the opposite side along the railway of the RC box structure, a RC retaining wall with a maximum height of approximately 5.4 m had been constructed at the same time as the GRS-RW. This wall is supported by a row of bored piles despite the similar subsoil conditions as the GRS-RW. Therefore, the construction cost per wall length of the RC retaining wall was approximately double to triple of that of the GRS-RW. Despite these differences, the RC retaining wall displaced similarity to the GRS-RW: i.e. at the interface with the RC box structure, the outward lateral displacement of the retaining wall was 21.5 cm at the top and 10 cm at ground level. Thirdly, the length of geogrid reinforcement used in GRS-RWs with a FHR facing is generally much shorter than that for most metal strip-reinforced soil retaining walls and other types of GRS retaining walls with a deformable facing. For conservatism, most of the GRS-RWs with FHR facing that have been constructed to date have several longer reinforcement layers at higher levels of the wall, as typically shown in Fig. 5b. With the Tanata wall, the length of all reinforcement layers was truncated to approximately the same length due to construction restraints at the site. This arrangement may have reduced the seismic stability of the wall; i.e., the wall would have tilted less if the top geogrid layers had been made longer. Model shaking table tests (1g) to evaluate the seismic stability of RWs on level ground To understand the different performances between the conventional RW types (Fig .1) and the GRS RW with FHR facing at Tanata (Fig. 7) and also to evaluate the effects of different arrangements of reinforcement layers on the stability of GRS RW, a series of model shaking table tests (in 1 g) were performed (Tatsuoka et al., 1998; Koseki et al., 2003, 2006). 140

155

155

SuSurchrge rch arge (1 kPa ) 1kPa

53 53

53 53

Air-dried ModelBackfill Toyoura sand (Dr=(Dr=90%) 90 %) 20

23

20

a.Model Cantilever type(C) C: cantilever

50 50

20

20

53 53

Air-dried Toyoura sand (D ModelBackfill r = 90 %) 18 18

20

23

b. Gravity Model G: type(G) gravity

c.Model Leaning type(L) L: leaning

140

140 Su rch arge1kPa (1 kPa ) Surchrge

20

Air-dried Toyoura sand ModelBackfill (Dr= 90 %)

d. Reinforced-soil type 1(R1) Model R1: GRS-RW type 1

SuSurchrge rch arge 1kPa (1 kPa )

Air-dried Toyoura ModelBackfill sand (D r= 90 %)

140

20

140 140

140

SuSurchrge rcharge (1kPa 1 kPa )

50 50

20

20

140

140 SuSur rchchrge arge 1kPa (1 kPa ) 80 80

35 35

45 45

50 50

Ext ended Reinforcement Extended ModelBackfill reinforcement

e. Reinforced-soil type 2(R2) Model R2: GRS-RW type 2

20

140 Su Sur rcharge kPa ) chr ge( 11kPa

Air-dried ModelBackfill Toyoura sand (Dr= 90 %)

f. Reinforced-soil type 3(R3) Model R3: GRS-RW type 3

Figure 8. Six models of RW on level ground for shaking table tests (all units in cm: Tatsuoka et al., 1998; Koseki et al., 2003, 2006) Fig. 8 shows six different models of RWs without a pile foundation on level ground. The models are 50 – 53 cm high. Models G and L simulate the collapsed prototype gravity type and cantilever RWs described in Fig. 1, while model R1 simulates the Tanata GRS RW (Fig. 7). The performance of model

261

R1 was compared with those of models R2 and R3 to evaluate the effects of reinforcement arrangement. The backfill was air-pluviated dense Toyoura sand (emax=0.977; emin=0.609; Gs= 2.64; and D50=0.23 mm). Each model RW was constructed on a level 20 cm-thick sand layer. The model reinforcement was a grid made of phosphor-bronze strips. A surcharge of 1.0 kPa was applied on the backfill to simulate the railway ballast. The models were subjected to several shaking steps of horizontal excitation with an irregular time history of acceleration shown in Fig. 9a, made from the time history of ground acceleration recorded during the 1995 Kobe Earthquake. The maximum amplitude of the base acceleration amax was initially set 100 gals and increased at an increment of about 100 gals. The time scale was reduced by a factor of 0.3 so that the relationship between the predominant frequency of the original acceleration record and the natural frequency of the prototype structures is as much as possible maintained in the model tests.

Wall top displacement, dtop(mm)

100 -1.0

Base acceleration (g)

-0.8

amax Predoninant frequency: adjusted to 5 Hz

-0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0

Modified from N-S component at Kobe Marine Meteorological Observation Station during the 1995 Kobe EQ.

dtop 80 Subsoil

2

3

4

5

6

50 or 53 cm

20 cm

G

L

R1

R3

C

60

R2

40

20

GRS-RW with a FHR facing 0 0.0

1

Backfill

5 cm

7

8

0.2

0.4

0.6

0.8

1.0

1.2

1.4

Seismic coefficient, k h=amax/g

Seismic force

Bearing capacity, P Seismic Force

Tensile force, T

Tensile force, T (N)

Level

Total normal force, P (kN/m)

Time (sec) a) b) Figure 9. a) Table acceleration; and b) relationships between the residual wall displacement and the seismic coefficient from the model shaking table tests (see Fig. 8 for model names (Tatsuoka et al., 1998; Koseki et al., 1999, 2006).

Leaning

Slope

Level

a)

Slope

Smaller P in the slope

Seismic coefficient Shear band

Level

b)

Smaller T by failure in the slope

Slope Reinforced Seismic coefficient G

Figure 10. a) Total normal force P at the footing base of leaning RW; and b) total tensile reinforcement force T of GRS RW type 2 versus input load, level ground and slope. Fig. 9b shows the relationships between the residual lateral outward displacements 5 cm down

from the crest of the facing observed after the respective shaking stages and the seismic coefficient, kh, defined as amax /g, where amax is the maximum acceleration at the shaking table. Fig. 10 shows the total normal force at the bottom face of the footing of model L (leaning type RW) measured with a set of local load cells and the total tensile force at the connection between the reinforcement layers and the facing of model R2 (GRS RW type 2). The results from the tests on the RWs on slope presented in this figure are explained later. The following trends of behaviour may be seen from Figs. 9 and 10:

262

1. GRS RW model R1 is much more stable than model G (gravity type RW), model L (leaning type RW) and model C (cantilever RW). This trend is consistent with the performances of the prototype RWs during the 1995 Kobe Earthquake (Figs. 1 and 7). 2. The failure of model R1 (i.e., GRS RW with FHR facing) was much more ductile than models G, L and C (i.e., conventional type RWs). This is because, as seen from Fig. 10, the tensile force in the reinforcement of the GRS RW increased with an increase in the input acceleration, while, with model L (conventional leaning type RW without a pile foundation) as well as models G and C, the bearing capacity was suddenly lost, which resulted in brittle failure of these RWs. 3. The GRS RW becomes more stable by increasing the length of reinforcement. The effects of increasing a limited number of layers at high levels of the wall are equivalent to those by equally increasing the length of all the reinforcement layers from 20 cm to 35 cm. The former arrangement using a less total amount of reinforcement is more cost-effective. SOIL RETAINING WALLS ON SLOPE 1999 ChiChi Earthquake (Taiwan) A great number of RWs among those that had been constructed on slope in the mountain areas affected by the earthquake collapsed as typically shown in Fig. 11a (e.g., Huang, 2005). All the collapsed RWs were a kind of gravity type leaning on the backfill (i.e., leaning type). The inferred major cause for the failure is the bearing capacity failure of the slope supporting the RWs (Fig. 11b). In comparison, a number of gravity RWs leaning on the cut slope did not exhibit any noticeable damage (Fig. 11c), which is due seemingly to a high bearing capacity of the supporting ground. Asphalt pavement

Original slope

0.6m

3.6m

about 7.0 m

0.7m

2.5m

5.0m

Backfill Nearly no damage

sandy gravel Estimated failure surface

Excavation

Serious damage

Bearing capacity failure

a) b) c) Figure 11 a) Typical collapsed gravity RW on slope; b) failure of RW on slope; and c) different performances of RWs, 1999 ChiChi Earthquake, Taiwan (Tatsuoka et al., 2007a). Model shaking table tests Another series of shaking table tests were performed on model RWs described in Fig. 12, similar to those on RWs on level ground described in Figs. 8 and 9. The differences are; 1) the supporting ground in front of the respective model RWs was sloped at 1:2 (V:H); and 2) a different model combination was adopted. Namely, model L (leaning RW), model C (cantilever RW) and model R2 (GRS RW with FHR facing) in Fig. 12 are the same as models L, C and R2 in Fig. 8, while model G (gravity) in Fig. 8 was not tested in this series. The short reinforcement layers at the lower levels of model R2 is realistic when constructed on slope: i.e., if long reinforcement layers are to be arranged at the lower levels of the wall, too much excavation of the slope becomes necessary. In this series, model L-N (leaning type RW with nails) and model R2-N (GRS RW with nails) were also prepared to evaluate the effects of reinforcing the backfill with nails of RWs on slope. Four model large-diameter nails, which were 4 cm-in diameter and 40 cm-long mortar columns with a 3 mm-in diameter steel reinforcement at the center, were placed at a horizontal center-to-center spacing of about 10 cm in each layer. Fig. 13 shows the test results. It may be seen that the seismic stability of the three RW types, the leaning and cantilever RWs as well as the GRS RW, constructed on slope is significantly lower than the respective RW types constructed on level ground. With the leaning and cantilever RWs, the deformation of the supporting slope and the normal load measured at the foundation base of the facing

263

showed that the bearing capacity failure took place in the slope below the RWs at lower input acceleration than in the level ground supporting the respective RWs. This trend can be seen also from Fig. 10a. With the GRS RW on slope, the reinforcement force suddenly decreased when the input seismic load reached some level (Fig. 10b). This was due to the development of a shear band starting from below the bottom of the facing extending toward back of the reinforced backfill zone as illustrated in Fig. 10. It may also be seen from Fig. 13 that nailing the backfill can effectively increase the seismic stability of RWs on slope. With the leaning RW, the upper nails anchored the top part of the facing preventing it from overturning failure while the lower nails increased the bearing capacity of the slope supporting the facing. Such nailing as above can effectively increase the seismic stability of leaning RWs already existing on slope. With the GRS RW, the nails prevented the development of a shear band starting from below the facing. When a RW is to be newly constructed on slope, a GRS RW with nails is very cost-effective with very high seismic stability.

L)

C)

R2)

Horizontal accelerometer Horizontal & vertical acc. Disp. Displacement transducer Electric-resistance strain gauge (all units in mm)

R2-N)

L-N)

Figure 12. RW models on slope: L) leaning RW; C) cantilever RW; R2) GRS RW; L-N) leaning RW with nails; and R2-N) GRS RW with nails (Kato et al., 2002).

Wall top displacement. dtop (mm)

200 C: Cantilever (slope)

45 cm

C: Cantilever (level) 150 R2: Reinfoced (slope) 100

R2-N: Reinforced with nails (slope)

R2: Reinforced (level)

L: Leaning (slope) L: Leaning (level 50

L-N: Leaning with nails (slope) 0 0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

Seismic coefficient, kh =amax/g

Figure 13. Relationships between residual wall displacement and seismic coefficient from the model shaking table tests on RWs on slope (see Fig. 12 for model names) (Kato et al., 2002). 2004 Niigataken-Chuetsu Earthquake A great number of embankments, RWs and bridge abutments with backfill for railways, highways and residential areas were seriously damaged during the mainshock with a magnitude of 6.8 and several aftershocks (M= 6.5 and less) of the 2004 Niigata-ken Chuetsu Earthquake. Compared to the damage to RC structures, the scale and extent of the damage to soil structures and its effect on the transportation system and associated civil life were much more extensive. In particular, a number of embankments that had been constructed in narrow valleys and on slopes, where ground and surface

264

water tends to concentrate, were most seriously failed. The effects of heavy rainfall a couple of days before the earthquake had made the backfill saturated more than usual conditions. After remedy work GRS-RW with a full-height rigid facing Slope: 1:0.3 (V:H); height= 13.2 m Vertical spacing of geogrid= 30 cm Joetsu line backfill of sand including round-shaped gravel on sedimentary soft rock (weathered, more at shallow places) After remedy After work failure Failed gravity wall

13.18 m

Before failure:

. 1:2 0

Silt rock

Shinano riiver

Rock blots

Sand rock

a)

b)

Gravel-filled steel wire mesh basket

) (V:H 1:4

c)

d)

Figure 14. a) Slope profiles before and after failure and after reconstruction to a GRS RW; b) failed embankment; and c) & d) the GRS RW under and after construction. Jo-etsu Line, JR East, 2004 Niigata-ken Chuetsu Earthquake (Tatsuoka et al., 2007a). Three totally failed embankments for the Jo-etsu line of the East Japan Railway Company that had been constructed on the slope on the right bank of Shinano River were typical of the above. These were reconstructed to three GRS RWs with FHR facing (Fig. 2), as typically presented in Fig. 14, for the following reasons. Firstly, several GRS RWs of this type performed very satisfactorily during the 1995 Kobe Earthquake (e.g., Fig. 7). On the other hand, a number of conventional type RWs (gravity, leaning, masonry and cantilever RC) were failed and may of them were reconstructed to this type of GRS RWs within a short duration (Tatsuoka et al., 1997, 1998). Secondly, it was considered that this type of GRS RW is superior over other types of soil structures (i.e., conventional sloped embankment supported by a RC cantilever RW with a pile foundation; bridges and so on) in terms of construction cost, construction period and performance in terms of deformation and ultimate stability. In fact, these three railway embankments were reconstructed within two months after the failure. This new case history validated again that the GRS RW (Fig. 2) could be very competitive to construct wall structures for such important infrastructures (i.e., railway and highway).

BRIDGE ABUTMENTS Several problems with the conventional type bridge abutment with backfill Despite a wide use over the world, such conventional type bridge abutment as described in Fig. 15 have several drawbacks as listed below, which are due mostly to that the backfill is not reinforced and the abutment is a cantilever structure on which a girder is placed via movable and fixed supports: 1) As the abutment is a cantilever structure, a pile foundation is usually necessary (Fig. 3a).

265

2) Despite that the RC abutment is not allowed to noticeably displace once constructed, associated with the backfill construction, it is subjected to earth pressure and effects of the settlement and lateral flow in the subsoil via its effects on the piles. To alleviate these negative effects, it may become necessary to increase the number and size of piles. 3) The construction and long-term maintenance of girder-supports is generally costly. 4) The seismic stability of the unreinforced backfill as well as the abutment supporting the girder via a fixed-support is generally low, as observed in many previous major earthquakes. Watanabe et al. (2002) and Aizawa et al. (2007) confirmed this trend by model shaking table tests. 5) A bump may be formed behind the abutment by residual deformation of the backfill due to its selfweight as well as traffic and seismic loads. To develop bridge systems that are more cost-effective than the conventional bridge types while alleviating these problems described above, several solutions have been proposed and attempted as described in Fig. 16. (7) Long-term settlement by self-weight, traffic load & seismic load (6) Girder (3) Backfill

(5) Girder support (fixed or movable); (7) Long-term maintenance (2) RC abutment

Displacement due to earth pressure

(4) Earth pressure

Ground The numbers, (1) – (7) indicate the event sequence.

Ground settlement and lateral flow due to the weight of backfill, and associated negative effects (i.e., negative friction & bending) on the piles.

(1) Piles

Figure 15. Several technical problems with conventional type bridge abutments g

Ordinary backfill without improvement

a1)

p p

Well-graded gravel

a2)

Cementmixed gravel

Improvement of the backfill (already adopted)

Conventional type bridge abutment

Connected Geosynthetic reinforcement

b1)

Connected

b2)

Reinforcing of the backfill (b1: already adopted) Connected Connected

c1)

G. R.

G. R.

c3)

c2)

Cement-mixed gravel

Gravel with PL/PS

G. R.

Connected

Combined measures

Figure 16. Different proposed new bridge abutments (Tatsuoka et al., 2005). Improving the backfill One of the earliest attempts employed by the Japanese railway engineers is to construct a trapezoidal zone of well-compacted well-graded gravelly soil behind the abutment (type a1 in Fig. 16). However, the performance of this abutment type during several previous earthquakes in Japan was unsatisfactory. Watanabe et al. (2002) and Tatsuoka et al. (2005) confirmed the above by model

266

shaking table tests. They also showed that the seismic stability of another similar type constructing a trapezoidal zone of cement-mixed gravel (type a2, Fig. 16) is also low. Reinforcing the backfill A number of bridges comprising of a pair of GRS RWs with FHR facing (described in Fig. 2) that support a girder (type b1 in Fig. 16) were constructed (Tatsuoka et al., 1997). This bridge type (Fig. 17) is herein called the GRS RW bridge. Although this bridge type is structurally simpler and more cost-effective than the conventional type, it has the following limitations. Firstly, the girder cannot be very long due to low stiffness and potential large residual deformation of the backfill supporting the girder. Secondly, the construction and long-term maintenance of movable and fixed girder-supports is costly. This is the common problem with all of the bridge types presented in Fig. 16. Lastly, despite that the seismic stability of GRS RWs with FHR facing is generally very high (e.g., Tatsuoka et al., 1998; Koseki et al., 2003), the seismic stability of the sill beam on which a fixed girder-support is placed is not (Aizawa et al., 2007; Tatsuoka et al., 2007b). This is because the mass of the sill beam is much smaller than the inertia force of the girder while the anchorage capacity of the reinforcement layers connected to its back is small due to their shallow depths. 3

5. Girder

3: Sill beam

4. Movable and fixed girder-supports 1

2: FHR facing

1: GRS RW

- The numbers indicate construction sequence. - Not to scale

Figure 17. GRS RW bridge (type b1 in Fig. 16) Type b2 (Fig. 16), placing a girder on the crest of the FHR facing, is more stable than type b1 (Watanabe et al., 2002; Tatsuoka et al., 2005). However, they also showed that, with type b2, the reinforced backfill behind the facing supporting the girder via a fixed-support would exhibit too large deformation when subjected to level II design seismic load. Combining multiple-measures To substantially decrease long-term residual deformation of the backfill, it is very effective to vertically preload the reinforced backfill and then maintain some vertical prestress that is about a half of the preload in the backfill during long-term service (i.e., the PL & PS technology). The above was validated by laboratory model tests (Shinoda et al., 2003a&b) and long-term performance of a prototype railway bridge pier (Uchimura et al., 2003). Moreover, Uchimura et al. (2003) and Tatsuoka et al. (2005) showed that the seismic stability of PL & PS reinforced bridge pier and abutment is very high. It is in particular the case if high prestress is maintained during dynamic loading and this can be ensured by using a ratchet mechanism as shown by model shaking table tests (Shinoda et al., 2003a&b). Type c3 in Fig. 16 consists of a PL & PS GRS RW with a ratchet system supporting a girder via a fixed-support. Its high seismic stability was validated by laboratory shaking table tests (i.e., Nakarai et al., 2002). Despite the above, any prototype bridge of this type has not been constructed, because possible long-term maintenance works of the ratchet system were not preferred by practicing engineers. Types c1 and c2 were then proposed, which are combining types b2 and b1 with type a2. The first prototype of type c1 was constructed for a new bullet train line in Kyushu (Fig. 18: Tatsuoka, 2004; Tatsuoka et al., 2005). The conventional RC abutment (Fig. 16) supports the backfill with the earth pressure activated on its back. In comparison, with type c1 as well as type b2, the reinforced backfill zone laterally supports the RC parapet (i.e., facing) that is supporting a girder without dynamic earth pressure activated on its back. Type c1 abutments are constructed by the staged procedure presented in

267

Fig. 18b, which is basically the same as the one described in Fig. 2a. A number of similar bridge abutments of type c1 are now at the design stage.

Soil backfill .

b)

3

1

Soil backfill .

2

Cement-mixed gravel

Cement-mixed gravel

c)

a) Figure 18. a) New type bridge abutment using cement-mixed gravel constructed at Takada (type c1): b) staged construction; and c) the completed new type bridge abutment, mid 2003.

Settlement by long-term traffic load and seismic effects

Distance (cm) from the back of the facing Lateral displacement Displacement-controlled loading system

4. Integration

3. Girder 4. Backfill

Increase in the earth pressure due to T

2. RC parapet

Box width: 40

8 polyester reinforcement layers

Hinge support Air-dried Toyoura sand (Dr = about 90%)

1. Pile

a)

Load cell

Settlement of the crest 5 10 20 35

50.5

T: Seasonal cyclic expansion and contraction by thermal effects

- The numbers indicate construction sequence. - Not to scale

9 separated load cells: LCs

8.5

Settlement due to T

129.5 [unit : cm]

b) Figure 19. Conventional integral bridge; and b) model test setup (Tatsuoka et al., 2007b).

Integral bridge A pair of RC parapets (or facing) is integrated with a bridge girder, without using girder supports (Fig. 19a). This type is very popular in the UK and the USA due mainly to low construction and maintenance cost resulting from no use of girder supports. However, the backfill may exhibit large residual settlements by self-weight as well as traffic and seismic loads, while the seismic stability of both girder-parapet system and backfill is relatively low (Aizawa et al., 2007; Hirakawa et al., 2007b; Tatsuoka et al., 2007b). Moreover, as the girder is integrated with the parapets, seasonal thermal expansion and contraction of the girder results into cyclic lateral displacements at the top of the facings, which results in a gradual increase in the earth pressure on the facing and residual settlements in the backfill as shown below (Hirakawa et al., 2006, 2007a). Effects of cyclic displacements of the facing Small-scale model tests (Fig. 19b) were performed in the laboratory to evaluate the problems described above and also to examine whether they can be alleviated by reinforcing the backfill. The backfill was air-dried Toyoura sand produced by air-pluviation for the unreinforced backfill while by hand-tamping for the reinforced backfill. The reinforcement was a Polyester grid (strand diameter= 1 mm; spacing between the adjacent strands= 18 mm; covering ratio= 9.5 %; and rupture tensile strength at an axial strain rate of 1.0 %/min.= 19.6 kN/m). The FHR facing was cyclically displaced about the bottom hinge at a rotational displacement rate of 0.00053 degree/min. Fig. 20a summarizes the peak

268

earth-pressure coefficients in the respective cycles, Kpeak= 2Qpeak/H2 , where Qpeak is the peak total earth pressure per wall width in each cycle; H is the wall height (50.5 cm); and is the dry unit weight of the backfill (16 kN/m3), plotted against the ratio of the double amplitude of cyclic displacement at the facing top to the facing height, (DA)/H, at selected numbers of loading cycle, N. The facing top was allowed to move up to about 0.2 mm ( /H= 0.04 %) toward the active direction associated with an increase in the earth pressure. The solid squares represent the cycles when the active failure plane developed in the backfill. The earth pressure increased with an increase in (DA)/H and N. These test results are consistent with previous laboratory model tests (Ng et al., 1998; England et al., 2000) as well as full-scale behaviour of a prototype RW for three seasons (i.e., N= 3). This earth pressure increase may result in structural damage to the facing and may push out the bottom of the facing. By reinforcing the backfill, this earth pressure increase does not reduce, but the facing is not structurally damaged and not pushed out at the bottom, as the FHR facing behaves like a continuous beam supported by a number of reinforcement layers at a small spacing (Fig. 3b). 2.0

Earth pressure coefficient, K and K peak

Active failure in the model backfill

26

Backfill in the model tests: unreinforced Toyoura sand

1.8

7

1.6

1.2 1.0

N: number of loading cycles

158

Likely behaviour of prototype wall for three seasons: seasonal (DA) vs. estimated Kprototype

1.4

N= 10

2,200

N=1

0.8

10

0.6

N= 2,000

0.4

1 0

3,000 0

0.2 0.0 1E-3

10 1

1

0

0

K peak during

1

0

respective cyclics 0

K0 in the model walls

0.01

0.1

1

10

Lateral displacement ratio, (DA) /H [%]

a) Residual settlement of the backfill at 5 cm back of the facing, Sg/H (%)

-0.5 R & C: D/H= 0.6 % 0.0 Reinforced & Connected (R & C): D/H= 0.2 %

0.5 1.0 1.5 2.0

Reinforced & No Connected (R & NoC): D/H= 0.2 % NR: D/H= 0.6 %

NR (no reinforcement): D/H= 0.2 %

5 cm

2.5 3.0

R & NoC: D/H= 0.6 %

S5

0

50

100

150

200

Number of loading cycles, N (cycles)

b) Figure 20. a) Peak earth pressure coefficients in the model tests and a field full-scale case; and b) residual settlements of the backfill (when =0) by cyclic displacements of the facing and effects of reinforcing the backfill (Hirakawa et al., 2006, 2007a). The other detrimental effect of cyclic displacement of the facing with unreinforced backfill is gradual but eventually large settlements in the backfill associated with the development of an active failure plane in the backfill (case NR in Fig. 20b). In the experiments, the backfill settlement increased with an increase in the cyclic facing displacement, (DA)/H. On the other hand, the backfill settlement became nearly null when the backfill was reinforced with reinforcement layers connected to the back of the facing (case R&C). Even slight heaving of the backfill took place, which was due seemingly to dilatation of the backfill associated with repeated passive movements of the facing. The benefits of reinforcing the backfill with reinforcement layers connected to the facing are as follow. Firstly, for the

269

same thermal thrust from the girder, the displacements of the facing become smaller due to higher stiffness of the reinforced backfill. Secondly, for the same cyclic facing displacement, the residual settlement in the backfill decreases drastically due to higher confining pressure in the backfill and membrane effects of reinforcement layers connected to the facing. It may also be seen from Fig.20b that these positive effects of reinforcing the backfill are largely lost when the reinforcement layers are not connected to the facing (case R & NoC). This is because the deformation of the active zone cannot be effectively restrained by the reinforcement layers.

5. Integration 2

4. Girder 2: GRS RW

3: FHR facing

a)

1. Pile

Bridge type

- The numbers indicate construction sequence. - Not to scale

Cost & period Maintenance of construction cost

C, D

252 gal*

GRS RW

C

589 gal*

Integral

D, E

Conventional (gravity)

GRS Integral

A, B

Seismic stability

Total

641 gal*

1,048 gal*

(* Acceleration at failure in model shaking table test)

b) Figure 21. a) GRS integral bridge; and b) features of four different bridge types. GRS integral bridge Fig. 21a shows a new bridge type (called the GRS integral bridge), which is most cost-effective and most dynamically stable among those described in this paper. This type combines the GRS RW bridge (Fig. 17) and the integral bridge (Fig. 19) taking their advantages: i.e., the backfill and the facing are stabilized by reinforcing the backfill with geosynthetic reinforcement connected to the facing (the GRS RW bridge); and a simple and cost-effective RC bridge structure without using girder-supports (the integral bridge), while alleviating their inherent problems. A GRS integral bridge may also need a pile foundation to support the girder, but a lighter one than the integral bridge may be sufficient, as needs for a pile foundation are usually low with GRS RWs. In particular, as seen from Fig. 20b, the residual settlement of the backfill reinforced with reinforcement layers connected to the facing is very small. Moreover, a high seismic stability with small deformation and displacements can be expected because of integrated performance of the whole bridge system, as shown below. Fig. 21b compares the advantages and disadvantages in the three factors listed in the top line of the four bridge types: i.e., conventional gravity type, GRS RW, integral and GRS integral. The accelerations shown in the second column from the right are those at which the respective bridge models collapsed in the shaking table tests described below. Letters A through E denote negative factors other than seismic stability with the respective bridge types: A= heavy abutment structure because of a cantilever structure; B= need for a pile foundation; C= high cost for construction and long-term maintenance of girder-supports; D= bump due to settlement of backfill by self-weight,

270

traffic load and seismic load; and E= settlement of the backfill and structural damage to the facing by cyclic lateral displacements of facing due to seasonal thermal expansion and contraction of the girder. The full point assigned to each factor is equal to three, which is reduced one by one when these negative factors are relevant. In addition, the seismic stability is classified to three levels with points equal to 1, 2 and 3. So, the total full point is equal to nine when free from all these negative factors A – E with the highest seismic stability, which is assigned only to the GRS integral bridge. Model shaking table tests Shaking table tests of the four bridge types listed in Fig. 21b were performed to validate a highseismic stability of the GRS integral bridge (Aizawa et al., 2007; Hirakawa et al., 2007b; Tatsuoka et al., 2007b). Fig. 22a shows the GRS integral bridge model before shaking. Assuming a length similitude ratio equal to 1/10, the facings were 51 cm-high and the girder was 61 cm-long. By adding a mass of 200 kg at the center of the girder, the equivalent length became 2 m (i.e., 20 m in the assumed prototype). Twenty sinusoidal waves with a frequency of 5 Hz was applied at the table while step by step increasing the maximum acceleration max with an increment of 100 gal. Fig. 22b shows the backfill settlements at 5 cm back of either the sill beam supporting the girder via a fixed support with the GRS RW type (Fig. 17) or the facing with the other three types. Fig. 22c shows the lateral displacements at the top and bottom of the facing. In Fig. 22c, with the GRS RW type, dt is the displacement of the sill beam and in Figs. 22b and c, with the gravity and GRS RW types, the displacements on the side supporting the girder via a fixed-support are presented. It can be readily seen that the GRS integral bridge is much more stable than the other types, while the gravity type is least stable. Moreover, the pushing out of the facing bottom is the major failure mode with the integral and GRS integral bridges.

Lateral displacement at top, dT (mm)

Girder 200 kg (an equivalent length = 2 m)

Box width: 60 cm

61 cm

51 cm

Grid reinforcement L= 35 cm 35 cm

Air-dried Toyoura sand (D r = 90 % )

0

Lateral displacement at bottom, dB (mm)

b)

Settlement of the backfill, S5 (mm)

a)

GRS Integral

GRS RW 20 Integral 40 Conventional (gravity)

S5

Out of measument range

5 cm

60 0

200

400

600

Base acceleration,

800 max

(gal)

1000

60

40

Dislodging of girder Conventional (gravity)

GRS RW (displacement at the sill beam) GRS Integral

20 Integral 0

60

dT

40

dB

Integral GRS Integral

Conventional (gravity)

20

0

1200

GRS RW 0

c)

200 400 600 Base acceleration,

800 1000 (gal)

1200

max

Figure 22. a) GRS integral bridge model; b) backfill settlement; and c) outward lateral displacements of the facing in laboratory shaking table tests.

5 CONCLUSIONS The recent developments of new types of soil retaining structures and bridge with backfill made in Japan aiming at not only a high cost-effectiveness but also a high seismic stability were reviewed. Several case histories of prototype soil structures, including those during earthquakes, and results from

271

model shaking table tests that validate high advantages of geosynthetic-reinforced soil retaining walls with staged-constructed full-height rigid facing are presented. Among a couple of new bridge systems with reinforced backfill, the advantageous features of the GRS integral bridge, which comprises of an integral bridge and backfill geosynthetic-reinforced with geosynthetic reinforcement layers connected to the facing are described. AKNOWLEDGEMENTS The authors express their sincere thanks to their past and present colleagues for their help in performing this long-term research program at the University of Tokyo, Railway Technical Research Institute and the Tokyo University of Science. The financial support from The Ministry of Education, Culture, Sports, Science and Technology, the Japanese Government, and Japan Railway

Construction, Transport and Technology Agency is deeply appreciated. REFERENCES Aizawa, H., Nojiri, M., Hirakawa, D., Nishikiori, H., Tatsuoka, F., Tateyama, M. and Watanabe, K. “Validation of high seismic stability of A new type integral bridge consisting of geosyntheticreinforced soil walls,” Proc. of 5th Int. Sym. on Earth Reinforcement (IS Kyushu 2007), 2007. England, G, L., Neil, C, M. and Bush, D, I. 2000, Integral Bridges, A fundamental approach to the time-temperature loading problem, Thomas Telford. Hirakawa, D., Nojiri, M., Aizawa, H., Tatsuoka, F., Sumiyoshi, T. and Uchimura, T. “Behaviour of geosynthetic-reinforced soil retaining wall subjected to forced cyclic horizontal displacement at wall face,” Proc. 8th Int. Conf. on Geosynthetics, Yokohama, 3, 1075-1078, 2006 Hirakawa, D., Nojiri, M., Aizawa, H., Tatsuoka, F., Sumiyoshi, T. and Uchimura, T. “Residual earth pressure on a retaining wall with sand backfill subjected to forced cyclic lateral displacements, Soil Stress-Strain Behavior: Measurement, Modeling and Analysis, Geotech. Symposium in Roma 2006 (Ling et al., eds.), 2007a. Hirakawa, D., Nojiri, M., Aizawa, H. , Nishikiori, H., Tatsuoka, F., Tateyama, M. and Watanabe, K. “Effects of the tensile resistance of reinforcement embedded in the backfill on the seismic stability of GRS integral bridge,” Proc. of 5th Int. Sym. on Earth Reinforcement (IS Kyushu 2007), 2007b. Huang, C.C. “Seismic displacements of soil retaining walls situated on slope, Journal of Geotechnical and Geoenvironmental Engineering, ASCE, Vol.131, No. 9, pp.1108-1117, 2005 Koseki, J., Tatsuoka, F., Watanabe, K., Tateyama. M., Kojima, K. and Munaf, Y. “Model tests of seismic stability of several types of retaining walls,” Reinforced soil engineering, Advances in Research and Practice, Marcel Dekker, Inc. (Ling et al. eds.), pp. 317-358, 2003. Koseki, J., Bathurst, R.J., Guler, E., Kuwano, J. and Maugeri, M. “Seismic Stability of Reinforced Soil Walls, Proc. of 8th International Conference on Geosynthetics, Yokohama, Vol.1, pp. 51-77, 2006. Nakarai, K., Uchimura, T., Tatsuoka, F., Shinoda, M., Watanabe, K. and Tateyama, M. “Seismic stability of geosynthetic-reinforced soil bridge abutment,” Proc. of 7th Int. Conf. on Geosynthetics, Nice, 1, 249-252, 2002. Ng, C., Springman, S. and Norrish, A. “Soil-structure interaction of spread-base integral bridge abutments,” Soils and Foundations, Vol.38, No.1, 145-162, 1998. Shinoda, M., Uchimura, T. and Tatsuoka, F. “Increasing the stiffness of mechanically reinforced backfill by preloading and prestressing,” Soils and Foundations, Vol.43, No.1, 75-92, 2003a. Shinoda, M., Uchimura, T and Tatsuoka, F. “Improving the dynamic performance of preloaded and prestressed mechanically reinforced backfill by using a ratchet connection,” Soils and Foundations, Vol.43, No.2, 33-54, 2003b. Tatsuoka, F. “Roles of facing rigidity in soil reinforcing”, Keynote Lecture, Proc. Earth Reinforcement Practice, IS-Kyushu ‘92 (Ochiai et al. eds.), II, 831-870, 1992. Tatsuoka, F., Tateyama, M, Uchimura, T. and Koseki, J. “Geosynthetic-Reinforced Soil Retaining Walls as Important Permanent Structures,” Geosynthetic International, Vol.4, No.2, 81-136, 1997.

272

Tatsuoka, F., Koseki, J., Tateyama, M., Munaf, Y. and Horii, N. “Seismic stability against high seismic loads of geosynthetic-reinforced soil retaining structures” Keynote Lecture, Proc. 6th Int. Conf. on Geosynthetics, Atlanta, 1, 103-142, 1998. Tatsuoka, F. “Cement-mixed soil for Trans-Tokyo Bay Highway and railway bridge abutments, Geotechnical Engineering for Transportation Projects,” Proc. of GeoTrans 04, GI, Los Angels, ASCE GSP No. 126 (Yegian & Kavazanjian eds.), pp.18-76, 2004. Tatsuoka, F., Tateyama, M., Aoki, H. and Watanabe, K. “Bridge abutment made of cement-mixed gravel backfill,” Ground Improvement, Case Histories, Elesevier Geo-Engineering Book Series, Vol. 3 (Indradratna & Chu eds.), pp.829-873, 2005. Tatsuoka, F., Tateyama, M., Mohri, Y. and Matsushima, K. “Remedial treatment of soil structures using geosynthetic-reinforcing technology,” Geotextiles and Geomembranes, Vol.25, 2007a. Tatsuoka, F., Hirakawa, D., Nojiri, M. & Aizawa, H., Tateyama, M. and Watanabe, K. “A New Type Integral Bridge Comprising of Geosynthetic-Reinforced Soil Walls,” Proc. of 5th Int. Sym. on Earth Reinforcement (IS Kyushu 2007), 2007b. Uchimura, T., Tateyama, M., Koga, T. and Tatsuoka, F. “Performance of a preloaded-prestressed geogrid-reinforced soil pier for a railway bridge,” Soils and Foundations, Vol.43, No.6, 33-50, 2003. Watanabe, K., Tateyama, M., Yonezawa, T., Aoki, H., Tatsuoka, F. and Koseki, J. “Shaking table tests on a new type bridge abutment with geogrid-reinforced cement treated backfill,” Proc. of 7th Int. Conf. on Geosynthetics, Nice, 1, 119-122, 2002.

273

4th International Conference on Earthquake Geotechnical Engineering June 25-28, 2007 Paper No W1-1002 .

LARGE-SCALE SHAKE TABLE TESTS FOR EARTHQUAKE GEOTECHNICAL ENGINEERING AT NCREE, TAIWAN Tzou-Shin UENG 1, Meei-Ling LIN 1, Wen-Jong CHANG2, Chia-Han CHEN3, and Kuo-Lung WANG4 ABSTRACT This paper describes the experiments and researches performed in the area of geotechnical earthquake engineering using the large 5 m × 5 m shake table at the National Research Center for Earthquake Engineering (NCREE) in Taiwan. A biaxial laminar shear box was designed and manufactured for tests on a soil specimen of 1.880 m (W) × 1.880 m (L) × 1.520 m (H) under one- and multi-directional earthquake shakings. With densely placed instruments, the soil responses, including pore pressure changes, acceleration, strains, and settlements could be observed spatially and temporally. Researches on soil-structure interaction and development of new in situ measurement devices are under way using this laminar shear box. Another model box, 4.4 m (L) × 1.3 m (W) × 1.2 m (H), with fixed but transparent boundaries was used to contain a model slope of a size of 0.5 m in height, 1.3 m in width, and with a slope angle of 30° to study the slope stability under seismic shaking. The responses and failure of the slope were observed and analyzed. Image analyses of the slope test results are to be performed. Applications of these research results and future developments are also discussed. Keywords: Shaking table tests, Earthquake, Sand, Liquefaction, Model slope.

INTRODUCTION In order to study the soil behavior, such as liquefaction and soil-structure interaction, large soil specimens have been placed on a shaking table that can reproduce the actual seismic ground shaking according to the earthquake recording under either 1 g or centrifugal conditions. Thus, the responding phenomena and behavior of a soil stratum as a whole, instead of a small soil element, under the more realistic seismic loading conditions can be observed and analyzed. For a large size specimen, possibly denser sensor placements with respect to the size of soil specimen provide well-distributed measurements, spatially and temporally, for a better understanding of the behaviors of the modeled soil and structures, especially during soil liquefaction and slope failure under shakings. Large scale 1 g shaking table tests were performed on a 5 m × 5 m shake table at the National Center for Research on Earthquake Engineering (NCREE), Taiwan for researches in earthquake geotechnical engineering. Two types of soil model containers were developed and used for different types of experiments. One is a biaxial laminar shear box (1.880 m × 1.880 m in plan and 1.520 m in height) which was designed and manufactured for modeling a soil stratum of about 1.5 m in thickness under one- and multi-directional earthquake shakings on a horizontal plane. The shear box is used for the studies of soil liquefaction and soil-structure interaction during earthquakes. Another one is a fixed 1

Professors, Department of Civil Engineering, National Taiwan University, Taiwan, Email: [email protected] 2 Assistant Professor, Department of Civil Engineering, National Chi Nan University, Puli, Taiwan. 3 Assistant research fellow, National Center for Research on Earthquake Engineering, Taiwan. 4 Postdoctoral research, Department of Civil Engineering, National Taiwan University, Taiwan.

274

boundary box with a size of 4.4 m (L) × 1.3 m (W) × 1.2 m (H) to be used for tests on models of slopes and quay walls. The performances of these two boxes were tested and verified, and they were found satisfactory (Ueng, et al., 2006, Lin and Wang, 2006). Experiments which had been performed using these boxes on the shake table are described and the results of the tests and their analyses are presented briefly. Applications of the results, future research activities, and possible collaborations with other testing facilities are also discussed in this paper.

BIAXIAL LAMINAR SHEAR BOX Figure 1 is the biaxial laminar shear box on the shake table at NCREE. It is composed of 15 layers of sliding frames. Each layer consists of two nested frames, an inner frame (1880 mm × 1880 mm) and an outer frame (1940 mm × 2340 mm). Both frames are made of a special aluminum alloy with 30 mm in thickness and 80 mm in height, except the uppermost layer that has a height of 100 mm. These 15 layers of frames are separately supported on the surrounding rigid steel walls with a gap of 20 mm between adjacent layers. The 20-mm gap is provided to avoid rupture of the rubber membrane inside the box during a large relative deformation between layers. Thus, a sand specimen of 1880 mm × 1880 mm × 1520 mm can be placed inside the inner frames. A 2-mm thick silicone rubber membrane was placed inside the box to obtain a watertight container.

Figure 1. The biaxial laminar shear box on the shake table at NCREE

Linear guideways consisting of sliding rails and bearing blocks are used to allow an almost frictionless horizontal movement without vertical motions. Each outer frame is supported by the sliding rails built on two opposite sides of the outer rigid walls. The bearing blocks on the outer frame allow its movement in the X direction. Similarly, sliding rails are also provided for each outer frame to support the inner frame of the same layer such that the inner frame can move in the Y direction with respect to the outer frame. With these 15 nested layers of inner and outer frames supported independently on the rigid walls, the soil at each depth can move multidirectionally in the horizontal plane without torsion. Figure 2 shows the schematic drawings of the biaxial laminar shear box.

275

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

sliding rails

inner frames soil specimen

rigid base

Z

outer frames

Y

shake table

X

rigid walls

X (b) side view

(a) plan view

Figure 2. Schematic drawings of the biaxial laminar shear box

LIQUEFACTION TESTS Tests on clean sand Series of shaking table tests on clean sand in the shear box have been conducted at NCREE since August 2002. A fine silica sand from Vietnam was used to prepare the sand specimen for the shaking table tests at NCREE. The maximum void ratio, emax, and the minimum void ratio, emin, range from 0.887 to 0.912 and 0.569 to 0.610, respectively, for different batches of sand used in the tests. The permeability of the sand ranges from 0.04 cm/s to 0.1 cm/s for relative densities from 35% to 90%, respectively. One- and multi-directional shakings including sinusoidal waves (with frequencies from 1 to 8 Hz and amplitudes, A max, from 0.03 to 0.15 g) and recorded accelerations, full and reduced amplitudes, from various earthquakes were applied. Miniature piezometers and accelerometers were installed within the sand specimen for pore water pressure and acceleration measurements at different locations and depths in the soil during shaking. Transducers for displacements and accelerations were also placed on different layers of the frames to record the movements of the specimen. Figure 3 are the layout of instrumentation inside the sand specimen and on the inner frames. Large amount of data were obtained from the densely placed sensors inside and outside the sand specimen. Sand Surface

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

20 30

20

40 30 30 30

30

60 30 60 50

50

30

30

(a) within soil specimen Accelerometer

Piezometer

(b) on inner frame Settlement Plate

Coupled Sensor

LDT

Figure 3. Layout of instrumentation within the specimen and on the frames (unit = cm) The test results showed that a multidirectional shaking induced a higher excess pore water pressure and it took a longer time to dissipate in a multidirectional shaking test than under a one-directional shaking. Figure 4 shows the excess pore water pressure distribution along the depth of the specimen at various time during 1-D and 2-D shaking tests. It indicates that one-directional shaking induced less

276

excess pore pressure and probably caused only a shallower liquefied layer, while the multidirectional shaking caused a deeper liquefaction zone. A clear liquefaction depth can be seen in the large specimen from the rather dense piezometer measurements. The depth of the liquefied sand can be better determined to a degree of accuracy based on the measurements of mini-piezometers and accelerometers on the inner frames as describe in Ueng et al. 2007b. Exce ss p ore pressure head (mm) 0

200

400

600

Excess pore pressure head (mm)

800

1000

0

0

200

400

600

800

1000

0

Time 200

Tim e 1s 3s 5s 7s 10 s 13 s 16 s

400

600

1s 3s 5s 7s 10 s 13 s 16 s

200

400

600

800

800

1000

1000

1200

1200

(a) 1D shaking, A max = 0.075g

(b) 2D shaking, A max = 0.075g

Figure 4. Excess pressure heads inside the specimen during shaking test, October 2004 According to the settlement measurements during and after the shaking tests, the settlements during ose when there is liquefaction of the soil. The settlements resulted from the multidirectional shaking were larger than those under one-directional shaking in both cases of liquefaction and non-liquefaction of the soil. The volumetric strain of the sand after liquefaction caused by shaking was calculated considering the depth of liquefaction depth. With consideration of thickness of the liquefied sand, the test results show that the volumetric strain after liquefaction, under sinusoidal shakings decreases with the relative density of the sand regardless of the amplitude, frequency and directions of shaking. Figure 5 shows volumetric strains after liquefaction under sinusoidal shakings with durations of 5, 10, 20 and 30 seconds in this study (Ueng et al., 2007b). It can be seen that the volumetric strain after liquefaction increases with the shaking duration. The results show a good comparison with those given by Tokimatsu & Seed (1987). 10 9

Shaking duration 5s 10 s 20 s 30 s D = 10 line D=5s D = 20 s D = 30 s

8 7 6 Duration = 10 s

5 Duratio n = 20 s

4 3

Duration = 30 s

2 1

Duration = 5 s

0 0

20

40

60

80

100

Dr (%)

Figure 5. Volumetric strain after liquefaction versus relative density of sand under various shaking durations

277

Preparation and testing of Mailiao silty sand Mailiao silty sand, a typical soil of the reclaimed lands in the western coastal area in Taiwan, was used in this study. The particle shape of the Mailiao silty sand is mainly sub-angular and flaky. The sand is more compressible and less dilatable than silica sand. Sample preparation method of the Mailiao silty sand specimen in a large scale laminar shear box was evaluated. The fines content and the large size of the sand specimen are the main concerns for obtaining a reasonably homogeneous and representative sample of the in situ soil. A series of experiments in a smaller glass container were conducted using various preparation methods, including wet sedimentation method using dry and moist sand, staged sedimentation method and pre-mixture filled method. This study used the staged sedimentation method that rather uniformly distributed fines contents at different depths except a top thin layer of sand with a higher fines content. The moist Mailiao silty sand with a mass of 1.0 Mg and an average water content of 7.2% were dumped down evenly through a specially designed pluviator into the laminar box filled with water to a pre-calculated depth at first. After an interval of about 30 minutes for the primary sedimentation, the laminar box was filled with water slowly to the calculated depth based on the measurements of water level and elevation of the sand specimen after the first step. Then another stage of soil placement proceeded in the same manner. It took seven stages to complete a Mailiao silty sand specimen of about 1.350 m in height. The uniformity, density and saturation of the sand were checked by the thin-walled tube sampling and P-wave velocity measurements. The results showed that both uniformity and saturation of the specimen were satisfactory by this preparation method (Ueng et al., 2007a). Two series of one- and multi-directional shaking table tests were performed on the saturated Mailiao silty sand in the laminar shear box. The preliminary results showed that the relatively low permeability layer of the Mailiao silty sand resulted in a longer duration of dissipation of generated pore water pressure and surface settlement compared with those of the clean Vietnam silica sand. The sand boils were observed when there was liquefaction of the Mailiao silty sand. Further analyses of the test results and more shaking table tests are under way.

SOIL-PILE INTERACTION For the study of the soil-structure interaction in a liquefiable soil during one- and two-dimensional shaking, physical pile model tests in the large biaxial laminar shear box is under way at NCREE. The model pile is made of stainless steel with an outer diameter of 100 mm and the thickness of 3 mm. The pile tip was fixed at the laminar shear box to simulate the condition of the pile foundation embedded in the rock. Strain gauges and accelerometers will also be installed on pile surface to obtain data relating to soil-pile interaction. The soil responses, including pore pressure changes, acceleration, strains, and settlements will also be measured. The first stage of tests is the lateral loading tests from reaction walls. The lateral loads will be applied on the pile head, monotonically and cyclically, with and without soil in the shear box. The same physical models will also be placed on the shaking table under one- and multi-directional shakings to study the soil-pile interaction under 1D and 2D shakings. A mass or a multi-degree-of-freedom structure will be added on top of a single pile or a pile group to evaluate the possible inertial effect on pile performances. An inclined shear box will be set up to simulate the possible lateral spreading effect on the pile under 2-D shaking in order to clarify mechanisms of soil-pile interaction in liquefied and laterally spreading ground.

DEVELOPMENT OF COUPLED EMBEDDED SENSOR The spatial variations of soil motions raise concerns of the shear strains evaluated of the ground under seismic loading. In addition, pore pressures and shear strains measured at different locations can not precisely capture the coupled responses between the induced shear strain and the generated pore pressure due to time lag and/or non-uniformity of the soil. Therefore, development of an instrument

278

that can capture the coupled shear strain-pore pressure responses locally will be beneficial to understanding of the process of pore water pressure generation and its relation to the shear strain induced by earthquake loading. To capture the coupled shear strain-pore pressure responses of saturated soils subjected to horizontal shaking, a coupled sensor, which integrates a triaxial accelerometer and a miniature pore pressure transducer in an acrylic case, is fabricated as shown in Figure 6 (Chang et al., 2007). The small size (12 cm long, 5.5 cm in diameter) of the coupled sensor enables it to simultaneously measure both particle accelerations and water pressure changes at the same location. Four coupled sensors are embedded in the soil specimen in the laminar box to form a 0.6m by 0.6m square array on the vertical plane parallel to the wave propagation direction, as shown in Figure 3(a). This array of sensors is used to evaluate the shear strain within the array. The densely embedded single-function sensors and the external instrumentation on the frames of the shear box at NCREE provide a testing platform to verify the performance of the developed coupled sensors.

Figure 6. Coupled sensor Soil accelerations and frame movements were processed to calculate shear strains. The 2D displacement-based method (2DBM) uses the isoparametric formulation in the finite element framework to calculate strains at any point within the instrument array (Rathje et al., 2005). The nodal displacement components are computed by numerically integrating the vertical and horizontal accelerations obtained by the coupled sensors. In a 1D shear wave propagation condition as the test condition in this study, a soil deposit can be modelled as a stack of shear beams and the average shear strain between two depths ( 12) can be evaluated by: 12

u y1 u y 2 z2 z1

(1)

where uyi represents the horizontal displacement of point i at depth of zi. It was found that the accelerations measured by two coupled sensors at the same depth are essentially the same. That is, the 1D shear wave condition did occur as intended in the laminar shear box. The horizontal displacements are computed by double integrations of accelerations measured by embedded accelerometers and the strains within the sensor array can be obtained. The results were compared with the shear strains calculated using the direct measurements of frame displacements by LDTs. The comparisons revealed that two methods agree well for cases with shear strain level less than 2 × 10-2 %, when only small excess pore pressure was generated. Two methods gave different strains when there is liquefaction. Pore pressure data were processed to highlight characteristics of pore water pressure changes during and after the shaking. The shear strains and pore pressure changes at the same location were

279

Excess Pore pressure ratio, ru

compared to observe their coupling behavior in the liquefiable sand. Time histories of excess pore pressure ratio (ru = u/ v', where u = excess pore pressure and v' = effective vertical stress) at one of the sensors and the shear strain computed according to the measurements by the coupled sensors in a liquefied case are shown in Figure 7.

Stage 1

Stage 2

Stage 3

Stage 4

Shear strain (%)

(sec)

Note: Calculated by 2DBM

Time (sec)

Figure 7. Coupled response of a liquefied case measured by a coupled sensor The developed embedded coupled sensors are being used in the large field soil-structure interaction tests in the previously liquefied area at Taichung harbor. A pilot test that used the embedded sensors to measure the free-field responses due to surface shaking had been conducted in 2006. The coupled sensors were deployed in a 1D fashion and an eccentric vibrator with maximum output of 17 tons at a frequency of 7 Hz is used to generate downward propagating shear waves. Although only about 20% of excess pore pressure was generated at a depth of 6 m, the measured coupled behaviors during shaking and subsequent dissipations are observed. Further tests in the future will include the pile-soil interaction measurements in the liquefied reconstituted soils and field dynamic responses of wharf due to surface dynamic sources.

SEISMIC SLOPE STABILITY Figure 8 illustrates the fixed boundary box for the model slope on the shaking table at NCREE. The inner dimension of the box is 120cm (H) x 130cm (W) x 440cm (L) and the weight of the box is 39,800 kN. The box has fixed boundaries with reinforced glass panels mounted in the longitudinal axis. Sensors were installed at the four corners of the box to observe the performance of the box including acceleration, velocity and displacement.

280

Figure 8. Fixed boundary box for the model slope The specimen was prepared by two methods: compaction and pluviation. For the compaction methods, the soil was prepared by mixing with water to 8 % water content, and then cured for 24 hours. The soil specimen was compacted as shown in Figure 9(a) into the model box using the controlled-volume method to obtain a unit weight of 16.6 kN/m3. The slope surface was compacted by a modeling tool to keep the slope angle at the designated value. For the pluviation method, the soil was pluviated through the pluviation box (Figure 9(b)), which contains several layers of sieves to assure that the granular particles drop vertically and evenly into the model box.

(a) Compaction method

(b) Dry pluviation method

Figure 9. Construction of the slope specimen Significant nonlinear responses and amplifications were observed with acceleration amplitude up to 0.6g in the specimen prepared with compaction method. The observed slope failure surface appeared to be quite shallow and in the upper part of the slope. Although the specimen was carefully excavated after the test, the shear surface inside the specimen could not be well-defined. For the specimens prepared by the pluviation method, the amplification effect was not obvious and the slip surface is shallower than the specimen prepared by the compaction method. However, the measured accelerations inside the specimens showed that slip surface could reach deeper than that observed through the transparent glass. The testing processes were recorded by digital video (DV) and charged-couple device (CCD) during experiment. In order to study the displacement behavior of slope under seismic force, the Particle Image Velocimetry (PIV) analysis is adopted in this research. PIV is based on the processed images recorded during the experiment. The general concept of PIV is to find the displacement vector of

281

particle from one image to another. The displacement vector can be transferred to velocity when the duration between two images is known. The computer program used in this research is PIVview 2C. The purpose of this analysis is to understand the displacement behavior and the landslip initiation of slope during seismic loading. The recorded images by digital video (DV) and charged-couple device (CCD) are 740 480 pixels and 640 480 pixels, respectively. The images recorded of specimen prepared by compaction method are 740 by 480 pixels. The PIV result for 32-33 seconds of a loading sequence is illustrated in Figure 10. The down-slope direction is to the right hand side and the distance between dots is 10 cm. The slope starts moving from the slope surface on the bottom of the image.

Figure 10. PIV results of specimen from 32 to 33 seconds (compaction method)

CONCLUDING REMARKS Various types of large scale shaking table test for earthquake geotechnical engineering were conducted using the biaxial laminar shear box and the fixed boundary box at NCREE. The large-scale tests with larger soil specimens simulate closely the in situ situations under earthquake loading. With proper instrumentation, these tests give important information for the understanding of the phenomena occurring in the field. Appropriate analyses can then be performed according to the findings from the large-scale tests. Future studies at NCREE will include: Studies of the liquefaction characteristics of soils with high fines contents, especially the local soils in Taiwan; Effect of multidirectional shakings on pile foundations, including bridge foundations; Development of sensors including optical devices to be used in the field utilizing the shear box on the shaking table to simulate the ground conditions for the purposes of verification and calibration; Further development of image analysis techniques, hardware and software, for large scale tests. Collaborations with other institutes can be achieved by: Conducting specially designed experiments at NCREE for comparisons with the results obtained by the centrifuge tests and full scale tests. It is hoped that the scale effects in these

282

tests can be better understood via these comparisons so that the testing and interpretation of physical model tests can be improved; Comparing the results of experiments using the biaxial laminar shear box at NCREE with those from large one-dimensional shaking tests at other institutes to study the effect of multidirectional shaking on geotechnical structures.

ACKNOWLEDGEMENTS These researches are partly supported by the National Science Council, Taiwan and NCREE. The technical supports and operational assistances in the shaking table testing including large specimen preparation by the engineers at NCREE are gratefully acknowledged.

REFERENCES Chang WJ, Ueng TS, Chen CH and Yang CW. Embedded instrumentation for coupled shear strainpore pressure response in multidirectional shaking table test, Proc. 4th International Conference on Earthquake Geotechnical Engineering, Paper 1213, 2007. Iai S. Similutude for shaking table tests on soil-structure-fluid model in 1-g gravitational field , Soils and Foundations, 29(1), 105-118, 1989. Lin ML and Wang KL. Seismic slope behavior in a large-scale shaking table model test, Engineering Geology, 86, 118-133, 2006. Rathje EM, Chang WJ, and Stokoe KHII. in situ dynamic liquefaction test ASTM Geotechnical Testing Journal, 28(1), 65-76, 2005. Ueng TS, Chen CH and Tsou CF. Preparation of a large Mailiao silty sand specimen for shaking table test, Proc. 4 th International Conference on Earthquake Geotechnical Engineering, Paper 1339, 2007a. Ueng TS, Wang MH, Chen MH, Chen CH and Peng LH. A large biaxial shear box for shaking table tests on saturated sand, Geotechnical Testing Journal, ASTM International, 29(1), 1-8, 2006. Ueng TS, Wu CW, Cheng HW and Chen CH. Settlements of Saturated Clean Sand Deposits in Shaking Table Tests, in progress, 2007b.

283

4th International Conference on Earthquake Geotechnical Engineering

WORKSHOP 2 Geotechnical earthquake engineering related to monuments and historical centres

4th International Conference on Earthquake Geotechnical Engineering June 25-28, 2007 Paper No. W2-1012

SEISMIC RESPONSE ANALYSIS OF ANCIENT COLUMNS Nikolaos ARGYRIOU 1 , Olga-Joan KTENIDOU2, Maria MANAKOU3 , Pashalis APOSTOLIDIS4, Francisco J. CHAVEZ GARCIA5 and Kyriazis PITILAKIS6 ABSTRACT

This paper presents a numerical and an experimental study of the seismic response of ancient columns. The multi-drum column analysed here corresponds to the Hellenistic portico of Lindos acropolis. This structure was modelled using a finite element model. The simulations were made in three dimensions. Extensive time domain parametric analyses were performed in order to examine the response of the column subjected to seismic motion having different values of peak ground acceleration and frequency content. The seismic input used consisted of horizontal components of three real earthquakes with different frequency content. Three different systems connecting the drums of the column are analysed in order to examine their influence on the seismic response of the column. The analysis takes into account the complex behaviour of the structure with the aim to determine the threshold PGA value before its collapse. In a second stage we discuss available procedures to estimate the fundamental period of a monolithic, a multi-drum, and two multidrum columns connected with an architrave using microtremors measurements. The results of the microtremor measurements are compared with the numerical simulations to assess the effectiveness of the procedure. It is shown that microtremor measurements may be a useful tool to estimate the modal shapes of ancient columns. Keywords: ancient columns, numerical analysis, seismic response, microtremors, HVSR, SSR

INTRODUCTION Ancient Roman and Greek temples used to be systems made of curved stones (drums) laid one on top of the other without any connecting mortar. Mostly, gravity loads were carried to the ground by means of monolithic and multi-block columns and colonnades, connected at the top with architraves. Recently, these systems have received increasing scientific attention. The reason is their interesting static and dynamic behaviour which derived from the ‘ductility’ of the whole system. In addition, shear transfer is achieved through friction and through wooden or metal pins –shear keys- (called ‘polos’) between the drums. In fact, this element should have a finite shear strength that does not exceed the shear strength of the interface of the drums; it is a sort of early version of the capacity 1

Civil Engineer MSc, Department of Civil Engineering, University of Thessaloniki, Greece, Email: [email protected] 2 PhD Candidate, Department of Civil Engineering, University of Thessaloniki, Greece, Email: [email protected] 3 Dr Geologist, Department of Civil Engineering, University of Thessaloniki, Greece, Email: [email protected] 4 Dr Geologist, Department of Civil Engineering, University of Thessaloniki, Greece, Email: [email protected] 5 Professor, Instituto de Ingeniería, Universidad Nacional Autónoma de México, México D.F. 04510 México, Email: [email protected] 6 Professor, Department of Civil Engineering, University of Thessaloniki, Greece, Email: [email protected]

284

design process currently used for the seismic protection of structures. The effectiveness of this process is demonstrated by the fact that a large number of these monuments have survived historic earthquakes which must have had undergone large peak ground accelerations. The seismic response of these systems is a multi-parametric problem involving both material and geometrical non-linearities. The structure response may go through several modes during the same excitation, thus no modes exist ‘with the classical meaning that is used in continuum mechanics medium’ (Papantonopoulos et al., 2002). As a matter of fact, the modes of such systems depend on the nature of contact existing between the interfaces of adjacent drums (after Dr. Kostas Zambas, personal communication). As a result, the response is highly non-linear including relative sliding of the drums and rocking. This complex response is caused by the ‘polos’ behaviour and strength, the non-uniform friction and contact surface and the dispersion of axial load developed during excitation at the contact surface between the drums. Over the past two decades many researchers have approached the phenomenon by studying the behaviour of a rigid 2D block resting on a horizontal plane either experimentally, analytically and numerically (Zambas, 1994; Shenton, 1996; Sinopoli, 1989; Yim et al., 1980, Zhang and Makris, 2001). In addition, complex numerical models ‘simulating’ real cases were examined (Mouzakis et al., 2002; Papantonopoulos et al., 2002; Manos and Demosthenous, 1997, Psycharis et al. 2000; Psycharis et al., 2003; Makris and Konstantinidis, 2005). The main conclusions of these researches can be summarized as follows: The response of the system is very sensitive to very small changes in the parameters of the system or the parameters of the excitation. The results of a 2D study of the problem can be more conservative than those of a 3D study Pure sliding, pure rocking, as well as a slide-rock mode may be observed during the response of these systems during dynamic excitation. An in-plane excitation can produce large displacements in the out-of-plane direction The predominant period of the ground motion and the size of the structure are the main parameters for stability, with low-frequency pulses being more dangerous as they increase the possibility of collapse. Moreover smaller columns, while having the same aspect ratio as larger columns, can be more unstable. Free standing columns are more susceptible to earthquakes than colonnades connected with architraves Another interesting feature of the detailed study of the damage patterns and displacements of ancient columns is that these analyses may improve the existing knowledge on the seismic history of an area (Bottari, 2003) and through back analysis may allow estimation of the monuments’ current seismic resistance level. In this paper a comprehensive dynamic analysis of a multidrum column is presented using the finite element method. As previously mentioned, it is extremely difficult to simulate the exact geometry of the interfaces between the drums; thus, any attempt to study such a case with numerical or even experimental approaches, remains a rough approximation. But still, these methods are always useful in order to estimate the response even in a qualitative way. We examined the differences in the response of the column to earthquakes having different frequency content and different PGA values. We also investigated the differences in the response of the column due to the presence or not of several ‘polosempolio’ systems. In addition, we made an effort to determine the threshold PGA value that would cause the column to topple. Finally, microtremor measurements were conducted in three different cases: a multidrum column, a monolithic column and a system of three multidrum columns connected with an architrave. We propose a procedure for estimating the modal periods of the columns by comparing numerical results with the results from the analysis of the microtremor measurements. The aim of this paper is to contribute to our understanding of the seismic response of ancient columns, which is an exciting topic of mechanics and dynamics.

285

NUMERICAL STUDY Selected case study The multidrum column of the Hellenistic Portico at Lindos Acropolis, Rhodes was selected for the analyses. The column is 5.0 m high and is composed of seven drums and a capital. The diameters of the drums range from 0.78m at the base to 0.62m at the top and each drum is around 0.68m high. In ancient times, drums were connected with wooden or metal pins, a system called ‘polos-empolio’. This system was used as a guide in order to achieve perfect contact between the drums. Recent researches (Manos, 2006; Psycharis et al., 2003) have highlighted the role of these pins as shear keys, able to dissipate seismic energy through fracture. The connection between drums is also important for modern restoration works. In the present study two different ‘polos’ systems were selected for the analyses assuming different materials, either wood or titanium. The latter is used in current restoration projects. In addition, the case without any ‘polos’ system was also studied. Analysis overview Most of the numerical studies examining the seismic behaviour of ancient columns have used the discrete element method. The reasons are the small computing cost and the possibility to efficiently simulate large displacements. Under this approximation, the drums of each column are assumed to behave as rigid bodies (Psycharis, 2003). In this paper, all the analyses were conducted using the finite element program Ansys (2004) noting that advances in computing software and hardware have enormously decreased the computing costs associated to non-linear analyses. Contact elements were used to simulate the interfaces between the drums. Static and dynamic coefficients of friction and cohesion can be used while the Mohr Coulomb criterion was adopted for the analyses. Contact elements can also simulate the rocking response of the drums, making it possible to describe accurately all possible motions of these blocks (Argyriou et al., 2006). In addition, friction stress can be evaluated at each step of the computation. Table 1 shows the parameters used to simulate the interfaces between drums and the mechanical parameters of the drums themselves for the multidrum column of Lindos Acropolis (Eleftheriou, 2002). For the dynamic coefficient of friction, a generally accepted value equal to 0.7 was considered. Table 1. Mechanical properties of the interface and the stone Cohesion

Dynamic coefficient of friction

Modulus of elasticity E for stone (MPa)

Density (ton/m3)

0

0.7

1044

2

Shear resistance for wood (KN/cm2) 0.7

Shear resistance for titanium (KN/cm2) 31

The time step for the numerical integration was set to 10 -3s. Larger values led to numerical instabilities (extremely large displacements or overlap between bodies). It was verified that the spatial discretization step affected only the computing cost and not the accuracy of the results. The numerical analysis performed in two stages. In the first one, the model was subjected to the acceleration of gravity. This allowed for modelling the weight of the system, imposing the friction forces at each contact element. In the second stage, the model was subjected to the input seismic signal, incident from below, and the seismic response of the system was computed. The influence of the ‘polos’ system was included indirectly in the models by the artificial increase of the static coefficient of friction s to a value equal to the shear resistance of the poles q (1).

's

s

w( z) q w( z)

(1)

w(z) is the weight of the column (in KN) (according to the analysed interface) and q (in KN) is the shear strength of the pole. A diameter of 4.0cm was considered for hard pine wood and of 1.0cm for titanium. This simplification reduces the cost of the computation and allows a qualitative evaluation of

286

the effect of each system (Makris and Konstantinidis, 2005). Clearly, the physical phenomenon is more complex (even fracture may occur) due to inelastic behaviour under recycling loading and the possible appearance of gaps between the polos and its stone enclosure. The influence of the ‘polos system’ case to the shear resistance of each interface is shown in Figure 1 through the ratios of static to dynamic coefficient of friction. The damping was set equal to zero based partially to experimental results (Mouzakis et al. 2002) and partially to the observed oscillation of the drum column. Without polos 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0

Wooden polos 8.72 5.81 4.44 3.60 3.08 2.70 2.41 2.20

Titanium polos 22.63 14.50 10.65 8.28 6.83 5.76 4.95 4.37

Figure 1. Ratios of static to dynamic coefficient of friction for each interface Selection of the seismic input The seismic input applied at the base of the model consisted of the horizontal component records of selected earthquakes. Three earthquakes were chosen: Aigion: 15/6/1995 00:15:49, Ms=6.2, Mw=6.5, Ml=5.8. Station: Telecommunication building, Re=15km Kalamata: 13/9/1986 17:24:35, Ms=5.8, Mw=5.9, Ml=5.5. Station: Old Telecommunication building, Re=10km Erzincan: 13/3/1992, Mw 6.9, Ms=6.8. Station: 95 Erzincan, Rrup=2km, Re=1km The criterion for this choice was their different frequency content, as they will give information about the response of the column in different period ranges. Figure 2 shows one of the horizontal components of acceleration for each record. The records were retrieved from PEER and ESMD online databases. The accelerogram selected from the Aigion, Greece 1995, was recorded 15km from the epicentre. Its PGA is approximately 0.5g, it was recorded on medium rigid soils and it is dominated by a 0.5sec period pulse. The second record was obtained during the Kalamata, Greece, 1986 earthquake, was recorded on medium to loose ground at a distance of 10km and has PGA of 0.27g. Its largest amplitudes occur at three periods: 0.4, 0.7 and 1.3 sec. This earthquake caused considerable damage to the buildings of Kalamata city. The final record was obtained during the Erzincan, Turkey, 1992, earthquake. The record, obtained at 1km from the epicentre, has a PGA of 0.42g and its largest amplitudes occur at periods at 0.3, 1 and 2 sec. In order to estimate the threshold PGA value to collapse, a set of parametric analyses was carried out. The column was subjected to increasing levels of horizontal acceleration. 0,5

0,3

0,3

0,2 0,1

0,1 -0,1 0

0 5

10

15

20

25

30 time (sec)

-0,3

-0,1

0

5

10

15

2 time (sec)

-0,2 -0,3

-0,5

(a)

(b)

287

0,6 0,4 0,2 0 -0,2 -0,4

0

5

10

15

20

25 time (sec)

-0,6

(c) Figure 2. One of the two horizontal components of the earthquakes used as input: (a) Aigion 1995, (b) Kalamata 1986, (c) Erzincan 1992 Analysis results Because the response of a multidrum column can vary between several modes during excitation, a reliable index of damage should be introduced in order to describe the final mode of the column. The maximum permanent displacement relative to the base, as calculated at the end of each analysis, was chosen as such an index. Figure 3 summarizes final displacements estimated for the three earthquake records. For cases where the column did not collapse (stable modes), the drums where maximum displacement occurred are shown. For cases where it collapsed (collapse mode), the drums from which collapse begins are shown. In order to ‘classify’ the possible response of a column for increasing PGA values and thus define ‘levels of damage’, the column is subdivided along its height into two parts. The first part contains all drums from the 6th up to the capital (green circles), while the second part contains all drums from the 3rd up to 6th (red circles). It is noted that this response refers either to the permanent displacement of a drum (for stable mode cases) or to the drum that collapsed (collapse mode cases). As can be seen for the Erzincan case (figure 3a), the column is stable for PGA values up to 0.2g, with cases of failure beginning to occur for higher levels of excitation. The same is observed for all ‘polos system’ cases. The wooden ‘polos’ system seems to have minor influence on the response of the column, as the total displacements are not affected by it. The titanium ‘polos’ system affects results by reducing permanent displacements in most PGA levels. Moreover, for small values of PGA (scale factor f=0.20-0.35) displacements are only noticed in the upper drums, namely the capital and 7th drum, while the lower drums are excited only for higher values of horizontal acceleration (f=0.400.80). In general it can be said that this second level of damage (red circles) is introduced when PGA values are increased. In Figure 4, the final response modes of the column are depicted for the three system cases when excited by the Erzincan record. It is shown that for low PGA values all drums enter the rocking mode, while for larger PGA values rocking occurs only in one drum, whereas the upper drums behave as a monolithic body. Moreover, small twisting rotation of the capital was observed during the first seconds of excitation for PGA values equal to 0.15g. The rotation was due to rocking. During the next seconds little rocking of the 4th drum was noticed, which induced sliding and twisting rotation of the 3rd and 5th drum. For PGA values higher than 0.2g, the failure mechanism in the case of the ‘no polos’ system is due to extensive rocking of the 3 rd or 4 th drum. It must be noted that in this case failure occurred at the drum above the one exhibiting the extensive rocking. Contrary to that, for the titanium ‘polos’ system, the failure mechanism started from the drum with the extensive rocking. In general it is observed that the induced seismic force was dissipated between the 3 rd and 5th drum for most levels of PGA.

288

capital 7th

Collapse modes

6th 5th

Response concern 3rd -5th drum Response concern 6th drum - capital

4th 3rd 2nd 1st

Displacements (mm)

Stable modes

base

(a)

capital 7th 6th 5th 4th 3rd 2nd 1st base

(b) capital 7th 6th 5th 4th 3rd 2nd 1st base

(c) Figure 3. Comparative results of collapse and stable modes for (a) Erzincan 1992, (b) Aigion 1995, (c) Kalamata 1986

289

Figure 4. Response shapes of the column for several PGA values for Erzincan case

290

PGA : 0.15g The response is almost the same as with ‘no polos system’ case. Few centimeters sliding of the 5th drum is produced in this case.

PGA : 0.15g The response is almost the same as with ‘no polos system’ case. Sliding of the 5th drum is not produced here.

Rotation of the 4th drum produce sliding and rotation of the 3rd and 5th drum in the next few seconds

PGA : 0.15g Rotation of the capital at first seconds of the excitation.

PGA : 0.41g Intense rocking of the 3rd drum. The motion doesn’t transmit to the upper drums. A small bound of the 3rd drum was also produced.

PGA : 0.20g The response is almost the same as with ‘no polos system’ case. The rocking is more intense and produces the collapse of the column earlier.

PGA : 0.18g Intense rocking of the 3rd drum. Contrary to PGA=0.17g the motion transmits to upper drum and produces the collapse of the 4th drum.

The motion transmits fast to the 4th and 5th drum and produce intense rocking and their collapse

Intense rocking of the 3rd drum after a while which produce its collapse.

PGA : 0.20g Same response with ‘wooden polos system’ case. The 3rd drum observes the maximum energy of the excitation.

PGA : 0.26g Tilting begins at 3rd drum at first.

PGA : 0.20g Intense rocking of the 2nd drum at first.

Without polos Wooden polos Titanium polos

Figure 5. Response shapes of the column for several PGA values for Aigion case

291

PGA : 0.45g The rocking of the 7th drum produces sliding of the 6th drum and the capital. The rocking of lower drums coming afterwards produce a sinusoidal mode shape. PGA : 0.45g Severe rocking of 5th drum that produces collapse of the 6th. The rocking of lower drums coming afterwards produce smaller displacements contrary to ‘wooden polos’ case.

PGA : 0.40g Contrary to ‘no polos system’ case, more severe rocking of the 6th drum produces the collapse of the 7th drum and the capital.

PGA : 0.35g Intense rocking of 6th drum that produces rocking and sliding of the lower drums.

In the next seconds rocking transmits to the 4th drum and produces small displacement.

PGA : 0.45g Almost the same mechanism with the previous PGA=0.40g case. Small differences are due to the fact that capital dissipates the energy and not the 7th drum.

PGA : 0.40g Rocking of the 6th and 7th drum produces rotation of the capital.

PGA : 0.50g Rocking and rotation of the 6th drum. Sliding due to rocking is produced to lower drums.

PGA : 0.60g Intense rocking of the 6th drum. Sliding of the lower drums Collapse of the 7th drum and the capital.

PGA : 0.60g Intense rocking of the 7th drum drives to its collapse. This motion produces severe rocking of the 5th and 6th drum and finally their collapse too.

Without polos Wooden polos Titanium polos

Figure 6. Response shapes of the column for several PGA values for Kalamata case

292

PGA : 0.33g Rocking and sliding of the 6th drum but with higher displacements (see no polos case). Rocking of the 2nd drum which follows doesn’t produce many displacements to the lower drums. PGA : 0.33g Rocking, sliding and rotation of 6th drum. Soft rocking is produced also to the 4th drum

It produces rotation of the 7th drum and of the capital due to small tilting

PGA : 0.29g Rocking begins at 5th drum and transmits to the upper drums

PGA : 0.40g Rocking, sliding and rotation of 6th drum. The final displacements that produced are 110% bigger than those of ‘no polos’ case.

PGA : 0.36g Sliding and rocking of the 6th drum. Intense rocking of the 2nd drum produces rocking of the 3rd drum (contrary to no polos case) and finally the collapse.

PGA : 0.36g Sliding and rocking of the 6th drum at first seconds of motion. Intense rocking of the 2nd drum after a while which produce rocking of the 4th drum and finally the collapse.

PGA : 0.55g Intense rocking of the 6th and the 7th drums produces their collapse. The rocking of the lower drums that follows collapse is softer and produces a sinusoidal shape response of drums.

PGA : 0.55g Intense rocking of the 6th drum produces the collapse of the 7th drum and the capital. The rocking of the lower drums that follows is softer and produces a sinusoidal shape response of drums.

PGA : 0.40g Sliding and rocking of the 6th produced small displacement of the 5th drum and the capital. Rocking of 2nd drum is not so intense (contrary to PGA=0.36 case) and shared also to the upper drums. The column not collapses.

Without polos Wooden polos Titanium polos

In the case of Aigion earthquake, it is observed that the threshold PGA value for collapse cannot be precisely defined. Collapse of the column is observed for a PGA value around 0.5g, while for higher values the column could also prove stable (Figure 3b). The titanium poles reduced the displacements for most PGA levels. From Figure 5 it is clear that a great part of the seismic energy is dissipated at the capital and upper drums even for low values of PGA. The mode response shows rotation of the capital by about 45 o due to rocking after the end of the excitation. Because of the intense rocking, the 7th and 6th drums also exhibit some displacement, of the order of 10cm (Figure 5). For greater values of PGA (>0.40g), rocking is induced to lower drums, in particular at the 3rd one Finally, it was also observed as in the previous case (Erzincan) that the 2 nd level of damage was reached for increasing values of PGA. But here, the higher frequency content of the earthquake, contrary to the previous case of Erzincan, distributed the seismic energy across each drum of the column. This is indicated by the final displacement pattern which is also distributed over all drums. Finally, neither for the Kalamata earthquake is it possible to determine a specific value of PGA for which the column collapses. Collapse in most cases happens due to rocking of the lower drums, namely the 5th and 6th, as is presented in Figure 3c. It has been also observed that upper drums are more excited for PGA values lower than 0.35g.On the contrary for higher PGA values the lower drums are excited the most. In the case of a system ‘without polos’, extensive rocking is observed at the 6th drum during the first seconds of strong motion. During the next few seconds deflections are transmitted to lower drums. In many cases that produced a sort of sinusoidal shape of the column. The same was observed for the wooden ‘polos’ system case (Figure 6). The titanium poles reduced permanent displacements for most PGA levels. Extensive rocking was observed for PGA values greater than 0.40g and more precisely to the lower drums, namely the 4th and 5th, which finally produced the collapse of the column. For the other two ‘polos’ cases the failure process started with the collapse of the capital, followed by the 7 th or 6th drum. Distribution of energy over all drums was also observed, probably due to the high frequency content of the input motion. That is probably the reason for the sinusoidal shape observed.

EXPERIMENTAL ESTIMATION OF COLUMN RESPONSE THROUGH MICROTREMOR MEASUREMENTS AND COMPARISON WITH NUMERICAL RESULTS Selection of the case study Ancient columns located at two different sites were used as case studies for microtremor measurements. The first one is a free multidrum column and the second one is a system of three multidrum columns connected with architraves. Both are located in the Acropolis of Lindos, Rhodes (Figure 7a). In addition to these two cases, a monolithic column located in the ancient Agora of Thessaloniki is the third case investigated (Figure 7b). Geological setting of the investigated sites Lindos Acropolis is situated near the middle of the west coast of Rhodes Island. It is laid on a limestone rock outcrop hill of 80m high. The Ancient agora of Thessaloniki is extended on stiff clay with average Vs of 450 to 550 m/sec. This layer overlies the schist rock basement at relatively low depth, around 50m (Apostolidis 2002). Ambient noise tests Single station ambient noise measurements is an interesting method traditionally used for site response analysis. It has been recently applied in structure response analysis as well. Microtremors include random vibration of the ground caused by physical phenomena (wind, ocean waves, etc.) or human activities (traffic, industrial noise and other ambient disturbances). It was found that it can accurately describe a few basic vibration characteristics of ground and structures. (Aki, 1957; 1965). The ratio of the horizontal to vertical Fourier spectra (HVSR) is a usual way to estimate the fundamental frequency of a soil profile (Nogoshi and Igarashi, 1971). Although the significance of the HVSR peak amplitude is still a subject of debate (Nakamura, 1989, proposed that it gave reliable

293

estimates of soil amplification, something which was later questioned by Kudo, 1995), the frequency at which the peaks appear is considered reliable (Bard, 1999). Lermo and Chávez-García (1993) introduced the use of HVSR computed for earthquake records to study site response and site effects. More recently, noise measurements have been used for estimating the seismic response of structures, investigating site-structure interaction and evaluating damage and vulnerability (Chávez-García and Cárdenas, 2002; Mucciarelli et al., 2004; Gallipoli et al., 2003; Irie and Nakamura, 2000; Volant et al., 2002; Nakamura et al., 1995; Nakamura, 1997). Galli et al. (2006) used noise at a historical site in Calabria, to estimate site effects and monument response. Noise measurement HVSR has been also performed to study the response of historical monuments, such as the Coliseum in Rome (Nakamura et al., 2000) and the leaning tower of Pisa (Nakamura et al.,1999). Another empirical technique used here, and which also initiated as a site response study tool, is the standard spectral ratio (SSR). This consists of the spectral ratio relative to a reference station. SSR has been used to evaluate site effects since Borcherdt (1970). It is assumed that a record on the bedrock or a nearby rock outcrop can be considered representative of the input motion below the sediments and thus the ratio is related to site response of the soil profile. Again, this method’s scope is extended in this case so as to study the response of the structure, rather than the sediments. The sites used as reference sites lay on rock outcrop, or almost.

5.70 m 6.70 m

(b) (a) Figure 7.(a) Archeological site of Lindos in Rhodes and column case studies. (b) Archeological site of Ancient Agora in Thessaloniki and column case study. The column heights and locations of the recording instruments are also shown. In the present study, ambient noise was recorded at the top of the columns under study and at a “reference” site nearby. In the case of the Acropolis of Lindos in Rhodes, instruments were placed in the configurations shown in Figure 8a and 8b: one at the top of a single column (a), one on top of the architrave of the three-column-system (b) and one on the ground, rock outcrop (reference instrument), at a distance roughly equal to the height of the column. In the case of the Agora in Thessaloniki, instruments were placed as shown in Figure 8a: one at the top of a single column and one on the ground, at a distance again roughly equal to the column height. The exact locations of the instruments are shown with red arrows in Figure 7. It should be noted here that in the case of the architrave,

294

because only one instrument was placed on the structure, there is no way of estimating rotational motion. Thus, all results can only be interpreted in terms of translation. In both cases, ambient noise was recorded continuously for 30 to 35 min using three component broad-band seismometers Guralp (CMG-40T of 30 sec natural frequency), coupled to a high resolution Reftek recording system (DAS-130) and a GPS unit. The sampling interval of the records was 0.008sec.

(a)

(b)

Figure 8. Schematic representation of the place of instruments H/V Spectra Ratios have been computed using SAC (Goldstein et al., 2003). First, the records were corrected for linear trend and constant offset. A band-pass filter with frequency corners from 0.1Hz to 25 Hz was applied and a representative noise window of 400sec time duration was chosen, avoiding the obvious parasites and spikes in the records. This window was divided into 19 windows of 40sec duration with a 50% overlap. Each window was cosine-tapered (10%) and Fourier transformed and its amplitude spectrum smoothed with a running Hanning filter. Finally, the ratio between the two horizontal components (EW, NS) relative to the vertical one was computed for every time window. In agreement with previous results (Mucciarelli et al., 2005), it was proved that wind does not affect the frequencies indicated by the HVSR while it increases the amplitude on all components. Thus the results can be considered unaffected by any wind present during the measurements, at least regarding the resonant frequencies. For all calculations related to the architrave case in Lindos, the NS and

EW components of noise recorded were rotated parallel and perpendicular to the architrave direction. We applied both HVSR and the classic SSR technique, where the fundamental response characteristics are estimated through the ratio of the two horizontal components recorded on the structure and at the reference site. Figure 9 shows the HVSR for the reference sites. Theoretically, a reference site lying on rock outcrop is unaffected by any site effects which is reflected by an HVSR near unity over a significant frequency range. However this is not always the case, as it is observed in the case of Lindos outcrop site where the mean ratio is about 2. 10

10

1

1

0.1 0.1

1

f [hz]

(a)

10

0.1 0.1

1

10

f [Hz]

(b)

Figure 9. HVSR in case of (a) Agora Thessaloniki , (b)Lindos Numerical response

295

Modal analyses were performed to investigate the eigen characteristics of the examined columns which in both case studies were considered as monolithic Figures 10b and 11b present the results of the modal analysis of the single column and the columns with architrave in Lindos. The range of values of the fundamental frequencies is due to the range of values assumed for the elastic modulus and the mass density of the column material. The parametric analyses considered different combinations of the limestone’s elastic modulus (E) (considered to take any value between 1.40 and 1.55 GPa) and its density ( ) (taking values between 1.7 and 2.0 ton/m3). . Comparison of experimental and numerical results: Lindos Single column Figure 10a shows the results derived from noise measurements using SSR for both horizontal components. As expected, the two components are almost identical. Three distinctive peaks appear at frequencies 4.0Hz, 6.5Hz and 7.0Hz. On the same figure the results from numerical simulation (violet shaded rectangles) are shown, indicating two distinguished frequencies range between 3.8-4.2Hz for the first mode and between 1721Hz for the second. The results for HVSR are shown in Figure 10c. SSR and HVSR agree well for the peaks that appear at 6.5Hz and around 18Hz, while the HVSR method completely misses the 4Hz peak which is visible only through SSR. This frequency is also depicted numerically as the equivalent fundamental frequency (3.4Hz). One possible explanation is that, for frequencies smaller than 5Hz, the vertical component is so well correlated to the horizontal ones due to flexibility of the column that the ratio is not useful. Moreover it should be expected that the fundamental mode and overtones should follow the fundamental expression fn=fo*(2n+1). However the two peaks at 4.0 Hz and at 6.5 Hz do not follow this relation, suggesting that they correspond to different vibration mechanisms. SSR is perhaps more appropriate to capture these complex response compared to HVSR which has some inherent shortcomings.

f 1=3.25 – 4.25 Hz

100 90

HVSR(2)

80 70 60 50 40

HVSR(3)

30 20 10 0 0

5

10 f (Hz)

(c)

15

20

f 2=17.0 – 19.8Hz

(b)

Figure 10. (a) SSR results (b) mode shapes of the free column (c) HVSR results of the free column Columns with an architrave

296

Figure 11a shows the SSR results obtained with ambient noise measurements for the architrave; motion is oriented in parallel and transversely to the architrave (N32E, N58W respectively). The peaks appear at frequencies of 4.0 and 7.0 Hz. On the same graph the numerical results are shown as ranges for the first three modal shapes. The red shaded rectangle indicates the range of the frequency for the 1st mode. The green shaded rectangle indicates the range of the frequency for the 2nd mode. Finally, the possible range for the frequency of the 3 rd mode is indicated with the blue shaded rectangle. The parametric analyses accounted for the same combinations of the limestone’s mechanical properties as for the single column case. l. Figure 11c, finally, shows the results from HVSR for both components oriented, like the SSR, parallel and transversely to the architrave (N32E, N58W respectively). The peaks appear at frequencies 4.0, 7.0, 11.0 and 17.0 Hz. It is of interest to note that the peak at 7.0Hz disappears completely for the transverse component. 1200 SSR(2)

1000

SSR(3) 800 600 400 200 0 0

5

10 f (Hz)

15

20

1st mode shape f1=2.73Hz

2nd mode shape f2 =2.77 Hz

(a) 35 HVSR(2)-N58W

30

HVSR(3)-N32E

25 20 15 10 5 0 0

5

10 f (Hz)

(c)

15

20

3rd mode shape F3=4.38Hz

4th mode shape f2 =10.0 Hz

(b)

Figure 11. (a) SSR results (b) mode shapes of the free column (c) HVSR results of the three column case Comparison of experimental and numerical results: Thessaloniki ancient Agora site Figure 12a shows the results derived from noise measurements using SSR ratios for both components, for the single monolithic column. As expected, the two components (2 and 3) are almost identical. As has been already mentioned, the range of amplitudes is not significant; and we rely only on the recorded frequency values. Two clear peaks appear at frequencies of 5.2 Hz and 6.8Hz. In the same figure the computed results shown a range of frequency between 7.0-10.0Hz (green shaded rectangle). This frequency range is reached through several parametric analyses, for different combinations of the marble’s mechanical properties. Namely, the elastic modulus (E) ranges from 45 to 60GPa while mass density ( ) ranges between 2.5 and 5.0ton/m3. The results using HVSR are shown in Figure 12b. The experimental frequency of the first peak lies around 6.4Hz, a value close to the second one given by SSR. Finally it must also remarked that the experimental ambient noise measurements might include certain site effects due to the fact that the site taken as reference does not show a flat response for HVSR in a frequency range of 8-10Hz.

297

100

SSR(2)

Top-HVSR(2)

SSR(3)

Top-HVSR(3) 10

10

1

1

0,1 0,1

0.1

1

f (Hz) 10

0.1

1

10

(b) (a) Figure 12. (a) SSR results (b) HVSR results of the free column

CONCLUSIONS The paper presents a set of numerical and experimental analyses in order to investigate the seismic response of monolithic and multidrum ancient columns. The main conclusions are summarised as follows: From the three different input signals used, two of them did not allow to determine a specific “yielding” value of PGA that would lead to the collapse of the columns. Rocking occurs at all drum interfaces for PGA values around 0.2g-0.4g for rather highfrequency, near-field earthquakes (Aigion and Kalamata case). Larger PGA values lead to extensive rocking of the lower drums while the response of the upper part is rather monolithic. The seismic energy is dissipated by the lower drums for increasing values of PGA. For Erzincan earthquake record, showing an energy distribution over a wide range of periods, the maximum dissipation of seismic energy through extensive rocking occurs between the 2nd and 4th drum for most PGA values. The collapse mode that was observed (between the 2nd and the 3rd drum) indicates a monolithic behaviour of the column. For the Kalamata record, whose maximum energy occurs in the period range between 0.3 and 0.7 sec, the 4th and 5th drums also seem to be affected by the excitation. However, the higher frequency content of this input motion distributes the energy along the height of the column by dislocating more drums. Finally, for the Aegion record, with a fundamental period of 0.5 sec, it was observed that the 6th drum suffered the largest dislocation and its rocking affected the motion of the other drums. The titanium poles reduced the displacements in most of the cases. For PGA values larger than 0.5g, however, the use of titanium produced a monolithic collapse of the column which is generally undesired. The observed twisting rotation was mainly due to rocking. It is observed that the twisting rotations of the lower drums were enhanced by the extensive rocking of the upper ones. Finally, a set of microtremor measurements were performed at two different archaeological sites for three different column configurations. The goal was to estimate the eigen-properties of the column and compare it with the numerical modal analysis. Both HVSR and SSR methods were applied. Distinct peaks were clearly visible over specific frequency values which were quite close to the theoretical modal analysis at least for the SSR. With HVSR it is more difficult to capture the probable fundamental mode frequency. This is probably due to the fact that the vertical motion at the top of the column is very similar to the horizontal one because of the large flexibility of the column. The results derived from the numerical analysis were very similar to the results from SSR, especially in the case of Lindos. Some discrepancies were noted, possibly due to site effects at the sites taken as reference. In general it is concluded that noise measurements may give interesting tool to study the seismic response of monuments and in specific for ancient columns. Given the low cost of this method, as well

298

as the fact that it does not cause any damage to the monuments, it may be a very useful tool to investigate the seismic response of monuments and complement numerical calculations.

AKNOWLEDGEMENTS The field surveys described in this paper were financially supported through the Pythagoras project (Contract no. 97436). The authors express their thanks to Prof. Michalis Tiverios (Department of History and Archaeology, AUTH), to archaeologist Lilian Acheilara (16 th Ephorate of Prehistoric and Classical Antiquities, Thessaloniki) and to civil engineer Maria Pikoula (Committee of Restoration Works Progress in Lindos Acropolis, Rhodes) for the measurements in the archaeological sites of Thessaloniki and Rhodes. The first author (NA) would like to express his gratitude to Dr. Anastasios Sextos (Lecturer, Department of Civil Engineering, AUTH) for his overall contribution; His initial inspiration and the fruitful discussions proved very helpful in producing this paper. The first author also thanks Prof. Marco Mucciarelli (Universita della Basilicata, Italy) for supplying relevant literature. The second author (OJK) wishes to thank the Propondis Foundation for its financial support through scholarship.

REFERENCES Aki K., 1957. Space and Time Spectra of Stationary Stochastic Waves, with Special Reference to Microtremors. Bull. Earthquake Res. Inst. Tokyo Univ. 25, pp. 415-457. Aki K., A note on the use of microseismic in determining the shallow structure of the earth’s crust, Geophysics Vol. 30, pp. 665-666, 1965. ANSYS 8.1 (2004). Users Manual Apostolidis P. “Determination of ground structure with the use of microtremors Application:Estimation of the dynamic properties and geometry of Thessaloniki soil formations”. Phd Thesis, Department of Civil Engineering, AUTH, 2002 (in Greek), Argyriou N, Pitilakis K. Sextos A., ‘Numerical study of the seismic behavior of ancient columns, Proceedings of 1st Hellenic conference on Scientific Restoration Works, Thessaloniki 14-17 June 2006. (in Greek),. Bard P.-Y. "Local effects on strong ground motion: Physical basis and estimation methods in view of microzoning studies." Proc. of Advanced Study Course in 'Seismotectonic and Microzonation techniques in Earthquake engineering', Kefalonia, Greece 1999. Borcherdt, R.D. “Effects of local geology on ground motion near San Francisco Bay”, Bull. Seis. Soc. Am. 60, 29-61, 1970. Bottari C. “Ancient constructions as markers of tectonic deformation and of strong seismic motions”, Proceedings of 11th FIG Symposium on Deformation Measurements, Santorini, Greece, 2003. Chávez-García FJ and Cárdenas M, “The contribution of the build environment to the “free-field” ground motion”, Soil Dyn & Earthq. Engrg., 22, 773-780, 2002. Eleftheriou V. Committee for the restoration of monuments at Lindos Acropolis, Rhodes. “Reconstruction works on the Hellenistic Portico”. Athens 2002 (in Greek) European Strong Motion Database (ESMD): http://www.isesd.hi.is/ESD_Local/frameset.htm, Mar. 2007 Galli P., Ruga A., Scionti V., Spadea R., “Archaeoseismic evidence for a Late Roman earthquake in the Crotone area (Ionian Calabria, Southern Italy)”:Seismotectonic implications. J. Seismol. 10, pp. 443-458, 2006 Gallipoli M.R., Mucciarelli M., Gallicchio S., Tropeano M., Lizza C. “Structure, soil-structure response and effects of damage based on observations of horizontal-to-vertical spectral ratio of microtremors”, Soil Dyn. Earthquake Eng. 24, pp. 487-495, 2003. Goldstein P, Dodge D, Firpo M, Lee Minner "SAC2000: Signal processing and analysis tools for seismologists and engineers. Invited contribution to "The IASPEI International Handbook of Earthquake and Engineering Seismology", Edited by WHK Lee, H. Kanamori, P.C. Jennings, and C. Kisslinger, Academic Press, London, 2003.

299

Irie Y., Nakamura Y. (). “Dynamic characteristics of a r/c building of five stories based on microtremor measurements and earthquake observations”, Proc. of the 12th World Conf. on Earthquake Engineering, Wellington (NZ), pp. 500-508, 2000. Kudo K., “Practical estimates of site response”, State of the art report, in: Proc. of the 5th intern. conf. on seismic zonation, Nice, Ouest editions Nantes, 3, pp 1878-1907, 1995 Lermo, J., and F.J. Chavez-Garcia. “Site effects evaluation using spectral ratios with only one station”, Bull. Seis. Soc. Am. 83, 1574-1594, 1993. Makris N Konstantinidis D,. Seismic response analysis of multidrum classical columns. Earthquake Engng Struct. Dyn.,DOI:10.1002/eqe.378, 2005 Manos G., Demosthenous M., Models of ancient columns and colonnades subjected to horizontal base motions – Study of their dynamic and earthquake behaviour. Transactions on the Built Environment vol 26, © 1997 WIT Press Manos G. ‘The recycling behavior of polos-empolio, as a joint of sliding drums under earthquake excitations, Proceedings of 1st Hellenic conference on Scientific Restoration Works, Thessaloniki 14-17 June 2006. (in Greek) Mucciarelly M., Masi A., Rosaria Gallipoli M., Harabaglia P., Vona M., Ponzo F., Dolce M. “Analysis of RC building dynamic response and soil-building resonance based on data recorded during a damaging earthquake” (Molise, Italy, 2002). BSSA 94(5), pp. 1943-1953, 2004. Mucciarelli M., Gallipoli M.R., Giacomo D., Nota F., Nino E., “The influence of wind on measurements of seimic noise”, Geophysical Journal International, Vol. 61, issue 2, pages 303-308, 2005 Mouzakis HP, Psycharis IN, Papastamatiou DY, Carydis PG, Papantonopoulos C, Zambas C. Experimental investigation of the earthquake response of the model of a marble classical column. Earthquake Engineering and Structural Dynamics; 31(9):1681-1698, 2002. Nakamura Y. A method for dynamic characteristics estimation of subsurface using microtremors on the ground surface, QR Railway Technical Resource Institute 30, 1, 1989. Nakamura Y., Hidaka K., Sato S., Tachibana M. “Porposition of a method for pier inspection using microtremor”, Quarterly Report of RTRI, Vol. 36, No. 1, ISSN 0033-9008, 1995. Nakamura, Y., “Seismic vulnerability indices for ground and structures using microtremor”, World Congress on Railway Research in Florence, Italy, November 1997. Nakamura, Y., Gurler, E.D. and Saita, J., “Dynamic Characteristics of Leaning Tower of Pisa Using Microtremor-Preliminary Results”, Proc. 25th JSCE Earthquake Eng. Symposium, Vol. 2, 921-924, 1999 Nakamura, Y., Gurler, E.D., Saita, J., Rovelli A. and Donati S., “Vulnerability investigation of Roman Colliseum using microtremor”, Proc. 12th WCEE, paper no. 2660, 2000 Nogoshi M and Igarashi T. ”On the amplitude characteristics of microtremor (Part 2)”, Journal of Seismological Society of Japan, 24, 26–40, 1971 (in Japanese with English abstract). Orlandos AK. “Construction Materials of the Ancient Greeks”, vol.2. Archeological Society of Athens: Greece, 1958 (in Greek) Papantonopoulos C, Psycharis IN, Papastamatiou DY, Lemos JV, Mouzakis H., “Numerical prediction of the earthquake response of classical columns using the distinct element method”. Earthquake Engineering and Structural Dynamics; 31(9):1699 –1717, 2002. PEER Strong Motion Database (http://peer.berkeley.edu/smcat/search.html), Mar. 2007 Psycharis IN, Papastamatiou DY, Alexandris AP. “Parametric investigation of the stability of classical columns under harmonic and earthquake excitations”. Earthquake Engineering and Structural Dynamics; 29(8):1093–1109, 2000. Psycharis I, Lemos J, Papastamatiou D, Zambas C, Papantonopoulos C, “Numerical study of the seismic behaviour of a part of the Parthenon Pronaos”. Earthquake Engng Struct. 32:2063-2084, 2003 Shenton III HW. “Criteria for initiation of slide, rock, and slide-rock rigid body modes”. Journal of EngineeringMechanics (ASCE); 122:690–693, 1996. Sinopoli A. “Kinematic approach in the impact problem of rigid bodies”. Applied Mechanics Reviews (ASME); 42(11):S233–S244, 1989.

300

Volant P.H., Orbovic N., Dunand F., Seismic evaluation of existing nuclear facility using ambient vibration test to characterise dynamic behaviour of the structure and microtremor measurements to characterise the soil: a case study’, Soil Dyn. Earthquake Eng. 22, pp. 1159-1167, 2002. Yim CK, Chopra AK, Penzien J “Rocking response of rigid blocks to earthquakes”. Earthquake engineering and structural dynamics; 8: 565-587, 1980 Zambas K. “Study for the restoration of the Parthenon”. Committee for the preservation of the acropolis monuments, Ministry of Culture, Athens 1994 Zhang J, Makris N., “Rocking response of free-standing blocks under cycloidal pulses”. Journal of Engineering Mechanics (ASCE); 127(5):473–483, 2001.

301

4th International Conference on Earthquake Geotechnical Engineering June 25-28, 2007 Paper No. W2-1015

Design and Implementation of Engineering Measures for the Protection of a Historical Monument at the Seismic Area of Mount Athos Peninsula, Greece Stavros C Bandis1, Christos Schinas2, Elias Bakasis3 ABSTRACT The paper refers to a major project of geotechnical studies, design and implementation of engineering measures for the protection of the understructure of an important historical monument under restoration, the Stavronikita Monastery at Mount Athos, Greece dating back to the 10 th century. The area is part of a highly seismic zone of the Greek territory and earthquakes of intensity >7 on the Richter scale are known to have occurred over the last century. The superstructure of the monument has suffered considerable damages due to extensive cracking associated with slope movements. Site investigation works were carried out for the recognition of the principal causes of instability and the design of appropriate and adequate engineering measures. The implementation of the project presented significant challenges because of the extent of the works, the construction limitations associated with the site and the requirement to minimize environmental impact. The project was completed in 2000 and is being monitored by an appropriate network of instrumentation. Keywords : seismic slope stability, Stavronikita Monastery INTRODUCTION Integrated engineering schemes for long-term protection of historical monuments invariably encompass design issues associated with the effects of time and processes on the natural structural materials of the superstructure and the ground forming the monument understructure. The monastery of Stavronikita is located at the eastern coast of the Athos peninsula in northern Greece and is an important historical monument dating back to the 10th century (Figure 1). It is a masonry wall compound occupying an area 60x35m and is founded on the edges of a 35-45m high rock cliff forming slopes 30o-40 o to the east and 50o-70o to the north.

FIGURE 1. Views of the northern side of the cliff of the Stavronikita Monastery, Mount Athos, 1

Professor, Department of Civil Engineering, Aristotle University, Thessaloniki, GR Lecturer, Department of Civil Engineering, Aristotle University, Thessaloniki, GR 3 Lecturer, Department of Civil Engineering, Aristotle University, Thessaloniki, GR 2

302

The superstructure of the monument has suffered considerable damages due to extensive cracking associated with slope movements both towards the northern and the eastern directions. Crack widths were 50-350mm and exhibiting periodical widening. The cracks had a preferential orientation in two principal directions namely E-W and N-S as shown in Figure 2, which includes typical views of major cracking in the masonry walls and the foundation bedrock. Such effects were attributed to earthquake damages, permanent deformations of the rock cliffs, natural wear of construction materials and the construction history of the complex.

(a)

< (b)

10m

(a)

(b) (c)

FIGURE 2 . Plan view of the ground floor of the Stavronikita Monastery compound with network of major cracks. Photographs illustrate cracks in (a) the northern wall (b) the western wall and (c) in the rock mass at foundation level of the north wall basement.

GEOTECHNICAL INVESTIGATIONS The site investigation program comprised exploratory boreholes with orientated core recovery, joint surveying, piezometric monitoring and extensive laboratory tests. Special in-situ tests were also executed to derive shear strength parameters for critical discontinuities in the rock mass. The rock cliff materials are schists and gneisses, with well-developed schistosity planes and joints. Attributes of engineering significance were the variable degree of weathering / alteration, the intensive jointing and the presence of continuous, open planes of schistosity. Typical strength parameters for the rock mass components (rock material and discontinuities) were measured and assessed as summarized in Table 1.

303

Table 1. Summary of rock mechanical properties of rock materials and discontinuities Rock materials Depth WG (m) 0-5 IV

UCS (MPa) 1-5

E (MPa) 2000

5-10

III

5-20

12000

10-20

II-III

40-70

40000

Rock discontinuities Description Schistosity in Grade II-III rock Schistosity in Grade II-III rock Joints

peak

res

(deg) 40o

(deg) 30o

35o

25o

50 o

30o

IN-SITU TESTS (Fig. 3a &b) Roughness coefficient of schistosity JRC=6-7 Weathered schist c=60 kPa =35o-40 o

WG : weathering grade UCS : Uniaxial Compressive Strength, E: Modulus of Elasticity peak : Angle of peak friction, res : Angle of residual friction

(a)

Application of normal load by hydraulic jack resting against an anchored reaction beam

Anchors retaining the reaction beam Hydraulic jack for application of shear load

(b)

MULTI-STAGE TEST Normal stress Shear stress 73 kPa 27 kPa 115 kPa 45 kPa 155 kPa 71 kPA

Figure 3. (a) In situ shear test on a schistosity plane for the determination of large scale roughness and shear strength (block size 1000x600 mm) (b) In situ shear test on 500x600 mm block for measurement of the shear strength of weathered schist

304

SLOPE STABILITY MECHANISMS From evaluation of all information the following conclusions were drawn relating to the principal stability mechanisms: The stability of the cliff slopes was controlled predominantly by the rock mass structure. A major unstable wedge was identified at the eastern slope formed by schistosity and two major structural features acting as release planes. The base plane (schistosity) of the wedge was below sea level. Multi-planar kinematic sliding mechanisms controlled the stability of the northern slope. The structure of the rock mass has a dominant control of the overall deformation modes of the rock cliff. Several normal faults running N-S and E-W intersect the rock mass and three faults were identified cutting across the foundation bedrock of the monastery. Jointing is well-developed and comprises three main sets, namely J1:0 o–020 o/65o-75o, J2:0o-020o/45o-55 o, J3:320o/30o-40o. The network of discontinuities combined with the slope faces creates different kinematic mechanisms towards the north and the east sides. Eastern slope Orientated core recovery revealed several schistocity planes (dip direction / dip : 140 /25 ) and joints (225 / 80 ) with slickensided surfaces at depth of ca 20 m (Figure 4c). The orientation of the slickensides was 100 -130 /15 -25 and almost coincided with the line of intersection between the two sets of discontinuities. The geometrical combination of the planes 140 /25 , 225 /80 and a known fault isolated a 5-plane wedge volume sliding along the line of intersection of the above planes. The outline of the unstable wedge at the eastern slope is illustrated in Figure 3. The Factor of Safety (FoS) of the wedge with the toe terminating below sea level was calculated to be close to limiting state (FoS=0.9) for seismic loading. 225 /80

(b)

FAULT PLANE

N-S FAULT DIRECTION OF WEDGE MOVEMENT

Schistosity 140 /25

Direction of slickensides coincides with the line o intersection of schistosity with joint set 225 /80

SCHISTOSITY 140 /25

(a)

FIGURE 4 (a) Diagrammatic illustration of major wedge at eastern slope (b) Kinematic mechanism of slope instability (c) Slickensided schistosity plane

Scistocity plane (140 /25 )

(c)

305

Northern slope The stability is controlled by the joint sets J1 and J2 which promote planar or multi-planar sliding mechanisms. The schistocity planes daylighting at the northern slope face, form the lowermost plane of sliding with a shallow opposite dip (i.e. towards the slope), thus achieving a marginal static equilibrium. However, the extensive sliding failures that have occurred in the past (evident from the slope face overbreaks as seen on the photograph of Figure 1) imply that seismic loadings and local toe overstressing under static load may activate instability. The latter can be envisaged to develop rapidly as a result of the structure controlled and gravity driven mechanism influencing a slope zone that penetrates 3-4 m from the face and endangering the largest part of the northern wall of the compound. Permanent deformation of the rock mass under past seismic events has resulted in notable loosening (loss of interlocking) of the rock blocks close to the surface with associated significant reduction of the shear strength and stiffness. A view of the “disturbed” rock mass is provided by the photograph in Figure 2c, where the open “cracks” are dislocated joints of the set J1. The disturbed rock mass zone extends at least down to the level of the slope undercutting (see Fig. 1) created by a local zone of poor materials and wave erosion. The illustrations in Figure 5 present the distributions of the rock materials and structure as modelled in the UDEC analyses, which will be referred to later.

Highly disturbed rock mass zone Slope undercut

Weathered schist

Schistosity

Gneissic Schist moderately weathered J1

(a)

Schist Slightlymoderately weathered

J2

(b)

Figure 5. Sections of the northern slope (a) Distribution of geological materials and extent of disturbed rock mass zones (b) Rock mass structure model simulated in the UDEC analyses.

STUDY OF AREAL SEISMICITY The particular area is highly seismic as shown by the accumulation of epicentres during the last 100 years in Figure 6. Studies of the areal seismicity assuming a radius of 150 km yielded the seismic parameters summarized in Table 2.

306

ZONE 17

Figure 6. Zonal map of seismicity and earthquake epicentres in the area of study

Table 2. Seismic characteristics of the area within 150 km from the site Seismic characteristics I (MM) amax Vmax (cm/sec)

5 years

10 years

20 years

50 years

V 0.04 2.60

V-VI 0.08 5.50

VI-VII 0.12 8.40

VII-VIII 1.17 12.20

SHALLOW EARTHQUAKES Time period Epicentre Magnitude 1901-1985 6.5-7.7 1911-1985

5.5.- 6.4

1911-1985

5.2 – 5.4

1950-1985

4.8 – 5.1

1964-1985

4.5 – 4.7

The seismic design parameters were selected as follows: a = 0.12 Vhorizontal = 10 cm/sec (1932 earthquake) Cs( shear wave velocity)= 2000 m/sec STABILITY ANALYSES Stability calculations were carried out on representative sections of the slopes in parallel with wedge analyses. Factors of safety under static/ seismic loadings (a=0.12) were derived for the northern slope as summarized in Table 3.

307

Table 3. Factors of safety of northern slope against multi-planar sliding failure Initial state (undisturbed1 rock mass): 100t/m

FoS (static) = 1.2 - 1.9 (ave 1.35) FoS (seismic)= 0.9 - 1.4 (ave 1.04) Current state (disturbed rock mass):

100t/m 100t/m

600

600

2

100t/m

100t/m

130 0

60

600

1300

0

700

580 600

FoS (static) = 0.95 - 1.6 FoS (seismic3) < 1.0

700

(ave 1.1)

1300

1300

0

20 00 100 100

1000 500 1000 600 200

00

1000

1

2 3

The term “undisturbed” rock mass implies an initial inferred condition prior to the permanent deformation state and associated loosening as currently observed. The “disturbed”condition was represented by assigning reduced shear strength and stiffness parameters to the rock mass. The design stability calulations were carried out on the basis of Fhor =2 W Fvert =(2 W)/3 as suggested by the Code for Antiseismic Design valid at the time (1991). At construction stage all calculations were carried out on the basis of the revised Code (NEAK 1995) where Fhor =0.5 W and Fvert= 0.25 W.

At the eastern slope the 2-D and 3-D analyses demonstrated that the Factors of Safety were: Static FoS = 1.20 Seismic FoS 1.00 The engineering assessment of all analytical studies converged to the conclusion that the stability state and geotechnical conditions as the northern slope presented high risk of a catastrophic failure. At the eastern slope, although the geotechnical conditions did not favour sudden destabilization, it was evident that the periodical mobilization of slope movement due to earthquake activity and creeping could prompt critical conditions to the structural integrity of the eastern wing of the compound. DESIGN OF STABILIZATION MEASURES The following design criteria were set for the selection of the stabilization scheme: FoS (static) >2.0 FoS (seismic+ground water pressure) > 1.4 Control deformations under seismic loadings to < 20mm (ca 5o/oo x H) The FoS were considerably higher than would be considered for a typical civil engineering project. However, it was deemed necessary to upgrade the safety reserves due to the special nature of the structure, the adverse seismic loading environment and the highly disturbed nature of the rock mass. The stabilization measures selected after a rigorous investigation of different options comprised:

308

Perimetrical retention system consisting of rows of high capacity pre-tensioned anchors. Reinforcement of the foundation rock mass with micropiles and grouting. Underpinning of masonry structures and part of the wing walls. Special retaining structures to support critical slope zones, especially at the northern slope Figure 7 illustrates the general layout of the slope anchoring system.

NORTHERN SLOPE

EASTERN SLOPE

Figure 7. General layout of the perimetrical anchoring and micro-piling systems

309

(a) Layout of micropiles for strengthening of the north wing foundation subgrade

(b) Section at basement of compound with details of micro-piling system

(c) Section of micropile founded pedestal structure at undercutting of the northern slope

Figure 8. Details of engineering measures for improvement of the rock mass strength and stiffness and provision of local support.

DESIGN VALIDATION AND IMPLEMENTATION Due to the limited access to various sectors of the site during the SI and subsequent design stages, systematic monitoring and interpretation of the conditions was undertaken during construction to validate the design assumptions and engineering solutions. It was thus possible to optimize the engineering measures at local scale by adapting the design to the conditions as exposed, particularly for strengthening the foundation subgrade at the previously unaccessible basements of the northern

310

wing of the monastery. However, no modifications were required for the principal stabilization scheme of slope anchoring. The validated geotechnical models provided a reliable basis for further computational verification of the adequacy of the engineering measures using state-of-art numerical modelling techniques. The Distinct Element Method (DEM) was applied for rigorous seismic stability analyses of the slopestabilization system. The DEM is implemented in the Finite Difference Software UDEC-BB (Universal Distinct Element Code coupled with the Barton–Bandis constitutive model for interface shear and normal stiffness behavior. A set of characteristic plots from the UDEC analyses demonstrating the predicted stabilization effects of the remedial measures under seismic loading of the northern slope is presented in Figures 9 and 10.

0.6g

Instant of seismic excitation of model at displacement stage illustrated below

20 mm

FAULT

ANCHORS

SLIDING PLANE OF WEDGE AT EASTERN SLOPE

22 mm

Figure 9. UDEC model predicted displacement state of eastern slope with fully deployed Stabilization measures.

311

(b) (c)

(a) The model predicted slope collapse without stabilization measures (b) Stable state was predicted (albeit below displacement threshold of 500 m/s) is very variable, and that it is often hard to identify a firm bedrock between the Pleistocenic conglomerates and the Pliocenic clay formation.

Vs (m/s) 800

650

500

360

180

Figure 3. 3D shear wave velocity model for the subsoil of Benevento (after Penna et al., 2007). Table 2 summarizes the soil properties, either measured or assumed on the basis of literature data, of the geotechnical units. The initial damping, D0, and the linear and volumetric threshold strain levels, l and v, were used to define the G/G0- and D- curves relevant for the seismic response analyses (Santucci de Magistris et al., 2004).

Reference input motion Benevento falls inside the seismogenic zone Z927 Sannio - Irpinia – Basilicata (MPS Working Group, 2004), that represents the most hazardous earthquake source for the city. Earthquakes in zone Z927 have a mode focal depth of 10 km and normal faulting mechanism. For a return period of 475 yrs, the maximum acceleration amax expected in the city on outcropping bedrock ranges between 0.250 and 0.275 g (cf. Fig. 1).

322

Table 2. Geotechnical properties of Benevento subsoil. Soil type Man-made ground (R) Debris colluvium (Dt) Recent alluvium (ALG) Terraced alluvium (GS) Fine fluvio lacustrine (FLF) Coarse fluvio lacustrine (FLG) Weathered Rissian conglomerate (CR) Cemented Shallow Pliocenic clay (AGA) Deep

(kN/m3) 17 18-19 18-19 19-21 19 20 20 23 21.3 22.5

VS (m/s) 130 – 300 250 -330 200 – 500 630 -750 700 300 – 500 600 – 800 800 – 900 450 – 600 800

D0 (%) 5 3 2 1 3 1 1 0.5 3 2

l

v

(%) 0.001 0.0029 0.001 0.002 0.005 0.002 0.0015 0.02 0.009 0.009

(%) 0.01 0.0371 0.01 0.02 0.05 0.02 0.042 0.2 0.103 0.106

The maximum historical event that struck the city was the 1688 Sannio earthquake with I max=XI MCS, moment magnitude MW= 6.72 and epicenter located around 40 km N of Benevento. This earthquake, featured by a return period Tr 380 years, was adopted as the reference input motion for dynamic seismic response analyses. The ground motion at rock outcrop was simulated using a hybrid statisticaldeterministic approach, producing three sets of 150 synthetic seismograms, one for each motion component (NS, EW and UD). The average response spectrum of the NS components of the full dataset is plotted in Fig. 4, together with the standard deviation (red lines). An appropriate ‘cluster analysis’ (Aliperti, 2006) allowed to reduce this numerous dataset to 8 most representative accelerograms, the average spectrum of which ( ) is also plotted in the same figure (blue lines). Note that the higher the number of signals, the lower the standard deviation. 2.00 1.75 1.50 8 signals; avg. ±

1.25

150 signals - average 8 signals - average

1.00 150 signals; avg. ± 0.75 0.50 0.25 0.00 0

0.5

1

1.5

2

Period, T(sec)

Figure 4. Average response spectra and confidence intervals for the accelerograms simulating the 1688 Sannio earthquake (after Aliperti, 2006). Seismic response analysis The 3D geostatistical subsoil model in Fig. 3 was used for a preliminary seismic zonation of Benevento, based only on site classification criteria (Grade-2 according to ISSMGE-TC4, 1999). As suggested by Eurocode 8 (EN 1998-1 2003), each surface point was associated to a site-specific acceleration spectrum, consistent with the equivalent shear wave velocity VS,30 , this latter computed along the vertical profiles of the geostatistical model. Indeed, this procedure lead to a quite rough microzonation map, consisting in an almost homogeneous distribution of site-specific seismic actions throughout the whole urban area (Santucci de Magistris et al., 2004; Penna, 2005). As of matter of

323

fact, the EC8 ground types do not cover adequately and uniquely all soil conditions which are often encountered in practice, in terms of both typical soil layering and definition of the classification parameter, VS,30. Following the suggestions by Bouckovalas et al. (2006) addressing some possible modifications of the EC8 provisions (Fig. 5a), an upgraded zonation of the Benevento territory resulted as mapped in Fig. 5b. a)

b)

A1 A2 B

C D E

Figure 5. Definition of ground types according to Bouckovalas et al (2006) (a) and site classification map of Benevento (b). A Grade 3 zonation (ISSMGE-TC4, 1999) was finally carried out, by dynamic response analyses with reference to the deterministic seismic scenario above defined. 1D linear equivalent analyses were carried out using EERA code (Bardet et al., 2000) on 28 sites widespread all around the city (see Fig. 6). 16 of them corresponded to the locations of DH VS measurements (red flags in Fig. 6), while in the remaining sites (green triangles) the VS profile was obtained from the geostatistical model. Sitespecific response spectra were obtained by averaging the results of seismic response analyses with the clusterized accelerometric dataset as input motion signals. Finally, seven homogeneous zones were identified by grouping together the sites with similar average spectra. Zone 1 is characterized by relatively low amplification in the whole range of periods, due to the limited seismic impedance ratio. In Zone 1 *, higher peak spectral accelerations occur for T = 0.25 0.4 s, since the deformable soil layers are deeper than in zone 1. Moderate spectral amplifications over a wide range of periods are computed in Zone 2, where deformable soils are presents at higher depth, and a medium impedance ratio can be evaluated along the shear wave velocity profiles. Zone 3 is characterized by some significant spectral amplification for periods spanning between 0.2 and 0.6 s, while the VS profiles indicate large impedance ratios. In Zone 4, where deeper deformable formations and large impedance ratios are detected, the spectral acceleration reaches even 2g for periods T= 0.2 0.3 s. A quite large spectral amplification, concentrated in a narrow frequency range characterizes the spectral response of Zone 5, due to the shallow bedrock and large seismic impedance ratio. A similar situation is finally found in Zone 6, where the highest spectral accelerations (up to 2.5g!) have been computed for a period of 0.2s. By comparing Figs. 5 and 6, it appears that some important details in the zonation were not appreciated even using an improved Grade 2 approach. This could be due to the presence of the stiff levels somewhere overlying the Pliocenic clay formation assumed as bedrock. Its is worth nothing that the Grade 3 zonation map resulted quite compatible with the indications of standard spectral ratios of micro-earthquakes and ambient noise recordings (Improta et al., 2005) and the site intensity distribution observed after 1688 Sannio earthquakes (Castenetto and Romeo, 1992).

324

3.00

Zone 1

2.50

Ponte Galanti Asse S10 Asse S3 Viad. SS3 Ponte 2 Input motion

2.00 1.50 1.00 0.50 0.00 0

0.5

1 Period, T (sec)

1.5

2

3.00

Zone 1*

2.50 2.00

Rufina Asse S1 Input motion

1.50 1.00 0.50 0.00 0 3.00

0.5

1 Period, T (sec)

1.5

2

3.00

Zone 2

2.50

Zone 3

2.50

2.00

2.00

Avellola AT1 Avellola AT3 AvellolaAT2 Input motion

1.50

IND_SOND 113 IND_SOND 190 IND_SOND 260-261 Input motion

1.50

1.00

1.00

0.50

0.50

0.00

0.00 0

0.5

1 Period, T (sec)

1.5

2

0

3.00

0.5

1 Period, T (sec)

1.5

2

3.00

Zone 4

2.50

Zone 5

2.50 Avellola AT6 Viad. SS11 Avellola AT8 Viad. SS9 Avellola AT4 Input motion

2.00 1.50 1.00

2.00 IND_SOND 243 IND_SOND 253 IND_SOND 250 Input motion

1.50 1.00

0.50

0.50

0.00

0.00 0

0.5

1 Period, T (sec)

1.5

2

0

0.5

1 Period, T (sec)

3.00

1.5

2

Zone 6

2.50 2.00

Guerrazzi IND_SOND 8 IND_SOND 182 IND_SOND 16 Input motion

1.50 1.00 0.50 0.00 0

0.5

1 Period, T (sec)

1.5

2

Figure 6. Grade-3 seismic zonation of Benevento: map and site response spectra grouped for different zones.

325

THE CASE HISTORY OF ORVIETO

Stat e Um road b ro -C a se n tin e se

CHP

A'

ero

gu ll

y

S7

m s.l.m.

A

S

an

Z

Fosso C avaro

ne

San Benedetto gully

Site description and subsoil model The upper part of the Orvieto hill is formed by a slab of lithic tuff and weakly cemented pozzolana delimited by subvertical cliffs up to 60 m high. The slab is approximately elliptical in shape, with maximum length and width of 1500 m and 700 m in the WE and NS directions, respectively (Fig. 7). The lower part rests on a truncated cone carved in an overconsolidated clay formation, with a natural slope about 100 m high and an average inclination of 18 , overlying the cenozoic bedrock located at a minimum depth of 200 m. A 5 to 15 m thick succession of weakly cemented silts and dense sands/gravels (Albornoz formation) is interposed between the slab and the clay deposit. The slow deformation and yielding of the clay slope produced by the erosion induces severe tensile stresses at the cliff foot, determining the fracturing of the slab margin which is subsequently involved in instability phenomena. Major instabilities consist of collapses of pozzolanic spurs and by sliding of large regularly jointed portions of the cliff. More frequently, block falls are produced by toppling, sliding and tensile failure along sub-horizontal planes of tuff and pozzolana slices (Manfredini et al., 1980; Tommasi et al., 1996). The interpretation of historical data and a comparative analysis of seismic and landslide catalogues reveal that a number of falls can be related to seismic events. However, damages to structures are the most frequent seismic effects reported in historical sources.

RU 7 Pozzo d i San Patrizi o San D omen ico

A

Porta Ro mana San Fran cesco ese

RU 1

C i ve

100 tta

gu lly

de ll a

Ab ba d ia

BR CATHED RAL

S ta Um te R bro oad -C as ent in

POZZO DI S. PATRIZIO

200 Fo ss o

RU 10 Pal azzo de i Ca pitan i del Po pol o Porta Mag gio re

A' RU10

300

N

CHR

Clay

Pyroclastic materials

Albornoz

m s .l.m.

B

300

CHR RU1

Landslide debris

RU10

B'

RU7 S7 CHP

200 100 0

B

0

400

800

1200

1600

2000 m

Figure 7. Plan and geological sections of Orvieto hill (modified after Manfredini et al., 1980). Geotechnical data on pyroclastic materials were collected at the southern cliff: boreholes were drilled, static and dynamic properties of materials were determined on the recovered cores and shear wave velocities were measured by a cross-hole test (Tommasi and Ribacchi, 1998; Rotonda et al., 2002). The stiff in-situ clay formation is covered by a slide debris blanket of remoulded and oxidized clay. At the top, the stiff clay is softened and fissured, with closely spaced joints which progressively disappear with depth. The thickness of the different clay “layers” varies widely throughout the slope. The geotechnical parameters of pyroclastic and clayey materials are summarised in Tables 3-4. Table 3. Geotechnical properties of the Orvieto cliff materials Red Tuff Bulk dry density (Mg/m3) 1.18 laboratory 1150 Shear wave velocity (m/s) in situ Material

Competent pozzolana 1.15 620 550-650

Weak pozzolana 1.06 540 470

Table 4. Geotechnical properties of the Orvieto slope materials Clay material Bulk density (Mg/m3) Natural water content (%) Liquid limit Plasticity Index Shear wave velocity (m/s)

Stiff 2.14 20 50 21 560-590

Softened 2.01 23.5 50 20 300-450

Remoulded 2.0 27 53 24 200-210

326

The geotechnical model along the longitudinal (WE) and transversal (NS) directions is sketched in Fig. 8. From top to bottom, the slab consists of a 20 m weak pozzolana layer, resting on a 10 m layer of competent pozzolana, which in turn overlies a 30 m tuff layer. Beneath the slab, a 200 m thick stiff clayey substratum is superimposed on the bedrock. The different layers were characterised by constant values of bulk unit weight, , and shear wave velocity, VS, assigned on the basis of the above mentioned laboratory and cross-hole tests. It is worth noting that the product of Vs and values yields singular variations of impedance ratios between in-contact materials. More in detail, the clay/tuff impedance ratio is 1.1; conversely, impedance ratio values for tuff/pozzolana and competent/weak pozzolana are 1.24 and 1.27, respectively. This implies that the larger impedance contrast is not at the clay-slab interface, but between the volcanic materials within the slab. The G/G0- and D- curves for each material were taken from experimental data published in the literature on similar materials (Muzzi et al., 2001). 1400 m (E-W cross-section) 700 m (N-S cross-section) 60 m

300 m 100 m

100 m 3 Bedrock 25,5 kN/m , VS =2200 m/s More cemented pozzolana Pozzolana 13,3 kN/m3 , VS =660 m/s 13,3 kN/m 3 ,VS =530 m/s Tuff Stiff clay 3 12,5 kN/m3 ,VS =900 m/s 21,0 kN/m , VS =600 m/s

Figure 8. Geotechnical model for numerical analyses (modified after Muzzi et al., 2001). Reference input motion As documented by historical sources, Orvieto suffered the effects of near-field as well as far-field earthquakes. The strongest near-field event was the 1695 Bagnoregio earthquake (de = 10 km, I 0 = IX MCS) which also produced the highest site intensity at Orvieto (I S = VIII MCS). Most of the events were triggered by far-field sources (de = 70-80 km), located in the Umbria Apennines. An analysis of the historical seismicity of Orvieto and a seismotectonic study of the region were carried out for evaluating a credible rock outcropping motion (Muzzi et al., 2001). Two controlling near-field (NF) and far-field (FF) seismogenic zones were identified, which are characterised by de 20 km, Mmax = 5.9 (NF) and d e 70 km, Mmax = 6.7 (FF). Several rock outcropping accelerograms were selected from the world-wide databases matching the required Mmax and de values. Two records producing the maximum ground motion at the site in the frequency range of interest were chosen: for the near-field, the NS horizontal component recorded at Cascia station during the 1979 Valnerina earthquake (Italy), and the NS horizontal component at Butler Valley station during the 1992 Cape Mendocino earthquake (USA) for the far-field (Fig. 9). 0.2

Cascia NS (Valnerina 1979)

0.8

0.1

Cascia NS (near-field)

0

0.6

-0.1 near-field

0.15 g

-0.2 0 0.2

5

10

15

20

0.4 Butler V. NS (far-field)

Butler Valley NS (Cape Mendocino 1992) 0.067g

0.1

0.2

0 -0.1 fa r-field

-0.2 0

5

10

Time (s)

15

20

c)

0 0.01

0.1

1

10

Period (s)

Figure 9. Time histories and response spectra of the reference input motions.

327

Seismic response analysis One- and two-dimensional total stress analyses were respectively performed by means of SHAKE91 (Idriss and Sun, 1992) and QUAD4M (Hudson et al., 1994), along the two sections of Fig. 7. In 2-D analyses, to minimise the effects of wave reflection from the side boundaries, they were extended horizontally 900 m away from the slope foot. As expected, the results obtained on the transversal (NS) cross-section were more significantly affected by geometrical factors (Fig. 10). In fact, 2D analyses yielded higher peak accelerations with respect to 1D analyses (Fig. 10a), especially in the slab, with the maximum difference at the ground surface: peak acceleration raises from 0.12g to 0.21g and from 0.22g to 0.34g in the far and near-field, respectively. These results, influenced by the relatively small width of the slab in the NS direction, suggest that 1D analysis do not adequately account for the actual geometry. On the other hand, the peak shear strain profiles obtained at the hill center from 1D and 2D analyses are in fair agreement (Fig. 10b). Even though the peak shear strain is higher in the clay than in the slab (Fig. 10b), its magnitude does not exceed 0.04%. This value is smaller than the volumetric threshold for clays of similar plasticity (0.05%), thus suggesting that the increase of pore pressure related to shaking is negligible, and therefore justifying a total stress analysis. The horizontal profile of amax along the hill surface is plotted in Fig. 10c. It can be noted that the 2D geometry greatly affects the amax distribution, with maximum values generally at the center and at the edge of the slab. Under the near- field input motion, amax reaches 0.35g at the center and at the edge of the cliff, while a value of 0.2g is reached at the slab center in the far-field condition. At the cliff foot, amax abruptly decreases down to about 0.15g and 0.1g for near- and far-field motions respectively. An estimation of the overall topographic effects can be expressed by the 2D over 1D ratio of horizontal peak acceleration. The resulting topographic amplification factor, AT, is about 1.7 for both input motions. For the longitudinal (WE) section, characterised by a higher slab width, the comparison between 2D and 1D analyses at the hill center is very satisfactory in terms of amax and max profiles, while the amax profile computed along surface by 2D analyses show higher values close to the cliff edge (Muzzi et al., 2001). Estimated topographic amplification factors AT are lower than NS section, being 1.4 for both near-field and far-field conditions. 0

0.1

amax (g) 0.2 0.3

0.4 0

max (%) 0.02 0.04

0.06 0.4

0

Shake (NF) Quad4m (NF) Shake (FF) Quad4m (FF)

40 80

NF FF

0.3 0.2 0.1

(c)

120

0 0

100

200

300

400

500

600

Distance from the hill center

160 NS section hill center

200

(a) 240

NS Section

(b)

Figure 10. NS section: vertical profiles of peak acceleration (a) and shear strain (b) by 1D and 2D analyses at the hill center, and horizontal amax profiles computed by 2D analyses (c) (modified after Muzzi et al., 2001)..

328

THE CASE HISTORY OF NICASTRO Site description and subsoil model Nicastro cliff is located in Calabria (Southern Italy), one of the most active seismic regions of the Mediterranean area. The cliff is elongated in the NE-SW direction, about 60 m high; the width at the crest varies between 30 and 60 m moving from NE to SW, while it is about 200 m at the base (Fig. 11). The ridge flanks are quite irregular, with slope angles varying generally between 20° and 50°, but almost vertical in the northern side. Precious remains of an ancient Norman-Swabian castle lie on the crest. It was built in the period 1100-1500 and severely damaged during two strong-motion earthquakes in 1638 and 1783. In the last two centuries the castle was completely abandoned, and only recently has been subjected to consolidation works.

Figure 11. Plan view and sections of the Nicastro cliff. The cliff is constituted by metabasitic rocks and schists characterised by extremely variable degree of weathering and jointing, which strongly affect the mechanical behaviour of the cliff. From 1998 to 2003, several geotechnical investigations were carried out (Fig. 11), including 31 boreholes drilled to a maximum depth of about 40 m, 5 seismic refraction surveys, and 2 down-hole tests (Pagliaroli, 2006). From the sections 1-1 and 2-2 reported in Fig. 11, it can been seen that the cliff is locally covered by fill or debris, mainly gravel in a silty matrix, with a maximum thickness of 5 m. Despite the high degree of heterogeneity, it is possible to recognise some “macro-zones” from bottom to surface: namely, moderately weathered and jointed rock, moderately weathered and highly jointed rock, highly weathered and jointed rock. The physical and mechanical material properties of the above materials are reported in Table 5. The shear wave velocity VS was taken from down-hole tests results (Pagliaroli et al., 2006). Table 5. Geotechnical properties of the Nicastro subsoil. Material cover highly weathered and jointed rock moderately weathered, highly jointed rock moderately weathered and jointed rock bedrock

(kN/m3) 21 23 24 26 27

0.35 0.25 0.25 0.25 0.25

VS (m/s) 350 700 900-1000 1200 1500

D0 (%) 1.0 1.0 0.4 0.4 0.1

329

Reference input motion The historical seismicity of the area was characterised by setting up a site-specific seismic catalogue, based on the intensities either felt at Nicastro, or estimated through attenuation relationships (Pagliaroli, 2006). The most destructive event was the earthquake of March 27, 1638 (M=7.1), producing a site intensity IS = XI MCS. Two reference seismogenic zones were identified: the Catanzaro strait (near-field condition) characterised by a minimum source-to-site distance de = 10 km and maximum historically observed magnitude MS = 7.1; the Reggio Calabria fault system (far-field condition), characterised by de = 100 km and MS = 7.3. The peak ground acceleration resulted 0.35g and 0.04g, using the attenuation relationships by Ambraseys and Douglas (2003) and Ambraseys et al. (1996), for the near-field and far-field conditions respectively. The corresponding estimated mean response spectra and their standard deviations are shown in Fig. 12, together with those relevant to ten accelerograms (five for near- and five for far-field conditions), selected from European and world-wide records on rock outcrops, matching the required MS - D values and scaled to the expected amax. 1.75 1.5

near-field

1.25

average +

(attenuation model)

a)

0.25

b)

real accelograms real accelograms average

0.2

far-field

average (attenuation model)

1

0.15

0.75

0.1

0.5 0.25 0 0.01

0.05 average -

(attenuation model)

0.1 Period, T (s)

1

0 0.01

0.1 Period, T (s)

1

Figure 12. Real accelerograms selected for near-field (a) and far-field (b) conditions. Seismic response analyses The seismic response of the cliff to both reference earthquakes was simulated by the finite difference code FLAC (Itasca, 2002). The input motions were applied to the base of the model as vertically incident SV waves. Due to the high stiffness of the materials, only linear analyses were carried out. The dynamic response of the cliff was investigated for both sections in Fig. 11, characterised by different shape ratios H/L (where H is the average cliff height and L is the half-width at the base), respectively equal to 0.55 and 0.8 for cross-sections 1-1 and 2-2. The steeper section 2-2 is also characterised by stiffer material with respect to the other. The representative finite difference meshes shown in Fig. 13 were discretised in order to model frequencies up to 20 Hz, while radiation damping was simulated by viscous dashpots at the bottom and ‘free-field boundaries’ at the both vertical sides. The same figure shows the horizontal profiles of the peak surface acceleration, normalized to the corresponding value at the outcropping bedrock, with reference to all the selected input motions. For both sections, the amplification increases from the base to the crest of the cliff, with maximum values generally between 2 and 2.5. In order to evaluate the magnitude of the stratigraphic amplification, 1D analyses were carried out for the vertical soil columns corresponding to two representative nodes at the crest of the cliff (1A and 2A in Fig. 13). In the time domain, the topographic amplification factor, AT, was then evaluated as the 2D over 1D ratio of horizontal peak acceleration. The values calculated by averaging the amplification factors resulting for all the selected near-field and far-field accelerograms were about 1.5 for section 1-1 and 1.9 for the steeper section 2-2. Overall, such factors are 15% higher than those calculated for a simple homogeneous subsoil model (Pagliaroli et al., 2006).

330

3.5 3 2.5 2 1.5 1 0.5 0

3.5 3 2.5 2 1.5 1 0.5 0

a)

0

40

80

120

160 200 distance (m)

240

0

40

80

120 160 distance (m)

2A

1A

1-1 cross-section

near field far field

b)

200

2-2 cross-section

cover (VS = 350 m/s)

moderately weathered and jointed rock (VS= 1200 m/s)

highly weathered and jointed rock (VS= 700 m/s)

bedrock (V S= 1500 m/s)

moderately weathered, highly jointed rock (V S= 900-1000 m/s)

Figure 13. Normalised horizontal peak accelerations computed at the surface of cross-sections 11 (a) and 2-2 (b) for the near-field and far-field input motions. In the frequency domain, the topographic amplification was estimated by the so-called TAF function, expressed in terms of the ratio between 2D over 1D smoothed Fourier spectra at the crest nodes 1A and 2A (Fig. 14). It can be noted that the input motion does not affect the variation of TAF with frequency, for all the selected input motions. The maximum peak spectral amplification occurs between 3-5 Hz for both sections. This peak corresponds to the fundamental frequency of vibration (f2D) of the cliff, estimated by the Rayleigh method applied to homogeneous triangle-shaped asymmetric ridges (Paolucci, 2002): f 2D

f SV

VS 2L

(1)

where f SV is a factor usually variable between 0.6 and 1.0, according to geometric features of the cliff and the Poisson ratio. For the case of Nicastro, being 2L=180 m and VS=1200 m/s (representative value of the average stiffness of the cliff), equation (1) yields f 2D=4-6 Hz. Therefore, topographic amplification can be interpreted as a two-dimensional resonance phenomenon, as also shown by the simplified homogeneous model (Pagliaroli, 2006). The amplitude of spectral amplification at crest is about 1.8 and 2.5 for sections 1-1 and 2-2 respectively, thus increasing with H/L ratio analogously to amplification factor A T. 4 3.5 3 2.5 2 1.5 1 0.5 0

1-1 section

0.1

1A

near field far field mean

a) 1 Frequency, f (Hz)

10

4 3.5 3 2.5 2 1.5 1 0.5 0

2-2 section

2A

b) 0.1

1

10

Frequency, f (Hz)

Figure 14. Topographic aggravation factor (TAF) computed at the crest of cross-sections 1-1 (a) and 2-2 (b) for the near-field and far-field input motions.

331

THE CASE HISTORY OF GERACE Site description and susbsoil model Gerace is located at about 480m a.s.l. on a cliff oriented in direction NW-SE (Fig. 15), between the basins of two rivers, characterised by continuous gully erosion affecting the slope instability of the valley borders. The upper part of the cliff is formed by a soft rock slab, constituted by calcarenites and sandstones, overlying a deep formation of clayey marls, which in turn rests on a relatively thinner layer of interbedded sands and conglomerates. The slab floats on a deep layer varicoloured clay shales, with an estimated thickness of about 200m. The current geomorphological setting is the result of the intense erosion processes, producing the progressive removal of the soils overlying the clay formation (Monteleone, 1993).

Figure 15. Geological map of Gerace (modified after Monteleone, 1993). The longitudinal and transversal subsoil models for the seismic response analyses (see Figs. 17a,b) were characterised combining data collected from previous investigations and experimental results of on-purpose dynamic in situ (Down-Hole, MASW and SASW) and laboratory tests (RC and TS) (Costanzo, 2007). A synthetic picture of physical and mechanical material properties, used in the analyses, is reported in Table 6. Table 6. Geotechnical properties of Gerace subsoil. Material Fractured calcarenites Calcarenites Sandstones Clayey marls Conglomerates Varicolour clay (cliff) Varicolour clay (valley) Bedrock

(kN/m3)

VS (m/s)

16.69

400

16.69 16.94 18.05 19.51 21.00 21.00 20.87 20.87 22.00

750 400 800 750 658 725 454 643 1500

G (MPa)

K (MPa)

D0 (%)

' (°)

c' (kPa)

0.250

267

445

0.5

35

7

0.250 0.322 0.332 0.285

939 271 1155 1097 909 1104 430 863 4950

1565 671 3053 2186 4243 5151 1404 2816 10725

0.5 0.5 1.5 0.5 1.5 1.5 3.0 3.0

40 35 32 30

700 54 34 1000

25

0

25

0

0.400 0.361 0.300

r

v

Linear 0.100 0.140 0.300 0.164 0.164 0.103 0.103

0.048 0.060 0.160 0.068 0.040

Linear

332

Reference input motion In less than sixty days of 1783, since February, 5 to March, 28, Southern Calabria was struck by five strong earthquakes, with magnitudes between 5.9 and 6.9. The sequence caused the destruction of many towns (Carbone-Grio, 1887), as well as widespread environmental damage (Cotecchia et al., 1986), including significant ground movements and maybe slope failure in the historical town of Gerace (Romeo & Delfino, 1997). From the analysis of the site seismic history (Costanzo, 2007), the first event of the sequence could be assumed as the maximum historical earthquake. The reference seismic scenario was simulated through two methods: one based on macroseismic data (estimated epicentre and magnitude) and the other on seismogenic parameters (position and geometry of the possible sources). The acceleration and frequency content of the reference motion were evaluated by means of different attenuation relationships, yielding very close evaluations of the motion amplitude. The estimate of synthetic ground motion parameters allowed to sort out a set of compatible acceleration time histories (Costanzo, 2007). On the basis of the reference parameters, fixing suitably restricted ranges of magnitude and distance, compatible seismic records were selected as possible input motions from seismic databases. To select the most reliable accelerogram, the approach suggested by Bommer & Acevedo (2004) was followed, computing the amplitude scale factor, F sc, and the root mean square parameter, Drms, that supplies a quantitative evaluation of the similarity between the frequency contents. The most suitable seismic motion was the record of the Landers earthquake of 1992 (M=7.5, d e=21km) at the Morongo Valley Fire Stn 461 (Fig. 16), that showed the better spectral compatibility to the reference response spectrum, estimated through the relationship by Pugliese & Sabetta (1989). 1.00

0.25

(a)

0.15

P&S(89) mean

(b)

P&S(89) mean ± 0.80

0.05

0.60

-0.05

0.40

-0.15

0.20

-0.25

LANDERS 1992

0.00

0

10

20

30 40 time, t (s)

50

60

70

0

1

2 period, T (s)

3

4

Figure 16. Landers earthquake of 1992: acceleration time history (a) and spectral compatibility with the reference motion (b). Seismic response analysis The numerical 2D simulations of the seismic response of the cliff to the sequence have been carried out by FLAC 5.0 (Itasca, 2005), along the two orthogonal sections of the cliff. The bedrock depth was calibrated on the basis of far-field earthquake records taken by a mobile accelerometric station installed in the cathedral, at the cross-point between the two sections (Costanzo, 2007). The plots in Fig. 17a,b report with blue lines the horizontal peak acceleration profiles computed at surface, while the green lines refer to the values at the top of the varicolour clay formation; the coloured dots indicate the stratigraphic contacts between different soils. Along the longitudinal hill section (Fig. 17a), the amplitudes at the top of the clay are comparable or not significantly lower than the reference input motion (black horizontal lines). For the transversal section, instead, an apparent amplification is shown at the toes of the cliff (Fig. 17b), which was shown to affect the clay slope stability along this direction (Costanzo et al., 2007a,b).

333

0.60 0.50 0.40

(a)

surface top of varicolour clay bedrock

0.30 0.20 0.10 0.00 -1800 -1600 -1400 -1200 -1000 -800

50 0 m

-600

-400 -200 0 200 distance, d (m)

400

fractured calcarenites calcarenites

800

1000 1200 1400

0

C1

NW

600

(435ms.l.m.)

0.5

1km

S

SE

sandstones clayey marls

25 0 conglomerates & sands varicolour clay shales

0 0.60

(b)

surface 0.50

top of varicolour clay

0.40

bedrock

0.30 0.20 0.10 0.00 -1600 -1400 -1200 -1000 -800

-600 -400 -200

0

200

400

600

800 1000 1200 1400 1600

distance, d (m)

0

C2 500m

250

0.5

1 km

(435ms.l.m.)

SW

NE

0

Figure 17. Horizontal peak acceleration profiles along NW-SE (a) and SW-NE (b) sections. The surface motion amplitudes (blue lines) along the NW-SE section (Fig. 17a), characterised by a smoother topographic profile, show that the amplification is visibly affected by the impedance contrasts, and enhances across the more deformable sandstone layer. Also, the interaction between incident and diffracted waves induces oscillating amplification factors and out-of-phase motion. In the narrower NE-SW section, wave focalisation into the hilltop induces a pronounced topographic amplification, which results into A T of the order of 1.9 (Costanzo, 2007), i.e. a little higher than the maximum values evaluated for Orvieto (1.7) and equal to those relevant for Nicastro. As for the case of Nicastro cliff, the net topographic effect was evaluated by the TAF parameter, again defined as 2D over 1D Fourier spectra ratio (Fig. 18). At the hilltop (sites C1 and C2) and on the sandstones outcropping (site S), the TAF results sensibly higher than unity in the range between 2 and 8 Hz. At the crest of the longitudinal cliff section (C1), the TAF is about 2.5 around 5 Hz, which falls in the range of resonance periods of typical Gerace buildings. At the apex of the narrow section (C2), the TAF shows a peak of 3.8 for frequency of 3.5Hz and an apparent tendency to increase beyond

334

6Hz, due to numerical oscillations of low values of 2D and 1D Fourier amplitudes in the high frequency range. 5 C1

4

C2 3

S

2 1 0 0.1

1

10

frequency, f (Hz)

Figure 18. Topographic aggravation factory (TAF) computed at the crest of both sections and at the outcrop of sandstones along the section NW-SE. The vertical ground motion profiles and the plastic strains distribution, not reported here (see Costanzo et al., 2007a,b), indicate a tendency to slippage of the calcarenite layers along the contact with the underlying weak sandstone. Irrecoverable strains as high as 0.1% concentrate at the NW and SE toes of the hill and across the soft rock slab, evidencing an asynchronous motion of independent calcarenite blocks: this could be responsible for the opening of sub-vertical cracks across the cliff and of a rigid sliding of the calcarenite above the sand layer at the hilltop (see Costanzo et al., 2007a). Also, the final deformation pattern of the transversal section showed residual displacement values higher than 1.5 m in correspondence of the swelled varicolour clay, just beneath the contact with conglomerates: this deformation pattern seems to highlight the triggering of a rotational slope instability at the base of the hill. Therefore, the results of the non-linear 2D numerical analyses indicated not only a frequencydependent amplification of the ground motion at the cliff surface, but also a significant ground deformation under the maximum historical earthquake; both effects have to be accounted for the protection of Gerace and its cultural heritage from the seismic risk.

THE CASE HISTORY OF SAN GIULIANO DI PUGLIA Site description and subsoil model San Giuliano di Puglia is a little middle-age town in the Molise region, edified on a narrow hill with the ridge crest oriented SSE-NNW. The urban centre shows the typical characteristics of the Southern Apennines villages: buildings are mostly located on outstanding positions, and frequently connected in heterogeneous structural aggregates. The historical residential houses have typically rubble masonry walls, while in the newer part, built after the ‘40s, rubble and brick masonry are mostly present, with sometimes recent r.c. additions (Mucciarelli et al., 2003). Overall, both historical and newer buildings of San Giuliano di Puglia are usually 2-4 storeys high, resulting in fundamental periods of 0.1 to 0.5s (Baranello et al., 2003; Vona, 2007). As a consequence of the historical development of the town, the older part was edified on a soft rock formation (Faeto flysch) outcropping in the southern part of the town (Fig. 20a); in the progressive extension towards North, the newer buildings were raised up along the ridge, on a thick fine-grained soil (Toppo Capuana marly clays). Until 2002, the town was classified as not hazardous by the Seismic Italian Code. On October 31st, 2002, a moderate earthquake of local magnitude ML=5.6 (CPTI Working Group, 2004) shook the area. The epicentral distance (D) of San Giuliano di Puglia was only about 5 km (Pacor, 2007) and a degree of damage observed was VIII-IX MCS, while it never resulted larger than VII in the other towns close to the epicenter (Stucchi et al., 2007). Detailed surveys on the buildings (Baranello et al., 2003; Dolce

335

et al., 2004) indicated that the damage distribution was strongly non-uniform, with resulting VII-VIII EMS + in the recent part of the town, and VI in the ancient zone (see Fig. 20b). Such evidence suggested that differential site amplification significantly affected the seismic response in the town; the hypothesis was confirmed by the observation of the aftershock sequence recorded by two mobile accelerometric stations, located on clay and flysch outcrops, shown in Fig. 20c (Puglia et al., 2007). For the seismic microzonation of the town, the Department of Civil Protection (DPC) of the Italian Government committed a comprehensive study on the subsoil properties (Baranello et al., 2003). The field investigation included a large number of boreholes, sampling pits, in situ piezometric measurements, CPT, cross-hole and down-hole tests. The laboratory experimental programme involved oedometer, triaxial and cyclic/dynamic torsional shear tests on undisturbed samples of finegrained soils, analyzed in detail by Silvestri et al. (2006a). Table 7 describes the geotechnical properties for the marly clay and flysch formations, as reported by Silvestri et al. (2006b). Table 7. Geotechnical properties of the San Giuliano di Puglia subsoil Toppo Capuana marly clays

Debris cover Tawny clay Grey clay 1 Grey clay 2 Grey clay 3 Grey clay 4 Grey clay 5 Grey clay 6 Faeto flysch

Depth, z [m] 0 - 2.4 2.4 - 7.3 7.3 - 15 15 - 30 30 - 60 60 - 120 120 - 240 > 240 -

[kN/m3] 19.6 21.15

21.2

22

VS [m/s] 122 250 339 364 391 421 454 483 1350

D0 [%] 3 2.3

2.5

0.5

0.493 0.489 0.485 0.483 0.481 0.479 0.477 0.475 0.392

VP [m/s] 1010 1700 1970 2000 2050 2100 2160 2210 3200

Although very accurate and detailed in terms of characterisation of lithological, physical and mechanical properties of the geomaterials, neither the above investigation nor the numerous surface geological surveys could assess for certain the actual geometry of the stratigraphic contact between flysch and marly clay. On purpose, in the framework of the INGV-S3 National Research Project, deep geophysical surveys were carried out: three geo-electric resistivity tomographies (see Silvestri et al., 2006b), a gravimetric (Palmieri et al., 2006) and two deep seismic surveys (under processing) seem to indicate a complex 3D flysch-clay interface, characterised by planar boundaries rather than a basin geometry, as formulated by the earlier studies (Baranello et al., 2003). To date, the subsoil model most effective in matching the results of geophysical surveys and reproducing the subsoil response recorded during the aftershocks (Puglia et al., 2007) is that reported in Fig. 20c, i.e. a wedge-shaped clay deposit, fully bounded by the soft rock and deep about 300m. Reference input motion The mainshock of the 2002 Molise seismic sequence was simulated by Franceschina et al. (2006) through the Hybrid Integral-Composite source model (HIC), considering the fault mechanism formulated by Basili and Vannoli (2005). Fig. 19a shows the acceleration time history of the horizontal component on rock outcrop resulting from the HIC simulation, after projection along the section in Fig. 20. A 7Hz low-pass filter was applied for consistency with the mesh discretisation used in the numerical analyses described below. A typical ‘engineering’ approach was also used to define the seismic input motion, by using the most significant aftershock (2002.XI.12, ML=5.2) recorded by the mobile station on the rock outcrop. The horizontal and vertical peak acceleration were scaled to the values predicted by regional attenuation laws, calibrated on the records of several temporary and permanent networks operating in the area affected by the seismic sequence (Luzi et al., 2006). Fig, 19b shows the time history of the scaled aftershock after projection along the section and 7Hz low-pass filtering. The comparison between the two input motions considered shows that the simulated mainshock has a peak amplitude of about +

Note that such values in EMS-98 (European Macroseismic Scale) are roughly equivalent to those corresponding in the MCS scale.

336

0.05g, against 0.03g of the scaled aftershock. On the other hand, the response spectra (Fig. 19c) show comparable frequency contents, with peak values around 5 Hz (0.2 s), and similar trends throughout the range of the resonant periods (0.1 to 0.5s) of most of the buildings.

Figure 19. Horizontal time histories of simulated mainshock (a), scaled aftershock (b) and relevant pseudo-acceleration response spectra (c).

Seismic response analyses Simulations of the seismic response of San Giuliano di Puglia during the mainshock were carried out along a representative longitudinal section traced along the hill (Fig. 20a); this direction approximately follows the alignment of the most severe damage distribution, the locations of the two accelerometer stations and those of main geotechnical investigations. 2D numerical analyses were executed by QUAD4M (Hudson et al., 1994) on the ‘wedge-shaped’ section drawn in Fig. 20c, geometrically defined with the criteria described in the paper by Puglia et al. (2007). The horizontal profiles of peak ground acceleration (plotted in Fig. 20c with red lines, right scale) reproduce the differences in amplitudes between the simulated mainshock and the scaled aftershock; instead, the amplification factors on the marly clay deposit are similar, varying between 1.5 and 3, with significant oscillations characterized by a ‘surface wavelength’ of about 300 m, i.e. equal to the depth of the clay deposit. It must be noted that the predicted peak ground acceleration cannot directly be referred to the observed damage distribution, since they pertain to frequencies higher than those typical of the existing structures. Therefore, to assess the damage distribution occurred in the actual earthquake, the output of the analyses were processed in terms of Housner intensity (SI), i.e. the integral of the pseudovelocity response spectrum, PSV(T), calculated in the range of periods of San Giuliano di Puglia buildings (0.1 to 0.5 s): SI

0. 5 0 .1

PSV , T dT

(2)

and assuming a structural damping =5%. The horizontal profile of this integral ground motion parameter is also plotted in Fig. 20c with black lines (left scale). Note that the ratio between the SI values pertaining to the two ground motions, as well as their trends with local oscillations, are much similar to those observed for the peak ground acceleration. In such a way, the predicted ground motion profiles can be reliably compared to the observed degree of damage, this latter expressed in terms of ‘standardized damage’, as shown in Fig. 20a (Dolce et al., 2004). This index assumes values between 0 (no damage), and 1 (building collapse), and it does not depend on the vulnerability class of the building; therefore, it can be viewed as a significant variable

337

to empirically measure the local amplification effects†. A closer analysis of the variation of such parameter at the building size scale appears promising to reveal the likely correlation existing between the predicted ground motion and the observed damage.

Figure 20. Geological map (from Puglia et al., 2007) and standardized damage of buildings in San Giuliano di Puglia (from Dolce et al., 2004) (a); average degree of damage in terms of EMS (from Dolce et al., 2004) (b); horizontal profiles of peak ground acceleration and Housner intensity for the ‘wedge-shaped’ subsoil model subjected to two input motions (c).

CONCLUSIONS In all the above examples, the expected seismic response of urban areas under maximum historical earthquakes was accurately analysed with comparable methodologies, although with different specific addresses. In fact, the studies were more or less oriented to prediction or assessment, depending on the availability of data on the historical damages. For each case study, Table 8 summarizes the main aspects of input motion and subsoil modelling which significantly influenced the seismic response analyses.

†

The overall intensity in terms of EMS (Fig. 20b) was instead obtained through the comparison between average damage index, as calculated by Damage Probability Matrices (DPM) referred to the vulnerability class of the buildings, and observed for specific zones (Dolce et al., 2004).

338

Table 8. Main aspects of the analysis of the selected case studies. Input motion Case study

amax (hazard map)

amax (this study)

Benevento

0.250 – 0.275

0.335±0.129

Orvieto

0.125 – 0.150

Nicastro

0.250 – 0.275

Gerace

0.200 – 0.225

0.200

S. Giuliano di Puglia

0.200 – 0.225

0.03-0.05

0.150 (nf) 0.067 (ff) 0.350 (nf) 0.040 (ff)

Bedrock

Soil layering

Topography

Assumed at variable depth Assumed at fixed depth Assumed at fixed depth Back-analysed (far field records) Back-analysed Under deeper investigation

Complex with some impedance inversions Regular with constant impedance Variable with weathering at surface Regular with some impedance inversions

Gently sloping hill

Continuous variation with depth

Gently sloping hill

Plateau Peak Plateau

In most cases (except for San Giuliano), no representative seismic signals were already available for the definition of suitable reference rock outcropping motions: therefore, these were defined on the basis of detailed studies on the historical seismicity and seismotectonic outlines of the regions. Table 8 compares the peak ground accelerations predicted by the official hazard map by MPS Working Group (2004) to those used in this study. Actually, for Nicastro and Benevento, the ground motion amplitudes considered herein overestimate the specified values. For San Giuliano, the values of amax assumed in the analysis significantly underestimate that prescribed by the hazard map, since the recent earthquake considered is not the maximum historical event. Anyway, in most cases the results presented herein might be taken as references for seismic microzonation maps for urban planning. In such perspective, when the available subsoil data are numerous, like in the case of Benevento, the use of GIS maybe very promising. The geo-morphological setting and the susbsoil layering were sometimes quite complex; thus, their study resulted of critical importance for the subsoil modelling. The bedrock depth and geometry were in most cases unknown; their definition strongly influences the low-frequency response and the variation of amplification at surface, like verified for San Giuliano (see Puglia et al., 2007). In such cases, deep geophysical profiling can be of primary utility; also far-field or weak motion records can help to back-figure the bedrock depth, as shown by Costanzo (2007) and Puglia et al. (2007) for the two cases of Gerace and San Giuliano, respectively. The degree of accuracy in the description of soil layering must be adequate to the frequency content of the input motion, especially when sharp variations of seismic impedance are present. The response to near-field earthquakes, with significant energy content at high frequencies, can be very sensitive to the stratigraphic details, such in most of the cases presented. The example of Benevento shows that unusually excessive spectral accelerations at high frequencies may be induced by the use of synthetic accelerograms as input motion; as a matter of fact, in the case of San Giuliano, the simulated input motion was preliminarly low-pass filtered, to be comparable to that of a recorded signal. The site amplification was seen to be strongly affected by surface topography for the towns lying on flat (Orvieto, Gerace) to steep (Nicastro) cliffs. It must be noted that, in such cases, an irregular morphology is often associated with a stratigraphic heterogeneity, because erosion processes that can affect different materials with variable intensity. Therefore, the interplay between geometry and layering can be misunderstood, if simple homogeneous linear analysis models are used; only 2D numerical FEM or FDM simulations, either with linear equivalent or non-linear analysis, are capable to yield reliable predictions or assessment of dynamic ground motion and deformation. Summing up, in order to preserve the high value of the cultural heritage from the damage induced by strong-motion earthquakes, the strongest efforts are needed in terms of definitions of seismic actions, subsoil characterisation and methods of analysis.

339

ACKNOWLEDGMENTS The town municipalities of Benevento, Orvieto, Nicastro, Gerace and San Giuliano are kindly acknowledged for providing the available data on the subsoil, even when not sponsoring the research projects. The authors are indebted to the Department of Civil Protection (DPC) of the Italian Government and the National Institute of Geophysics and Vulcanology (INGV) for funding the research works on Benevento, Gerace and San Giuliano di Puglia. The coordinators of the INGV-S3 Research Project, Dr. Francesca Pacor and Prof. M. Mucciarelli, are warmly acknowledged for their valuable scientific and administrative support. The oldest Authors are indebted to their former supervisors, Profs. B. D’Elia, R. Ribacchi and F. Vinale, for having introduced them to these fascinating research studies.

REFERENCES (pr)EN 1998-1 (2003). “Eurocode 8: Design of structures for earthquake resistance – Part 1: General rules, seismic actions and rules for buildings”, CEN European Committee for Standardization, Bruxelles, Belgium. Aliperti D. (2006). “Un contributo alla zonazione sismica del comune di Benevento”, Undergraduate thesis, Università di Napoli Federico II (in Italian). Ambraseys N. N., Simpson K. A., and Bommer J. J. (1996) “Prediction of horizontal response spectra in Europe” Earthquake Engineering Structural Dynamics, 25:371-400. Ambraseys N. N., and Douglas J. (2003) “Near-field horizontal and vertical earthquake ground motions” Soil Dynamics and Earthquake Engineering, 23:1-18. Baranello S., Bernabini M., Dolce M., Pappone G., Rosskopf C., Sanò T., Cara P.L., De Nardis R., Di Pasquale G., Goretti A., Gorini A., Lembo P., Marcucci S., Marsan P., Martini M.G. and Naso G. (2003) “Rapporto finale sulla Microzonazione Sismica del centro abitato di San Giuliano di Puglia” Dipartimento di Protezione Civile, Roma (in Italian). Bardet J.P., Ichii K. and Lin C.H. (2000) “EERA a Computer Program for Equivalent-linear Earthquake site Response Analyses of Layered Soil Deposits”, Univ. of Southern California, Dep. of Civil Eng. Basili R. and Vannoli P. (2005) “Source ITGG052 San Giuliano di Puglia”. In: DISS Working Group, Database of Individual Seismogenic Sources (DISS), Version 3.0.1, “A compilation of potential sources for earthquakes larger than M 5.5 in Italy and surrounding areas”, Istituto Nazionale di Geofisica e Vulcanologia (INGV), internet: http://www.ingv.it/DISS/ . Bommer J. J. and Acevedo A. B. (2004) “The use of real earthquake accelerograms as input to dynamic analysis,” Journal of Earthquake Engineering, 8, special issue 1: 43-91. Bouckovalas G.D., Papadimitriou A.G. and Karamitros D. (2006). “Compatibility of EC-8 Ground Types and Site Effects with 1-D Wave Propagation Theory”, ISSMGE ETC-12 workshop, NTUA Athens, Greece, 53-67, Bouckovalas Ed. Carbone Grio D. (1884) “Terremoti di Calabria e Sicilia nel secolo XVIII,” Barbaro editore, Oppido Mamertina (RC) (in Italian).Castenetto R. and Romeo R. (1992). “II terremoto del Sannio del 5 Giugno 1688: Analisi del danneggiamento subito dalla città di Benevento”, Proc. XI C.G.N.G.T.S., Roma, 21–38 (in italian). Costanzo A. (2007) “Analisi di fenomeni deformativi di pendii e rilievi in condizioni sismiche : il caso di Gerace”, Ph.D. thesis, Università della Calabria (in Italian). Costanzo A., D’Onofrio A. and Silvestri F. (2007a) “Numerical simulations of the ground deformation recorded in the historical town of Gerace during the seismic events in Calabria (1783)”, IV International Conference on Earthquake Geotechnical Engineering, Thessaloniki, Greece, paper no. 1613. Costanzo A., D’Onofrio A., Silvestri F. (2007b) “Analisi dei danni registrati nel borgo di Gerace durante gli eventi sismici della Calabria del 1783”, XII National Conference “L’Ingegneria Sismica in Italia”, Pisa (in italian). ANIDIS, Roma.

340

Cotecchia V., Guerricchio A. and Melidoro G. (1986) “The Geomorphogenetic Crisis Triggered by the 1783 Earthquake in Calabria (Southern Italy),” Symposium: Engineering Geology Problems in Seismic Areas, 6, Bari. CPTI Working Group (2004) “Catalogo Parametrico dei Terremoti Italiani, versione 2004 (CPTI04)”, INGV, Bologna, internet: http://emidius.mi.ingv.it/CPTI04/ (in Italian). Dolce M., Masi A., Samela C., Santarsiero G., Vona M., Zuccaro G., Cacace F. and Papa F. (2004) “Esame delle caratteristiche tipologiche e del danneggiamento del patrimonio edilizio di San Giuliano di Puglia”, XI National Conference “L’ingegneria Sismica in Italia”, Genova (in Italian). ANIDIS, Roma. Fenelli G.B., Pellegrino A. and Picarelli L. (1998) “Stability problems of old towns built on relict plateaux on clay deposits”, Proc. Arrigo Croce Memorial Symp. on Geotechnical Engineering for the Preservation of Monuments and Historic Sites, Napoli, 163-173, Rotterdam: Balkema. Franceschina G., Pacor F., Cultrera G., Emolo A. and Gallovic F. (2006) “Modelling directivity effects of the October 31, 2002 (MW =5.8), Molise, Southern Italy, earthquake” First European Conference on Earthquake Engineering and Seismology (1st ECEES), Geneva, Switzerland, paper no. 1424. Hudson M., Idriss I. M. and Beikae M. (1994). “QUAD4M - A computer program to evaluate the seismic response of soil structures using finite element procedures and incorporating a compliant base”, University of California, Davis. Idriss J. and Sun J.I. (1992). “SHAKE91- A computer program for conducting equivalent linear seismic response analyses of horizontally layered soils deposits”, University of California, Davis. Improta L., Di Giulio G. and Iannaccone G. (2005). “Variations of local seismic response in Benevento (Southern Italy) using earthquakes and ambient noise recordings”, Journal of Seismology, 9:191-210. ISSMGE-TC4 (1999) “Manual for zonation on seismic geotechnical hazards”, The Japanese Society of Soil Mechanics and Foundation Engineering. Itasca (2002) “FLAC – Fast Lagrangian Analysis of Continua – Version 4.0” User’s Guide, Itasca Consulting Group, Minneapolis, USA. Itasca Consulting Group (2005) ”Flac 5.0 USER’S Manual,” Minneapolis, Minnesota, USA. Kallou P.V., Gazetas G., and Psarropoulos P.N. (2001) “A case history on soil and topographic effects in the 7th September 1999 Athens earthquake” Proc. of 4th Int. Conf. on Recent Advances in Geotechnical Earthquake Engineering and Soil Dynamics, San Diego, California. Luzi L., Morasca P., Zolezzi F., Bindi D., Pacor F., Spallarossa D. and Franceschina G. (2006) “Ground motion models for Molise region (Southern Italy)” First European Conference on Earthquake Engineering and Seismology (1st ECEES), Geneva, Switzerland, paper no. 938. Manfredini M., Martinetti S., Ribacchi R., Sciotti M. (1980) “Problemi di stabilità della rupe di Orvieto” XXI Convegno Nazionale di Geotecnica, Firenze, 1, 231-246 (in Italian). Marcellini A., Bard P.-Y., Iannaccone G., Meneroud J.P., Mouroux P., Romeo R., Silvestri F., Duval A., Martin C., and Tento A.. (1995) “The Benevento seismic risk project: The microzonation” Proc. of the V Int. Conf. on Seismic Zonation, Nice, France, I, pp.810–814. Monteleone S. (1993) “Deformazioni gravitative profonde e grandi frane indotte dalle argille varicolori scagliose a Gerace e liquefazione nei terreni del territorio di Locri,” D.Sc. thesis, Università della Calabria (in Italian). MPS Working Group (2004). “Redazione della mappa di pericolosità sismica prevista dall'Ordinanza PCM 3274 del 20 marzo 2003”, Final Report for DPC, INGV, Milano-Roma, http://zonesismiche.mi.ingv.it/ (in Italian). Mucciarelli M., Masi A., Vona M., Gallipoli M. R., Harabaglia P. and Caputo R. (2003) “Quick survey of the possible causes of damage enhancement observed in San Giuliano after the 2002 Molise, Italy seismic sequences”, Journal of Earthquake Engineering, Vol. 7, No. 4, pp. 599-614. Muzzi M., Lanzo G., Tommasi P. and Ribacchi R. (2001). “Analisi numerica della risposta sismica del colle di Orvieto”, X National Conference “L’ingegneria Sismica in Italia”, Potenza (in Italian). ANIDIS, Roma. Pacor F. (2007) Personal communication. Palmieri F., Marello L. and Priolo E. (2006) “Rilievo gravimetrico di dettaglio nell’area di San Giuliano di Puglia (CB)”, INOGS Research Report, S3 Project, internet: http://esse3.mi.ingv.it/ (in italian).

341

Pagliaroli A. (2006) “Studio numerico e sperimentale dei fenomeni di amplificazione sismica locale di rilievi isolati” Ph.D. Thesis, Università di Roma “La Sapienza”, Roma (in italian). Pagliaroli A., Lanzo G., D’Elia B. (2006) “Numerical study of the topography effects at the Nicastro (Southern Italy) cliff and comparison with EC8 recommendations”, Proc. ETC-12 Workshop, NTUA, Athens, Greece, January 20-21, paper #9. Paolucci R., (2002) “Amplification of earthquake ground motion by steep topographic irregularities” Earthquake Engineering and Structural Dynamics, Vol. 31, pp. 1831-1853. Penna A. (2005) “Geostatistical application for the local seismic response: seismic zonation of the town of Benevento”. Ph.D Thesis, University of Naples Federico II (in Italian). Penna A., d’Onofrio A. and Santucci de Magistris F. (2007) “Geostatistical analysis in the evaluation of local seismic response” Proc. of the XIV European Conf. in Soil Mech. and Geotech. Eng., Madrid (in print). Puglia R., Lanzo G., Pagliaroli A., Sica S. and Silvestri F. (2007) “Ground motion amplification in San Giuliano di Puglia (Southern Italy) during the 2002 Molise earthquake”, IV International Conference on Earthquake Geotechnical Engineering, Thessaloniki, Greece, paper no. 1611. Pugliese A. and Sabetta F. (1989) “Stima di spettri di risposta da registrazioni di forti terremoti italiani”, Ingegneria Sismica, 6:2 (in Italian). Romeo R. and Delfino L. (1997) “CEDIT: Catalogo Nazionale degli effetti deformativi del suolo indotti da forti terremoti,” Technical report SSN/RT/97/04, Department of Civil Protection, Rome (in Italian) Rotonda T., Tommasi P. and Ribacchi R. (2002) “Physical and mechanical characterization of the soft pyroclastic rocks forming the Orvieto cliff”, Int. Work. on Mechanical Characterization of Volcanic Rocks, Madeira. Santucci de Magistris F., d’Onofrio A. and Sica S. (2004). “A step into the definition of the seismic risk for the city of Benevento (Italy)”. Proc. of the Fifth Inter. Conference on Case Histories in Geotechnical Engineering, New York City, N.Y. Silvestri F., Vitone C., d’Onofrio A., Cotecchia F., Puglia R. and Santucci de Magistris F. (2006a) “The influence of meso-structure on the mechanical behaviour of a marly clay from low to high strains”, Symposium to celebrate Prof. Tatsuoka's 60th birthday, Roma, Italy. Silvestri F., D’Onofrio A., Guerricchio A., Lanzo G., Pagliaroli A., Puglia R., Santucci de Magistris F., Sica S., Eva C., Ferretti G. and Di Capua G. (2006b) “Modelli geotecnici 1D e/o 2D per i comuni di San Giuliano di Puglia, Bonefro, Ripabottoni, Colletorto e Santa Croce di Magliano”, Deliverable D8, S3 Project, internet: http://esse3.mi.ingv.it/S3_doc.html (in Italian). Stucchi et al. (2007) “DBMI04, il database delle osservazioni macrosismiche dei terremoti italiani utilizzate per la compilazione del catalogo parametrico CPTI04”, Quaderni di Geofisica, INGV, internet: http://emidius.mi.ingv.it/DBMI04/ (in Italian). Tommasi P., Ribacchi R., Sciotti M. (1996). Geotechnical aspects in the preservation of the historical town of Orvieto. Proc. Arrigo Croce Memorial Symposium on Geotechnical Engineering for the Preservation of Monuments and Historic Sites, Napoli, 849-858. Balkema, Rotterdam. Tommasi P. and Ribacchi R. (1998). “Mechanical behaviour of the Orvieto tuff”, Proc. Int. Conf. “The Geotechnics of Hard Soils-Soft Rocks”, Napoli, 901-909. Balkema, Rotterdam. Vona M. (2007) Personal communication.

342

4 th International Conference on Earthquake Geotechnical Engineering June 25-28, 2007 Paper No. W2-1001

A RESEARCH ON THE PERFORMANCE OF THE CONCRETE STRUCTURES AND THE REASONS OF THEIR FAILURE IN BAM EARTHQUAKE AND DESIGN SUGGESTIONS Roozbeh Ettehad ¹, Hamed Jahangiri ²

ABSTRACT The Bam earth quake occurred on December 25th at 05: 26: 26 according to the local time (at 01:26: 26 according to the international time GMT) in the historical city of Bam in the south east Iran (south east Kerman).This earth quake is considered to be the third destroying earth quake of the country after the Tabas and Manjil earth quakes .Most parts of the city had been destroyed completely and rest parts had been damaged between 30% to 70%. The earthquake was caused due to the move of the Bam…..which is located between Baravat village and Bam city and has a north –south direction and a length of 15 km. Investigations on the land showed small scale cracks in the south and north of the ….. This corresponded the rightward movement of it .This ….. Were on e of the few…… of the country which had not experienced earth quakes at all. Site inspections from the suspected structures in Bam area show that considering the consumed material and current performing methods most of them were considerably suseptable against the earth quake. Concrete structures located in Bam in comparison with the steal constructions had a relatively acceptable performance because of rigid connections .In the following article along with investigation on the concrete structures behavior against the earthquakes, the reason of the mentioned failures, design and site work problems will be discussed as well. Keywords: concrete structure, Bam earth quake, design problems, site work problems and code regulations.

INTRODUCTION Bam area is located in the south east Iran which suffered a severe earth quake .No historical earthquake had been reported up to the December 25 th 2003 .The mentioned earth quake was reported in the early morning hours that most of the city residents were asleep and may be that can be a cause of high civilian death .Most parts of the city like Arg-e-Bam (a famous historical place) was destroyed completely and other parts the city were susepted between 30% to 70% . After the earth quake some pats of the city specially city bazzar were on fire ,however on unfortunately the fire engines were down due to the destruction of the fire station building .The greatness of the earth quake were 9 in mercalie’s scale . In the vertical direction to the fault fast reduction of the severe movement and greatness of the earth quake is noticeable in a way that all the destructions are limited to the urban parts of the Bam .researches done show that a considerable force is being imposed to the homogenous buildings which usually is more than the elastic limit .If the studied structure is unable to tolerate elastic deflections a significant damage will be mad.

B.Sc of Civil Engireening, Young Researchers Club, Islamic Azad University of Bonab. Add: East Azerbaijan Science and Technology Park, Golgasht Cross, Azady Ave, Tabriz, IRAN P.O.Box: 51745-153. Email: [email protected]

343

CONCRETE STRUCTURES Concrete structure in comparison with the other structures have got a higher flexibility .So they can resist more easily against the earthquakes(they need less resistance).The ability mentioned for concrete structures would be actual when the site work regulations of this structure system are obeyed correctly .The most important elements of the system that can resist against the lateral forces are rigid connections and concrete shear walls .If a frame has rigid connections ,it can bear lateral forces by a bending reaction and transfer it to the earth. The concrete structures existing in Bam had rigid connections and their relatively good performance is an approval of it .The suseption which at times were caused in their structure was due to not obeying the design and site work regulations according to the code. Some of the problems are as follows: 1) Producing unqualified concrete 2 )un proper performance of concrete 3 )using simple bars instead of lined ones 4)using long bars 5)the high distance between struts 6)not having enough cover on the long bars 7) improper connection between beam and columns 8) performing a weak column against a strong beam

DESIGN PROBLEMS The most important problems are as follows: A) Wrong estimation of weather effects: 1) not considering the repetitions of freezing and melting periods 2) not considering the repetitions of getting wet and drying 3) not considering the fluctuations of temperature and damp and their effects on deflections made on concrete. B) Wrong estimations of environmental corrosion 1)not considering the unpleasant effects of the soil ,water and atmosphere carrying salts and corrosive gases 2) selecting improper construction material for example improper cement or bar 3) not predicting the proper cover on the bars in corrosive environment C) Problem in providing performance plans and documents 1) not defining the type of the consumed material in performance plans and opening the way to further performance problems 2)planning problems and drawing plans that does not correspond with the calculations 3) insufficient performance documents 4) not having harmony between the different parts of one plan 5) not controlling the plans before performance or inaccurate controlling D)not considering performance details 1)not having enough internal length different elements of a structure 2)not having enough struts ,connection at the end of the bars 3)not predicting distribution bars in the effect point of the concentrated loads 4)not predicting strut in the bending points of the tensile bars on the inner corners 5)cutting the attachments or big parts of a main bar in one section 6)not considering the curvature radius of thick bars

344

7)not paying attention to the location of bars around the openings in the structure and not having diagnol bar on the corners of the openings 8)not putting bars on the corners of the slabs that are on the soil and there are walls on them 9)the condition of the bars in the connections between beams and columns which prevents the concrete flow 10)not continuing the bars of the foundation under the columns in to the columns 11)placing too much bar in one section with low width and therefore not allowing the concrete to flow properly 12)not connecting and joining the surrounding walls to the frame 13)the connection screws were too close that in case of an earthquake ,they’re dangerous 14)locating the walls and partitions on one beam 15)not providing continius connection of the walls E)not considering the codes and regulations during preparing the performance plans 1) not putting cheek bar 2) not considering the minimum strut required 3) not continuing the minimum bar required onto the supports 4)not continuing the column struts in connection with beams and producing weak points during an earthquake 5) not continuing the negative moment bars required from the supports to the center of the beams 6)not defining vertical ,horizontal connectors in the masonary structures particularly in seismic areas 7) using simple bars as main bars and many other problems

PERFORMANCE PROBLEMS a)problem in framing 1)not considering the section dimensions and distraction from what was predicted 2)the frame wasn’t water proof that helped concrete lose water and have a low quality 3) no stability in the frame form specially the slide ones 4)no stability in the permanent supports b)bar placing problems 1)not considering type, diameter, length, form and the number of the bars 2)not considering the connection length and curvatures 3)not considering the sequence of placing the bars in the joints 4)not considering the performance details in the joints and connections 5)not considering bending the struts end and connecting the in concrete volume 6) not considering the location and length of the waiting bars 7)long strut distance 8) not considering the minimum connection for the column bars c) problems in pouring the concrete 1) using improper stone material 2) using improper water and cement to prepare concrete 3) not considering the optimum ratios of mixing in the concrete 4) using unknown complements specially using several complements together without considering the adaptability 5) not considering concrete homogenousity in transportation and using improper transportation methods 6) using improper methods and equipment to pour and place the concrete, not paying attention to the corners and around the bars 7) not caring or caring incompletely using ineffective equipment and methods 8) smoothing the concrete surface in a hurry before the main part of the initial shrinkage has settled

345

9) not paying attention to the weather conditions while pouring concrete 10) no homogenousity between the new and the old concrete d) problem in picking the frames 1)picking the frames before the proper time when the concrete hasn’t gained its complete strength 2)picking the frames by pressure and knocking the new concrete elements 3)picking all the frame and reinstalling the safety supports or not installing them in the middle of long beams 4)leaving wooden parts in the concrete e)not considering technical principles and regulations: 1)not preparing flow surfaces in the construction openings before restarting to pour the concrete 2)making the bottom part of the column or the wall with a height of 10cm using concrete or mason 3) doing construction work in improper weather conditions 4) not considering the sequence of the performance 5) discribtion and possible changes in plan without the designers permission 6) not considering the stability of the performed section under site work loading 7) not doing obvious things to prevent enviormental conditions effect 8) not considering the levelness of the ceiling concrete 9) not considering the continuity in different concrete parts 10)making defected flow surface (sloped , improper place , improper form) 11) using bars with different strength 12) not connecting the walls upper part in to the under part of the beam 13) not cleaning the frames before pouring the concrete 14)filling the walls made of concrete blocks by cracked blocks 15) not cleaning rusty bars and not cutting their connection with defected concrete 16)improper isolation of the parts related with soil 17) not considering the freezing and leaving openings which reduces the connection strength f) problem in quality control: 1) not having control or having weak control over the work quality of the workers 2) not having effective and proper internal controle by the builders not outer control by the owner 3) having improper or incorrect quality control methods and misleading the controller about the quality of the work done DESIGN ADVICES Obeying these simple cases has been suggested by Iranian earthquake code (2800): 1)the building plan must be plain and symmetric 2)the columns in different stories must be on eachother 3) the elements resistance against the lateral forces must be in a vertical plain 4) the correspondence between the gravity center and the rigidness center 5)balconies with a length lower than 1.5 m 6) not placing heavy equipment or facility on the balconies 7) reducing the gravity center of the building 8)using high strength structural material and low weighed unstructural material 9)increasing structure ductility 10) designing on the basis of the first plastic joint formation in the beams before columns

CONCLUSION

346

According to the existing codes, instructionsand issues and because most of the destructions were due to not obeying them in the building it is suggested to the following to improve the quality of the works done: a)educating all the project workers b)quality control for construction material 1)providing technical and professional labs to control the quality of the construction material 2)obligatory standard for all the material 3)proper keeping of the material up to the using time 4)using light and resistant material 5)using modern construction system c)control in all the constructing places 1)controlling all the phases by resident observer 2)controlling all the phases and preparing required reports 3) performing required test in different phases 4) controlling the builder groups in relation with required profession 5)control and correct use of the proper machinery d) correct maintenance 1)providing documents for all the buildings 2)professional annual control for all the buildings 3)performing the required repairs on time 4)recording the changes and repairs in the documents provided for the building

REFERENCES Iranian concrete code (ABA) Iranian earthquake code 2800 Instructions for national seismic improvement

347

4th International Conference on Earthquake Geotechnical Engineering June 25-28, 2007 Paper No. W2-1013

Underground monuments (Catacombs) in Alexandria, Egypt. Sayed HEMEDA1, Kyriazis PITILAKIS 2, Ioanna PAPAYIANNI 2, Stavros BANDIS 2, Mohamed GAMAL3 ABSTRACT The Catacombs in Alexandria, Egypt from the Greek-Roman era represent cultural heritage of outstanding universal values. They suffer weathering – aging as well as multiple geotechnical and earthquake problems. A pilot study has been carried out for on the Catacombs of Kom El-Shoqafa in order (a) to define the pathology and the causes (b) to study the seismic performance (c) to assess the global risk due to several factors and (d) to define the appropriate retrofitting techniques to be applied. In the paper a general outline of the various tests, surveys and analyses is presented, highlighting the most important issues related to the static and seismic stability of underground structures like Catacombs in particularly unfavorable geotechnical and environmental conditions. Keywords. Catacombs of Kom El-Shoqafa, Alexandria underground monuments, geotechnical problems, geotechnical investigations, microtremors, creep tests, stability and seismic analysis. INTRODUCTION Alexandria is located in eastern part of the Mediterranean Basin (Northern Egypt) and it is a place of great historical and religious interest. Numerous Catacombs and cemeteries for Greek-roman were erected in Greek-roman and Christian era have been found. They represent actually a large complex of an underground necropolis. The aim of the present study is the investigation and documentation of the existing stability conditions of the geological formations at a site of a catacomb in order to define the instability problems to interpret the pathology and to propose the best retrofitting procedure On 28 September 1900 the ground on the Hill of potsherds (Kom El- Shoqafa) spontaneously opened, and a donkey disappeared into the crevasse. The unfortunate beast had inadvertently discovered one of Alexandria’s most important archaeological sites, the principal hypogeum of a funerary complex dating from the end of the first century of the Christian era and still in use at the beginning of the fourth(Empereur,2003). It had been known for some time that this area held antique tombs, since the hill has being extensively quarried to provide building materials for a fast –expanding modern Alexandria. Much had already been destroyed, though certain archaeologists of the late 19th century had been able to record other tombs that were subsequently to disappear. These reports have descriptions and drawings, which show that the complex that can visit today was part of a vast necropolis, traces of which must still exist under the foundations of the neighboring buildings. The stability conditions of the historical monuments are of crucial interest, especially in regions like the Mediterranean Basin and particularly Alexandria, Egypt, where the seismotectonic and weathering regime is active and the geological structure is complex .Phenomena like settlement and slope movements as well as earthquakes and tectonic activity contribute to the damage of the historical buildings. The ground water activity is also an important factor, especially in cases underground monuments the environmental factor is also necessary to be taken in mind, when different protection measures are decided to apply. 1

Doctoral Candidate, Department of Civil Engineering, Aristotle University of Thessaloniki, Greece. E-mail : [email protected] 2 Professors, Department of Civil Engineering, Aristotle University of Thessaloniki, Greece. 3 Associate Professor, Geophysics Department, Cairo University, Egypt.

348

PROBLEM DESCRIPTION The site: The Catacombs of Kom El-Shoqafa lies in the district of Karmouz on the south-west of Alexandra, not far from the so-called Pompey, s pillar, on the south slope of the hill. The area was called Kom El- Shoqafa or a pile of shards. This catacomb lies at about 2-2.5 kilometersfrom the seashore, and it is higher in topography than Amod El- Sawari area, Fig (1). The structure: The catacombes under study was most likely initially a private tomb and later converted to a public cemetery. It consists of 3 levels cut into the rock, a staircase, a rotunda, the triclinium or banquette hall, a vestibule, an antechamber and the burial chamber with three recesses in, where in each recess there is a sarcophagus. The Catacombs also contains a large number of Luculi or grooves cut in the bedrock.

Figure 1. Location maps of Catacombs of kom El-Shoqafa.

The Entrance

Catacombs of kom El-shoqafa site.

349

El-Shatbi tombs site.

Moustafa kamil tombs site.

Amod El-Sawari site (library and temple). Figure . 2. Underground monuments (Catacombs) in Alexandria, (present condition). Pathology and causes : The Alexandria Catacombs show some clear indications of yielding and partial collapse at several locations, Weathering as indicated by in particular honeycomb ,stone surface scaling disintegration of construction material, intense rock meal damp surfaces in particular for semi-sheltered parts of the excavation, white salt efflorescence and yellowish brown iron staining can be noted at many parts. The structural damage is represented by ceiling cracking, stone surface decay and partial collapse of some parts of the ceilings and walls, rock exfoliation especially noted in the ceiling of the narrow corridors that are found at the deepest parts and mass wasting from its ceiling and walls of corridors. In conclusion the present state of conservation of the great Catacomb of Kom El-Shoqafa, the best known and most highly prized testimony of Alexandrine funerary architecture culture is now at the utmost limit of degradation.

350

The majority of structural damages have been caused from one or the combination of the following factors: Progressive weakening of rock material due to intrinsic sensitivity to weathering factors especially the underground water and salt weathering effect. Earthquake and other man made dynamic loading Permanent deformation of the rock mass. Natural wear of material. Construction history in the area

0 2 4 6 8 10 M

Figure 3.Plan and a typical cross section A-A/ of the Catacombs of kom El-shoqafa.

Figure 4 .Catacombs of kom El-shoqafa.present conditions.

351

GEOTECHNICAL CONDITIONS The Catacombs are carved in the oolitic sandy limestone (calcareous cemented sand); it is yellowish white massive, fine to medium grained cross-bedded sandstone cemented with calcareous cement. Intersected conjugated joints filled with very fine friable sand saturated with water in the lower parts. This unit is underlined by loose calcareous sandstone. It is brownish medium to fine grained calcareous limestone over saturated with ground water. It overlies unconformable the El Hagif formation (Pliocene) or the older Miocene. Surface quaternary deposits obscure actual contact (Adel ElFouly 2000, Girgis 1956, Said1990). Bulk structure of rock and construction materials. Different tests have been made to assess the durability and the weathering effects on the bulk structure of the rock mass used as building material. The following set of mineralogical analyses have been performed for the soft rock and the building materials. -X-ray diffraction -Thermal analyses DTA&TGA -X-ray Fluorescence analysis -Chemical analyses (for plaster and painting layers collected samples) -Transmitted plane polarized light (for the collected rock samples). -Scanning electron microscopy (SEM), attached with EDX Microprobe (energy dispersive Xray) microanalyses. -Porous media characterization for weathered and sound rock samples. -Pore size measurement and Specific surface area by nitrogen BET-TPV. -Determination of the specific surface area (SSA). -Grain size characteristics of the weathered (salt contaminated) rock samples. -Saturation coefficient, S. -Capillary water uptake measurements. Based on this extended set of tests a detailed mineralogy description of the rock mass is achieved together with other construction materials (plaster). A detailed presentation of these tests may be found in S.Hemeda (2007) (Progress report of the PhD thesis, AUTH). Selected results are presented herein. Catacombs of Kom El-Shoqafa are carved in the fossiliferous sandy oolitic limestone composed of calcite CaCO3 (47 %),quartz SiO2 (31%) ,halite NaCl (12 %),gypsum Ca SO4.2H2O (10 %).The rock is yellowish and can be characterized as medium grained with uniform relative grain size, angular to sub angular grain shape with equidimensional form and rough surface texture. Sound pieces of rock can be characterized of medium compactness and durability and the weathered pieces characterized of low compactness and durability. It must be mentioned that weathering attacked strongly the rock materials, started from the surface and continuing inward thus loosing the mineral fabric. Diagrams of representative rocks samples are given in figures (5). From this analysis we noticed that, % W.T of CaCO3 for the collected rock samples is 64, 25 and for the % W.T. free water is 0, 54 .and for the % W.T Ca (OH) 2 the results are zero. The elemental arrangement for the sandy oolitic limestone samples collected from Catacombs of Kom El-shoqafa can be put in a decreasing order according to their concentration:CaO(51.54),SiO2(14.55),MgO(1.07),Al2O3(2.51),SO3(0.59)Fe2O3(1.05),Na2O(0 .15) ,K2O(0.02),TiO2(0.04),P2O5(0.03)and L.O.I(28.20) total is (99.75). In the internal structure we can observe the dominant components, which are bioclasts of gastropods, foraminifera, algae, and shell debris, most of them are with test wall of neomorphic microspar, while the tests are internally filled with micrite and microspar.

352

Long contact

Long cracks

Figure .5. TGA-DTA results,Thin section blue dyed X40, the blue is the porosity.

Figure .6. Photomicrographs and (EDX) micro analysis of the tested sandy Oolitic limestone samples collected from catacombs of kom El-Shoqafa. The pore diameter distribution of these rock samples is ,10-20A (4%) ,20-30A(9.17 %) ,3050A(11.1%) ,50-100A(15.847%) ,100-200A(17.80%) ,200-1960 A (41.97 %) ,and nm 2.26599 E-05 , and BET(m2/gr) 2.21098, and TPV (ml/gr) 0.00992,and micro porosity % is 1.79327 ,fig(7) . and for the Adsorption / desorption isotherm diagram. But the rock sample which collected from the sound layers from the same site ( COM2) we noticed that : the pore diameter distribution of this rock is ,10-20A (9.46%) , 20-30A(12.5%) , 30-50A(14.98%) , 50-100A(12.72%) , 100-200A(7.93%) , 200-2040A(42.4%) ,and nm 1.5522E-05, BET(m2/gr) 1.51452 ,TPV (ml/gr) 0.00232 ,and micro porosity % is 0.42542 . Catacom b o f kom El-s hoqafa

catacomb of kom El-shoqafa 45

% PORES

45 40 35 30 25 20 15 10 5 0

42. 4

40

41.97

35 30

% 15.847 9.17

17.8

11.1

20 15 10

4

12. 5

14. 98

12. 72

9. 46

7. 93

5

0.12