• Author / Uploaded
• Facundo

Citation preview

A guide to safe and cost effective spillways Dr Peter J Mason MWH Ltd Terriers House 201 Amersham Road High Wycombe Bucks HP13 5AJ UK

Introduction Hydropower dams will all require means of dissipating the energy of any surplus flows. As a minimum these may represent a power station by-pass facility for use in the event of a sudden load rejection. The spillway and energy dissipation capacity installed to achieve that will correspond to the installed station capacity with the same amounts of power and energy to be dissipated. There may also be a need to pass amenity or surplus flows when one or more turbines are inoperative. More significant energy dissipation requirements will generally be required to safely discharge surplus flood water. This will relate not to turbine capacity and flows but rather to those of major flood events. For example in the case of Mangla dam in Pakistan the installed capacity of the power station is 1,000 MW whereas at times of peak floods the power dissipation requirement of the main flood spillway is approximately 25,000 MW. There are various standardised types of stilling basins and other energy dissipaters and these are listed in various textbooks and design guides. However the guidance is generally centred around hydraulic performance based on model testing. Their practical large-scale application, taking into account operational problems which may have occurred with them over the years, is often not covered with the same thoroughness. This paper attempts to go beyond theoretical hydraulic performance and guide the reader in terms of safe, while at the same time economic, practical application. For brevity the paper focuses on the requirements of energy dissipating structures rather than those of upstream conveyance works such as weirs, intakes and chutes. An exception is made where chutes are designed to act as energy dissipaters. Energy dissipation for large flows typically relies on one of the following general methods: Stepped or baffled chutes  Rock basins  Simple stilling basins including cascade basins  Baffle basins  Roller buckets  Flip buckets and plunge pools  Crest splitters and impact basins

Stepped and Baffled Chutes Both stepped and baffled chutes dissipate energy as the flows pass along them. Thus, when the flow eventually reaches tailwater the residual energy left to be dissipated is minimal. Stepped chutes follow two distinct flow regimes. The first is referred to as nappe flow and the second as skimming flow. In the first the water passes from step to step as a series of small waterfalls, dissipating energy as it does so. In the second, at higher unit flows, the water "skims" over the steps at high velocity, trapping hydraulic “rollers” between each step. Much more residual energy remains in this form of flow although some is still dissipated along the chute. More importantly significant pressure differentials can build up between adjacent parts of the steps and the chute walls. This can lead to severe damage and even collapse in the case of masonry spillways, see Fig 1. Stepped chutes have become increasingly popular for use on roller compacted concrete (RCC) dams, where the nature of the construction naturally suits their formation. They certainly have a reliable track record up to unit flows of approximately 12 to 15 m3/s/m.

Figure 1 – Major damage to the stepped masonry spillway at Boltby dam in the UK Baffle chutes too dissipate energy over the whole chute and feature baffles covering the entire chute surface. Well established guidelines exist for baffled chutes and successful large-scale applications include the Driekloof dam spillway in South Africa and the 1,600 m3/s emergency bypass spillway at the Ghazi-Barotha power station in Pakistan. They are effective but best used to fit specific needs as the limited unit flows, large area and the large numbers of reinforced concrete baffles required can make them an expensive option. The baffles may also be prone to damage where significant amounts of large floating debris are involved.

Hydraulic Jump Basins The use of hydraulic jumps for energy dissipation includes rock basins, simple stilling basins, cascade basins and baffle basins. Standardised guidelines, such as those used by the United States Bureau of Reclamation (USBR) are available to proportion such basins, generally in terms of the Froude number of the incoming flow. The basins will also require sufficient conjugate tailwater depths to force the jump to occur and to stop it sweeping out of the basin. Both Froude number and conjugate depth requirements are fairly straightforward to calculate. However operational experience has revealed at least three distinct types of problem can occur with such basins: Under-slab pressurization and displacement  Abrasion due to ball mill action  Cavitation damage to baffle blocks. A survey carried out by the writer some years ago, Mason (1982), indicated some generally safe operational ranges for these types of basins and these are shown in figure 2. Broadly speaking, simple hydraulic jump basins operate most successfully where the head difference between upstream reservoir level and downstream tailwater level is less than 50m. Above that there have been a notable number of basin problems either in terms of lifted slabs or abrasion due to debris circulating in the pool. This does not necessarily mean such basins cannot successfully operate at higher heads, only that special care may be needed with their design, detailing and operation. Baffle basins seem to work best at upstream to downstream head differences of between 10 and 30 m. Within that head range the baffles will intensified the jump, increasing the efficiency of energy dissipation and hence reducing the required basin size. At below 10 m the Foude number of the incoming flow is insufficient to form an effective jump and the energy is dissipated in swirls and eddies and at quite a low efficiency. Baffles are ineffective at increasing the turbulence levels of such flows. For this reason, at head differences below 10 m, simple basins are generally preferred. At upstream to downstream head differences of above 30 m the associated velocities have in the past caused significant damage to protruding baffles and splitter teeth, see figure 3, and so for upstream to downstream head differences of between 30 and 50 m, again simple basins are preferable.

Figure 2 - Typical safe ranges of use for various forms of energy dissipater

Figure 3 – Cavitation damage to splitter teeth at the high-head, Pit No 7 dam in the USA Cascade basins are especially useful where a considerable length of gravity dam can be devoted to overspilling. Simple collector basins along the downstream toe of such an arrangement can collect the flows and discharge them laterally in the forms of cascades, down the valley flanks and in towards a central main basin. The increased hydraulic turbulence involved in such an arrangement can make it especially efficient irrespective of the need for baffles.

Roller Buckets Good guidelines exist for the design of roller buckets, both solid and slotted. Where deep tailwaters are involved they can be a cost-effective option in place of conventional hydraulic jump stilling basins. However, because the energy dissipation takes place immediately downstream and over foundation material, there have been cases of damage due to circulating and entrained rock. This problem is accentuated if the buckets have operated symmetrically for any reason. It is probably true to say that the use of such buckets has been less popular in recent years.

Flip Buckets and Plunge Pools For high head flows flip buckets and plunge pools are generally a preferred option as costs are minimised and there is plenty of good and safe operational experience. Figure 2 indicates the typical head and flow ranges where such prototype arrangements were found to operate successfully in the survey on which the figure was based. Conceptually the flow is conveyed to such buckets or ski jumps with minimal energy loss. The bucket then deflects the flow as a free jet to some point downstream where it can excavate a scour hole or plunge pool in the downstream river bed. The maximum projection length for such jets is achieved when the projection angle is 45° to the horizontal. However economies in bucket size can be achieved in practice by reducing the angle to 35° and with very little difference in projection distance. Jet impact angles of below 25° are generally not preferred as the jet may fail to plunge fully and instead generate high lateral return currents. The present writer has already given guidance on design aspects associated with such buckets and also with associated downstream plunge pools, Mason (1993) and so these will not be repeated in detail here. Broadly speaking the bucket radius should be approximately 5 times the flow depth in order to avoid streamline separation. However some tightening of the exit radius to, say, 3 times the flow depth can be useful in shearing the flow to increase jet turbulence. On some ski jumps, such as at L’ Aigle in France, the ski jump chutes are super-elevated causing the flow to spiral and again leading to internal turbulence. Jets from counter opposing ski jumps can also be made to collide prior to impact, increasing energy dissipation in the air. This may also reduce scour depths. One disadvantage of such increased turbulence can be the generation of excessive spray. This in turn can lead to adjacent hillside instability through saturation and/or affect adjacent electrical installations such as switch yards. In practice where the buckets and plunge pools have given rise to problems, studies have shown that these have tended to be related to either: Asymmetrical gate operation in the case of wide chutes  Variable geology in plunge pool, again leading to asymmetrical scour development  Greater use than hitherto planned due, for example, to prolonged power station outages Fortunately such problems are rare and very much the exception rather than the rule. One query that arises quite regularly is whether or not to pre-excavate the plunge pool. Some prefer this on the basis that the pool geometry is then defined and controlled. Certainly if the rock can be used elsewhere, such as for rockfill or concrete aggregate, then pre-excavation may be useful, cost-effective and justified. In the absence of that, where the rock would simply go to waste, it may be preferable to simply let spillway flows excavate the plunge pool in due course. In such cases a rock bar will form downstream, raising the tailwater levels. This may be useful in minimising further scour but will be detrimental where such tailwater levels also affect power station generation. In such cases it is best to either site the plunge pool upstream of the power station tailrace, or in a separate channel to it, or to make some on-going maintenance allowance to ensure periodic rock removal to maintain generation capacity. Some degree of pre-splitting the foundation rock in the plunge pool area is another option to “guide” the development of the subsequent scour.

Crest Splitters and Impact Basins A very effective, but not widely known, method of high-capacity and high head energy dissipation is by the use of crest splitter arrangements in conjunction with a downstream impact basin. Such arrangements were originally developed in the 1930s and are often referred to as "Robert’s splitters" in recognition of the man who pioneered their use. They have been used successfully on many dams since, see figure 4.

Figure 4 – Roberts type crest splitter teeth and deflector lip with downstream impact basin at the Wadi Dayqah dam in Oman They are particularly suited to concrete dams where the overspill weir is fitted with both downstream deflector teeth and, below those, a continuous deflecting lip. In such a location, water velocities are not high enough to cause cavitation concerns. The combination of teeth and lip breaks up the flow at high level such that the falling flow is projected downstream in a less concentrated form than that of a solid jet. Impact basins receiving the dissipated falling jet can either be unlined rock or, more generally for high heads, concrete basins. In spite of the high heads sometimes involved there seem to be no reported problems with the use of these basins, possibly because the flow is so broken and aerated prior to impact. Furthermore the depth of tailwater required in the basin to avoid flows "sweeping out" is far less than would be the case with the same flow entering a conventional hydraulic jump basin. Successful uses of this arrangement have included the 108m high Vanderkloof dam in South Africa, the 122 m high Victoria dam in Sri Lanka and the 75m high Wadi Dayqah dam in Oman. Associated design flows for these were 8,500, 9,500 and 13,500 m3/s respectively. Wadi Daydah is an RCC dam and Robert’s splitters were used as an alternative to downstream steps in view of the high unit flows involved.

Dissipater Selection and Efficiency The appropriate form of stilling basin or hydraulic energy dissipater is often selected on the basis of personal past experience and preference. However it is recommended that reference can, and should, also be made to the guidance provided in figure 2. Another useful consideration is how efficient the basin might be in safely dissipating the power and energy of the flow. Clearly a form of dissipation which has a high value of power in relation to retained water involved will be more efficient, and so should require a more compact and less expensive construction. The approximate “power densities” of the various types of dissipater already listed are given below:   

Simple hydraulic jump basins Hydraulic jump basins with baffles Simple cascade basins Crest splitter teeth impact basins

5 to 10 kW/m3 15 to 20 kW/m3 30 to 40 kW/m3 50 to 60 kW/m3

By comparison plunge pools in rock below flip buckets have been shown to, typically, have a power density of around 7 to 9 kW/m3 as they approach stability. However, the fact that construction costs associated with these are limited to just that of the concrete flip-bucket, means that they are particularly cost-effective.

Other Factors – Limiting Total Dissolved Gases All the above discussions focus on the hydraulic efficiency and performance of various basins and other forms of energy dissipater. In all cases aeration is involved as part of the dissipation process. Indeed, it can be argued that the more efficient the dissipater the higher the degree of aeration. Increasingly engineers are required to consider the environmental aspects of their work and under some circumstances such high levels of aeration can be a disadvantage. In the case of large energy dissipaters, with stilling basin depths of 20 m or more, the combination of depth and aeration can cause significant levels of gas to go into solution. Total Dissolved Gas (TDG) saturation levels of 120% are generally accepted as an upper safe limit for downstream fish. Levels much above that can result in large-scale fish kills as gases come out of solution inside the fish. The phenomenon is the same as that experienced by divers who depressurised too quickly and develop “the bends”. On the Rio Parana in Argentina, large fish kills occurred downstream of the Yacytera dam due to supersaturation of dissolved gases in a particularly large spillway flow. Gases went into solution due to the incoming jet plunging down 20 m in the downstream stilling basin, but the process of gases coming out of solution was much slower. Measurements of 150% saturation immediately downstream of Yacetera stilling basin still corresponded to values of 107 to 110%, 90 km downstream. A significant factor was the wide and shallow nature of the Rio Parana downstream of Yacyreta. The fish had nowhere to dive and recover. Where rivers have much greater depths under flood the fish have places where they can stay deep and “de-compress” gradually. Eliminating artificial aeration on the spillway chute was tested and had only a small effect (a reduction of approx 7 to 10% in TDG levels). In the event, high level deflectors were tested and added to the main branch spillway to stop the jet plunging. This was effective at low flows but high flows still plunged. The design was a compromise between conflicting requirements, but generally successful. The downstream profile of the piers was also changed to reduce the “rooster tail” potential for aerating the flow. In practice it can be difficult to avoid this situation in the case of very large floods and shallow downstream rivers. Indeed the phenomena often occurs naturally downstream of waterfalls. Current mitigation practice in the US is to design spillways so that the one in 10 year flood, occurring over seven days, does not cause TDG levels in excess of 120%. One way of achieving this is to limit stilling basin flow depths to not more than 8 to 10 m. This avoids the TDG levels associated with higher pressures. However in practice such basins will also need to be sized for much larger flows. Limiting flow depths within them during low flows may not be practicable. An alternative, which has some successful precedent, is to deflect the flows horizontally at tailwater level to ensure that low flows enter only the upper part of the stilling basin water. Figure 5 shows the model testing of such a deflector lip on the downstream slope of a conventional spillway. It can be seen that the associated aeration in the stilling basin is mainly limited to the upper regions of the retained water. More interestingly when the same basin was operated at full flow the effect of the lip seemed to improve the dissipation as compared to that of a conventional basin.

Figure 5 – Deflected flow limiting air entrainment to the upper zones of water in a conventional stilling basin

Structural and Detailed Design Obvious factors in terms of structural and detailed design include the hydraulic and hydro-dynamic pressures to which various parts of the stilling basin or stilling device will be subject. Associated aspects include the location and details of any joints, water bars, under-drains and anchor bars. A typical hydraulic jump stilling basin fed from an upstream, sloping spillway chute will experience centrifugal pressures at the upstream juncture between the chute and the basin. Flow depths may be small but the effective water density will be high due to the change in flow direction. Pressures will then drop in that upstream part of the basin subject to only to the depth of the incoming flow but rise to tailwater levels at, and immediately after, the hydraulic jump. Aeration depths will also tend make effective water depths in the basin higher than indicated in model tests. High aerated flow depths acting on the spillway sidewall will exert the same total loading on the wall as an equivalent, non-aerated flow. However the higher aerated flow depths will exert a greater bending moment on the wall and should be assumed as the design case. Joints are best not located in the upstream zone subject to high centrifugal pressures. This avoids the possibility of these pressures being transmitted elsewhere. On small, river barrage type structures it is not uncommon to perforate stilling basin apron floors so that the pressures above and below the apron can locally balance. However, in major high head stilling basins it is generally preferable to isolate the water retained in the basin and make the basin watertight using water bars at joints. Under-drains are also generally provided and these are best made reasonably large to help dissipate any small high head leakages and pressure surges which may enter. On some basins drainage is only provided laterally and connected to risers in the stilling basin sidewall. This effectively prevents the upstream-downstream connection between the different pressure zones of the basin. Others prefer to interconnect the drains to create greater redundancy but, where this is done, sometimes the upstream, low pressure zone is kept separate from a downstream high-pressure one. For interconnected systems, drain exits should generally be located where external pressures are known and stable and not subject to hydrodynamic pressure fluctuations. Anchor bars are used for two reasons. The first is to provide yet another line of redundancy in case pressures are transmitted below the basin. In such cases the focus is generally on reinforcing those parts of the basin subject to reduced upper loadings against sub-pressures being transmitted from other parts of the basin where the upper loadings are higher. Large diameter bars are generally more economic than small ones given that two thirds of the cost of providing them is likely to be in hole drilling and installation. There are also good reasons to stagger their lengths, with every second bar taken to just half depth. The second reason anchor bars are used is to bond the upper slabs to the foundations thus making them less susceptible to vibration under the action of any large, local, hydrodynamic load fluctuations within the basin. For the same reason it is generally better to make basin apron slabs as large as practically possible to dissipate such local pressure fluctuations over larger areas. Distribution reinforcement will clearly be needed over the upper surfaces such slabs. It is generally also best to be generous with the concrete cover depths provided to such reinforcement. This will give some additional protection against abrasion damage over the basin floor. Given all the above some designers then avoid the use of shear keys between adjacent stilling basin floor slabs so that, if one slab does fail, adjacent slabs remain unaffected. However in the writer's experience and opinion the more common practice is to provide shear keys at joints so that any weak or susceptible floor slabs can gain additional support from adjacent ones.

Optimum Sizing The spillways and stilling basins for major dam and hydropower works are generally sized to be able to pass major floods such as the Probable Maximum Flood (PMF) or the 1 in 10,000 year flood. This does not, however, mean that the hydraulic performance needs to be optimised about such rare and infrequent events. The hydraulic performance of stilling basins is often better optimised around a lower figure, say, the 1in 500 or 1 in 1,000 year event. It is important too to consider how all such basins and other dissipaters will operate at much lower and more frequent flood events. Significant economies can also be made by shortening the length of hydraulic jump stilling basins. While standardised figures are published for the theoretical length of hydraulic jumps, generally based on the incoming Froude number, such jumps are often quite inefficient, with much of the energy being passed downstream. It is often acceptable to provide stilling basin with a length of approximately 60% of the

theoretical jump length, provided the works, including any downstream scour effects, are appropriately modelled and assessed. A downstream sill is also important on such basins as it will generate a reverse hydraulic roller immediately downstream. This will tend to push any material excavated immediately downstream of the basin back upstream against the sill rather than lead to a progressive deepening and eventual undermining.

Conclusions Simple conclusions from all of the above are that good guidelines exist for the hydraulic sizing of a whole range of alternative forms of energy dissipation devices for major spillways dams. However operational experience has also shown many cases of failure. Oversizing the energy dissipater is not necessarily the answer in such cases, in fact in many cases economies can safely be made by reducing the size of some such structures. The key to successful design is to better appreciate and understand the reasons for past failures and ensure they are not repeated. It should be remembered that simply copying past precedent is not always the safest option. Many stilling basins are submerged and have never been dewatered and inspected. Others may have been in existence for some years but never really tested under significant flood conditions. The broad recommended usage ranges recommended in Figure 2 are a good starting point towards a successful and safe design. This then needs to be coupled with the use of proven and tested, good design practices, such as those discussed in this paper, coupled with equally good follow-up on site to ensure that those practices and details are implemented.

Selected References and Bibliography Chow, V.T. (1983). Open Chanel Hydraulics, McGraw-Hill, New York. Khatsuria, R. M., (2005). Hydraulics of Spillways and Energy Dissipaters. Marcel Dekker, New York Mason P J, "The choice of Hydraulic Energy Dissipater for Dam Outlet Works based on a Survey of Prototype Usage", Proceedings of the Institution of Civil Engineers, Part1, (72), May 1982 Mason P J, "Practical Guidelines for the Design of Flip Buckets and Plunge Pools", International Water Power & Dam Construction, 45 (9/10), September/October 1993 Novak, P, Moffat, A. I. B., Nalluri, C, Narayanan, R. (2001). Hydraulic Structures. Spon, London. Thomas, H. H. (1976). The Engineering of Large Dams: Parts 1 & 2. John Wiley & Sons, London. US Army Corps of Engineers web site (www.140.194.76.129/publications/eng-manuals) USBR web site (www.usbr.gov/pmts/hydraulics_lab/pubs/index.cfm) USBR (1987). Design of Small Dams. United States Department of the Interior, Bureau of Reclamation (available on web as a free download, see above) Zipparo, V.J., Hansen, H. (1993). Davis’ Handbook of Applied Hydraulics, 4th ed, McGraw Hill, New York.

  The Author Dr Peter J Mason graduated with a BSc in Civil Engineering from the Woolwich Polytechnic and with MSc and PhD degrees in Applied Hydraulics from the City University, London. He is Technical Director, International Dams & Hydropower for MWH Ltd with a career spanning over 43 years and over 40 countries. He has worked on major international dam & hydropower projects in Africa, Asia, Europe and North and South America. Among other roles he currently chairs the Board of Management for a major, 969 MW hydropower project under construction in Asia. Recent Chairman of the British Dam Society he has also authored over 60 technical papers on all aspects of dams and hydropower.