• Author / Uploaded
• Leonardo Paliza

Citation preview

Proceedings

Volume 2

Pyrometallurgy I

Conference organized by GDMB, IIMCh, MetSoc, MMIJ, SME and TMS

Editor GDMB Paul-Ernst-Straße 10, D-38678 Clausthal-Zellerfeld Internet: www.GDMB.de Volume 1 Volume 2 Volume 3 Volume 4 Volume 5 Volume 6 Volume 7

ISBN 978-3-940276-25-4 ISBN 978-3-940276-26-1 ISBN 978-3-940276-27-8 ISBN 978-3-940276-28-5 ISBN 978-3-940276-29-2 ISBN 978-3-940276-30-8 ISBN 978-3-940276-31-5

Set (Volume 1+2+3+4+5+6+7)

ISBN 978-3-940276-32-2

All rights reserved. No part of this publication may be reproduced or electronically processed, copied or distributed without the prior consent by the editor. The content of the papers is the sole responsibility of the authors. All papers were peer reviewed by the corresponding members of the technical groups of the organizing societies.

Editorial staff: Production and marketing:

Dipl.-Ing. Jens Harre GDMB Informationsgesellschaft mbH

Printed by:

Papierflieger

© GDMB

Clausthal-Zellerfeld 2010

Bibliographische Information Der Deutschen Bibliothek Die Deutsche Bibliothek verzeichnet diese Publikation in der Deutschen Nationalbibliographie; detaillierte bibliographische Daten sind im Internet über http://dnb.ddb.de abrufbar. Bibliographic information published by Die Deutsche Bibliothek Die Deutsche Bibliothek lists this publication in the Deutsche Nationalbibliographie; detailed bibliografic data is available in the internet at http://dnb.ddb.de.

Proceedings

Volume 2

Pyrometallurgy I

The Copper 2010-Proceedings are friendly supported by

The Organizing Society: GDMB Society for Mining, Metallurgy, Resource and Environmental Technology The GDMB is a non-profit organization. Its activities focus on combining science with practical experience in the fields of mining, engineering, tunnelling, mineral processing, extraction, recycling and refining of metals, as well as on the manufacturing of semi and finished products. There is an increasing emphasis on associated environmental issues. The GDMB is internationally active with a European basis and covers a wide variety of topics from applied geology via processing to recycling. These include many important areas of chemistry, especially the complex metallurgical chemistry and, last not least, also analytical chemistry. As a consequence of their increasing importance, aspects of industrial minerals are addressed in addition to the traditional fields of metals and alloys. In order to remain a vibrant and attractive professional society, the GDMB draws on the experience and interests of its worldwide members.

The Co-Organizing Societies and their Representatives Institutos de Ingenieros de Minas de Chile (IIMCh) Enrique Miranda S., Gerente IIMCh, Chile

The Metallurgical Society of the Canadian Institute of Mining, Metallurgy, and Petroleum (MetSoc) Joël Kapusta, Ph.D., Air Liquide Canada Inc., Canada Dr. Phillip Mackey, Xstrata Process Support Centre, Canada

The Mining and Materials Processing Institute of Japan (MMIJ) Dr. Takahiko Okura, The University of Tokyo, Institute of Industrial Science, Japan Yasuo Tamura, Japan Mining Industry Association, Japan

Society for Mining, Metallurgy, and Exploration (SME) Dr. John L. Uhrie, Newmont Mining Corporation, USA

The Minerals, Metals & Materials Society (TMS) Dr.-Ing. Andreas Siegmund, LanMetCon, USA

IV

Proceedings of Copper 2010

Conference Chairman Dipl.-Ing. Michael Kopke

Aurubis AG, Germany

Technical Programme Chair Dipl.-Ing. Jo Rogiers

Aurubis AG, Belgium

Conference Chair Assistance Dipl.-Ing. Jürgen Zuchowski

GDMB Gesellschaft für Bergbau, Metallurgie, Rohstoff- und Umwelttechnik e. V.

Session Chairs 1

Plenary lessons of general interest for all conference members

Dipl.-Ing. Norbert L. Piret, Piret & Stolberg Partners, Germany

2

Economics

Dr. Patricio Barrios, Aurubis AG, Germany

3

Downstream Fabrication, Application and New Products

Dr.-Ing. Hans Achim Kuhn, Wieland Werke AG,Germany

4

Mineral Processing

Assoc. Prof. Sadan Kelebek, Queen’s University Canada

5

Pyrometallurgy

David B. George, Rio Tinto, USA

6

Hydrometallurgy

Dr.-Ing. Andreas Siegmund, LanMetCon, USA

7

Electrowinning and -refining

Dr.-Ing. Heinrich Traulsen, Germany Mike Murphy, Xstrata Technology, Australia Mike Hourn, Xstrata Technology, Australia

8

Process Control, Automatization and Optimization

Prof. Dr. Markus Andreas Reuter, Outotec Ausmelt, Australia

9

Recycling

Dipl.-Ing. Jörg Wallner, Austria

10

Sustainable Development / Health, Safety and Environmental Control

Dipl.-Ing. Miguel Palacios Atlantic Copper S.A., Spain

Proceedings of Copper 2010

V

Technical Group Chairs GDMB

(region: Europe, Russia, near Orient)

Dipl.-Ing. Jo Rogiers, Aurubis AG, Belgium

IIMCh

(region: South America)

Sergio Demetrio, IIMCh, Chile

MetSoc

(region: Canada, Australia, Africa)

Ass. Prof. Edouard Asselin, University of British Columbia, Canada

MMIJ

(region: Japan, China, South East Asia)

Dr. Takahiko Okura, The University of Tokyo, Japan

SME / TMS (region: USA, Mexico)

Dr. Andreas Siegmund, LanMetCon, USA

Technical Group Members Full information you will find in the internet at: www.Cu2010.GDMB.de

Short Course Organizing Committee Dipl.-Ing. Michael Kopke (Chair), Aurubis AG, Germany Dipl.-Ing. Miguel Palacios, Atlantic Copper S.A., Spain Prof. Dr. mont. Peter Paschen, Austria Dipl.-Ing. Norbert L. Piret, Piret & Stolberg Partners, Germany

Organizing Committee Dipl.-Ing. Jürgen Zuchowski

GDMB Gesellschaft für Bergbau, Metallurgie, Rohstoffund Umwelttechnik e. V. (Copper2010 Organizing Committee Chairman)

Mareike Hahn

GDMB Gesellschaft für Bergbau, Metallurgie, Rohstoffund Umwelttechnik e. V.

Thomas Marbach

GDMB Gesellschaft für Bergbau, Metallurgie, Rohstoffund Umwelttechnik e. V.

Dipl.-Ing. Jens Harre

GDMB Informationsgesellschaft mbH

Mareike Müller

GDMB Informationsgesellschaft mbH

Ulrich Waschki

GDMB Informationsgesellschaft mbH

VI

Proceedings of Copper 2010

Preface Copper – Indicator of the progress of civilization This is the motto for the 7th international copper conference, the most important copper seminar in the world, which has been organized by the GDMB, the German based Society for Mining, Metallurgy, Resource and Environmental Technology, together with IIMCh from Chile, MetSoc from Canada, TMS, SME from USA and MMIJ from Japan. The copper conferences bring together the highest level of science and technology: universities, metal producers, manufacturing companies, suppliers and finally the people who work with copper: scientists, technicians, engineers, traders and many more. An extensive programme has been arranged for this conference and an abundance of contributions from all over the world dealing with the different aspects of copper making and its use are registered already, for which we gratefully thank the authors. Apart from plenary addresses, separate sessions will be held for economics, mineral processing, pyrometallurgy, hydrometallurgy, electrowinning and -refining, downstream fabrication and application, process control and automation, recycling and sustainable development, environmental control, health and safety. Copper, one of the oldest metals used by mankind, is still today one of the most important industrial metals and indispensable for modern life. It is the indicator of industrialization and progress in every country. It is used everywhere, where electricity flows and thus is still valued so highly today. The increased economic potential of newly industrialized countries, above all East Asia and China, has increased the significance of the red metal once again. More recent technologies in production, processing and application often provide new answers to old questions. From the middle of the last century there was another innovation surge resulting in totally new technologies, a trend still going on. This has made each copper conference into an exciting adventure. It is positive and reassuring that particularly the high industrialized countries have become the vanguard, not just in technical innovation, but also protection of environment and nature and preserving resources. They repeatedly prove that ecology and economy may go hand in hand. Many sponsors have contributed to the conference’s success, for which I would like to express my sincere thanks! Hamburg is expecting its guests! Hamburg, the old Hanseatic city with a 1200 year long tradition, one of the biggest and most beautiful cities in Germany, combines a wonderful mixture of industry, commerce, nature and culture. Not only the “Copper 2010” in the Congress Centre awaits you, but rich offerings of sightseeing and shopping in a cosmopolitan city, one of the largest harbours in Europe, the Alster lake in the city centre, green parks and plenty of cultural events, some of which we hope to show you in the companions programme. We are delighted that you will join us and look forward to a highly interesting conference! Michael Kopke Chairman Copper 2010 Proceedings of Copper 2010

VII

Structure of the Proceedings • Volume 1: Downstream Fabrication, Application and New Products Sustainable Development / Health, Safety and Environmental Control • Volume 2: Pyrometallurgy I • Volume 3: Pyrometallurgy II • Volume 4: Electrowinning and -refining • Volume 5: Hydrometallurgy • Volume 6: Economics Process Control, Automatization and Optimization • Volume 7: Plenary lessons of general interest for all conference members Mineral Processing Recycling Posters Authors Index Keywords Index

VIII

Proceedings of Copper 2010

Plenary Lectures (Abstracts) Some of the full papers will be published in Volume 7 of the Proceedings of Copper 2010 and in World of Metallurgy – ERZMETALL.

Is the Copper Industry Fit for the Future? Dr.-Ing. Bernd Drouven, CEO, Aurubis AG, Hamburg, Germany The copper world has already changed a great deal during recent years. But what still lies ahead of us? What are the changes in conditions that we have to cope with and how will our solutions look? The rise in the demand for raw materials is unrelenting. At the same time, both the primary and secondary feed materials have a specific complexity, the customers’ needs are becoming increasingly differentiated and their orders are placed at increasingly short notice. Production times for metals have to be faster and inventories minimised. The volatility of the metal prices has increased significantly in recent years and we will probably have to live with that in the future as well. The LME functioned well in the crisis, but the copper price is being influenced more and more by funds. How can the value added chain change to adapt to this? Which consequences will that have for process technology, production planning and logistics? How will the consolidation of our industry continue? This keynote will go into the various effects and challenges – in particular from the perspective of a European custom smelter and fabricator – and present possible approaches for solutions.

Sustainable Growth Strategy for Japanese Copper Business Toshinori Kato, Managing Director, Mitsubishi Materials Corporation, Tokyo, Japan Business environment, for any industry, has changed dramatically over the last decade. Copper industry is no exception and those involved are experiencing an unprecedented period of upheaval. The landscape of the market has completely altered, with large-scale M&As taking place among miners - creating an oligopoly situation - and a rapid expansion of smelting capacity within developing countries, driven by strong economic growth. Traditional copper smelters and fabricators have been facing challenges, but a long-term sustainability of Japanese copper business is achievable. Firstly, an immense amount of effort has been made over the years to develop the clean and one of the most environmentally-friendly processes in the industry. Mitsubishi Materials (MMC) particularly plays a great role in establishing Japanese smelters’ reputation as the most energy-efficient operations in the world. It is our strong commitment to be a leader in this expertise by sharing our technologies with the industry.

Proceedings of Copper 2010

IX

Secondly, infrastructures for copper smelting have been utilized in developing the recycling business. Exemplified by the operations at Onahama smelter, which has the world’s largest furnace for treatment of shredder residue, the industry has worked closely with other sectors and municipalities. Building an effective structure to make the best use of our facilities is a key for success in this field. It is fair to say that Japanese copper smelters, including MMC, are now regarded as an indispensable part of national environmental policy. Finally, progression of integration in the copper fabrication sector has strengthened the industry’s capability of providing a variety of high-value-added products to end-users. Tight relationships with the consumers have been beneficial in developing new use of copper. Evidence of a promising future can be seen in increasing use of copper in the growing sectors such as hybrid automobiles and renewable energy.

Copper Sulphide Smelting: Past Achievements and Current Challenges Dr. Carlos M. Díaz, Adjunct Professor, University of Toronto, Private Consultant, Toronto, Canada In the last three decades, increased oxygen consumption in copper sulphide smelting and converting and the implementation of computerized process control, among other factors, have led to higher process intensity and smelter concentrate processing capacity, decreased smelting energy consumption and improved SO2 capture from process gas streams. An annual primary smelting furnace throughput of one million tonnes of concentrate is the new industry standard. Top submerged lance smelting has become an important processing route in recent years. Two new continuous converting routes were commercialized in the 1990s. However, due to substantially improved converting practice and larger converters, Peirce-Smith converting still maintains its position as the dominant technology. A major realignment of world copper smelting has taken place in the last 30 years. Spurred by rapid economic expansion and the resulting huge increase in demand for basic materials, a number of modern, large capacity copper smelters have been built in China, India and other Asian countries. Only moderate growth in smelting and electrorefining capacity has taken place elsewhere. Moreover, ER cathode output in the USA has substantially decreased. In this paper, the author examines recent technological advances and industry changes and highlights issues, such as energy consumption and the corresponding greenhouse gas emissions that will become the focus of future discussion.

Energy as a Key Factor of Sustainability Javier Targhetta, Vice President, Freeport McMoRan, Phoenix, Arizona, USA Energy must be seen as the prime mover for development and is therefore vital for economic equilibrium and social welfare. Energy will continue to play a key role in the coming decades being not only an environmental challenge but also a fundamental issue in terms of the progress of humanity. Nevertheless, harmonizing several aspects related to energy management will be of essence for the X

Proceedings of Copper 2010

near future and will make energy one of the greatest challenges of the 21st century. It is becoming increasingly important to maintain the appropriate equilibrium between environmental issues (global warming, the future of nuclear waste, the integration of renewable energies…), coordinated governments policies to ensure that global energy costs are kept competitive and companies finding a balanced, rational and fair model of energy utilization based on the review of ethical business aspects and the establishment of social responsibility principles. This paper develops the idea that the rules and procedures for future energy use must be based on a reasonable equilibrium between technical aspects mainly associated to environmental issues, political decisions made by governments and public administrations and social responsibility programmes that companies themselves must assume and implement.

The Supply and Copper Producer Response to a Growing Demand Scenario Ricardo Alvarez, General Manager, Codelco El Teniente Division, Santiago, Chile The prospects for medium and long-term consumption of copper are promising. A recovery projected for developed economies, starting 2010, joins the growing impact on demand generated from the process of development and urbanization of emerging countries. The intensity of copper use will also maintain the positive growth started a decade ago based on factors such as providing solutions to combat global warming. The objective of this presentation is to analyze how supply may react and adjust to envisaged demand scenarios, bearing in mind some distinctive elements of the copper industry like: - Historical supply reaction to demand and price levels - Availability of resources and copper ore reserves, incorporating a dynamic analysis and the effects of geological exploration and possible technological changes. - Pipeline of probable and possible projects, analysing the effects of projects in less developed and riskier geographical locations - Technological changes under development that may positively impact the amount of reserves available and the competitiveness of projects. - The growing role of scrap as a supply source for copper. The analysis of these points will allow us to confirm the capability of supply response to growing demand, ruling out revisited hypothesis of insufficient reserves

Implementation of Recent Global Copper Projects Tim J. A. Smith, Vice President, Copper, SNC-Lavalin UK Limited, Croydon, Surrey, UK Despite the recent global recession, a relatively strong copper price combined with continuing supply shortfalls continues to drive the implementation of a number of important new worldwide copper projects.

Proceedings of Copper 2010

XI

Along with base metals projects in general, the scale and complexity of such projects has increased such that multibillion dollar projects are increasingly common. Together with expansion projects in the older traditional copper producing regions, new geographic areas are still being opened up. These frequently require major infrastructural development, environmental and global procurement capabilities as major components of such projects. This plenary session address will examine and discuss the many project management challenges and skills needed to deliver successful projects worldwide, as viewed from the perspective of one of the world’s leading metallurgical plant engineers and constructors.

XII

Proceedings of Copper 2010

Table of Contents – Volume 2 Pyrometallurgy I

(Authors A-L)

Design of Copper-Cobalt Sulphating Roasters for Katanga Mining Limited in D.R. Congo

587

Dr. Kamal Adham, Tuisko Buchholz, Alex Kokourine, Rayson Lu, Jim Sarvinis, Andrew Tohn, Stephane Girouard

Filsulfor and Gypsulfor: Modern Design Concepts for Weak Acid Treatment

601

Dr. Angela Ante

Feasibility to Profitability with Copper ISASMELT™

615

G. R. Alvear F., P. Arthur, P. Partington

Present and Future Modernization of Metallurgical Production Lines of the Głogów Copper Smelter

631

Leszek Byszyński, Leszek Garycki, Zbigniew Gostyński, Tomasz Stodulski, Jerzy Urbanowski

Energy Consumption in Copper Sulphide Smelting

649

P. Coursol, P. J. Mackey, C. M. Díaz

Problems of Lead and Arsenic Removal from Copper Production in a One-Stage Flash-Smelting Process

669

J. Czernecki, Z. Śmieszek, Z. Miczkowski, G. Krawiec, S. Gizicki

Control of Fugitive Emissions in a Continuous Mitsubishi C-Furnace during Limestone Fluxing

685

Adam Salomon de Friedberg, Alan Hyde, Mark Coleman

Modern Flash Smelting Cooling Systems

699

K. Fagerlund, M. Lindgren, M. Jåfs

Proceedings of Copper 2010

XIII

Introduction of a Slide Gate System for Copper Anode Furnaces

713

Dipl.-Ing. Klaus Gamweger

New Highly Efficient Rotary Furnace for Environmentally Friendly Refining Process

721

Dr. Bernhard Hanusch

Sulphur Capacity of the “FeO”-CaO-SiO2 Slag of Interest to the Copper Smelting Process

731

H. M. Henao, P. C. Hayes, E. Jak

Changes in the ISASMELTTM Slag Chemistry at Southern Peru Ilo Smelter

749

Enrique Herrera, Leopoldo Mariscal

Experimental Study of Phase Equilibria of Silicate Slag Systems

761

M. Phil. T. Hidayat, Dr. H. M. Henao, Prof. P. C. Hayes, Prof. E. Jak

Dryer Fuel Reduction and Recent Operation of the Flash Smelting Furnace at Saganoseki Smelter & Refinery after the SPI Project

779

Mitsumasa Hoshi, Katsuya Toda, Tatsuya Motomura, Masaharu Takahashi, Yushiro Hirai

Development of the Continuous Copper Converting Using Two-Zone Vaniukov Converter

793

E. Jak, L. Tsymbulov

Liquidus Temperature in Calcium Ferrite Slags in Ca2Fe2O5 and Ca2SiO4 Primary Phase Fields with Cu and Fixed Po2

811

E. Jak, B. Zhao, C. Nexhip, D. P. George-Kennedy, P. C. Hayes

Numerical Simulation of Fluid Flow and Melt Temperature in Settler

823

Zhou Jun, Zhou Ping, Chen Zhuo, Liu Anming, Meichi

XIV

Proceedings of Copper 2010

Profit Enhancement through Steam Selling

831

Kyoung-Soo Jung, Gun-Woong Byun, Sung-Ho Shin

Gas Injection Phenomena in Converters – An Update on Buoyancy Power and Bath Slopping

839

Dr. Joël P. T. Kapusta

Recovery of Valuable Metals from Converter Slags by Reduction with Iron

863

Dipl.-Ing. Stefan Konetschnik, Dipl.-Ing. Helmut Paulitsch, Dipl.-Ing. Dr.mont. Josef Pesl, Ao.Univ.Prof. Dipl.-Ing. Dr.mont. Helmut Antrekowitsch

Waste heat boilers for Copper Smelting Applications

879

Dipl.-Ing. Stefan Köster

Processing of High-Silicon Copper Sulfide Concentrates by Vanyukov Smelting

893

S. M. Kozhakhmetov, S. A. Kvyatkovskiy, E. A. Ospanov, Z. S. Abisheva, A. N. Zagorodnyaya

Boiler Tube Cooling of TSL-Furnace Walls

907

Heikki Lankinen, Rauno Peippo

Experimental Estimation of the Residence Time Distribution in a P-S Converter

919

C. López, A. Almaraz, R. Cuenca, B. Hernández, F. Reyes, G. Plascencia

Numerical Simulation of Air Blowing into a Copper Matte in a P-S Converter Using a Convergent / Divergent Nozzle

931

C. López, A. Almaraz, I. Arellano, E. Martínez, M. A. Barrón, T. A. Utigard, G. Plascencia

Proceedings of Copper 2010

XV

Table of Contents – Volume 3 Pyrometallurgy II

(Authors M-Z)

Sulphide Bath Smelting: 19th Century Concept and Hollway’s Legacy

945

P. J. Mackey, A. E. Wraith

Large Scale Copper Smelting Using Ausmelt TSL Technology at the Tongling Jinchang Smelter

961

R. W. Matusewicz, Prof. M. A. Reuter, S. P. Hughes, Shengdao Lin, Laisheng Sun

Extending Copper Smelting and Converting Furnace Campaign Life through Technology

971

K. McKenna, C. Newman, N. Voermann, R. Veenstra, M. King, J. Bryant

High Intensity Cast Cooling Element Design and Fabrication Considerations

987

K. McKenna, N. Voermann, R. Veenstra, C. Newman

Management of Copper Flash Smelting Off-Gas Line Gas Flow and Oxygen Potential

1003

Dr. Elli Miettinen, Tapio Ahokainen, Kaj Eklund

The Teniente Converter: A High Smelting Rate and Versatile Reactor

1013

Alex Moyano, Fernando Rojas, Carlos Caballero, Jonkion Font, Marco Rosales, Hugo Jara

Development of Sumitomo Concentrate Burner

1025

K. Nagai, K. Kawanaka, K. Yamamoto, S. Sasai

Review of Process Options to Treat Enargite Concentrates

1035

J. G. Peacey, M. Z. Gupta, K. J. R. Ford

XVI

Proceedings of Copper 2010

Customized Burner Concepts for the Copper Industry

1051

Dipl.-Ing. Michael Potesser, Dipl.-Ing. Burkhardt Holleis, Dipl.-Ing. Dr. Mont. Martin Demuth, Dipl.-Ing. Davor Spoljaric MBA, Dipl.-Ing. Johannes Zauner

Process Optimization by Means of Heat and Mass Balance Based Modelling at Olympic Dam

1063

D. J. Ranasinghe, R. Russell, R. Muthuraman, Z. Dryga

Clyde-WorleyParsons’ Flash Furnace Feed System: The Development Cycle

1079

Michael E. Reed, Charles U. Jones, Brian Snowdon, Mark Coleman

A Fluid-Dynamic Review of the Teniente Converter

1095

M. Rosales, J. Font, R. Fuentes, A. Moyano, F. Rojas, C. Caballero, R. Mackay

Usage of Colemanite in Copper Matte Smelting

1115

Aydın Rüşen, Ahmet Geveci, Yavuz Ali Topkaya

Furnace Condition Assessment and Monitoring by Utilization of Innovative Non-Destructive Testing (NDT) Techniques

1123

Afshin Sadri, Ehsan Shameli, Pawel Gebski, David George-Kennedy

Gresik Operation: Past, Present and Future

1143

Hideya Sato, Djoko S. Adji, Antonius Prayoga, Bouman Tiroi S.

FSF Online Process Advisor

1155

Ville Suontaka, Peter Björklund

The Development of the Chinese Copper Industry and Copper Extraction Technology

1167

Yao SuPing, Wang Wei

Proceedings of Copper 2010

XVII

Gas-Gas Cooler as Off-Gas Duct for a Slag Cleaning Furnace: Example of HSE Progress and Engineering Excellence

1173

Michael Ströder, Miguel Palacios

Kumera Technology for Copper Smelters

1183

Jyri Talja, Dr. Shaolong Chen, Hannu Mansikkaviita

The Initial Years of the O-SR Process

1199

F. Tanaka, K. Kiyotani, O. Iida

Latest Results of the Intensive Slag Cleaning Reactor for Metal Recovery on the Basis of Copper

1213

A. Warczok, G. Riveros, R. Degel, J. Kunze, M. Kalisch, H.Oterdoom

3D-Refractory Engineering Using the Example of a MAERZ Tilting Furnace

1233

Dr. Christine Wenzl, Dipl.-Ing. Ladislav Koncik, Stefan Ruhs, Dr. Andreas Filzwieser

Ionic Liquids – The New Way in Cooling Technology

1247

Dr. Christine Wenzl, Dr. Andreas Filzwieser, Dr. Iris Filzwieser, Eva Raaber

Gold Extraction from Copper Ferrite Residue Produced by Oxidizing Roasting Copper Matte

1259

I. Wilkomirsky, N. Rojas, E. Balladares

Smelting of High-Arsenic Copper Concentrates

1273

I. Wilkomirsky, R. Parra, F. Parada, E. Balladares, Carlos Caballero, Andrés Reghezza, Jorge Zúñiga

Distribution of Precious Metals between Matte and Slag and Precious Metal Solubility in Slag

1287

Katsunori Yamaguchi

XVIII

Proceedings of Copper 2010

Effects of SiO2, Al2O3, MgO and Na2O on Spinel Liquidus in Calcium Ferrite Slags with Cu and Fixed Po2

1297

B. Zhao, C. Nexhip, D. P. George-Kennedy, P. Hayes, E. Jak

Simulation Study of Intensified Flash Smelting Process

1313

Chen Zhuo, Zhou Jun, Wang Yunxiao, Liu Anming, Meichi

Proceedings of Copper 2010

XIX

Pyrometallurgy I

Proceedings of Copper 2010

585

586

Proceedings of Copper 2010

Design of Copper-Cobalt Sulphating Roasters for Katanga Mining Limited in D.R. Congo Stephane Girouard

Dr. Kamal Adham, Tuisko Buchholz, Alex Kokourine Rayson Lu, Jim Sarvinis, Andrew Tohn Hatch Ltd. 2800 Speakman Drive Mississauga, Canada

Katanga Mining Limited Avenue Industrielle 3207 Kolwezi, D.R. Congo

Keywords: Sulfide ore, copper-cobalt concentrate, sulphating roasting, fluid bed, Finite Element Analysis (FEA)

Abstract In late 2006, Hatch was awarded a contract to supply fluid bed technology for the Katanga Mining Limited (KML) copper-cobalt processing project in D.R. Congo (DRC). The new roasting facility is part of the revitalization of an existing facility in the Katanga Province of the DRC being undertaken by KML. The sulphide ore fed to the roasting plant is treated under the “sulphating roast” conditions, followed by hydrometallurgical processing. Equipment for the first roasting line has been installed and the plant commissioning is currently (October 2009) underway. Hatch's Fluidization Technology Group (FTG) in Mississauga, Canada supplied its proprietary equipment and technology to KML under a license agreement, which included the provision of detailed engineering services for the roaster equipment. Hatch offices in Canada and South Africa also conducted the required engineering for the balance of the plant, including off-gas dust and SO2 scrubbers design, civil and structural, piping and ducting, electrical and process controls. This paper only describes the process conditions planned for the sulphating roast operation, and the custom-design features of the fluid bed equipment employed.

1

Introduction

Hatch’s scope of work involved the design of two eight meter inside diameter (I.D.) copper/cobalt concentrate fluid bed roasters, two fluid bed calcine coolers for product cooling, and the corresponding feed and product handling systems for the roaster area. Design of the roaster included complete thermal modeling of the roaster shell, roof, tuyere plate and associated expansion joint, grillage beam, and windbox, as well as design of the feed injection nozzles, cooling water guns, piping

Proceedings of Copper 2010

587

Adham, Bucholz, Kokourine, Lu, Sarvinis, Tohn, Girouard and ducting, expansion joints, and discharge solid valves (seal legs). Stress analyses and load conditions for several cases were examined, including start-up and upset conditions. Design of the new roaster lines was based on the KML’s operating experience with the three existing roaster lines, two comprising 4.9 meter I.D. roasters, and one comprising a 7.3 meter I.D. roaster. Design of the two new lines also included testwork of the representative feed samples, performed in the Republic of South Africa. Several improvements were made compared to the original roaster design, such as the elimination of transfer boxes for direct solids discharge through seal legs. The transfer boxes were a significant source of maintenance and downtime for KML due to excessive blockages. Seal legs are expected to offer less flow-path restriction, thereby reducing the frequency of blockages. The installation of the first line is expected to be complete by October 2009, with start-up of the roaster to follow immediately. Design of the second roaster line is well advanced, with the construction work planning currently under consideration. Alternative technologies and processes were considered during the initial phases of the project. Among them, the use of an autoclave for sulphation of the slurry feed material was considered. However, following technical reviews, KML decided to design the new plant similar to the existing roast-leach concept that had been operating for several decades. As such, fluidized bed technology was selected for the partial roaster design. The following challenges were addressed during the course of the fluid bed design: • • • •

High viscosity and solids content of slurry feed, Hot shell design, to avoid acid condensation, Large roaster diameter, for maximum capacity per roaster, Novel tuyere plate design for improved thermal stress relief.

Due to the shutdown of the existing plant in the late 1990’s, much of the 50-year long operational data was unavailable. Limited representative feed ore samples were available, as the rehabilitation of the mine had just begun. Many feed properties needed to be inferred from a limited pool of data. The initial design had to be based on these corollaries until the exact properties could be confirmed through testwork. In order to optimize the heat balance, it was desired to limit any auxiliary fuel (coal) addition to the roaster. As such, feed slurry was maintained at up to 75 wt. % solids to minimize heat consumption through evaporation of water to superheated vapor. This resulted in a slurry that presented a significant challenge in the roaster feed system design due to both the high apparent viscosity and the very high solids content. The presence of sulphur gases in sulphation reactors necessitates a high accuracy of design calculations and precise monitoring of roaster shell temperatures. In order to maintain a sufficient 588

Proceedings of Copper 2010

Design of Copper-Cobalt Sulphating Roasters Katanga Mining Limited vessel shell temperature to prevent condensation and associated corrosion, while complying with safety requirements for the external surface temperature, the refractory and external insulation arrangement and materials were optimized. This optimization was carried out by an iterative steady state heat transfer analysis of the metal cylindrical shell with external and internal thermal insulation and forced/free convection on the internal/external surfaces. Design of a high temperature fluid bed vessel is governed by the vessel size and key operating conditions. Due to the high throughput requirements, a large diameter roaster (8 m I.D.) was selected. This presented a significant challenge through increased deflection of the tuyere plate/grillage beam assembly, due to vertical and horizontal temperature gradients. Several novel design solutions were implemented to overcome this problem including the specialized structural supports for the tuyere plate and the main beams, and the specialized expansion joint between the tuyere plate and the vessel shell. Design of the tuyere plate, supporting grillage beam, main beam bearing blocks and tuyere plate expansion joint presents a novel approach to the design of fluid bed vessels. Design offers improved distribution of the tuyeres for a more homogeneous introduction of fluidizing air to the roaster while maintaining complete structural integrity of the vessel elements at operating, start-up, shutdown and upset conditions. The developed methodology of analytical and Finite Element (FE) calculations integrated the steady state and transient heat transfer analyses for a hot suspended/slumped bed inside the refractory lined and externally insulated fluid bed vessel. Every effort was made throughout the project to comply with the highest standards of health, safety, and the environment. Design was performed to specification according to the South African Bureau of Standards (SABS). Design of the roaster off-gas system was performed to the requirements laid out by the World Bank.

2

Copper/cobalt process overview

Ore from two KML mine sites is sent to the Luilu Metallurgical Plant for treatment; an oxide ore originating from open pit mines and a sulphide ore from an underground mine. The average copper and cobalt grades of the sulphide ore from the underground mine is 4.21 % and 0.37 %, respectively. Run-of-mine sulphide ore is milled and then concentrated in froth flotation cells. Flotation concentrate is pumped to a thickener located in proximity to the roaster building. Thickener underflow is filtered and the cake sent to temporary storage. The 75 wt. % solids roaster feed slurry is produced from the stored filter cake in a modified ball mill and fed to the roaster area containing approximately 43 % copper and 4 % cobalt (dry basis). Analysis of the major sulphide components of the feed slurry is given in Table 1.

Proceedings of Copper 2010

589

Adham, Bucholz, Kokourine, Lu, Sarvinis, Tohn, Girouard Table 1: Element wt. %

Approximate composition of sulphides in roaster feed (dry basis) Cu2S

CuS

CoS

FeS

37

15

6

2

Roasting of the sulphide slurry feed produces a dried calcine product that is highly amenable to leaching in the form of oxysulphates of copper and sulphates of cobalt. Cooled calcine product is then sent to the leach tanks, where it is mixed with the recycled spent electrolyte for leaching. Product of leaching is then filtered prior to the electrolysis. Downstream processing of sour roast product leads to a significant net generation of sulphuric acid; approximately 220 kg of sulphuric acid is generated for every ton of calcine product produced. This is a large profit driver for the treatment of the open pit (oxide) ore.

2.1 Roaster operation The roasting area consists of a bubbling fluid bed roaster, fluidized calcine cooler for product cooling to 150 °C, and ancillary feed and product distribution system. Figure 1 shows a simplified process flow diagram. Thickened and filtered concentrate is fed to a repulping ball mill ahead of a slurry storage tank. Operation of the ball mill is intermittent, presenting a design challenge in agitation of the tanks. A significant hold-up of slurry is necessary to ensure continuous feed to the roaster during ball mill downtime. As such, the slurry tanks were oversized at a design storage volume of 100 m3, allowing for a 20 hour retention time. Given this large size of tank and high solids content of the slurry to be agitated, design of the agitation tank was carefully performed to ensure the solid particles remained in suspension. Slurry is then pumped to a roof-mounted surge tank, which provides a density check of the slurry prior to feeding the roaster. Additionally, the surge tank has been designed to maintain a constant head, so as to ensure a continuous, steady flow to the roaster. The roaster roof-mounted feed injection guns atomize the feed to approximately 250 µm diameter droplets. Atomized slurry feed droplets undergo partial drying in the freeboard prior to entering the fluidized bed region. The bed is controlled to a temperature in the range of 650 to 700 °C through the use of water sprays mounted on the roof of the roaster for hot bed conditions, and through the use of airassisted coal injection for cold bed conditions. Under normal operating conditions, the roaster is thermally self-sufficient, with some water required to cool the bed. A dedicated 800 horsepower fluidizing blower is used to supply fluidizing air to both the roaster and the calcine cooler. Fluidizing air flowrate to the roaster is set at an adequate level of excess air to ensure complete sulphation. The fluidizing blower is designed with a substantial turndown capability through the use of controllable dampers and an outlet vent for low-flow operation. 590

Proceedings of Copper 2010

Design of Copper-Cobalt Sulphating Roasters Katanga Mining Limited

Figure 1:

Katanga roasting process flow diagram

Roaster off-gas is treated in two parallel, high efficiency cyclones for dust capture before being sent to a scrubber, while the captured dust is fed back to the roaster. The off-gas scrubber uses spent electrolyte for scrubbing to maximize copper/cobalt recovery. The use of water for scrubbing of offgas dust would dilute the concentration of copper and cobalt in the spent electrolyte, and has a negative impact on the acid balance in the plant. Solid product is discharged from the roaster via a seal leg and fed to the calcine cooler freeboard, where water sprays maintain a bed temperature of 150 °C to cool the calcined product, while maintaining a sufficiently high temperature to avoid localized bed wetting and de-fluidization. Cooler off-gas is passed through a single high-efficiency cyclone for dust removal, prior to being combined with roaster off-gas for scrubbing. Dust removed from the off-gas by the scrubber is sent to a quench tank, where it is mixed with product discharged from the calcine cooler and cooler cyclone capture. The quench tank uses spent electrolyte to maximize recoveries and optimize spent electrolyte use. A sketch of the fluid bed roaster and calcine cooler, along with auxiliary equipment, is given in Figure 2. Proceedings of Copper 2010

591

Adham, Bucholz, Kokourine, Lu, Sarvinis, Tohn, Girouard

Figure 2:

Sketch of fluid bed roaster and auxiliary equipment

2.2 Important chemical reactions The principle chemical reactions taking place in the roaster are the sulphation of the various sulphides present in the feed slurry. These include: 4CuS + 7O2 → 2CuSO4 · CuO + 2SO2

∆Hrxn= -2,226 kJ

(1)

2Cu2S + 5O2 → 2CuSO4 · CuO

∆Hrxn= -1,703 kJ

(2)

CoS + 2O2 → CoSO4

∆Hrxn= -795 kJ

(3)

4FeS + 7O2 → 2Fe2O3 + 4SO2

∆Hrxn= -2,439 kJ

(4)

592

Proceedings of Copper 2010

Design of Copper-Cobalt Sulphating Roasters Katanga Mining Limited Additionally, the decomposition of MgCO3 to MgO and CO2 gas, according to reaction 5, is important with regards to the thermal balance in the roaster. MgCO3 → MgO + CO2

∆Hrxn= 117 kJ

(5)

All reactions were modeled to thermodynamic equilibrium.

3

Roaster design

3.1 Design basis Complete heat and mass balances were performed using the software METSIM® Version 14.04 for the fluid bed roaster, calcine cooler and ancillary equipment. The design case for the roaster was based on a solids feed rate of approximately 20 t/h (dry basis). Based on historical research performed on samples of Katanga ore and concentrate feed slurry as well as on previous plant experience, a bed operating temperature of 680 °C was selected. Based on thermodynamic data, the partial pressure of sulphur dioxide in the roaster bed and the bed gas temperature range have been selected to maximize production of soluble copper and cobalt and minimize production of soluble iron. Heat input to the roaster is necessary during start-up and is achieved by three removable preheat burners mounted to the roaster sidewall. Once a bed temperature of about 700 °C has been achieved, reaction of the sulphide feed slurry with oxygen in the fluidizing air will provide adequate heat to maintain operating conditions, with cooling water needed to limit the bed overheat. Residence time of solids in the roaster is in the range of 3 to 4 hours for the design case. Research and original pilot plant data performed on representative feed samples indicated a minimum retention time of 3 hours for solids in the fluid bed. An increase in retention time has been shown to increase the decomposition of sulphates; this is especially prevalent for iron sulphates and to a lesser extent for copper and cobalt sulphates.

3.2 Fluid bed sizing Results of the METSIM® heat and mass balance provided the basis for sizing of the equipment. Based on the particle size distribution of the feed as well as particle and gas properties, a minimum fluidization velocity of 0.03 m/s was identified. An actual fluidization velocity of 0.7 m/s was selected to ensure complete fluidization of the bed in the bubbling bed regime, while minimizing solids elutriation to the cyclone. Based on gas flowrate requirements and fluidization velocities, a roaster diameter of 8 m I.D. was selected. Based on retention time requirements, a bed height of 1.8 m was chosen. Dimensions of the fluid bed were selected to account for design and turndown operation. Proceedings of Copper 2010

593

Adham, Bucholz, Kokourine, Lu, Sarvinis, Tohn, Girouard As the bed temperature of the calcine cooler is lower than that of the roaster, differing gas properties result in a smaller diameter fluid bed. Consequently, an actual fluidization velocity of 0.67 m/s was selected for the 3 m I.D. calcine cooler at a bed height of 1 m. Water addition to the calcine cooler must be well controlled due to the proximity of the operating temperature to the evaporation point of water. The large flowrate of water to the calcine cooler presents the potential for cold spots, which can lead to agglomeration of particles and slumping of the bed. Roaster and cooler freeboard heights were selected based on the transport disengagement height for elutriated bed material. Resultant roaster and cooler internal dimensions are given in Table 2. Table 2:

Roaster and cooler internal dimensions

Parameter

Unit

Roaster

Calcine Cooler

Internal Height (Freeboard and Bed)

m

6.4

4.6

Internal Bed Diameter

m

8.0

3.0

Internal Maximum Freeboard Diameter

m

9.0

3.7

Bed Height

m

1.8

1.0

Splash Zone

m

1.0

1.0

Freeboard Height

m

3.6

2.6

4

Design features

4.1 Roaster vessel The presence of sulphur in the roaster feed necessitated consideration of sulphur-bearing compounds and their associated potential for corrosive attack. Of particular concern is the formation of H2SO4 below the sulphuric acid dew point of 177 °C under normal operating conditions. This served as the basis for shell temperature design for the roaster, roaster cyclone, and interconnecting ductwork. Additionally, while a high shell temperature minimizes corrosion, the strength of steel decreases above 200 °C. Thus, shell design temperatures were carefully selected in order to minimize corrosion while maintaining strength. Water condensation on the steel shell can induce corrosion and therefore shell temperatures must also be kept above the water dew point of 71 °C. Design of the calcine cooler shell considered the water dew point due to the lower operating temperature of 150 °C. In order to meet these shell temperature requirements, design of the roaster was based on analytical and Finite Element thermo-mechanical calculations performed using MathCAD-12, Solid Edge-20 and ANSYS Workbench-11.0 software packages. 594

Proceedings of Copper 2010

Design of Copper-Cobalt Sulphating Roasters Katanga Mining Limited The novel design methodology integrated for the first time the steady state and transient heat transfer analyses for a hot suspended/slumped bed inside a refractory lined, externally insulated/noninsulated fluid bed vessel. The analysis also encompassed heat diffusion through the refractory lined tuyere plate and windbox, considering forced cooling inside the tuyere stems and participating/nonparticipating radiative heat transfer gas inside the wind-box. Finite Element Analysis (FEA) was utilized in evaluating the thermo-mechanical stresses in the vessel’s structural members. The essential components of this methodology are as follows: • Development of a physical model of heat transfer for the suspended/slumped hot fluid bed and refractory lined tuyere plate with forced cooling of the tuyere stems. • Determination of typical load cases including normal operating, start-up, shutdown and upset conditions. • Development of an iterative method of calculations for the system of equations describing the physical model to define the boundary conditions for the FEA simulations. • Calibration of the model by iterative modification of key-values such that the calculated temperature field matches the experimentally measured numbers for the existing FB vessels. • Development of preferred design options for the vessel structural elements. • Design enhancement of the vessel components to allow safe operating within allowable stress intensities and deflections, as set out by international design codes, for the vessel structural members (tuyere plate, vessel shell, grillage beams, supporting brackets, and tuyere plate expansion joint) with a minimum life time of 20 years.

4.2 Tuyere and tuyere plate design As solids throughput requirements necessitated a large diameter (8 meter I.D.) fluid bed roaster, design of the tuyere plate, grillage beam, and associated supports presented a significant challenge. The tuyere plate consists of over 800 tuyeres to maintain fluidization in the large diameter bed. It was decided to arrange the tuyeres in an evenly distributed, equilateral triangle formation for improved fluidization. Tuyeres are mounted to the tuyere plate through the use of a customized tuyere coupling. This coupling offers a significant reduction in maintenance time and tuyere plate damage during replacement of the tuyeres. High temperature fluid bed vessels operate at a complex thermo-mechanical load during normal operation, start-up, shutdown and upset conditions. The large size of the KML roaster increases the risk of elevated stresses and large deformations of its structural elements. For example, the radial thermal expansion of the tuyere plate at hot shutdown or malfunction conditions could be a reason for structural failure of the tuyere plate or vessel shell, thus a specially designed expansion joint was required.

Proceedings of Copper 2010

595

Adham, Bucholz, Kokourine, Lu, Sarvinis, Tohn, Girouard

Figure 3:

Finite element analysis (FEA) stress intensity modeling

Deformation of the tuyere plate main beams due to the vertical temperature gradient can damage the tuyere plate and its refractory lining. Thermo-mechanical analysis of the roaster revealed that vertical deflection of the tuyere plate for the hot upset case (no fluidization, hot slumped bed in direct contact with tuyere plate refractory, cold fluidizing air flowing through tuyeres) is beyond an acceptable value. Thus, a specialized arrangement for the grillage beam to support the tuyere plate was implemented. Design of structural supports for the main beams was a challenge due to a large horizontal and vertical deformation under combined thermal and mechanical load. Specially designed bearing blocks were developed in order to resolve this issue. Due to the complexity of thermo-mechanical interactions between the vessel structural members, vessel structural elements were combined in one large 3D model. The defined load cases were successfully FEA simulated and the vessel design was optimized. A schematic representation of an FEA stress-intensity model is given in Figure 3. 596

Proceedings of Copper 2010

Design of Copper-Cobalt Sulphating Roasters Katanga Mining Limited

4.3 Removable preheat burners During normal operation, the preheat burner ports on the roaster sidewall will be plugged and the burners stored nearby. During start-up, the burners are mounted to the ports for firing with diesel. Combustion air is supplied to all three burners by a single, dedicated blower. Start-up of the roaster consists of firing the burners to achieve the operating temperature of about 700 °C. This is performed with an empty roaster and no fluidizing air. Once the set-point is reached, inert seeding material (dried, roasted product or sand) is introduced to the roaster via the coal bin system and slightly fluidized. With a hot bed of inert material, the combustible feed slurry is introduced and the burners are gradually taken off-line. After a period of steady operation, the burners are removed and their ports are plugged. The selection and use of removable preheat burners rather than fixed burners was due to the potentially corrosive nature of the sulphation process. Similar to the steel shell of the roaster, the preheat burners are subject to potential acid corrosion in the form of sulphuric acid. Removal of the preheat burners during normal operation limits the exposure of the burners to the corrosive environment. Design includes the use of monorails on the burner level deck to facilitate removal and safe storage of the burners.

4.4 Roof-mounted slurry feed injection system The roaster feed system consists of four slurry feed ports, three operating and one standby, mounted to the roof of the roaster. Each feed port consists of a single nozzle used for the atomization of slurry feed with plant air. Design includes control over atomization air flowrate and pressure and an installed purge air delivery system for the prevention of dust ingress to the gun. Spraying of the feed slurry onto the bed from roof-mounted injection guns allows for significant evaporation of water in the freeboard. This is an important design parameter for the roaster, as low moisture in the bed limits the following agglomerating reaction: 4CuSO4 + Cu2S → 6CuO + 5SO2

(6)

This reaction occurs between 420 °C and 540 °C, inhibiting the primary sulphation reaction (2) given above. The presence of moisture surrounding the fine sulphide particles in the feed slurry reduces the particle heating rate. Therefore, the lower the residual surface moisture at the time of contact between the feed particles and the bed, the greater the heating rate of the incoming particles in the bed. This leads to a shorter length of time the particles are in the 420 °C to 540 °C temperature range, thereby limiting the agglomerating reaction. Partial agglomeration can negatively affect conditions in the fluid bed, potentially leading to defluidization. While the fluidized bed comprising both solids and interstitial fluidizing gas is maintained at a temperature of up to 700 °C, roaster off-gas rising from the bed undergoes significant cooling in the Proceedings of Copper 2010

597

Adham, Bucholz, Kokourine, Lu, Sarvinis, Tohn, Girouard freeboard. This drop in temperature is a direct result of the injection of water from the roof-mounted water sprays. Due to the nature of the fine-particulate, viscous slurry, selection and design of the roaster feed injection nozzles was critical to the successful operation of the roaster. The particle size distribution of the bed is significantly higher than that of the slurry feed and is a direct consequence of the design of the atomized feed system. Specifically, the bed particle size distribution is determined by the spray properties of the injection nozzles. As such, design of the nozzles was critical to ensure the solid particles had the desired size distribution prior to contact with the bed. Hatch undertook the design of a custom, vertical down flow, multiphase nozzle for this application. Based on past experience at KML where nozzles were changed every several months, Hatch selected a much harder material of construction to withstand the abrasion erosion caused by the high velocity of solid particulate through the nozzle. Features of this customized nozzle include a jacketed cooling air circuit, mixing of the slurry feed and atomizing air in the injector shaft, and a shallow spray angle. Testing of the custom-designed nozzle confirmed its viability for use in the copper sulphide roaster.

4.5 Product discharge seal legs Both the roaster and calcine cooler discharge their solid product via seal legs to the calcine cooler and quench tank, respectively. The seal legs permit removal of bed material from the fluid beds at a controlled rate, while also providing a gas seal between the fluid beds and downstream equipment. Seal legs were designed to be larger than required for product discharge to permit the removal of large pieces of refractory that will inevitably break off the wall due to attrition. Thermal swings arising during start-up and shutdown of the fluid beds typically lead to gradual attrition of refractory. Seal leg design consisted of minimizing plant air consumption, while maintaining sufficient fluidizing air to ensure a reasonable margin between operating flowrate and minimum fluidizing conditions throughout both nominal and turndown operation. Seal leg rodding ports are available to facilitate maintenance in the event of a blockage. Blockages can be caused by excessively large pieces of refractory breaking off the roaster lining, as well as by settling of solids due to unexpected downtime. Rodding ports allow operators to quickly rod the seal leg, which is often enough to relieve a blockage. Seal legs were also equipped with strainers, which are designed to prevent large solids from entering and blocking the seal legs.

598

Proceedings of Copper 2010

Design of Copper-Cobalt Sulphating Roasters Katanga Mining Limited

4.6 Windbox cleaning system During normal operation, erosion or blockage of tuyeres can lead to the accumulation of siftings in the windbox. These siftings are generally fine solid particulate from the fluidized bed. High sifting levels in the windbox can affect the pressure distribution across the tuyere plate and through the freeboard. As such, removal of these siftings is necessary and typically requires shutdown of the operation to allow manual solids removal from the windbox by an operator. To increase plant availability and facilitate operation, design of the roaster and calcine cooler incorporated a fully automatic, online windbox cleaning system. Features of the system include level detection in the windbox to obtain online information on the amount of solids in the windbox. Upon reaching a defined setpoint, the solids are discharged from the windbox via a double dump valve, which maintains the windbox operating pressure and therefore does not disrupt the process.

5

Process control philosophy

The roaster plant operation is designed as fully automated, with minimum operator intervention requirements. All start-up sequences and their requirements are programmed in a step-by-step manner, through an interactive human-machine-interface (HMI). At the start of each sequence, the operator will be promoted to verify the readiness of the subsystems that are required for the initiation. Once all the requirements are verified as complete, the operator can remotely action the start. All the outstanding requirements for system start-up or function are prompted to the operator’s attention. Similar, warnings and alarms are displayed through the HMI. If the operator does not take remedial actions with regards to any critical alarms, the HMI can take step-by-step actions to relieve the alarm, e.g. by turning off a faulty system and switching to a stand-by unit, by lowering the operating load of the roaster, or by placing the faulty system in the idle mode.

6

Start-up and commissioning

Construction of the roaster and calcine cooler has been completed in the summer of 2009, with some delays due to prevailing economic conditions in the 2008/2009 period. At the time of the preparation of this paper, installation of the refractory lining is scheduled for completion at the end of October 2009, with start-up and commissioning scheduled to follow immediately.

Proceedings of Copper 2010

599

Adham, Bucholz, Kokourine, Lu, Sarvinis, Tohn, Girouard

References [1] THEYS, L.F. and LEE, L.V. (1958): Sulfate roasting copper-cobalt sulfide concentrates: Journal of Metals: 134-136 [2] THOUMSIN, F.J. and COUSSEMENT R. (1964): Fluid-bed roasting reactions of copper and cobalt sulfide concentrates: Journal of Metals: 831-834

600

Proceedings of Copper 2010

Filsulfor and Gypsulfor: Modern Design Concepts for Weak Acid Treatment

Filsulfor and Gypsulfor: Modern Design Concepts for Weak Acid Treatment Dr. Angela Ante Bamag GmbH Zum Oberwerk 6 Butzbach, Germany

Keywords: Acid treatment, off-gas, off-gas cleaning, sulfuric acid, SO2, SO3

Abstract The copper smelting process generates a process gas with high SO2, SO3, volatile heavy metals and arsenic loads which is routed to a byproduct sulphuric acid plant for the recovery of marketable sulphuric acid. In the off-gas cleaning system, SO2, volatile heavy metals and arsenic are removed from the off-gas by wet scrubbing, generating a liquid scrubbing effluent referred to as scrubbing acid or weak acid. The weak acid contains sulphuric acid as well as high concentrations of arsenic and heavy metals. The conventional treatment process for this liquid effluent stream uses a first neutralization stage to convert the sulphuric acid to gypsum by adding Ca(OH)2 and precipitate the bulk of the arsenic load as calcium arsenate. In a second step, the remaining arsenic is precipitated as ferrous arsenate by adding ferrous sulphate. Because of the high sulphuric acid concentration of this effluent stream (50 to 100 g/l), the conventional weak acid treatment process has the drawback of generating large amounts of gypsum. Moreover, the gypsum produced is contaminated with heavy metal impurities, predominantly arsenic. Compounding the problem is that the heavy metal-contaminated gypsum is not leach-resistant so that heavy metals may be resolubilized and released to the environment. The process concept here presented is a further development of this technology with the aim of quantitatively recovering the sulphuric acid present in the weak acid for reuse as secondary raw material in the sulphuric acid production process.

Proceedings of Copper 2010

601

Ante

1

Introduction

The copper ores contain heavy metals and arsenic. During smelting, a process gas is produced consisting of SO2, traces of SO3, HF, HCl, flue dusts containing heavy metals oxides and volatile arsenic trioxide (As2O3). This process gas is treated to catalytically oxidise SO2 to SO3 for production of marketable sulphuric acid. The metallurgical gases are cleaned of their harmful constituents in a scrubber system to protect the catalysts and to ensure the required quality of the H2SO4. All constituents soluble in water, especially the SO3, flue dusts and arsenic trioxide, are collected as wastewater known as weak acid. As in the current conventional treatment process, the sulphuric acid is neutralised to gypsum with the aid of Ca(OH)2 in the first cleaning stage and the majority of the arsenic precipitates as calcium arsenate. In an optional second stage, the residual arsenic is precipitated as iron arsenate through the addition of iron sulphate. Large amounts of gypsum accrue here, this arising through neutralisation of the sulphuric acid. This gypsum is contaminated with heavy metals, predominantly arsenic, and must undergo expensive disposal. The heavy metals and arsenic can be leached from this gypsum. BAMAG has therefore developed a process for a Spanish copper producer, in which marketable gypsum can be obtained by means of fractionated precipitation. The process concept presented here represents a further development with the aim of returning the reusable material (sulphuric acid) originally contained in the wash acid quantitatively or to win a pure gypsum supplying it for reuse as a secondary product. The arsenic occurring in anionic form at very low pH values is precipitated almost quantitatively by means of sulphide, the majority of the accompanying heavy metals also being separated as precipitation products. In the “Filsulfor” process the remaining heavy metals occurring in cationic form are retained by nanofiltration. The sulphuric acid obtained exhibits a high marketable quality. Apart from saving disposal space and cost reductions for the sludge treatment, there is also no need for lime and the space requirement is lower. Another treatment concept is “Gypsulfor”, the combination of the sulphide precipitation with the neutralization of the decontaminated sulphuric acid with lime producing pure gypsum usable for a wide range of application purposes.

2

Atlantic Copper: technical gypsum production

At the Spanish copper smelting plant, approximately 300,000 tons of copper are produced annually from sulphidic copper ore or concentrates. 602

Proceedings of Copper 2010

Filsulfor and Gypsulfor: Modern Design Concepts for Weak Acid Treatment

2.1 Process stages The process described below comprises the process stages shown in Figure 1. The process is divided into technical gypsum production (Part 1; items 1 to 5) and conventional residuals precipitation (Part 2; items 6 to 12). Microfiltration

Ca(OH)2 liquid

1

6

Weak acid

Polyelectrolyte

Polyelectrolyte

FeCl3, HCl

12

7

Discharge

Ca(OH)2 powdered

Precipitation stage 1

Clarifier 1

Precipitation stage 2

Clarifier 2

8

5 Precipitation stage 0

Polyelectrolyte Polyelectrolyte

2

Sludge holding tank

Vacuum belt filter

9

3 Belt filter press

10 Gypsum for sale to industry

Secured landfill

4

Pure gypsum store

Pure gypsum production

Figure 1:

11

Filter cake store

Residuals precipitation

Flow sheet for weak acid treatment plant with technical gypsum production [1]

1 Microfiltration: 2 Precipitation stage 0:

Separation of coloured solids for the production of a white gypsum Pure gypsum precipitation takes place by partial neutralisation of the weak acid down to a pH smaller than 1 3 Vacuum belt filter (VBF): The gypsum slurry from precipitation stage 0 is dewatered and washed to a residual moisture content below 35 % 4 Pure gypsum store 5 Filtrate: Filtrate accumulating from dewatering of the gypsum slurry is routed to precipitation stage 1 6 Precipitation stage 1: Gypsum precipitation, coarse As precipitation and heavy metal precipitation with milk of lime 7 Precipitation stage 2: Final As and heavy metal precipitation as hydroxides 8 Clarifier 1 and 2 9 Sludge holding tank 10 Belt filter press (BFP): Dewatering of dirty gypsum 11 Filter cake store 12 Discharge Proceedings of Copper 2010

603

Ante In fact, approx. 85 % of the produced solids are sold as technical gypsum and only 15 % are to undergo costly disposal at a secured landfill [1].

2.2 Disadvantages In order to obtain marketable gypsum (u. a. < 100 ppm arsenic), a part of the sulphuric acid is required to set a low pH value in the technical gypsum precipitation stage so as to retain the arsenic in soluble form corresponding to the dissociation weight. The residual amount of sulphuric acid is not converted to gypsum until the neutralisation stage together with the heavy metals and the arsenic and must be disposed of cost-intensively as "waste sludge". Moreover, technical gypsum is a "bulk product" whose use is restricted to consumers located nearby owing to relatively high transport costs. The disposal safety is therefore restricted from the viewpoint of the producer. Technical gypsum must occasionally be disposed of cost intensively. Economic reasons mean that application of the technical gypsum process is restricted to weak acids streams with a considerably high content of sulphuric acid and low concentrations of contaminants. In addition, the heavy metals can be leached from this gypsum contaminated with heavy metals. Furthermore, the potential recovery of valuable heavy metal hydroxides from the calcium arsenate embedded in a gypsum matrix is technically difficult and therefore not economically feasible.

3

Process development

3.1 Process requirements Due to these disadvantages BAMAG searched for a new process design. Our innovative project bases on the following process requirements:
Separation of the sulphuric acid from heavy metals and Arsenic
Process with high driving force
Recovery of a technical sulphuric acid for use of low or moderate quality requirements if ever possible In Table 1 the processes are summarized which were discarded after theoretical evaluation of their principal technical applicability. Some of them were additionally investigated by simple lab tests.

604

Proceedings of Copper 2010

Filsulfor and Gypsulfor: Modern Design Concepts for Weak Acid Treatment Table 1: Discarded Processes [2] Process Adsorption process Iron hydroxide process Pisolite Retardation RGS-11 Precipitation Thiopaques

Membrane separation process Diffusion dialysis Electro dialysis Nanofiltration Miscellaneous Solvent extraction (TBP)

Reason for exclusion pH 5-9 pH 6.5 As removal Na2SO4 + H2S 606

(1) Proceedings of Copper 2010

Filsulfor and Gypsulfor: Modern Design Concepts for Weak Acid Treatment This is followed by the actual sulphide precipitation in Equation 2 with the example of arsenic As: 3 H2S + 2 H3AsO3 ==> As2S3 + 6 H2O

(2)

When using NaHS the available hydrogen equalises the basic effect of the sodium, the pH value only shifting minimally during the addition of NaHS. The use of gaseous H2S would be highly advantageous from a technical point of view, as excess educt would be trapped and hence reusable. However, this gas is highly toxic and therefore requires extremely high safety standards, not only during application in the process, but also during transport and storage, thereby making its use infeasible [4].

3.2.2 Nanofiltration The nanofiltration was tested, as it is capable of retaining 2-valent cations at relatively moderate pressures (5-30 bar or rather 5·105-30·105 Pa) owing to its specific separating effect, while, however, letting water and sulphuric acid through. Nevertheless, arsenic cannot be separated via nanofiltration because the selectivity is much too low, as 3-valent arseniate is present uncharged or with a single negative charge and can hence pass through the membrane in the same way as sulphuric acid at very low pH values. A certain degree of Arsenic retention can be explained by the fact that the arsenic molecule together with its hydrate envelope is larger than the sulphuric acid molecule and can hence pass through the membrane. Furthermore, the arsenic concentration in the example examined is so high that scaling occurs at a retention of 50 % and higher, as the solubility limit of around 20 g/l is exceeded. The nanofiltration is therefore suitable as an after-treatment stage after the sulphide precipitation in order to clean the remaining heavy metals from the sulphuric acid [2].

4

Investigations by BAMAG

All investigations were carried out with a sample of original weak acid from a copper smelter of Chile. All theoretical and practical investigations as well as mass balances were carried out on the basis of the composition of the weak acid of this smelter.

4.1 Sulphide precipitation In Table 3 some typical precipitation results are summarized. With these pre-investigations without any optimization the residual concentration of copper, lead, tin and molybdenum met already the requirements, whereas the parameters has to be optimized for Proceedings of Copper 2010

607

Ante meeting the maximum values for As before the realization of this process. Nevertheless more than 99.9 % of the As load had been precipitated. Table 3:

Depiction of some analysis of precipitation

Parameter

Start value [ppm]

End value [ppm]

Cleaning [%]

Arsenic

11,000

6.9

100

Copper

634

0

100

Lead

86

0.1

100

Tin

430

0.2

100

Molybdenum

109

0.2

100

189

160

15

Aluminum

49

49

0

Zink

219

210

0

n.a.

n.a.

Precipitable

Partially precipitable Iron Not precipitable

Unknown Cadmium Chromium

0 kg/h Me+ = 12 kg/h H2 SO4 = 3,550 kg/h

Waste water V = 2.1 m3/h As = > 0 kg/h Me + = 167 kg/h (Zn and Fe) H2 SO4 = 390 kg/h

Waste sludge 1 t/h contaminated with heavy metals (60 % DS)

Figure 6:

612

Overall mass balance for Filsulfor

Proceedings of Copper 2010

Filsulfor and Gypsulfor: Modern Design Concepts for Weak Acid Treatment

NaHS 1 t/h (DS = 40 %)

Pure gypsum (for reuse)

0.84 t/h As < 1 mg/kg

Weak acid V = 20 m3/h As = 300 kg/h Me+= 244 kg/h H2 SO4 = 3,940 kg/h

WWTP Waste water V = As
120 Combinations

2.3 Bronze Strip Casting – 3 Lines Melting & Casting Furnaces Withdrawal Unit

Figure 3:

Cutter

Coiler

Bronze Plant, melting and casting

• Products: CopperTinBronce CS04 / 06 / 08 • Casting Performance: 1.2-2.3 t/h per line • Width: 720-1280 mm

Proceedings of Copper 2010

725

Hanusch

3

Change in Foundry Layout due to Diffusive Emissions

3.1 Refinery layout until 2007 Asarco & Holding Furnace Thomas Furnaces 1-3

Holding Furnaces Continuous Casting

Figure 4:

726

diffusive emissions

Holding Furnace Semi Continuous Casting

Layout Copper Refinery until 2007

Proceedings of Copper 2010

New Highly Efficient Rotary Furnace for Environmentally Friendly Refining Process

3.2 Environmental Situation in the Copper Foundry until 2007 3.2.1 Thomas Furnace Process • Tiltable vessel • 35 t copper capacity per batch • Gas / Air / Oxygen fired • Melting raw materials and scraps • Oxidizing / Slagging • Deoxidizing, Desulphurization • Storage and holding Furnace

Figure 5:

Copper Refinery, Thomas Furnace Process

Proceedings of Copper 2010

727

Hanusch

3.2.2 Process steps with potential for environmental improvement • Charging Scrap ⇒ partial off gas capture in upright position ⇒ diffusive emissions • Deoxidizing ⇒ poling with trees ⇒ partial off-gas capture Developing an alternative to the existing process was urgently necessary.

3.2.3 Targets for an optimized refining process • Fundamental improvement of the hygiene situation • High-efficiency process in terms of productivity and costs • Expanding the scrap inflow towards higher impurity levels • Optimized metal result • Treatment of internal slags • Developing a new process for deoxidizing – replace poling

728

Proceedings of Copper 2010

New Highly Efficient Rotary Furnace for Environmentally Friendly Refining Process

3.3 New Refining Furnace Concept for Copper Refinery 3.3.1 Process and Furnace Layout

off gas side

gas air oxygen

burner side

Figure 6:

Copper Refinery: New Refining Furnace Process

3.3.2 Features • rotating & tilting • Oxygen metallurgy • Porous plugs • Flexibility (time & raw materials)

Proceedings of Copper 2010

729

Hanusch

3.3.3 First Improvements and Results • Fundamental improvement of the hygiene situation o Optimized off gas capture by efficient hood systems o Total combustion of all components inside the furnace by counter flow • High-efficiency process in terms of productivity (new refining furnace vs. Thomas Furnace) o Natural gas consumption (specific): - 30 % o Productivity (t/h): + 20 % • Expanding the scrap inflow towards higher impurity levels o higher organic content possible (0.5 % ⇒ 2 %) o intensive processing of scraps from external suppliers o higher metallic and mineral impurity level is acceptable • Treatment of internal slags o Processing of slags optimized (from fine granules up to bigger pieces) o Slags to be sold reduced from 1.1% to 0.5 % o Copper content in slag lowered from ~ 60 % Cu to ~ 25 % Cu

3.3.4 Targets and future activities for an optimized refining process • Developing a new process for deoxidizing – replace poling o actual: deoxidizing with inert gas and charcoal o option: alternative reducing agent o in combination with adapted process technology

730

Proceedings of Copper 2010

Sulphur Capacity of the “FeO”-CaO-SiO2 Slag of Interest to the Copper Smelting Process H. M. Henao, P. C. Hayes, E. Jak The University of Queensland Pyrometallurgy Research Centre (PYROSEARCH) Brisbane, Australia

Keywords: Slag, copper smelting, liquidus, phase diagram

Abstract Optimal control of the slag chemistry in the copper smelting is essential for high recovery and productivity and requires detailed knowledge of the slag phase equilibria. However, limited data are available on the thermochemistry of the multi-component slag system “Cu2O”-FeO-Fe2O3-SiO2CaO-MgO-Al2O3-S used in copper smelting. New experimental procedures have been developed by the Pyrometallurgy Research Centre (PYROSEARCH) at the University of Queensland that have resolved a number of experimental difficulties associated with phase equilibria determination in these systems. The experimental procedures involve high temperature equilibration and quenching followed by electron probe X-ray microanalysis. This technique has been used in the present study to construct the phase equilibrium diagrams in the multi-component Fe-Si-Ca-O-S system in a range of controlled oxygen and sulphur partial pressures and temperatures directly relevant to the copper smelting operation as part of a large of fundamental experimental and modelling program on gas/slag/matte/metal system in support of sustainable copper smelting and converting technologies. A comprehensive set of experimental phase diagrams will be presented demonstrating the effect of sulphur concentration on the phase diagrams. Sulphur capacities of slag were measured and a mathematical relationship between sulphur capacity and slag composition was obtained. Accurate information on the behaviour of this sulphur-containing system Fe-Si-Ca-O-S is essential to characterization of the more complex copper and sulphur-containing system.

1

Introduction

Optimal control of the slag chemistry in the copper smelting and converting is one of the key issues influencing efficient and stable operation. Further improvements to industrial copper production processes require detailed knowledge of the slag properties. Sulphur is present in many metal exProceedings of Copper 2010

731

Henao, Hayes, Jak traction processes. In view of this, extensive experimental information regarding the sulphur solubility in a wide variety of metallurgical slags has been reported [1-11]. In spite of this, there is little information available at the temperatures and oxygen partial pressures of interest for nonferrous pyrometallurgical processing of copper sulphide concentrates. Most of the information on the sulphur content of slags corresponds to gas/slag/matte equilibrium obtained both from industrial practice and from laboratory studies [12-21]. However, it is clear from an analysis of the experimental results in gas/slag/matte equilibrium studies that sulphide matte entrainment is difficult to avoid resulting in uncertainties in slag sulphur concentration [12]. In addition, the solubilities of copper and sulphur in slag affect each other. Accurate information on this is essential for the optimization of the processes to achieve high copper recovery. A literature review indicated that only two papers are available on sulphur solubility measured by a gas/slag equilibration technique under controlled oxygen and sulphur potentials encountered in the pyrometallurgical processing of nonferrous metals [22-23]. Those papers report the studies of the Fe-O-Si-(Ca-Mg-Al)-O-S slag system equilibrated with matte in quartz, magnesium oxide or alumina crucible with data obtained by bulk analysis of slag and matte phases after physical separation of these phases. The above mentioned techniques of the gas/slag and gas/slag/matte characterization techniques have the following limitations: 1. Different oxide crystalline phases usually are suspended inside the liquid slag in different proportions. In addition, it is possible that the matte is also suspended inside the slag. It is practically impossible to achieve a complete physical separation of liquid slag from the crystalline oxide and the entrapped matte phases in order to analyse the liquid slag phase by a wet chemical method. Thus, the results obtained from the bulk analysis have uncertainties depending on the proportion and composition of solid and matte phases suspended in the slag. 2. The use of the SiO2, MgO (or Al2O3) crucibles put significant restrictions on the range of CaO/SiO2 and SiO2/Fe where the experiments can be carried out. Recently experimental procedures have been developed and successfully applied to characterize a number of complex industrial slag systems. The experimental procedures involve high temperature equilibration and quenching followed by the Electron Probe Micro-Analyzer (EPMA). This technique has been used previously by the PYROSEARCH centre at the University of Queensland to construct a number of the phase equilibrium diagrams including the one for the FeO-Fe2O3-SiO2 slag system at controlled oxygen partial pressure and metallic copper saturation at the conditions relevant to the copper converting slag system [24-27]. The technique was further developed in the present study to overpass the above mentioned experimental limitations in the previous gas/slag equilibrium. The present study addresses three specific targets.

732

Proceedings of Copper 2010

Sulphur Capacity of the “FeO”-CaO-SiO2 Slag of Interest to Copper Smelting 1 To quantify the concentration of sulphur in slag at controlled conditions of temperature, oxygen and sulphur partial pressures, and slag composition. 2 To investigate the effect of concentration of sulphur in slag on the liquidus in the FeO-Fe2O3CaO-SiO2-S system. 3 To determine the sulphur capacity behaviour in terms of slag composition at the temperatures and gas conditions relevant to the copper smelting. 4 To provide background for further characterization of Cu partitioning in presence of sulphur in the system. As part of a broader fundamental experimental and modelling program on gas/slag/matte/metal system in support of sustainable copper smelting and converting technologies, the experimental information obtained in the present study will give a better characterization of the Fe-Si-Ca-O-S system, assist in resolving discrepancies between previous sulphur capacities experimental results, generate the basic information in order to clarify the quite controversial role of the so called “sulphidic solubility” of copper when the present results be compared with a future experimental data on the gas/slag/matte equilibrium, and generate enough experimental information for a critical assessment of the available computer simulation models of sulphur capacities and phase diagram [28-30, 31, 35].

2

Experimental technique and procedure

There are a number of difficulties in obtaining accurate chemical equilibrium data for complex slag systems at high temperatures at controlled laboratory conditions. Experimental procedures have been developed that have resolved a number of experimental difficulties and have been successfully applied by the Pyrometallurgy Research Centre at the University of Queensland to a number of complex industrial slags [24-27]. The technique for phase equilibrium measurements is based on the high temperature equilibration of the synthetic slag sample in a well controlled gas atmosphere and temperature followed up by rapid quenching and analysis of the compositions of phases present. The experimental technique can be explained with reference to Figure 1. The liquid slag phase is converted on cooling into glass and crystalline solids present at high temperature remain unaltered. The quenched samples are then mounted, polished, and compositions of the liquid and solid phases are measured by Electron Probe Micro-Analyzer (EPMA) using Wavelength Dispersive Detectors (WDD).

Proceedings of Copper 2010

733

Henao, Hayes, Jak

Figure 1:

Example of initial mixture for phase equilibrium determination using the present technique.

All the equilibration experiments were conducted in a vertical reaction tube (impervious recrystallized alumina, 30 mm I.D.) in electrical resistance silicon carbide (SiC) heated furnaces. The furnace temperature was controlled to ± 1K by an alumina shielded Pt/Pt-13 wt. % Rh thermocouple placed immediately adjacent to the sample. This thermocouple was periodically calibrated against standard thermocouple (supplied by National Measurement Institute of Australia). The overall absolute temperature accuracy is estimated to be within 5 K. The atmosphere within the reaction tube was maintained at fixed oxygen and SO2 partial pressures using CO/CO2/SO2 (CO 99.5 wt. % pure, CO2 99.99 wt. % pure, 99.9 wt. % SO2) mixtures. The flow rates of gases to the furnace were controlled using glass capillary flow meters with the gas flowing from the bottom to the top of the furnace. The volumetric ratio of the gases used to achieve the selected thermodynamic oxygen partial pressure at a set temperature was calculated using the FactSage™ 5.3 thermodynamic package [31]. A DS-type oxygen probe supplied by Australian Oxygen Fabricators (AOF, Melbourne, Australia) was used to confirm the oxygen partial pressure of the experiment. This was done by directing the output gas from the equilibration furnace into a separate vertical tube furnace equipped with the DStype oxygen probe. Information regarding the gas composition and calculated and measured oxygen partial pressure measurement are provided in Table 1. It was confirmed that the results of the measurements in the present study are within the accuracy of the DS-type oxygen probe, i.e. close to log PO2 of + 0.1 units (PO2 in atm) [32].

734

Proceedings of Copper 2010

Sulphur Capacity of the “FeO”-CaO-SiO2 Slag of Interest to Copper Smelting Table 1:

Gas composition and results of DS-type oxygen probe measurements of CO/CO2 gas mixtures. Gas Mixture

Calculated Partial Pressure

LogPO2 Measured

(%Vol.)

(FactSage™ 5.3) P(atm)

by probe P(atm.)

Temperature (oC) CO

CO2

SO2

1300

12.5

76.5

11.0

1250

1.9

88.0

1250

6.0

1250 1250

PSO2

logPO2

logPS2

0.1

-8.0

-2.4

-7.98

10.1

0.1

-7.0

-5.2

-6.98

83.9

10.1

0.1

-8.0

-3.2

-7.93

10.5

27.3

62.2

0.6

-8.0

-1.7

-7.84

19.2

57.6

23.2

0.18

-8.5

-1.6

-8.41

The starting mixtures were made from CaO, SiO2, Fe2O3 powders (99.9 wt. % purity). The mixtures, selected for each sample were weighed and mixed with an agate mortar and pestle, and then pelletized. In order to confirm achievement of equilibria, it was approached from different compositions. Variable amounts of metallic Fe powder were used to alter FeO/Fe2O3 in the starting mixture. No differences were observed in the experimental results. The pellets (0.3 g) were placed inside open platinum crucible. The size of the platinum envelope, made from 0.025 mm thick foil, was 10 mm x 12 mm. All crucibles were suspended by platinum wire within the reaction tube in the furnace. A holding time of 12 hours was used to obtain equilibrium in the condensed phases. Selected experiments were repeated at different equilibration times to test and ensure achievement of equilibria. After the equilibration time, the base of the reaction tube was immersed in water and ice, and the lower rubber stopper sealing the tube removed. Then, the sample was allowed to fall directly into the ice-water. The samples were dried and mounted in epoxy resin, and polished for metallographic observation and microanalysis. Measurement of the composition of the various phases within the sample was undertaken using a JEOL JXA 8200L (trademark of Japan Electron Optics Ltd., Tokyo) Electron Probe Micro-Analyzer (EPMA) with Wavelength Dispersive Detectors (WDD). An acceleration voltage of 15 kV and a probe current of 15 nA were used. The Duncumb-Philibert ZAF correction procedure supplied with the JEOL JXA 8200L probe was applied. The standards (Charles M. Taylor, Stanford, CA) were used in the EPMA measurements were: wollastonite (CaSiO3) for Si and Ca, hematite (Fe2O3) for Fe and CaSO4 for S. The compositions were measured to accuracy within 1 wt. % for SiO2, CaO and Fe2O3 and an accuracy of 0.1 wt. % for S.

Proceedings of Copper 2010

735

Henao, Hayes, Jak Iron oxide is present in both 2+ and 3+ states, but only the Fe cation concentration was measured with EPMA. For the presentation purposes in this article, all the iron was recalculated to the ferrous state 2+ (FeO) to ensure that the projected points are unambiguously defined. Thus, all compositions are projected onto the “FeO”-SiO2-CaO sections.

3

Experimental results

Examples of the microstructures observed in the quenched samples at the gas/slag equilibrium are presented in Figure 2. The experiments were conducted in the temperature, oxygen and SO2 partial pressures ranges of interest for copper smelting. Two temperatures 1300 and 1250 oC were selected. Oxygen partial pressure of 10-8 atm and sulphur partial pressure of 0.1 atm were employed for the experiments at 1300 oC. Four sets of experiments were carried out at 1250 oC for a range of oxygen partial pressures from 10-7 to 10-8.5 atm and PSO2 from 0.1 to 0.6 atm as shown in Table 1.

Figure 2:

Backscattered SEM micrograph of quenched sample containing spinel, wollastonite, dicalcium silicate, and liquid phase.

3.1 Experimental results at 1300 oC One set of experiments was carried out at 1300 oC, PO2 of 10-8 atm and PSO2 of 0.1 atm. The pseudoternary section constructed by projection of the slag compositions onto the “FeO”-SiO2-CaO plane is given in Figure 3. The recalculated liquid compositions were determined by normalizing the original compositions measured with EPMA considering only the relative “FeO”, CaO and SiO2 736

Proceedings of Copper 2010

Sulphur Capacity of the “FeO”-CaO-SiO2 Slag of Interest to Copper Smelting concentrations. The concentration of sulphur is indicated next to the experimental points and the liquidus lines were constructed for each crystalline primary phase investigated. The experimental results showed that the concentration of sulphur in the liquid slag, at a given CaO/SiO2 or at a given SiO2/FeO wt. % ratio, increases with increasing content of FeO and CaO wt. %, respectively. The liquidus isotherms obtained at the oxygen partial pressure of 10-8 atm without SO2 gas and the wustite liquidus line at the condition of iron saturation taken from the Slag Atlas are also plotted in Figure 3 to evaluate the effect of sulphur on the liquidus [33, 35]. Using this representation, the influence of sulphur content on the liquidus and stabilities of condensed phases can be clearly seen. Stability of tridymite increases as the sulphur content in the liquid increases. The solid iron oxide, however, is unstable at these experimental conditions.

Figure 3:

Phase diagram and sulphur concentration in slag: Fe-Si-Ca-S-O system at fixed temperature of 1300 oC, PO2 of 10-8 atm and PSO2 of 0.1 atm.

Proceedings of Copper 2010

737

Henao, Hayes, Jak

3.2 Experimental results at 1250 oC Experiments at 1250 oC were conducted at PO2 varied from 10-7 to 10-8.5 atm, and PSO2 from 0.1 to 0.6 atm to evaluate the influence of PO2 and SO2 on the solubility of sulphur in slag and on the liquidus lines. The experimental results are show in Figures 4, 5 and 6. It is observed that the concentration of sulphur in slag at PO2 of 10-7 atm is < 0.1 wt. % (the limit of detection for the EPMA). The concentration of sulphur in slag at 10-8 atm increases by approximately five times with increasing PSO2 partial pressure from 0.1 to 0.6 atm. Also, it was observed that spinel is unstable at PO2 of 10-8 atm and PSO2 of 0.6 atm and at PO2 of 10-8.5 atm, and PSO2 of 0.18 atm.

4

Discussion

4.1 Liquidus It was observed in Figures 3 to 6 that PSO2 partial pressure has a small effect on the tridymite and wollastonite liquidus. However, the effect on the spinel liquidus is so strong that at some partial pressures of PSO2 spinel is unstable. This behaviour has important implications on the copper smelting operation from the viewpoint of process control and refractory protection. The diagrams presented in Figures 3 to 6 can now be used to take into account effect of sulphur in process improvement.

4.2 Sulphur capacity 4.2.1 Definition Sulphur may dissolve in slag according to the exchange reaction suggested by Richardson and Withers [1], (O2-)(slag) +

738

1 1 S2(gas) = (S2-)slag+ O2(gas) 2 2

(1)

Proceedings of Copper 2010

Sulphur Capacity of the “FeO”-CaO-SiO2 Slag of Interest to Copper Smelting

Figure 4:

Phase diagram and sulphur concentration in slag: Fe-Si-Ca-S-O system at fixed temperature of 1250 oC, PO2 of 10-7 atm and PSO2 of 0.1 atm.

which gives, within the limits of dilute solution, the relation: Cs2-=(wt. % S2-)√(PO2/PS2)=(K×aO2-)/f s2-

(2)

where Cs2- is the sulphide capacity of slag, wt. % S2- is weight pct of sulphur in the slag, PO2 and PS2 are the respective partial pressures of oxygen and sulphur in the gas phase in equilibrium with the slag and K, aO2- and f s2- represent the equilibrium constant of Equation 1, the activity of O2- in the slag and activity coefficient of S2- in the slag respectively. The values of K, aO2- and f s2- are not measurable and the sulphide capacity in the present study is obtained by chemical analysis of the wt. % S at a given temperature and partial pressures of sulphur and oxygen in the gas phase.

Proceedings of Copper 2010

739

Henao, Hayes, Jak

Figure 5:

740

Phase diagram and sulphur concentration in slag: Fe-Si-Ca-S-O system at fixed temperature of 1250 oC, PO2 of 10-8 atm and PSO2 of 0.1 and 0.6 atm.

Proceedings of Copper 2010

Sulphur Capacity of the “FeO”-CaO-SiO2 Slag of Interest to Copper Smelting

Figure 6:

Phase diagram and sulphur concentration in slag: Fe-Si-Ca-S-O system at fixed temperature of 1250 oC, PO2 of 10-8.5 atm and PSO2 of 0.18 atm.

4.2.2 Sulphur capacity at 1300 oC The values of sulphur capacity at 1300 oC calculated using Equation 2 are represented in Figure 7 showing the logarithm of Cs2- vs. SiO2/FeO wt. % ratio at four levels of CaO/SiO2 wt. % ratios.

Proceedings of Copper 2010

741

Henao, Hayes, Jak

Figure 7:

Relationship between sulphur capacity and SiO2/FeO wt. % ratio.

It is apparent that there is an approximately linear relationship. A similar linear tendency was observed for the correlation between the logarithm of Cs2- vs. CaO/SiO2 wt. % ratio. Thus, from the experimental data for a range of compositions, a mathematical relationship was obtained to express the logCs2- as a linear function of SiO2/FeO wt. % ratio and CaO/SiO2 wt. % ratio as is indicated in Equation 3, log Cs2-(1300 °C) = -1.14 × (wt. % SiO2/wt. % FeO) + 0.21 × (wt. % CaO/wt. % SiO2) - 2.18

(3)

R2 = 0.97 where -1.14, 0.21 and -2.18 are parameters empirically determined for the 1300 oC and the range of SiO2/FeO wt. % ratio from 0.05 to 1.9, and the range of CaO/SiO2 wt. % ratio from 0 to 1.3. Note that regression coefficient R2 = 0.97 is close to 1 which confirms that the relationship of log Cs2-(1300 °C) is close to linear in terms of these compositional ratios. The optimized parameters were then used to interpolate the position of iso-sulphur capacity lines in the pseudo-ternary system “FeO”-SiO2-CaO at 1300 oC. The iso-sulphur capacity (1000 × Cs) lines and the experimental results are shown in Figure 8.

742

Proceedings of Copper 2010

Sulphur Capacity of the “FeO”-CaO-SiO2 Slag of Interest to Copper Smelting

Figure 8:

Effect of slag composition on sulphur capacity: Fe-Si-Ca-S-O system at fixed temperature of 1300 oC, PO2 of 10-8 atm and PSO2 of 0.1 atm.

4.2.3 Sulphur capacity at 1250 oC Figure 9 shows the experimental results at 1250 oC for the whole range of PO2 and PSO2 investigated. The tendency presented in the figure indicates that, in the range of conditions investigated in the present experimental study, sulphur capacity is independent of oxygen and sulphur partial pressures. Thus, a mathematical equation, similar to Equation 3, was obtained to express the relationship between the sulphur capacity and slag composition.

Proceedings of Copper 2010

743

Henao, Hayes, Jak log Cs2-(1250 °C) = -1.28 × (wt. % SiO2/wt. % FeO) + 0.015 × (wt. % CaO/wt. % SiO2) - 2.41

(4)

R2 = 0.95 where -1.28, 0.015 and -2.41 are parameters empirically determined for the 1250 oC and range of SiO2/FeO from 0.2 to 1.5, and the range of CaO/SiO2 wt. % ratio from 0 to 1.3. Again, R2 close to 1 confirms linear character of the relationship between logCs2-(1250 °C) and selected composition ratios.

Figure 9:

Relationship between sulphur capacity and SiO2/FeO wt. % ratio.

Fe-Si-Ca-S-O system at the following fixed conditions of PO2 and PSO2 : a) -PO2 of 10-8 and PSO2 of 0.1 atm. b) -PO2 of 10-8 and PSO2 of 0.6 atm. c) -PO2 of 10-8.5 and PSO2 of 0.18 atm. Similar to the previous temperature, the optimised parameters were then used to interpolate and construct iso-sulphur capacity lines in the pseudo-ternary system “FeO”-SiO2-CaO at 1250 oC. The iso-sulphur capacity lines along with the obtained experimental results are shown in Figure 10.

744

Proceedings of Copper 2010

Sulphur Capacity of the “FeO”-CaO-SiO2 Slag of Interest to Copper Smelting

Figure 10: Effect of slag composition on sulphur capacity: Fe-Si-Ca-S-O system at the following fixed conditions of PO2 and PSO2: a) - PO2 of 10-8 and PSO2 of 0.1 atm. b) - PO2 of 10-8 and PSO2 of 0.6 atm. c) - PO2 of 10-8.5 and PSO2 of 0.18 atm.

4.2.4 Comparison and previous investigations and thermodynamic models The trends of Cs2- indicated by Equations 3 and 4 are similar to those determined at 1500 oC by Fincham and Richardson [5]. The agreement is good, within experimental uncertainties, with the experimental data reported by Simeonov for the system “FeO”-SiO2 at the condition of silica saturation as indicated in Figures 8 and 10 at 1300 oC and 1250 oC, respectively [23].

Proceedings of Copper 2010

745

Henao, Hayes, Jak The observed difference in the position of the liquidus points reported by Simenovmay indicate that tridymite was suspended in the liquid in that experimental study [23]. This difference must be considered in the interpretation of the Simenov’s data. The experimental results obtained by Jalkanen at 1250 oC are plotted in Figure 10 to illustrate the difficulties to obtain accurate results on sulphur capacities using the gas/slag/matte equilibrium [12]. Their experimental sulphur capacity ranges between 2 and 23, a large deviation from the results obtained by Simenov and from the present study [23]. Simenov suggested that HSC (Outokumpu, Finland) database used by Jalkanen to calculate the gas partial pressures could be in error [23]. However, this factor does not explain the Jarkanen results wide scattering in the position of the liquidus and the sulphur capacities at similar oxygen and sulphur partial pressures [12]. Large scattering in the content of sulphur in slag is also found in the gas/slag/matte equilibrium experimental results reported by Tavera, Nagamori and Roghani [13-15]. Previous studies [5, 22] have indicated that for a given slag composition, plots of logCs2- vs. the reciprocal temperature are linear with a negative slope. The data represented in Figures 7 and 9 indicate that at a given slag composition, Cs2- increases with increasing temperature. This is consistent with the previously reported trend. The results of sulphur capacity calculated using IRSD model (Slag Atlas Figure 6.6 [35]) for the “FeO”-SiO2-CaO ternary system at 1600 oC agree with the trends identified in the present study. In both studies logCs2- is a linear function of SiO2/FeO and CaO/SiO2 wt. % ratios.

5

Conclusions

The established equilibration/quenching/EPMA technique has been used to accurately experimentally investigate phase equilibria of the Fe-Si-Ca-O-S slag system at conditions relevant to the copper smelting operation. A range of conditions has been selected to address industrial interest required to characterise and describe copper slag chemistries. The experimental results have been reported in the form of the “FeO”-CaO-SiO2 phase diagrams. The influences of SO2 gas atmosphere on the phase equilibria and chemistries of silicate slags have been demonstrated. It was observed that the spinel liquidus is strongly affected by PSO2 partial pressure. However, the effects on the tridymite and wollastonite liquidus are rather small. The obtained concentration of sulphur in the liquid slag phase was used to calculate the sulphur capacity in a wide range of the slag compositions. A regression analysis gives the equations for the relationship between sulphur capacity and slag composition. It was possible to calculate the sulphur capacity in the investigated ranges of slag and gas compositions using these equations. The present results were compared directly in terms of temperature (1250 oC and 1300 oC) and slag composition (silica saturation) with published experimental results. The sulphur capacities calculated with the present experimental results are in a reasonably good agreement with those measured using the gas/slag equilibrium and in disagreement with the experimental results obtained using the gas/slag/matte equilibrium. 746

Proceedings of Copper 2010

Sulphur Capacity of the “FeO”-CaO-SiO2 Slag of Interest to Copper Smelting

Acknowledgments The authors would like to thank the following for their assistance in this project: Ms. Suping Huang and Ms. Yan Chen from the Pyrometallurgy Research Centre (PYROSEARCH) within the University of Queensland; Mr. Ron Rasch and Mr. Ying Yu from the Centre of Microscopy and Microanalysis (CMM) within the University of Queensland.

References [1]

Richardson F.D. and Withers G. (1959): J. Iron Steel Inst., Vol.165, 66.

[2]

St. Pierre G.R. and Chipman J. (1956): Trans. AIME, Vol. 206, 1474-80.

[3]

Nzotta M.M. (1997): Scan. J. Metall., Vol. 26, 169-77.

[4]

Hino M., Kitagawa S. and Ban-Ya S.: ISIJ International, Vol. 33, 36-42.

[5]

Fincham C.J.B.and Richardson F.D. (1954): Proc. R. Soc. A, 40-62.

[6]

Abraham K.P. and Richardson F.D. (1960): J. Iron. Steel Inst., Vol. 196, 309-17.

[7]

Venkatadri A.S. and Bell H.B. (1969): J. Iron. Steel Inst., Vol. 207, 1110-1115.

[8]

Bronson A. and St. Pierre G.R. (1981): Metall. Trans. B, Vol. 12B, 729-731.

[9]

Kasrud K. (1984): Scan. J. Metall., Vol. 13, 144-50.

[10] Chang A.H. and Fruehan R.J. (1990): Metall. Trans. B, Vol. 20B, 71-76. [11] Susaki K., Maeda M. and Sano N. (1990): Metall. Trans. B, Vol. 21B, 121-29. [12] Jalkanen H. (1981): Scan. J. Metall., Vol. 10, 177-184. [13] Tavera F. and Davenport W.G. (1979): Metall. Trans. B, Vol. 10B, 237-41. [14] M. Nagamori (1974): Metall. Trans. B, Vol. 5, 531-38. [15] Roghani G., Takeda Y. and Itagaki K. (2000): Metall. Trans. B, Vol. 31B, 705-12. [16] Kaiser D.L. and Elliot J.F. (1986): Metall. Trans. B, Vol. 17B, 147-57. [17] Shimpo R., Goto S., Ogawa O. and Asakura I. (1986): Can. Metall. Q., Vol. 25 (2), 113-21. [18] Biswas A.K. and Davenport W.G. (1976): Extractive metallurgy of copper, Permagon International Library, Oxford, United Kingdom. [19] Kaiser D.L. and Elliot J.F. (1986): Metall. Trans. B., Vol. 17B, 147-57. [20] Abraham K.P., Davies M.W., and Richardson F.D. (1960): J. Iron Steel Inst., Vol. 196, 309-12. [21] Davenport W.G. and Partelpoeg E.H. (1987): Flash Smelting, Pergamon Press, Oxford, United Kingdom, 4-5.

Proceedings of Copper 2010

747

Henao, Hayes, Jak [22] Li H. and Rankin W.J. (1994): Metall. Trans. B, Vol. 25B, 79-89. [23] Simenov R.S., Sridhar R. and Toguri J.M. (1995): Metall. Trans. B, Vol. 26B, 325-34. [24] Jak E. and Hayes P.C. and Lee H.G. (1995): IMM J., 1-8. [25] E. Jak, Zhao B., Nikolic, Hayes S.P.C. (2007): European Metallurgy Conference 2007. Düsseldorf. [26] Jak E. and Hayes P. (2009): VIII Intl. Conf. on Molten Slags, Fluxes and Salts, Santiago, Chile, 473-490. [27] Nikolic S., Hayes P.C. and Jak E.: Metall. Trans. B, Vol. 39B, 210-217. [28] Reddy R.G. and Blander M. (1987): Metall. Trans. B, Vol. 18B, 591-96. [29] Nilsson R., Sichen D. and Seetharaman S. (1984): Scan. J. Metall., Vol. 13, 144-50. [30] Pelton A.D., Erikson G.E., and Serano A.R.: Metall. Trans. B, Vol. 24B, 817-25. [31] Bale C.W., Chartrand P., Decterov S.A., Eriksson E., Hack K., Mahfoud R.B., Mclancon J., Pelton A.D and Petersen S. (2002): Calphad, Vol. 26, 189-228. [32] Mendybaev R.A., Becket J.R., Stopler E. and Grossman L. (1998): Geochim. Cosmochim. Acta, Vol. 62, 3131-3139. [33] Hidayat T., Henao H.M., Hayes P.C. and Jak E.: Paper submitted to Copper 2010. [34] Slag Atlas Slag Atlas (1995): Verein Deutscher Eisenhütten (VDEh), Page 126. [35] Slag Atlas Slag Atlas (1995): Verein Deutscher Eisenhütten (VDEh), Page 260.

748

Proceedings of Copper 2010

Changes in the ISASMELTTM Slag Chemistry at Southern Peru Ilo Smelter Enrique Herrera, Leopoldo Mariscal Southern Peru Copper Corporation (“SPCC”) Fundición Ilo, Punta Tablones S/N, Pacocha, Ilo Moquegua, Peru

Keywords: Pyrometallurgy, copper, slag chemistry, ISASMELTTM

Abstract Ilo Copper Smelter has been operating since February 2007 with an ISASMELTTM furnace as a single smelting unit (1,200,000 tpy). Due to the agitated state of the smelting bath, the matte and slag (fayalite) are periodically tapped out together into two Rotary Holding Furnaces. These vessels are required to provide phases separation, allowing discard quality slag and matte to be poured separately. The Ilo smelter has reduced in a progressive way the addition both lime and silica in order to decrease the total amount of slag produced in the ISASMELTTM furnace. The “sea shell” flux addition was suspended while the SiO2/Fe ratio was reduced from 0.82 to around 0.72 in order to obtain magnetite content in the slag between 8 to 10 %. The bath temperature was kept around 1185 °C and the matte grade between 61-62 % Cu. This change in the slag chemistry of the ISASMELTTM has led to lower copper losses and get a higher thermal availability in the furnace. In this paper the data and results on the slag chemistry modification are presented.

1

Introduction

Southern Peru has commissioned in February 2007 an ISASMELTTM furnace as a single primary smelting unit, which is associated with two Rotary Holding Furnaces “RHFs”. The Figure 1 shows the smelter flow diagram.

Proceedings of Copper 2010

749

Herrera, Mariscal

Figure 1:

Flow diagram of the Ilo Smelter

The ISASMELTTM technology is a bath-smelting process in a vertical refractory-lined vessel in which a specially designed submerged-combustion lance is inserted into the bath of molten material. The furnace is fed continuously with copper concentrates and fluxes; oxygen-enriched air is injected into the bath through the lance, creating very intense bath agitation and a rapid reaction rate. The bath principally consists of molten iron-silicate slag and molten copper matte. Due to the agitated state of the bath, the matte and slag are tapped out together periodically through a single tap hole to either of two RHF via water cooled copper launders. In this way RHFs are required to provide a phase separation, allowing clean slag and matte to be poured separately. RHFs also provide surge capacity between the continuous operation of the ISASMELTTM furnace and the batch Peirce Smith Converters (PSC) cycles. The design specifications of the ISASMELTTM furnace slag considered silica to iron ratio of 0.88, and a silica to lime ratio of 7.0. The addition of “sea shell” as lime flux was considered in order to reduce the slag viscosity. A matte grade of 62 % Cu and a bath temperature around 1180 °C also were fixed by considerations of productivity, bricks wear, and copper losses in slag. In the second half of 2007 was done the first change of the ISASMELTTM slag chemistry. The main purpose of this change was to avoid leaving the undissolved silica from ISASMELTTM furnace.

750

Proceedings of Copper 2010

Changes in the ISASMELTTM Slag Chemistry at Southern Peru Ilo Smelter

2

Fundamental Considerations

The slag chemistry modifications of the ISASMLETTM furnace were made on basis of the diagram showed in the Figure 2 [1]. In this diagram the slag liquidus temperature could be predicted in at oxygen partial pressure of 10-8.4 atm and at fixed Al2O3 of 6 wt. % with focus on the range of SiO2 concentrations from 30 to 50 wt. %. All iron oxide is recalculated to “FeO” for presentation purposes, the compositions in the diagram are represented using weight ratios of CaO/(CaO+FeO+SiO2), FeO/(CaO+FeO+SiO2) and SiO2/(CaO+FeO+SiO2). The CaO/SiO2 and SiO2/Fe are also shown for convenience.

Figure 2:

Liquidus in the system Al2O3-CaO-’FeO’SiO2 at PO2 = 10-8.4 atm and Al2O3 = 6 wt. %

The main results of the first SPCC slag chemistry modification are shown in Figure 2. The black circle represents the design composition (SiO2/Fe = 0.88 and SiO2/CaO = 7.0) and the liquidus Proceedings of Copper 2010

751

Herrera, Mariscal temperature is predicted to be around 1200 °C. With the first slag chemistry modification the SiO2/Fe ratio was reduced to 0.82 and it was necessary to raise the ratio SiO2/CaO from 7.0 to 7.5 in order to keep the liquidus temperature close to 1200 °C (red circle). This change meant a reduction in the addition of both fluxes to the ISASMELTTM furnace, silica and “sea shell”. During this change, the operating bath temperature was kept around of 1185 °C and the matte grade around 61 % Cu. After the change the copper content in slag was kept around historical values (0.83 ± 0.4 %). The main benefit of this change was that the amount of slag was reduced significantly, however it was noted that it could be possible to achieve a greater benefit if the silica and lime were reduced even more in the slag. In this way, the green circle is shown on the ternary diagram of Figure 2 represents the composition goal, of a new slag chemistry modification at the ISASMELTTM furnace. The decreased of the silica and lime without increasing liquidus temperature of the slag can be explained by the interaction of lime with silica [2]. When the CaO is increased, it consumes some silica in the slag which usually prevents magnetite formation. This can be explained by the fact that the CaO-SiO2 interaction in the liquid slag is stronger than the FeO-SiO2 interaction. In other words, the CaO increases the slag liquidus by enhancing the magnetite formation if there is not enough silica. 2 [CaO] slag + [SiO2] slag = [Ca2SiO4] slag

(1)

2 [FeO] slag + [SiO2] slag = [Fe2SiO4] slag

(2)

3 [FeO] slag + O2 = (Fe3O4) solid

(3)

By the other hand, the CaO is well known because it increases the amount of liquid slag at a lower temperature and therefore potentially promotes the mechanisms of mass transfer of oxygen in the ISASMELTTM furnace, increasing the efficiency of the oxygen utilization in the smelting bath.

3

IsasmeltTM Slag Chemistry Modifications

In order to decrease the amount of slag generated in the ISASMELTTM furnace, SOUTHERN PERU has carried out trials plant consisting in reduce the CaO content and decreasing simultaneously the silica added to the furnace, but maintaining both the temperature bath and the matte grade at normal values operating.

3.1 Methodology of plant trials The reduction of the CaO and SiO2 in the slag was carried out in a step by step methodology. The Table 1 shows the reduction program of the addition of “sea shell” flux to the ISASMELTTM furnace until the total suspension.

752

Proceedings of Copper 2010

Changes in the ISASMELTTM Slag Chemistry at Southern Peru Ilo Smelter Table 1:

Program to reduce the addition of the sea shell flux, for a concentrate feed rate of 160 tph Sea Shell [tph]

SiO2 / CaO

Base

6.8

7.7

Step 1

4.0

12

Step 2

0

> 20

The silica flux was added to the furnace in order to maintain the content of magnetite in the slag within the range of 8-10 % (historical reference values obtained through SATMAGAN Magnetic Analyzer). The bath temperature was kept around 1180 °C, since the liquidus temperature for the new composition of the slag was expected to be around 1200 °C, according to the model showed in Figure 2. The slag viscosity was monitored to avoid problems with tapping and copper losses, with a careful examination of the bath temperature. Since the slag mass in the furnace would be reduced, the oxygen efficiency also was monitored.

3.2 Results and discussion The Figure 3 shows the SiO2/Fe ratio versus the CaO content in the RHF slag in each phase of the plant trials. It is noted that reducing the content of CaO in the slag decreased the consumption of silica in the furnace ISASMELTTM, which shows the interaction between them.

Proceedings of Copper 2010

753

Herrera, Mariscal

Figure 3:

Silica to iron ratio versus CaO content in RHF slag

The Figure 4 shows the CaO content versus the copper and magnetite contents in slag. In this figure it can be seen a slight reduction of the magnetite and copper contents in slag when the CaO tends to decrease. After the stopping of the addition of the “sea shell” flux to the furnace, the SiO2/Fe ratio was reduced to 0.72 on average, while the SiO2/CaO ratio was higher than 20, and depended mainly of the lime content in the concentrates smelted. The lower consumption of fluxes in the ISASMELTTM furnace resulted in a slag mass reduction of around 6 %. This less amount of slag also resulted in an increase of its residence time in the Rotary Holding Furnaces.

754

Proceedings of Copper 2010

Changes in the ISASMELTTM Slag Chemistry at Southern Peru Ilo Smelter

Figure 4:

CaO versus copper and magnetite contents in RHF slag

In the Figure 5 it can be seen that when the silica/iron ratio in slag is decreased, as consequence of the reduction of the CaO content in slag, the magnetite content also is slightly decreased. This effect could be explained by the higher availability of silica free in the molten slag, which shows that CaO added to the furnace “robs” the SiO2 from the slag.

Proceedings of Copper 2010

755

Herrera, Mariscal

Figure 5:

The SiO2/Fe ratio versus magnetite contents in ISASMELTTM slag

In the first histogram of the Figure 6, it can note that with the interruption of the “sea shell” flux added to the ISASMELTTM furnace, the copper content in the slag was reduced on average from 0.87 to 0.77. The second histogram also shows the slight reduction of magnetite in the slag after stopping the lime addition.

Figure 6:

756

Histograms of copper and magnetite in ISASMELTTM slag

Proceedings of Copper 2010

Changes in the ISASMELTTM Slag Chemistry at Southern Peru Ilo Smelter The Figure 7 shows the copper content in the discarded slag from the ISASMELTTM furnace. The plant trials beginning at the end of December last year, and on March the addition of the “sea shell” was stopped.

Figure 7:

Histograms of copper and magnetite in ISASMELTTM slag

The lower copper losses could be explained by the increased residence time of the slag in RHFs as a result of their lower mass, as well as by the slight lower slag viscosity as a result of its lower content of magnetite. However the slag viscosity has not changed significantly, the “tapping times” were the same before and after of changing the slag chemistry.

Proceedings of Copper 2010

757

Herrera, Mariscal

Figure 8:

ISASMELTTM Furnace Refractory Wear Trend – II Campaign

Figure 8 shows that the rate bricks wear in the ISASMELTTM furnace has not changed with the change of the slag chemistry, which means that the liquidus temperature has not changed significantly, as bath temperature was maintained around 1180 °C. By the other hand the reduction of the mass slag did not affect the efficiency of the oxygen utilization in the smelting bath. The oxygen efficiency was monitored through the variation of the oxygen/concentrate ratio, and the oxygen content in the off-gases process. However at the end of the plant trials SPCC decided to operate the ISASMELTTM furnace with more bath height, which increased significantly the oxygen efficiency. Another important result that was observed with the modified slag chemistry, was the improvement of the heat balance of the ISASMELTTM furnace, oxygen demand and coal were reduced significantly.

4

Conclusions

The reduction of the CaO content in the ISASMELTTM slag by dropping simultaneously the SiO2 added to the furnace has proven its effectiveness and negative effect of the CaO-SiO2 interaction in the liquid slag. With this new chemistry of the ISASMELTTM slag, its quantity was reduced by ~6 % compared with the last change. Simultaneously the content of copper in the slag was reduced by ~0.05 %. 758

Proceedings of Copper 2010

Changes in the ISASMELTTM Slag Chemistry at Southern Peru Ilo Smelter There was also a significant saving in the consumption of fluxes and hence improvements the heat balances of the furnace, by requiring less oxygen and/or fuels. With the new slag chemistry the liquidus slag did not change significantly. Although the slag viscosity and associated copper losses were slightly decreased, the bricks wear rate did not increase, since the bath operating temperature did not change significantly as well as the matte grade. Finally the reduction of slag amount by reducing the CaO did not affect the oxygen efficiency of the ISASMELTTM furnace.

Acknowledgements The authors would like to thank the Southern Peru Copper Corporation for permission to develop this work as well as the publication of this paper.

References [1] HERRERA E. and MARISCAL L. (2009): “IsasmeltTM Slag Chemistry and Copper Losses in the Rotary Holding Furnaces Slag at Ilo Smelter”, Molten2009, VIII International Conference on Molten Slag, Fluxes & Salts, Santiago de Chile, Chile. [2] COURSOL P., MACKEY P.J., PREVOST Y., ZAMALLOA M (2007): “Noranda Process Reactor at Xstrata Copper – Impact of Minor Slag Components (CaO, Al2O3, MgO, ZnO) on the Optimum %Fe/SiO2 in Slag and Operating Temperature”, Cu2007 – Volume III (Book 2), The Carlos Diaz Symposium on Pyrometallurgy, Toronto, Canada.

Proceedings of Copper 2010

759

760

Proceedings of Copper 2010

Experimental Study of Phase Equilibria of Silicate Slag Systems M. Phil. T. Hidayat, Dr. H. M. Henao, Prof. P. C. Hayes, Prof. E. Jak The University of Queensland Pyrometallurgy Research Centre (PYROSEARCH) Frank White Building Brisbane, QLD 4072, Australia

Keywords: Slags, copper, silicate, liquidus temperature, phase diagrams

Abstract Phase equilibria of silicate slag systems relevant to the copper smelting operation have been experimentally studied over a wide range of slag compositions, temperatures and atmospheric conditions. The systems selected are of industrial interest and fill gaps in fundamental information required to characterise and describe copper slag chemistries. The experimental procedures developed at the Pyrometallurgy Research Centre (PYROSEARCH) have been used, which involve equilibration of mixtures at high temperatures, rapid quenching of resulting phases, and accurate measurement of phase compositions using electron probe X-ray microanalyses. It has been shown that ferrous calcium silicate (“FeO”-CaO-SiO2) slags can be characterised over a range of temperatures and oxygen partial pressures relevant to copper production operations. The influences of Al2O3 and MgO on the phase equilibria of this slag system are also demonstrated. Furthermore, silicate slag systems containing copper can be experimentally studied to investigate the influence of copper on the chemistry and phase equilibria of the silicate slags. Differences between the present measurements and previously published data are discussed.

1

Introduction

One of the important objectives of smelting/converting process is to remove impurities from the rich-metal phase (molten metal, alloy or matte) through the formation of slag phase. Most of nonferrous smelting and converting processes operate with the use of slag systems containing principally ferrous calcium silicate components (FeO-Fe2O3-CaO-SiO2 or “FeO”-CaO-SiO2). Other components can also present in the slags, such as impurities that are introduced in by the process feeds (Al2O3, MgO) or minor elements and valuable metal that are oxidized during the operations (“Cu2O”). Proceedings of Copper 2010

761

Hidayat, Henao, Hayes, Jak Knowledge of the chemistries and phase equilibria information of this slag system is required if improvement on the current operations and development of new processes are to be carried out. Each smelting/converting process operates under specific temperature and atmosphere conditions. Information of the chemistries and phase equilibria of ferrous calcium silicate slags in air or at iron metal saturation alone is not sufficient since commercial metal smelting/converting processes are carried out at intermediate oxygen partial pressures between 10-5 atm and 10-8 atm. Limited information is available at these specific conditions. The aim of the present study is to demonstrate the methodology that is now being used to provide information on the slag chemistries of these silicate slags, and to show that the gaps in fundamental information, encompassing the conditions of industrial interest, in particular copper production operations, can be filled through careful systematic laboratory-based studies. Examples of new phase equilibrium information are provided in this study on important ferrous calcium silicate slag systems without the presence of copper and silicate slag systems containing copper.

2

Experimental technique

Determination of phase equilibria of complex slag system is not straight forward. The difficulties are sometimes related with: (1) the problems in controlling the equilibration condition, (2) the contamination of the slag by the containment materials, (3) difficulties in retaining the properties of slag equilibrated at high temperature; and (4) the time-consuming sample analysis. At the Pyrometallurgy Research Centre (PYROSEARCH), an experimental technique that enables systematic investigation of phase equilibria of multi-phase metallurgical slag systems has been developed. The technique involves equilibration of mixtures at high temperatures, rapid quenching of resulting phases, and accurate measurement of phase compositions using electron probe X-ray microanalyses (EPMA). It has been shown that the technique can be applied to a wide variety of complex pyrometallurgical slag systems [5]. In the present study the investigation of the silicate slags relevant to copper production operations has been carried using this novel technique. The detailed description of the technique will be delivered in the following sections.

2.1 Preparation of oxide mixtures and containment crucible High purity oxide and metal were used as starting materials, i.e. CaO powder (calcined at 900 oC from 99.0 wt. % pure CaCO3 powder), SiO2 (99.99 wt. % pure 1-3 mm fused lump that had been ground with an agate mortar and pestle), Fe2O3 powder (99.99 wt. % pure), MgO powder (99.95 wt. % pure), Al2O3 (99.99 wt. % pure), Fe powder (99.9 wt. % pure), and Cu powder (99.7 wt. % pure). Mixtures of various compositions were prepared by accurately weighing the oxide/metal powders and mixing them thoroughly using an agate mortar and pestle. The initial compositions of the mixtures were selected such that at equilibrium there would be liquid phase in equilib-

762

Proceedings of Copper 2010

Experimental Study of Phase Equilibria of Silicate Slag Systems rium with one or more condensed phases as shown in Figure 1. Each mixture was then pressed with pressure of 40 MPa to produce a pellet weighing less than 0.2 grams.

Figure 1:

Example of initial mixture for phase equilibrium determination using the present technique [5]

To avoid contamination of the slag by the crucibles, appropriate type of containment materials must be selected. For slag systems without copper metal, platinum crucible/envelopes were used. The size of the platinum envelope was 10 mm x 12 mm, made from 0.025 mm-thick platinum foil as shown in Figure 2a. The use of platinum envelope ensures that there will not be any unexpected component introduce into the slag. It was found that under the conditions investigated only a small amount of iron was dissolved in the platinum. Although this would slightly change the bulk composition of the mixture (point b, in Figure 1), the compositions of phases at equilibrium in the multiphase equilibrium conditions (compositions of liquid, point a, and solid phases, point c, in Figure 1) will be unchanged.

Proceedings of Copper 2010

763

Hidayat, Henao, Hayes, Jak

Figure 2:

Crucible designs and suspension: (a) platinum crucible; (b) tridymite/CaSiO3 crucible; (c) spinel crucible

If investigation of slag in equilibrium with molten copper metal is to be carried out, the use of platinum crucibles should be avoided. The use of platinum leads to the progressive attack on the crucible material and the formation of copper-platinum alloy that will alter the activity of copper metal [8]. As a consequence the target equilibrium condition will be altered as well. In this case, different type of crucible that can endure the attack of molten copper metal must be selected. It has been shown previously that crucibles/substrates made from the solid of primary phase field being investigated can be successful in containing slag with copper metal [3, 4]. Depending on the initial composition of the mixture, there would be dissolution of substrate to the liquid or precipitation of solid from the liquid. In both cases accurate information on phase equilibria can still be obtained since components transferring into or from the liquid are the components under investigation. Various primary phase crucibles were used, i.e. tridymite, CaSiO3 and spinel. The tridymite crucible was made from SiO2 powder that was pressed at 30 MPa to a 13 mm-diameter, 1.2 mm thick crucible as shown in Figure 2b. The crucible was then heated in air at 1500 oC for 24 hours. The CaSiO3 crucible was made by mixing an equimolar proportion of CaO and SiO2 powders. This mixture was pressed into the same shape as that for tridymite crucible. The CaSiO3 crucible was then heated at around 1450 oC for 1 day. The spinel (Fe3O4) substrate was prepared from 99.5 wt. % pure iron foil with thickness of 0.1 mm. The pure iron foil was folded to an envelope with an open bottom as shown in Figure 2c. This envelope was oxidized for 1 hour at 1300 oC in a CO/CO2 stream resulting in an oxygen partial pressure of 10-6 atm.

764

Proceedings of Copper 2010

Experimental Study of Phase Equilibria of Silicate Slag Systems

2.2 Control of temperature and oxygen partial pressure To monitor the actual temperature surrounding the sample, a calibrated working thermocouple was placed immediately adjacent to the sample in a re-crystallised alumina thermocouple sheath as shown in Figure 3. The temperature of the experiment was continuously controlled within + 1 oC of the target temperature. It is estimated that the overall absolute temperature accuracy of the experiment is + 3 oC.

Figure 3:

Furnace design for equilibrium experiments

The oxygen partial pressure inside a closed-system furnace was controlled by introducing gas with a specific CO/CO2 ratio. In this study, equilibrations were carried out at oxygen partial pressures between 10-5 and 10-8 atm. These oxygen partial pressures required very low CO/CO2 ratios, thus premixed CO and Ar gases were used to achieve the target CO/CO2 ratios. 5 % and 20 % CO diluted in high purity Argon (Beta standards, 0.02 % uncertainty) mixtures were supplied by BOC and high purity CO2 (99.995 % pure) was supplied by Coregas. The proportion of each gas was controlled using pressure differential type flow-meter. The total flow-rate of the gas inside the furnace was

Proceedings of Copper 2010

765

Hidayat, Henao, Hayes, Jak between 400-900 ml/min and the fluctuation of the gas flow-rate was found less than 1 % of the total flow-rate. A DS-type oxygen probe supplied by Australian Oxygen Fabricators (AOF, Melbourne, Australia) was used to confirm the oxygen partial pressure of the experiment. This was done by directing the output gas from the equilibration furnace into a separate vertical tube furnace equipped with the DStype oxygen probe. By this arrangement, direct monitoring of the oxygen partial pressure during the equilibration was possible. Results of oxygen partial pressure measurement for selected experiments are provided in Table 1. The results of the measurements in the present study are within the accuracy of the DS-type oxygen probe, i.e. + 0.1 Log ( ) units [9]. For experiment on systems at metallic copper saturation, an inert atmosphere in the closed-system furnace was used to make sure that the equilibrium was regulated by the condensed phase in the system through the reaction: 2 CuMetal + Oslag, Gas = Cu2OSlag

(1)

For this purpose, high purity Ar gas (99.999 % pure) supplied by Coregas was used. In the case of experiments in air, an open system furnace was used to equilibrate sample with air from the atmosphere. This was simply done by opening the bottom end and the gas outlet connection at the top of the furnace. Table 1:

Results of DS-type oxygen probe measurements of CO/CO2 gas mixtures Log ( (oC)

)

Target

Temperature Gas Mixture

Log (

)

Measured by Probe

1200

Stream of CO2

-3.9

-3.8

1200

CO/CO2 = 0.00316

-6.0

-5.9

1200

CO/CO2 = 0.03164

-8.0

-8.0

1300

Stream of CO2

-3.5

-3.4

1300

CO/CO2 = 0.00429

-5.0

-4.9

1300

CO/CO2 = 0.01356

-6.0

-5.9

1300

CO/CO2 = 0.04288

-7.0

-7.0

1300

CO/CO2 = 0.1356

-8.0

-8.0

2.3 Equilibration technique Equilibration experiments were carried out using a vertical tube furnace with design shown in Figure 3. The specimen with suitable container material was introduced from the bottom of a vertical 766

Proceedings of Copper 2010

Experimental Study of Phase Equilibria of Silicate Slag Systems tube furnace. The specimen was suspended on a sample holder made of platinum wire. Before the specimen was raised into the hot zone of the furnace, the reaction tube was preconditioned for 30 minutes to the required temperature and oxygen partial pressure/gas compositions. The specimen was then raised into the hot zone of the furnace and pre-melting of the sample was carried out by increasing the temperature 25 oC above the target temperature. After 30 minutes of pre-melting, the temperature of the furnace was adjusted back to the target temperature. Equilibration was then carried out for 24 hours at the target temperature and oxygen partial pressure/gas condition. After the equilibration process was completed, the bottom end of the furnace was released and the base of the furnace was immersed in water. The specimen was rapidly quenched into the water by pulling the sample holder upward until the specimen was released from the sample holder. The quenched sample was dried on a hot plate, crushed into smaller pieces and mounted in epoxy resin. The mounted sample was then ground for metallographic examination with silicon carbide paper, polished using diamond paste from 6 to 0.25 µm, and carbon coated for micro-analysis.

2.4 Analysis technique

Figure 4:

Microstructures of the quenched ferrous calcium silicate slags in equilibrium with various condensed phases: (a) spinel (“FeO”), (b) tridymite (SiO2), (c) pseudo-wollastonite (CaSiO3) and (d) rankinite (Ca3Si2O7)

Proceedings of Copper 2010

767

Hidayat, Henao, Hayes, Jak Typical backscattered scanning electron microscope (SEM) micrographs of the quenched slags in equilibrium with various condensed phases are shown in Figure 4. The rapid quenching technique successfully retains the liquid slag as a homogenous glassy phase. Using JEOL 8200L EPMA with wavelength dispersive detectors, measurement of the compositions of various phases were undertaken. 15-kV accelerating voltage and 15 nA probe current were selected for the micro-analyzer operation. The Fe2O3, MgO, Al2O3, CaSiO3 (from the Charles M. Taylor Co., Stanford, CA), and Cu2O (prepared in-house using Cu2O, 99.99 wt. % purity) standards were used for calibrations. Corrections based on the Duncumb–Philibert atomic number, absorption, and fluorescence were applied. It was shown that the accuracy of the EPMA measurements was within 1 wt. % (Jak et al., 1995). It is worth noting that the EPMA measurements only provide information on the total metal cation concentrations; no information on the proportions of the same element having difference valence states can be obtained using this technique. Consequently, all iron and copper oxide concentrations in the present study are reported as “FeO” and “Cu2O”, respectively.

3

Results and discussion

In the present study, accurate information on the phase equilibria of silicate slag systems relevant to copper production has been obtained by ensuring a strict control of equilibration conditions, avoiding contamination on the specimen, and utilizing EPMA for accurate measurement of compositions of equilibrium phases. The results of the present experimental study of phase equilibria of silicate slag systems will be delivered systematically starting from “FeO”-CaO-SiO2 systems without the presence of copper to silicate slag systems containing copper.

3.1 Silicate slag systems without the presence of copper 3.1.1 “FeO”-CaO-SiO2 system at fixed oxygen partial pressures and temperatures Significant work on “FeO”-CaO-SiO2 at temperatures between 1200 oC-1350 oC, at

= 10-5-10-6 atm

was carried out by Nikolic et al. [10]. Using the established equilibration/quenching/EPMA technique, liquidus isotherms at fixed oxygen partial pressure for spinel, tridymite, and CaSiO3 (wollastonite/pseudo-wollastonite) primary phase fields were accurately obtained. It can be seen from phase diagram of “FeO”-CaO-SiO2 at =10-6 atm shown in Figure 5, the fully liquid region expands with the increasing temperature. The change of the fully liquid region is dependent critically on the primary phase. It can be seen that the positions of the liquidus in the the CaSiO3 and spinel phase phase fields both change significantly with increasing temperature. In the case of tridymite phase, the composition of the liquiduschanges only slightly with increasing temperature. Several experiments were carried out in the present study to confirm liquidus points previously reported [10], good agreement was obtained. Further measurements have been carried out to complete the liquidus

768

Proceedings of Copper 2010

Experimental Study of Phase Equilibria of Silicate Slag Systems isotherm at 1300 oC in the rankinite and dicalcium silicate phase fields, and at 1350 oC in the silica phase field.

Figure 5:

Liquidus isotherms in the “FeO”-CaO-SiO2 system at 1200 oC, 1250 oC, 1300 oC, and =10-6 atm (Data from the present study indicated by open circle and 1350 oC, at a data from Nikolic et al. [10] indicated by closed diamond)

The measurements by Nikolic et al. [10] did not cover all primary phase fields. In the present work, experiments at higher lime to silica ratios were undertaken to complete the phase diagram of “FeO”CaO-SiO2 system at temperature 1300 oC and a =10-6 atm. At lime to silica ratios higher than 1.2, rankinite or dicalcium silicate will appear in equilibrium with the liquid slag. Difficulty arises with the presence of dicalcium silicate. Dicalcium silicate undergoes a phase transformation from α’ phase to β phase during cooling and leads to disintegration of sample into small particles (the socalled “dusting” phenomenon) [11]. The ability of EPMA to carry out analysis on small area up to 5 µm facilitates the examination of these small particles. Figure 6 shows backscattered scanning Proceedings of Copper 2010

769

Hidayat, Henao, Hayes, Jak electron microscope (SEM) micrographs of quenched sample, with liquid in equilibrium with spinel and dicalcium silicate (Ca2SiO4). It can be seen that although “dusting” took place, relatively large area for EPMA analysis can still be found and composition of each phase can be obtained accurately. This example shows one of the advantageous of using the equilibration/quenching/EPMA technique.

Figure 6:

Backscattered SEM micrograph of quenched sample containing spinel, dicalcium silicate, =10-6 atm) and liquid phases (Equilibration carried out for 24 hours at 1300 oC and

The effect of oxygen partial pressure on the liquidus isotherm of “FeO”-CaO-SiO2 system at 1300 oC is shown in Figure 7 for in the range 10-5-10-8 atm. Data of Nikolic at al. [10] are indicated by closed diamonds and data of the present study are indicated by open symbols. The oxygen partial pressure determines the crystal structure of solid iron oxide presents in the system and their =10-8 atm wustite is stable; with the increase of oxygen partial pressure spinel stabilities. At gradually becomes more stable and starts to significantly reduce the fully liquid region. The effect of oxygen partial pressure on the stabilities of tridymite and pseudo-wollastonite is less pronounced. Slight changes in the liquidus of pseudo-wollastonite primary phase field start to be noticed at lime to silica ratios higher than unity. The two diagrams are relevant to copper production operations, i.e. copper smelting that runs at between 10-8 atm and 10-9 atm and copper converting that operates at

between 10-5 atm and 10-6

atm. In actual processes other components may also present, such as sulphur, alumina, magnesia, and copper. The “FeO”-CaO-SiO2 phase diagram forms the basis of the equilibria involving those elements and knowing the nature of this basic diagram facilitates understanding on the more complex multi-component systems.

770

Proceedings of Copper 2010

Experimental Study of Phase Equilibria of Silicate Slag Systems

Figure 7:

=10-5 atm, 10-6 atm, 10-7 atm, and 10-8 atm “FeO”-CaO-SiO2 system at 1300 oC and (Data from the present study indicated by open symbols and data from Nikolic et al. [10] indicated by closed diamond.

3.1.2 The influences of Al2O3 and MgO Experimental results of phase equilibria of “FeO”-CaO-SiO2-Al2O3 system at 1300 oC and

=10-8 atm

are provided in Figure 8. The results are presented as projections onto the pseudo-ternary diagram of “FeO”-CaO-SiO2. Using this representation, the influence of Al2O3 concentration in the liquid on the liquidus and stabilities of condensed phases can be seen clearly. Stabilities of tridymite and pseudo-wollastonite decrease as the Al2O3 concentration in the liquid increases. Conversely, the stability of solid iron oxide increases with the increase of Al2O3 concentration in the liquid. Figure 9 shows the experimental results of phase equilibria of “FeO”-CaO-SiO2-MgO system at 1300 oC and =10-8 atm. Again, the results are presented in pseudo-ternary diagram of “FeO”-CaO-SiO2. The Proceedings of Copper 2010

771

Hidayat, Henao, Hayes, Jak minor addition of MgO clearly moves the liquidus isotherms of “FeO”-CaO-SiO2 slag. The liquidus of tridymite and pseudo-wollastonite primary phases appear on MgO addition to behave in a similar way to Al2O3. The information of the influence of Al2O3 and MgO on the phase equilibria of “FeO”-CaO-SiO2 system is relevant to copper smelting since at most smelting operations there are significant amounts of the two components present.

Figure 8:

Liquidus isotherm in the “FeO”-CaO-SiO2-Al2O3 system at 1300 oC and

=10-8 atm

projected into “FeO”-CaO-SiO2 plane

772

Proceedings of Copper 2010

Experimental Study of Phase Equilibria of Silicate Slag Systems

Figure 9:

Liquidus isotherm in the “FeO”-CaO-SiO2-MgO system at 1300 oC and

=10-8 atm

projected into “FeO”-CaO-SiO2 plane

3.2 Silicate slag systems containing copper During copper smelting/refining operations, loss of copper into the slag is unavoidable. The loss of copper into the slag can be due to: (1) molten copper metal that is physically entrained in the slag phase; (2) copper metal that is chemically oxidized, which then combines with the slag. The latter can be understood by investigating the phase equilibria of slag systems containing copper.

3.2.1 “Cu2O”-SiO2 systems Better understanding on the behaviour of silicate slag systems containing copper can be obtained by analysing the systems starting from the lowest order system, i.e. “Cu2O”-SiO2. The pseudo-binary Proceedings of Copper 2010

773

Hidayat, Henao, Hayes, Jak phase diagrams of the “Cu2O”-SiO2 in equilibrium with air and molten copper metal are provided in Figure 10. The different conditions mean that the samples must be contained in different crucibles. In the case of equilibration in air, platinum envelope is sufficient to contain the slag. In the case of equilibration at copper metal saturation, crucibles made of primary phase solid must be used to avoid aggressive attack of the molten copper metal.

Figure 10: Pseudo-binary phase diagrams of the “Cu2O”-SiO2 system: =0.21 atm), (b) equilibrated with molten copper metal (a) equilibrated in air ( The phase diagram for “Cu2O”-SiO2 equilibrated with air is provided in Figure 10(a). In the subsolidus region, SiO2 may present in equilibrium with CuO or Cu2O. The transformation of CuO to Cu2O was measured using differential scanning calorimetry technique and was found to be 1030.6 oC. The phase assemblage between Cu2O and SiO2 is stable up to the eutectic point of 1054 oC, determined by using quenching and microscopy examination technique. Slightly different measurements of the eutectic temperature of 1060 + 10 oC have been reported by Berezhnoi et al. [1] and Gadalla et al. [2]. In term of copper oxide content, the liquidus and eutectic obtained by previous investigators are significantly higher than the present result. For example, Berezhnoi et al. [1] reported eutectic composition of around 92 wt. % Cu2O-8 wt. % SiO2, while the present result shows eutectic located at around 87 wt. % Cu2O-13 wt. % SiO2 as shown in Figure 10(a). A well quenched slag in the cuprite primary phase is difficult to obtain. Berezhnoi et al. [1] appears to carry 774

Proceedings of Copper 2010

Experimental Study of Phase Equilibria of Silicate Slag Systems out some measurements at this primary phase. He reported melting point of pure cuprite in air to be 1230 oC; this is not in agreement with the Cu-O system constructed by Robert and Smith [12] from which it is expected that the highest temperature where pure cuprite is stable in air is around 1135 oC. Figure 10(b) shows the “Cu2O”-SiO2 system in equilibrium with copper metal. The liquidus points in tridymite primary phase obtained by the present study, within the experimental uncertainty, agree with those of previous investigators [6, 7]. It was found that the presence of molten copper metal significantly changes the liquidus and eutectic of “Cu2O”-SiO2. The stability area of fully liquid slag equilibrated with molten copper metal is smaller than that in air. For example at 1300 oC in the presence of molten copper, tridymite starts to precipitate at 13.8 wt. % SiO2. In air at the same temperature, the tridymite starts to precipitate after slag contains silica higher than 16.9 wt. %. Phase assemblage between “Cu2O”-SiO2 at copper saturation stabilizes to higher eutectic temperature of 1186 oC [6]. The increase of eutectic point was confirmed by the present experiment at 1150 oC showing the presence of only Cu2O and SiO2 solids. Measurement of the liquidus at cuprite primary phase field was not carried out by the present and previous studies. Kuxmann and Kurre [6], however, predicted that the liquidus line would connect the eutectic point and the melting point of pure Cu2O at copper saturation, i.e. 1236 oC. The investigation of system at metallic copper saturation is important since it dictates the oxygen potential limit for the converting process, above which all copper will be oxidized into the slag.

3.2.2 “Cu2O”-“Fe2O3”-SiO2 systems The phase diagram of “Cu2O”-“Fe2O3”-SiO2 system in equilibrium with air is shown in Figure 11. The primary phases present in the system include tridymite, cristobalite, cuprite, spinel and delafossite. The spinel and delafossite phase boundaries for the diagram were constructed using data from the “Cu2O”-“Fe2O3” join [13]. The ternary eutectic temperature was determined to be 1040 +10 oC, using the quenching and microscopy examination technique. Extensive measurements were carried out in the tridymite primary phase fields. The boundaries between tridymite, cristobalite, and two liquids were predicted from the pseudo-binary data to give an approximate view of the systems. It is clear that the liquidus for the silica primary phase field runs almost parallel to the “Cu2O”-“Fe2O3” join. Further work is now being undertaken to characterise these complex “Cu2O”-FeO-Fe2O3-SiO2 and “Cu2O”-FeO-Fe2O3-CaO-SiO2 systems at metallic copper saturation for a range of temperatures and oxygen partial pressures.This will provide new fundamental data on these systems that will be of use to metallurgical practice and the development of improved thermodynamic models of these slag systems.

Proceedings of Copper 2010

775

Hidayat, Henao, Hayes, Jak

Figure 11: Liquidus isotherms in the “Cu2O”-“Fe2O3”-SiO2 system in equilibrium with air =0.21 atm) (

4

Conclusions

It has been shown that the established equilibration/quenching/EPMA technique can be used to accurately study phase equilibria of silicate slag systems relevant to the copper smelting/refining operations. It has been demonstrated that a wide range of conditions can be characterised to address issues of industrial interest and fill gaps in essential information required to characterise and describe copper slag chemistries. The important influences of gas atmosphere, temperature, and additional components on the phase equilibria and chemistries of silicate slags have been demonstrated. The experimental results have been reported in the form of phase diagrams of “FeO”-CaO-SiO2, “Cu2O”-SiO2, and “Cu2O”-“Fe2O3”-SiO2. Moreover, the influences of Al2O3 and MgO on the ferrous calcium silicate slag system have been shown in the form of projections onto the 776

Proceedings of Copper 2010

Experimental Study of Phase Equilibria of Silicate Slag Systems “FeO”-CaO-SiO2 plane. The information can be used to improve the industrial operations and for the refinement of thermodynamic databases.

Acknowledgements The authors would like to thank the following for their assistance in this project: Ms. Belinda Chen and Dr. Baojun Zhao from the Pyrosearch Centre at The University of Queensland; Mr. Ron Rasch and Ms. Ying Yu from the Centre for Microscopy and Microanalysis (CMM) at the University of Queensland. The Australian Research Council Linkage program, BHP Billiton Olympic Dam Operations, Rio Tinto Kennecott Utah Copper Corporation, Xstrata Copper and Xstrata Technology for their financial support for the project.

References [1]

BEREZHNOI A.S., KARYAKIN L.I., DUDAVSKII I.E. (1952): Doklady akademii nauk SSSR, 83: 399-401.

[2]

GADALLA A.M., FORD W.F., WHITE J. (1963): Equilibrium relations in the system CuOCu2O-SiO2, Transactions of the British Ceramic Society, 62: 42-66.

[3]

ILYUSHECHKIN A., HAYES P.C., JAK E. (2004): Liquidus temperatures in calcium ferrite slags in equilibrium with molten copper, Metallurgical and materials transactions B: process metallurgy and materials processing science, 35B: 203-215.

[4]

JAK E., HAYES P.C., LEE H.G. (1995): Improved methodologies for the determination of high temperature phase equilibria, Kor. IMM J., 1: 1-8.

[5]

JAK E., ZHAO B., NIKOLIC S., HAYES P.C. (2007): Experimental measurement and prediction of complex phase equilibria in industrial non-ferrous slag systems, European metallurgy conference 2007; Düsseldorf.

[6]

KUXMANN U., KURRE K. (1968): Miscibility gap in the system copper-oxygen and its alteration by the oxides CaO, SiO2, Al2O3, MgO.Al2O3, and ZrO2, Zeitschrift fuer erzbergbau und metallhuettenwesen, 21: 199-209.

[7]

LANDOLT C. (1969): Equilibrium studies in the system copper-silicon-oxygen, Materials science; Pennsylvania State University.

[8]

LANDOLT C., MUAN A. (1969): Activity-composition relations in solid Cu-Pt alloys as derived from equilibrium measurements in the system Cu-Pt-O at 1000 and 1200 oC, Transactions of the metallurgical society of AIME, 245: 791-796.

Proceedings of Copper 2010

777

Hidayat, Henao, Hayes, Jak [9]

MENDYBAEV R.A., BECKET J.R., STOPLER E., GROSSMAN L. (1998): Measurement of oxygen fugacities under reducing conditions: Non-nernstian behavior of Y2O3-doped zirconia oxygen sensors, Geochim. Cosmochim. Acta, 62: 3131-3139

[10] NIKOLIC S., HAYES P.C., JAK E. (2008a): Phase equilibria in ferrous calcium silicate slags: Part I. Intermediate oxygen partial pressures in the temperature range 1200 oC to 1350 oC, Metallurgical and materials transactions B, 39: 179-188 [11] NIKOLIC S., HAYES P.C., JAK E. (2008b): Phase equilibria in ferrous calcium silicate slags: Part IV. liquidus temperatures and solubility of copper in “Cu2O”-FeO-Fe2O3-CaOSiO2 slags at 1250 oC and 1300 oC at an oxygen partial pressure of 10-6 atm, Metallurgical and materials transactions B, 39: 210-217 [12] ROBERTS H.S., SMYTH F.H. (1921): The system copper: cupric oxide: oxygen, Journal of the American Chemical Society, 43: 1061-1070 [13] YAMAGUCHI, T. (1966): Phase relations in the ferrite region of the system Cu-Fe-O in air, Fujihara Mem. Fac. Eng. Keio University; Tokyo.

778

Proceedings of Copper 2010

Dryer Fuel Reduction and Recent Operation of the Flash Smelting Furnace at Saganoseki Smelter & Refinery after the SPI Project Mitsumasa Hoshi, Katsuya Toda, Tatsuya Motomura, Masaharu Takahashi, Yushiro Hirai Saganoseki Smelter & Refinery Nikko Smelting & Refining Co., Ltd. 3-3382 Saganoseki, Oita, Japan

Keywords: Flash dryer, heavy oil consumption, energy saving, waste heat recovery

Abstract In 1996, Saganoseki Smelter & Refinery integrated its two flash smelting furnaces into one while maintaining smelting capacity to reduce operating costs and maintain global competitiveness. Many improvements to expand the feeding capacity to the flash smelting furnace resulted in a feed of 160 dry-mt/h and a smelting capacity of 450,000 mt/a in 1999. However, the copper content of the concentrates decreased and we estimated that anode production was limited. Therefore, the Saganoseki Process Innovation project was carried out from 2003 to 2007 to expand smelting capacity. As a result of modifications and improvements, the feeding capacity to the flash smelting furnace reached 215 dry-mt/h and the smelting capacity reached 480,000 mt/a. This paper introduces the recent operation of the flash smelting furnace (hereinafter FSF) and some improvements of concentrate flash dryer (hereinafter FSF dryer) for energy saving.

1

Introduction

In 1916, Saganoseki Smelter & Refinery started operation as a custom smelter of Nippon Mining Co., Ltd. (currently NIPPON MINING HOLDINGS, INC.) The No.1 FSF started operation in 1970, followed by the start of the No.2 FSF in 1973. Their total smelting capacity of copper was 330,000 mt/a in 1989. In 1996, two FSFs were integrated into one, while maintaining smelting capacity 330,000 mt/a of copper [1]. Since then, the smelting capacity had been executed to 450,000 mt/a in 1999 [2]. Proceedings of Copper 2010

779

Hoshi, Toda, Motomura, Takahashi, Hirai Thereafter, because the decrease of copper grade in concentrates were expected, Saganoseki Process Innovation project (hereinafter SPI project) was executed to expand smelting capacity [3]. After SPI Project, various improvements were made at Saganoseki FSF dryer to reduce the heavy oil consumption and CO2 gases. This paper introduces outline of the improvements made for FSF by the SPI project and some modification for the FSF dryer to reduce the heavy oil consumption.

2

Schematic Flow Sheet of Saganoseki Smelter

Figure 1 shows the schematic flow of Saganoseki Smelter & Refinery. Its smelting process adopts the FSF and PS (Peirce Smith) converters. One FSF operation is conducted at the maximum feed rate of 215 dry-t/h. Recent matte grade is 63–68 %. Saganoseki Smelter & Refinery has four converters, which are operated under the 3 hot-2 blowing system. The anode furnace operation is conducted with three anode furnaces and two anode casting machines. A sulfuric acid plant processes exhaust gases from the FSF and the converters. The sulfuric acid plant has two lines (FSF off gas and converter off gas) of gas scrubber & cooler and three systems of sulfuric acid production plant. Conc. F lux Coke Powder 3

3

8 ,0 0 0 N m /h 2 4 ,0 0 0 Nm /h C yclone

PSA

B lending Ya rd Oxygen Pla nt A cid Pl a nt

Dryer

Oxygen W aste He at

2 0 0 t /h Slag Flotation Concentrate 6 8 %-C u Matte

Bo ile r

El ect ri c Furna ce

Slag Fl a s h S m el t i ng Furna ce Wa s te Hea t B oi l er

Converter Slag

Blister Copper

Slag Refined Copper

Anode

Air+Oxygen C onvert er

A node Furna ce

C a s ti ng W heel

Flot a t i on

Electrolytic Copper T ailing

Figure 1: 780

Ta nk Hous e

Schematic flow sheet of Saganoseki Smelter & Refinery Proceedings of Copper 2010

Dryer Fuel Reduction and Recent Operation of FSF at Saganoseki

3

Recent Operation of FSF at Saganoseki Smelter & Refinery

Saganoseki Smelter & Refinery has increased smelting capacity through large investments such as on the integration of its two flash furnaces into one (in 1996) and the SPI project. Figure 2 below shows changes of copper production at Saganoseki smelter. The smelting capacity of the smelter reached 450,000 mt/a in 1999, sustained by the integration of the furnaces mentioned above and other measures to increase capacity. However, the copper content of the concentrates tended to decline (see Figure 3), so we estimated that it would be difficult to maintain the preceding production of copper. The SPI project was carried out from 2003 to 2007 to expand smelting capacity. 500

C opper P roduction [103 t/year ]

450 400 350 300 250 200 150 100 50 0 1994

2000

’02

’04

’06

’08

Annual production of copper at Saganoseki Smelter Cu Content in Concentrates S/Cu ratio in Concentrates 35.2

1.150 1.073 32.6

Cu％

38.0 37.0 36.0 35.0 34.0 33.0 32.0 31.0 30.0 29.0 28.0 27.0 26.0

32.6

32.1

0.995

0.952

31.4 29.7

0.920 0.900

28.5

0.848

1.24 1.20 1.16 1.12 1.08 1.04 1.00 0.96 0.92 0.88 0.84 0.80 0.76

2008

2007

2006

2005

2004

2003

2002

2001

Figure 3:

’98

S/Cu

Figure 2:

’96

Changes in the copper content of concentrates and S/Cu ratio

Proceedings of Copper 2010

781

Hoshi, Toda, Motomura, Takahashi, Hirai Table 1 below shows major items executed in the SPI project. According to the major targets, the capacities of the FSF and the sulfuric acid production process were increased to against with the declining copper content of the concentrates; facilities were consolidated for the converter and sulfuric acid gas scrubbing processes with the target of environmental preservation and reducing equipment maintenance costs. The ISA Process was applied at the tank house to improve the quality of copper cathode. Table 1: Major items executed in the SPI project Process Items

Target

Flash smelting Installation of a new flux mill furnace Increase the cooling capacity of FSF Increase the feed rate Increase matte grade (65 % → 68 %) Increase the waste heat boiler(WHB) capacity Converter furnace

Reduce the number of converters (6 → 4 furnaces) Change operating cycle of converters Cost saving (4 hot-2 blow →3 hot-2 blow) Simplify layout of off gas line Modification to converter boiler

Energy saving

Installation of SO3 coolers and pre-converters

Increase the feed rate and energy saving

Modification to the gas scrubber and cooler

Cost saving

Refinery

Introduction of the ISA process

Quality improvement

Utility

Increase the oxygen production capacity

Increase the feed rate

Sulfuric acid

(PSA x 2 units installed) Table 2 below compares FSF operating conditions before and after capacity was increased. As a result of the actual projects in the table above, the maximum smelting capacity was increased to 480,000 mt/a and the maximum feed rate to 215 dry-mt/h, respectively.

782

Proceedings of Copper 2010

Dryer Fuel Reduction and Recent Operation of FSF at Saganoseki Table 2:

Comparison of operation conditions 1995 Unit

Feed rate of FSF S/Cu in Feed Conc. Matte grade Total flow rate Oxygen enrichment Process air temperature Smelting capacity

1996

Two FSF One FSF operation operation

Dry-mt/h

70 0.90 60 730 24 953 330

% Nm3/min % ºC mt/a

1999

2003

2007

High Intensive operation I

High Intensive operation II

After SPI project

160 0.87 65 503 72 25 450

170 0.93 67 585 69 25 450

215 1.03 68 684 76 25 480

140 0.91 61 350 81 25 330

Figure 4 is a flow diagram of the concentrates drying system at Saganoseki Smelter & Refinery. Heavy oil is combusted by the burner installed at the upper end of the horizontal combustion furnace, whereas combustion air and diluted air for temperature adjustment are fed from the side of the combustion furnace. Conventionally, we replaced part of the diluted air with off gas issued from the post-process, anode furnace (approx. 260 ºC) to reduce heavy oil consumption. In recent years, the unit consumption of heavy oil has been further reduced using waste heat recovered from the SO3 cooler at the sulfuric acid plant and steam from the waste heat boiler of the FSF. Big stack No.1 Bag filter off gas fan Combustion air fan 200～300 Nm3/min

Bag filter

Dr outlet fan

Diluted air fan

off gas Receiving bin Heavy oil 800～1100 l/h

Heat exchanger

Feed rate : max 213 wet-mt/h Water content in feed ：8～ ～10%

Steam 4.0～6.0 t/h

Preheat air from SO3 cooler

Temperature of Dr ： 75℃ ℃ Moisture in Concentrates ：0.5%

Pneumatic transportation tube

Flow conveyor

Concentrates feeder

Combustion Chamber Cage mill

Anode furnace off gas fan

Rotary kiln 500℃

Concentrates burner

200℃

FSF

Figure 4:

Schematic drawings of FSF dryer

Proceedings of Copper 2010

783

Hoshi, Toda, Motomura, Takahashi, Hirai

4

Recovery of waste heat from the SO3 cooler at the acid plant

The sulfuric acid plant of Saganoseki should ensure an average conversion rate of 99.60 % or more, to keep the environmental regulation. On the other hand, SO2 concentration of inlet gas during operation should be increased for the energy saving and the increasing production capacity. Saganoseki owned three lines of acid production system. These three systems are called B, D, and E respectively. Largest system E had to set the SO2 concentration of gases to low level, compared to the other two systems. If the SO2 concentration of the gas is increased, temperature in the 3rd and 4th catalyst layers is increased. In this case, the conversion rate is decrease drastically. Consequently, to maintain the temperature of the catalyst layers during actual operation, SO2 concentrations of inlet gas had to keep low level, which considerably restricted the production capacity in system E. In addition, the unit consumption of electricity of the SO2 blower was high due to plant operation at low SO2 concentrations. Under these circumstances, the SO3 cooler was installed the inlet of the third catalyst layer in December 2004 (see Figure 5 below). Feeding air preheated by the SO3 cooler into the FSF dryer to reduce heavy oil consumption.

Figure 5:

5

Flow diagram of E system at sulfuric acid plant

Installed a steam heat exchanger for dryer diluted air

In February 2008, it’s installed a steam heat exchanger to reduce the unit consumption of heavy oil by preheating the diluted air. Figure 6 shows the flow diagram of the steam heat exchanger.

784

Proceedings of Copper 2010

Dryer Fuel Reduction and Recent Operation of FSF at Saganoseki It is designed so that steam produced by cooling the off gas from the FSF with the waste heat boiler can be used for power generation. The diluted air is preheated by the steam heat exchanger using steam (3.0 MPa, 4.0~6.0 t/h) that is branched from the main steam pipework for the power generating turbine. The horizontal fin-tube type heat exchanger has a heat transfer area of 2184 m2 and its maximum heat exchanger duty is 18,440 MJ/h. That heat exchanger making 200 ºC diluted air and supply to FSF dryer for reduce the unit consumption of heavy oil.

Steam Off gas

4.0MPa 50~52 t/h W aste He at Bo ile r

Power generating turbine

Flash Smelting furnace

Waste Heat

Steam

Boiler

3.0MPa 4.0～6.0 t/h

Steam heat exchanger Preheated air Flash furnace dryer

200℃ ℃

Drain

Condenser Atmosphere air 600～1000 Nm3/min

Diluted air fan Figure 6:

6

Flow diagram of the steam heat exchanger for preheating diluted air in the dryer

Increasing the size of the rotary kiln

To increase the feed rate of the FSF, by the dryer rotary kiln expanded in November 2008. Increasing the size of the kiln has enabled to reduce the unit consumption of heavy oil by improving the drying efficiency. Table 3 below compares the specifications of the rotary kiln between before and after kiln size was increased. Furukawa Industrial Machinery Co., Ltd was designed and manufactured this kiln.

Proceedings of Copper 2010

785

Hoshi, Toda, Motomura, Takahashi, Hirai Table 3:

Comparison of specifications of rotary kiln between before and after increasing kiln size Before increasing kiln size After increasing kiln size (ø 2400 mm) (ø 3100 mm) Internal diameter ø 2400 mm ø 3100 mm Length 11,000 mm 10,995 mm Volume 50 m3 83 m3 Average gas flow velocity 21 m/s 12.6 m/s inside the kiln Evaporated water content 15.8 t/h 20.3 t/h

7

Reasons for selecting a larger for kiln size

Figure 7 shows an image of the movements of charged concentrates in the kiln. The charged concentrates placed at the bottom of the kiln are transported and lifted up by a lift bar and dropped on the upper portion of the work-in-process concentrates that are retained or are passing under. These operations are repeated until the concentrates fall on the cage mill at the exit of the kiln. The charged concentrates are then stirred and have good condition to make contact with hot blasts when they drop. As the dropping height increases, the contact time of the hot air with concentrates increases, thereby improving drying efficiency. Figure 8 shows comparisons of the movements of the charged concentrates in the kiln before and after kiln size was increased. Before kiln size was increased, the dropping height of the ores was 840 mm, and stirring and dispersion of work-in-process concentrates in the kiln were considered to be insufficient. Raising the hot blast temperature was studied as a means to improve drying efficiency. In this case, it was considered the heat transfer from hot blast to the moisture was not enough because the dropping height was low. On the other hand, increasing hot blast was also studied as another method aimed at increasing the caloric content of air. However, it was considered that this option might be likely to reduce drying efficiency due to the shorter residence time inside the overall dryer (especially in the pneumatic transportation tube). Taking into account the above factors, we selected the option of increasing kiln size to raise the drying efficiency of the charged concentrates.

786

Proceedings of Copper 2010

Dryer Fuel Reduction and Recent Operation of FSF at Saganoseki

Lifter bar

An image 鉱石の動きイ drawing メージ of concentrates moving

Figure 7:

Image view of concentrates movement in the kiln

落下高さ 840㎜ 840 mm 落下高さ 850㎜ 1,850 1, mm

Before

Figure 8:

After

Comparison of movements of concentrates in the kiln before and after increased

Proceedings of Copper 2010

787

Hoshi, Toda, Motomura, Takahashi, Hirai

8

Calculation of kiln volume

When determining kiln volume, the following formula was used for calculations. The caloric content required for evaporating the water content was calculated on the supposition that the kiln’s contribution to drying, which accounts for the overall dryer facilities, is 40 % and the water content of concentrates in the kiln should be reduced by 3.5 %, where, the kiln’s contribution to drying means the ratio of drying degree that is accounted for by the kiln in the overall concentrates drying system. In contrast, regarding the contributions of other facilities, the cage mill’s contribution to drying was assumed to be 36 % and that of the pneumatic transportation tube was assumed to be 26 %. V[m3] =

Q [kcal/hr] 3

q [kcal·m ·hr·ºC] × ∆T [ºC] (1)

V: Dryer capacity Q: Caloric content due to water evaporation q: Comprehensive thermal capacity coefficient ∆T: Average temperature difference Figure 9 below shows changes in the unit consumption of heavy oil at the dryer before and after the improvements were made. After commencing waste heat recovery from the SO3 cooler at the acid plant in December 2004, the unit consumption of heavy oil could be reduced from 8.09 L/t (annual average for 2002 and 2003) to 6.60 L/t (annual average for 2005). Due to the operational start of the steam heat exchanger in February 2008, it was also reduced to 5.00 L/t (semiannual average for the first half of 2008). Furthermore, it could be reduced to 4.57 L/t (semiannual average for the first half of 2009), sustained by the increased size of the rotary kiln in December 2008 and increased steam volume at the steam heat exchanger since March 2009 (4.1 mt/h → 5.1 mt/h).

788

Proceedings of Copper 2010

Dryer Fuel Reduction and Recent Operation of FSF at Saganoseki

Unit consumption: 8.09 L/t (annual average for 2002 and 2003)

9.50

9.00

Recovery of waste heat from the SO3 cooler (system E) at the sulfuric acid plant: Started in December 2004

S pecific consum ption of heavy oil (L /t – Total charges)

8.50

Steam heat exchanger started operation: February 2008

8.00

Regular repair in November 2008: Increasing kiln size and increasing steam consumption in heat exchanger (4.1 → 5.1 t/h)

7.50

7.00

6.50

6.00

Unit consumption: 6.60 L/t (annual average for 2005)

5.50 5.00

Steam heat exchanger started operation: Reduction of specific consumption 5.00 L/t (semiannual average for the first half of 2008)

4.50

4.00 09/6

08/12

08/6

07/12

07/6

06/12

06/6

05/12

05/6

04/12

04/6

02/4

Figure 9:

Changes in Unit Consumption of Heavy Oil at the Dryer

Figure 10 shows the caloric content required for drying (incoming caloric content) and changes to the breakdown of the caloric content, which resulted from the above improvements to reduce the unit consumption of heavy oil. In 2002, the combustion heat of heavy oil accounted for approximately 90 % of the caloric content required for drying (17,800 Mcal/h). However, the ratio could be reduced to 59 % due to recovery of waste heat from the SO3 cooler at the acid plant and installation of the steam heat exchanger. In addition, the improvement of drying efficiency by increasing the size of the dryer kiln in December 2008 also contributed to reducing the caloric content required for drying by 4 % (17,800 Mcal/h → 17,080 Mcal/h). Furthermore, the ratio has been reduced to 57 % at present, reflecting the increased steam volume at the steam heat exchanger since March 2009 (4.1 mt/h → 5.1 mt/h).

Proceedings of Copper 2010

789

Hoshi, Toda, Motomura, Takahashi, Hirai

Drying conditions: Hot blast temperature: 1000°C Hot airflow: 1800 Nm3/min Water content before drying: 9.3% Water content after drying: 0.5%

Caloric content required for drying: 17,800 Mcal/h Accounted for 90% of the total calorific content Specific consumption of 8.09 L/t (annual average for 2002)

Combustion heat of heavy oil: 16,000 Total charges: 210 t/h (in conversion) Heavy oil volume: 1,600 L/h

Recovery of waste heat from refining furnace: 1,800

Caloric content required for drying: 17,800 Mcal/h

Conventionally, the combustion heat of heavy oil accounted for approximately 90% of total caloric content required for drying.

①SO3 cooler (system E) at sulfuric acid plant ②Heat recovery using heat exchanger

Accounted for 59% the total calorific content Caloric content required for drying: 17,800 Mcal/h

Combustion heat of heavy oil: 10,500

①2,100 ②2,200

Total charges: 210 t/h (in conversion) Heavy oil volume: 1,050 L/h

③ Increasing size of dryer kiln ②’Increasing steam volume at steam heat exchanger

Caloric content required for drying: 17,080 Mcal/h Accounted for 57 57% the total calorific content Specific consumption of 4.46 L/t (during May 2009)

Combustion heat of heavy oil: 9,780 ①2,100

②’ 2,700

The combustion heat of heavy oil has been reduced to a level that accounts for 57% of total caloric content required for drying at present.

Total charges: 210 t/h (in conversion) Heavy oil volume: 978 L/h

③Reduction of 720 Mcal (accounted for 4% of total calorific content)

Figure 10: Changes to the Breakdown of Caloric Content

9

Conclusion

Saganoseki Smelter & Refinery has established a high-efficiency, production-increasing system through several large-scale investments such as integration of its two FSFs into one and the SPI project. As a result, the maximum feed rate to the flash furnace reached 215 dry-mt/h and the maximum smelting capacity reached 480,000 mt/a. We also addressed efficiency improvements after the SPI project with measures such as recovering waste heat from the SO3 cooler at the acid plant, using steam generated from the waste heat recovery boiler of the FSF by installing a steam heat exchanger and increasing rotary kiln. As a consequence, we successfully reduced the unit consumption of heavy oil for drying ores from 8.09 L/mt to 4.57 L/mt.

790

Proceedings of Copper 2010

Dryer Fuel Reduction and Recent Operation of FSF at Saganoseki In the future, we will continue to address stable operation of the respective facilities including the FSF on the condition that safety and environmental preservation are fully ensured in the smelter. At the same time, we intend to pursue further efficiency improvements of different processes and cost reductions to raise global competitiveness.

References [1] M.Ishikawa (1998): Journal of MMIJ, 114, 447-454 [2] Chikashi Suenaga, Takayoshi Fujii, Yoshiaki Suzuki and Mitsumasa Hoshi: Second International Conference on PMP (2000), 879-884 [3] Katsuya Toda, Toshihiro Kamegai, Masaharu Takahashi, Mitsumasa Hoshi, Takayoshi Fujii: 12th International Flash Smelting Congress (2008)

Proceedings of Copper 2010

791

792

Proceedings of Copper 2010

Development of the Continuous Copper Converting Using Two-Zone Vaniukov Converter E. Jak

L. Tsymbulov

The University of Queensland Pyrometallurgy Research Centre (PYROSEARCH) Brisbane, Australia

Norilsk Nickel RJS Gipronickel Institute JS 11 Grazhdansky prospect Saint-Peterburg, Russia, 195220

Keywords: Vaniukov converter, copper, nickel, oxidation, reduction, continuous converting

Abstract Gipronickel Institute, JS is developing a process of continuous converting of copper concentrates as well as complex sulphide feeds, in particular Ni-containing copper matte, resulting in the production of blister copper and fluid slag in a two-zone Vaniukov furnace. The process under development is an alternative for the existing technology of blister copper production to improve sulphur capture, copper recovery and enable treatment of complex feeds. Mass balance, chemical partitioning, distribution of major elements as well as heat balance are analysed using customised model based on the thermodynamic computer package FactSage. Results of the pilot plant trials with copper concentrate and of the laboratory tests have been used to develop and tune the model. The model is then used to develop feed and fluxing strategy for the full scale industrial process for the co-treatment of the copper concentrate with Ni-containing copper matte. Analysis of some of the output parameters including chemical distribution and heat balance are presented as a function of main input production characteristics such as feed, flux, fuel and oxygen rates.

1

Introduction

Gipronickel Institute, JS is developing a process of continuous converting of copper concentrates as well as complex feeds, in particular Ni-containing copper-sulphide materials like copper matte using a two-zone Vaniukov furnace. Intensive development program has been carried out to date starting from laboratory investigations and progressing to the pilot plant tests of the two-zone Vaniukov furnace with area of the base of 11.4 m2 (results of these tests are presented in previous publication [1]).

Proceedings of Copper 2010

793

Jak, Tsymbulov The compositions of the treated concentrates and matte used in different operations vary in a wide range of Cu/Ni and Cu/Fe ratios. The resulting copper and nickel partitioning between copper blister and slag, the amount of the produced slag, fluxes and their proportions, as well as the thermal balance also vary in a wide range. The Vaniukov converting belongs to the strongly agitated bath type of processes like Isasmelt, Ausmelt, Mitsubishi, Noranda, Teniente. Thermodynamic equilibrium can be useful in analyses of chemical partitioning of elements between phases reacting in these processes with intensive interaction between reacting phases at high temperatures. Development of the thermodynamic model of the Vaniukov two-zone continuous converting of the copper Ni-containing sulphide feeds is developed for detailed trend analysis for further use in optimisation of the process parameters. The model is based on the results of the experimental studies, and in particular, on the results of the pilot plant trials [1].

Figure 1:

Schematic drawing of the two-zone Vaniukov continuous converting for the treatment of the copper nickel-containing matte and concentrates producing copper blister.

The Vaniukov two-zone continuous converting furnace (see Figure 1) consists of the oxidation zone and the reduction zone. The original design combined a siphon used for continuous blister and slag tapping. Copper matte or/and enriched concentrate (simultaneous treatment is possible), fluxes and coal are fed into the oxidising zone. SiO2 and Ca-containing fluxes are used. Oxygen enriched to 80-95 % is injected through tuyeres. Coal is used to balance heat if needed. The reducing/oxidizing level is controlled by adjustment of the sulphur / oxygen ratio or coal. The slag formed in the oxidis794

Proceedings of Copper 2010

Development of the Continuous Copper Converting Using Two-Zone Vaniukov Converter ing zone flows into the reduction zone. The reduction is achieved by oxygen-deficient combustion of the gas or liquid fuel. The required reducing/oxidizing level in the reduction zone can be adjusted by the coal addition. The metal droplets settle down from the slag into the blister which is tapped trough the siphon into the mixer – the furnace for collection of the blister before it is passed further into anode furnaces for refining. The reduction zone slag contains an insignificant part of the copper, a significant part of the nickel and practically all the cobalt; this slag is continuously tapped through the siphon barrier located at the 2000-2200 mm height from the furnace base. The slag is granulated and passed to nickel smelting for further treatment. The oxidation and reduction zones are separated by a water-cooled wall. The gas cleaning systems are separate. The oxidation zone SO2-containing off-gas may be used for elemental sulphur, sulphuric acid or liquid SO2 production. The reduction zone CO- and H2-containing off-gas are post-combusted with air or oxygen-enriched blast through special tuyeres located in the top row of the water jackets (see Figure 1) and then cleaned before release into atmosphere. The properties of the slags are particularly important for effective processing in this actively agitated bath, and the selection of the fluxing strategies therefore are the key issues for successful and efficient operation of the smelting in the Vaniukov furnace. The continuous converting slags contain copper, nickel, cobalt and iron oxides together with the fluxing components CaO and SiO2. Small Al2O3, MgO and Cr2O3 concentrations can also be present. The properties of such slags have not been properly investigated in practice. In particular, the fully-liquid composition area at fixed temperature, the crystallisation sequence and such important properties as viscosity and surface tension have not been determined. Analysis of the key properties of the slag system, in particular, effect of the amount and CaO/SiO2 ratio of the fluxes on copper and nickel partitioning between phases, on the Fe3+ and Fe2+ ratio, on copper and nickel partitioning into blister copper, and on other process parameters is essential.

2

Thermodynamic modelling

The thermodynamic modelling of oxide systems was carried out using thermodynamic computer package FactSage [3, 4] (version 6.1). Previously developed thermodynamic model of the slag phase [5] was tested together with the copper alloy and matte thermodynamic models using experimental copper and nickel distribution data obtained in the pilot plant tests [2]. The alloy phase was developed through optimisation fitting in the experimental pilot plant data [2] on Ni and Cu partitioning between blister and slag phases. Parameters of the alloy model were adjusted to achieve best agreement between predictions and the results of the pilot plant trials. Agreement between predicted and experimental copper and nickel partitioning data is demonstrated in Figure 2.

Proceedings of Copper 2010

795

Jak, Tsymbulov Further modelling was carried out with the slag and matte models from FactSage [4] combined with the thermodynamic model of the new alloy phase developed in this study. The spinel, wustite, wollastonite, dicalcium silicate, olivine and mullite solid phases were also considered in modelling. The model of the overall two-zone Vaniukov continuous converting process was then developed with these thermodynamic models of the individual phases using FactSage. Figure 3 presents schematic of this model of the two-zone process. In brief, equilibrium between input material was assumed in the oxidation zone, The gas from the oxidation process was removed, the condensed phases – slag and blister – were passed as an input for the reduction zone. Equilibrium was then predicted between the slag and blister from the oxidation zone and additional input streams – blast and oil – in the reduction zone. 15 Results of Optimisation of t/d of Cu alloy

R

10

e on gZ cin du Re

Ni in Cu blister [wt%]

R

R

Base case 5

R

Exp Ox zone Exp Red zone Calc Ox zone Calc Red Zone

R

Base case

Ox Ox R Ox

Oxidising Zone

0 0

5

10

15

Ox R

Ox Ox

20

25

Ox

30

35

Cu2O in slag [wt %] Figure 2:

796

Nickel and copper partitioning (FactSage predictions).

Proceedings of Copper 2010

Development of the Continuous Copper Converting Using Two-Zone Vaniukov Converter

Figure 3:

Schematic of the thermodynamic model of the two-zone Vaniukov furnace.

The Input for the oxidation zone is given in Table 1. The temperature of 1350 °C was taken for the oxidation and reduction zones. Equilibrium was assumed, and phase compositions and amounts formed in the oxidation zone were predicted with the model. The slag and copper blister phases from the oxidation zone were the input for the reduction zone along with the enriched air and oil (see Table 1). It is essential to note that the present model predicts separate equilibria in the oxidation and reduction zones with different individual compositions of the slag and copper blister products for each zone. The actual process produces one slag and one copper blister product. It may be argued that the model takes into account physical separation of two zones and predicts continuous, sequential change of the slag and blister from oxidation to reduction zones.

Proceedings of Copper 2010

797

Jak, Tsymbulov Table 1:

Input of the model. Cu concentrate SiO2 flux CaO flux

Oxidation zone

Blast

Oil

total Kg

4000.0

470.0

290.0

110.0

3237.0

512.0

total %

106.5

103.0

103.0

103.0

100.0

100.0

Cu2S

%

88.1

0.0

0.0

0.0

0.0

0.0

Ni3S2

%

4.1

0.0

0.0

0.0

0.0

0.0

FeS

%

6.9

0.0

0.0

0.0

0.0

0.0

Ni

%

0.7

0.0

0.0

0.0

0.0

0.0

Co

%

0.2

0.0

0.0

0.0

0.0

0.0

SiO2

%

0.0

75.0

0.0

6.8

0.0

0.0

Al2O3

%

0.0

15.0

0.0

7.1

0.0

0.0

MgO

%

0.0

6.0

2.9

0.0

0.0

0.0

Fe2O3

%

0.0

4.0

1.4

1.0

0.0

0.0

CaO

%

0.0

0.0

0.0

2.3

0.0

0.0

CaCO3

%

0.0

0.0

95.7

0.0

0.0

0.0

H2O

%

6.5

3.0

3.0

3.0

0.0

0.0

C

%

0.0

0.0

0.0

79.4

0.0

0.0

H

%

0.0

0.0

0.0

3.0

0.0

0.0

S

%

0.0

0.0

0.0

0.5

0.0

0.0

O2

%

0.0

0.0

0.0

0.0

92.5

0.0

N2

%

0.0

0.0

0.0

0.0

7.5

0.0

C3H8

%

0.0

0.0

0.0

0.0

0.0

38.3

C4H10

%

0.0

0.0

0.0

0.0

0.0

61.7

Slag-ox-zone Copper-ox-zone

Reduction zone

798

Coal

Blast

Oil

total Kg

Model

model

1697.0

577.0

total %

100.0

100.0

100.0

100.0

Slag

Model

model

model

model

Blister

Model

model

model

model

O2

%

0.0

0.0

92.5

0.0

N2

%

0.0

0.0

7.5

0.0

C3H8

%

0.0

0.0

0.0

38.3

C4H10

%

0.0

0.0

0.0

61.7

Proceedings of Copper 2010

Development of the Continuous Copper Converting Using Two-Zone Vaniukov Converter

3

Results of the thermodynamic modelling and discussions

An extensive predictions program has been carried out to investigate the effect of major parameters outlining fluxing and oxygen injection rates on the outcomes of the two-zone Vaniukov continuous converting process in terms of chemical partitioning of major elements between major gas and condensed phases and in terms of heat balance. Some selected results of the study are given in Figures 4 through 19 to present analysis of the major trends and to demonstrate capabilities of this approach of thermodynamic analysis of the complex metallurgical processes. Figure 2 reports the Ni concentration in blister as a function of the copper content in the slag phase. The discrepancy in the 13-18 wt. % Cu2O concentration range for the oxidation zone is in the range of much lower PO2 than is practically achieved during copper blister production. Reasonable agreement between experimental and predicted values is observed in the range of conditions of practical interest (see Figure 2).

3.1 Effect of SiO2 and CaO fluxing with the SiO2/CaO ratio of 2.2 Effect of SiO2 and CaO fluxing with the SiO2/CaO ratio of 2.2 was investigated first. It is apparent that the amount of the fluxes should be minimised to reduce the volume of produced slag. At the same time, the amount of fluxes should be sufficient to achieve slag with the required properties necessary for stable processing. For the oxidation zone, it was predicted that the amount of the fluxes practically does not influence the concentration of Ni in blister (see Figure 4) which is determined by the effective oxygen partial pressure during the process, i.e. the oxygen/concentrate and coal ratio. The Cu and Ni concentration in the slag decrease with the increase of the fluxes (see Figures 5 and 6). Figure 7 indicates that the iron in the slag is present predominantly in the ferrous Fe2+ form (60-70 %) at this SiO2/CaO ratio of 2.2. The increase of the flux amount reduces Fe3+ fraction. The rate of the decrease of the nickel and copper partitioning into blister with the increase of the flux amount is presented in Figure 8 and 9 respectively. For the reduction zone, the proportion of the ferric iron Fe3+ is significantly lower than in the oxidation zone and is decreasing from 0.24 to 0.18 (see Figure 7) with the increase of the flux addition. The increase of the added fluxes practically does not effect the nickel and copper concentrations in the slag resulting from the reduction process (see Figures 5 and 6 respectively). The nickel partitioning into the blister increases by approximately 5 % during the reduction stage of the process (see Figure 8). It is important to note that the addition of the SiO2 and CaO fluxes reduce the proportion of the solid phase (see Figure 10) – presence of the solid phases during Vaniukov process is not acceptable due to the higher probability of foaming noticed to be related to the precipitation of crystalline phases – the fraction of flux additions for the investigated SiO2/CaO ratio of 2.2 is predicted to be not less than 0.35. Proceedings of Copper 2010

799

Jak, Tsymbulov

6

Ni in Cu blister [wt%]

5

4

g Zone n i c u d e R

3

2

1

Oxidising Zone

0 0.15

0.2

0.25

0.3

0.35

0.4

0.45

CaO&SiO2 flux to concentrate ratio[wt]

Figure 4:

Concentration of Ni in blister as a function of total flux. 30

Oxidising Zo ne

Cu2O in slag [wt%]

25

20

15

10

Reducing Zone

5 0.15

0.2

0.25

0.3

0.35

0.4

0.45

CaO&SiO2 flux to concentrate ratio[wt]

Figure 5:

800

Concentration of Cu2O in slag as a function of total flux.

Proceedings of Copper 2010

Development of the Continuous Copper Converting Using Two-Zone Vaniukov Converter

10

Ni in slag [wt%]

9.5

9

Oxi disi ng

8.5

8

Zon e

7.5

Reducing Zone 7 0.15

0.2

0.25

0.3

0.35

0.4

0.45

CaO&SiO2 flux to concentrate ratio[wt]

Figure 6:

Concentration of Ni in slag as a function of total flux. 0.7

Fe

3+ + /Fe2 in slag

0.6

Oxidis ing Zo ne

0.5

0.4

0.3

Reducing Z one

0.2

0.1 0.15

0.2

0.25

0.3

0.35

0.4

0.45

CaO&SiO2 flux to concentrate ratio[wt]

Figure 7:

Fe3+/Fe2+ ratio in slag as a function of total flux.

Proceedings of Copper 2010

801

Jak, Tsymbulov

Ni in Cu blister / total Ni [wt%]

13

12

Reducin g Zone

11

10

9

8

Oxidising Z one

7

6 0.15

0.2

0.25

0.3

0.35

0.4

0.45

0.4

0.45

CaO&SiO2 flux to concentrate ratio[wt]

Cu in Cu blister / total Cu [wt%]

Figure 8:

Ni partitioning to blister as a function of total flux. 90

Oxidising Zone

85

80

75

70

Reducing Zone

65

60 0.15

0.2

0.25

0.3

0.35

CaO&SiO2 flux to concentrate ratio[wt]

Figure 9:

802

Cu partitioning to blister as a function of total flux.

Proceedings of Copper 2010

Development of the Continuous Copper Converting Using Two-Zone Vaniukov Converter

Proportion of solid phases [wt%]

10

8

6

Red ucin g

4

2

Zon e

Oxidising Zone 0 0.15

0.2

0.25

0.35

0.3

0.4

0.45

CaO&SiO2 flux to concentrate ratio[wt]

Figure 10: Proportion of solids as a function of total flux.

Ni in Cu blister [wt%]

6

Reducing Zone

5

4

3

2

1

Oxidising Zone

0 0

1

2

3

4

5

6

7

SiO2/CaO in oxidising slag [wt]

Figure 11: Concentration of Ni in blister as a function of SiO2/CaO ratio.

Proceedings of Copper 2010

803

Jak, Tsymbulov

3.2 Effect of SiO2/CaO ratio in the flux Further analysis is focused on the investigation of the effect of the SiO2/CaO ratio at fixed amount of the flux addition. It is important to note that investigation of this issue is of considerable practical interest for technologists since there is no firm information of the optimum SiO2/CaO ratio in the slag. For the oxidising zone, it was predicted that the SiO2/CaO ratio in the slag practically does not effect the nickel concentration in blister (see Figure 11). Significant effect has the SiO2/CaO ratio on the ferric/ferrous slag ratio (see Figure 12). As the Fe3+ fraction is significantly higher than Fe2+ for the SiO2/CaO ratio of 0.8, the ferric iron Fe3+ fraction decreases to 0.2 in the high-silica slag. It was predicted that the Cu2O concentration in the slag practically does not change as a function of the SiO2/CaO ratio (see Figure 13). The exception is the range of SiO2/CaO ratios between 0.8 and 2 that is related to the formation of the solid phase (see Figure 14). The high-Ca slag with the SiO2/CaO ratios between 0.8 and 2 therefore cannot be recommended for the continuous copper converting of the nickel-containing matte and concentrates. Lower nickel partitioning into blister also suggests preference of the slags with SiO2/CaO ratio over 2 (see Figure 15). For the reduction zone, it was predicted that the ferric iron fraction is relatively higher even at reducing conditions for the SiO2/CaO ratios between 0.8 and 2 (see Figure 12). The increase of the SiO2/CaO ratio in the slag results in some increase of the copper concentration in the slag and in quite significant increase of the nickel concentration in the slag (see Figures 13 and 16 for Cu and Ni oxides respectively). The recovery of the copper into blister, therefore, decreases with the increase of the SiO2 fraction in the slag (see Figure 17), which is the negative factor of the high-SiO2 slags. The nickel recovery into blister, however, decreases with the increase of the SiO2 fraction (see Figure 15), which is an advantage of the high-silica slags since lower concentration of nickel in blister makes the copper refining easier.

3.3 Effect of the oxygen blast rate The influence of the oxygen blast rate on the main process parameters has also been investigated in the course of this study; only selected plots are reported in this paper. The main trends with the change of the oxygen injection rate are obvious – the Cu2O concentration and the Fe3+/Fe2+ ratio in the slag increase, and the concentration of nickel in the blister and partitioning of the copper and nickel into the blister decrease with the increase of the oxygen blast rate. The concentration of the nickel in slag as a function of oxygen injection rate is of particular interest (see Figure 18) – this relationship has a maximum. This maximum is a result of the preferential oxidation of nickel compared to copper at low PO2 resulting in the increase of the nickel concentration in the slag. Significant partitioning of the copper oxide into slag however is predicted to occur as the concentration of

804

Proceedings of Copper 2010

Development of the Continuous Copper Converting Using Two-Zone Vaniukov Converter nickel in blister reaches 0.4-0.5 wt. % which results in the decrease of the concentration of other components in the slag including decrease of the nickel concentration. The effect of the oxygen blast rate, and therefore of the level of oxidation of the copper sulphide feed on the heat balance is analysed with reference to Figure 19. The heat balance strongly depends on the oxygen enrichment and on the achieved PO2. For example, at 92.5 % enrichment the heat balance is achieved at approximately 0.4 wt. % nickel concentration in the blister. 1.8 1.6

Fe

3+ + /Fe2 in slag

1.4 1.2 1 0.8

Oxidi sing Z one

0.6

Red uci ng

0.4 0.2

0 0

1

Zon e 2

3

4

5

6

7

SiO2/CaO in oxidising slag [wt]

Figure 12: Fe3+/Fe2+ in slag as a function of SiO2/CaO ratio.

Proceedings of Copper 2010

805

Jak, Tsymbulov

30

Cu2O in slag [wt%]

25

Oxidising Zone

20

15

Reducing Zone 10 0

1

2

3

4

5

6

7

6

7

SiO2/CaO in oxidising slag [wt]

Figure 13: Concentration of Cu2O in slag as a function of SiO2/CaO ratio.

10

e on gZ cin du Re

8

6

g sin idi Ox

4

2

ne Zo

Proportion of solid phases [wt%]

12

0 0

1

2

3

4

5

SiO2/CaO in oxidising slag [wt]

Figure 14: Proportion of solids as a function of SiO2/CaO ratio.

806

Proceedings of Copper 2010

Development of the Continuous Copper Converting Using Two-Zone Vaniukov Converter

Ni in Cu blister / total Ni [wt%]

20

18 16 14 12

Reducing Zone

10 8

Oxidising Zone

6

4 0

1

2

3

4

5

6

7

SiO2/CaO in oxidising slag [wt]

Figure 15: Partitioning of Ni to blister as a function of SiO2/CaO ratio. 10

Ni in slag [wt%]

9

Oxidising Zone

ne o Z g in c du e R

8

7

6

5 0

1

2

3

4

5

6

7

SiO2/CaO in oxidising slag [wt]

Figure 16: Concentration of Ni in slag as a function of SiO2/CaO ratio.

Proceedings of Copper 2010

807

Jak, Tsymbulov

Cu in Cu blister / total Cu [wt%]

90

Oxidising Zone

85

80

75

70

Reducing Zone

65

60 0

1

2

3

4

5

6

7

SiO2/CaO in oxidising slag [wt]

Figure 17: Partitioning of Cu to blister as a function of SiO2/CaO ratio. 10

Base case

Ni in slag [wt%]

9

8

7

6

85

90 5

95

92.5

Oxidising Zone

4 0.94

0.96

0.98

1

1.02

1.04

1.06

1.08

1.1

O2+N2 wt frac.of base case(labls-O/ON)

Figure 18: Concentration of Ni in slag as a function of oxygen blast rate.

808

Proceedings of Copper 2010

Development of the Continuous Copper Converting Using Two-Zone Vaniukov Converter

-1.82E+07

85

-1.84E+07

95

Base case

∆H loss [KJ]

90

92.5

-1.86E+07

-1.88E+07

-1.90E+07

-1.92E+07

Oxidising Zone

-1.94E+07

-1.96E+07

-1.98E+07

0.94

0.96

0.98

1

1.02

1.04

1.06

1.08

1.1

O2+N2 wt frac.of base case(labls-O/ON)

Figure 19: Heat balance as a function of oxygen blast rate.

4

Conclusions

Thermodynamic modelling of the two-zone Vaniukov continuous converting process has indicated that the most promising for this technology are slags with the SiO2/CaO ratios above 2 that enable to obtain slags with low concentration of the ferric iron Fe3+ and lower probability of the spinel precipitation. Higher nickel partitioning into slag is another advantage of the high-silica slags since this makes further copper refining easier. It was predicted that the amount of the SiO2 and CaO fluxes do not strongly influence the partitioning of the copper and nickel between blister and slag, however this partitioning is strongly effected by the SiO2/CaO ratio. It is recommended to add 35-40 % flux relative to the copper sulphide concentrate to avoid precipitation of solid phases.

Proceedings of Copper 2010

809

Jak, Tsymbulov

References [1] KNYAZEV M.V., RYABKO A.G., TSYMBULOV L.B. ET AL.,” Two-Zone Vaniukov Furnace: New Potential Copper and Nickel Production”, Sohn International Symposium, San Diego, USA, August 27-31, 2006, Vol. 8, 327 334. [2] L. B. TSYMBULOV, M. V. KNYAZEV, L. SH. TSEMEKHMAN AND A. N. GOLOV, “Pilot testiing of a process developed for treatment of Ni-containing copper concentrate after highgrade matte separation resulting in blister copper production in two-zone vaniukov furnace”, Cu2007- The 6th International Copper/Cobre Conference, Toronto, Canada, Aug 2007. [3] C. W. BALE; E. BÉLISLE; P. CHARTRAND; S. A. DECTEROV; G. ERIKSSON; K. HACK; I. H. JUNG; Y. B. KANG; J. MELANÇON; A. D. PELTON; C. ROBELIN; S. PETERSEN, “FactSage thermochemical software and databases -- recent developments”, Calphad, Volume 33, Issue 2, June 2009, Pages 295-311. [4] FactSage, Version 6.1. Ecole Polytechnique, Montréal. http://www.factsage.com/, 2009. [5] DECTEROV S., JUNG I. H., JAK E., HAYES P. AND PELTON A.D., “Thermodynamic Modeling of the Al2O3-CaO-CrO-Cr2O3-FeO-Fe2O3-MgO-MnO-SiO2-S System and Applications in Ferrous Process Metallurgy”, VII Int.Conf. on Molten Slags, Fluxes and Salts, Capetown, publ. SAIMM, Johannesburg, South Africa, 2004, pp. 839-850.

810

Proceedings of Copper 2010

Liquidus Temperature in Calcium Ferrite Slags in Ca2Fe2O5 and Ca2SiO4 Primary Phase Fields with Cu and Fixed Po2 E. Jak, B. Zhao, P. C. Hayes The University of Queensland Pyrometallurgy Research Centre (PYROSEARCH) Brisbane, Australia

C. Nexhip, D. P. George-Kennedy Rio Tinto Kennecott Utah Copper LLC Magna, Utah, USA

Keywords: Liquidus temperature, calcium ferrite slag, copper converting, EPMA

Abstract Calcium ferrite slags, with the major components Cu2O, "Fe2O3" and CaO are used in continuous copper-converting processes. A number of other impurities including SiO2 are also present in the slag. CaO is added to flux Fe, to reduce liquidus temperatures in the spinel (magnetite) primary phase field to obtain liquid slag at Cu converting temperatures. The extent of the Fe fluxing by CaO is limited by the appearance of the high-calcium phases and depends on a number of factors including the SiO2 concentration that is the focus of this study. Knowledge of phase equilibria in this slag system is important for optimal control of the furnace performance. Extensive investigations therefore are being undertaken to characterise the liquidus temperatures and phase equilibria in the "Cu2O"-Fe2O3-FeO-CaO-SiO2 system. Most of the previous studies in this system have been focused on the spinel primary phase field with relatively low concentration of calcium oxide. Present paper is focused on the investigation of phase equilibria at high-calcium oxide concentrations in the range of conditions relevant to the copper converter slag at Rio Tinto -Kennecott Utah Copper LLC (KUC). The experimental methodology has been developed and is based on the application of the equilibration/quenching/Electron Probe X-Ray Microanalysis (EPMA) techniques using oxide substrates. Dicalcium ferrite (Ca2Fe2O5) and dicalcium silicate (Ca2SiO4) were found to be primary phases at high-CaO concentrations. The effects of SiO2 on the liquidus temperatures and phase equilibria of calcium ferrite slags in primary phase fields of Ca2Fe2O5 and Ca2SiO4 have been investigated at 1250 °C and the oxygen partial pressures of 10-5.5 to 10-4.8 atm in equilibrium with metallic copper. The liquidus temperatures in the Ca2Fe2O5 primary phase field decrease with increasing SiO2 concentration in the liquid. Further increase of SiO2 results in the appearance of Ca2SiO4 as a primary phase. The liquidus temperatures in the Ca2SiO4 primary phase field first increase and then decrease Proceedings of Copper 2010

811

Jak, Zhao, Nexhip, George-Kennedy, Hayes with increasing SiO2 concentration in liquid. The addition of Al2O3 can decrease the liquidus temperature in Ca2SiO4 primary phase field.

1

Introduction

Calcium ferrite slags are used in continuous copper converting processes, such as the Mitsubishi and the Kennecott flash converters have a number of advantages including faster kinetics, lower slag volumes and better impurity removal [1-4]. The major components of these slags are Cu2O, CaO and iron oxides. CaO is added to flux Fe, to reduce liquidus temperatures in the spinel (magnetite) primary phase field to obtain liquid slag at Cu converting temperatures. The extent of the Fe fluxing by CaO is limited by appearance of the high-calcium phases and depends on a number of factors including the SiO2 concentration that is the focus of this study. The stability of the highcalcium phases strongly depends on the SiO2 concentration. Knowledge of phase equilibria in this slag system is important for optimal control of the furnace performance. Extensive investigations therefore are being undertaken to characterize the liquidus temperatures and phase equilibria in the "Cu2O"-Fe2O3-FeO-CaO-SiO2 system. Most of the previous studies have mostly been focused on the spinel primary phase field with relatively low concentration of calcium oxide [5-8]. Present paper is focused on the investigation of phase equilibria at high-calcium oxide concentrations in the range of conditions relevant to the copper converter slag at Rio Tinto -Kennecott Utah Copper LLC (KUC). SiO2 is present in copper converting slags in addition to the major components Cu2O, CaO and iron oxide as a result of entrainment of smelting slag during matte tapping [3]. There is no information available in literature on the effect of SiO2 on phase equilibria of calcium ferrite slag at high-CaO concentrations at copper saturation and controlled oxygen partial pressures. The aim of this study is to provide information on the stabilities of high-CaO phases including Ca2Fe2O5 and Ca2SiO4 that can be used to directly assist and guide the selection of process conditions in plant operations. The experiments, at this stage, are focused on the range of composition and temperature directly related to industrial practice at KUC.

2

Results and discussions

The experimental procedure used in this project is similar to the equilibriation/quenching/Electron Probe X-Ray Microanalysis (EPMA) technique which has been described in previous papers by the authors [7, 9]. In the present study the primary phases are high-CaO phases Ca2Fe2O5 and Ca2SiO4. One of the main difficulties in this study was to prepare suitable substrate. The experimental methodology therefore has been further developed for the primary phase fields of Ca2Fe2O5 and Ca2SiO4. The procedures for preparation of Ca2Fe2O5 and Ca2SiO4 substrates are different from that for spinel. 812

Proceedings of Copper 2010

Liquidus in Calcium Ferrite Slags in Ca2Fe2O5 and Ca2SiO4 Primary Phase Fields Ca2Fe2O5 substrate was prepared by mixing pure CaO and Fe2O3 powders in required composition and pelletizing the mixture to approximately 0.2 g. The final substrate was obtained by machining the pellet to the desired shape and then sintering at 1400 oC in air. Ca2SiO4 substrate cannot be prepared by sintering the mixture of CaO and SiO2 at high temperature. The reason is that Ca2SiO4 formed at high temperature disintegrates to powder on cooling as a result of transformation of α'-Ca2SiO4 to γ-Ca2SiO4. After a series of different tests and attempts a strong “Ca2SiO4” substrate that could further be used in experiments has been successfully prepared at relatively low temperature with chemical additives. Typical microstructure of the “Ca2SiO4” substrate used in this project is shown in Figure 1. It can be seen from the figure that this partially sintered substrate is dense although the Ca2SiO4 phase is actually not formed. At equilibration temperature (1250 oC in this project) the Ca2SiO4 phase is formed and surrounded by liquid phase.

Cu2O

CaO

SiO2

Figure 1:

Typical microstructure of the substrate material for the Ca2SiO4 primary phase field experiments (also contains 1 wt. % Fe2O3 and 0.5 wt. % Cu2O).

2.1 Experimental results Typical microstructures of the equilibrated and then quenched samples are presented in Figures 2 and 3, showing well quenched liquid slag in equilibrium with primary phases and metallic copper. Figure 2 shows a typical microstructure of a sample quenched from 1250 °C and Po2 = 10−5.0 atm in the Ca2Fe2O5 primary phase field and Figure 3 shows a typical microstructure of a sample quenched from 1250 °C and Po2 = 10−5.0 atm in the Ca2SiO4 primary phase field. It can be seen from the figures that small dendrites were formed on cooling in the samples. In the free surface of the sample Proceedings of Copper 2010

813

Jak, Zhao, Nexhip, George-Kennedy, Hayes there is well quenched area in which the composition of the glass is measured; the composition in that area represents the liquid composition at the equilibration temperature. Most of experiments, at this stage, have been carried out at 1250 °C and fixed Po2 of 10−5 atm. That is directly related to the copper converting operations at KUC. A number of experiments were also carried out at Po2 of 10−4.8 and 10−5.5 atm to examine the effect of oxygen partial pressure on the 1250 °C isotherms.

Ca2Fe2O5 Cu metal

glass

Figure 2:

814

Typical microstructure of a sample quenched from 1250 °C and Po2 = 10−5.0 atm in the Ca2Fe2O5 primary phase field at Cu metal saturation.

Proceedings of Copper 2010

Liquidus in Calcium Ferrite Slags in Ca2Fe2O5 and Ca2SiO4 Primary Phase Fields

Ca2SiO4

Cu metal

glass

Figure 3:

Typical microstructure of a sample quenched from 1250 °C and Po2 = 10−5.0 atm in the Ca2SiO4 primary phase field.

It has been previously reported that the liquidus isotherms of high-CaO phase Ca2Fe2O5 are approximately parallel to the join “Cu2O”-“Fe2O3”, i.e. the liquidus temperatures in these primary phase fields are mainly dependent on CaO concentration for a given oxygen partial pressure [5]. Based on this fact, the effect of SiO2 concentration on the position of the liquidus isotherm for these high-CaO phases can be examined by plotting the CaO concentration in the liquid phase against to the SiO2 concentration in the liquid phase for the liquidus at a given temperature and oxygen partial pressure. Figure 4 shows the observed relations between CaO and SiO2 concentrations in the liquid slag at 1250 °C and Po2 =10−5 atm in the Ca2Fe2O5 and Ca2SiO4 primary phase fields at Cu saturation. The solid squares and circles represent experimental results in the Ca2Fe2O5 and Ca2SiO4 primary phase fields respectively from the present study. The open square represents experimental result reported by Nikolic on Ca2Fe2O5 liquidus without SiO2 [11]. It can be seen that when SiO2 concentration is below 2.2 wt. % the primary phase of the slag is Ca2Fe2O5 and the liquidus temperatures slightly decrease with increasing SiO2 concentration in the liquid phase. Above 2.2 wt. % SiO2 in the liquid phase, the primary phase of the slag is Ca2SiO4 and the liquidus temperatures increase with increasing SiO2 concentration up to 5 wt. %.

Proceedings of Copper 2010

815

Jak, Zhao, Nexhip, George-Kennedy, Hayes

liquid + 2CaO.F2O3 liquid + Ca2SiO4

liquid

Figure 4:

Observed relations between CaO and SiO2 concentrations in the liquid at 1250 °C and Po2 = 10−5 atm in Ca2Fe2O5 and Ca2SiO4 primary phase fields at Cu saturation ■ present liquidus in Ca2Fe2O5 primary phase field, ● present liquidus in Ca2SiO4 primary phase field, □ Nikolic result without SiO2 [11].

Figure 5 shows correlations between CaO and SiO2 concentrations in the liquid slag at 1250 °C and Po2 =10−5 atm in Ca2SiO4 primary phase field at Cu saturation. The solid squares and circles represent experimental results in the Ca2Fe2O5 and Ca2SiO4 primary phase fields respectively in the system “Cu2O”-CaO-“Fe2O3”-SiO2 from the present study. The open square represents experimental result reported by Nikolic on Ca2Fe2O5 liquidus without SiO2 [11]. The dashed line represent the predictions by the FactSage™ computer package [10]. Two points with Al2O3-containing slags from the present study are also shown in the figure. It can be seen from the figure that in the Ca2SiO4 primary phase field the CaO concentrations in the liquid phase decrease with increasing SiO2 concentration in liquid when SiO2 concentration is below 5 wt. %. This means that the liquidus temperatures in the Ca2SiO4 primary phase field increase with increasing SiO2 concentration in liquid at SiO2 concentration below 5 wt. %. When SiO2 concentration in liquid is above 5 wt. %, the liquidus temperatures in the Ca2SiO4 primary phase field decrease with increasing SiO2 concentration. FactSage™ calculation gives the same trends but different absolute values (note that the FactSage™ databases have been developed when these data were not available – further work is necessary to further improve the FactSage™ databases).

816

Proceedings of Copper 2010

Liquidus in Calcium Ferrite Slags in Ca2Fe2O5 and Ca2SiO4 Primary Phase Fields When Al2O3 is present in the slag, it can be seen from Figure 5 that liquidus at 1250 oC and PO2 = 10-5 atm moves towards the higher CaO concentrations, i.e., the liquidus temperatures are decreased by addition of Al2O3.

liquid + Ca2SiO4

liquid

Figure 5:

Observed relationships between CaO and SiO2 concentrations in the liquid at 1250 °C and Po2 =10−5 atm in Ca2SiO4 primary phase field at Cu saturation in the “Cu2O”-CaO-“Fe2O3”-SiO2-Al2O3 system. Solid symbols: present results. Solid line: best fitted line for the present results. Dash line: calculated results by FactSage™ 5.3.

Figures 6a and 6b show the projection of the 1250 °C isotherms in Ca2Fe2O5 and Ca2SiO4 primary phase fields at Cu saturation. The solid lines represent isotherms in the Ca2Fe2O5 and Ca2SiO4 primary phase fields and isobars experimentally determined in the present study, and dashed lines show recent data reported by Nikolic et al. [11]. In these primary phase fields the Cu2O concentration is fixed at a given temperature and oxygen partial pressure. It can be seen from the figure that the 1250 °C isotherm in the Ca2SiO4 primary phase field moves towards lower CaO concentrations (the Ca2SiO4 liquidus temperature increases) with increasing SiO2 concentration in liquid from 2.2 to 2.5 wt. %. In contrast, with increasing SiO2 concentration in liquid from 1.5 to 2.2 wt. %, the isotherm of 1250 °C in the Ca2Fe2O5 primary phase field moves towards high CaO direction (the Ca2Fe2O5 liquidus temperature decreases). The effect of SiO2 concentration on liquidus temperature is more significant in the Ca2SiO4 primary phase field than that in the Ca2Fe2O5 primary phase field. At 1250 °C and Cu saturation, increase of Po2 from10−6.0 to 10−5.0 atm results in the increase of over ~12 wt. % Cu2O in slag. Similarly, concentration of Cu2O increases as Po2 increases further from 10−5.0 to 10−4.8 atm. Proceedings of Copper 2010

817

Jak, Zhao, Nexhip, George-Kennedy, Hayes

Figure 6a: Experimentally determined 1250 °C isotherms (solid lines) in Ca2Fe2O5 and Ca2SiO4 primary phase fields at Cu saturation. Dashed lines: isotherms and isobars by Nikolic [11] ■ Ca2Fe2O5 liquidus at Po2 = 10−5 atm, ● Ca2SiO4 liquidus at Po2 = 10−5 atm, □ Ca2Fe2O5 liquidus at Po2 = 10−5.5 atm , ○ Ca2SiO4 liquidus at Po2 = 10−4.8 atm.

Figure 6b: Enlarged 1250 °C isotherms (solid lines) in Ca2Fe2O5 and Ca2SiO4 primary phase fields at Cu saturation. Dashed lines: isotherms and isobars by Nikolic [11] ■ Ca2Fe2O5 liquidus at Po2 = 10−5 atm, ● Ca2SiO4 liquidus at Po2 = 10−5 atm, ○ Ca2SiO4 liquidus at Po2 = 10−4.8 atm. □ Ca2Fe2O5 liquidus at Po2 = 10−5.5 atm , 818

Proceedings of Copper 2010

Liquidus in Calcium Ferrite Slags in Ca2Fe2O5 and Ca2SiO4 Primary Phase Fields

2.2 Industrial implications CaO/Fe weight ratio has been used in industrial practice to control fluxing conditions [3]. The data from Figures 3 and 4 are combined to give another type of presentation: correlation between CaO/Fe ratio and SiO2 concentration in liquid phase at 1250 °C and Po2 = 10−5 atm in Ca2Fe2O5 and Ca2SiO4 primary phase fields at Cu saturation. It can be seen from Figure 7 that at SiO2 concentration below 2.2 wt. % (Ca2Fe2O5 primary phase field) the CaO/Fe weight ratios increase with increasing SiO2 concentration. This means that liquidus temperatures in the Ca2Fe2O5 primary phase field are decreased by increasing SiO2 concentration. However, if SiO2 concentration is between 2.2 and 5 wt. % in liquid, the CaO/Fe weight ratio in liquid significantly decreases with increasing SiO2 concentration. To attain a liquidus temperature below 1250 °C for a slag with high SiO2 concentration, CaO/Fe weight ratio has to be increased.

Figure 7:

Correlation between CaO/Fe ratio and SiO2 concentration in liquid phase at 1250 °C and Po2 = 10−5 atm in Ca2Fe2O5 and Ca2SiO4 primary phase fields at Cu saturation.

The data obtained in the present study demonstrate the need for careful control of CaO/Fe ratio in the feed to avoid excessive accretion formation. The liquidus temperatures are found to vary in the complex way as a function of SiO2 concentration in slag, the effect being dependent on the primary phase formed.

Proceedings of Copper 2010

819

Jak, Zhao, Nexhip, George-Kennedy, Hayes

3

Summary

Experiments have been carried out in the system CaO-“Fe2O3”-“Cu2O”-SiO2-Al2O3 at controlled oxygen partial pressures and metallic copper saturation to characterise the liquidus temperatures and phase equilibria relevant to the copper converter slag at Rio Tinto - Kennecott Utah Copper LLC (KUC). It was found that Ca2Fe2O5 and Ca2SiO4 are primary phases of the calcium ferrite slags at high-CaO region. The liquidus temperatures in Ca2SiO4 primary phase field first increase and then decrease with increasing SiO2 concentration in liquid. The liquidus temperatures in Ca2Fe2O5 primary phase field decrease with increasing SiO2 concentration in liquid. Addition of Al2O3 can decrease the liquidus temperature in Ca2SiO4 primary phase field.

Acknowledgments The authors wish to thank Rio Tinto - Kennecott Utah Copper LLC for providing the financial support to enable this research to be carried out. Ms Jie Yu, who provided general laboratory assistance and undertook much of the careful sample preparation and measurement work.

References [1]

George D.B. (2002): Continuous copper converting – a perspective and view of the future Sulfide Smelting 2002, Proceedings of a symposium held during the TMS annual meeting: 313, Seattle, WA, USA, TMS, Warrendale.

[2]

Kojo I., Lahtinen M. and Miettinen E. (2009): Flash converting – sustainable technology now and in the future – International Peirce-Smith Converting Contennial held during the TMS 2009, annual meeting & exhibition: 383-395, San Francisco, CA, USA, TMS, Warrendale.

[3]

Walton R., Foster R. and George-Kennedy D. (2005): Converter and fire refining practices, Proceedings of a symposium held at the TMS annual meeting: 283-294, San Francisco, CA, USA, TMS, Warrendale.

[4]

Davenport W.G.L., King M., Schlesinger and M., Biswas A.K. (2002): Extractive metallurgy of copper, 4th ed.; Pergamon, Oxford, United Kingdom, pp. 155-71.

[5]

Hino J., Itagaki K. and Yazawa, A. (1989): Phase relations in the CaO-FeOn-Cu2O and CaOFeOn-Cu2O-SiO2 systems at 1200-1300 °C - Shigen to Sozai, 105: 315-320.

[6]

Takeda Y. (2003): Phase diagram of CaO-FeO-Cu2O slag under copper saturation – Yazawa Int’l Symposium: 211-225, San Diego, CA, USA, TMS, Warrendale.

[7]

Ilyushechkin A., Hayes P.C. and Jak E. (2004): Liquidus temperatures in calcium ferrite slags in equilibrium with molten copper - Metall. Trans. B., 35B: 203-215.

820

Proceedings of Copper 2010

Liquidus in Calcium Ferrite Slags in Ca2Fe2O5 and Ca2SiO4 Primary Phase Fields [8]

Nikolic S, Hayes P.C. and Jak E. (2007): Liquidus temperatures in calcium ferrite slags equilibrated with molten copper at fixed oxygen partial pressures – Copper 07, Vol. III, Book 1: The Carlos Diaz Symposium on Pyrometallurgy, A.E.M. Warner, C.J. Newman, A. Vahed, D.B. George, P.J. Mackey and A. Warczok, eds., MetSoc, Toronto, CA, CIM, Montreal: 77-92.

[9]

Zhao B.H., Nexhip C., George-Kennedy D.P., Hayes, P.C. and Jak E. (2010): Effects of SiO2, Al2O3, MgO and Na2O on liquidus temperature of calcium ferrite slags at Cu saturation and fixed Po2 – Copper 2010, Hamburg, Germany.

[10] Bale C.W., Pelton A.D. and Thompson W.T.: Facility for the analysis of chemical thermodynamics (FactSage), Ecole Polytechnique, Montreal, Canada, available at: www.crct.polymtl.ca. [11] Nikolic S (2008): Investigation into the high temperature phase equilibria in the system“Cu2O”-FeO-Fe2O3-CaO-SiO2 at metallic copper saturation: an investigation of copper converting slags – PhD Thesis, The University of Queensland.

Proceedings of Copper 2010

821

822

Proceedings of Copper 2010

Numerical Simulation of Fluid Flow and Melt Temperature in Settler Zhou Jun Tongling Non-ferrous Metals Group Cooperation Ltd. Tongling, P.R. China, 244000

Zhou Ping, Chen Zhuo, Meichi Central South University School of Energy Science and Engineering Changsha, P.R. China, 410083

Liu Anming Jinlong Copper Co., Ltd. Tongling, P.R. China, 244021

Keywords: Numerical simulation, fluid flow, temperature distribution, settler, flash smelting furnace

Abstract With its great technical advantages and economic benefits, flash smelting technology seems appealing to the copper smelting industries and research institute. Research of the technology has been very active in the past decades. However, most of the research has been concentrated on the reaction shaft, while few has been reported on the settler of the flash furnace. Here, a simulation of fluid flow and temperature distribution in the settler is considered by authors with commercial software CFX4.3. In the simulation, a numerical model of high viscous melt in the settler is developed; and a FORTRAN program is self-developed to simulate the inlet boundary condition how melt particles fall down and scatter into the settler. The results indicate that the melt flow is chaotic at the inlet of settler, and it would to some extend influence the separation rate of the matte and slag. Meanwhile, the velocity in the main flow of the melt is quite low in the center, whilst it is relatively high near both side walls. Temperature of the melt is greatly influenced by the flow field and exhibits the same distribution pattern as the melt velocity. The uneven distribution of melt velocity and temperature thus causes decrease of the separation efficiency of matte and slag. Simulation is also carried out for cases in which frozen melt are formed on the side walls of the settler. The results show that fluid flow in such cases is similar to that without frozen melt; but temperature becomes more uniform due to the existence of frozen melt. This however accelerates the melt flow and decreases the retention time of slag in the settler, which therefore occurs against the settling process of slag and matte. Proceedings of Copper 2010

823

Zhou J., Zhou P., Chen, Liu, Meichi

1

Introduction

The pyrometallurgical flash smelting process is the main production method employed for extracting copper from its sulphide ores. It is perceived that Outokumpu installations account for 35-50 % of installed smelting capacity worldwide [1]. In the flash furnace, concentrate, flux and returned dust are introduced through the burner into the top of the reaction shaft in a stream of oxygenenriched air. Some fuel is also added to trim the heat balance. After ignition, intense reaction occurs in the shaft. The molten particles fall into the settler where slag and matte form and are tapped separately. Off-gas from the furnace is cleaned in a waste heat boiler and electrostatic precipitators before passing to the acid plant. The slag is cleaned in an electric furnace that is an integral part of the main furnace. In the past decades, the smelting intensity and production capacity of flash smelting furnace are increasing due to the application of the “Four Highs” technique-high productivity, high matte grade, high oxygen enrichment of the process air and high heat load [2]. However, it brings us many problems in the running and management of the settler. With the high concentrate loading, the molten metal flows more quickly and the separating time of matte from slag becomes relatively shorter, which causes the increase of copper losing in slag. Some factors, such as high oxygen enrichment, high grade of matte, and uneven velocity and temperature of molten metal in settler, lead to the formation of Fe3O4 and the decrease of effective volume of settler which is disadvantageous to the separation of matte from slag. Therefore, the research into the velocity and temperature fields of molten metal in the settler is helpful to optimize the geometry and operation of the settler, which will improve the separation of matte from slag. It is well known that the numerical simulation is an effective method for analyzing the transfer processes in metallurgical furnaces and also applied to study flow, temperature, concentration and heat releasing fields in copper flash furnaces. However, most of the research has been concentrated on the reaction shaft [3-6], while few has been reported on the settler of the flash furnace. Here, the fluid flow and temperature distribution in the settler is simulated with commercial software CFX4.3.

2

Physical and mathematical model

The geometry of the settler is shown as Figure 1. After falling into the settler from the reaction shaft, the molten particles move toward the outlets. Meanwhile, during the flow process of the melt, the matte drops down to the settler bottom through the slag layer, and two layers of fluid – slag layer and matte are formed due to their density difference. Considering the thickness of slag layer is much less than that of matte, the slag layer is ignored and the fluid is assumed as to include solely the matte. Although four matte exits and two slag exits (see Figure 2) run alternatively, every outlet of matte or slag generally works continuously over two hours, thus operation of the settler is thought to be steady-state, which means that the flow of the molten is incompressible and steady-state. 824

Proceedings of Copper 2010

Numerical Simulation of Fluid Flow and Melt Temperature in Settler

Inlet

M1,M2,M3,M4- Matte Outlet; SS,SM-Slag Outlet

Figure 1:

Geometry of the settler

The fluid at the downstream of the settler flows slowly and has a velocity of round 4-10 m/s. However, the inlet velocity of the molten is 28 m/s, and the fluid at the region near the inlet appears to be turbulent. Moreover, both exit of matte and slag can be considered as convergent tube, the correspond flow belongs to turbulence. Therefore, the standard k-ε turbulent model is chosen in the simulation. The velocity and temperature distribution can be described by the following general control equation.

∇ ⋅ ( ρUΦ) = ∇ ⋅ (ΓΦ ∇Φ) + S Φ

(1)

Where, SΦ is the source term. Replacement of Φ with the value 1 gives the continuity equation, while for Φ=U and H, the momentum and enthalpy equations are obtained respectively. Also for Φ=k and ε, Equation 1 results in the appropriate expressions for the turbulent kinetic energy (k) and its rate of dissipation (ε).

3

Computational mesh

The multi-block uneven structured mesh is adopted. The denser mesh is utilized at the regions of inlet and exits of fluid where lager velocity gradient occur, and the coarser mesh is employed at the central section of the settler where there is an evener velocity distribution. Meanwhile, the mesh number decrease as much as possible on the premise that calculating accuracy is guaranteed. Here, the overall number of mesh is 136420, and Figure 2 is the sketch of mesh at the cross section of computational region.

Proceedings of Copper 2010

825

Zhou J., Zhou P., Chen, Liu, Meichi

M1, M2, M3,M4- Matte exits; SN,SS-Slag exits

Figure 2:

4

Computational mesh of settler

Parameters and boundary conditions

4.1 Parameters of matte The parameters of matte required in the simulation are listed in Table 1. Table 1:

The parameters of matte

Parameter name

Data Source

Value

Density / kg·m-3

3880 + 404[Cu ] M + 4590[Cu ] 2M − 3750[Cu ]3M

4964

Viscosity / Pa·s

3.36 × 10 −4 exp(5000TM )

0.009

Melting point / K

Reference [7]

1353

Surface tensor / N⋅m-1

Reference [7]

0.236

Specific heat / W·m-1·K-1

Reference [7]

8.916

Grade / %

X-ray analysis

60

Average temperature / °C

Measurement

1240

4.2 Boundary conditions 4.2.1 Inlet Because the melt with high temperature falls into the settler as a dispersed phase, the inlet boundary conditions cannot be given as usual. If the real velocity of the fluid is taken, the mass flux of the fluid is much larger than the real one. On the other hand, if the real mass flux of the fluid is taken, the fluid velocity at the inlet is much less than real one. To ensure that the mass flux and velocity of the fluid is accordant with the real parameters, a code to define the inlet boundary conditions is de826

Proceedings of Copper 2010

Numerical Simulation of Fluid Flow and Melt Temperature in Settler veloped based on Fortran program language. The mass flux and velocity of the molten are 24.2 kg/s and 2.8 m/s respectively.

4.2.2 Outlet Outlet is treated as pressure boundary and specified as 0 Pa.

4.2.3 Top surface Flue gas fills up the space above the settler, so the top surface of the melt is an interface between the melt and off-gas, and can be treated as free surface. Usually, free surface is simplified as symmetry boundary, which means that the gradient of parameters is equal to 0, that is,

∂u ∂v ∂w ∂p ∂T ∂k ∂ε = = = = = = ∂n ∂n ∂n ∂n ∂n ∂n ∂n within, n refers to the normal direction of the symmetry face.

4.2.4 Side wall and bottom The velocity is no-slip boundary condition, k and ε is calculated with wall function, and the temperature is thought to be the melting point of the molten.

5

The results and discussion

Here, it is noteworthy that M1 and SS are specified as the exit of matte and slag in the simulation. The main results are shown in Figure 3 to 7.

5.1 Velocity distribution 5.1.1 The flow at the inlet and its vicinity After the melt falls into the settler at a speed of 2.8 m/s, its vertical velocity is greatly reduced due to the resistance of the melt in the settler. At the depth of 0.5 m away from the inlet, the vertical velocity is 0.003 m/s, which is almost the horizontal velocity of the main flow. It can be seen from Figure 3 that the molten metal in the settler moves away from the inlet under the impingement of the molten droplets with high velocity, and flow toward the exit when reaching the wall. For this reason, an interesting phenomena then takes place, that is, the fluid near the wall has a larger velocity than that in the central one (refer to Figure 4). As a whole, the melt flow is chaotic at the inlet of settler.

Proceedings of Copper 2010

827

Zhou J., Zhou P., Chen, Liu, Meichi

Figure 3:

Velocity vector distribution at the cross-section of z=-0.1 m

Figure 4:

Velocity vector distribution at the cross-section of z=-0.5 m

5.1.2 The main flow region At the region from the central to downstream, most melt in the settler move towards the exit, and exhibits the behaviour of a plug flow. However, the velocity in the main flow of the melt is quite low in the centre, while it is relatively high near both side walls. At the central and the corner near the side of exit, there are some regions where the melt velocity is extremely small (refer to Figure 5). These regions result in the reduction of effective space for separation of matte and slag and higher copper losing in slag. Generally, these regions with greatly small velocity are called as “dead zone”.

Figure 5:

828

Velocity contour at multi-section

Proceedings of Copper 2010

Numerical Simulation of Fluid Flow and Melt Temperature in Settler

5.2 Temperature distribution From Figure 6 and Figure 7, it can be seen that high temperature regions mainly locate at the inlet of the melt and the fluid at this region is of larger temperature gradient. This is because the temperature field of the melt in the settler depends on the velocity distribution. The chaotic flow field in the inlet is helpful to heat transfer between the fallen droplets and the melt in the settler. Thus the larger temperature gradient is formed.

Figure 6:

Temperature distribution at the cross-section of z=-0.5 m

Figure 7:

Temperature contour at multi-section

The molten in the main flow region has a relatively lower temperature and a relatively small temperature gradient. It is obvious the temperature of main flow gradually drops down, which results from the influence of the heat loss through side walls.

6

Conclusion

A numerical simulation is applied to analyze the flow and temperature fields in the settler of copper flash smelting furnace. In the simulation, a numerical model of high viscous melt in the settler is established; and a FORTRAN program is self-developed to simulate how melt droplets fall into the settler at the inlet boundary at a mass flux and velocity in good accordance with the real data in production. The results indicate that the molten flow is chaotic at the inlet of settler, and it would to some extend influence the separation rate of the matte and slag. Meanwhile, the velocity in the main flow of the melt is quite low in the center, whilst it is relatively high near both side walls. Temperature of the melt is greatly influenced by the flow field and exhibits the same distribution pattern as the molProceedings of Copper 2010

829

Zhou J., Zhou P., Chen, Liu, Meichi ten velocity. The uneven distribution of melt velocity and temperature thus causes decrease of the separation efficiency of matte and slag.

Reference [1] R. R. Moskalyk, A M Alfantazi. Review of copper pyrometallurgical practice: today and tomorrow. Minerals Engineering[J]. 2003, 16:893-919 [2] Xie Kai. Several theory and operation optimization challenges in the development of modern copper flash smelters[D]. Central South University. 2006 [3] Chen Hong-rong, Mei Chi. Operation optimization of concentrate burner in copper flash smelting furnace[J]. Transactions of Nonferrous Metals Society of China. 2004, 16(03):382-386 [4] Christopher B Solnordal, Frank R A Jorgensen, Peter T L Koh, Arthur Hunt. CFD modeling of the flow and reactions in the Olympic Dam flash furnace smelter reaction shaft [J]. Applied Mathematical Modeling. 2006, 30:1310-1325 [5] D. R. Higgins, N B Gray, M R Davidson. Simulating particle agglomeration in the flash smelting reaction shaft[J]. Minerals Engineering. 2009, 22:1251-1265 [6] Li Xin-feng, Peng Shi-heng, Han Xiang-li, Mei Chi, Xiao Tian-yuan. Influence of operation parameters on flash smelting furnace based on CFD[J]. Journal of University of Science and Technology Beijing: Mineral Metallurgy Materials. 2004, 11(2):115-119 [7] Edit committee for Manual for Non-ferrous Metal Extraction Metallurgy. Manual for Nonferrous Metal Extraction Metallurgy (Copper and Nickel Volume). Beijing: Metallurgical Industrial Press, 2000

830

Proceedings of Copper 2010

Profit Enhancement through Steam Selling Kyoung-Soo Jung, Gun-Woong Byun, Sung-Ho Shin LS-Nikko Copper Inc. 70, Daejung-ri, Onsan-eup, Ulju-gun Ulsan, 689-892, Korea

Keywords: Steam selling, energy saving, steam generation

Abstract Onsan plant of LS-Nikko Copper started operation in 1979 by Outokumpu Flash Smelting Process and its capacity has been increased to 560,000 tons/year of Cu cathode through several expansions. Previously, Onsan plant produced steam from Waste Heat Boiler of smelting furnace in order to operate plant and surplus steam was vented. In 2004, Onsan plant began to sell some amount of steam to a nearby company which previously produced steam from oil boiler. It brought positive result for both the companies from financial point of view. Since then, steam production and selling have been increased by improving several processes such as expansion of WHB at Flash Smelting Furnace, installation of WHB at Converter Furnace and Heat Recovery System at Sulfuric acid plant. It has been introduced as a good model which increases a profit about 18 billion won per year and contributes to the environmental policy by way of decreasing CO2 emission approximately 65,000 tons per year.

1

Introduction

Waste heat generation is an integral part of many processes in Onsan plant such as smelting process producing copper anode and sulfuric acid (Figure 1). High temperature waste heat is being used for generating electric power, heating and it is own purpose by producing steam through the WHB.

Proceedings of Copper 2010

831

Jung, Byun, Shin

Figure 1:

Typical steam production and use scheme

Onsan smelter has been reducing unit operating cost through steady cost-reducing activities for energy. But it has been faced with difficulties in cost-reducing activities because of continual increase of oil price and energy cost. Subsequently, Onsan plant has recently carried out new project to motivate itself and overcome difficulties as a part of our paradigm shift. As a result of this project within a year, Onsan plant has succeeded in recovering additional waste heat and finally began to sell some amount of steam to neighbouring company which has used to produce steam from oil boiler. It has improved the balance sheet of both the companies and contributed to environment aspects too.

2

Background

The profit from the recovery system of waste heat was about 2.2 million dollar in 2004. It has been dramatically increased since we had additional steam and sold it to another company in 2005 (Figure 2).

832

Proceedings of Copper 2010

Profit Enhancement through Steam Selling

Figure 2:

The trend of profit for steam sale

Since Onsan plant began to sell steam to another company in 2005, we have gained many experiences and know-how for improving process. On the basis of these, Onsan plant carried forward a scheme to create new profit source. From those experiences, we realized that we could create more value by supplying steam to others.

Figure 3:

The profit structure for steam sale

Proceedings of Copper 2010

833

Jung, Byun, Shin We found that the neighboring company wanted to use low price steam produced by Onsan plant instead of expensive steam from oil boiler. It made us review to recover additional waste heat from Converter furnace and Sulfuric acid plant by improving process steadily.

3

Study on the possibility of the steam supply

Onsan plant immediately set about marking inquiries of steam demand in Onsan complex to supply steam consistently. After all, we found an oil refinery having many oil boilers in order to generate an amount of steam. The company is one of the biggest oil-refining-company producing an amount of steam by burning B-C oil in order to minimize power trouble in the process. So, we expected that the oil refining-company could show dramatic reduction of energy unit cost. But, there are three major issues we have to deal with for the oil refinery as follows. 1) Additional steam production. 2) Search of heat source for production of saturated steam (over 390 °C). 3) Minimization of heat loss and pressure drop loss (over 42 kg/cm2 for long steam pipe (3 km).

3.1 Additional steam generation According to the result of energy recovery project in Onsan plant, there were two ways of process improvement expected to be more efficient in producing additional steam. But those had some problems as follows. • Recovery of waste heat from Sulfuric Acid Process. Selecting materials of process facilities were in difficulties on the condition in Sulfuric acid process. • Recovery of waste heat from Converter Furnace. Excess of maintenance cost because of declining efficiency of boiler caused by dust and fume from the furnace. Precaution of corrosion from drop of gas temperature in batch process.

3.2 Search of heat source to make superheated steam (over 390 °C) Installation of new facilities is needed to increase steam quality (Now 260 °C and 42 kg/cm2)

834

Proceedings of Copper 2010

Profit Enhancement through Steam Selling

3.3 Minimization of heat loss for the long steam pipe (3 km) • Minimization of pressure drop (over 42 kg/cm2) • Minimization of radiant steam heat (keeping up temperature over 390 °C) • Special design for pipe is needed for above conditions

4

Improvement

4.1

Additional steam generation

• Recovery of waste heat in Sulfuric Acid plant In SO3 absorption process of Sulfuric Acid, there is a lot of energy available from acid formation, condensation and dilution. But in conventional plants it has been rejected to cooling tower through absorption towers and acid coolers. With improved materials of construction (310SS) and high temperature absorption, most of this energy is now recoverable at useful levels of temperature. To generate additional steam from Acid plant, Heat Recovery System was introduced by MECS (HRS). Today, HRS systems are operated reliably, generating steady income from steam. As a result, Onsan plant has been producing additional steam of about 30 t/h. The following scheme shows simplified Heat Recovery System (HRS) in Sulfuric Acid plant (Figure 4).

Figure 4:

Simplified HRS flow scheme

Proceedings of Copper 2010

835

Jung, Byun, Shin • Recovery of waste heat from converter furnaces Existing long jacket-type gas cooler which consumed electric energy and demanded a lot of maintenance costs more than we thought could be eliminated. To solve this problem and recover waste heat from converter furnaces, Onsan plant replaced existing long jacket-type gas cooler with new waste heat boiler (Figure 5). Also, the by-pass duct of new WHB to prevent the corrosion caused by gas temperature drop was installed. As a result, Onsan plant has succeeded in producing additional steam of about 8 tph in this process.

Figure 5:

Simplified CF WHB scheme

4.2 Search of heat source to make superheated steam (over 390 °C) Saturated steam of 260 °C has been produced in Onsan plant. However, oil refinery has been using superheated steam over 400 °C. Thus, new review for superheating saturation steam was started. Finally, we found the way of making superheated steam through heat source of SO3 Converter in acid plant. The process gas stream leaving the 1st bed converter is approximately 630 °C and cooled to approximately 420 °C at 2nd bed of converter by hot gas heat exchanger and waste heat boiler #1. New steam super-heater was installed at the outlet of the first bed to the waste heat boiler #1 to superheat saturation steam.

4.3 Minimization of heat loss for the long steam pipe (3 km) • Minimization of heat loss and pressure drop for long steam pipe. • Keeping up pressure 42 kg/cm2 and minimization of radiation heat. 836

Proceedings of Copper 2010

Profit Enhancement through Steam Selling Considering the length of new steam piping from Onsan plant to oil refinery nearby, it was impossible to keep 42 kg/cm2 steam pressure as required. Accordingly, special piping design for long distance was required to minimize pressure drop. Generally, conventional steam pipe was used to solve problem with pipe expansion by composing expansion loop. However, it could not meet customer’s need about required pressure at the end point because of long distance pipe. So slip-joint was used to minimize pressure drop of pipe and solve pipe expansion (Figure 6). Onsan plant is the first company using slip-joint on the condition of high pressure steam in Korea.

ㅁ ㄴ ㄷ ㄱ Figure 6:

Simplified slip-joint scheme

Also, we could not meet customer need about keeping steam temperature over 390 °C, if the steam pipe was just insulated by current conventional lagging with excessive radiant heat. To minimize radiant heat, we searched new lagging made of U-Brid to keep up temperature over 390 °C at the end point (Figure 7).

Thermal Conduction Rate

(W/m.K) 0.45 0.40 0.35

U- Brid

0.30 0.25

Perlite

0.20

Glass wool

0.15 0.10 0.05 0.00

Figure 7:

24

60

100

200

300

400

500

600

700

800

(℃)

Thermal Conduction Rate

Proceedings of Copper 2010

837

Jung, Byun, Shin Finally, we satisfied customer’s requirement for both steam temperature and pressure.

5

Results

1. Additional steam production: about 40 tonnes per hour. 2. As a result of supplying super-heated steam to oil refinery nearby, the profit was approximately 14 million dollar in 2008. 3. CO2 emission is expected to be reduced up to 100,000 tonnes per year through saving B-C oil in oil refining-company.

6

Conclusions

This project had a great difficulty in making a contract with consumer because of the difference of culture between the two. But, the difficulty was overcome by continuous and passionate effort. This project is a win-win situation for both the parties from the financial point of view. Also, this contributes to the environmental policy of Korean government by way of decreasing CO2 emission of 65,000 tonnes per year.

838

Proceedings of Copper 2010

Gas Injection Phenomena in Converters – An Update on Buoyancy Power and Bath Slopping Dr. Joël P. T. Kapusta Air Liquide Canada Inc. 90, boul. Marie-Victorin Boucherville, Québec, Canada J4B 1V6

Keywords: Converters, tuyere submergence, buoyancy power, bath slopping, critical blast flow rate

Abstract Numerous investigations were carried out worldwide in the field of submerged gas injection in nonferrous pyrometallurgy from 1975 to 1985. Major breakthroughs were made in Canada towards both the understanding of the underlying injection phenomena and the discovery of possible solutions to remedy the inherent hindrances of these pyrometallurgical processes. At a recent short course held in San Francisco in February 2009 as part of the International Peirce Smith Converting Centennial Symposium, course attendees were asked to help gather industrial data on current converter operations, essentially launching a mini converting survey, in an attempt to re-evaluate and refine the concept of slopping behavior based on sound operating data. This paper presents an analysis of the survey data received, together with an updated relationship between converter dimensions, tuyere submergence, blast rate characteristics and the critical slopping conditions, as well as a new plot of the slopping behavior diagram.

1

Introduction

In light of the recent shift of university research towards ‘materials’, for an author who has been involved with gas injection in molten metals, both in academia and industry, the years between 1975 and 1985 appear as a golden decade for Canadian non-ferrous metallurgical research. During that period, numerous investigations were carried out worldwide in the field of submerged gas injection in non-ferrous pyrometallurgy, which has been well documented in a comprehensive review by Brimacombe et al. [1]. Major breakthroughs were made in Canada towards both the understanding of the underlying injection phenomena and the discovery of possible solutions to overcome the inherent limitations on these pyrometallurgical processes [2-12]. Some of the advances and discoveries related to the century old copper converting process were revisited in San Francisco in February 2009 at the International Peirce Smith Converting Centennial Symposium during the short course Proceedings of Copper 2010

839

Kapusta “Injection Phenomena in the PS Converter – The Teachings of J. Keith Brimacombe and His UBC Research Team”. Bath slopping, the critical blast flow rate and the so called ‘slopping behavior diagram’ published by Brimacombe et al. [7] were amongst the topics of discussion during the course. The assumptions used by Richards et al. [9] to develop a relationship of buoyancy power as a function of tuyere submergence were reviewed and the limitations due to the uncertainties in the set of data used – the survey of Johnson et al. from 1979 [13] – were pointed out. A proposal was presented to the course attendees requesting their help in gathering industrial data on current converter operations, essentially launching a mini converting survey, in an attempt to re-evaluate the concept of slopping behavior based on sound operating data. As a result data from ten copper smelters and four nickel smelters was obtained, somewhat less than anticipated. With little new data to work with, it was decided to revisit the theoretical and mathematical description of buoyancy power in horizontal submerged injection, re-analyze the data from the surveys of Johnson et al. [13] and Lehner et al. from 1993 [14] and finally put the new analysis to test with the 2009 survey data. In this regards, the smelters that provided responses to the mini survey of 2009, their personnel and everyone that helped in gathering those questionnaires are gratefully acknowledged. This paper presents an analysis of the combined data set, and examines the implication of injection phenomena for the actual amount of power that can be physically transferred to the bath in the type of bubbling regime seen in converters. A revised diagram indicating limiting blowing conditions for operation is developed reflecting the new understanding of the implications of bath slopping.

2

Previous Work & Theoretical Analysis

In submerged gas injection in molten baths, buoyancy power has typically been evaluated in published metallurgical literature using a ‘simplified’ expression of the form:

εb = 2

 P + ρ bath g h  Q Pa Ln  a  M bath Pa  

(1)

where ε b is the buoyancy power (W/kg, which is equivalent to kW/t), Q is the volumetric gas flow rate at bath temperature (m3/s), Pa is the atmospheric pressure (Pa), M bath is the mass of the bath (kg), ρ bath is the density of the bath (kg/m3), g is the gravitational acceleration (9.81 m/s2) and h is the tuyere submergence below the bath surface (m) [7, 9]. Nakanishi et al. [15] derived a different form of the buoyancy power which they called ‘stirring power’ and which was expressed as follows:

840

Proceedings of Copper 2010

Gas Injection Phenomena in Converters



Q*

 RT

 Pa + ρ bath g h   Pa 

gas o  Ln  ε b = 2  -3   22.4 10  M bath 

(2)

where Q*o is the ‘normal’ volumetric gas flow rate (Nm3/s at 0 oC and 1 atm.), R is the universal gas constant (J/mol. K) and Tgas is the gas temperature (K). Tgas is assumed to be constant from the point of injection to the bath surface. All other parameters are similarly defined as those in Equation (1). Equations (1) and (2) have been used for metallurgical systems by assuming that the gas temperature reaches the bath temperature immediately at the gas discharge point, which in itself is a debatable assumption. Although rather convenient in providing a simple mathematical expression of the buoyancy power in a molten bath, this assumption may generate a significant error in the calculation of buoyancy power transferred from the gas. In a critical review of the phenomena involved in bath mixing by gas injection, Wraith [16] explained his view that the term ‘Tgas’ is problematic and the assumption that the gas temperature is equal to the bath temperature is most likely flawed. Wraith’s most interesting theory has yet to be published. Moreover, the above buoyancy power expressions only describe the power transferred to the bath during bubble rise without any consideration for the power developed during bubble growth at the tip of the submerged injection device and transferred to the bath prior to detachment. In this sense they would seem to apply more to a ‘jetting’ injection condition in which there is a continuous flow of gas from the tuyere. In 1979, Robertson and Sabharwal [17] developed a model of gas mixing in steel ladles induced by submerged lancing of argon. The novelty in their approach was their assumption that the temperature of the gas increased linearly with height during its rise within the bath and attained the bath temperature upon reaching the bath surface. They derived an overall power input expression by splitting the calculations in two parts: 1) the work done by the gas while it is growing an attached bubble (‘bubble growth’), and 2) the work done while the gas is in bubbles rising in the bath (‘bubble rise’ or buoyancy power). The overall power expression they developed was as follows:

ε total = ε growth + ε rise where

(3)

ε growth =

Q o Pa TI  T  1 − o  M bath To  TI 

ε rise = −

 T  Pa   Pa + ρ bath g h  Q o Pa TI  TS Q P T T  − 1 + 2 o a I  S +  S − 1  Ln   M bath To  TI M bath To  TI Pa   TI  ρ bath g h   

where ε total is the overall input power (W/kg), ε growth is the input power during ‘bubble growth’ (W/kg), ε rise is the buoyancy power during bubble rise (W/kg), Q o is the ‘standard’ volumetric gas

Proceedings of Copper 2010

841

Kapusta flow rate (Sm3/s at 15 °C and 1 atm.), Pa is the atmospheric pressure (Pa), M bath is the mass of the bath (kg), ρ bath is the density of the bath (kg/m3), g is the gravitational acceleration (9.81 m/s2), h is the tuyere submergence below the bath surface (m), To is the ‘standard’ gas temperature (288.15 K), TI is the gas temperature in the bubble at detachment (K), and TS is the gas temperature in the bubble at the bath surface (K). For convenience, since the 1979 manuscript is no longer easily accessible, the complete derivation from Robertson and Sabharwal is reproduced in this manuscript as an appendix. It is worth pointing out that if the gas temperature is assumed to reach the bath temperature upon detachment and remains constant during bubble rise (i.e. TI = TS = Tbath), then the buoyancy portion, ε rise, in Equation (3) reduces to:

ε rise = ε b = 2 Since Q o

 P + ρ bath g h  Q o Pa TI Ln  a  M bath To Pa  

(4)

TI = Q (at Tbath ) , then Equation (4) simplifies to: To

ε rise = ε b = 2

 P + ρ bath g h  Q Pa Ln  a  M bath Pa  

(5)

which is the same expression as Equation (1). In theory, the actual ‘total’ energy input imparted by gas injection into a bath of molten matte or metal should be described by an expression that includes a kinetic power term such that (6)

ε Total = ε kinetic + ε growth + ε rise The kinetic power input, ε kinetic, is defined by the following expression:

ε kinetic

2 1 ρ gas Q u gas = 2 M bath

 Q   ρ gas Q    A 1 tuyeres   = 2 M bath

2

(7)

where ρ gas is the gas density in the tuyeres, u gas is the gas velocity at the tuyeres exit, and A tuyeres is the total cross sectional area of the tuyeres (total gas flow area at the tuyeres exit). An example of the relative proportions of ε kinetic, ε buoyancy and ε growth is presented in Table 1 for air injection into a copper converter with the following ‘average’ operating parameters: 120 t of copper matte with a density of 5500 kg/m3, a tuyere submergence of 0.5 m, an air flow rate of 600 Nm3/min injected through 50 tuyeres (5 cm inner diameter). The gas temperature is assumed to be 15 °C in the tuyere (To = 15 °C) and to reach the bath temperature of 1200 °C at detachment (TI = 1200 °C).

842

Proceedings of Copper 2010

Gas Injection Phenomena in Converters Table 1:

Comparison of relative proportion of kinetic, bubble growth and buoyancy powers for a ‘typical’ copper converting operation

ε buoyancy = ε rise

ε kinetic  Q   ρ gas Q   A 1 tuyeres   2 M bath

ε growth

2

0.61 W/kg

2

 P + ρ bath g h  Q Pa Ln  a  M bath Pa  

21.5 W/kg

Q o Pa TI  T  1 − o  M bath To  TI  36.6 W/kg

As illustrated by the values in Table 1 for an ‘average’ copper converter slag blow, the kinetic power input is two orders of magnitude smaller than the buoyancy power input. This therefore justifies that the kinetic power input for conventional (low pressure) gas injection is typically ignored in the power input analysis of copper and nickel converters. However, the power imparted to the bath during bubble growth up until detachment, ε growth, is of the same order of magnitude as the buoyancy power. In fact, ε growth attains its maximum value under the assumption that, at detachment, the gas inside the bubble has reached the bath temperature. The value of ε growth is primarily temperature dependent and corresponds to a ‘thermal expansion power’ input.

3

Analysis of Converter Survey Data

Deriving correlations from survey data is a simple matter of number crunching made possible by reasonable assumptions. Explaining and making some practical use of such correlations is a much more daunting task, the ‘usability’ of the correlations being dependent on and limited by the ‘validity’ of the assumptions. For buoyancy calculations from survey data, in addition to the ‘Tgas’ concern mentioned above and which will be discussed later, tuyere submergence and mass of bath are the two other critical parameters with a significant impact on the results. Tuyere submergence is typically estimated rather than actually measured in converting operations while the mass of bath varies significantly from start to end of the converting cycle as the bath density increases from just above 5000 kg/m3 to just below 8000 kg/m3. Let’s first review the assumptions related to bath mass. Richards et al. [9] analyzed the data from Johnson et al. [13] by assuming a constant converter fill of 35 % from which the bath mass was calculated. Bath densities of 5200 and 7800 kg/m3 for the slag and copper blows respectively were used. Buoyancy power was calculated using Equation (1) on the basis of blast flow rate at bath temperature (i.e. Tgas = Tbath). This analysis was reproduced in this manuscript for the copper blows only and assuming a refractory thickness of 45.7 cm (18 inch) for both end walls and barrel. The results, in the form of a correlation of buoyancy power as a function of tuyere submergence, are presented in Figure 1.

Proceedings of Copper 2010

843

Kapusta

80

Buoyancy Power, W / kg

70 y = 30.3 x + 2.83 R2 = 0.76

60 50 40 30 20 10

Bath mass from constant 35% fill

0 0.00

0.20

0.40

0.60

0.80

1.00

1.20

Tuyere Submergence, m

Figure 1:

Buoyancy power calculated from data from Johnson et al. [13] using Equation (1) on the basis of blast flow rate at bath temperature (i.e. Tgas = Tbath), a constant fill of 35 % and a refractory thickness of 45.7 cm (18 inch) for both end wall and barrel

A similar analysis was performed using the reported mass of blister produced as the mass of bath for copper blows rather than a constant 35 % fill. The results are presented in Figure 2. This comparison shows that assuming a constant percent fill artificially improves the correlation. For this reason, the reported mass of blister produced per cycle will be used in the present analysis from this point forward as it more accurately reflects the actual operating conditions. This implies that in this review, buoyancy calculations will be performed for blowing conditions at the end of the copper-making final blow where very little slag is present. The relationship in Figure 2 shows the effect of increasing tuyere submergence on the buoyant power per unit mass as a result of gas injection. In general, there is a reasonably linear relationship which matched theory as illustrated in Figure 3. The dashed line is the theoretical line from Equation 1 with a constant mass of bath of 77.3 t (= average mass of blister produced by all smelters) and constant flowrate calculated at bath temperature of 42.0 m3/s (= average flowrate in the copper blow for all smelters). The data indicates that copper converters in the copper making blow tend to operate along a limiting condition defined by the dashed line. This limit is thought to derive from the power input required to sustain a standing wave on the surface of the bath [9].

844

Proceedings of Copper 2010

Gas Injection Phenomena in Converters

80

Buoyancy Power, W / kg

70 60 50 40 y = 54.2 x + 9.04 R2 = 0.56

30 20 10

Bath mass = mass of blister produced

0 0.00

0.20

0.40

0.60

0.80

1.00

1.20

Tuyere Submergence, m

Figure 2:

Buoyancy power calculated from data from Johnson et al. [13] with Equation (1) on the basis of blast flow rate at bath temperature (i.e. Tgas = Tbath) using the reported mass of blister produced as the mass of bath

At the point where a standing wave forms, the converter bath enters a slopping condition which periodically exposes the tuyere line to atmosphere causing excessive splashing and ejection of material through the mouth of the vessel. Generally, the power of the standing wave increases with wave height, and therefore, for increasing tuyere submergence, the splashing threshold is expected to occur at greater input power levels. The line therefore separates the non-slopping region in the lower area from the slopping region above. For a given submergence it is possible to calculate the buoyant power input, and hence the limiting blowing rate. However, before addressing this quantitatively it is important to complete the review of the industrial data.

Proceedings of Copper 2010

845

Kapusta

80

Buoyancy Power, W / kg

70 60 50 y = 55.3 x + 6.71 R2 = 0.99

40 y = 54.2 x + 9.04 R2 = 0.56

30 20 10

Bath mass = mass of blister produced

0 0.00

0.20

0.40

0.60

0.80

1.00

1.20

Tuyere Submergence, m

Figure 3:

Comparison of buoyancy power per unit mass from plant data from copper blows (solid line) with theoretical relationship for averaged converter conditions (dashed line)

The same approach was taken to an analysis of the data from the survey undertaken by Lehner et al. in 1993 [14]. The data and regression line, shown in Figure 4, follow the same pattern as seen for the Johnson et al. data in Figure 2. The slope and position of the 1993 line are similar to the 1979 results and confirm the general conclusions noted above.

846

Proceedings of Copper 2010

Gas Injection Phenomena in Converters

90 80

Buoyancy Power, W / kg

70 60 50 y = 51.9 x + 2.61 R2 = 0.52

40 30 20 10

Bath mass = mass of blister produced

0 0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

Tuyere Submergence, m

Figure 4:

Buoyancy power calculated from data from Lehner et al. [14] with Equation (1) on the basis of blast flow rate at bath temperature (i.e. Tgas = Tbath) using the reported mass of blister produced as the mass of bath

The 2009 survey results are presented in Figure 5, and show the same positive relationship as was apparent in 1979 and 1993. However, in this case the slope of the regression line is substantially lower than the 50 to 60 W/kg/m that had been estimated on the basis of the earlier data. The reason for this lies with a careful evaluation of the tuyere submergence. For each case, the submergence was calculated based on the reported bath mass and height of tuyeres from the bottom of the converter. The data in the figure is for 14 different operations covering slag and copper blows in copper converting, and slag and finishing blows in nickel making (19 data points). For the slag blows the bath mass used was the initial amount of matte charged (no slag), and for copper and nickel blows, the final mass of blister and Bessemer matte respectively.

Proceedings of Copper 2010

847

Kapusta

60

Buoyancy Power, W / kg

50 y = 34.1 x + 3.26 R2 = 0.68

40

30

20

10

0 0.00

0.20

0.40

0.60

0.80

1.00

1.20

Tuyere Submergence, m

Figure 5:

Buoyancy power versus tuyere submergence for the 2009 survey data using recalculated submergence levels

A re-evaluation of the data in the 1979 and 1983 surveys was undertaken to screen out the data where the reported submergence could not be reconciled with the reported bath mass. When the inconsistent data points are removed and the remaining data are combined with the 2009 data, Figure 6 is obtained. It is clear from this figure that the data from all surveys generally falls along the same line, with a reasonably good correlation. The slope of this line is 39.1, close to the value of 34.1 determined for the 2009 data alone.

848

Proceedings of Copper 2010

Gas Injection Phenomena in Converters

60 y = 39.1 x + 6.11 R2 = 0.58

Buoyancy Power, W / kg

50

40

30

20

10

0 0.00

0.20

0.40

0.60

0.80

1.00

1.20

Tuyere Submergence, m

Figure 6:

Buoyancy power versus tuyere submergence for the ‘screened out’ 1979 and 1993 survey data using reported submergence levels together with the 2009 survey data using recalculated submergence levels (67 data points)

As noted above, this line defines a limiting condition for converter operation. The implication of this relationship can be taken a step further by analyzing the slope of the regression line in Figure 6:

ε b = 39.1 h + 6.11

(R2 = 0.58)

(8)

The slope of the line is about 39 W/kg/m indicating that the buoyant energy input to the bath increases by this amount for every metre of submergence. The theoretical slope of this line (derivative of the right hand side) is also seen in Equation (1) and is given by the expression: Slope =

2 Q Pa ρ bath g M bath (Pa + ρ bath g h )

(9)

Although this is, in part, a function of submergence (h), if an ‘average’ value for the submergence is used (0.53 m for the present case) and the slope set to the critical value from the plant data (39.1) then it gives the critical relationship between blowing rate (Q) and bath mass (Mbath): Q = 39.1

(Pa

+ 0.53 ρ bath g ) M bath 2 Pa ρ bath g

Proceedings of Copper 2010

(10)

849

Kapusta For the copper blow this equation becomes: Q (m3/s) = 0.36 Mbath (t)

(11)

where Q is the limiting blast flowrate at bath temperature. A more practical form of Equation (11) can easily be written in terms of the normal gas flow rate: Q (Nm3/min) = 4.0 M bath (t)

(12)

The slope of this line suggests that a blowing rate increase of 4 Nm3/min/t of bath is possible (at an assumed bath temperature of 1200 °C). This relationship is shown in Figure 7 which represents a new ‘slopping behavior diagram’ of the limiting blowing rate as a function of bath mass (and hence, indirectly as a function of tuyere submergence). 500

Blast Flow Rate, Nm 3/min

400

Slopping 300

Non Slopping 200

y = 4.0 x 100

0 0

20

40

60

80

100

120

140

Mass of Bath, tonnes

Figure 7:

Relationship of blowing rate (at bath temperature) to bath mass for copper blows at the slopping limit from plant data assuming an averaged submergence of 0.53 m

The practical application of Equation (12) for converter operation is that for a constant blast rate, splashing could be reduced by increasing converter filling level. Alternatively, for a non-splashing converter, an increase in blast flow rate could be implemented without exceeding the splashing threshold provided the converter filling is increased proportionately. From the survey data for copper blows, it was determined that converter filling ranged between 10.5 to 37.0 % which indicates that a number of smelters may still have room to increase their filling level. Increasing converter filling is 850

Proceedings of Copper 2010

Gas Injection Phenomena in Converters often limited, amongst other restrictions, by the maximum number of crane moves that an aisle can handle during a converting cycle. One of the options to remedy this operational limitation, as already implemented or evaluated at several smelters, is to increase the size of transfer ladles. For converters already at their maximum practical filling level, one of the next options to increase the overall blast rate (and thus throughput) without generating or worsening splashing is to consider sonic gas injection through either single pipe tuyeres [6, 11] or shrouded injectors if high oxygen enrichment is also sought [18-22]. These operational matters and alternative gas injection modes are beyond the scope of the present paper but will however be examined in a future publication once additional new survey data are obtained.

4

Buoyant Power with Gas Bubbling

As noted in the Introduction this analysis can be extended by accounting for the bubbling nature of the gas injection. In this scenario, the simple buoyant power equation is not able to account for the complex phenomena, and the Robertson and Sabharwal method should be used. In order to apply this approach it is necessary to characterize the temperature of the gas at the end of each stage in the sequence, bubble growth and rise. Using a heat transfer model for bubble growth at the tip of a tuyere in a copper converter, Ashman et al. [4] determined that the gas temperature at bubble detachment is around 312 °C (585 K) for a submergence of 0.4 m. The temperature of the gas at the surface of the bath likely varies with tuyere submergence, and is even more difficult to estimate. However, as a first approximation the gas was assumed to reach 35 % of the bath temperature by the time it reaches the surface. This puts the gas temperature in the range of 360 to 450 °C. Application of the Robertson equation to the revised survey data on this basis is shown in Figure 8. The results indicate that under the assumptions described above, there is essentially no correlation between power imparted by the gas and the submergence of the tuyeres. Increasing the detachment temperature and/or the surface temperature of the gas does not induce the development of any trend or correlation. Upon deeper analysis of the calculated thermal power and buoyancy power, it was apparent that the thermal power (power input during bubble growth) contributed to 30 to 50 % of the total power input. Since thermal power per kilogram is strongly dependent on gas temperature and bath mass while being independent of tuyere submergence, the result is a ‘flattening’ of the correlation between the overall power input and tuyere submergence.

Proceedings of Copper 2010

851

Kapusta

80

Total Power Input, W / kg

70 60 50 y = 10.0 x + 18.28 R2 = 0.09

40 30 20 10 0 0.00

0.20

0.40

0.60

0.80

1.00

1.20

Tuyere Submergence, m

Figure 8:

Total power input calculated from the revised data set using Equation (3) with a detachment temperature of 312 °C and a surface temperature of 35 % of bath temperature

Upon analyzing the revised survey data one step further, assuming a detachment temperature of 312 °C and a bubble frequency of 10 Hz, it was calculated that the bubble volume at detachment for the 67 data points ranges from 13.4 to 58.6 l. As a first approximation, if the bubbles are considered spherical, then the corresponding bubble diameter at detachment ranges between 0.29 and 0.48 m, which is often in the order of or larger than the tuyere submergence. With these considerations, the buoyancy power was recalculated by comparing the bubble diameter, d b, to the reported tuyere submergence, h. If d b was larger than h, then the buoyancy power (ε b = ε rise) was set to zero as the bubble collapses at the surface without having to rise through the bath. If d b is smaller than h, then an ‘effective’ tuyere submergence, h eff, is defined as h eff = h – d b. This effective submergence, in essence the remaining length of path for the bubble top to reach the bath surface, is then used in the calculation of the buoyancy power. The results of this analysis are presented in Figure 9 which is the total power input as a function of the reported tuyere submergence (physical position of the tuyere with respect to the bath). Although the correlation in Figure 9 is not very strong (R2 = 0.3), it seems to be in agreement with the general trend as seen in Figure 6. If the gas is assumed to reach 50 % of the bath temperature upon reaching the surface, the correlation improves (R2 = 0.48) as shown in Figure 10. 852

Proceedings of Copper 2010

Gas Injection Phenomena in Converters

40

Total Power Input, W / kg

Tsurface = 35% Tbath

30 y = 12.2 x + 7.62 R2 = 0.30

20

10

0 0.00

0.20

0.40

0.60

0.80

1.00

1.20

Tuyere Submergence, m

Figure 9:

Total power input calculated from the revised data set using Equation (3) with a detachment temperature of 312 °C and a surface temperature of 35 % of bath temperature 40 Tsurface = 50% Tbath

Total Power Input, W / kg

y = 23.0 x + 5.79 R2 = 0.48

30

20

10

0 0.00

0.20

0.40

0.60

0.80

1.00

1.20

Tuyere Submergence, m

Figure 10: Total power input calculated from the revised data set using Equation (3) with a detachment temperature of 312 °C and a surface temperature of 50 % of bath temperature Proceedings of Copper 2010

853

Kapusta Although this approach to the data analysis has not yielded a significantly improved correlation at this stage, it holds promise in developing a more complete understanding of the gas injection power input and therefore a stronger basis for the optimization of operating parameters. It would be better to extend this analysis with a larger number of data points from the detailed survey of operations before drawing too many conclusions. In order to develop more confidence in this approach, and work toward the specification of optimum operating conditions, it would be important to have a means of establishing the temperature of the gas at the bath surface. It is expected that this would involve the formulation of a better heat transfer model of the gas-liquid interaction, and some plant measurements to validate assumptions. The injection process is also complicated by the fact that tuyeres are closely spaced and interfere with each other through a gas envelope along the tuyere line (well explained by Wraith et al. [23]). As a result, discrete bubbles are not typical, and therefore gas injection power calculations should ideally reflect this condition. Clearly, the analysis is not complete at this stage but is an indication of how power from gas injection is imparted to the bath taking into account tuyere submergence, bath temperature and bath mass. More updated data would also be useful in refining the analysis and developing a more sophisticated model of the process.

5

Conclusions

Gas injection phenomena in non-ferrous converting are complex phenomena with a range of chemical and physical implications for the process. The limiting blowing rate is an important factor in determining converter capacity, and efforts to optimize existing operations and design a next generation of converters rely on improving our understanding of this issue. This paper has attempted to move in this direction by bringing together the available industrial data and the fundamental science. As a result the following conclusions have been reached. 1. Analysis of the fundamentals of gas injection power input to the converter bath shows that the kinetic energy term is negligible. However, the energy available from bubble growth and rise are of the same order of magnitude. The actual energy imparted to the bath during bubble growth is strongly dependent on the temperature increase in the gas during this phase. If the gas temperature reaches the bath temperature then the growth power will likely exceed the buoyant power released by the bubble rise. 2. The 1979 and 1993 industrial survey data of converter operations contains operating data which is not internally consistent. As a result, screening of the data is necessary in order to extract meaningful relationships. When this is done, and combined with the detailed data obtained in 2009, a consistent picture of an operating limit for converters is seen. The critical limiting relationship for converter operation, based on a simple calculation of gas injection power, is seen as: Gas flow rate (Nm3/min) = 4.0 Bath mass (t). 854

Proceedings of Copper 2010

Gas Injection Phenomena in Converters If the blast exceeds the level indicated by this equation, then the bath is expected to enter a slopping condition leading to excessive wave motion on the surface and uncontrolled splashing. 3. More sophisticated approaches to the analysis of gas injection power inputs to the converter are possible, and are expected to yield deeper insights into the operation. This work would take into account factors such as bubble size relative to tuyere submergence, heat transfer to the gas as influenced by growth and rise time. Preliminary work in this direction shows that it is consistent with the general relationships in the simple buoyant power analysis.

Acknowledgements The author again wishes to express his appreciation to those companies that contributed to the 2009 survey data and provided detailed information on their converting parameters. In addition, the author acknowledges helpful and insightful discussions with Greg Richards on various issues addressed in the paper, and deeply appreciates his additional help in reviewing and contributing to this manuscript on his private time. The permission of Air Liquide to publish this paper is also greatly appreciated.

Proceedings of Copper 2010

855

Kapusta

List of Symbols g

Gravitational acceleration (= 9.81 m/s2 )

h

Tuyere submergence below bath surface, m

M bath

Mass of bath, kg

Pa

Atmospheric pressure, Pa (= 101,325 Pa)

Q

Gas flowrate at gas temperature (Tgas) and 1 atm, m3/s

Qo

Gas flowrate at standard conditions (15 °C & 1 atm), Sm3/s

Q *o

Gas flowrate at normal conditions (0 °C & 1 atm), Nm3/s

R

Universal gas constant (8.3143 J/mol K)

T

Temperature, K

Tbath

Bath temperature, K

Tgas

Gas temperature, K

To

Standard gas temperature (288.15 K)

TI

Gas temperature in the gas bubble at detachment, K

TS

Gas temperature in the gas bubble at the bath surface, K

V

Gas volume, m3

Vo

Molar volume of gas at standard temperature, Sm3

Wgrowth

Work done by the gas during bubble growth, J

Wrise

Work done by the gas during bubble rise, J

Wtotal

Total work done by injected gas, J

Greek letters

εb

Buoyancy power, W/kg

ε growth

Input power during bubble growth, W/kg

ε rise

Input power during bubble rise (= buoyancy power), W/kg

ε total

Total input power imparted by injected gas, W/kg

ρ bath

Bath density, kg/m3

856

Proceedings of Copper 2010

Gas Injection Phenomena in Converters

References [1]

BRIMACOMBE J.K., NAKANISHI K., ANAGBO P.E., RICHARDS G.G. (1990): Process Dynamics: Gas-Liquid – Proceedings of the Elliott Symposium on Chemical Process Metallurgy: 343-412, M.I.T., Cambridge, USA

[2]

ORYALL G.N., BRIMACOMBE J.K. (1976): Towards a Basic Understanding of Injection Phenomena in the Copper Converters – Metallurgical Transactions B, Vol. 7B: 391-403

[3]

HOEFELE E.O., BRIMACOMBE J.K. (1979): Flow Regimes in Submerged Gas Injection – Metallurgical Transactions B, Vol. 10B: 631-648

[4]

ASHMAN D.W., McKELLIGET J.W., BRIMACOMBE J.K. (1981): Mathematical Model of Bubble Formation at the Tuyeres of a Copper Converter – Canadian Metallurgical Quarterly, Vol. 20, No. 4: 387-395

[5]

BUSTOS A.A., RICHARDS G.G., GRAY N.B., BRIMACOMBE J.K. (1984): Injection Phenomena in Nonferrous Processes – Metallurgical Transactions B, Vol. 15B: 77-89

[6]

BRIMACOMBE J.K., MEREDITH S.E., LEE R.G.H. (1984): High Pressure Injection of Air into a Peirce-Smith Copper Converter – Metallurgical Transactions B, Vol. 15B: 243-250

[7]

BRIMACOMBE J.K., BUSTOS A.A., JORGENSEN D., RICHARDS G.G. (1985): Towards a Basic Understanding of Injection Phenomena in the Copper Converters – Proceedings of the Physical Chemistry of Extractive Metallurgy Symposium, AIME Meeting: 327-351, New York, USA

[8]

BRIMACOMBE J.K., BUSTOS A.A., RICHARDS G.G. (1985): Development of Punchless Operation of Peirce-Smith Converters – Final Report – Contract No. 24 St. 2344-0-4-9259

[9]

RICHARDS G.G., LEGEARD K.J., BUSTOS A.A., BRIMACOMBE J.K., JORGENSEN D. (1986): Bath Slopping and Splashing in the Copper Converter – Proceedings of the Reinhardt Schuhmann International Symposium on Innovative Technology and Reactor Design in Extraction Metallurgy, TMS-AIME Fall Meeting: 385-402, Colorado Springs, USA

[10] BUSTOS A.A., BRIMACOMBE J.K., RICHARDS G.G. (1986): Heat Flow in Copper Converters – Metallurgical Transactions B, Vol. 17B: 677-685 [11] BUSTOS A.A., BRIMACOMBE J.K., RICHARDS G.G., VAHED A., PELLETIER A. (1987): Development of Punchless Operation of Peirce-Smith Converters – Proceedings of the COPPER 87 – COBRE 87 International Conference, Vol. IV – Pyrometallurgy of Copper: 347-373, Santiago, Chile: 347-373 [12] ANAGBO P.E., BRIMACOMBE J.K., CASTILLEJOS A.H. (1990): A Unified Representation of Gas Dispersion in Upwardly Injected Submerged Gas Jets – Canadian Metallurgical Quarterly, Vol. 28, No. 4: 323-330

Proceedings of Copper 2010

857

Kapusta [13] JOHNSON R.E., THEMELIS N.J., ELTRINGHAM G.A. (1979): A Survey of Worldwide Copper Converter Practices – Proceedings of the Symposium on Copper and Nickel Converters, TMS-AIME Meeting: 1-32, New Orleans, USA [14] LEHNER T., ISHIKAWA O., SMITH T., FLOYD J., MACKEY P., LANDLOT C. (1993): The 1993 Survey of Worldwide Copper and Nickel Converter Practices – Proceedings of the Converting, Fire Refining & Casting Symposium, TMS Meeting: 1-58, San Francisco, USA [15] NAKANISHI K., FUJII T., SZEKELY J. (1975): Possible Relationship between Energy Dissipation and Agitation in Steel Processing Operations – Ironmaking & Steelmaking (Quarterly), No. 3: 193-197 [16] WRAITH A.E. (2009): Review Notes on Mixing Energy and Mixing Times in Gas Driven Bath Agitation – Personal Communication [17] ROBERTSON T., SABHARWAL A.K. (1979): A Physical Modelling-Based Approach to some Problems Associated with Submerged Gas Injection into Liquid Metal Melts – Proceedings of the Symposium on Gas Injection into Liquid Metals: I1-I29, University of Newcastle upon Tyne, England [18] BUSTOS A.A., CARDOEN M., JANSSENS B. (1995): High Oxygen Enrichment at UMHoboken Converters – Proceedings of the Copper 95-Cobre 95 International Conference, Vol. IV – Pyrometallurgy of Copper: 255-269, Santiago, Chile [19] BUSTOS A.A., KAPUSTA J.P., MACNAMARA B.R., COFFIN M.R. (1999): High Oxygen Shrouded Injection at Falconbridge – Proceedings of the Copper 99-Cobre 99 International Conference, Vol. VI – Smelting, Technology Development, Process Modeling and Fundamentals: 93-107, Phoenix, USA [20] BUSTOS A.A., KAPUSTA J.P. (2000): High Oxygen Shrouded Injection in Copper and Nickel Converters – Proceedings of the Brimacombe Memorial Symposium: 107-124, Vancouver, Canada [21] KAPUSTA J.P., STICKLING H., TAI W. (2005): High Oxygen Shrouded Injection at Falconbridge: Five Years of Operation – Proceedings of the Symposium on Converter and Fire Refining Practices: 47-60, San Francisco, USA [22] KAPUSTA J.P., WACHGAMA N., PAGADOR R.U. (2007): Implementation of the Air Liquide Shrouded Injector (ALSI) Technology at the Thai Copper Industries Smelter – Proceedings of Cu2007, the Sixth International Copper-Cobre Conference, Vol. III (Book 1) – The Carlos Diaz Symposium on Pyrometallurgy: 483-500, Toronto, Canada [23] WRAITH A.E., HARRIS C.J., MACKEY P.J., LEVAC C. (1994): On Factors Affecting Tuyere Flow and Splash in Converters and Bath Smelting Reactors – Proceedings of the European Metallurgical Conference 1994 (EMC’94), Vol. I: 50-78, Germany

858

Proceedings of Copper 2010

Gas Injection Phenomena in Converters

Appendix – Derivation of Robertson & Sabharwal Gas Injection Input Power Expression 1) Derivation of ‘bubble rise’ component Assuming that the temperature of the gas in a bubble increases linearly with the height y, then

 T T  y T  T − TI  = 1 − S  + S T = −  S  y + TS or TI TI  h TI h   

(A1)

The temperature dependence of the gas volume is defined as follows:

V = Vo

T T T = Vo I To To TI

(A2)

T  T  y T  = Vo I 1 − S  + S To  TI  h TI  Once the pressure effect on the volume is included, it follows:

V = Vo

 TI  Pa  To  Pa + ρ bath g y 

 TS  y T   + S 1 − TI  h TI  

(A3)

The rate of change of volume with height is given by     TS  1 T  y T   dy − 1 − S  + S  ρ bath g dy   (Pa + ρ bath g y ) 1 − TI  h TI  h TI  T    dV = Vo Pa I  2   To (Pa + ρ bath g y )    

(A4)

The work done by the gas in rising a height ‘dy’ is given by: dWrise = P dV − ρ bath g V dy

(A5)

where P = Pa + ρ bath g y

(A6)

Substituting equations (A3), (A4) and (A6) into equation (A5) yields:

dWrise

    TS  1 T  y T   dy − 1 − S  + S  ρ bath g dy   (Pa + ρ bath g y ) 1 − TI  h TI  h TI  T    = Vo Pa I    To Pa + ρ bath g y     − ρ bath g Vo

TI To

  Pa P + ρ g y bath  a 

Proceedings of Copper 2010

(A7)

 TS  y T   + S  dy 1 − TI  h TI   859

Kapusta The work done by the gas when rising to the liquid metal surface is obtained as:

o

∫ dW

rise

= Vo Pa

h

TI To

o



∫ 1 − h

TS  1 T  dy − 2 ρ bath g Vo Pa I TI  h To

o

∫ h

 TS  y T   + S  dy 1 − TI  h TI   Pa + ρ bath g y

(A8)

Integration of Equation (A8) yields:

Wrise

 o TI  TS  TI  1 −  − 2 Vo Pa = − Vo Pa To  TI  To  ∫h   = Vo Pa

TI To

+ 2 Vo Pa

 TS  T  − 1 − 2 Vo Pa I To  TI  TI To

 TS Pa −  ρ bath  TI

o  T 1 1 − S  dy + ∫ TI  h  h

TS Pa  T 1  1 − S  −  TI ρ bath g  TI  h  dy Pa  y +  ρ bath g 

 TS   − 1  TI 

(A9)

  P + ρ bath g h  T  1 1 − S   Ln  a  g TI  h  Pa  

After rearranging, the work done by the gas rising to the liquid metal surface is given by:

Wrise = − Vo Pa

TI To

 Pa + ρ bath g h   TS  T  Pa  T T  − 1 + 2 Vo Pa I  S +  S − 1  Ln   To  TI Pa  TI   TI  ρ bath g h   

(A10)

2) Derivation of ‘bubble growth’ component dWgrowth = P dV − ρ bath g V dy

(A11)

Since the bubble is stationary (i.e. the work done by the bubble is independent of depth), then dWgrowth = P dV (A12) Applying the gas law,

PV is constant, and for a fixed bubble P is also constant, then T

T dV − V dT PV V d = 0  = P d  = P T2  T  T

(A13)

Equation (A13) implies that T dV = V dT or by rearranging that dV =

860

V dT T

(A14)

Proceedings of Copper 2010

Gas Injection Phenomena in Converters Substituting Equation (A14) into Equation (A12) yields: dWgrowth =

PV P V dT = a o dT T To

(A15)

The work of the bubble before detachment is then obtained by integration between the temperature of the gas at the tip of the injection device and the temperature just prior to detachment: TI

∫ dW

growth

To

TI

P V = a o To

∫ dT

=

To

Pa Vo [TI − To ] To

(A16)

Finally, upon rearranging, the work of the bubble prior to detachment is defined as:

Wgrowth = Vo Pa

TI To

 To  1 − T   I 

(A17)

3) Total work during ‘bubble growth’ and ‘bubble rise’ The total work done by a gas during bubble growth at the tip of the injection device followed by bubble rise to the liquid metal surface is given by:

Wtotal = Wgrowth + Wrise = Vo Pa

TI To

 To  1 − T  I  

− Vo Pa

TI To

 TS  T  Pa   Pa + ρ bath g h  T T  − 1 + 2 Vo Pa I  S +  S − 1  Ln   To  TI Pa  TI   TI  ρ bath g h   

(A18)

4) Total input power during ‘bubble growth’ and ‘bubble rise’ The total input power of a gas during bubble growth at the tip of the injection device followed by bubble rise to the liquid metal surface is obtained by replacing the molar volume (Vo) by the gas flow rate (Q o). The overall input power per kilogram of bath, ε total, is then given by:

ε total = ε growth + ε rise =

Q o Pa TI  T  1 − o  M bath To  TI 

−

 T  Pa   Pa + ρ bath g h  Qo Pa TI  TS Q P T T  − 1 + 2 o a I  S +  S − 1  Ln   M bath To  TI M bath To  TI Pa   TI  ρ bath g h   

Proceedings of Copper 2010

(A19)

861

862

Proceedings of Copper 2010

Recovery of Valuable Metals from Converter Slags by Reduction with Iron Dipl.-Ing. Stefan Konetschnik, Dipl.-Ing. Helmut Paulitsch Ao.Univ.Prof. Dipl.-Ing. Dr.mont. Helmut Antrekowitsch University of Leoben Nonferrous Metallurgy Franz-Josef-Straße 18 8700 Leoben, Austria

Dipl.-Ing. Dr.mont. Josef Pesl Montanwerke Brixlegg AG Werkstraße 1-3 6230 Brixlegg, Austria

Keywords: Slag treatment, metal recovery, converter slag, copper, nickel, lead, tin

Abstract For the reduction of the internal material flow in the secondary copper metallurgy, a separate converter slag treatment is necessary. With iron as a reduction agent, the metals copper, lead, tin, etc. can be reduced out of the slag. They form a metal phase, which can be separated from the slag after settling. Finally, the process should be done twice, to form a copper (and nickel) rich phase as well as lead and tin. The basic idea of the project is to reduce the circulation of material, which is generated at the different furnaces. Converter slags contain up to 50 per cent metal losses and have to be fed in the smelting furnace. By treating slags in a separate process, the capacity of the smelting aggregate for low grade raw material and/or the throughput of the overall production process will be increased. Within the project, a thermodynamical model by using HSC chemistry and FactSage was developed. So it was possible to calculate the necessary amount of reduction agent and silica as well as to evaluate the success and/or deviation of the experimental investigations. In these multiple series of tests based on industrial input material, the influence of parameters like temperature, slag composition, amount and composition of the reduction agent, crucible material, etc. was determined. In a stepwise approximation, an optimised process should be developed for reaching two separate metal phases and a slag without valuable metals.

Proceedings of Copper 2010

863

Konetschnik, Paulitsch, Pesl, Antrekowitsch

1

Introduction

In the secondary copper metallurgy, it is possible to process different materials like residues or scrap by pyrometallurgical operations. At the Montanwerke Brixlegg AG, the melting process is done in a shaft furnace. There, black copper is produced, which is processed to raw copper in a Peirce-Smith converter. The unrefined metal phase contains about 96 per cent copper. Due to the distribution equilibrium and the bad settling properties, the losses of valuable metals in the slag are high (about 30 per cent copper in the slag as well as metals like nickel, lead and tin). To minimise the metal losses, the slag is circulated back to the shaft furnace, where the final slag contains about 1-2 per cent copper. The disadvantage of this circulation is the lower capacity of the furnace for the adequate raw materials. Instead of this conventional operation, the converter slag at the Montanwerke Brixlegg AG should be reduced with an iron-copper scrap to lower the content of the valuable elements in a final slag without circulation. The process is planned in two stages. In the first step, copper and nickel oxides are reduced to gain black or raw copper; the second stage should produce a lead-tin alloy (see Figure 1).

Figure 1:

Process for the extraction of valuable metals from converter slag

The trials were performed with converter slag and silica (to define the Fe/SiO2 ratio). Preliminary to experimental investigations, a thermodynamic modelling of the process using HSC Chemistry and FactSage was done.

2

Possibilities of slag treatment

Basically, it is possible to reduce metal oxides out of a slag phase with various reduction agents. Examples are carbon, hydrogen, carbon monoxide, metals and methane. The investigated possibility is the reduction with iron. Here, iron oxide (FeO) is built, which is bound in a fayalitic slag by the addition of silica [1, 2, 3].

864

Proceedings of Copper 2010

Recovery of Valuable Metals form Converter Slags by Reduction with Iron

2.1 Application of iron as reduction agent The use of iron as a reduction agent is already known and was investigated by Metallo-Chimique N.V. among others. The company has implemented Top Blown Rotary Converters (TBRC) in the production process to recycle copper and the accompanying elements tin, lead, nickel and zinc. Following, the investigations conformable to [1] are described shortly. A rotating furnace causes a good bath agitation, and this provides a good mass exchange between metal and slag. There are no hot-spots and the temperature allocation is very homogenous. As a result, zinc and lead are hardly volatilised and the recycling of these elements is improved. In addition, the good mass transfer allows a low process temperature (1180 °C). At the beginning of the process, a stoichiometric amount of iron is given into the bath. Viscosity and relative density are lowered by adding bases and flux. Chemical bound copper and other metal compounds are reduced by iron, which is added in form of scrap below the bath surface. Due to this, a fayalitic slag is formed and the following reactions occur: MeO + Fe
FeO + Me

(1)

(MeO)x·SiO2 + xFe
(FeO)x·SiO2 + xMe

(2)

xFeO + SiO2
(FeO)x·SiO2

(3)

Because of these exothermic reactions the temperature is getting higher and the volatilisation of lead and zinc will come into favour. It is better to win lead and tin in a separate process step. Therefore, iron is added again to the process. The resulting slag has to be recycled and is given back to the melting process.

2.2 Application of gas as reducing agent At Olympic Dam Operations of WMC Resources Ltd., a process has been tested, where a methaneair mixture for the reduction of valuable metals was blown into the “direct to blister” slag in an electro furnace. On the bath surface, a layer of coke ensures the reducing conditions. The metal is settling down through the slag to the bottom of the furnace. This process needs plenty of time because of the bad coagulation of the fine metal droplets in the contact area between the coke and the slag. The methan-air mixture improves the kinetics of the process, which is described in detail in [3].

3

Modelling with HSC Chemistry and FactSage

To get some information about the two-staged process, a thermodynamic model has been developed using HSC Chemistry and FactSage. With HSC Chemistry, the energy and material requirements were calculated and balanced. The calculation with FactSage was to determine the distribution coefficients of the equilibrium for a non-ideal dissolving behaviour of the individual phases. Proceedings of Copper 2010

865

Konetschnik, Paulitsch, Pesl, Antrekowitsch

3.1 Build-up of the model The scheme of the model is shown in Figure 1. Input streams are the converter slag, a reduction agent, silica, natural gas and air. Two metal phases as well as slag and off gas are defined as output streams. The composition and amount of the input materials as well as the excess of the reduction agent can be varied. A post-combustion and a defined Fe/SiO2 ratio are also implemented in the model.

3.2 Results The modelling was done with varying amount of reduction agent and temperature. In the calculations, a stoichiometry of 1 was described as followed: 90 per cent copper, 50 per cent nickel, 25 per cent lead, 30 per cent tin, 10 per cent zinc and 30 per cent arsenic reduced out of the slag. For these values, an addition of about 34 per cent FeCu80 to the slag is necessary. The effect of the stoichiometry on the reduction degree of the individual metals is shown in Figure 2. At a value of 0.5, half of the copper oxide is reduced and at stoichiometry 1.25 there is more than 95 % of the copper in the metal phase. A significant amount of lead and tin are reduced, too. A decreasing temperature sets the thermodynamic equilibrium to a higher reduction degree, although the influence is rather minor. It is important to mention that the influence of the kinetics and problems with settling down of the metal phase through the slag at lower temperatures are not considered in a thermodynamic model. There is also an impact on the equilibrium by the basicity. It has to be kept in mind that the Fe/SiO2 ratio also has an influence on the viscosity of the slag. The higher the amount of silica and consequently the lower the basicity, the higher is the total amount of the slag. So, the activity of the valuable metals in the slag is less and metal losses are higher.

866

Proceedings of Copper 2010

Recovery of Valuable Metals form Converter Slags by Reduction with Iron

Figure 2:

Dependence of stoichiometry to the reduction degree of Cu, Ni, Pb and Sn

As a result of a larger amount of the reductant, metal oxides are reduced in a higher grade and the copper content of the metal phase is decreasing. The dependence of the stoichiometry on the composition of the metal phase is shown in Figure 3.

Proceedings of Copper 2010

867

Konetschnik, Paulitsch, Pesl, Antrekowitsch

Figure 3:

Dependence of the black copper on the stoichiometry

The more reducing agent is added the higher is the amount of iron oxide in the slag. Resulting from this, the silica addition has to be raised because of the required constant ratio of Fe/SiO2.

4

Experimental investigations

The first series of trials was done in a muffle furnace with samples of 50 g converter slag. This laboratory test should give information about the necessary mixture of feedstock to reach a successful reduction of the valuable metals. Additional to these trials industry-oriented experiments were done with a sample weight of 5 kg using an induction furnace. The metal phase was tapped twice and the reduction agent was added four times. After each addition the slag was sampled.

4.1 Muffle furnace For this first evaluation of the possibility of recovering valuable metals from a converter slag by iron, the composition of the raw material was varied as well as the temperature and the stoichiometry. Crucibles made of alumina provided the necessary temperature durability. For a part of the trials, iron crucibles simulated an excess of reductant. 868

Proceedings of Copper 2010

Recovery of Valuable Metals form Converter Slags by Reduction with Iron The necessary amount of the reductant was calculated from the model according the designated stoichiometry. Analogical to the model, a stoichiometry of 1 was defined as the following reduction degrees: 90 per cent copper, 50 per cent nickel, 25 per cent lead, 30 per cent tin, 10 per cent zinc and 30 per cent arsenic. Also, a stepwise variation of the content of tin and lead oxide in the slag should show the influence of a changing slag composition. Metallic copper has been placed on the bottom of the crucible at some trials to improve the coalescence of the metal phase. For better kinetics the pulverised material had to be compressed. The mixture of slag, reductant and fluxes was heated in a Nabertherm high temperature furnace to 1200 and 1300 °C respectively and held there for accordingly four and two hours. For an inert atmosphere nitrogen was used. The metal phase was weighted and analysed. Several problems occurred within the experiments: At trials with a low stoichiometry there was no single metal regulus formed and the metal droplets were distributed in the slag in a very fine form. So, a complete separation of slag and metal phase was not possible. The more reducing agent has been used the better the coalescence of the metal phase was. Also, there was no possibility to separate the slag and the crucible completely. As a result, the amount of the slag phase had to be calculated by the material balance. Another problem was that the reaction between the iron crucibles and the slag was very strong, mainly at 1300 °C.

4.2 Induction furnace The planning of these experiments happened iterative, so that problems should be avoided in the following trials. The starting point was defined as follows: A temperature of 1250 °C and at the end of the two-staged process a double stoichiometric amount of the reducing agent (FeCu80 and Fe respectively). To gain an approximately equal amount of lead and tin in the slag, about 5 per cent of lead oxide had to be added. At the first trial, the final slag was highly viscous due to a varying Fe/SiO2 ratio. Some thermocouples took care of the exact temperature control. However, a temperature deviation of 50 °C between the surface and bottom could not be avoided. To realise a reducing atmosphere, the clay graphite crucible was covered with a graphite plate. An additional purging with nitrogen improved the avoidance of contact with oxygen. After tapping the first metal phase, the viscosity of the slag increased clearly, which lead to a wrong Fe/SiO2 ratio. So the temperature was increased to 1300 °C. Also, a part of the added iron did not dissolve. Due to this, the later trials were done at a final stoichiometry of 1.75 (1.5). Two insolvable metal phases were built in the second stage according to thermodynamic calculations. The chronology of the experiments was as follows:

Proceedings of Copper 2010

869

Konetschnik, Paulitsch, Pesl, Antrekowitsch

1 Charging the cold crucible with converter slag, reduction agent and silica (as well as copper master alloy for the third trial) 2 Heating on process temperature, 30 min holding time at temperature, sampling 3 Charging further reductant and according silica for designated Fe/SiO2 ratio 4 90 min holding time, followed by sampling 5 Tapping of the metal phase 6 Charging of reduction agent and silica (as well as Pb/Sn master alloy for the third trial) 7 Reheating on process temperature, 30 min holding time, sampling. 8 Charging of reduction agent and silica 9 90 min holding time, followed by sampling. 10 Tapping, solidification and separation of the slag and the metal phase.

5

Results

In this chapter, the results of the experimental investigations are described and discussed. The reduction degree of the elements copper, lead and tin is calculated as follows: Cureduction degree = (Cumetal - Cureducing agent - Cumaster alloy)/Cuconverter slag

(5)

Pb|Snreduction degree = Pb|Snmetal/Pb|Snconverter slag

(6)

5.1 Muffle furnace Figure 4 shows the reduction degree of copper, lead and tin in dependence of the amount of reduction agent and the stoichiometry respectively.

870

Proceedings of Copper 2010

Recovery of Valuable Metals form Converter Slags by Reduction with Iron

Figure 4:

Reduction degree of Cu, Pb and Sn by using a copper master alloy

Despite the problems of settling down through the slag phase and the infiltration of the iron crucible, following statements can be made: •

At a stoichiometric value of the reductant, the generation of the metal phase is insufficient. The reduction degree according to equation (5) is about 20-30 per cent.

•

A stoichiometry of about 1.25 results in a disproportional higher metal recovery. So, the calculated reduction degree reaches a value of 75 per cent.

•

The reduction degree of lead is significantly lower than that of copper.

•

Iron crucibles are highly infiltrated by the metal phase during the process. To avoid these procedural problems, the simulation of an excess of reductant should be realised in another way.

•

Due to the lower viscosity, the coalescence of metal droplets is better at higher temperatures. This improves the recovery of copper despite of the thermodynamic equilibrium, which is contrary.

•

A copper master alloy improves the coagulation of the metal droplets especially at a low stoichiometry.

•

Higher amounts of lead and tin in the slag result in a lower reduction degree of copper.

Proceedings of Copper 2010

871

Konetschnik, Paulitsch, Pesl, Antrekowitsch Slag analysing is another way for proving the success of the experiments. Following statements can be made: •

The higher the amount of the reduction agent, the more valuable metals can be removed from the slag.

•

The influence of the temperature on the metal content of the slag is given but rather low.

•

By this first step of reduction, the copper content can be decreased to about 5 per cent. The lead and tin content can be lowered to 2 per cent.

•

A copper master alloy has no influence on the content of valuable metals in the slag after separation of the metallic inclusions.

•

The addition of lead and tin oxide results in a larger metal volume but also in higher absolute losses of lead and tin in the slag.

5.2 Induction furnace For evaluating the two-staged trials, the two produced metal phases as well as four slag samples were analysed. An exact material balance was not possible because of the lack of possible measurement of the slag amount. The reasons are the losses of slag on the crucible wall, thermocouples and so on. So, the black copper phase and the Pb/Sn alloy have been weighted and the slag amount could roughly be evaluated by the difference of the sum of the metal phases to the input material. Due to the circumstance that sampling of the slag is more representative, the reduction degree of the valuable metals was calculated by the remaining metal amounts in the slag. Trial 2 shows that at a temperature of 1300 °C and a stoichiometry of 1.25, about 4 kg of black copper out of 5 kg converter slag can be produced. At this point, it has to be mentioned that the clay graphite crucible was involved in the reduction. Beside copper significant amounts of lead, tin and nickel were reduced. Because of the iron content of about 10 per cent, an excess of reductant because of the clay graphite crucible is probable. The amounts of copper, lead and tin during the process are shown in Figure 5.

872

Proceedings of Copper 2010

Recovery of Valuable Metals form Converter Slags by Reduction with Iron

Figure 5:

Developing of Cu, Pb and Sn in the slag during the individual trials

Most of the reduction work is done at the beginning of the process. Therefore, a strict separation of copper and nickel from tin and lead is not possible. Beside this, the content of these valuable metals in the slag is consistently below 1 per cent. These values are below the concentration in the shaft furnace slag. Trial 2 shows a negative material balance for tin and lead, which is caused by the volatilisation during the experiment. The reduction degrees of the considered metals are mostly distinctly more than 97 per cent. This differs noticeable from the values reached in the muffle furnace. The reasons are the higher temperature and the better possibility for nodulising at the significant larger sample size. As a result of the equilibrium of [Me]/(MeOx), the use of a master alloy leads to higher losses of these metals in the slag. Because most of the copper is removed after the stage 2, there is only a marginal influence on the copper content of the final slag by using a master alloy before the first tapping.

6

Comparison of the thermodynamic modelling and the trials

In this chapter, the thermodynamic calculations and the experiments will be compared. A complete comparison of model and the experimental work in the muffle furnace as well as in the induction furnace is not possible due to the different parameters for the individual series of investigation.

Proceedings of Copper 2010

873

Konetschnik, Paulitsch, Pesl, Antrekowitsch

6.1 Comparison model – muffle furnace The indicator for the achievement of the process is the reduction degree. In Figure 6, the calculated and experimentally determined values are shown for a temperature of 1200 °C.

Figure 6:

Comparison of the reduction degree of Cu, Pb and Sn between model and trial in dependence of the stoichiometry

The values for lead and tin agree between model and experiment. The copper content in the metal phase is lower than the thermodynamically calculated one. One reason is the problem of forming a complete regulus because of incomplete settling.

6.2 Comparison model – trials in the crucible induction furnace The model is calculated two-staged which is equivalent to stage 2 and 4 of the experiments in the induction furnace. Thermodynamic software packages calculate the equilibrium and do not consider the kinetics or incomplete settling. Also, the influence of nitrogen which is blown on the surface of the slag is very difficult to include in thermodynamic models. Despite these problems, following statements can be made:

874

Proceedings of Copper 2010

Recovery of Valuable Metals form Converter Slags by Reduction with Iron

•

Phase separation: The thermodynamic model indicates a phase separation of the metal phase after the first tapping. A Pb-Sn-Cu phase is in thermodynamic equilibrium with a Fe-Sn-Cu one. This corresponds with observations during the experimental investigations.

•

Amount of the phases: The amounts of the phases which are generated in trial 1 are similar to the calculations. Experiments 2 and 3 show a significantly higher amount of metal than the model. The reason is the reduction force of the clay graphite crucible. At process temperature, the crucible has a higher reduction potential than iron.

•

Copper: The copper content of the black copper is lower (55-60 per cent) than the model shows (80-85 per cent). In the slag, the copper content equates the calculated one.

•

Lead: In the first reduction stage, significantly more lead is reduced than predicted. This also indicates that the carbon of the crucibles acts as a reductant.

•

Tin: This element shows the same behaviour as lead, but in a smaller dimension. This means that the absolute tin content in the metal phase is too high.

•

Iron: Here, the biggest difference between real and ideal behaviour occurs. According to the equilibrium, the iron should be completely slagged. In contrast, the experiments show a considerable solution of iron in the black copper.

•

Fe/SiO2 ratio: Because of a low slag viscosity, the Fe/SiO2 ratio should be in an area around 1.6. Therefore, the necessary silica addition was calculated for this value. Because of the high iron content in the black copper, the amount of FeOx in the slag is rather low. At the end of the process, a ratio of 1.0 was determined.

At the experimental investigations in the induction furnace, the reduction degree is also an important parameter for evaluating the success of the process. Table 1 shows the comparison of the reduction degrees after the last process step. Table 1:

Comparison of the reduction degree after the fourth process step Trial 1

Trial 2

Trial 3

Experiment

Model

Experiment

Model

Experiment

Model

Cu

97.25 %

99.66 %

98.15 %

99.58 %

99.47 %

99.75 %

Pb

97.86 %

93.59 %

98.52 %

92.49 %

88.65 %

93.08 %

Sn

98.04 %

83.24 %

98.68 %

78.47 %

92.88 %

66.02 %

In spite of the availability of carbon, the experiments reached a reduction degree of copper lower than the calculated one. One reason could be that a part of the metallic copper was distributed in the slag. The absolute amount of reduced copper cannot be compared with industrial scale. Contrary, the reduction degree of lead and tin are distinctly better than predicted.

Proceedings of Copper 2010

875

Konetschnik, Paulitsch, Pesl, Antrekowitsch

7

Conclusion

This paper deals with a method to reduce valuable metals out of a converter slag by iron and an iron copper alloy respectively. The two-staged process should form a black copper melt as well as a lead/tin alloy. Within the project, a thermodynamic modelling and two series of experimental investigations were done. For the first trials, a muffle furnace with small alumina as well as iron crucibles was used. The process was done one-staged for evaluating the principal possibility of the method. Some problems concerning the settling of the metal phase occurred, primarily by lower amounts of the reductant. Lead and tin react ideally as predicted by thermodynamic calculations. The losses of metallic copper in the slag are high, especially at lower temperatures. Iron crucibles simulate an excess of reductant, but the dissolution by the metal phase results in procedural problems. The higher the amount of the reduction agent, the more copper and valuable metal oxides can be reduced, and the reduction degree is increasing to 80-90 per cent. Due to a higher temperature, the process of settling is better caused by the lower viscosity and the faster coagulation. Therefore, the reduction degree of copper is better at higher temperatures, contrary to the thermodynamic equilibrium. Using copper as a master alloy, an improvement of the coagulation of the metal droplets can be observed. Higher amounts of tin and lead oxides in the converter slag cause a lower reduction degree for copper. Also the losses of lead and tin in the final slag are higher in this case, although the amount of reduced metal increases. The second series of investigations was done two-staged in an induction furnace with a clay graphite crucible. According to the model, the whole iron should be slagged when reducing the valuable metals. However, up to 18 per cent of iron was measured in the metal phase although the reduction degrees of the valuable metals accord much better to the predicted values. This is caused by the reduction potential of the clay graphite crucible which participates in the reduction. Due to these circumstances a changing of the Fe/SiO2 ratio and an increasing slag viscosity took place. Fluxes and an exact Fe/SiO2 ratio should improve the process at lower temperatures. A lower amount of reduction agent causes a better separation of copper and nickel from lead and tin, however, the copper losses in the Pb/Sn alloy are high. After both stages, a copper, lead and tin recovery rate of 97 per cent could be achieved. The copper content of the final slag is below 1 per cent and is adequate for a secondary copper facility. Further investigations will be done to improve the separation of copper and nickel from lead and tin to form a marketable Pb/Sn alloy. So, alternative reduction agents have to be considered as well as changes in the slag composition and process flow.

876

Proceedings of Copper 2010

Recovery of Valuable Metals form Converter Slags by Reduction with Iron

References [1] DIERCKX, L., BEERSE & FERON, D. (1972): Copper Refining Process. Aug. 1972. Patent No. 3,682,623 [2] TARASOV, A. V & KLUSHIN, S. D. (2000): Removal of copper from slag with the aid of reducing and sufiding gas mixtures, edited by Stewart D. L., Daley J. C., Stephens R. L., TMS (The Minerals, Metals & Materials Society) [3] AN, X., LI, N. & GRIMSEY E. J. (1998): Recovery of copper and cobalt from industrial slag by top-submerged injection of gaseous reductants. EPD Congress 1998, Western Australia, The Minerals, Metals & Materials Society 1998 [4] MALONE, J. G., LEAHY, G .J., PLAYER, R .L. & WRIGHT, D. J. (1985): The mechanisms of reduction of liquid slag by carbon. 24th Annual Conference of Metallurgists, Symposium on Quality Control in Non-Ferrous Pyrometallurgical Processes, Vancouver, Canada, August 1985, The Metallurgical Society of CIM, The Canadian Institute of Mining and Metallurgy, 216-273

Proceedings of Copper 2010

877

878

Proceedings of Copper 2010

Waste heat boilers for Copper Smelting Applications Dipl.-Ing. Stefan Köster Oschatz GmbH Westendhof 10-12 Essen, Germany

Keywords: Waste heat boiler, flash smelting furnace, cyclone reactor, gas flow, optimization

Abstract The increasing energy price level due to lack of resources require plant operators in the metallurgical industry worldwide to utilise the full energy potential of their plants. Therefore, the process gas cooling with waste heat boilers (WHB) downstream pyro-metallurgical processes and the utilisation of the recovered heat is a basic condition and it is also important with regard to environmental aspects. Furthermore the process operation has to be efficient and economical. As a part of the plant a waste heat boiler system has to be designed to achieve these requirements. The gas cooling concept downstream Flash Smelting Furnace (FSF) for copper is a horizontal WHB consisting of a section for radiation and for convectional heat transfer. By understanding the specific process requirements based on the experience with plants in operation, the design of this kind of WHB was improved. The application of computer flow modelling led to further improvements by optimizing the gas flow profile. The development of this WHB design will be described in the example of the Aurubis AG (earlier Norddeutsche Affinerie AG) Flash Smelting plant which was extended and modernized several times from the start up in 1972 till today. A development has also taken place for the WHB cleaning systems installed to prevent the fouling on the heating surfaces and to improve the cooling efficiency. The different technologies will be described and compared. Another objective of this article is cyclone smelters. These smelters apply the CONTOP® technology which was originally developed for the treatment of copper concentrates by KHD Wedag AG in Cologne, Germany. In the smelting reactor copper, zinc and lead oxide are recovered from secondary oxide material with its origin in the steel and galvanizing industry. This technology will be described in the example of conducted plants for Asarco, Codelco and Harzer Zink.

Proceedings of Copper 2010

879

Köster

1

Introduction

The Oschatz GmbH was founded in 1849 and is one of the last private owned WHB companies in the world. Oschatz’s range of products comprises process gas cooling solutions for the metallurgical and chemical industry and turnkey solutions for the thermal utilisation of residues, biomasses and alternative fuels. Oschatz is always in close dialogue with plant operators and considers their individual operational requirements. With 100 % subsidiaries for standardized detail engineering and manufacturing and with representatives all over the world it has a high degree of flexibility and it has achieved an independent market position. In the last decades Oschatz has developed from a subsupplier of general contractors to an innovative partner of plant operators. For the copper industry Oschatz develops and supplies WHB downstream processes like Isasmelt, Flash and Cyclone Smelting. Some of the corresponding WHB geometries are shown in Figure 1.

Figure 1:

2

Typical WHB geometries

Industrial requirements

The primary off gas treatment downstream smelting process for copper is done in a WHB. The boiler has to drop the process gas temperature to a level suitable for further treatment e.g. in an electrostatic precipitator, in a bag house or for sulphur recovery. The dust coming from the copper smelting process is sticky. Before a further treatment is possible, the sticky components have to be converted in the WHB from the molten to the solid phase. The heating surfaces have to be cleaned from the dust and the dust has to be discharged continuously. To avoid the risk of sulphur corrosion on the WHB tubes the conversion of the process off gas to SO3 has to be minimized. As an effective measure to protect the WHB from the corrosion, the heating surface walls and elements have to be kept above the dew point. Thus the water/ steam operating pressure is usually chosen between 40-70 bar to get a saturated steam temperature between 250 and 285 °C. An advantage in using a WHB for the process gas cooling is the possibility of heat recovery and converting this heat into power. The power can be used for the process itself (e.g. concentrate drying) to reduce the process costs and/ or for profitable network electricity supply.

880

Proceedings of Copper 2010

Waste Heat Boilers for Copper Smelting Applications

3

Basics water circulation system

The WHB water circulation system is a combination of natural and forced circulation. The natural circulation system works self-sustaining with the off gas heat. The connection of the heating surfaces to the natural circulation system leads to smaller pump capacities and electrical power consumption for the forced circulation pump drives. Thus investment and operation costs are saved. The water flow starts at the steam drum through the suction line to the circulation pump group which delivers the water through the different evaporator heating surfaces and through the main discharge lines. Two pumps are usually used, one pump is in continuous operation and the other one is in stand by function. The operation pump is driven by an electric motor and the stand-by pump by a steam turbine fed by external steam or also by an electric motor with a separate energy supply. By the heat absorption the circulation water partly evaporates and the resulting water/ steam mixture flows from the heating surfaces back to the steam drum via the main piping lines. Water and steam are separated in the steam drum. The water returns to the circulation system, the saturated steam flows through the steam drum demister in which the steam is dried. The droplets formed on the steel meshes of the demister fall back into the water. After the demister the steam leaves the drum and flows through the saturated steam line. The evaporated water consumption is compensated by water coming from the feed water system so that the water level of the steam drum is constant. Each heating surface is equipped with one distributor and one collector. The water flow direction is from the distributor through the heating surface to the collector. The inlet of the tubes is equipped with orifices or nozzles to get uniform water flow for each tube which is a necessary condition for a sufficient cooling effect.

4

Waste heat boiler downstream flash smelting process

The WHB downstream FSF usually has to handle an off gas flow of up to 140,000 Nm³/h. The off gas inlet temperature range is approximately 1250 °C up to 1350 °C. The WHB downstream flash smelting process shown in Figure 2 is divided into two parts. The first part of the WHB is for the heat transmission by radiation. The second part with heating surfaces in form of bundles in the gas flow cross section is for the convectional heat transmission. The temperature at the outlet of the radiation section is approximately 650 °C up to 700 °C and at the outlet of the convection section it is usually between 350 °C and 400 °C.

Proceedings of Copper 2010

881

Köster

Figure 2:

WHB downstream Flash Smelting process

Table 1:

WHB downstream Flash Smelting process (Figure 2)

No.

Item

No.

Item

1

Flash Smelting Furnace

7

Buckstay, reinforcement

2

Transition to WHB

8

Support beam

3

Radiation section

9

Convection screen bundle

4

Radiation screen bundle

10

Convection evaporator bundle

5

Steam drum

11

Drag chain conveyor

6

Hammering cleaning device

In the hopper area of the convection section a separation of the dust load into a recycle and a product portion can be realised. In Figure 3 the WHB downstream FSF for Aurubis AG (earlier Norddeutsche Affinerie AG) is shown. The WHB was built in 1972. Since then it was modernized several times until today. In the radiation section a baffle wall element was installed as a flow disturbance to direct the off gas flow to the lower part of the WHB radiation section. A better heat transmission should be achieved by this. During the years the radiation section was extended two times. The convection section was extended as well and one evaporator bundle was installed at the WHB end. Two evaporator bundles at the convection section inlet had been replaced by screen bundles which are coiled parallel to the off gas flow. Three screen bundles are now installed in the radiation section.

882

Proceedings of Copper 2010

Waste Heat Boilers for Copper Smelting Applications

Figure 3:

5

WHB downstream Flash Smelting process for the Aurubis AG in Hamburg, Germany

Heating surface cleaning systems

The application of a heating surface cleaning system for the WHB walls and bundles is necessary because of the high dust load of gases generated from copper smelting processes and their sticky characteristic. During the process the dust deposits on the boiler tubes, especially on the tubes of the evaporator screens and bundles. This influences the heat transfer to the water circulation system and in consequence the gas outlet temperature increases. A regular and effective cleaning on the gas side of the WHB is obligatory. Three different types of WHB cleaning devices with different functional designs exist: • mechanical rapping cleaning devices operated by pneumatic cylinders, • mechanical hammering cleaning devices operated by electric motors and • pneumatic hammering cleaning devices operated by pneumatic actuators. In Figure 4 the mechanical rapping cleaning device for a screen bundle is shown. This system is fixed at the tube wall casing and connected through the walls with the screen bundle layer. The movement of the rapping devices and the screen bundle respectively is activated by a pneumatic cylinder. The bundle movement forces the dust deposits to fall off and a sufficient heat transfer is ensured.

Proceedings of Copper 2010

883

Köster

Figure 4:

Side view of a mechanical rapping cleaning device for a screen bundle

The different installation places at the WHB heating surfaces influence the number of hammering cleaning devices operated by one electric motor: • single hammering cleaning devices with one motor drive and one shaft, • multiple hammering cleaning devices with one motor drive and one shaft with two or three devices and • bundle hammering cleaning devices with one motor drive and one shaft with up to 20 devices next to each other.

884

Proceedings of Copper 2010

Waste Heat Boilers for Copper Smelting Applications

Figure 5:

Side view of a single mechanical hammering cleaning device

Table 2:

Mechanical hammering cleaning device (Figure 5)

No.

Item

No.

Item

1

Bolt wheel

6

Hammer weight

2

Holes for the bolts (1 - 7)

7

Torque bracket

3

Protected bolt (2)

8

Bearing bracket

4

Cogwheel

9

Driving shaft

5

Hammer shaft

The mechanical hammering cleaning device is fixed at a cogwheel which can be rotated. The bracket for this construction is welded to a web plate on the tube wall. The shaft above the cogwheel is driven by an electrical motor. Its rotation is transferred to the wheel via a disc which is fixed on the shaft. This disc has seven bores for bolts and the bolts engage in the cogwheel when it rotates. By this mechanism the hammer is lifted. The hammer rotates back when the last bolt leaves the cogwheel and impacts onto the anvil or pestle. The vibration of the impact is transmitted to the tube walls or bundles. This effect loosens the dust deposits and the dust falls off. For a multiple hammering cleaning device the angles of the shaft discs are different so that the hammers are not lifted at the same time. This is done to protect the electrical motor against overloading. The number and position of the fitted bolts determine the impact force of the hammer on the anvil or pestle. More bolts Proceedings of Copper 2010

885

Köster mean a longer distance of acceleration for the hammer and this is followed by a stronger impact. As the bolts are subjected to wear it should be aspired to keep the impact force as low as possible. An appropriate adjustment is determined by the experience with the dust load of plants in operation. Inclined tube walls are the application for pneumatically operated hammering cleaning devices. Depending on the inclination angle the pneumatically operated cleaning device is more effective than the mechanical one which needs more space because of the acceleration distance. If this space is not sufficient the cleaning effect is too low. These devices are composed of: • a pestle mounted on a special bearing plate installed on the tube wall and • a pneumatic impactor for the direct pestle actuating.

Figure 6:

Side view of a single pneumatic hammering cleaning device

Table 3:

Pneumatic hammering cleaning device (Figure 6)

No.

Item

No.

Item

1

Piston

4

Pestle

2

Air hose

5

Tube

3

Impact plate

The impactor of the heating surface cleaning device in Figure 6 is a pneumatic hammer. The pneumatic piston is pushed against one or two springs by compressed air. The hammering effect is induced by a fast releasing of the air under the piston through a solenoid valve. The piston is shot by springs against an impact plate and this transmits the impact to the pestle. 886

Proceedings of Copper 2010

Waste Heat Boilers for Copper Smelting Applications In Table 4 a comparison between the rapping and the hammering cleaning system is summarised. The hammering cleaning system has clear advantages to the rapping system. As a consequence Oschatz today mainly focus on this system. Table 4:

Comparison rapping and hammering cleaning device

Criterion

Rapping cleaning system

Hammering cleaning system

Vibration on boiler casing

high

low

Bundle blocking

high

minor

Cleaning effectiveness

good

very good

Installation time

quick

very quick

complex

simple

limited

wide range

no

yes

high

low

Design complexity Modification possibility Upgrade capability Operation costs

6

CONTOP cyclone technology

The Klöckner-Humboldt-Deutz AG in Cologne, Germany developed the CONTOP® process applied in a cyclone smelter. This technology was originally foreseen for the treatment of copper concentrate. In the iron and steel industry it is used primarily for the treatment of zinc residues and for the shredder light fraction in the automotive recycling industry. Oschatz had built since today three CONTOP® cyclones. Engineering has been done for a fourth cyclone. Two plants were built for ASARCO in El Paso, Texas, USA. With a capacity of 32 t/h of copper-concentrate mix for each cyclone they are the largest in the world. The third CONTOP® cyclone was for Codelco in Chile and the fourth one was for Harzer Zink GmbH in Goslar, Germany. Due to the increasing zinc content in recycled steel scrap the application of the CONTOP® cyclone is an important environmental technology. Zinc accumulates in the sludge and dust of the steelmaking process. Especially for Electric Arc Furnaces (EAF) the complete scrap is charged. The zinc content is extracted in the cyclone reactor with a capacity range of 5000 t/a to approximately 100,000 t/a. Larger cyclones are suitable for the zinc recycling of the regional ferrous industry, smaller units are more applicable for single steel plants. The CONTOP® technology is also important for the automobile recycling industry where shredder light fraction has to be treated. Over 50 million automobiles are produced every year. The recycling Proceedings of Copper 2010

887

Köster of an automobile usually takes place after an average service lifetime of twelve years. 70-80 % of the automobile weight is metal and reused in the metals industry. The rest (glass, rubber, plastics) has to be disposed under consideration of environmental aspects. The combustion of the shredder light fraction in a CONTOP® cyclone is an alternative to the incineration or dumping of these residues.

Figure 7:

Cyclone Reactor

Table 5:

Technical data for the WHB downstream Cyclone Reactor (Figure 7)

Parameter Off gas rate Fume and dust inlet Temperature WHB inlet Dust WHB outlet Temperature WHB outlet Steam pressure Steam rate 888

Value

Unit

5000-50,000 Nm³/h 0.7-7 t/h 1200-2000 °C 0.8-8 t/h 240-380 °C 30-70 bar 6-60 t/h Proceedings of Copper 2010

Waste Heat Boilers for Copper Smelting Applications Table 6:

Technical data for the Cyclone Reactor (Figure 7)

Parameter

Value

Unit

Cyclone diameter

1-3 m

Feed rate of solids

10-100 t/h

Off gas temperature at outlet Steam pressure Steam rate

1200-2000 °C 30-70 bar 3-10 t/h

The feed material with oxygen and solid fuel is tangentially injected into the CONTOP® cyclone. The cyclone smelting reactor is evaporative-cooled and the water circulation system pressure is high enough to keep the tube wall temperature above the off gas dew point. The process temperature range is between 1200 °C and 2000 °C. In the reactor, molten droplets are generated and projected centrifugally onto the cyclone tube walls. The resulting slag forms a protective lining layer. A continuous slag flow downwards the cyclone outlet together with the gas can be observed. The intense tangential gas motion leads to a carry-over rate of the input material of 2-3 %. In another chamber, a separation of the gas and the molten slag takes place. After this the off gas is post combusted with air to ensure a complete oxidation of the vaporized zinc and CO. In a bag filter the zinc oxide is removed from the off gas. This product can directly be sold to zinc producers. The slag can also be used for several purposes, for example in the construction industry.

Proceedings of Copper 2010

889

Köster

Figure 8:

Schematic drawing of a Cyclone Reactor [3]

The off gas heat is recovered in the cyclone cooling system by producing steam which can be used for the process or for electricity generation in a turbine. A schematic drawing of the CONTOP® process is shown in Figure 8. The CONTOP® plant at Harzer Zink GmbH in Goslar, Germany is operating since 1996. The plant was foreseen to treat retort residues from the production of zinc from the so called New Jersey process. During plant operation these retort residues were substituted step by step by zinc-bearing dusts and sludges generated in the iron and steel industry. After shut down of the New Jersey plant in the mid-2000 dust and sludges from several German EAF steel plants were treated in the CONTOP® cyclone. The off gas heat is recovered in the cyclone by steam generation which is used for the transmission to electrical power. The main benefits of the CONTOP® technology are as follows: • copper and zinc recycling, • waste heat recovery for power generation, • low emissions, • environmental products and • clean and safe working conditions. 890

Proceedings of Copper 2010

Waste Heat Boilers for Copper Smelting Applications

7

Summary

WHB for copper smelting applications have been introduced in this article on examples of WHB downstream flash smelting process and the CONTOP® cyclone for the treatment of copper, zinc and lead residues. The industrial requirements have been described and the differences between the various heating surface cleaning systems for WHB in the copper industry have been pointed out. The sticky characteristic of gases generated in copper smelting applications requires innovative gas cooling solutions. WHB systems in the copper industry have to be adaptable to process capacity increases with concentrate changes, higher off gas flows and inlet temperatures. The several modifications on the WHB for Aurubis AG in Hamburg, Germany show this very clear. The experience with plants in operation is the basis for a continuous development of WHB in this field. The copper industry gives various starting points for further improvements in the WHB off gas treatment. The focus has to be set on the optimisation of existing and the development of new technologies.

References [1] S. Köster: Modern waste heat recovery systems for the non-ferrous industry. GDMB - Global Growth of Nonferrous Metals Production, Volume 1, S. 103–115. Presentation at the European Metallurgical Conference 2009, Innsbruck, Austria, June 28 - July 1, 2009 [2] F. Sauert: CONTOP® - A Cost-Effective Recycling Technology for the Steel and AutomotiveScrap Industry. Siemens VAI Metals Technologies GmbH & Co Linz / Austria [3] Ulrich Kerney: The Harzer Zink Smelting Reactor Process for Zn Bearing Secondaries. Intensivierung metallurgischer Prozesse, Heft 87 der Schriftenreihe der GDMB Gesellschaft für Bergbau, Metallurgie, Rohstoff- und Umwelttechnik, Clausthal-Zellerfeld 2000, S. 155–169.

Proceedings of Copper 2010

891

892

Proceedings of Copper 2010

Processing of High-Silicon Copper Sulfide Concentrates by Vanyukov Smelting S. M. Kozhakhmetov, S. A. Kvyatkovskiy, E. A. Ospanov, Z. S. Abisheva, A. N. Zagorodnyaya Center of Earth Sciences, Metallurgy and Ore Beneficiation Shevchenko Str., 29/133 Almaty, Republic of Kazakhstan

Keywords: Copper concentrates, Vanyukov furnace, fluxes, dust, rhenium recovery, leaching, extraction

Abstract Today pyrometallurgy is the most widely applied industrial practice of copper recovery in Kazakhstan. The ores occurred in Kazakhstan may be conventionally classified in two groups by basic components in produced concentrates, i.e., Zhezkazgan ores concentrates currently processed by electric smelting methods and East Kazakhstan ores concentrates processed by autogenous smelting technique. The first ones are differed for their increased concentrations of copper and silicon dioxide while the second ones contain relatively increased concentrations of iron and sulfur. Besides, the abovementioned technologies are based on the use of different sub-products and fluxes without valuable components. Autogenous smelting of balanced mixtures of both concentrates provides a partially or totally fluxes-free processing resulting in reduced bulk slag volumes and valuable components losses. Changing in Zhezkazgan ores compositions results in decreasing of iron concentrations and increasing of zinc, lead and arsenic sulfides concentrations in concentrates. Reduced concentrations of iron sulfides lead to shifting heat balance of autogenous smelting process that requires increased desulfuration degree which, in its turn, results in changing of slag compositions. Moreover, substantial technological difficulties in high-silicon raw materials processing are due to increased volumes of used calcic fluxes and iron-containing additions. Also, it shall be considered that Zhezkazgan concentrates contain rhenium. As autogenous smelting process runs at higher degrees of oxidation than electric smelting process increased rhenium concentrations in exhaust gases must be much higher. It requires of development of technology for rhenium recovery from lead dusts of copper production. Preliminary study data proved that the using of available ores deposits in Kazakhstan gives the possibility in providing source materials for autogenous smelting with minimum fluxes consumption to producing mattes of different compositions.

Proceedings of Copper 2010

893

Kozhakhmetov, Kvyatkovskiy, Ospanov, Abisheva, Zagorodnyaya

1

Introduction

Copper industry of the Republic of Kazakhstan features two types of deposits with the ore composition differing considerably. Concentrates produced from the ore mines at Zhezkazgan group of deposits contain high copper and silica dioxide. Besides, the peculiarity of those concentrates is the presence of rhenium. Concentrates produced from Eastern Kazakhstan ore contain higher sulfur and iron. Two different technologies are used for processing these concentrates. Zhezkazgan concentrates are processed in electric furnaces using calcium-containing flux. For Eastern Kazakhstan concentrates autogenous smelting in Vanyukov furnace with silica flux is used. Using flux in both technologies causes increase in feed tonnage as well as slag volume per ton of copper. One of the ways to solve this problem could be combination of these concentrates for pyrometallurgical treatment. Besides, there is possibility to determine the optimum portions of Zhezkazgan and Eastern concentrates in the feed which will allow using minimum flux. However, the existing technology for processing of high-copper concentrates – electric smelting – cannot be used for combination of different types of concentrates since it will produce low-copper matte. The main objective of this report was to confirm the possibility of processing different types of Kazakhstan concentrates by autogenous smelting in Vanyukov furnace, select optimum portions of the concentrates in the feed and investigate complex recovery of valuable components including rhenium. During review of studies related to processing of high-copper and high-silica concentrates a general trend of using autogenous smelting for processing of complex copper sulfide feed was revealed. The studies were mainly focused to reduction of losses of valuable metals into the discard slag by different slag cleaning techniques as well as directly during smelting by selecting specific feed and slag composition [1-5]. Specially distinguished here are studies done by Queensland University, Australia regarding the equilibrium composition of slag systems similar in composition to the slag produced at Kazakhstan smelters [5]. Critical review of the information on processing of high-silica and high-copper concentrates and technologies for production of blister copper has revealed a range of problematic issues, specifically selection of optimum feed and slag composition. Processing of high-silica Re-containing concentrate using autogenous smelting allows 94 % increase of rhenium recovery into the gas phase with totally up to 26 % of rhenium in the feed distributed in the metallurgical dust. Dry ESP dust is marketable product while other dusts are recycled being identical in composition to the feed. Dusts produced at one of Kazakhstan copper smelters containing, in average (mass %) 44.5 Pb; 11.7 Stot; 6.2 Cu; 5.7 Zn; 1.2 F; 1.1 As; 0.04 Bi; 0.017 Re and 0.016 Ag are fed to a lead refinery. In our opinion it could be quite reasonable to process these dusts in-house producing rhenium and salts of lead as well as zinc, cadmium and copper. In order to produce lead and zinc both pyrometallurgical and hydrometallurgical dust treatment techniques were proposed [6-10]. Pyrometallurgical processes feature a range of serious disadvantages such as generating secondary fumes, requirements for additional treatment, generating consi-

894

Proceedings of Copper 2010

Processing of High-Silicon Copper Sulfide Concentrates by Vanyukov Smelting derable volumes of gas and dust mixture, which requires gas cleaning and neutralization, low rate of recovery of rare earth metals and considerable environmental costs. Hydrometallurgical treatment is more rational from both technological and environmental points of view. Review of scientific papers and patents on dusts break-down confirmed the solutions of mineral acids, alkalis, ammonium, iron (III) and sodium salts could be used as leaching agents [6, 11, 12]. There are some works published regarding processing of the dusts with the above composition allowing recovery of rhenium. However, there are no works available on complex dust treatment. Thus, the main objective of this work is to select optimum composition of the feed using two types of concentrates available in Kazakhstan for autogenous smelting in Vanyukov furnace and further combined recovery of valuable components including rhenium.

2

Experimental

It is quite difficult to simulate autogenous smelting in laboratory conditions. However, the chemical processes occurring during smelting with oxidation by oxygen-enriched gas mixture blown through the feed preliminarily smelted in the pot can be easily described. Study of smelting and distribution of metals during processing of different copper concentrates has been undertaken. The trial plant is a furnace with a crucible installed within and the feed batched in 100 g portions. After feed smelting the melt was blown with the ambient air allowing practically complete trapping of the off-gas and fumes (Figure 1) or with air-oxygen mixture through submerged alundum pipe using other techniques (Figure 2). The experiments carried out on this plant have ultimately confirmed the estimates on oxidative smelting of concentrates mixture given below. Note that the oxygen utilization factor was not less than 90 %.

Proceedings of Copper 2010

895

Kozhakhmetov, Kvyatkovskiy, Ospanov, Abisheva, Zagorodnyaya

Figure 1:

Laboratory plant used for simulation of autogenous smelting / ambient air blowing (1 – silit furnace; 2 − graphite unit; 3 − alundum case; 4 – alundum pipe; 5 – pressure gauge; 6 – filter; 7 – rheometer; 8 – vacuum pump)

Figure 2:

Laboratory plant used for simulation of autogenous smelting / oxygen blowing (1 –furnace; 2 – alundum case; 3 – Pt-Pt-Rh thermocouple; 4 – alundum case; 5 – alundum pipe; 6 – millivoltmeter; 7 – rubber hose; 8 – rheometer; 9 – gasometer)

896

Proceedings of Copper 2010

Processing of High-Silicon Copper Sulfide Concentrates by Vanyukov Smelting The dust was leached by stirring in temperature-controlled cell. After leaching the solid and liquid phases were separated by vacuum filtration with the cake flushed with water acidated to рН=1. Cake and filtrate were analyzed for Re, Pb, Cd, Zn, Cu, S. Grades of these elements in the middlings depending on concentration were determined using known techniques. The phase composition of the products of trial smelting as well as elements in dust and cakes were determined using chemical and X-ray phase techniques. The X-ray patterns were produced using Dron-1 device with cobalt anode. Crystal optic analysis was done using MIN-8 microscope while optic coefficients were measured using standard set of immersion liquids.

3

Results and discussion

3.1 Combined processing of Zhezkazgan and Eastern Kazakhstan concentrates In order to simulate chemical and heat transformations during autogenous smelting the feed parameters were calculated based on high-copper Zhezkazgan concentrate an high-sulfur East Kazakhstan concentrate produced at Copper Chemical Plant, Eastern Kazakhstan (Table 1). Extensive experience in operating Vanyukov furnaces allowed determination of the main requirements to the feed composition shown in Table 2. Besides, Table 2 shows the proportion of the concentrates in the mixture as 66.85 % of Eastern Kazakhstan concentrate and 33.15 % of Zhezkazgan concentrate that is the closest one to the required composition as well as deviations of main components. Table 1:

Chemical composition of concentrates

Concentrate

Cu [%]

Fe [%]

S [%]

SiO2 [%]

CaO [%]

Pb [%]

Zn [%]

Zhezkazgan

37.50

6.65

14.50

24.20

0.60

1.47

1.20

East Kazakhstan

15.61

24.23

39.79

7.20

0.19

2.00

2.00

Table 2:

Chemical composition of concentrate mixture (66.85 % of East Kazakhstan concentrate and 33.15 % of Zhezkazgan concentrate)

Concentrate mixture Required composition Deviation

Cu [%]

Fe [%]

S [%]

SiO2 [%]

CaO [%]

Pb [%]

Zn [%]

23.21

18.12

31.0

13.11

0.33

1.82

1.72

25

23

31

15

2

2

2

2.14

4.60

0.40

2.16

1.67

0.18

0.27

The deviations of estimated concentrate mixture from the required feed composition in terms of main components are shown in Figure 3 below. Based on the above calculations the feed consisting of 67 % of Zhezkazgan concentrate and 33 % of Eastern Kazakhstan concentrate with no flux added shall be considered satisfactory since it is in the Proceedings of Copper 2010

897

Kozhakhmetov, Kvyatkovskiy, Ospanov, Abisheva, Zagorodnyaya best compliance with the main components grade requirements. Thus, the feed composed from these concentrates virtually will not require flux and may allow to achieve higher grade matte and acceptable slag composition that will ensure reasonably complete separation of matte and slag and good recovery of copper and precious metals into the matte. In order to confirm this, the process parameters of smelting in Vanyukov furnace were calculated and laboratory tests were done with the products assayed.

Figure 3:

Total deviation of concentrate mixture from required composition in terms of absolute value

For the purpose of modeling of chemical composition of the smelting products and average melt temperature a number of smelting tests were done and matte and slag composition was determined at different values of specific oxygen flow in the gas mixture per mass unit of the feed. In addition, the material balance of smelting in Vanyukov furnace given the throughput capacity of 100 metric tons of concentrate per hour was calculated. The results of the trials as well as material balance calculations are given in Tables 3 and 4. Smelting temperature was 1300 °С. Approximate material balance for such smelting is given in Table 4. Table 3:

Matte and slag composition at 58 % Cu matte Cu Fe S O Zn Matte [%] 58.08 14.64 22.55 0.90 1.13 Slag [%] 0.58 29.91 0.44 9.19 2.44

898

Pb 2.38 0.59

SiO2 0.23 31.75

CaO − 0.80

Others − 24.16

Proceedings of Copper 2010

Processing of High-Silicon Copper Sulfide Concentrates by Vanyukov Smelting Table 4:

Material balance for 58 % Cu matte process FEED [t/h] Concentrate 100.000 Coal 0.500 Oxygen 23.454 Air 2.575 Total: 126.529 OUTCOME: Matte 36.619 Slag 38.162 Gas 50.529 Dust 1.219 Total: 126.529

‘000 [m3/h]

16.500 2.00

24.373

The calculations of smelting at 55 % Cu matte are given in Tables 5 and 6 below. Smelting temperature is 1300 °С. Table 5:

Matte and slag composition at 55 % Cu matte Cu Fe S O Zn Pb Matte [%] 55.09 16.68 22.79 1.03 1.42 2.64 Slag [%] 0.55 28.20 0.51 8.60 2.42 0.53

As 0.09 0.18

Material balance for 55 % Cu matte process FEED [t/h] Concentrate 100.000 Coal 0.500 Oxygen 22.459 Air 2.575 Total: 125.534 OUTCOME: Matte 38.645 Slag 36,639 Gas 49.031 Dust 1.219 Total: 125.534

CaO − 0.83

SiO2 0.26 33.02

Others − 25.17

Table 6:

‘000 [m3/h]

15.800 2.000

23.896

Matte and slag composition for 62 % Cu matte process is given in Tables 7 and 8. Smelting temperature is 1330 °С. Table 7:

Matte and slag composition at 62 % Cu matte Cu Fe S O Zn Matte [%] 62.01 11.97 22.24 0.72 0.79 Slag [%] 0.62 31.78 0.36 9.82 2.36

Proceedings of Copper 2010

Pb 2.00 0.67

SiO2 0.19 30.42

CaO − 0.77

Others − 23.11

899

Kozhakhmetov, Kvyatkovskiy, Ospanov, Abisheva, Zagorodnyaya

Table 8:

Material balance for 62 % Cu matte process FEED [t/h] Concentrate 100.000 Coal 0.500 Oxygen 24.690 Air 2.575 Total: 127.765 OUTCOME: Matte 34.258 Slag 39.891 Gas 52.397 Dust 1.219 Total: 127.765

‘000 [m3/h]

17.370 2.000

24.959

The results of the trial smelting and calculations confirmed possibility of smelting of the feed composed of two concentrates mentioned above in Vanyukov furnace with no flux added. The heat balance in such case will allow producing 55 % to 62 % Cu matte while keeping the slag composition satisfactory.

3.2 Copper smelter dust treatment Based on the elemental and material composition of the dust as well as on the chemical properties of lead, zinc, copper, rhenium and cadmium compounds the ways of breaking down the dusts and recovery of the above elements from solutions were outlined. Main operations are as follows: • Dust leaching with soda solution; • Carbonate cake leaching with nitric acid solution; • Precipitation of lead sulfate by mixing carbonation and leaching solutions and producing tribasic lead sulfate (TBLS); • Selective extraction of rhenium by trialkylamine (ТАА) from nitrate-sulfate master solutions; • Extraction of zinc by di-2-ethylgexylphosphoric acid; • Precipitation of cadmium and copper carbonates mixture from raffinates by soda. Each process was laboratory assayed. The dust tested had the following composition (mass %): 39.3 Pb; 6.5 Zn; 5.5 Cu; 10.8 Stotal; 1.05 As; 0.6 SiO2; 0.8 Cd; 0.04 Bi; 0.02 Ag; 0.03 Re. According to the chemical and X-ray phase analyses the lead, zinc and copper are mainly represented by sulfates as well as, to different extent, by oxides, sulfide and silicates (Table 9). About 90 % of sulfur (based on chemical analysis) is in sulfates while the remaining ~10 % in sulfides.

900

Proceedings of Copper 2010

Processing of High-Silicon Copper Sulfide Concentrates by Vanyukov Smelting

Table 9: Phase composition of lead, zinc, copper and sulfur in the dust Grades [%] Elements Pb Zn Cu Cd Sulfates 81.43 81.54 58.18 82.72 Oxides 6.10 6.20 27.70 5.43 Sulfides 5.85 10.77 12.73 11.45 Metal 6.36 1.82 Silicates 1.54 0.52

S 89.6 10.4

3.2.1 Dust leaching by soda solutions The effect of soda consumption (100-240 % from theoretically required, taking into account content of Pb, Zn, Cu and Cd sulfates), contact time (30-120 min), solid-to-liquid ratio (1:2-6) and temperature (20-70 °С) on behaviour of lead, zinc, copper, cadmium and rhenium was investigated at carbonation stage. The quantity of soda theoretically required to convert metal sulfates into carbonates was calculated based on the following reaction: MeSO4+ Na2CO3 = MeCO3 +Na2SO4.

(1)

In order to examine the effect of soda consumption on carbonation process a series of solutions was prepared containing 74, 90, 106, 134, 159, 164 and 180 g/l of soda. Test conditions were: solid-toliquid ratio=1:4, temperature=50 °С, duration=1 hour. The degree of carbonation of lead, copper, zinc and cadmium compounds was assessed based on content of carbonates of the above elements in the cake as well as sulfur content in the solution. Based on the experiments and taking into account conversion of metal sulfates into carbonates and residual soda content in filtrates the optimum consumption was determine to be 140 %. Increase in solid-to-liquid ratio, as returned by the experiments, affects forming of zinc, cadmium, copper and, especially, lead carbonates and recovery of rhenium into the solution in different extent. Maximum recovery rates were achieved at solid-to-liquid ratio of 1:4. The duration study was based on time intervals of 20, 40, 60, 90 and 120 minutes. The Zn and Cd carbonates are mainly (about 75 % each) formed during the first 20 minutes of contact between dust an soda solution while the same time for copper and lead is 1 hour. Besides, during one hour more than 70 % of rhenium was extracted into the solution. The temperature varied from 20 to 70 °С in 10 °С intervals (solid-to-liquid ratio 1:4, soda consumption=140 %, duration=1 hour). Temperature increase influenced positively the conversion of base metals sulfates into carbonates and, especially, extraction of rhenium into the solution. At 60 °С the sulfates converted into carbonates almost completely (80 % of Pb, Cd and Zn; 53.2% Cu) with 83 % of rhenium extracted into the solution. Based on the experiments the optimum conditions for conversion of lead, copper, cadmium and zinc sulfates into carbonates were determined being as follows: soda consumption=140 % (taking into Proceedings of Copper 2010

901

Kozhakhmetov, Kvyatkovskiy, Ospanov, Abisheva, Zagorodnyaya account content of the metals in form of sulfates), solid-to-liquid ratio=1:4, contact time=60 minutes, temperature=60 °С.

3.2.2 Carbonate cake leaching by nitric acid solutions The nitric acid was selected as leaching reagent due to a number of reasons. First, the lead nitrate is the most soluble among common inorganic lead salt [Pb(NO3)2 – 52.2 g vs. PbCl2 – 0.67 g in 100 g of water]. Secondly, nitric acid being a strong oxidant reacts with Re, Zn, Cd and Cu sulfides and the lowest rhenium oxides (IV, VI) forming water-soluble compounds. Thus, use of nitric acid is reasonable in view of increase in metals recovery into the solution and, consequently, reduction of the volume of solution during carbonate cake leaching. The optimum conditions for carbonate cake leaching by nitric acid solutions were determined to be as follows: nitric acid concentration=4 mol/l, solid-to-liquid ratio=1:4, contact time=30 minutes, temperature=50 °С. At these conditions 97 % of Zn, 82 % Pb, 93 % Cu and almost all remaining rhenium (20 %) were extracted into the solution.

3.2.3 Lead sulfate production Lead sulfate was produced by way of mixing of the solutions from dust carbonation and carbonate cake leaching processes. The effect of рН (4 to 0.8) of the solutions mixture adjusted by adding of sulfuric acid solution, temperature (20-70 °С) and stirring time (20-60 minutes) on the degree of lead precipitation and salt composition was investigated. The experiments showed almost complete (99.99 %) lead precipitation despite variations of the above factors. According to X-ray phase analysis the residues are represented only by lead sulfate. Lead sulfate can be considered as the feed for production of tribasic lead or other metal sulfate.

3.2.4 Production of ammonium perrhenate After lead sulfate precipitation the solutions contain, in average, 34 g/l of nitrate ions, 48 g/l sulfate ions, 8 g/l Zn, 12 g/l Cu, 40 mg/l Re and 93 mg/l Cd. In order to recover rhenium from these solutions solvent extraction was used. The extraction was done by 10 % TAA kerosene solution with higher alcohols added. This extractant is widely used for rhenium extraction from acidic sulfate and nitrate solutions [13, 14]. The effect of organic-to-water ratio (1:5-30) and contact time (1-20 minutes) on the rhenium extraction was investigated. Using the optimum conditions determined (organic-to-water ratio=1:20, contact time and phase sedimentation – 5 minutes) the 4-stage counterflow extraction process was simulated. The rhenium recovery rate at extraction stage made 99 %. The rhenium was further re-extracted from the extract containing 790 mg/l Re by 5 mol/l ammonia water solution at organic-to-water ratio of 15:1. Ammoniac Re-containing re-extracts were evaporated to Re grade of 55 g/l followed by cooling down to 5 °С. Rough ammonium perrhenate precipitated in such conditions. The rhenium recovery at re-extraction

902

Proceedings of Copper 2010

Processing of High-Silicon Copper Sulfide Concentrates by Vanyukov Smelting stage made 99 % while rhenium recovery from the solution into rough salt was 96 %. Finally, the ammonium perrhenate is produced from this salt using recrystallization [13] or electrodialysis [15] process.

3.2.5 Precipitation of carbonates The raffinates produced from the rhenium extraction contain 20 g/l nitrate ions, 38 g/l sulfate ions, 8 g/l Zn, 12 g/l Cu, 0.5 mg/l Re and 93 mg/l Cd. These raffinates were neutralized by soda solution in order to produce carbonate cake which was further added to the copper smelter feed. Based on the obtained information the procedure sheet for complex dust processing was developed (Figure 4).

Figure 4:

The procedure sheet of complex processing of lead dust from copper smelter

Proceedings of Copper 2010

903

Kozhakhmetov, Kvyatkovskiy, Ospanov, Abisheva, Zagorodnyaya

4

Conclusion

1. The mixture of high Si and high S concentrates can be processed by autogenous smelting with no flux added; 2. Based on the experiments the recommendations were made to Kazahmys to feed high Cu (up to 37 %) concentrates to Vanyukov furnace that will allow increase in Cu grade in matte and reduction of slag per 1 metric ton of copper. 3. The dusts from copper smelter can be processed by consecutive leaching by soda and nitric acid water solutions and metals extraction from the solutions producing lead sulfate and tribasic lead sulfate (TBLS), ammonium perrhenate, zinc and copper-cadmium cake. 4. The combination of two dust treatment techniques at determined optimum conditions ensures extraction of 98 % Re, 82 % Pb, 92 % Cu, 96 % Zn and 96 % Cd into the solution. The rhenium is mainly extracted into the solution at carbonation stage (∼80 %) while other metals are extracted during nitric acid leaching. 5. Based on the experiments the flowsheet of dust processing was proposed to produce saleable product, namely lead, rhenium salts and base metals carbonates.

References [1]

TARASOV, A.V. & PARETSKY, V.M.: Development of Autogenous Copper Smelting Processes in Russia and CIS Countries: Cobre 2003 the 5th International conference, Santiago, Chile, 2003.

[2]

BYSTROV, V. P., KOMKOV, A.A., FYODOROV, A.N. & LADYGO, E.A.: Use of Vanyukov process and furnace for comprehensive treatment of non-ferrous metal slag, various wastes and intermediate products: Recycling and Waste Treatment in Mineral and Metal processing: Technical and Economic Aspects, proceedings of TMS Fall 2002 Extraction and Processing Division Meeting Luleå, Sweden, 2002, vol.2, pp. 445-456.

[3]

RAMACHANDRAN, V., DIAZ, C., ELTRINGHAM, T., JIANG, C.Y., LEHNER, M., MACKEY, NEWMAN, C.J. & TARASOV A.V.: Primari Copper Production- A Survey of Operating World Copper Smelters: Cobre 2003 the 5 the international conference, 2003, Santiago, Chile.

[4]

DIAZ, C. & UTIGARD, M.A. Copper Smelting in the Americas – 1995-2003 Changes in a Diverse Technological landscape: Cobre 2003 the 5 the international conference, 2003, Santiago, Chile.

[5]

NIKOLIC, S., HAYES, P.C. & JAK, E. Phase equilibria in ferrous calcium silicate slags. Part IV: Liquidus temperatures and solubility of copper in Cu2O-FeO-Fe2O3-CaO-SiO2 slags at 1250 °C and 1300 °C at an oxygen partial pressure of 10-6 atm.: Pyrometallurgical Research Centre. The School of Engineering, The University of Queensland, Brisbane, Australia, 2008.

904

Proceedings of Copper 2010

Processing of High-Silicon Copper Sulfide Concentrates by Vanyukov Smelting [6]

KARELOV, S.V., MAMYACHENKOV, S.V., NABOICHENKO, S.S., YAKORNOV, S.A. & USOV С.P.: Comprehensive processing of nonferrous smelter zinc- and lead-containing dusts, Institute of Non-Ferrous Metals Economy and Information Moscow, 1996. (in Russian).

[7]

KE, JIA-JAN, QIU, RUI-YUN & CHEN, CHIA-YUNG: Recovery of metal values from copper smelter flue dust, Hydrometallurgy, 1984, 12, pp. 217-224.

[8]

TER-ARAKELYAN, K.A., AVAKYAN, G.S. BAGDASARYAN, K.A.: Comprehensive use of copper production fine converter dusts, Izv. Vuzov. Nonferrous Metallurgy, 1991, № 4. pp. 52-56. (in Russian).

[9]

ANABLE, W.E., PAIGE, J.I. & PAULSON, D.L.: Copper recovery from primary smelter dust, U.S. Bur. Mines Rep. Invest. 1981. № 8659.

[10] MOHRI, E. & YAMADA, M.: Recovery of metals from the dusts of flash smelting furnace. World Mining and Metals Technology, American Institute of Mining, Mtallurgical, and Petroleum Engineers, New York, 1976, pp, 520 – 533. [11] BEN’EASH, E. YA., GETSKIN, L.S., & FISHMAN, M.A.: Production of chemical compounds from wastes and by-products of non-ferrous metallurgy, Proceeding of East Institute of Nonferrous Metals, Publishing house Metallurgy, Moscow, 1962. pp. 45-50. (in Russian) [12] KOKUSHEVA, А.А., DAIRABAEVA, G.А., USABEKOVA, А.SH., PERFILJEV, N.А.: Recovering rhenium from Dzhezkazgan copper smelter sulfuric acid slimes, Nonferrous Metals, 1992. № 5, pp. 14-15. (in Russian). [13] ABISHEVA, Z.S., ZAGORODNYAYA, A.N.: Hydrometallurgy in rare metal production technology in Kazakhstan, Hydrometallurgy, 2002, 63, pp.55 - 63. [14] KUNAEV, А.M., NEREZOV, V.M., DADABAEV, A. YU.: New hydrometallurgical processes for the production of molybdenum, tungsten, and rhenium, Publishing house Science, Alma-Ata, Алма-Ата: Наука. 1985, 149 p. (in Russian). [15] AGAPOVA, L.YA., PONOMAREVA, E.I., ABISHEVA, Z.S.: Production of concentrated rhenium acid by electrodialysis of rhenium salts solutions, Hydrometallurgy 2000, 60, pp. 117-122.

Proceedings of Copper 2010

905

906

Proceedings of Copper 2010

Boiler Tube Cooling of TSL-Furnace Walls Heikki Lankinen, Rauno Peippo Foster Wheeler Energia Oy Relanderinkatu 2 FIN-78200 Varkaus, Finland

Keywords: Top Submerged Lance (TSL) furnace, cooling, boiler tube, ISASMELTTM

Abstract New pyrometallurgical processes have emerged and gained popularity. Especially the number of new Top Submerged Lance (TSL) - process lines (Ausmelt, Isasmelt™) has increased, for treatment of primary and secondary feed materials. Increased unit sizes and the aim to extend campaign life have raised the need to improve furnace wall cooling and an increasing number of furnaces are built with water cooled walls from bottom to top. Foster Wheeler has built the upper furnace freeboard sidewalls and furnace roof (“Furnace Hood”) for several TSL furnaces using membrane boiler tube wall design. In some cases only the TSL-furnace roof has been built of boiler tube wall. Closed high pressure cooling water circuit has also been used, not only evaporating boiler water. The furnace process sets requirements and challenges for tube wall furnace hood layout. Hood geometry, design and supports must consider splashing and accretions as well as static and dynamic loads from process equipment, operation and maintenance. Smooth transition from round furnace vessel to rectangular WHB uptake is required. Ports and openings for lance, feed, burner, sampling, maintenance hatch etc. give complex design and cause challenges to tube panel manufacturing. The experience from the operating furnaces with Foster Wheeler boiler tube hood, uptake and downcomer have indicated the better than expected availability and decreased maintenance needs, along with the extended campaign life. Eight completed or ongoing Foster Wheeler deliveries to TSL furnaces in the past decade have shown the boiler tube cooling of TSL furnace walls and/or roof being proven technology and feasible option to be considered in the quest to improve plant operation and investment payback.

Proceedings of Copper 2010

907

Lankinen, Peippo

1

Introduction

Foster Wheeler Energia Oy is an independent Finnish subsidiary of Foster Wheeler A.G. (formerly Foster Wheeler Ltd). Together it’s predecessor A.Ahlstrom Corporation it has been a pathfinder and one of the leading suppliers of Waste Heat Boilers (WHB) for non-ferrous metallurgical industry for 5 decades. Since the early 1950’s over 100 WHB units have been supplied. WHB’s are tailored to meet client and process specific needs. Experience of Foster Wheeler (FW) has accumulated from boiler deliveries to processes like Flash Smelting, TSL (Ausmelt, Isasmelt™), Mitsubishi Process, Kivcet, Kaldo, Zinc or Pyrite Roasters, and other processes. FW has also carried out several succesful modernization projects to improve the performance of its older generation WHB’s but also WHB’s delivered originally by others. Gas cooling and handling equipment downstream the furnace form a significant portion of the plant investment and operation cost. The WHB has an important task in the gas train. On the other hand the reliability of furnace cooling plays a major role in the smelting line productivity and campaign life. WHB technology can be extended to furnace side cooling. This paper address the development and challenges of the quest for improving furnace availability and decreasing maintenance needs by using boiler water /steam cooling.

2

Background and historical overview

New pyrometallurgical processes have emerged and gained popularity in the non-ferrous industry. Especially the number of new TSL process lines (Ausmelt, Isasmelt™) has increased for treatment of primary and secondary feed materials. In the early TSL projects the furnace shell was often made of brick and uncooled steel shell. However, the need for increased unit sizes and longer campaign life have raised the need to adopt cooling. Initially the cooled furnace area was limited to the lower melt region of the furnace walls, but an increasing number of furnaces are built with cooled walls from bottom to top. Waste heat boiler circulation water has been previously used for cooling in various types of furnaces so it was a natural step to introduce similar cooling principle to the new type TSL furnaces also. Foster Wheeler has built the upper furnace freeboard side walls and roof (“Hood”) for several TSL furnaces using membrane boiler tube wall design. In some cases only the furnace roof has been built of boiler tube wall. High pressure hot water circuit has been used as cooling medium, in addition to evaporating boiler water. Small foot print in layout is one of the drivers for TSL-process selection and the WHB development has followed the same layout criteria for Hood, Uptake & Downcomer design. By using forced circulation, a robust structure with flexible geometry and compact layout has been achieved.

908

Proceedings of Copper 2010

Boiler Tube Cooling of TSL-Furnace Walls Design of vertical type Foster Wheeler WHB had been developed in deliveries, such as Samim, Portovesme, Italy (Kivcet,1984) and Cyprus Miami, Arizona (1991, Isasmelt™) (see Figure 1).

Figure 1:

Samim, Portovesme Italy, Kivcet 1984 (left). Cyprus Miami, Arizona, Isasmelt™ 1991.

However, the original supply of the early WHB’s excluded the cooled furnace walls. These projects gave solid ground for introduction of boiler water cooled furnace roof and freeboard walls. This development was first fully utilized in the WHB’s for two converting furnaces ( Ausmelt process ) in Anglo Platinum‘s ACP-Project at Rustenburg , South Africa (Stage A: 2001 and Stage B: 2004).

3

Design Features

3.1 Process specific requirements The development of water cooled hood had first to consider the requirements from process point of view together with boiler specific requirements. From process point of view Hood geometry plays a significant role. Splash and accretion handling require smooth geometry; the valley angle (corner) of the slope section must be steep enough to allow molten splash flowing back to furnace. The geometry of the tube wall hood must also provide streamlined transition from round furnace vessel to rectangular WHB uptake (Figure 2).

Proceedings of Copper 2010

909

Lankinen, Peippo

Figure 2:

Example of Hood geometry.

Required ports and openings for lance, feed, burner, sampling, maintenance hatch etc. have to be accommodated into the hood roof design. In certain cases the whole roof panel is removable for furnace maintenance e.g. re-bricking or refractory repair. The hood cooling circuit must be designed to accommodate very wide range of possible heat flux in different processes or process stages. Typical average level may be from 20 kW/m2 up to 80 kW/m2 throughout the hood, but may locally have essentially higher spot heat flux figures. Due to possibility of refractory peeling off in some spots, the tube cooling and water circuit velocity may require design for peak heat loads of up to 400 kW/m2. Anchors / studs must be attached to hood walls & roof for holding the refractory lining. Furnace hood structures must be designed to stand vibrations and dynamic loads coming from furnace and lance operation, as well as the static loads of refractory lining and possible accretions on the roof, walls and especially on the slope transition section. Hood support structure shall be designed so that the thermal expansion can take place.

3.2 Boiler design aspect From boiler point of view one very significant criterion is to guarantee sufficient cooling water circulation in the limits set by the hood geometry. Boiler tubing design must consider differences of e.g. round bottom ring and sloped transition section compared to conventional boiler tube sections. The hood refractory life is extended by cooled wall tubes, yet the fixing method of refractory anchor 910

Proceedings of Copper 2010

Boiler Tube Cooling of TSL-Furnace Walls design should limit excessive heat transfer via anchor to boiler wall to minimize hot spots and stress on boiler tubes. As an alternative for conventional steam generating boiler, closed hot water circulation can be utilized in cases where heat recovery is not an issue or where hot water is better applicable than steam. Heat extraction/heat transfer takes place by means of air cooler (fin fan cooler) in case of dumping the heat or heat exchanger (plate/tube exchanger) in case of heat utilization for process purposes or district heating. An important aspect in WHB design is the temperature along the gas stream and consequently minimizing material and dust sticking problems but still eliminating acid dew point on cooled walls. This is achieved by designing the WHB operating temperature window above acid dew point but below material temperature limitations or dust sticking range. This is valid also in furnace hood walls to certain degree. It is important that acid condensation cannot be formed behind refractory lining if cracks or porosity allow gas to diffuse into contact with the cooled wall. In steam generating boiler the steam pressure set point is automatically keeping the saturation temperature. In closed hot water circulation a safety margin between operating water temperature and saturation point must be maintained and this requires higher operating pressure for similar boiler water temperature (Figure 3).

Figure 3:

Schematic T-s diagram: Steam boiler a), closed hot water circulation b).

Roof ports for lance, burner and feed are often equipped with collars made of boiler tubing. These tubes have similar forced circulation cooling as the hood itself being however made of invidual cirProceedings of Copper 2010

911

Lankinen, Peippo cuits. This enables fast continuity of operation even if the lance port collar is damaged by e.g. manual cleaning of accretions around lance. In addition to lance, burner and feed port arrangement to roof layout, a maintenance hatch for furnace re-bricking purposes has been required. This hatch is part of the roof cooling circuits and its size must exceed the standard brick pallet. This makes the roof design even more challenging. (Figure 4).

Figure 4:

Simplified example of typical roof layout arrangement with ports and maintenance hatch.

3.3 Customization Customized WHB design and a flexible product range is a strength in a demanding market. Furnace hood development is one good example of such continuing product development. Parallel to the development of the hood for the specific requirements of TSL technology there are other related development areas which have emerged. WHB uptake design needs to be considered together with the transition area and the connecting joint design. An omega tube panel section with smooth gas side surface has been successfully used at the lower part of uptake for decreasing splash and sticky particles adhering to lower uptake walls (Figure 5).

912

Proceedings of Copper 2010

Boiler Tube Cooling of TSL-Furnace Walls

Figure 5:

Lower uptake omega tube panel section. Lifting of prefabricated section at site (left). Sample of inner surface (middle). Samples of tube panel sections (right).

Uptake layout must also consider the space requirement for lance handling equipment located above the hood roof. The connection of the TSL furnace hood to the WHB uptake has numerous design challenges, like splash, accretion risk, thermal expansions, access, shut-off gate arrangements, etc. Specific attention must also be paid to hood thermal expansion and sufficient support even in the worst accretion scenario. Depending on the case the hood can be supported on furnace top flange when the load is considered in furnace foundations or alternatively from the building steel structure. The uptake and downcomer are designed to hang from building steel and allow free thermal expansion downwards. The intersection between the furnace hood top flange and uptake bottom has to be sealed but also access for maintenance is required. Several designs e.g. gate slot covers have been used for this area. Design typically includes the possibility to insert a damper gate. Gate slot cover types vary from simple refractory lined uncooled design to cold water cooled and further to high pressure boiler water cooled panels. Specific attention shall be paid to the gas tightness of the covers and minimizing ingress air. Damper gate can be cold water cooled with by-pass vent for burner gas or plain uncooled plate, depending on whether it is to be used on hot or only cooled-down furnace (Figure 6).

Proceedings of Copper 2010

913

Lankinen, Peippo

Figure 6:

Example of gate slot covers and damper gate arrangement (left). Detail of corner area.

3.4 From design to fabrication TSL furnace being round shaped, typically 3-5 meters I.D. and uptake shaft cross section being rectangular approx. 1.5-3 m x 3-5 m, the design and fabrication of both round and transition sections become challenging. With round/polygon bottom ring automated tube panel welding is partially possible, but the slope transition section is typically made of individually bent tubes and all fins between tubes are manually welded. This is time consuming and sets high quality control requirements for both dimensional and weld quality. Uninterrupted hood operation being essential, further site installation must follow the same strict quality criteria (Figure 7).

Figure 7: 914

Hood slope section under fabrication. Proceedings of Copper 2010

Boiler Tube Cooling of TSL-Furnace Walls Refractory anchors are considered partially sacrificial i.e. they will wear/burn down. In the anchor fixing design, the limitation of heat transfer in order to avoid hot spots to tube surface has to be balanced with the requirement of cooling the anchors. V-type anchor have been successfully used instead of simple studding regardless of the larger extent of manual welding (Figure 8).

Figure 8:

Principle arrangement of tube panel with V-anchor.

The hood slope section protruding outside the hood centre of gravity together with worst accretion scenario over the slope section dominates the structural design. Solutions, such as sliding plates with small friction factor and pre-tensioned horizontal hangers have been used for balancing the hood in all design conditions. Complete hood sections are oversized for standard overseas transportation methods so the shipping block sizes shall be designed considering both cost effective transportation and smooth assembly at site. This requires specific shop pre-assembly arrangements before shipments. The hood dimensional check during pre-assembly is a challenging task for workshop personnel. Both safety issues and rigidity of this temporarily built even 10 m high structure are demanding. Furhermore the verified & recorded dimensions and match marked interfaces must be maintained during dismantling, loading and overseas transportation. The same internal jigs used as transportation supports can be used for site pre-assembly on ground level and for further lifting into the furnace building (Figure 9).

Proceedings of Copper 2010

915

Lankinen, Peippo

Figure 9:

4

Hood pre-assembly arrangement at shop (left). Hood front section at site; pre-assembly on ground level (top right) and ready for lifting.

Future development

TSL furnace boiler tube walls without refractory lining are in limited use but the design suffers from excess splash/dust build-up. If furnace heat balance allows, possible development trend is furnace hood built of polygon shape omega-tube wall without any refractory lining. This arrangement will have the advantages of smooth surface needing no refractory layer/anchoring and allowing Spring Hammers to be used for accretions removal (Figure 10). Simpler and more reliable pneumatic Spring Hammer rapping system is under development.

916

Proceedings of Copper 2010

Boiler Tube Cooling of TSL-Furnace Walls

Figure 10: Example of possible future arrangement of polygon shaped hood.

5

Cost profile aspects

Compared to alternatives such as copper cooling elements or uncooled construction in general, boiler tube cooling is a cost effective construction having also longer service life than uncooled elements. Considering the favourable effect on operation & maintenance, the influence of boiler tube cooling of TSL furnace walls on overall plant investment payback time should not be ignored.

6

Conclusion

The experience from the operating furnaces with Foster Wheeler boiler tube hood, uptake and downcomer have indicated the better than expected availability and decreased maintenance needs, along with the extended campaign life. Eight completed or ongoing Foster Wheeler deliveries to TSL furnaces in the past decade have shown the boiler tube cooling of TSL furnace walls and/or roof being proven technology and a feasible option to be considered in the quest to improve plant operation.

Proceedings of Copper 2010

917

Lankinen, Peippo

References [1] PEIPPO R., HOLOPAINEN H., NOKELAINEN J.: “Copper smelter waste heat boiler technology for the next millennium“, Proceedings of Copper 99-Cobre 99 International Conference, Volume V – Smelting Operations and Advances. The Minerals, Metals & Materials Society, 1999, p71 - 82 [2] W.G. DAVENPORT, M. KING, M. SCHLESINGER, A.K. BISWAS: Extractive Metallurgy of Copper, fourth edition, Elsevier Science Ltd, Oxford, UK 2002 , chapter 8 “Ausmelt/Isasmelt Matte Smelting“ p119-129 [3] SOFRA J., MATUSEWICZ R.: Ausmelt technology – copper production technology for the 21st century. Proceedings of Copper 2003 – Cobre 2003, Volume IV (Book 1), Pyrometallurgy of Copper, Santiago, Chile 2003. Edited by C. Diaz, J. Kapusta, C. Newman 2003, p157-172 [4] ARTHUR P.S.: Isasmelt™ - “6,000,000 TPA and rising“, Sohn International Symposium, Advanced Processing of Metals and Materials, Volume 8, International Symposium of Sulfide Smelting. The Minerals, Metals & Materials Society 2006, p275-290 [5] MATUSEWICZ R., HUGHES S., HOANG J.: The Ausmelt Continuous Converting (C3) Process, Cu2007, Volume III (Book 2), The Carlos Diaz Symposium of Pyrometallurgy. Edited by A.E.M Warner, C.J. Newman, A.Vahed, D.B. George, P.J. Mackey, A. Warczok, MetSoc 2007, p29-47

918

Proceedings of Copper 2010

Experimental Estimation of the Residence Time Distribution in a P-S Converter C. López, A. Almaraz, R. Cuenca, G. Plascencia CIITEC – IPN Cerrada Cecati s/n México, D.F. C.P. 02250 México

B. Hernández, F. Reyes Facultad de Química, Ed. D, UNAM Depto. de Ing. Química Metalúrgica Circuito de la Investigación Científica s/n México, D.F. C.P. 04510 México

Keywords: P-S converter, residence time distribution, experimental estimation, modelling

Abstract The aim of this research is to evaluate the flow patterns within the converter in order to find out how efficiently the reactor is used during blowing operations. To do so, a set of C curves were constructed by injecting an acidic solution into a Peirce-Smith converter plexiglass model. The C curves were constructed after recording with the aid of a data acquisition system (connected to a personal computer) the acid concentration in at least 18 different locations throughout the converter model and by varying the ration of height between the injection point and the water level. From these curves; it was possible to estimate the residence time distribution within such reactor. The evaluation of the C curves obtained was also compared with pictures and video taken during the injection (in an independent set of experiments) of a colour tracer into the plexiglass model. Additionally, the results were compared with some numerical calculations of the blowing operation previously reported by this research group. It was found that nearly 50 % of the reactor works as mixing unit, thus rendering the total capacity of the vessel to perform more readily the converting of copper. The data collected can be used to improve the design of the tuyeres used for the injection of gases into the copper matte.

1

Introduction

Copper matte converting is commonly carried out in Peirce - Smith converters (P-S) and/or El Teniente reactor. The P-S converter is a long cylindrical vessel with an opening on the top of it for charging the molten matte and silica flux and also, it is used for the removal of off-gases. Alongside the vessel, several tuyeres are aligned so air can be injected into the molten matte. During blowing, the vessel rotates on its axis so the opening on its top lies beneath an emission capture system. After blowing and during matte charging, the vessel rotates to its starting position. Proceedings of Copper 2010

919

López, Almaraz, Cuenca, Hernández, Reyes, Plascencia One consequence of the blowing operation is the creation of back mixing zones within the converter; these zones are responsible for the fluid flow patterns within the reactor which in turn have an important effect on the overall performance of the converter. In spite of the large amount of research conducted and published on the jet/matte interaction, there is not enough information available on how does the jetting affect the residence time within the reactor. Therefore, in this paper we attempt to estimate the residence time within a P-S converter by means of a scale model.

2

Residence time distribution and the C curve

Fluid flow in a reactor may be characterized as plug or mixed [1]; however, it is often found in reactors deviations from these two ideal flows. Such deviations may be caused by fluid channelling, fluid stagnation or the creation recirculation zones within the reactor. As expected, different fluid elements in any reactor would take different paths to flow throughout the vessel; consequently each fluid element will remain in the vessel for different lengths of time. This feature will result in a Residence Time Distribution (RTD). The residence time distribution function for a fluid in a reactor is given by the C function [1, 2]. The C function represents the age distribution of a fluid leaving a reactor; thus the units of the age distribution (C curve) are time-1. It is convenient to represent the RTD in a normalized manner, thus the C curve is defines as:

∫

∞

0

C ⋅ dt = 1

(1)

And the mean residence time of the fluid in the vessel is defined by: ∞

τ = ∫ t ⋅ C ⋅ dt 0

3

(2)

Experimental

The experimental procedure for this research can be divided into two main stages, the first one deals with the scaling of the model, whereas the second stage deals with the measurement and recording of the pH in the water before during and after the injection of an acidic tracer.

3.1 Scaling of the plexiglass model The P-S converter plexiglass model shown in Figure 1, was filled with water up to different heights. Compressed air was injected at different flowrates. The gas flowrates were established by scaling

920

Proceedings of Copper 2010

Experimental Estimation of the Residence Time Distribution in a P-S Converter down those used in industrial practice [3-5]. The scaling was done by means of the modified Froude number: Fr ' =

ρ gas u2 ⋅ g ⋅ D (ρ liquid − ρ gas )

(3)

Where u is the velocity of the injecting gas (m/s), g is the acceleration due to gravity (9.81 m/s2), D is the diameter of the tuyere from which air is injected into the liquid phase (m), ρgas is the density of the injecting gas (kg/m3) and ρliquid is the density of the liquid in the vessel (kg/m3). After gathering data presented on literature, we came up with a modified Froude number of 8, which is close to the value of 13.5 used by Vaarno et al. [3-6]. Table 1:

Data used for scaling the plexiglass model [3-5].

Variable

Value

Units

Gas velocity

125

m/s

Tuyere diameter

0.05

m

g

9.81

m/s2

ρmatte

4800

kg/m3

1.2

kg/m3

ρair

Fr’ = 8 With this value of the modified Froude number, the air flow rate needed for our tests was back calculated. The value of this parameter was 12.5 L/min. To do this back calculation, a tuyere diameter of 4 mm and the density of liquid (water) were assumed. Table 2 shows our experimental flow conditions. Table 2:

Experimental flow conditions

Variable

Value

Units

8.0 Gas flow rate

12.5

L/min

25.0 10.6 Gas velocity

16.6

m/s

33.2 Height from injection point

Proceedings of Copper 2010

0.14 0.07

m

921

López, Almaraz, Cuenca, Hernández, Reyes, Plascencia

3.2 Estimation of the residence time distribution To construct the C curves for the plexiglass model, an acidic solution 1 M of sulphuric acid was prepared. This solution was injected by means of a syringe of 5 mL through an injection port. Before each injection, air was blown into the model for about 5 minutes to allow good mixing within the vessel, during this mixing period, a pH-meter was connected to a personal computer through a data aquisition system; pH of the water was measured and recorded continuously before during and after the acid injection. Gas kept blowing into the system until the pH of the resulting solution was estabilized; at this point, the measurement and recording of the pH was stopped. Once this happened, sodium hyroxide 1 M solution was added into the vessel to neutralize the pH in the plexiglass model. When the pH in the vessel was neutralized and the level of water was set to the previously established height, the blowing and pH measurement operations were repeated for each of the points selected for measurements. In an independent set of tests, a colour tracer was used to verify the findings from the acid injection tests. An organic colorant was dissolved into water and once the coloured solution was obtained, it was injected through the same injection point as the acidic solution. The amount of colorant added was of 10 mL. Pictures of the vessel during and after the colour injection occurred were taken.

922

Proceedings of Copper 2010

LOCATION OF SAMPLING POINTS

K G

F

B

A D E

Injection point

L

J I

44 cm

H

C

40 cm

Experimental Estimation of the Residence Time Distribution in a P-S Converter

Figure 1:

Experimental set up. (A) Pictures showing the plexiglass model and the measurement devices. (B) Sketch showing the measurement points for the injection tests.

Proceedings of Copper 2010

923

López, Almaraz, Cuenca, Hernández, Reyes, Plascencia

4

Results and discussion

Figure 2, shows the C curves obtained for the gas injection at different locations with a gas flow rate of 25 L/min; whereas Figures 3 and 4 show the C curves for air injection at flow rates of 12.5 and 8 L/min, respectively. The locations shown in Figures 2 to 4 correspond to those shown in Figure 1 (B).

0.024

0.028 point A point B point C

0.020

point G point I point L

0.024

C (1/s)

C (1/s)

0.020 0.016

0.012

0.016 0.012

0.008

0.008

0.004

0.004 0

20

40

60

80

100

120

0

20

40

120

0.024 point A point B point C

point G point I point L

0.020

C (1/s)

0.016

C (1/s)

100

C curves for injection tests at 25 L/min. (A) curves for sampling points in the back of the plexiglass model. (B) Curves for sampling points in front of the injection port.

0.020

0.012

0.008

0.016

0.012

0.008

0.004

0.004 0

20

40

60

Time (s)

Figure 3:

924

80

Time (s)

Time (s)

Figure 2:

60

80

100

120

0

20

40

60

80

100

120

Time (s)

C curves for injection tests at 12.5 L/min. (A) curves for sampling points in the back of the plexiglass model. (B) Curves for sampling points in front of the injection port.

Proceedings of Copper 2010

Experimental Estimation of the Residence Time Distribution in a P-S Converter

0.024

0.028 point A point B point C

0.020

point G point I point L

0.024

C (1/s)

C (1/s)

0.020 0.016

0.012

0.016 0.012

0.008

0.008

0.004

0.004 0

20

40

60

Time (s)

Figure 4:

80

100

120

0

20

40

60

80

100

120

Time (s)

C curves for injection tests at 8 L/min. (A) curves for sampling points in the back of the plexiglass model. (B) Curves for sampling points in front of the injection port.

It is clear from Figure 2 to 4 that in front of the injection point, there is a higher degree of mixing in contrast with the observations made at the back of the plexiglass model. Although the higher mixing is obviously expected near the to the injection port, it seems that the higher gas flow rate does not affect in a significant manner the degree of dispersion within the vessel. Actually, from the C curves shown, it seems that at lower air flow rates, the degree of mixing increases. Such apparent “incorrect” behaviour can be attributed to the actual degree of turbulence in the system. As the turbulence increases, the more erratic the flow becomes due to the random variation in time of the fluid velocity both in magnitude and in direction. Thus higher degree of turbulence not necessarily means better mixing or longer residence times, this can be illustrated in Figure 5 for the different measuring points. It can be seen from Figure 5 that the point with less dispersion of the mean residence time is the point G, which is located near to one of the walls of the vessel. This point shows that regardless of the gas flow rate, the fluid mean residence time at this point remained practically constant. In terms of point L, which is located in front of the injection point, it is evident that its mean residence time is longer for a gas flow rate of 25 L/min, whereas decreasing the flow rate does not seem to affect the value of the residence time.

Proceedings of Copper 2010

925

López, Almaraz, Cuenca, Hernández, Reyes, Plascencia

60.0

Mean residence time (s-1)

25 L/min

12.5 L/min

8 L/min

50.0 40.0 30.0 20.0 10.0 0.0 A

B

C

G

I

L

Measuring point Figure 5:

Mean residence time for the different measuring points at different gas flow rates.

For point I, which is geometrically symmetrical to point G, it is quite clear that as the gas flow rate decreases, the mean residence time increases. Regarding the back of the vessel; it can be seen in Figure 5 that higher flow rates result in higher mean residence times, except for location B, where the medium flow rate resulted in the longer mean residence time. Point A as seen in Figure 1; is located directly behind of point L. As expected, the mean residence times estimated at point A are longer than those for point L, but in the case of location A, the gas flow rate has a stronger effect on the mean residence time than at point L. As mentioned before, at point A, the higher the flow rate used for injection, the longer the mean residence time. But not so at location L, where the mean residence time remains fairly not affected by the change in the gas flow rate. These differences are more evident by directly comparing the C curves at both locations (A and L) with different gas flow rates. It is clear in Figure 6 that air injection at point L is not drastically affected as location A is when decreasing the gas flow rate.

926

Proceedings of Copper 2010

Experimental Estimation of the Residence Time Distribution in a P-S Converter

0.020 Point A, 25 L/min Point A, 8 L/min Point L, 25 L/min Point L, 8 L/min

C (1/s)

0.016

0.012

0.008

0.004 0

20

40

60

80

100

120

Time (s) Figure 6:

Comparison of C curves for locations A and L with gas flow rates of 25 and 8 L/min.

The results clearly show that the flow moves asymmetrically within the vessel regardless of the amount of air injected into the system. In other words, once the gas plume has developed, the liquid phase does not move following a well defined path. The liquid moves accordingly to the mean flow velocities, thus creating some recirculation zones and maybe some fluid motionless zones may develop in the plexiglass model and of course in the actual copper converter. Tests conducted with half of the water height from the injection point to the water surface showed similar behaviour. Previous results have demonstrated that within the converter, practically at the centre of the reactor, a large recirculation zone develops, resulting in an inefficient use of the reactor [7]. The recirculation zone is estimated to occupy nearly half of the volume of the reactor so it only works at half of its capacity, rendering its actual production capacity.

Proceedings of Copper 2010

927

López, Almaraz, Cuenca, Hernández, Reyes, Plascencia

Figure 7:

Computed recirculation zones within the P-S converter after Plascencia et al. [7]. (A) Blowing gas velocity = 5 m/s. (B) Blowing gas velocity = 50 m/s.

An independent set of tests in which a colour tracer was injected into the blowing stream were also conducted. Results from these trials revealed more clearly the presence of the recirculation zone at the middle of the vessel. Furthermore, the asymmetry of the flow became pretty clear upon the realization of these tests. Unfortunately, the colour tests only provide qualitative information regarding to the direction the flow takes upon gas blowing, and no local times were measured in order to estimate the accuracy of the measurements with the acidic solution experiments. In spite of this limitation, it becomes evident that the nature of the blowing operation along with the geometry of the P-S converter commands for the creation of recirculation zones within the vessel. In order to minimize such recirculation zones, it is imperative to continue research on the gas injection into the copper matte.

928

Proceedings of Copper 2010

Experimental Estimation of the Residence Time Distribution in a P-S Converter

(A)

(B)

(C) Figure 8:

5

Results from colour tracer tests. Screen shots at different times after injecting the colorant into the vessel, indicating the high recirculation zone at the centre of the vessel.

Conclusions

Residence time distribution has been measured in a scaled P-S converter. Results show that a high recirculation zone develops at the centre of the converter, rendering its overall performance. These results are supported with measurements made with colour tracers and with the use of numerical simulation of the P-S converting practice. Proceedings of Copper 2010

929

López, Almaraz, Cuenca, Hernández, Reyes, Plascencia

Acknowledgements The authors wish to thank SIP – IPN grants # 20080019 & 20090503 to conduct this research.

References [1] LEVENSPIEL O. (1999): Chemical Reaction Engineering 3rd Ed; New York (John Wiley and sons), 257 – 269. [2] BEEK W.J., MUTTZALL K.M.K., van HEUVEN J.W. (1999): Transport phenomena 2nd Ed; New York (John Wiley and sons) 133 – 139. [3] KAPUSTA J.P.T. (2004): World non-ferrous smelter survey, Part I: Copper; JOM, 21 - 27. [4] DAVENPORT W.G., KING M., SCHLESINGER M. and BISWAS A.K. (2002): Extractive Metallurgy of Copper 4th Ed; Oxford (Pergamon), 19 - 29. [5] LATHE F.E., HODNETT L. (1958): Data on copper converter practice in various contries; Transactions AIME, 603 – 617. [6] VAARNO J., PITKÄLÄ J., AHOKAINEN T. and JOKILAASKO A. (1998): Modelling gas injection of a Peirce - Smith Converter; Applied Mathematical Modelling, 907 - 920. [7] PLASCENCIA G., JARAMILLO D., López C., Barrón M.A. and González J. (2007): Computer simulation of the early stages of blown in a Peirce - Smith converter; 6th Intl. Conference Cobre/Copper 2007, Vol. III Book I, 457 – 469.

930

Proceedings of Copper 2010

Numerical Simulation of Air Blowing into a Copper Matte in a P-S Converter Using a Convergent / Divergent Nozzle C. López, A. Almaraz, I. Arellano, E. Martínez, G. Plascencia CIITEC – IPN Cerrada Cecati s/n México, D.F. C.P. 02250 México

M. A. Barrón

Depto. de Materiales, UAM-Azcapotzalco Av. San Pablo 180 México, D.F. C.P. 02200 México

T. A. Utigard Materials Science & Engineering University of Toronto 184 College St, Toronto ON, M5S 3E4, Canada

Keywords: P-S converter, convergent – divergent nozzle, copper matte, CFD

Abstract Current and conventional blowing practices in P-S converters are carried out by injecting air or oxygen enriched air into the copper matte through tuyeres. As a consequence of the injection, localized thermal gradients develop in the vicinity of the tuyere zone. In addition, both erosion and wear of the refractory lining have been reported along with clogging of the injection tubes. These operational setbacks are inherent to the blowing practice; they impact negatively on the performance of the converter and increase its operational costs. To prevent or minimize such negative effects, this paper analyzes the possibility of implementing convergent/divergent nozzles to inject gases into the matte. The analysis presented here was conducted after running CFD calculations using commercial software. The numerical simulations were conducted assuming different geometries for the convergent/divergent nozzles. In every case the hydrodynamics within the vessel showed some similarities with that found in previous calculations. The computed velocities and flow patterns were compared with those previously obtained for conventional tuyeres. It was found that there are some differences in the velocity fields as well as in the flow patterns. Recirculation of the molten material in the tuyere zone still is a problem; however these new nozzles seem to decrease the extent of the problem thus opening the door to explore in further detail the implementation of this type of devices.

Proceedings of Copper 2010

931

López, Almaraz, Arellano, Martínez, Barrón, Utigard, Plascencia

1

Introduction

Gas injection is used in many different metallurgical applications. The aim of injecting a gas into a molten bath is to enhance the rate of converting reactions by increasing the number of reaction sites. Gas injection is also used to promote the mixing of the different species into a furnace so the resulting metal should have a homogeneous chemical composition. In copper making, the converting stage is largely performed in P-S (Peirce - Smith) converters. These converters have a cylindrical shape and along one of their sides a line of tuyeres is located. Through the tuyeres, air or oxygen enriched air is injected into the molten matte so iron and sulphur are selectively removed from the matte forming blister copper as a product. In spite of its productivity, the P-S converter has some setbacks; some of them are closely related to the tuyere zone. Since oxygen is injected into the matte and the oxidation reactions are highly exothermic, temperature gradients are created in the vicinity of the tuyere zone, decreasing the service life of the refractory lining. Further, the blowing and the hydrodynamics within the converter result in further erosion and wear of the refractory lining around the tuyere zone. This problem not only impacts the service life of the converter´s lining but also increases its operating costs. Several actions have been taken to alleviate the negative effect on the lining. Mainly, there have been efforts to improve the quality of the refractories used in the P-S reactor. More recently a modification to the tuyeres has been proposed [2]. In this paper we present a study based on the numerical simulation of the hydrodynamics within the P-S converter using a convergent/divergent type of nozzle as gas injection medium to blow the matte within the P-S reactor.

Figure 1: 932

Schematics of a P-S converter, after Davenport et al. [1]. Proceedings of Copper 2010

Numerical Simulation of Air Blowing into a Copper Matte

2

Hydrodynamics of a convergent/divergent nozzle

A nozzle is a device in which a fluid flows through a continuously changing cross sectional area [2]. Generally, the cross sectional area decreases in the flow direction. The decrease in area, results in the ability of the nozzle to accelerate the flow up to the velocity of the sound (Ma=1, Ma is the Mach number). However, a fluid flowing through a convergent/divergent nozzle may decelerate in the diverging section of the nozzle rather than accelerate [2-4]. Flow conditions are determined by the pressure ratio Pb / P0, where Pb is the back pressure or pressure at the outlet and P0 is the inlet pressure. The convergent/divergent nozzle as shown in Figure 2 can be divided in three planes: Plane 1 corresponds to the flow inlet, Plane 2 designates the smallest cross sectional area of the nozzle, known as the throat, and Plane 3 designates the flow outlet. An energy balance on such type of nozzle, assuming that no work is done and also neglecting the terms of friction and potential energy, results in [3]: u ⋅ du + u s2 ⋅

dρ

ρ

=0

(1)

Where u is the mean velocity of the flow (m/s), us is the sonic velocity (m/s), defined by Ma=1 and ρ is the density of the fluid (kg/m3). On the other hand, the continuity equation applied to a duct with variable cross sectional area, has the general form [3]: d (u, ρ , A) = 0 dL

(2)

After some manipulation, equation (2) yields:

 dA du  u 2 ⋅  2 − 1 = u  us  A

(3)

And the ratio u/us defines the Mach number (Ma), then equation (3) can be expressed as:

du  1  dA = ⋅ u  Ma 2 − 1  A

(3a)

Equation (3a) implies that depending on the sonic or subsonic nature of the flow, the fluid passing through the nozzle will behave accordingly to it. When the velocity is subsonic (Ma