Copper Volume 2 PDF

Proceedings
Volume 2
Pyrometallurgy I

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

Editor
GDMB
Paul-Ernst-Strae 10, D-38678 Clausthal-Zellerfeld
Internet: www.GDMB.de
Volume 1 ISBN 978-3-940276-25-4
Volume 2 ISBN 978-3-940276-26-1
Volume 3 ISBN 978-3-940276-27-8
Volume 4 ISBN 978-3-940276-28-5
Volume 5 ISBN 978-3-940276-29-2
Volume 6 ISBN 978-3-940276-30-8
Volume 7 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: Dipl.-Ing. Jens Harre
Production and marketing: 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 ber 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

Proceedings of Copper 2010 IV
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 recy-
cling. 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 impor-
tance, 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 ex-
perience 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)
Jol 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

Proceedings of Copper 2010 V
Conference Chairman
Dipl.-Ing. Michael Kopke Aurubis AG, Germany
Technical Programme Chair
Dipl.-Ing. Jo Rogiers Aurubis AG, Belgium
Conference Chair Assistance
Dipl.-Ing. Jrgen Zuchowski GDMB Gesellschaft fr Bergbau, Metallurgie,
Rohstoff- und Umwelttechnik e. V.
Session Chairs
1 Plenary lessons of general interest Dipl.-Ing. Norbert L. Piret,
for all conference members Piret & Stolberg Partners, Germany
2 Economics Dr. Patricio Barrios, Aurubis AG, Germany
3 Downstream Fabrication, Application Dr.-Ing. Hans Achim Kuhn,
and New Products Wieland Werke AG,Germany
4 Mineral Processing Assoc. Prof. Sadan Kelebek,
Queens 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 Prof. Dr. Markus Andreas Reuter,
and Optimization Outotec Ausmelt, Australia
9 Recycling Dipl.-Ing. Jrg Wallner, Austria
10 Sustainable Development / Health, Dipl.-Ing. Miguel Palacios
Safety and Environmental Control Atlantic Copper S.A., Spain

Proceedings of Copper 2010 VI
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. Jrgen Zuchowski GDMB Gesellschaft fr Bergbau, Metallurgie, Rohstoff-
und Umwelttechnik e. V.
(Copper2010 Organizing Committee Chairman)
Mareike Hahn GDMB Gesellschaft fr Bergbau, Metallurgie, Rohstoff-
Thomas Marbach GDMB Gesellschaft fr Bergbau, Metallurgie, Rohstoff-
Dipl.-Ing. Jens Harre GDMB Informationsgesellschaft mbH
Mareike Mller GDMB Informationsgesellschaft mbH
Ulrich Waschki GDMB Informationsgesellschaft mbH

Proceedings of Copper 2010 VII
Preface
Copper Indicator of the progress of civilization
This is the motto for the 7
th
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: scien-
tists, 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 de-
velopment, 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 indi-
cator 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 coun-
tries, 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 ques-
tions. 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 conferences 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, com-
merce, 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 VIII
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

Proceedings of Copper 2010 IX
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 sec-
ondary 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 in-
dustry 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 de-
veloping 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 achiev-
able.
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) particu-
larly 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 X
Secondly, infrastructures for copper smelting have been utilized in developing the recycling busi-
ness. Exemplified by the operations at Onahama smelter, which has the worlds 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 indispensa-
ble part of national environmental policy.
Finally, progression of integration in the copper fabrication sector has strengthened the industrys
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 fu-
ture 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. Daz, 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 consump-
tion and improved SO
2
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 prac-
tice and larger converters, Peirce-Smith converting still maintains its position as the dominant tech-
nology.
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. More-
over, ER cathode output in the USA has substantially decreased.
In this paper, the author examines recent technological advances and industry changes and high-
lights 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 equi-
librium 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

Proceedings of Copper 2010 XI
near future and will make energy one of the greatest challenges of the 21
st
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 as-
pects 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, politi-
cal decisions made by governments and public administrations and social responsibility pro-
grammes 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 pro-
jected 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 de-
mand 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 ef-
fects 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 sup-
ply shortfalls continues to drive the implementation of a number of important new worldwide cop-
per projects.

Proceedings of Copper 2010 XII
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, envi-
ronmental 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 worlds leading metallurgical plant engineers and constructors.

Proceedings of Copper 2010 XIII
Table of Contents Volume 2
Pyrometallurgy I (Authors A-L)
Design of Copper-Cobalt Sulphating Roasters for Katanga Mining Limited 587
in D.R. Congo
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 631
Gogw Copper Smelter
Leszek Byszyski, Leszek Garycki, Zbigniew Gostyski, Tomasz Stodulski,
Jerzy Urbanowski
Energy Consumption in Copper Sulphide Smelting 649
P. Coursol, P. J. Mackey, C. M. Daz
Problems of Lead and Arsenic Removal from Copper Production in a 669
One-Stage Flash-Smelting Process
J. Czernecki, Z. mieszek, Z. Miczkowski, G. Krawiec, S. Gizicki
Control of Fugitive Emissions in a Continuous Mitsubishi C-Furnace 685
during Limestone Fluxing
Adam Salomon de Friedberg, Alan Hyde, Mark Coleman
Modern Flash Smelting Cooling Systems 699
K. Fagerlund, M. Lindgren, M. Jfs

Proceedings of Copper 2010 XIV
Introduction of a Slide Gate System for Copper Anode Furnaces 713
Dipl.-Ing. Klaus Gamweger
New Highly Efficient Rotary Furnace for Environmentally Friendly 721
Refining Process
Dr. Bernhard Hanusch
Sulphur Capacity of the FeO-CaO-SiO
2
Slag of Interest to the 731
Copper Smelting Process
H. M. Henao, P. C. Hayes, E. Jak
Changes in the ISASMELT
TM
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 779
at Saganoseki Smelter & Refinery after the SPI Project
Mitsumasa Hoshi, Katsuya Toda, Tatsuya Motomura, Masaharu Takahashi,
Yushiro Hirai
Development of the Continuous Copper Converting Using Two-Zone 793
Vaniukov Converter
E. Jak, L. Tsymbulov
Liquidus Temperature in Calcium Ferrite Slags in Ca
2
Fe
2
O
5
and Ca
2
SiO
4
811
Primary Phase Fields with Cu and Fixed Po
2

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

Proceedings of Copper 2010 XV
Profit Enhancement through Steam Selling 831
Kyoung-Soo Jung, Gun-Woong Byun, Sung-Ho Shin
Gas Injection Phenomena in Converters An Update on Buoyancy 839
Power and Bath Slopping
Dr. Jol 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 Kster
Processing of High-Silicon Copper Sulfide Concentrates by 893
Vanyukov Smelting
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 919
in a P-S Converter
C. Lpez, A. Almaraz, R. Cuenca, B. Hernndez, F. Reyes, G. Plascencia
Numerical Simulation of Air Blowing into a Copper Matte 931
in a P-S Converter Using a Convergent / Divergent Nozzle
C. Lpez, A. Almaraz, I. Arellano, E. Martnez, M. A. Barrn,
T. A. Utigard, G. Plascencia

Proceedings of Copper 2010 XVI
Table of Contents Volume 3
Pyrometallurgy II (Authors M-Z)
Sulphide Bath Smelting: 19
th
Century Concept and Hollways Legacy 945
P. J. Mackey, A. E. Wraith
Large Scale Copper Smelting Using Ausmelt TSL Technology at the 961
Tongling Jinchang Smelter
R. W. Matusewicz, Prof. M. A. Reuter, S. P. Hughes, Shengdao Lin, Laisheng Sun
Extending Copper Smelting and Converting Furnace Campaign Life 971
through Technology
K. McKenna, C. Newman, N. Voermann, R. Veenstra, M. King, J. Bryant
High Intensity Cast Cooling Element Design and 987
Fabrication Considerations
K. McKenna, N. Voermann, R. Veenstra, C. Newman
Management of Copper Flash Smelting Off-Gas Line Gas Flow 1003
and Oxygen Potential
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

Proceedings of Copper 2010 XVII
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 1063
Based Modelling at Olympic Dam
D. J. Ranasinghe, R. Russell, R. Muthuraman, Z. Dryga
Clyde-WorleyParsons Flash Furnace Feed System: 1079
The Development Cycle
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
Aydn Ren, Ahmet Geveci, Yavuz Ali Topkaya
Furnace Condition Assessment and Monitoring by Utilization of 1123
Innovative Non-Destructive Testing (NDT) Techniques
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 Bjrklund
The Development of the Chinese Copper Industry and 1167
Copper Extraction Technology
Yao SuPing, Wang Wei

Proceedings of Copper 2010 XVIII
Gas-Gas Cooler as Off-Gas Duct for a Slag Cleaning Furnace: 1173
Example of HSE Progress and Engineering Excellence
Michael Strder, 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 1213
on the Basis of Copper
A. Warczok, G. Riveros, R. Degel, J. Kunze, M. Kalisch, H.Oterdoom
3D-Refractory Engineering Using the Example of a 1233
MAERZ Tilting Furnace
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 1259
Oxidizing Roasting Copper Matte
I. Wilkomirsky, N. Rojas, E. Balladares
Smelting of High-Arsenic Copper Concentrates 1273
I. Wilkomirsky, R. Parra, F. Parada, E. Balladares,
Carlos Caballero, Andrs Reghezza, Jorge Ziga
Distribution of Precious Metals between Matte and Slag and 1287
Precious Metal Solubility in Slag
Katsunori Yamaguchi

Proceedings of Copper 2010 XIX
Effects of SiO
2
, Al
2
O
3
, MgO and Na
2
O on Spinel Liquidus in 1297
Calcium Ferrite Slags with Cu and Fixed Po
2

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 585

Pyrometallurgy I


Design of Copper-Cobalt Sulphating Roasters for
Katanga Mining Limited in D.R. Congo
Dr. Kamal Adham, Tuisko Buchholz, Alex Kokourine Stephane Girouard
Rayson Lu, Jim Sarvinis, Andrew Tohn
Hatch Ltd. Katanga Mining Limited
2800 Speakman Drive Avenue Industrielle 3207
Mississauga, Canada 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 Flui-
dization 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 SO
2
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
Hatchs 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 corres-
ponding feed and product handling systems for the roaster area. Design of the roaster included com-
plete thermal modeling of the roaster shell, roof, tuyere plate and associated expansion joint, gril-
lage beam, and windbox, as well as design of the feed injection nozzles, cooling water guns, piping
Adham, Bucholz, Kokourine, Lu, Sarvinis, Tohn, Girouard
and ducting, expansion joints, and discharge solid valves (seal legs). Stress analyses and load condi-
tions for several cases were examined, including start-up and upset conditions.
Design of the new roaster lines was based on the KMLs operating experience with the three exist-
ing roaster lines, two comprising 4.9 meter I.D. roasters, and one comprising a 7.3 meter I.D. roast-
er.
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 construc-
tion 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 1990s, 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 signifi-
cant 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
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 ar-
rangement 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 se-
lected. 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 in-
side 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 %, respective-
ly. 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 underf-
low is filtered and the cake sent to temporary storage. The 75 wt. % solids roaster feed slurry is pro-
duced from the stored filter cake in a modified ball mill and fed to the roaster area containing ap-
proximately 43 % copper and 4 % cobalt (dry basis). Analysis of the major sulphide components of
the feed slurry is given in Table 1.

Table 1: Approximate composition of sulphides in roaster feed (dry basis)
Element Cu
2
S CuS CoS FeS
wt. % 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. Prod-
uct 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 cool-
ing 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 m
3
, 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 injec-
tion guns atomize the feed to approximately 250 m 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 air-
assisted 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 capabili-
ty through the use of controllable dampers and an outlet vent for low-flow operation.
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 off-
gas dust would dilute the concentration of copper and cobalt in the spent electrolyte, and has a nega-
tive 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 main-
taining 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.


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 sul-
phides present in the feed slurry. These include:
4CuS + 7O
2
2CuSO
4
CuO + 2SO
2
H
rxn
= -2,226 kJ (1)
2Cu
2
S + 5O
2
2CuSO
4
CuO H
rxn
= -1,703 kJ (2)
CoS + 2O
2
CoSO
4
H
rxn
= -795 kJ (3)
4FeS + 7O
2
2Fe
2
O
3
+ 4SO
2
H
rxn
= -2,439 kJ (4)

Additionally, the decomposition of MgCO
3
to MgO and CO
2
gas, according to reaction 5, is
important with regards to the thermal balance in the roaster.
MgCO
3
MgO + CO
2
H
rxn
= 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 per-
formed on samples of Katanga ore and concentrate feed slurry as well as on previous plant expe-
rience, 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 reten-
tion 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 selec-
ted 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.
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 com-
pounds and their associated potential for corrosive attack. Of particular concern is the formation of
H
2
SO
4
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 minim-
ize 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.
The novel design methodology integrated for the first time the steady state and transient heat trans-
fer analyses for a hot suspended/slumped bed inside a refractory lined, externally insulated/non-
insulated 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/non-
participating radiative heat transfer gas inside the wind-box. Finite Element Analysis (FEA) was
utilized in evaluating the thermo-mechanical stresses in the vessels 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 expan-
sion 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.
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 vertic-
al 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 ac-
ceptable 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.

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 envi-
ronment. 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 slur-
ry 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:
4CuSO
4
+ Cu
2
S 6CuO + 5SO
2
(6)
This reaction occurs between 420 C and 540 C, inhibiting the primary sulphation reaction (2) giv-
en 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 tem-
perature of up to 700 C, roaster off-gas rising from the bed undergoes significant cooling in the
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 dis-
tribution 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 selec-
ted 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.
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 incor-
porated 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 man-
ner, through an interactive human-machine-interface (HMI). At the start of each sequence, the op-
erator 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 operators at-
tention. 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 oper-
ating 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.

References
[1] THEYS, L.F. and LEE, L.V. (1958): Sulfate roasting copper-cobalt sulfide concentrates: Jour-
nal 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
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, SO
2
, SO
3

Abstract
The copper smelting process generates a process gas with high SO
2
, SO
3
, volatile heavy metals and
arsenic loads which is routed to a byproduct sulphuric acid plant for the recovery of marketable sul-
phuric acid. In the off-gas cleaning system, SO
2
, 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 conven-
tional weak acid treatment process has the drawback of generating large amounts of gypsum. More-
over, 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 ma-
terial in the sulphuric acid production process.
Ante
1 Introduction
The copper ores contain heavy metals and arsenic. During smelting, a process gas is produced con-
sisting of SO
2
, traces of SO
3
, HF, HCl, flue dusts containing heavy metals oxides and volatile arse-
nic trioxide (As
2
O
3
). This process gas is treated to catalytically oxidise SO
2
to SO
3
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 H
2
SO
4
. All constituents soluble in water, espe-
cially the SO
3
, 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 precipita-
tion 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.
2.1 Process stages
The process described below comprises the process stages shown in Figure 1. The process is di-
vided into technical gypsum production (Part 1; items 1 to 5) and conventional residuals precipita-
tion (Part 2; items 6 to 12).
Figure 1: Flow sheet for weak acid treatment plant with technical gypsum production [1]
1 Microfiltration: Separation of coloured solids for the production of a white gypsum
2 Precipitation stage 0: 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
Weak acid
Ca(OH)
2
powdered
FeCl
3
, HCl
Ca(OH)
2
liquid Micro-
filtration
Vacuum belt
filter
Sludge
holding tank
Belt filter
press
Filter cake
store
Pure gypsum
store
Secured
landfill
Gypsum for
sale to
industry
Discharge
Precipitation
stage 1
Precipitation
stage 0
Polyelectrolyte
Clarifier
1
Polyelectrolyte
Polyelectrolyte
Poly-
electrolyte
Precipitation
stage 2
Clarifier
2
Pure gypsum production Residuals precipitation
1
2
3
4
5
6
7
8
9
10
11
12

Ante
In fact, approx. 85 % of the produced solids are sold as technical gypsum and only 15 % are to un-
dergo 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 re-
quired 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 view-
point 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.
Table 1: Discarded Processes [2]
Process Reason for exclusion
Adsorption process
Iron hydroxide process pH 5-9
Pisolite pH 6.5
Retardation As removal <75 %
RGS-11 High Wash water demand
Precipitation
Thiopaques High space demand
High chemical costs (H
2
source)
Restricted availability (biology process)
Membrane separation process
Diffusion dialysis Huge membrane surface
Electro dialysis High power consumption
Nanofiltration As removal < 50 %
Miscellaneous
Solvent extraction (TBP) As removal < 40 %
The theoretically most promising processes were tested practically for their principal applicability
without any claim of reproducibility.
For a new process design we chose the well known sulphide precipitation with a subsequent nano-
filtration named Filsulforor combined gypsum precipitation: Gypsulfor.
3.2 Theory
3.2.1 Sulphide precipitation
The sulphide precipitation for the above application is of interest, as most heavy metal sulphides
exhibit a very low solubility product as shown in Table 2 in comparison to the hydroxide. A precipi-
tation in acids is also possible, there being no need to neutralise the sulphuric acid. Figure 3 shows
how the solubility product of the metal sulphides depends on the pH value.
Ante
Table 2: Some solubility products of metals at 25 C [3, 4]
Formula Solubility product Formula Solubility product
PbS 3*10
-28
Pb(OH)
2
4.2*10
-15

CdS 5.1*10
-29
Cd(OH)
2
2*10
-14

FeS 3.7*10
-19
Fe(OH)
2
1.8*10
-15

CuS 8*10
-45
Cu(OH)
2
1.6*10
-19

NiS 10
-26
Ni(OH)
2
1.6*10
-16

Ag
2
S 1.6*10
-49
Ag(OH) 2*10
-8

ZnS 6.9*10
-26
Zn(OH)
2
4.5*10
-17

SnS 10
-20
Sn(OH)
2
3*10
-27

Figure 2 shows that the residual solubility of the sulphides at low pH values is significantly lower
than the corresponding residual solubility of the hydroxides. Above all, the main decontaminate
arsenic has a residual solubility below 3 mg/l up to a pH value of 2.
Figure 2: Dependency of the residual solubility on the pH value during H
2
S precipitation [5]
The sulphide precipitation of heavy metals and arsenic is, in principle, possible with Na
2
S, NaHS
and H
2
S as well as some organic sulphides. The use of Na
2
S is initially accompanied by neutralisa-
tion of the sulphuric acid (see Equation 1):
Na
2
S + H
2
SO
4
==> Na
2
SO
4
+ H
2
S (1)

This is followed by the actual sulphide precipitation in Equation 2 with the example of arsenic As:
3 H
2
S + 2 H
3
AsO
3
==> As
2
S
3
+ 6 H
2
O (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 H
2
S 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 510
5
-3010
5
Pa) owing to its specific separating effect, while, how-
ever, 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
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
[%]
Precipitable
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
Partially precipitable
Iron 189 160 15
Not precipitable
Aluminum 49 49 0
Zink 219 210 0
Unknown
Cadmium n.a. n.a.
Chromium < 1 0.2 n.a.
Nickel < 1 0.2 n.a.
Sodium 97 34,320 accumulation
4.2 Nanofiltration
Due to the advances on the field of material quality for membranes since membrane for nanofiltra-
tion are actually available allowing the application even for such a critical media as weak acid char-
acterized by a negative pH value and a concentration of fluoric acid up to than 3 g/l. Two of these
new membranes were investigated which were theoretically resistant to the medium.
The investigations took place in a dead-end batch filtration lab plant with an agitated reactor. The
nanofiltration system used was operated using a filter with 36.3 mm diameter and a filling volume
of 200 ml [6].
4.2.1 Resistance of the membranes
After 18 days, no significant change could be determined in the clarified water flow, which would
have occurred in the case of damage. This result provides a starting point for the acid stability, but
does not permit any conclusion regarding the long-term performance.
4.2.2 Removal of heavy metals
The efficiency of the nanofiltration to remove the contaminants of the weak acid was investigated
and the results are shown in Figure 3.
Figure 3: Depiction of the retention of heavy metals and sulphuric acid (SA)
The retention at 22 C was significantly higher than at 45 C for all heavy metals, as the temperature
dependency of the diffusion is considerably greater than the flow.
Fe, Zn, Cu, Mo und Al can be retained well at room temperature (> 95 %). The retention of Pb and
Sb is poor.
As the metals Fe, Zn and Al are primarily to be reduced after the sulphide precipitation, the nanofil-
tration is ideally for the production of high-quality sulphuric acid through after-cleaning.
-20
0
20
40
60
80
100
Fe Zn Al Cu Mo Sb Pb SA
Compound
R
e
t
e
n
t
i
o
n

[
%
]
22 C
45 C

Ante
In all tests, the retention increased with the pressure (not shown), as the flow increases through the
pressure increases. However, the diffusion speed, by means of which the dissolved contents can
pass the membranes, remains constant.
5 Modern design concepts
5.1 Filsulfor
In the subsequent Figure 4 the principle of the new process concept design Filsulfor is shown.
Figure 4: Flow diagram for weak acid treatment plant with sulphide precipitation
The process can be divided into two parts. Part 1 deals with the precipitation of the bulk of heavy
metals and arsenic (items 1 to 3) with the sludge handling (items 4 to 5) and part 2 with the subse-
quent nanofiltration for the water treatment (processes 6 and 7).
1 Precipitation: Precipitation of Arsenic and the majority of heavy metals with NaHS
2 Clarifier: Separation of the precipitation products from the weak acid
3 Aeration: Elimination of the by-product H
2
S
4 Belt press: Sludge dewatering
5 Waste sludge basin: Storage of waste sludge
6 Filtration: Protection of the membranes against solids in the overflow of 2
7 Nanofiltration: Separation of remaining heavy metals (concentrate) from the sulphuric acid
(permeate)

5.2 Gypsulfor
In the subsequent Figure 5 the principle of the new process concept design Gypsulfor is shown.
Figure 5: Flow diagram for weak acid treatment plant with sulphide precipitation
The process is divided as well into two parts. Part 1 remains the sulphide precipitation with the pre-
cipitation of the bulk of heavy metals and arsenic (items 1 to 5) and part 2 is the subsequent pure
gypsum precipitation (items 6 and 7) with the handling of the produced pure gypsum.
1 Precipitation: Precipitation of Arsenic and the majority of heavy metals with NaHS
2 Clarifier: Separation of the precipitation products from the weak acid
3 Aeration: Elimination of the by-product H
2
S
4 Belt press: Sludge dewatering
5 Waste sludge basin: Storage of waste sludge
6 Filtration: Protection of the secondary product against solids in the overflow of 2
7 Pure Gypsum Precipitation: Neutralization of the sulphuric acid with milk of lime
8 Vacuum belt filter: Gypsum washing and dewatering
9 Gypsum storage
NaHS
WeakAcid
Clarifier
Belt
Press
Aeration Precipitation
WasteSludge
Ca(OH)2
powdered
Micro-
filtration
Vacuumbelt
filter
Puregypsum
store
Gypsumfor
Saleto
Industry
Precipitation
stage0
1 6 2 3
4
5
7 8
9

Ante
6 Evaluation
6.1 Discussion of the process
In Figure 6 the overall mass balance for Filsulforfor a real copper plant is exemplarily shown and
in Figure 7 the mass balance for Gypsulfor.
Figure 6: Overall mass balance for Filsulfor
WWTP
Purified sulfuric acid
(for reuse)
Waste sludge
1 t/h
contaminated with
heavy metals
(60 % DS)
NaHS
1 t/h (DS = 40 %)
Weak acid
V = 20 m
3
/h
As = 300 kg/h
Me
+
= 244 kg/h
H
2
SO
4
= 3,940 kg/h
V = 18.8 m
3
/h
As = > 0 kg/h
Me
+
= 12 kg/h
H
2
SO
4
= 3,550 kg/h
V = 2.1 m
3
/h
As = > 0 kg/h
Me
+
= 167 kg/h (Zn and Fe)
H
2
SO
4
= 390 kg/h
Waste water

Figure 7: Overall mass balance for Gypsulfor
7 Outlook
7.1 Filsulfor
Sulphuric acid is anyway a product which is produced as by-product in large quantities by the cop-
per smelters. Therefore, the sale of sulphuric acid is a routine process and the number of clients is
much higher than for technical gypsum. An additional external client is not necessary.
Another potential use of the dissolved sulphuric acid produced involves the internal application for
the leaching process completely avoiding external sale. It therefore safeguards the supply.
7.2 Gypsulfor
The Gypsulfor treatment concept is suitable for sites where the secondary raw product is used on
site itself or where the secondary raw product gypsum is sellable. This is given mainly when the
copper plant is in vicinity to the sea offering the cost efficient transportation of the bulk product.
WWTP
Pure gypsum
(for reuse)
Waste sludge
1 t/h
contaminated with
heavy metals
(60 % DS)
NaHS
1 t/h (DS = 40 %)
Weak acid
V = 20 m
3
/h
As = 300 kg/h
Me
+
= 244 kg/h
H
2
SO
4
= 3,940 kg/h
0.84 t/h
As < 1 mg/kg
V = 53 m
3
/h
As < 1 mg/l
Waste water

Ante
The presented design concepts can contribute to the efforts of mankind to recycle secondary by-
products and to preserve natural non-renewable and exhaustible sources mainly for gypsum because
its winning causes extensive damages to the environment.
7.3 Further steps
For individual projects the cost reduction potential in accordance with local conditions has to be
considered such as the reduction of reagent costs of NaHS for the sulphide precipitation, the reduc-
tion of the membrane exchange and energy costs for the nanofiltration.
As the economy of the whole treatment plant is strongly dependant from the local conditions and
regulations these studies have to be done for each site separately taking into account the particular
situation. Mainly the treatment or disposal of the liquid and solid residues of the processes is to be
discussed, e.g. the concentrate (10 % of the input volume stream) deriving from the nanofiltration
step.
The different process concepts refer to the possible combination of treatment stages. The described
sulphide precipitation stages can be realized with different reagents and can be combined either with
the nanofiltration or with the technical gypsum precipitation. In the later case it has to be investi-
gated if the neutralisation and /or the precipitation with ferrous compound are still necessary to meet
the national and international requirements.
References
[1] ANTE, A., SCHNBRUNNER, S. (2005): Production of marketable gypsum from waste weak acid
- World of Metallurgy, 2: 75-82; Clausthal-Zellerfeld.
[2] EICHMANN, C. (2007): Trennung von Arsen und Schwermetallen aus Waschsure, master the-
sis, Butzbach.
[3] C. E. MORTIMER (2001): Chemie Basiswissen der Chemie, 7. korrigierte Auflage, Georg
Thieme Verlag.
[4] HARTINGER, L. (1991): Handbuch der Abwasser und Recyclingtechnik, Wilhelm Ernst & Sohn
[5] RAMACHANDRAN V. (2007): Processing of aqueous effluents in the copper industry, One day
short course in Toronto, Canada the 26
th
of Sep 2007; Toronto.
[6] RGENER, F. (2007): personal information and correspondance, BFI, the 20
th
of Mar 2007;
Dsseldorf.

Feasibility to Profitability
with Copper ISASMELT
G. R. Alvear F., P. Arthur, P. Partington
Xstrata Technology
Level 4, 307 Queen Street
Brisbane 4000, Australia

Keywords: Pyrometallurgy, Top Submerged Lance (TSL), ISASMELT
TM
, copper, southern Peru
Abstract
The ISASMELT Top Submerged Lance (TSL) bath smelting process was developed in Mount
Isa, Australia by Mount Isa Mines Limited (now a subsidiary of Xstrata plc) during the 1980s. It is
now successfully commercialized and has become the technology of choice for many new smelters
and smelter modernization projects, being extremely cost effective for smelting both copper concen-
trates and secondary materials. Ten copper smelters are now operating the process around the
world.1Two new furnaces were commissioned in 2009, while three more furnaces are under con-
struction and are scheduled to be commissioned during 2010 and 2011.
Production capacities of more than 330,000 tonnes per year of copper are being achieved through a
single ISASMELT furnace with instantaneous feed rates reaching up to 200 t/h.
Process development continues on the commercial scale plants at Mount Isa and elsewhere around
the world. By employing the design and operational experience gained by Xstrata over more than
25 years of process development ISASMELT plants can be constructed in remote locations in a
relatively short time and achieve rapid ramp up to design capacity with limited technical resource.
As the family of ISASMELT licensees grows the members are able to share and capitalize on the
knowledge and experience with their peers around the world.
This paper reports on copper ISASMELT projects completed in recent years and how licensees
have progressed from feasibility studies to profitable operations using Xstrata Technologys tech-
nology transfer process.
1 Introduction
It was in 1975 when Mount Isa Mines Limited (MIM) set its sights on a new development occurring
at CSIRO in Australia: the Top Submerged Lance (TSL) technology, a new concept for smelting
Alvear, Arthur, Partington
using the Sirosmelt lance. MIM joined forces with CSIRO and participated in lab scale trials of the
technology.
At that time, MIM was looking for new technologies that could be applied to its lead and copper
smelter operations to reduce operating costs while improving the environmental performance.
MIMs Mount Isa lead smelter used a sinter plant and blast furnace for lead bullion production,
while the copper smelter used a fluid bed roaster and two reverberatory furnaces to produce copper
matte. The matte was then converted to blister copper using Peirce-Smith converters [1].
MIM was seeking alternative smelting processes that would produce off-gases with higher SO
2
con-
tent so that the smelter gases could be treated and sulphur captured in an acid plant. MIM also
needed to find a process with lower operating costs to remain cost effective with the steady decline
in real metal prices.
While keeping an eye on the development of other technologies MIM, jointly with CSIRO, devel-
oped the concepts of the Lead ISASMELT and Copper ISASMELT processes. These were pi-
lot tested on a 250 kg/h test rig in Mount Isa in the early 1980s.
From the completion of the pilot scale tests a long history unfolded involving a multidisciplinary
team of visionaries, who saw the technical and business potential that ISASMELT technology
could offer. Over the years ISASMELT
TM
has progressed from a 250 kg/h pilot plant scale to indus-
trial facilities that can run up to 200 t/h and treat a variety of materials including nickel, lead and
copper concentrates and secondary materials [2-4].
Since the first plant was built at Mount Isa, twenty ISASMELT plants have either been con-
structed or are under construction. Figure 1 shows the location of the commercial plants that have
been licensed to date.
An important milestone in the market consolidation of the technology was when MIM became part
of Xstrata plc in 2003. At that time MIM Process Technologies, the division that was responsible
for the commercialization of the technology, became Xstrata Technology (XT). XTs mission was to
market the core technologies developed in Xstratas operating sites: IsaMill and Jameson Cell
technologies in mineral processing, ISAPROCESS for the electrorefining and electrowinning of
copper, the Albion atmospheric leaching process and the ISASMELT and ISACONVERT
technologies for the smelting and converting of non-ferrous materials.
Feasibility to Profitability with Copper ISASMELT
Figure 1: Location of ISASMELT plants licensed to date
During 30 years developing and operating ISASMELT technology on large scale plants, signifi-
cant technical improvements have occurred in areas such as furnace design, feed preparation sys-
tems, off-gas handling, operating and process control strategies, refractory management, operator
training and commissioning systems. The combination of experience led to the development of the
ISASMELT technology package that is licensed to external clients today [5]. Many of the im-
provements implemented by plant operators have been passed on to, and adopted by, other licen-
sees. Exchange of ideas and technical improvements occurs through visits to fellow licensee sites
and through regular licensee workshops arranged by XT at locations close to the ISASMELT
TM
in-
stallations, the most recent being at Arequipa, Peru in 2008 shown in Figure 2. It included a visit to
the Southern Peru Copper Corporation (SPCC) ISASMELT
TM
plant.
Figure 2: Licensee workshop in Arequipa, Peru, October 2008
2 The ISASMELT principle
2.1 ISASMELT concept
ISASMELT technology is based on the use of an elegant furnace design which is readily enclosed
to eliminate emissions to the surrounding environment. It uses submerged lance injection technolo-
gy to provide highly efficient mixing and reaction of feed materials in a molten slag bath. The use of
an advanced process control system results in the furnace operation being largely automated.1Being
a vertical furnace with a small footprint it can be easily retrofitted into existing smelters to either
augment or replace existing technology. The furnace concept is shown in Figure 3.

Figure 3: ISASMELT process concept
2.2 Copper ISASMELT reaction mechanism
The Copper ISASMELT process is a slag reaction process, where fresh feed is digested into the
molten slag phase. It is in this phase where the main chemical reactions occur and oxidation of the
feed takes place.
The oxygen transfer process is achieved through the controlled oxidation of the slag (FeO) and sub-
sequent formation of magnetite (Fe
3
O
4
) as shown schematically in Figure 4. It is the liquid oxy-
gen from the magnetite (Fe
2
O
3
FeO) that reacts with the concentrate and fluxes to form copper
matte, a fayalite slag and SO
2
gas.

Figure 4: Copper ISASMELT reaction mechanism
2.3 Slagmatte separation: the role of the Rotary Holding Furnace
Copper matte and slag generated from the reactions occurring in the Copper ISASMELT furnace
are tapped and mechanically settled in a separate furnace. The type of settling furnace will depend
on the economics of each operation. XT supplies a proprietary technology known as the Rotary
Holding Furnace (RHF). This process was developed at MIM and is currently used at the Mount Isa
smelter, the Sterlite smelter in India, and the SPCC smelter in Peru. The RHF is a horizontal, cylin-
drical, refractory-lined furnace that allows mechanical settling of the matte and the slag. Typically,
for ISASMELT
TM
furnaces with instantaneous feed rates greater than 150 t/h, the stream of slag and
copper matte tapped from the ISASMELT furnace is fed to either of two RHFs operating in pa-
rallel.
The RHF has the following three primary functions:
Holding function it provides a storage location for copper matte; this provides surge capacity
between the (upstream) continuous ISASMELT furnace and the (downstream) batch Peirce-
Smith converters.1
Settling function molten copper matte is separated via gravity separation from molten slag
Separation function molten slag is skimmed from the top of the RHF bath into a granulation
system or ladle; molten matte can be poured from the lower portion of the RHF bath into a ladle
for delivery to the converters.1

The RHF has an entry port, or feed port, to receive the mixed matte and slag stream tapped from
the ISASMELT furnace. Matte and slag products are poured out of the RHF via a matte spout and
slag spout respectively. These spouts are located at the far end of the RHF, resulting in maximum
residence time for separation to occur. The RHF can be tilted away from the converter aisle to allow
skimming of discard slag, or towards the aisle to allow pouring of matte. A hydraulic drive is used
to rotate the furnace via a heavy duty chain and sprocket system. When pouring slag an automated
system is used to provide slow and steady skimming of slag from the slag port. This system enables
a slag skim that minimizes the amount of matte carry-over to the slag.
The temperature of the RHF is maintained by means of burners located in the roof and the end wall
of the furnace.
2.4 Advantages of the RHF technology
The use of the Rotary Holding Furnace technology to hold, settle and separate copper matte and slag
generated in the ISASMELT furnace has a number of advantages:
No electrical energy required: The matte and slag are physically separated in the furnace with-
out the need of electrical energy for reduction. RHFs have been applied for copper matte
grades up to about 65 % Cu.
Low copper in slag can be achieved: The slag is slow poured via an automated system to mi-
nimize matte carry-over to the slag.
Avoids back-contamination of the matte with minor elements already deported to the slag
phase: By avoiding reduction of the slag, the minor element distribution between the slag,
matte and gas phase remains unchanged.
Ability to provide matte to the Peirce-Smith converters in a very short period of time: By tilt-
ing the furnace and pouring through the matte spout a 30 tonne matte ladle can be filled in one
minute.
Provision of a surge capacity between the smelting and converting processes.
3 Recent copper ISASMELT projects
Since the last review of the technology presented in the Copper/Cobre 2007 conference, in Toronto,
Canada, new projects and plants were added to the list of Copper ISASMELT plants around the
world. The new projects and plants are identified in Table 1. The SPCC project that was commis-
sioned in the first quarter of 2007 is also included in Table 1.
Table 1: New copper ISASMELT projects since 2007
Project Country Design Capacity
(t/y concentrate)
Commissioning
Year
Southern Peru Copper Corporation Peru 1,200,00 2007
YCC Chuxiong China 500,000 2009
YCC Chambishi Zambia 350,000 2009
Kazzinc Kazakhstan 290,000 2010
Doe Run Peru Peru 280,000 2011
Sterlite III India 1,360,000 2011
The new projects include plants of relatively small capacity like Doe Run Peru and Kazzinc with
important strategic value as both plants will treat complex sulphide concentrates containing high
contents of minor elements such as As, Pb, Zn and Sb. In these two cases the ISASMELT fur-
naces ability to eliminate highly volatile minor elements to the gas phase will contribute to produc-
ing a relatively clean copper matte compared with the product that could be produced by using al-
ternative smelting technologies [6]. The application of the ISASMELT technology in Yunnan
Copper Corporation (YCC) plants in Chuxiong, China and Chambishi, Zambia was the natural path
to follow by this company after the successful implementation of their first ISASMELT plant at
Kunming, China in 2002. Figure 5 shows a view of the ISASMELT plant in Chambishi that was
commissioned in early 2009.
Figure 5: View of the YCC Chambishi ISASMELT
TM
plant in Zambia

Kazzincs Ust-Kamenogorsk Metallurgical Complex in Kazakhstan is a fully integrated plant, in
which most by-products from each stage of the operation are enriched to form saleable metal prod-
ucts. A new copper smelter and refinery is being installed as part of the Kazzinc New Metallurgy
Project. Figure 6 shows a view of the Kazzinc Copper ISASMELT
TM
Plant. In addition, the lead
smelting process flowsheet will be upgraded by replacing the traditional Sinter Plant - Blast Furnace
process with more modern technology in the form of a Lead ISASMELT
TM
Furnace that will pro-
duce a lead rich slag to feed an existing Lead Blast Furnace.
The copper smelter will use an ISASMELT
TM
furnace to produce copper matte. The matte and slag
will be tapped from the ISASMELT
TM
furnace to an electric settling furnace. The matte will then be
converted and refined to produce anode in a conventional Peirce-Smith converter/anode furnace
aisle. Anode will then be refined using XTs ISAPROCESS
TM
technology. The Copper
ISASMELT plant is scheduled to be commissioned in 2010.
Sterlite Industries (India) Limited recently decided to expand their copper smelting and refining
complex at Tuticorin in southern India. They will build their third ISASMELT plant to continue
their expansion in the copper business. The new smelter will be located adjacent to the current facil-
ity and will have a design throughput capacity of 400,000 t/a of copper (equivalent to 1,360,000 t of
copper concentrate) through a single ISASMELT furnace. The matte and the slag from the
ISASMELT furnace will be separated in two RHFs and the matte will be converted in Peirce-
Smith converters. Anode copper will be refined using XTs ISAPROCESS
TM
technology.
Figure 6: View of the Kazzinc ISASMELT
TM
plant in Kazakhstan
As a result of these new plant installations the Copper ISASMELT
TM
family has continued to grow,
resulting in a transformation of the copper smelting industry. Figure 7 shows the increase in global
capacity of Copper ISASMELT plants since the first commercial plants were commissioned in
1992. The total capacity of the installed Copper ISASMELT plants in 2011 will be over 9 million
tonnes of copper bearing materials.

Figure 7: Copper ISASMELT capacity
4 The ISASMELT technology package
Behind this transformation of the copper smelting industry is the concept of delivering a technology
package where all the components need to work in concert the furnace, the lance system, the re-
fractories, feed preparation and product tapping systems, control system, off-gas cooling and clean-
ing systems, together with training and procedures for operations and maintenance personnel.
The ISASMELT technology package is an integrated assortment of technological research and
development, specialist process and mechanical design, proprietary equipment, know-how, training
programs, commissioning assistance and on-going technical support and collaboration that combine
to ensure successful smelter projects for XTs licensees. The technology transfer includes a unique
arrangement for training operators from new licensees they learn by operating the full scale pro-
duction smelter at Mount Isa. Clients from the USA, Belgium, Germany, China, Peru, India, Zambia
and Kazakhstan have been trained at the MIM smelter in Mount Isa over the years. During the train-
ing period, client operators and maintenance personnel are given the opportunity to operate and
maintain the Mount Isa plant. This real life training is a great advantage for the trainees when it
comes to operating their own plants. By applying accumulated operating and maintenance know-
how along with the full technology package, new users can very quickly achieve high production
levels and long furnace campaigns, avoiding the pitfalls that often plague new smelter projects and
reduce their profitability.
A further important part of the technology package is the interaction achieved through both informal
contact between technical personnel at the many operations and regular licensee workshops, where
0
1
2
3
4
5
6
7
8
9
10
1992 1996 1997 1998 2001 2002 2004 2005 2006 2007 2008 2009 2010 2011
M
i
l
l
i
o
n

t
o
n
n
e
s

p
e
r

a
n
n
u
m

f
e
e
d

m
a
t
e
r
i
a
l

ISASMELT users share innovations for mutual benefit. The companies can harness the ideas of
operators from all over the globe.
In addition, XTs engineers and metallurgists are involved from conceptual design through the de-
tailed design process, manufacturing of specialist equipment, commissioning, and then a long-term
ongoing relationship with licensees. So they are currently not only busy designing four major smel-
ters on three continents but are also actively "closing the loop" after witnessing the results of the
design process in operation.
5 SPCCs project: The technology package at work
XT provided a number of studies for SPCC during the 1990s when they were considering how best
to upgrade their smelter at Ilo, Peru. It was in early 2003, when SPCC was in the process of select-
ing a new technology that would meet their requirements while keeping capital and operating costs
low, when XT offered ISASMELT as an option for the modernization of the smelter. At that
time, SPCC was facing the challenge of modernizing its installations in order to comply with new
environmental regulations defined by the Peruvian government in terms of sulphur capture and par-
ticulate material emissions during a period of historically low copper prices.
XT in partnership with Fluor Chile S.A. (Fluor) presented an offer to evaluate, at conceptual level,
the use of the ISASMELT and RHFs as the core technology for processing 1,200,000 t/a of cop-
per concentrate.
After technical and economic evaluation of the conceptual study completed by XT and Fluor, SPCC
decided to select the ISASMELT and RHF technology package for the modernization of the Ilo
smelter in the middle of 2003. SPCC have stated that key criteria for selecting ISASMELT
TM
were
the fact that the ISASMELT technology was proven for the planned capacity (1,200,000 t/a
through a single furnace) and the lower capital and operating cost of the ISASMELT technology
compared with the other competing technologies [7]. Figure 8 shows a view of the SPCC
ISASMELT plant.
Figure 8: SPCC Ilo Smelter Copper ISASMELT plant view
5.1 The development of the SPCC project
The scope of work for XT was divided into the following stages:
Phase 1 Preliminary Engineering: Preliminary engineering for the ISASMELT and RHF fur-
naces was carried out at the XT office in Australia with active participation of Fluor and SPCC.
Several technical review meetings were scheduled to review the progress of the project in Brisbane
and Santiago. A project team from SPCC was established and located in Fluors Santiago office.
XT provided highly qualified technical support to the project team with design engineers allocated
to the project team in the Santiago office for an extended period. This model allowed working 24
hours a day by having people based in Santiago, Ilo and Brisbane. Regular use of advanced project
management tools allowed the three parties to share and review documents on line on both sides of
the Pacific Ocean. During this stage, the ISASMELT licensees network played an important role,
with XT arranging industrial scale tests of SPCCs Cuajone and Toquepala concentrates at the YCC
ISASMELT plant in China, in order to confirm relevant process information required for the de-
sign.

Phase 2 Detailed Engineering: Following evaluation of the estimated costs on completion of
Phase 1, SPCC decided to continue with the project. In February 2004 SPCC signed a Professional
Services Agreement with XT and an engineering, procurement and construction management
(EPCM) contract with Fluor. That month the Phase 2 kick-off meeting was held to start the detailed
design of the plant. During Phase 2 XT delivered the technical information required by Fluor to
continue with the detailed design process of the smelter. In addition, XT reviewed the mechanical
and structural engineering completed by Fluor during this stage.
XT Equipment Supply: As part of the services agreement with SPCC, XT was awarded with the
supply of core equipment for the ISASMELT plant. Lance hoisting, lances, specialist instrumen-
tation and control equipment and the hydraulic RHF drive system were part of the XT package.
Process Control System Design and Supply: An important part of the success of the ISASMELT
technology is its process control system. XT in conjunction with its partner MIPAC delivered a
complete process control package for the ISASMELT plant.
Training Program: Thorough training is essential to enable successful transfer of technology. The
process training for SPCC took place in four phases:
1. An initial familiarization training at the Freeport Copper Smelter in Arizona,
2. Full scale three month plant training at the MIM smelter in Australia,
3. A detailed training workshop at the Ilo smelter prior to plant startup, and
4. On-site training during commissioning of the plant where cold and hot commissioning of the
furnaces, cold test runs, start-up, shut-downs, and maintenance and trouble-shooting cases were
explained.
Technical Assistance: An important part of the project was the technical assistance provided by XT
to SPCC during the engineering, pre and post commissioning stages of the project. The technical
services included support of XT and MIM smelter personnel to Fluors Santiago office and the Ilo
smelter to carry out the following duties:
1. To assist Fluor with the detailed design of the plant.
2. To assist Fluor with installation of key equipment items.
3. To assist with training of SPCC personnel on the new ISASMELT plant.
4. To assist with the supervision of distributed control system installation.
5. To assist with supervision of plant commissioning and startup.
6. Supervision of the startup and hot commissioning of the integrated plant.
7. Regular support on site after finishing commissioning with XT process experts.
8. Joint task force work with SPCC to optimize furnace operation during the first two years of plant
operation.
The plant was commissioned in February 2007, 36 months after signing of the Professional Services
Agreement.
5.2 Commissioning of the plant and post commissioning assistance
In addition to the assistance from XT in equipment installation and cold commissioning activities
MIM smelter operators and XT metallurgists and engineers, with many years of operating expe-
rience in ISASMELT
TM
plants, worked on site with SPCCs project team and operations personnel
to carry out the hot commissioning of the plant. Hot commissioning commenced with heat up of the
ISASMELT
TM
furnace and RHFs, to bath make in the ISASMELT
TM
furnace, first feed on, first tap
of matte and slag to the Rotary Holding Furnaces and then over a number of weeks to stable smelt-
ing operations.
After stable operations were achieved XT and SPCC formed a taskforce to optimise the
process.1One example of the task forces activities was the optimization of the slag chemistry. Rela-
tively high alumina contents in the SPCC concentrates result in higher alumina contents in the slag
(4 to 7 %) than at ISASMELT operations elsewhere. Relatively high alumina contents in the slag
can cause not only a decrease in the liquidus temperature of the slag but also an increase in the vis-
cosity for a given temperature. New targets for the slag chemistry and operating temperature were
defined following joint efforts between SPCC and XT that included fundamental work on slag mod-
elling in conjunction with the University of Queensland and Xstrata Process Support in Sudbury,
Canada, plant observations and trials. As a result of the change in slag chemistry the operation was
improved. The benefit of modifying the ISASMELT
TM
slag chemistry was that a less viscous slag
was produced leading to improved tapping operations and a decrease in the copper content of the
RHF discard slag.
5.3 Current operation at SPCC
Table 2 shows key process parameters for the SPCC plant for June 2009. At time of writing this
paper the plant was running at high availability with a feed rate of up to 183 t/h (nominal feed rate:
165.2 t/h of copper concentrate, dry basis).
Table 2: Typical process parameters for SPCC ISASMELT
TM
plant for June 2009
Parameter Value Unit
Maximum concentrate feed rate 183 dry t/h
Average concentrate feed rate 165.2 dry t/h
Average Cu content in concentrate 27.3 %
Average moisture of the feed 9.2 %
Average silica flux feed rate 22.1 dry t/h
Average coal feed rate 1.1 dry t/h
Average reverts feed rate 6.3 dry t/h
Total average feed to the furnace 211.6 wet t/h
Average copper matte grade 62.5 %
Average SiO
2
/Fe in slag 0.7 --
Average ISASMELT
TM
lance air flow rate 15.4 Nm
3
/s
Average oxygen enrichment in ISASMELT
TM
lance air 66.7 %
Bath temperature range 1175 - 1185 C
6 Conclusions
The application of the Copper ISASMELT technology to process primary and secondary mate-
rials has steadily grown over the past 10 years, transforming the copper smelting industry. Copper
ISASMELT capacity is expected to exceed 9 million tonnes per year by 2011.
The basis for the success of the technology is a combination of the elegance of the process concept
with the diverse plant experience gained over more than 20 years operation of ISASMELT plants
at Mount Isa and a technology package that allows XT to provide its clients with process and engi-
neering designs, equipment supply and strong technical assistance through all the stages of the
project from initial conceptual studies at feasibility level through construction and the hot commis-
sioning of the plant and into the initial years of plant operation until the process is optimized for the
specific feed materials being treated. This process does not end there but continues through the life
of the plant with a continuous collaboration with the ISASMELT licensees through regular con-
tact and Licensees Workshops held on a regular basis and regular XT support to the clients when
required.
The Ilo smelter modernization project constitutes an example of how XT assists a licensee from
feasibility to profitability. A collaborative effort with a strong technical team comprising specialists
from both companies are continuing to work together to bring the Ilo ISASMELT plant to new
horizons.
Acknowledgments
The authors would like to acknowledge:
SPCC for its willingness to work jointly with XT throughout the project. The collaborative effort
and SPCCs openness to new ideas has been instrumental to the projects overall success.
Mount Isa Mines Copper Smelter for its continuous support.
Xstrata Technology for allowing the authors to publish this paper.
References
[1] FOUNTAIN C. (2002), The ISAMILL and ISASMELT models of successful R&D
11 AusIMM, Young Leaders Conference, Australia.
[2] BAKKER M.L., F. G.R. ALVEAR and KREUH M. (2009), ISASMELT TSL making a
splash for nickel,1Proceedings of the International Symposium Pyrometallurgy of Nickel and
Cobalt, MetSoc CIM-ICM, 181-193, Sudbury, Canada.
[3] ERRINGTON B., ARTHUR P., WANG J. and DONG Y. (2005), The ISA-YMG lead smelt-
ing process, Proceedings of the International Symposium on Lead and Zinc Processing, MMIJ,
581-599, Kyoto, Japan.
[4] ARTHUR P.S. and HUNT S.P. (2005), ISASMELT 25 years of continuous evolution,
Floyd International Symposium on Sustainable Development in Metals Processing, NCS Asso-
ciates (Australia), 73-94.
[5] BURFORD B. (2009), The ISASMELT technology package: Over 30 years of innovation The
AusIMM Bulletin, Journal of the Australasian Institute of Mining and Metallurgy, 1: 26-30.
[6] G.R. ALVEAR F., HUNT S.P. and ZHANG B. (2006), Copper ISASMELT dealing with
impurities, Sohn International Symposium Advanced Processing of Metals and Materials, Vol.
8: 673-686 San Diego, USA.
[7] MARISCAL S.L. and TORRES P.W. (2007), Modernization and start-up of the SPCC Ilo
smelter, Cu2007 - The Carlos Diaz Symposium on Pyrometallurgy, August 2007, Vol III: 193-
202, Toronto, Canada.

Present and Future Modernization of Metallurgical
Production Lines of the Gogw Copper Smelter
Leszek Byszyski, Leszek Garycki, Zbigniew Gostyski, Tomasz Stodulski, Jerzy Urbanowski
KGHM Polska Mied S. A.
ul. Marii Skodowskiej-Curie 48
59-301 Lubin, Poland

Keywords: Copper, process modernization, precious metals, lead, process heat balance
Abstract
This paper characterizes the current state of the metallurgical production lines of the Gogw Cop-
per Smelter. It discusses current and planned modernization projects, and also outlines the condi-
tions under which modernization is taking place.
KGHM, with annual production of 520 thousand tonnes of copper, holds a middle position in the
world ranking of copper producers. At the same time, in Europe it is the largest producer of copper
from its own ore deposits, possessing its own integrated production structure, from extraction to
final, top-quality product. KGHM Polska Mied is one of the largest Polish exporters.
Over 80 % of the electrolytic copper produced in KGHM, and 100 % of the precious metals associ-
ated with the copper ore, is produced in the Gogw Copper Smelter. The Gogw Copper Smelter
Complex is comprised of four smelters. Two major copper producing lines with different copper
concentrate smelting technologies are located in the Gogw I and II Copper Smelters. The other
two constituents of the complex are the Precious Metals Smelter and the Lead Smelter. Gogw I,
where shaft furnace technology is used, was commissioned in 1971. Gogw II launched in 1978
uses advanced flash smelting technology. The final product of both Smelters takes the form of elec-
trolytic copper cathode, containing a minimum of 99.99 % Cu. In the years 1990 to 1993 the Pre-
cious Metals Plant was built, where anode sludge, a by-product arising in the production of electro-
lytic copper, is treated, and silver, gold, palladium-platinum sludge and technical selenium are
recovered.
In 1973 the Lead Smelter came into operation, its production based on the materials recovered by
copper smelting technology. In the years 2001 to 2003 it underwent a total upgrade, and as a result
its production capacity increased from 17,000 to 22,000 tonnes of crude lead per year. Its target pro-
duction capacity after the end of the next stage of modernization will reach the level of 27,000 ton-
nes per year.
Byszyski, Garycki, Gostyski, Stodulski, Urbanowski
1 Gogw I Copper Smelter
Production at Gogw I is based on the smelting of copper concentrates in shaft furnaces.
Figure 1: A flow sheet of Gogw I Copper Smelter and Refinery
The manufacturing process includes five phases: the briquetting of copper concentrates, the smelt-
ing of briquettes in shaft furnaces with copper matte production, copper matte converting, fire refin-
ing in the anode furnaces with anode casting, and finally the electrorefining of copper anode, where
copper cathode is produced.
The blended and briquetted concentrate, containing approximately 25 % copper, including converter
slag and coke, is fed to the shaft furnace. The product obtained from melting this charge is copper
matte an alloy of copper and iron sulphides and waste slag. During the copper matte converting
process, iron and sulfur are oxidized to obtain the desired blister copper, with a pure copper content
of approximately 98.5 %. The remaining impurities are removed during fire refining in the anode
furnaces.
The cast anodes obtained from this stage are subjected to electrorefining, during which the copper
anode dissolves in the electrolyte, and under a constant current collects on cathode starter sheets,
creating the final product: 99.99 % pure cathode copper.

Present and Future Modernization of Production Lines of the Gogw Smelter
2 Gogw II Copper Smelter
Production at the Gogw II smelter is based on one-stage flash smelting technology in a flash
furnace, in which blister copper is produced directly from the dried concentrate. This technology
combines three phases: concentrate roasting, the smelting of copper matte and converting in a single
process.
Figure 2: A flow sheet of Gogw II Copper Smelter and Refinery
The dried copper concentrate is fed with an adequate amount of oxygen-enriched air through the
burner located in the roof of the furnace. Smelting takes place during the fall of the sprayed concen-
trate into the furnace settler, where a separation of the blister copper (containing approximately
99 % Cu) from the slag takes place. Slag, still containing approximately 11-15 % copper, is fed into
the electric furnace, where the slag cleaning process is performed. Meanwhile the blister copper,
identically as in Gogw I, is processed in the anode furnace and subjected to electrorefining. The
final product is 99.99 % pure cathode copper.

3 Precious Metals Plant
The Precious Metals Division at Gogw II produces silver, gold, palladium-platinum concentrate
and technical selenium.
The charge material is anode sludge a by-product of the copper electrorefining process, which
comes from the Gogw and Legnica Copper Smelters. After drying, the metal-rich sludge is sent to
the Kaldo furnace, where melting, converting and fire refining are performed. The product of this
process, Dor metal, is cast into silver anodes. These anodes are then electrorefined into silver cath-
odes of 99.99 % purity and cast into bars or grains.
The slime from the silver electrorefining process contains gold, platinum and palladium. Those pre-
cious metals are periodically recovered from the reaction of water and hydrochloric acid in the pres-
ence of gaseous chlorine. The gold is then recovered in another reactor in the form of gold sand.
During this stage a concentrate of platinum palladium is obtained. Afterwards the gold is melted
in a furnace and cast into gold bars.
Figure 3: A flow sheet of Precious Metals Plant

4 Crude Lead Plant
Apart from iron, lead accompanies the copper in the largest quantities. Its content in KGHM copper
concentrates reaches the level of 2.5 %.
Lead is concentrated mainly in the by-products from the metallurgical processing of copper concen-
trate, i.e. dust and slimes accumulating in technological gas cleaning installations. Dust and slimes
also are a major source of lead recovery, which takes place in the Lead Plant located at the Gogw
smelter.
Lead recovery from the by-products of copper smelting is based on reduction in a rotary-
reverberatory Drschel furnace. The reduction is performed cyclically and is a one-stage process. In
other words, the final product crude lead is produced in one operation.
After melting the lead is segregated from the slag in vats. Crude lead, with lead content of at least
98.5 %, is then cast in moulds. Currently, annual production of crude lead is about 21 thousand ton-
nes.
Figure 4: A flow sheet of Lead Plant

5 Production by the Gogw Copper Smelter
Production by the Gogw Copper Smelter in 2008:
Electrolytic copper: 427,304 tonnes
Silver: 1,192,751 kg
Gold: 902 kg
Crude Lead: 23,689 tonnes
Sulfuric Acid: 518,800 tonnes
Selenium: 81,857 kg
Pt-Pd concentrate: 93 kg
Nickel Sulfate: 1615 tonnes
Shaft Furnace slag: 551,440 tonnes
Granulated slag: 340,580 tonnes
5.1 Environmental Protection
The Gogw smelter is one of the cleanest copper smelters in the world, and today sets the environ-
mental standards for large industrial complexes throughout the world.
The modernity of the Gogw smelter may be seen not only in the modern technologies used in the
production process, but also in its strict adherence to environmental standards. Over the past
10 years, expenditure on environmental investments amounted to more than 300 mio. PLN
(75 mio. EUR). As a result of these investments the environmental impact of the Gogw smelter in
terms of the emission of harmful substances, sewage and waste products has been reduced to the
level of the leading companies in the world.
Thanks to technical and technological progress, and to investments (the modernization of shaft fur-
naces, the sulfuric acid plants, the lead plant, commissioning of the new rotary anode furnaces
equipped with a Dust Cleaning Installation and flue gas desulphurization installation) the Gogw
smelter has become one of the cleanest smelters in the world. In the forest protection zone the bio-
logical balance has been restored. In areas adjacent to the smelter, a variety of rare and wild plants
grow. Cranes, herons, storks and terns, as well as wild ducks, deer and foxes, also live there.
All of these environmental investments have been made in the interests of the surrounding environ-
ment and to ensure the best working conditions to the employees. The Gogw smelter meets the
requirements of environmental protection, both Polish as well as European standards. This means
that the Gogw smelter has all the required environmental law decisions and permits.
Figure 5: Reduction of dust emission in Gogw smelter
Figure 6: Reduction of SO
2
emission in Gogw smelter
14.442
2.017
125
122 136 134 179
177 130
34
0
2000
4000
6000
8000
10000
12000
14000
1980 1990 2001 2002 2003 2004 2005 2006 2007 2008

125,7
34,1
4,2 4,3 3,8
3,5 4,0 4,0 3,9
4,4
0
20
40
60
80
100
120
1980 2001 2003 2005 2007

6 Assumptions and directions of the development of metallurgy
in KGHM included in the strategy for the period 2008-2020
KGHMs currently implemented strategy is focused on the efficient processing of natural resources,
which should lead to an increase in the value of the company.
The strategy covers 5 guiding areas:
improvement of activity;
development of the resource base;
diversification of income sources and independence of energy prices;
support for the region;
organization skill and efficiency development.
Modernization of the Gogw II smelter in the years 2009-2013, and of Gogw I foreseen for the
year 2014, are among the most important elements in achieving the first area of KGHMs strategy.
The aim of the modernization of Gogw II is to implement changes in its technical infrastructure to
achieve annual capacity of 230 thousand tonnes of blister copper. The purpose of construction of a
new metallurgical complex at Gogw I with a capacity of 250 thousand tonnes is to lower substan-
tially the cost of processing.
Production volume depends on the availability of native copper concentrates as well as complemen-
tary processing of imported concentrates and scrap.
7 Modernization of the Gogw II Copper Smelter
7.1 Technical-technological limitations of copper production
Apart from the renewal of the technical infrastructure, the other direct cause of the decision to mod-
ernize the Gogw II copper smelter was the need to eliminate the following drawbacks:
Lack of concentrate storage;
Limited capacity of concentrate dryer;
Limited capacity of the flash smelting furnace;
Limited technological conditions for melting the concentrate mixture in the 4-burner system;
Limited reception of heat by the reaction shaft;
Limitations of FSF dust treatment;
WHB and the gas ducts build-up;
Constraints on the gas draft control;
Insufficient rate of furnace hearth cooling
Limited capacity of SAP.
Due to the need to resolve the above-mentioned issues, it was found necessary to implement
changes in the following areas:
Drying of concentrate.
Flash Smelting Furnace and related facilities.
Waste Heat Boiler.
Dust Treatment System.
Electric Furnace.
Sulphuric Acid Plant.
The occurrence of the current technological problems such as FSF energy balance, WHB build-up,
the limited domestic supply of concentrates and the high cost of all planned modernization has
forced the division of this task into two stages:
Stage I 2009, which aims to increase the availability of steel mills and change the heat balance of
the process, allowing concentrates mixture smelting with a higher-heat value than those presently
treated:
Stage II 2013, modernization of the sulfuric acid plant, enabling a production capacity to reach
230 thousand tons of blister copper per year and processing of sulfur at 16.4 ton/h.
7.2 Concentrate storage and drying facilities
Concentrate storage construction (planned realization year 2011 50 thousand tonnes) will
definitely improve the environmental conditions of concentrate storage and handling; the concen-
trate so far has been stored in an open landfill. The proposed system is an indoor warehouse with
selective off-take of the stored concentrates.
The higher planned level of FSF production approximately 230 thousand tonnes of blister cop-
per - and the poor operating results of the concentrate dryer to date, required its extensive mod-
ernization. The screening and crushing systems are also designated for modernization.
7.3 Flash Smelting Furnace
The modernization included the following elements:
Furnace Loading.
Concentrate burner.
Reaction Shaft.
Surrounding furnace facilities:
Furnace heating installation,
Hearth cooling installation,
Technological blow installation
Auxiliary materials loading installation.
7.4 Waste Heat Boiler - dust treatment
The scope of modernization works covers the construction of the baffle, as well as the oxidizing and
sulphatizing nozzles for dust treatment. The latter resulted in the oxidizing and sulphatizing of the
dusts. The dust treatment process was further improved by the introduction of a modified crusher
system. The dust transportation installation was expanded by a section for recycling the dust from
reserve landfills.
7.5 Electric Furnace
The scope of repair and modernization works concerning the electric furnace consisted mostly of
rebuilding its technical capacity to ensure the 4-year campaign between its repair shut-downs. The
main tasks of the modernization were to increase the efficiency of the hearth cooling installation and
further modification of the cooling system of the Electric Furnace walls.
7.6 Sulphuric Acid Plant
Two additional wet electrostatic precipitators were installed, which resulted in elimination of the
need to reduce the feed rate to the FF while rinsing the currently installed ones. Other tasks related
to upgrading SAP aimed at increasing the level of sulphur utilization to 16.4 t/h are foreseen for the
second stage of modernization in 2013.
7.7 Modernization of the Gogw I Smelter
In analyzing the process of modernization of the Gogw I smelter the following issues were em-
phasized:
the advantages and disadvantages of the shaft furnace process,
the possibility of applying available pyrometallurgical technologies,
the advantages and disadvantages of the flash smelting process currently applied by Gogw II,
the specific properties of domestic copper concentrates, with particular regard to their content of
Corg and Pb
the need to maintain an optimal level of copper production during the change of technology in
2014.
7.8 Advantages and disadvantages of the shaft smelting process
Advantages
1 High flexibility of the process in relation to the quality of concentrates;
2 High yield of copper and associated metals;
3 Precipitation of lead and zinc in concentrated PbS-ZnS form, from the stream of copper;
4 Possibility of melting of revert materials and contaminated charges;
5 Low cost of production facilities in the fire section;
6 Short shut-down period on the process line.
Disadvantages
1 Necessity of using waste black liquor;
2 Complicated and energy-consuming transport of input materials;
3 Expensive concentrate briquetting;
4 Necessity of using coke as a fuel;
5 Low energy efficiency of the shaft furnaces;
6 Generation of a large volume of lean process gases requiring combustion in the boiler and desul-
phurization of the flue gas;
7 An extremely difficult technological gases purification process with dusty and condensed frac-
tion;
8 Expansion installation of secondary gas extraction;
9 Relatively high transfer of pollutants into the atmosphere while transporting the phase products:
copper matte, slag, dust and reverts.
7.9 Pyrometallurgical technologies by % share of the global output
Outokumpu: 51 %
Noranda + Teniente: 16 %
Mitsubishi: 8 %
ISA: 6 %
Shaft Furnace: 4 %
Inco: 3 %
Other: 14 %
7.10 Advantages and disadvantages of the Flash Smelting Process
Advantages
1 Large flexibility of melting batch size;
2 Simple and economical process of concentrate drying and handling;
3 Concentration of chemical reactions in a single aggregate;
4 High efficiency of heat recovery from process gases in the waste heat boiler;
5 High concentration of SO
2
in the process gases;
6 High sulphur yield.
Disadvantages
1 Expensive slag cleaning process;
2 Short period between the repair campaigns (48 months);
3 Long repair period (50-60 days).
7.11 The specificity of Polish copper concentrates
Specific properties of Polish concentrates:
Relatively low iron content of 3.5-5 %;
High content of Pb and other metals 0.9-2.4 %;
Low sulfur content, 8-11 %
High organic carbon content of 5.5-8.3 %
Average copper content ranging from 17 to 30 %;
As demonstrated above, the high organic carbon content and the expected high level of blister cop-
per production projected in Gogw I and Gogw II has generated an urgent need to find solutions
balancing the amount of heat in both flash smelting furnaces.
Implementation of one of the following options will solve the problem:
change of concentrate flotation technology,
selection of an optimal batch mix (oxide concentrates),
use of heat-absorbing material in the reaction zone,
roasting of the concentrate,
recycling of process gases or slag,
loading of solid charge into the FSF settler.
Analysis is currently being performed regarding selections of the best possible technological direc-
tion for effectively eliminating the impact of Corg content on the projected higher FSF smelting
rates.
7.12 Optimum choice
As a result of analysis of the options listed above, the decision was made to introduce the one stage
copper smelting technology in the FSF at Gogw I. This technology differs from the technology of
Gogw II in the following areas:
Different ways of charge mixture blending nave system
Application of concentrate steam dryer
Foundations of the flash furnace dependent on the manner of slag evacuation from the electric
furnace
Introduction of the FSF charge additive in the form of slag from the lead smelting process
Application of a rectangular electric furnace equipped with and powered by 6 electrodes
(2 x 25 MVA).
Production of aggregates from slag arising from the electric furnace.
Substitution of dripping type EF cooling system by a sandwiched one.
Application of the upgraded technology of refining the Cu-Pb-Fe alloy.
7.13 Contingencies/Considerations in the planning process of
modernization tasks
In considering the concept of adding equipment to the modernized technological line, the need to
address the following factors was taken into account:
optimal link of the new facilities to the existing equipment of the production line (converters,
anode furnaces)
reduction of the total shut-down time of Gogw I
maximal utilization of the energy infrastructure to power the new facilities.
8 The Gogw I modernized technological stages
Figure 7: Gogw I Smelter plants to be modernized

8.1 Concentrate homogenizing and pretreatment
The Charge Preparation and Processing Department will be equipped with a steam dryer with a de-
dusting system, auxiliary materials gas drying system, a gallery transporting concentrates to the
steam dryer and a gallery of materials and auxiliary reverts.
The Homogenizing Warehouse is equipped with devices for improving the effect of homogenizing.
Concentrate unloading and part of the concentrate handling gallery will remain unchanged.
Raw materials Storage will be adapted to the storage and unloading of the target auxiliary materials,
reverts and concentrates.
8.2 Flash Furnace
The new flash furnace will be similar in design to that existing in Gogw II, nevertheless, the tech-
nological experience gained during operation of the latter will be used. Its designed annual capacity
is approximately 250 thousand tonnes of blister copper. The most significant difference between the
two units, apart from performance, will be the higher level position of the furnace hearth.
Figure 8: General dimensions and levels of a new Flash Furnace

8.3 Electric Furnace
The building of a new electric furnace, which will be similar to the flash furnace in terms of con-
struction and size, is planned. The furnace will have a rectangular cross section with an arrangement
of six electrodes (2 x 25 MVA).
From the waste slag formed in the furnace the road aggregates will be manufactured, and the Cu-Pb-
Fe alloy with lead content up to 14 % will be converted in 5 upgraded converters. Slag from the
slagging stage will be returned to the electric furnace, and slag from the copper stage will be re-
duced in a separate converter to obtain copper-and-lead-contaminated slag with high-lead and low-
copper content. The latter will be a charge material into the rotary-tilting furnaces for crude lead
production based on innovative technology.
Figure 9: Location of Flash and Electric Furnace
8.4 Sulphuric Acid Plant
The Sulphuric Acid Plant will be equipped with new production facilities, as the presently operated
washing-cooling section and wet electrostatic precipitators section were added in recent years as
elements of the new target production line.

8.5 Energy management of upgraded Gogw I Copper Smelter
In terms of the modernization of power equipment, the following investments are planned:
modernization of the Gogw I main transformer station construction of an Electric Furnace
power supply installation,
construction of a steam utilization station and reductive turbines with a total power of about 2.5
MW,
implementation of a media distribution network: oxygen, nitrogen, natural gas, fuel oil, com-
pressed air, de-ionized water,
implementation of new and modernization of existing metallurgical units cooling installations.
9 The proposed concept of lead balancing in upgraded
Gogw Smelters (two FF)
The properties of domestic copper concentrates, especially their high lead content, required a new
approach to the "closure" of its balance in the upgraded Gogw production line with two flash fur-
naces constructed. The increased possibility of lead production by the construction of an additional
rotary-tilting furnace forced the detailed analysis of the available technologies.
At the moment the innovative method developed by KGHMs engineers, solving the problem of the
treatment of copper concentrate with enhanced lead content, is being implemented. The solution
consists of an additional treatment of converter slag, enabling its applicability to further metallurgi-
cal processing.
The slag prepared in this way will become input material for the rotary-tilting furnaces as raw mate-
rial for the production of lead.
Increasing the capacity of the Doerschel furnaces will not only be used for the current processing of
the lead-bearing materials available but also for successive elimination of its accumulated stocks.


Energy Consumption in Copper Sulphide Smelting
P. Coursol, P. J. Mackey C. M. Daz
Xstrata Process Support Centre University of Toronto
Department of Materials Science and Engineering
6 Edison Road 210 Radley Road
Falconbridge, Canada, Ontario, P0M 1S0 Mississauga, Canada, Ontario, L5G 2R7

Keywords: Copper, smelting, converting, slag cleaning, fire refining, energy, emissions
Abstract
Using thermochemical modeling and industrial data, energy consumption in copper sulphide con-
centrate smelting was calculated for the following processing routes: a) flash smelting + flash con-
verting; b) Isasmelt smelting + Peirce-Smith (PS) converting; c) Mitsubishi continuous copper
smelting and converting; and d) Noranda/Teniente continuous bath smelting + PS converting. For
all routes, the battery limits of the exercise included value metals recovery from slag, heat recovery
from process gas streams, and abatement of SO
2
from both process gas streams and secondary
streams. A unit operation approach was adopted in this study, thus permitting rapid evaluation of
other flowsheets such as flash smelting + PS converting. Although natural gas was the main fuel
used in the computations, the model can be readily used to explore the impact of alternative fuels.
1 Introduction
It is more than 30 years since H. H. Kellogg and J. M. Henderson published their seminal paper on
Energy Use in Sulfide Smelting of Copper [1]. These authors not only calculated the direct fuel
and electrical energy consumption for nine different pyrometallurgical routes, but also evaluated the
energy required to produce main process supplies, thus pioneering copper life cycle assessment.
Since the publication of this classic study, the practice of anode copper production from sulphide
concentrates has experienced many changes. The reverb furnace, that was the dominant copper
smelting technology in the 1970s, has been replaced by far more energy efficient and environmen-
tally sound flash and bath smelting technologies. Top submerged lance smelting has become an im-
portant processing route in recent years. Continuous converting is currently challenging the one-
hundred-year old Peirce-Smith converter. Tonnage oxygen consumption in both smelting and con-
verting has substantially increased, thus fulfilling Paul Queneaus dream of lifting the dead hands
of nitrogen from oxidation reactions which utilize the oxygen in air [2]. Throughout the world,
increasingly more stringent environmental regulations are being imposed on the industry, and emis-
Coursol, Mackey, Daz
sions of greenhouse gases are being examined in depth as part of the worldwide strategy to slow
down climate change.
The present authors considered, then, that it was time to revisit the subject by examining some of
the more energy efficient and environmentally acceptable routes currently used for the production of
anode copper from sulphide concentrates. Toward this end, they selected the following four flow-
sheets: (a) flash smelting + flash converting; (b) Isasmelt
TM
smelting + Peirce-Smith (PS) convert-
ing; (c) Mitsubishi continuous copper smelting and converting; and (d) Noranda/Teniente conti-
nuous bath smelting + Peirce-Smith converting. In this study, the authors adopted the same unit
operation approach originally used by Kellogg and Henderson. This permits rapid evaluation of oth-
er flowsheets such as flash smelting + PS converting.
The objective of the exercise was to provide answers to questions such as: what has been the net
energy payout by substituting new technologies for reverb smelting? What has been the correspond-
ing decrease in CO
2
emissions? How big a price is industry paying in its quest for sustainability?
Which are the unit operations and secondary materials management techniques that offer further
significant energy savings?
Due to time constraints, the present authors have postponed undertaking the more ambitious goal of
determining the copper life cycle assessment from concentrate to anode copper. They will deal with
this matter in a future paper.
2 Methodology
The methodology adopted to develop smelting energy data in this study was based in part on the
approach used by Kellogg and Henderson [1]. A standardized flowsheet was developed for each of
the leading smelting process routes that were selected. A Metsim model previously developed by
Tripathi et al. was used as a basis for designing four new models for this study [3]. Heat and mass
balance, and energy consumption data were then computed for each process for subsequent analysis
and comparison. Process boundary limits were: Inputs - wet concentrate, flux and other consu-
mables delivered to the smelter day bins; and Outputs - copper anodes, sulphuric acid, acid plant tail
gas and cleaned fugitive gas released to atmosphere, and cleaned slag. Waste heat recovery from
process gas streams was also included as part of the study. As well, a literature review for previous
relevant information was conducted.
The four processing routes examined in this study are listed below. The respective flowsheets are
presented in the next section.
1. Outokumpu Flash Smelting + Kennecott-Outokumpu Flash Converting
2. Isasmelt Smelting/RHF Matte/Slag Settler + PS Converting
3. Mitsubishi Continuous Smelting Process
4. Noranda/Teniente Continuous Bath Smelting + PS Converting
Energy consumption in auxiliary unit operations and energy equivalent for process supplies were
computed from a set of unit energy consumption factors. These are presented in Table 1 below.
Table 1: Unit energy parameters used in this study
Item Unit energy Ref.
Steam dryer 2 tonne steam/tonne water evaporated Authors
Conversion of steam to electricity at smelter 6.25 kg of steam/kWh* Authors
Tonnage oxygen
(production + gaseous surge capacity)
285 kWh/tonne of oxygen [14]
Compress tuyere air (~110 kPa) 0.05 kWh/Nm
3
[1]
Compress lance air (~ 60 kPa) 0.03 kWh/Nm
3
Authors
Process off-gases handling 0.0085 kWh/Nm
3
Authors
Fan-secondary gases 0.002 kWh/Nm
3
[23]
Furnace cooling water 3 kWh/tonne Cu Authors
Matte granulation and handling 9 kWh/tonne of matte Authors
Slag granulation and handling 3 kWh/tonne of slag Authors
Matte comminution and handling 10 kWh/tonne of matte Authors
Lighting and misc power (allowance) 30 kWh/tonne of Cu [1]
Acid plant operation (double contact) [(646.8/% SO
2
) + 63.7] kWh/tonne of acid [1]
Energy - Flux (90 MJ + 3kWh)/tonne of flux [1]
Energy - Limestone calcination (CaO flux) 7000 MJ/tonne of CaO Authors
Energy - Wear steel in slag milling 20.7 MJ/kg of steel [21]
Energy - Pig iron 15.5 MJ/kg of pig iron Authors
Notes: * Based on a rate of 5 kg steam/kWh and an operational efficiency of 80 % to account for potential
losses on start-up/standby etc.
Appropriate data was taken from Kellogg and Henderson, and updated information was used in oth-
er cases [1]. Examples of auxiliary unit operations are: producing tonnage oxygen, compressing PS
converter injection air, delivering low-pressure air to burners, moving process off-gas, drying con-
centrate and other materials, transporting and injecting fine solids suspended in a stream of gas
(dense phase transportation of solid particulates).
3 Brief description of processes
Flowsheets of the four selected processes are presented in Figures 1 to 4. Specific references are
given in each brief process description. Data from the 2003 worldwide copper smelter survey was
used to fill gaps in these flowsheets [4]. Each process route also included the following standard
unit operations: (a) complete secondary gas collection and cleaning; (b) anode refining and casting;
and (c) process gas treatment in a double-contact acid plant with acid delivery to storage tanks. As
noted above, heat recovery from process off-gases was also considered. Concentrate and other solid
process streams requiring drying were treated in steam dryers using waste heat steam. Surplus steam
was assumed to generate electricity.
3.1 Outokumpu flash smelting + Kennecott-Outokumpu flash
converting
The Utah copper smelter flowsheet was used for developing the model for this process route (see
Figure 1 below) [5].
Figure 1: Outokumpu flash smelting + Kennecott-Outokumpu flash converting
Dry concentrate is fed to the flash furnace. The product matte (~70 % Cu) is granulated, ground and
dried. It is, then, fed to the flash converter to produce ~0.2 % S blister. A lime-ferrite slag, compati-
ble with the high oxygen potential required to produce relatively low-sulfur copper, is used in the
converter. This slag is granulated and returned to the flash furnace. In turn, the slag from the flash

furnace is controlled cooled and milled for copper recovery. The slag concentrate is also returned to
the flash furnace. Except for minor proportions, dust is recycled to the respective generating fur-
nace.
3.2 Isasmelt smelting + PS converting
Isasmelt is one of the two Top Submerged Lance (TSL) technologies currently practiced for cop-
per concentrate smelting. It was chosen for this exercise, because it has demonstrated annual
throughputs over one million tonnes of concentrate in various locations. The authors model was
built based on published flowsheet information on the modernization of the Ilo Smelter (Figure 2
below) [6-8].
Figure 2: Isasmelt smelting + Peirce-Smith converting
An agglomerated mixture of concentrate, flux, recycled dust and coal is fed to the vertical stationary
reactor, and oxygen-enriched air and supplementary gaseous or liquid fuel are injected through the
vertical lance. The product matte (60-65 % Cu) and slag are jointly transferred to a Rotary Holding
Furnace (RHF) settler. An electric furnace is used in some Isasmelt Smelters for matte/slag separa-
tion. The slag from the settler, usually containing <1 % Cu, is discarded. The matte is converted to
blister in PS converters. Copper is recovered as matte from the converter slag in a cylindrical-
horizontal tilting vessel, using pig iron as reductant. In the authors view, converter slag milling
could be considered as an alternative.

3.3 Mitsubishi continuous smelting process
The authors model is based on the Mitsubishi three-furnace operation as described by M. Goto
et al. (see Figure 3 below) [9].
In the smelting furnace, dry concentrate and other feed materials, including dust from the smelting
and converting furnaces, are injected directly into the matte in a stream of oxygen-enriched air via
vertical lances. Matte (68 % Cu) and slag flow jointly and continuously into an electric furnace sett-
ler. Molten matte is, in turn, continuously transferred from the settler to the converting furnace,
while the slag (~0.7 % Cu) is discarded. In the converting furnace, limestone flux is injected into the
molten bath in a stream of oxygen-enriched air also via vertical lances. Blister copper (0.75 % S)
and lime-ferrite slag, each at constant composition, coexist in the converting furnace. The two melts
are continuously and separately discharged from the converter. Granulated converter slag is recycled
to the smelting furnace.
Figure 3: Mitsubishi copper continuous smelting

3.4 Noranda/Teniente continuous bath smelting + PS converting
The Altonorte Smelter flowsheet was used as a basis for developing the model for this processing
route (see Figure 4 below) [10].
The smelting unit is a Noranda reactor-Teniente converter hybrid. Most of the concentrate is in-
jected dry through tuyeres into the molten bath. Wet concentrate, flux, coke and coarse internal
smelter reverts are fed to the reactor via a Garr-Gun. Dust is recycled to the reactor. The product
matte (~74 % Cu) is converted in PS converters. The reactor slag is controlled cooled and milled for
copper recovery, while solid converter slag is recycled to the reactor in the present case.
Figure 4: Noranda/Teniente continuous bath smelting + Peirce-Smith converting
4 Calculation of process energy requirements
4.1 Smelter throughput
As noted above, the four models developed in the present work used the initial model by Tripathi et
al. as a base [3]. Consistent with the trend towards higher smelter capacity, high tonnage plants were
considered in the present work as shown in Table 2 overleaf [4].

Table 2: Annual copper concentrate throughputs
Process route Technology Copper concentrate throughput,
million tonnes/year
1 Outokumpu 1.2
2 Isasmelt 1.2
3 Mitsubishi 0.75
4 Noranda/Teniente 0.85
Note: These tonnages are based on current proven industrial performance and adapted to handle this study
standard concentrate (see Table 3); 85 % smelter on-stream time or 7446 h/year was also assumed.
Specific furnace dimensions are found in the appropriate references.
4.2 Standard Conditions
The standard conditions that were used for the calculation of energy requirements for each of the
chosen processes are presented in Table 3 below. These include the assay of a standard copper
concentrate, and data on fluxes and fuels.
Table 3: Standard conditions used in the modified Metsim model
Item and data
Concentrate analysis (dry basis): 30.5 % Cu, 28.5 % Fe, 31.5 % S, 5 % SiO
2
, 2 % Al
2
O
3

Concentrate moisture content: 10 % H
2
O (Bone-dry feed charged to Outokumpu, Mitsubishi and Noran-
da/Teniente furnaces; wet feed charged to Isasmelt furnace)
Flux analysis (dry basis): 88 % SiO
2
, 2 % CaO, 6 % Al
2
O
3
, 2 % MgO and 2 % Fe
3
O
4

Flux moisture: 3 % H
2
O
Natural gas: 37.3 MJ/Nm
3

Coal: 28.4 MJ/Kg
Ambient conditions: 0 C, 760 mm Hg
4.3 Peirce-Smith converting
In this study, Peirce-Smith converting is used in two of the four flowsheets. It should be noted that
in most smelters that practice flash smelting, the product matte is also treated in PS converters. The
Metsim converter model developed here was based on standard converter practice, and for the ton-
nages used in the study a typical 4.6 m diameter by 13.4 m long vessel was considered. Unit energy
factors used for converting Isasmelt matte are presented in Table 4 below. Data for the Noran-
da/Teniente matte were developed in a similar way.
Table 4: Energy required for PS converting (Isasmelt matte)
Item Amount Unit energy
Electrical power
kWh/tonne Cu
Fossil fuel
MJ/tonne Cu
Compressed air 1419 Nm
3
/t Cu 0.05 kWh/Nm
3
71.0
Process off-gas han-
dling
3520 Nm
3
/t Cu 0.0085 kWh/Nm
3
29.9
Fuel for pre-heating
and standby
Estimated by au-
thors
385.2
Flux preparation 0.19 t flux/t Cu (90 MJ + 3 kWh)/t flux 0.6 17.1
Total before steam
credit
101.5 402.3
Steam production
12.5 t/h. It is noted that smelter and converter steam was used for drying, surplus smel-
ter steam was used to produce electricity
The above energy numbers compare well with those reported by Kellogg and Henderson in their
paper [1].
4.4 Anode refining and casting
This is a common unit operation for the four process routes covered in this study. A nominal
250 tonne capacity anode furnace was used in this study. A complete furnace cycle consists of load-
ing blister, oxidation, reduction and casting of anodes. The length of the oxidation time per tonne of
blister produced by PS converting is shorter than for the higher sulphur-containing blister produced
in both the Mitsubishi and the Kennecott Outokumpu converters. The length of the reduction time
per tonne of copper was assumed to be the same for the four process routes. The energy require-
ments for refining low-sulfur blister and anode casting per tonne of anode are presented in Table 5
overleaf.
Table 5: Energy requirements for anode refining and casting
Item Amount Unit energy
Electric energy
kWh/tonne of Cu
Fossil fuel
MJ/tonne Cu
Furnace heating
fuel (natural gas)
23.8 Nm
3
/tonne Cu 37.3 MJ/Nm
3
888
Natural gas for
poling
6 Nm
3
/tonne of Cu 37.3 MJ/Nm
3
224
Compressed air
and misc energy
Estimated by authors 14.7
Process off gas
handling
1494 Nm
3
/t Cu
0.0085
kWh/Nm
3

12.7
Total 27.4 1112
4.5 Slag cleaning
Milling is used for copper recovery from slags in two cases (Outokumpu flash smelting and Noran-
da/El Teniente bath smelting). An electric furnace (settler) is used for separating matte from slag in
Mitsubishi continuous smelting, while a Rotary Holding Furnace (RHF) is used in the Isasmelt
flowsheet for similar purposes. Finally, a cylindrical-horizontal-tilting furnace is used for converter
slag reduction with pig iron and copper recovery as matte in the latter process route. Unit energy
factors for each of these operations are summarized in Table 6 below.
Table 6: Equivalent energy requirements for slag treatment per tonne of slag
Item Slag milling
Converter slag
cleaning (Isa)
Slag/Matte settler
El. Fce. (Mitsubishi) RHF (Isasmelt)
Electric energy 66 kWh 45 kWh
Grinding media 0.78 kg steel
Electrode carbon 0.5 kg
Coke breeze 8 kg
Natural gas (heating) 28.3 Nm
3
15.8 Nm
3

Pig iron 23 kg
Miscellaneous power 6 kWh 6 kWh
Off gas handling 1197 Nm
3
666 Nm
3

Note: Data based on present model and additional information from Kellogg & Henderson [1]
4.6 Secondary and fugitive gases handling
A modern copper smelter includes facilities for collecting and handling secondary and fugitive gas-
es. Examples of recent installations can be found in papers by Toyo authors and by Boliden authors
[11, 12]. Based in part on these references, the present authors estimated volumes of unit secondary
and fugitive gases that are approximately linked to the number of furnaces employed in a process
routes. These unit volumes are estimated at 23,000 Nm
3
/tonne of copper for process routes (1) and
(3), and 38,000 Nm
3
/tonne of copper for process routes (2) and (4). The unit electrical power re-
quired to handle these gases is given in Table 1.
4.7 Process operating parameters
As noted, concentrate throughput and model operating parameters for each of the four processing
routes were based on published information [4-10]. Private communication with industry was used
to confirm or modify data. The values of these parameters were adapted for the standard concentrate
used in this study. Due to space constraints, only key process parameters will be given in succeeding
paragraphs. The reader is encouraged to check this paper references for further detail.
4.7.1 Outokumpu flash smelting + Kennecott-Outokumpu flash converting
A flowsheet of this process is presented in Figure 1. Operating parameters are based on data re-
ported in the 2003 Copper Smelter Survey and D. George et al.s paper on the modernization of the
Utah Smelter [4,5]. Following are the values of key parameters:
Smelting: (a) Feed rate - 161 tonnes dry concentrate/h; (b) Matte - Grade 70 % Cu, furnace temp.
1324 C; (c) Slag - Fe/SiO
2
1.28, 2.5 % Cu.
Converting: (a) Blister - 0.2 % S, furnace temperature 1270 C; (b) Slag 15 % CaO, 22 % Cu.
4.7.2 Isasmelt smelting + RHF matte/slag settler + PS converting
The Isasmelt process flowsheet presented in Figure 2, as well as dimensions and capacities of the
various furnaces, are based on published information on the modernization of the Ilo Smelter in
Peru [6-8]. Private communication with smelter personnel assisted in further developing operating
parameters [13]. Following are the values of key parameters:
Smelting: (a) Feed rate - 161 tonnes dry concentrate/h; (b) Matte - Grade: 64 % Cu, furnace
temp. 1185 C; (c) Slag - Fe/SiO
2
wt ratio 1.4, 0.8 % Cu (discharged from RHF).
Converting: (a) Slag - Fe/SiO
2
wt ratio 1.5, 4.4 % Cu.
4.7.3 Mitsubishi continuous smelting process
The flowsheet of this process route is presented in Figure 3. Operating data for the four smelters
currently practicing the Mitsubishi process are found in the 2003 Copper Smelter Survey [4]. In
addition, the metallurgy and commercial practice of this process are fully discussed in a Mitsubishi
monograph first published in 1987, and successively revised in 1998 and 2002 [9]. Following are
the values of key parameters:
Smelting: (a) Feed rate 100.7 tonnes dry concentrate/h; (b) Matte - Grade 70.1 % Cu, temp.
2
wt ratio 1.28, 0.8 % Cu (discard).
Converting: (a) Blister 0.75 % S, temp. 1250 C; (b) Slag 15 % CaO, 15 % Cu.
4.7.4 Noranda/Teniente continuous bath smelting + PS converting
As earlier discussed, the flowsheet presented in Figure 4 is based on the operation of the Altonorte
Smelter in Chile. Features and dimensions of the reactor, a hybrid Noranda reactor/Teniente conver-
ter, and other key components of the smelter flowsheet are discussed elsewhere [10]. Operating data
are presented in the 2003 Copper Smelter Survey [4]. In addition, two of the present authors have
had opportunity to familiarize with the operation of the Altonorte Smelter. Following are the values
of key parameters:
Smelting: (a) Feed rate 114.2 tonnes dry concentrate/h; (b) Matte - Grade 70 % Cu, temp.
2
wt ratio 1.4, 4.5 % Cu.
Converting: (a) Slag - Fe/SiO
2
wt ratio 1.5, 6.8 % Cu.
5 Results
The results of model calculations for process routes 1 to 4 are presented respectively in Tables 7 to
10. Separate entrances for electric energy, expressed in kWh/tonne of anode copper, and fuel, ex-
pressed in MJ/tonne of anode copper, are provided in these tables. The fuel equivalent of electric
energy has been calculated using a power plant efficiency of 38 % [15, 16]. In the tables, numbers
for items such as fuel, oxygen, compressed air, secondary and fugitive gases correspond to respec-
tively overall smelter consumption or production.
Table 7: Energy requirements for producing anode copper from concentrate using Outokumpu
Flash smelting + Kennecott-Outokumpu flash converting (Process 1)
Item
Electric energy/tonne anode Fuel MJ/tonne
anode
Total MJ/tonne
anode
kWh Equivalent MJ
Fossil fuel 1479 1479
Oxygen 312 2957 2957
High pressure air 10 95 95
Low pressure blowing air 16.3 154 154
Burner air & misc. 3.7 35 35
Process gas handling 27 256 256
Sec/Fugitive gas handling 46 436 436
Matte/Slag granulation & hand. 53 504 504
Supplies (flux, etc.) 2.1 20 39 59
Slag milling 159 1503 1503
Acid production 340 3216 3216
Lighting & miscellaneous 30 284 284
Steam (power) credit -20.5 -194 -194
Total 979 9266 1518 10,784

Table 8: Energy requirements for producing anode copper from concentrate using Isasmelt
Smelting/RHF Matte/Slag settler + PS converting (Process 2)
Item
anode
Total MJ/tonne
anode
kWh Equivalent MJ
Oxygen 227 2152 2152
Low pressure blowing air 33.4 317 317
Burner air & miscellaneous 18.7 177 177
Supplies (flux, steel, etc.) 1.8 17 213 230
Steam credit -154 -1456 -1456
Total 729 6903 4175 11,078
Table 9: Energy requirements for producing anode copper from concentrate using Mitsubishi
Continuous Smelting and Converting (Process 3)
Item
anode
Total MJ/tonne
anode
kWh Equivalent MJ
Oxygen 220 2081 2081
Low pressure blowing air 77 727 727
Burner air & miscellaneous 7.9 75 75
Slag granulation & handling 18.3 173 173
Supplies (flux, steel, electric
furnace electrode, etc.)
22 208 67 276
Electric fce. matte/slag settler 103 972 972
Steam (power) credit -28.7 -272 -272
Total 898 8508 2498 11,006
Table 10: Energy requirements for producing anode copper from concentrate using Noranda/El
Teniente bath smelting + PS converting (Process 4)
Item
anode
Total MJ/tonne
anode
kWh Equivalent MJ
Oxygen 167 1578 1578
Burner air & misc. 6.7 64 64
Sec/Fug gas handling 76 720 720
Supplies (flux, steel, etc.) 1.7 16 81 97
Slag milling 160 1519 1519
Steam credit -33.9 -321 -321
Total 1065 10,088 2658 12,746
6 Discussion
6.1 Energy consumption in copper smelting
6.1.1 This study vs. Kellogg and Henderson
The results of the calculated smelting energy requirements by process are summarized in Table 11
overleaf. Also shown in this table are energy requirements for selected processing routes covered in
the Kellogg and Henderson paper [1]. Separate entrances for electric energy and fossil fuel are
provided in Table 11 to facilitate the discussion.
Interestingly, the numbers in Table 11 show that three of the four processes investigated in this
study have similar energy requirements. In fact, the difference in energy consumption between
processes is within the margin of error of assumptions plus model calculations. There is, though,
quite a variation in relative amounts of electric energy and fossil fuel requirements, with process 2
having the lowest electric energy consumption, but a fossil fuel requirement almost 3 times higher
than process 1. The energy requirement of process 4, in particular the electric energy portion, is sig-
nificantly higher than for any of the other three processes; process 4 has the largest consumption of
high-pressure injection air and the highest acid making energy requirement. The data in Tables 7-10
help to explain other differences between processes.

Table 11: Energy requirements for producing anode copper from concentrate (MJ/tonne of anode)
Processes 1-4 and selected Kellogg and Hendersons processes [1]
Processing route Electric energy Fossil fuel Total
Process 1 (Flash-Flash) 9266 1518 10,784
Process 2 (Isasmelt) 6903 4175 11,078
Process 3 (Mitsubishi) 8508 2498 11,006
Process 4 (Noranda-Teniente) 10,088 2657 12,746
KH-Hot Calcine Reverb 2173 15,935 18,108
KH-Outokumpu Flash 7477 6760 14,237
KH-Mitsubishi 6904 9306 16,210
KH-Noranda 9045 5220 14,265
Notes on KHs numbers:
Power plant efficiency of 32.5 % used to calculate the fuel equivalent of electric energy.
WH recovered from all process gases. Steam raised in the WH boilers mostly used to generate power
that it is then credited to the electric energy required.
SO
2
recovered as acid from all process gases, except for reverb gas.
Flash furnace operated with 30 % O
2
enriched air. Matte grade 61.4 % Cu.
Noranda Reactor operated with 36 % O
2
enriched air. Matte grade 75 % Cu.
A comparison of the energy requirements of modern processes with those studied by Kellogg and
Henderson indicates that advances in technology and increased oxygen consumption have more than
compensated for increased energy requirements to meet more stringent environmental standards [1].
For instance, while in the Kellog and Henderson exercise process gases were treated in single-
contact acid plant, double-contact technology was used in the present study. The handling of sec-
ondary and fugitive gases, not an issue in the late 1970s, was also incorporated in this investigation.
On the other hand, a substantially lower energy requirement for producing tonnage oxygen
(285 vs. 397 kWh/tonne) and a significantly higher power plant efficiency (38 vs. 32.5 %) were
used in this study, a reflection that in addition to copper smelting other relevant technologies have
experienced substantial advances since the Kellogg and Henderson study was published. The latter
factors explain that, despite the environmentally driven demands on modern smelter operation and a
substantial increase in oxygen consumption, their average electric energy requirement (processes 1
to 3) is not substantially higher than for the three Kellogg and Henderson processes listed at the bot-
tom in Table 13. What it is truly different is the much lower consumption of fossil fuel in modern
smelters. Factors such as improved furnace design and operating practices, computerized process
control, higher furnace feed throughput have certainly contributed to these achievements.

6.1.2 Greenhouse gas emissions
As shown in Table 12 below, direct (heating fuel) and indirect (burning fuel for power generation)
CO
2
emissions are highly dependant on the type of fuel used for each purpose [15, 16].
This study has shown that most of the energy requirement in modern copper smelting processes
consists of electric energy. It is, then, the particular fuel mix used for the generation of power in
regions with a large concentration of copper smelters that is currently determining the carbon foot-
print of copper sulphide smelting.
Table 12: Direct (heating fuel) and indirect (power generation) CO
2
emissions for various types of fuel.
Type of fuel Direct CO
2
emission (tonne/MWh) Indirect CO
2
emission* (tonne/MWh)
Natural gas 0.20 0.53
Light oil 0.26 0.68
Heavy oil 0.28 0.74
Anthracite 0.33 0.87
Lignite 0.40 1.05
Assuming power plant generation efficiency of 38 %.
Using these numbers and data presented in Tables 7-10, and assuming natural gas for both direct
smelter heating and for electricity generation, the calculated average total CO
2
emissions for the
four smelting processes studied here, is 0.63 tonnes of CO
2
/tonne of anode copper. Direct heating
accounts for 24 % of these emissions and 76 % is due to power generation
Detailed data on greenhouse gases emitted by the seven Chilean smelters have been recently pub-
lished [17, 18]. The numbers are reported as CO
2
equivalent. Two of these smelters are connected to
the Northern Interconnected Power Generating System (SING), and the other five to the Central
Interconnected System (SIC). In 2007, 99.6 % of SING energy was generated from fossil fuel
(57.6 % coal, 22.6 % NG, 16.4 %, 3 % fuel oil), while SIC energy mix consisted of 47.1 % thermal
(less than one third of this was coal generated) and 52.9 % hydro. Not surprisingly, the CO
2
equiva-
lent emissions of SIC per MWh were about one third of the emissions of SING. Based on operating
data for SING and SIC, Cochilco estimated that, in 2007, the average CO
2
equivalent per tonne of
anode copper emitted by the Chilean smelters was 0.86 tonnes.
The Cochilco data illustrate the impact of the particular fuel mix used for power generation on
greenhouse gas emissions per MWh.
6.1.3 Other data
Other data on copper smelting energy requirements have been published in recent years. These are
presented in Table 13 below.
Table 13: Energy consumption in copper smelting (MJ/tonne ER cathode)
Reference Electric energy Fossil fuel Total
Cochilco [17, 18] 5090 (13,395) 3990 9080 (17,385)
Piret [15] (Smelting) 8411 5170 13,581
Piret [15] (Oxygen) 1989 1989
Marsden [21] 5046 (13,279) 6296 11,342 (19,575)
Note: Numbers in brackets correspond to the equivalent fuel of electric energy numbers reported in the
respective references, using a 38 % efficiency conversion factor.
Cochilcos (Comisin Chilena del Cobre-Chilean Copper Commission) smelting energy data are taken
from their comprehensive study on energy consumption in Chiles primary copper mining industry
[17, 18]. These are average numbers for the seven smelters currently operating in Chile. Four of them
are equipped with Teniente converters; one smelter operates a hybrid Noranda/Teniente Reactor; the
flowsheet of another one combines an Outokumpu flash furnace and Teniente converters; and one
operates an Outokumpu flash furnace as the sole primary smelting unit. All of them use PS converters
for processing high-grade matte to blister copper, but different slag cleaning technologies. The Cochil-
co numbers presented in Table 13 were slightly adjusted to account for copper losses in smelting. It
should be noted that waste heat recovery from process gases is not normal practice in Chilean smelters.
It should also be noted that five of these smelters get their electric energy from an interconnected pow-
er generation system with an important hydro component [17, 18].
Pirets data are based on a flash smelting operation with an annual copper production capacity of
150,000 tonnes [15]. The electric energy number in Table 13 corresponds to his calculated fuel
equivalent, also using 38 % conversion efficiency. As shown in the table, Piret reports a separate
energy consumption number for oxygen generation. Assuming that all of this oxygen is consumed in
smelting, his total energy requirement for smelting increases to 15,570 MJ/tonne of copper.
Marsden investigated energy consumption in Freeport-McMoran primary copper operations in the
USA, Chile and Peru [19]. These operations include various hydrometallurgical processing routes.
His smelting number was based on the Miami Isasmelt Smelter operation. Apparently, in his paper,
the electric energy number corresponds to direct conversion of kWh to MJ. On this assumption, a
fuel equivalent, shown in brackets, was also included in Table 13.
No information on smelting various unit operations is given in the papers and reports published by
the authors whose numbers are presented in Table 13. It is not possible, then, to comment on differ-
ences between their numbers and those calculated by the present authors. However, it should be
noted that the energy consumption numbers developed in this study most probably correspond to
ideal operations, and do not account for almost unavoidable real life inefficiencies. Factors such as
heating of launders and ladles, short partial or total smelter shut downs that require maintaining fur-
naces hot, generation of solid reverts, etc. contribute to increasing heating fuel requirements above
model calculations. In addition, neither Kellogg and Henderson nor the present authors consider
smelter maintenance in their respective smelter models, an item that it is most probably included in
industrial data. Nevertheless, the authors believe that results of studies such as the one they con-
ducted contribute to better understanding processes and may assist in identifying unit operations that
offer potential for improvement.
There are still some additional published data on energy consumption in copper smelting, such as
those reported in a paper by Norgate and Rankin on life cycle assessment of copper and nickel [20].
However, their basic assumptions are much different from those used by the present authors.
7 Concluding comment
Is there room for further energy savings in copper concentrate smelting? Recovering heat from a
variety of combustion gases (anode furnaces, melt holding furnaces, etc.), a rare practice in smelters
today, is an option. The corresponding decrease of greenhouse gas emissions may become an addi-
tional incentive for implementing these practices. New technological advances, such as the conti-
nuous copper fire refining process under investigation in Chile also offer opportunities for energy
savings [21]. Possible new reductions in energy requirement for producing oxygen have been an-
nounced [22]. In summary, it would not be surprising to see even more energy efficient smelters in
the near future.
Acknowledgements
Two of the authors, P. Coursol and P. J. Mackey, thank Xstrata Process Support for permission to
publish this paper. The three authors express their appreciation to numerous colleagues from indus-
try for valuable relevant data and comments. However, the authors take full responsibility for the
data presented in this paper.
References
[1] KELLOGG, H. H. & HENDERSON, J. M. (1976), Energy Use in Sulfide Smelting of Cop-
per, Extractive Metallurgy of Copper, Vol. I, Ed. J.C. Yannopoulos et al., TMS-AIME, 1976,
373-415.
[2] INCO Staff (QUENEAU, P. E. & SPROULE, W.K.), Oxygen Flash Smelting, JOM 7 (7),
1955, 742-750.
[3] TRIPATHI, N., COURSOL, P., MORISSETTE, S. & MACKEY, P.J., (2009), A new metal-
lurgical Model of the Horne Smelter of Xstrata Copper, Pyrometallurgy of Nickel and Cobalt
2009, Ed. J.Lin et al., MetSoc of CIM, 519-530.
[4] RAMACHANDRAN, V., DIAZ, C., ELTRINGHAM, T., JIAN, C.Y., LEHNER, T.,
MACKEY, P.J., NEWMAN, C.J. & TARASOV, A.V., Primary Copper Production A Sur-
vey of Operating World Copper Smelters, Copper-Cobre 2003, Vol. IV (Book 1), Ed. C. Daz
et al., MetSoc of CIM, 3-106.
[5] GEORGE, D. B., GOTTLING, R. J. & NEWMAN, C.J. (1995), Modernization of Kennecott
Utah Copper Smelter, Copper 95-Cobre 95, Vol. IV, Pyrometallurgy of Copper, Ed. W.J.
(Pete) Chen et al., MetSoc of CIM, 41-52.
[6] WALQUI, H., NORIEGA, C., PARTINGTON, P., & ALVEAR, G., SPCCs 1,200,000 TPA
Copper Isasmelt, Sohn International Symposium Proceedings, Ed. F. Kongoli et al., TMS,
2006, 199-209.
[7] ARTHUR, P., PARTINGTON, P., FAN, W. & LI, Y., ISASMELT Not just a flash in the
pan, Copper 2003-Cobre 2003, Vol. IV (Book 1), Ed. C. Daz et al., MetSoc of CIM, 139-156.
[8] MARISCAL, L. & HERRERA, E., Isasmelt Slag Chemistry and Copper Losses in the Rotary
Holding Furnaces Slag at Ilo Smelter, MOLTEN 2009, Gecamin, Santiago, Chile, January
2009, 1241-1250.
[9] GOTO, M. & HAYASHI, M., The Mitsubishi Continuous Process, Mitsubishi Materials Cor-
poration, Tokyo, Japan, June 2002.
[10] STRACHAN, M., SMITH, T., MIRANDA, S., ARRIAN, J., MAST, E. & MUNOZ, E., The
Phase III Expansion of the Altonorte Copper Smelter, Chile, Copper 2003-Cobre 2003, Vol.
IV (Book 1), Ed. C. Daz et al., MetSoc of CIM, 203-223.
[11] MORI, K., NAGAI, K., MORITA, K. & NAKANO, O., Recent operation and improvement at
the Sumitomo Toyo Peirce-Smith converters, International Peirce-Smith Converting Centen-
nial, Ed. J. Kapusta et al., TMS, 2009, 151-160.
[12] EK, M. & OLSSON, P., Converting and casting at Bolidens Ronnskar Smelter 2009 An
update, International Peirce-Smith Converting Centennial, Ed. J. Kapusta et al., TMS, 2009,
127-131.
[13] HERRERA, E. (Ilo Smelter), private communications with Pascal Coursol, 2009.
[14] REYES, P., MARTINICH, B., MALUENDA, R., FELMER, G., & ADELL, P., Start up and
operational optimization of the 800 MT Oxial oxygen plant at the Caletones Copper Smelter
El Teniente Division of Codelco Chile, Copper 2007, Vol. III (Book 2), Ed. A.E.M. Warner et
al., MetSoc of CIM, 2007, 309-321.
[15] PIRET, N.L., Will Todays Needs Promote Copper Concentrate Hydroprocessing? Update and
Perspectives, World of Metallurgy - ERZMETALL 62 (2009), No. 6, 344-365.
[16] OPFERMANN, A., GROSSE, A., BAUMGARTNER, S., AND RIEDINGER, D., Improve-
ment in Energy Efficiency, Stahl und Eisen, Sept. 2009, Nr. 129, 9, pp. 530-536.
[17] Consumo de Energa y Emisiones de gases de Efecto Invernadero de la Minera del Cobre de
Chile. Ao 2008 (Energy Consumption and Greenhouse Gas Emissions in Chiles Copper
Mining - 2008), July 2009.
[18] PIMENTEL, S., The Chilean Copper Mining Sector: Energy Consumption and Greenhouse
Gas Emission Profile (2001-2007), ENVIROMINE 2009, Santiago, Chile, Sept. 30-Oct. 2,
2009.
[19] MARSDEN, J.O., Energy Efficiency and Copper Hydrometallurgy, Hydrometallurgy 2008,
Ed. C.A. Young et al., SME, Phoenix, AZ, 2008, 29-42.
[20] NORGATE, T.E. & RANKIN, W.J., Life Cycle Assessment of Copper and Nickel Produc-
tion, MINPREX 2000, AusIMM, Melbourne, Vic., 11-13 Sept. 2000, 133-138.
[21] RIVEROS, G., WARCZOK, A., MARIN, T. & UTIGARD, T., A new Paipote process of con-
tinuous fire refining of copper, Copper 2007, Vol. III (Book 2), Ed. A.E.M. Warner et al.,
MetSoc of CIM, 2007, 633-645.
[22] FOGASH, K. (Air Products and Chemicals), Technical considerations for a very large scale
air separation unit for a coal fired power plant application, IEAGHG International Oxy-
Combustion Network, 3
rd
Workshop, Yokohama, Japan, March 5-6, 2008.
[23] LEHNER, T. (Boliden), Private communication with P.J. Mackey

Problems of Lead and Arsenic Removal from
Copper Production in a One-Stage
Flash-Smelting Process
J. Czernecki, Z. mieszek, Z. Miczkowski, G. Krawiec, S. Gizicki
Institute of Non-ferrous Metals
Sowiskiego 5
Gliwice, Poland

Keywords: Flash smelting, lead removal, arsenic removal
Abstract
Copper content in chalcocite-bornite concentrates of KGHM Polska Miedz S.A. is at the same level
as in the chalcopyrite concentrates 20-30 %, however their iron and sulphur content is several
times lower, 2-7 % and 9-12 %, respectively. The additional characteristic feature of the Polish
copper concentrates is presence of organic carbon (5-8 %) and significant content of impurities,
especially Pb (1.5-3.0 %) and As (0.05-0.4 %).
The mentioned above properties, especially low iron and sulphur content, resulted in application of
a one stage flash-smelting process in HM Glogow II in 1978. Currently implementation of an
analogous technology for HM Glogow I is considered, to replace shaft furnaces where briquettes of
copper concentrates are smelted.
Because of specific thermodynamic properties of lead and arsenic and their chemical compounds
the removal of the impurities from the one stage flash-smelting process requires special technologi-
cal operations and logistics. Pb and As concentrate in significant volumes in the CuPbFe alloy from
electric furnace and in the converter slag. To maintain a specific lead and arsenic level in CuPbFe
alloy it is necessary to direct a part or the whole of converter slag mass for separate treatment.
Currently a part of the converter slag from HM Glogow II is directed to the electric furnace for cop-
per removal from flash-smelting slag and another part for treatment in shaft furnaces of HM
Glogow I. With the elimination of the shaft process at HM Glogow I and replacing it with a one
stage flash-smelting process the converter slag will have to be treated by some other methods. Both
hydro- and pyrometallurgical technologies are considered for the treatment of converter slag.
The paper presents a description of lead and arsenic behaviour in the one stage flash-smelting pro-
cess as well as discussion on the methods for separation of those impurities from copper production
cycle.
Czernecki, mieszek, Miczkowski, Krawiec, Gizicki
1 Introduction
Polish copper concentrates differ in chemical and mineralogical compositions from the chalcopyrite
concentrates commonly produced elsewhere. Chalcopyrite concentrates usually contain between 25
and 35 % (mass percent unless otherwise denoted) copper, iron and sulphur each, and the various
gangue oxides typically do not exceed 10 % in the concentrate. Silica must be added to the copper
smelting process to convert the iron contained in the concentrates to a slag, and the slag mass usu-
ally exceeds by 50 % that of the chalcopyrite concentrates.
KGHM Polska Miedz S.A. concentrates are characterised by copper contents similar to that of chal-
copyrite concentrates (up to 30 %), but they contain several times less sulphur (below 12 %) and
iron (under 6 %). The main sulphide minerals in the Polish concentrates are chalcocite (Cu
2
S) and
bornite (Cu
5
FeS
4
). Moreover, an important component of the KGHM Polska Miedz S.A. concen-
trates is organic carbon, and the main impurities which are present in significant amounts are lead
(up to 2.5 %) and arsenic (up to 0.3 %). The gangue components (SiO
2
, CaO, MgO, Al
2
O
3
K
2
O,
Na
2
O etc.) make up 42-50 % of the concentrate mass and are present in such proportions that no
additives are required during smelting, and the mass of slag produced is even lower than that ob-
tained when smelting chalcopyrite concentrates. The low sulphur content of the Polish concentrates
is also advantageous because of the over-production of sulphuric acid in the Polish and global mar-
kets, and the production of sulphuric acid from pyrometallurgical processes is often unprofitable.
At KGHM Polska Miedz S.A., two copper production technologies are in use. The shaft furnace
process is used at the Legnica Copper Smelter and the Glogow I Copper Smelter. A single-stage
direct-to-blister Outokumpu flash smelting furnace is used at the Glogow II Copper Smelter.
2 Single-stage flash process
Due to the low iron and sulphur contents found in Polish concentrates, blister copper can be ob-
tained directly from the concentrate by flash smelting with acceptable process parameters. The flash
process aims to produce blister-quality copper and off-gases containing a high SO
2
concentration
suitable for processing in a sulphuric acid plant.
The flash process used at Glogow II Copper Smelter since 1978 includes the following operations
(Figure 1):
charge preparation, which consists primarily of drying the concentrate to less than 0.3 % mois-
ture,
blister copper smelting from concentrate in a flash furnace,
slag cleaning in an electric furnace,
converting the CuPbFe alloy produced in the electric furnace,
fire-refining of blister copper produced in the flash furnace and the converters in rotary and sta-
tionary anode furnaces, and
anode copper electrolysis.
Lead and Arsenic Removal in a One-Stage Flash-Smelting Process
Figure 1: HM Glogow II copper production flow sheet
One of the main advantages of the flash process is that charge preparation is simple and three times
less expensive than briquetting as it only requires concentrate drying in a rotary dryer. The flash
furnace also produces a steady stream of off-gas containing 12-16 vol. % SO
2
, which is directed to a
sulphuric acid plant.
The following products are obtained from the flash smelting furnace:
blister copper containing under 0.3 % of Pb and up to 0.5 % of oxygen,
slag containing up to 12-15 % copper, mainly as copper oxide dissolved in slag,
dusts containing sulphates of copper, calcium and magnesium that are recycled to the flash pro-
cess, and
gases containing SO
2
that are directed to sulphuric acid plant.
The flash furnace slag is cleaned in an 25 MVA electric furnace to which coke and limestone are
added. The products of slag cleaning are:
CuPbFe alloy containing 70-80 % Cu, 15-25 % Pb, and approximately 5 % Fe,
a discard slag containing between 0.4 and 0.7 % Cu,
dusts containing over 30 % Pb and approximately 15 % Zn which are used for lead and zinc pro-
duction, and
gases, which following oxidation of CO in the combustion chamber and dedusting are directed to
the atmosphere.
The CuPbFe alloy is converted to blister copper containing approximately 98 % Cu and under
0.3 % Pb, which is then fire refined in anode furnaces together with flash furnace blister copper.
1
2
3 4
5
6
7
8 9
copper
concentrate
natural gas
air for
burning
additional
air
exhaust
gas
concentrate
and flue dust
process air
fuel oil
fuel oil
air
FSF slag
flue dust
converter slag
blowing air
blister
blister
Cu-Pb-Fe alloy
process gas
to sulphuric acid plant
11
converter gas
converter
flue dust
1. Rotary dryer
2. Electrostatic precipitator
3. Dried charge bin
4. Flue dust bin
5. Flash smelting furnace
6. Waste heat bolier

8. Electric furnace
9. Convertor
10. Bag filter
11. Wet dedusting system
12. Anode refining furnace
13. Anode casting wheel
7. Electrostatic precipitator
waste
slag
exhaust gas
EF flue dust
10
natural gas, oil
air
exhaust gas
12
13
anode
copper
anods to electrorefinary
anode slag
3 Behaviour of lead and arsenic in flash smelting process
Lead and arsenic are present in the concentrates as sulphides. The predominant lead phase in a con-
centrate is galena (PbS), whilst arsenic forms sulphides with other metals such as arsenopyrite
(FeAsS
2
), enargite (Cu
3
AsS
4
) and tennantite (3Cu
2
S As
2
S
3
).
The processes used at HM Glogow II take place in an oxidising atmospheres, in the flash process
and CuPbFe alloy converting, in the reducing atmospheres for the slag cleaning, and in alternate
oxidising and reducing atmospheres in the copper fire-refining. In accordance with changing condi-
tions in which these processes take place, lead and arsenic are present in the process products as
oxides, metals or sulphates. The behaviour of lead and arsenic is determined by the properties of the
elements and their chemical compounds, and by the properties of their solutions produced during
technological processes.
Figure 2 shows ranges of phase stability for iron, copper, lead and arsenic at 1300 C and thermo-
dynamic equilibrium, and chemical composition ranges for gaseous phase for individual metallurgi-
cal processes. Phase stability ranges refer to pure substances and do not account for solutions that
may be formed by the components. However, they give a qualitative explanation of the behaviour
of lead and arsenic during the copper production cycle.

Figure 2: Sulphur-oxygen potential diagram for Cu, Fe, Pb, As-O-S systems at 1573 K
-12
-10
-8
-6
-4
-2
0
-16 -14 -12 -10 -8 -6 -4 -2 0
log (p
S2
/p)
l
o
g

(
p
O
2
/
p
)
CuSO
4
Cu
2
O
Cu
Cu
2
S
FeSO
4
Fe
2
O
3
Fe
3
O
4
FeO
Fe
FeS
Pb
PbO
PbSO
4
PbS
As
As
2
O
3
As
2
O
5
As
2
S
3
log p
SO2
= 0
Flash smelting process
Slag cleaning in electric furnace
Refining process and Cu-Pb-Fe
alloy converting
log p
SO2
= -2

3.1 Flash furnace
Quality of the blister copper produced in the flash furnace is determined by its lead content, which
should be lower than 0.3 % for the product to meet the requirements of fire refining in anode fur-
naces. The lead concentration in copper depends mainly on the thermodynamic equilibrium state of
the reaction:
[Pb] + (Cu2O) 2[Cu] + (PbO) (1)
which determines the content of that impurity in blister copper:
] Pb [ ) O Cu ( ) O Cu (
) PbO ( ) PbO (
) 1 ( p
] Pb [
1
X
X
K
1
X
2 2

= (2)
where:
X mole fraction of the component in the solution
activity coefficient
K
p(1)
constant of the thermodynamic equilibrium reaction (1)
() [] defines the component contained in slag or metal, respectively
Figure 3: Lead content in blister copper versus copper content in slag for industrial flash furnace
in Glogow II smelter
The equilibrium concentration of lead in copper is a function of cuprous oxide content in the slag
(in inverse proportion) and Pb content in the concentrate and, by extension, PbO content in the slag
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
9 10 11 12 13 14 15 16
Copper content in slag [%]
L
e
a
d

c
o
n
t
e
n
t

i
n

b
l
i
s
t
e
r

c
o
p
p
e
r

[
%
]

(in direct proportion). The other parameters of Equation (2) vary in a narrow range, therefore their
influence on the lead content in copper is less significant. The relationship between concentration of
lead in copper and copper in slag for industrial flash furnace at Glogow II smelter is presented in
Figure 3.
Arsenic is difficult to remove from metallic copper because of the formation of intermetallic com-
pounds in the Cu-As binary phase system that depress the activity coefficient for arsenic. Two val-
ues of
As
in copper at 1523 K, 510
-4
and 6.410
-4
, can be found in the literature [1, 2] and do not
differ significantly from each other. Due to arsenics very low activity coefficient, arsenic cannot be
removed from copper by evaporation in spite of a very high vapour pressure over pure arsenic metal
of 7.8610
2
atm at 1523 K [3].
Arsenic removal from copper to slag is described by the following reaction:
[As] +
2
O
2
(AsO
) (3)
which means that :
[ ]
( ) ( )
[ ]

p(3)
2 /
O As

AsO AsO
As
K
1

P
X
X
2

(4)
To remove arsenic from copper, it is necessary that arsenics activity coefficient in slag has a low
value and that the oxygen partial pressure be high. Yazawa [2] reported that the value of
AsO
for
a FeO-Fe
2
O
3
-CaO basic slag at 1523 K is seven times lower than for a silicate slag.
At the Glogow II smelter arsenic is partly removed to slag in the flash furnace and is refined from
copper in rotary anode furnaces. A mixture of Na
2
CO
3
and CaCO
3
flux is used, with the carbonates
amount and ratio depending on the composition, especially the arsenic level, of the charge to the
anode furnaces.
3.2 Electric furnace
The slag from flash process of concentrate smelting is characterised by high content of copper
(13-16 %) which is present in oxidised form, mainly (Cu
2
O). To recover the copper, the flash smelt-
ing slag is subjected to copper removal process in the electric furnace.
The process is based on the reduction of oxide compounds of copper and other metals, mainly lead,
iron and arsenic, and coalescence and sedimentation of the generated particles of metals and alloys.
In the electric furnace, mainly copper oxide (Cu
2
O) and lead oxide (PbO) are reduced.
Cu
2
O + CO 2 Cu + CO
2
(5)
PbO + CO Pb + CO
2
(6)
CO
CO
O Cu
2
Cu
) 5 ( p
P
P
a
a
K
2
2
= (7)
CO
CO
PbO
2
Pb
) 6 ( p
P
P
a
a
K
2
= (8)
One can define the dependence between Cu
2
O and PbO concentrations in the slag in different steps
of the slag cleaning process, which takes the form:
( )
( )
[ ] [ ]
[ ] [ ]
( )
( ) O Cu
PbO
Pb Pb
Cu
2
Cu
O Cu
) 6 ( p
) 5 ( p
PbO
2
2
X
X
X K
K
X
= (9)
The equilibrium concentration calculations for Cu
2
O and PbO were performed for flash furnace slag
containing 14 % Cu and 3 % Pb at 1350 C.
Thermodynamic data of OUTOKUMPU HSC software were used in the calculations for pure sub-
stances.
Figure 4: Relation between lead and copper equilibrium concentrations in slag
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
0 2 4 6 8 10 12 14
Copper content [%]
L
e
a
d

c
o
n
t
e
n
t

[
%
]

It was assumed that the activity coefficient values for the individual components have the following
values:
[ ]
1
Cu
=
( )
4
O Cu
2
= [3, 4]
[ ]
3
PbO
= [5]
( )
7 . 0
PbO
= [4]
The results of the calculations presented in Figure 4 show the dependences between the Cu
2
O and
PbO in the flash furnace slag during its reduction.
The relationships resulting from the conducted investigations between Cu and Pb as well as Cu and
As in the slag during its reduction in electric furnace are presented in Figures 5 and 6.
Figure 5: Dependence between Cu and Pb concentration in the slag during reduction

0
0.5
1
1.5
2
2.5
3
3.5
0 1 2 3 4 5 6 7 8 9 10
Copper in slag [%]
L
e
a
d

i
n

s
l
a
g

[
%
]

Figure 6: Dependence between Cu and As contents in the slag during reduction
They show different behaviour of oxide compounds of lead and arsenic during reduction. While
significant reduction of PbO starts when the copper content in the compound becomes reduced to
about 3 % (see Figure 5), arsenic oxides are reduced at the similar intensity as Cu
2
O, which can be
observed in the rectilinear relationship between Cu and As concentrations in the slag during reduc-
tion (see Figure 6).
Running the copper removal process until low final Cu concentrations in the waste slag are reached
(about 0.5 %) results in production of CuPbFe alloy of high lead (15-25 %) and arsenic (1.5-3.0 %)
concentrations.
3.3 Lead and arsenic balance in one-stage flash-smelting process
Figures 7 and 8 present lead and arsenic balance in the one-stage flash-smelting process in fully
completed copper production cycle.
3.3.1 Lead
Almost all of the lead contained in the concentrate is oxidised in the flash furnace and, as a result,
over 97 % of the lead passes to the slag. In the electric furnace used for flash furnace and converter
slag cleaning, almost 67 % of the lead reports to the CuPbFe alloy, while 17 % passes to the dusts,
and the remaining 16 % remains in the waste slag. During converting of the CuFePb alloy, about
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0.1
0 1 2 3 4 5 6 7 8 9 10
Copper in slag [%]
A
r
s
e
n
i
c

i
n

s
l
a
g

[
%
]

90 % of the lead passes to the slag and about 10 % reports to the off-gas as dust. The converter cop-
per and blister copper from the flash furnace contain 5 % of lead entering the process in blended
concentrate feed.
Figure 7: Lead balance in flash-smelting process (fully completed copper production cycle)
A characteristic feature of the technologies used at the Glogow II Copper Smelter is a tendency for
the lead to remain in the condensed phase semi-products, such as slags and CuPbFe alloy, with only
a small amount of lead reporting to the dusts generated in the processes. This results in high lead
contents in the semi-products, and the circulating amount of Pb exceeds the mass of lead contained
in the concentrate feed.
3.3.2 Arsenic
The average arsenic content in the blended flash smelting concentrate is 0.15 %. Due to the high
levels of alkaline components, such as CaO, MgO and K
2
O, and the high degree of concentrate oxi-
dation in the reaction shaft, the slag has a high ability to remove As from copper so that over 60 %
of As contained in the concentrate passes into the slag, and only approximately 20 % remains in the
blister. However, the reducing conditions in the electric furnace cause approximately 90 % of the
As present in the incoming slag to report to the CuPbFe alloy. Only about 5 % of the arsenic is re-
moved from the copper production cycle in dust with another 3 % exiting within the waste slag.

During converting of CuPbFe alloy, which partially takes place in the presence of a basic flux, al-
most 70 % of the As passes to the converter slag while over 10 % of the As reports to the furnace
off-gases as dust.
The characteristic feature of arsenic behaviour in the flash process used at the Glogow II Copper
Smelter is that this element is only slightly eliminated to dusts and waste slag during the pyrometal-
lurgical slag cleaning and alloy converting process.
Figure 8: Arsenic balance in flash-smelting process (fully completed copper production cycle)
The described lead and arsenic behaviour results in circulation of large masses of those impurities
in the system of electric furnace converters. Products of those processes, i.e. CuPbFe alloy and
converter slag are characterised by high concentration of Pb, Cu and As. The highest lead concen-
tration can be observed in the converter slag, which contains 25-45 % of Pb and 15-35 % of Cu.
Circulation of large masses of those types of materials brings several unfavourable phenomena, of
which the most important are:

lowering of copper removal efficiency because of the necessity to reduce not only cuprous oxide
but also large amounts of oxide compounds of lead,
limitation of electric furnace capacity because of the large amounts of converter slag being proc-
essed there,
consumption of additional masses of limestone and coke as well as electric energy,
higher wear of refractory lining in electric furnace and converters operating in the presence of
liquid solutions of high lead content (CuPbFe alloy, converter slag),
formation of large mass of dusts which are difficult in treatment in the lead production technol-
ogy,
necessity for sending some amounts of converter slag (10-20 %) from Glogow II smelter for
treatment in shaft furnaces of Glogow I smelter to maintain the Pb concentration in the alloy be-
low 15 %.
4 Methods for lead and arsenic removal from copper
production cycle
An efficient method for removal of lead and arsenic from copper production cycle is to send the
converter slag from oxidation of CuPbFe alloy for separate treatment.
Two technologies for converter slag treatment were developed:
hydrometallurgical method
pyrometallurgical method
The hydrometallurgical method is based on leaching of the converter slag with sulphuric acid. Two
basic products are produced in that way:
copper sulphate which contains about 32 % of Cu and about 21 % of S
leached out converter slag which contains lead (about 44 %), copper (about 2 %), sulphur (about
3.5 %) and arsenic (about 2.5 %)
Both products can be recycled in KGHM Polska Miedz S.A. The leached out slag can be used in
lead production together with other lead bearing materials from copper metallurgy (dusts and slime
from process gases dedusting). The impure copper sulphate can be utilised as a component of the
charge material for flash furnace.
Pyrometallurgical method uses the difference in chemical affinity of copper and lead to oxide and is
based on reduction of cuprous oxide from converter slag to the level of 2-4 %. Thus prepared slag
contains 35-50 % of Pb and is a good material for lead production together with other lead bearing
materials from KGHM Polska Miedz S.A.
The effects of sending the converter slag into the separate treatment are presented in Figures 9 and 10.
It can be observed that this method provides possibilities for fifty-fold reduction of the amount of
lead and fivefold reduction of the amount of arsenic circulating with converter slag between the
converter and electric furnace.
Figure 9: Lead balance in flash-smelting process (sending of converter slag to separate treatment)
Figure 10: Arsenic balance in flash-smelting process (sending of converter slag to separate treatment)

5 Summary
One-stage flash-smelting process is characterised by low level of Pb and As elimination from cop-
per production cycle in pyrometallurgical operations (melting of the concentrates in the flash fur-
nace, copper removal from flash-smelting slag in electric furnace, CuPbFe alloy converting).
The efficient method for separation of those impurities is to direct the converter slag from CuPbFe
oxidation of high Pb, Cu and As content for separate treatment to be processed into material which
can be used in lead production.
References
[1] Azukami T., Yazawa A.: Can. Metall. Quart., 1976, 15, p. 111
[2] Yazawa A.: A new slag system composed of ferrites, Nove Vedecko-Technicke Poznatky
v Hutnictve, Kosice, 1982, p. 32
[3] Ptak W., Czernecki J.: The cuprous oxide activity in the slag SiO
2
-CaO-Al
2
O
3
, Arch. Hut.
1981, p.213
[4] Pomianek T. et al.: Lead distribution between copper and flash furnace slag in thermodynamic
equilibrium conditions, IMN Proceedings, 1978, Nr 3
[5] Esdeile J. O., McAdam J. C.: Proc. Austral. Inst. Min. Met., 1971, Nr 239, p. 71


Control of Fugitive Emissions in a Continuous
Mitsubishi C-Furnace during Limestone Fluxing
Adam Salomon de Friedberg, Alan Hyde Mark Coleman
Xstrata Copper Canada, Xstrata Process Solutions (XPS) Clyde Materials Handling
Kidd Metallurgical Site
Timmins, Ontario, Canada, P4N 7K1 Doncaster, UK

Keywords: Mitsubishi process, fugitive emission, flux, limestone, RotoFeed

Abstract
The Xstrata Copper Kidd smelter operates a Mitsubishi Continuous Smelter. The converting
furnace requires granular limestone as flux for the converting process. Historically, the combination
of gas evolution from the calcining reaction and batch injection of the limestone produced
fluctuations in gas volume that made draft control in the furnace challenging. Fugitive gas
emissions from the converting furnace through the uptake shaft typically cycled in time with
limestone injection.
In 2006, a Clyde Materials Handling (Clyde) RotoFeed
was installed to provide continuous

injection of limestone to the converting furnace with the intent of smoothing out the gas generation
from the calcining of the flux and reduce uptake fugitive emissions. The project has been an
unqualified success resulting in improved draft control, higher blowing rates and a better than 85 %
reduction in lead and particulate emissions from the converting furnace uptake.
This paper will describe the issues surrounding limestone injection and the equipment used. The
paper will go on to detail the commissioning issues around the Clyde RotoFeed
. System
performance and the impact on converting furnace operation will also be reviewed. A resulting
opportunity for process control improvement will be outlined.
Clyde RotoFeeds offer precise and stable furnace injection systems for many materials. Its use will
be outlined in a general manner showing how the Clyde Solution can deliver control back to the
operator as seen in this example.
1 Introduction
The Xstrata Copper Kidd smelter uses granular limestone as flux for the converting furnace.
Historically, the combination of carbon dioxide gas evolution from the calcining reaction and batch
injection of the limestone produced fluctuations in gas volumes that made draft control in the
de Friedberg, Hyde, Coleman
furnace challenging. Furnace uptake fugitive gas emissions which cycled in time with limestone
injection (referred to as puffing) were common. In order to compensate for puffing, the
converting furnace was commonly operated at a lower than optimal draft set-point.
In 2005, a project was initiated to eliminate converting furnace puffing associated with limestone
fluxing. The following year, a Clyde RotoFeed
was installed to provide continuous injection of

limestone to the converting furnace, smooth out gas generation from the calcining of the flux and
reduced uptake fugitive emissions. The project has been very successful - resulting in improved
furnace draft control, higher blowing rates and an 85 % reduction in lead and particulate emissions
from the uptake.
This paper will describe the steps taken to address converting furnace puffing, as well as look at
commissioning issues around the Clyde RotoFeed
and the impact of this project on converting

furnace operation.
2 Problem definition converting furnace puffing
The granular limestone used as flux for the converting furnace contains roughly 43 % CO
2
bound as
calcium and magnesium carbonates as shown in Table 1.
Table 1: Limestone flux typical properties

Upon injection into the converting furnace, the limestone rapidly heats-up decomposing first the
magnesium carbonate and then the calcium carbonate into magnesia, lime, and carbon dioxide:
MgCO
3
(s) MgO(s) + CO
2
(g) decomposes at 623 K (1)
CaCO
3
(s) CaO(s) + CO
2
(g) decomposes at 1172 K (2)
At an operating temperature of 1523 K, the evolved carbon dioxide occupies a volume
approximately 3500 times greater than that of the limestone. Depending upon the fluxing rate and
blowing volumes, the CO
2
produced can account for more than 5 % of the total blowing volume as
shown in Figure 1.
Control of Fugitive Emissions in a Continuous Mitsubishi C-Furnace
Figure 1: CO
2
-generation from limestone injection
The volumes of carbon dioxide generated by the flux decomposition reactions could be
compensated for if they were constant. However, batch injection of the limestone with cycle times
of approximately one minute resulted in the flux being slug-fed to the furnace. The instantaneous
fluxing rates could easily be two, three or even more times greater than the average indicated. This
produced repeated surges in the furnace gas volume that coincided with the flux injection cycles.
The draft control dampers could not react fast enough to compensate for these surges and the
converting furnace would often puff in rhythm with the injection of limestone shown in Figure 2.
Figure 2 illustrates the fluctuations in converting furnace draft with batch injection of limestone.
The time period shown spans approximately 20 minutes. The columns at the bottom of Figure 2
show the times when the batch limestone injection system (VT-14) was conveying to the converting
furnace. The rate of injection was not constant during these periods but, in general, a longer
conveying cycle (wider column) would inject more material. The draft trace for the converting
furnace regularly spiked above 0 mm H
2
O indicating that the furnace had gone positive and was
puffing. The draft spikes also tended to coincide with the beginning of an injection cycle and peak
near the end of the cycle.
The fugitive emission from the converting furnace were a serious workplace hygiene concern with
respect to sulphur dioxide and lead fume. In addition, the ventilation fan in this area, BC105, was
identified as one of the Smelters major contributors to site ground-level concentrations of total
solid particulate and lead.
Figure 2: Converting furnace pressure cycling with batch limestone injection
3 Superior concept continuous injection of limestone
Several alternatives were considered to resolve the issue of converting furnace puffing. Initially,
attempts were made to tune the controls of the off-gas system dampers to compensate for the surges
in gas volumes. Due to the complexity of the off-gas system, the physical size of the dampers
involved and the short period of the draft swings, these attempts were not successful.
The second approach involved working with the existing batch injection tank (VT-14) to smooth
out the injection rates and reduce the fraction of time when limestone was not being conveyed to the
furnace. This dead time averaged 50 to 70 % resulting in actual feed rates that were two to three
times higher than the average. However, even with constant injection rates, the physical size of VT-
14 meant that cycle times could not be significantly lengthened and dead time would remain an
issue.
The next solution evaluated was to eliminate the issue of carbon dioxide evolution altogether by
using lime as a flux rather than limestone. This option was discarded due to the difference in
reagent costs (~ $500,000/year) and safety concerns with respect to handling the corrosive lime.
During the next stage of the project, new injection systems were considered. These included larger
batch injection vessels that would reduce dead time, multiple batch systems that could mimic
continuous injection and fully continuous systems (rotary feeders and screws). The factors
considered in selection included cost, space requirements, reliability, feed rate control, turn-down
capability and delivery. The Clyde RotoFeed
was selected as the superior concept. The operating

cycle of the Clyde system is shown in the schematic below:
Figure 3: Clyde RotoFeed
operating cycle
The main components of the Clyde RotoFeed
are two pressure vessels and a horizontal rotary

feeder. The dispensing vessel and the feeder operate at the same pressure as the conveying line
thereby eliminating the need for a gas seal at the feeder and reducing wear. The upper or lock vessel
depressurises to take on feed and then equalizes pressure with the dispense vessel before
transferring material below. The vent line between the two vessels allows displaced air to move
from the lower dispense vessel to the lock vessel as the material falls.
The configuration in Figure 3 above is for a volumetric-control feeding system. Once the feeder has
been calibrated, the feed rate is controlled by adjusting the rotation rate. The accuracy of this type
of feed control (quoted at +/- 2 %) is dependent upon the uniformity of the material fed and how
consistently the material fills the pockets of the rotary feeder.
The unit at Kidd is equipped with load cells for both the lock and dispense vessels. An additional
valve and expansion joints are used to isolate the two vessels and allow independent weighing. The
use of load cells provides more precise feed control (+/- 1 %) and avoids problems previously seen

at Kidd on other systems with level probes. The valves are very low maintenance Clyde Spheri
or
DomeValves
.
Figure 4: Lock vessel

Figure 5: Dispense vessel

Figure 6: Conveying line
Figures 4 through 6 show the installation of the RotoFeed
at the Kidd smelter. Granular limestone

is delivered to the lock vessel from a storage bin via a series of screw conveyors at the rate of 10 t/h.
The lock vessel fills immediately following transfer to the dispense vessel and waits in the full
condition for the next cycle. When full, each vessel can hold roughly 600 kg of limestone flux. Feed
rates from the dispense vessel can be varied between 0.75 and 7.5 t/h (1 to 5 t/h typical) with the
average limestone feed-rate being 1.8 t/h. The limestone is injected into the converting furnace
through two of the six lances at a pressure of approximately 100 kPa.
This system can also be used to inject coke into the converting furnace. However, since the bulk
density of the coke is half that of the limestone, an alternate control set-up with different weight
settings is used to avoid overfilling.

4 Commissioning
The existing batch injection system was retained as a back-up and alternate system for injection of
coke and addition of converting slag (coolant). The commissioning period coincided with start-up
after a maintenance shutdown and extended for several weeks. During this commissioning period,
the RotoFeed
was on-line over 90 % of the time. The following issues were dealt with:
1) Vent piping transferring load to lock vessel through the expansion joint: The vent piping
routing was changed during installation and new line supports had to be installed.
2) Load cell issues: The unit arrived pre-assembled at the site and the transport bolts were not
removed during installation. Also, one of the load cells had a faulty contact and had to be
replaced.
3) Control logic on the feed rate: A minor density set-point change was required to allow the unit
to function correctly in both operating modes (limestone and coke injection). In addition, a
software limitation on the Variable Frequency Drive was removed to allow the feed rate to be
adjusted down to 0.75 t/h.
4) Slow discharge of limestone from the lock vessel to the dispense vessel: Initially, the cause of
the limestone hanging up in the lock vessel was thought to be insufficient air flow through the
fluidizing air pads. The unit was still operable but cycle times were longer than planned to
allow for the slower material transfer. Ultimately, the cause turned out to be backwards wiring
of the pressure switches from lock vessel to the transition and transition to dispense vessel.
These pressure switches must make to show that the pressures between the two vessels have
equalized before the transfer of limestone is allowed.
5) Problems with tramp material in limestone bin jamming the rotary feeder: This occurred three
times in the first year. Twice, limestone has had to be vacuumed out to remove garbage
(alternative is to disconnect the line and drop the bottom of the tank/feeder/motor assembly).
The vacuuming process takes several hours since there is only a 2 inspection port on the
vessel. Recommendations to overcome this problem include installation of a screen in the
upstream feed system and installation of a 4 clean-out port in the dispense vessel near the
feeder.
5 System performance
Since the completion of commissioning, the Clyde RotoFeed
has performed extremely well.

Availability has been close to 99 % with the majority of downtime due to jamming of the feeder by
foreign material. Maintenance costs have been low and largely limited to preventive maintenance
tours.
With respect to process performance, puffing of the converting furnace has been virtually
eliminated. Area hygiene has seen a dramatic improvement and environmental emissions of
particulate, lead and sulphur dioxide from the fan ventilating the area (BC-105) have dropped by
more than 80 % as shown in Table 2.
Table 2: Emission testing at extraction fan BC-105
Flux utilization has also improved in the converting furnace. With batch injection, lime would, at
times, build up on top of the slag producing un-reacted floating solids. With continuous flux
injection, the lime is dissolved into the slag as it is injected.
Draft control in the converting furnace has also improved significantly. Figure 7 shows a
comparison of converting furnace draft traces over typical 45 minute periods with batch injection of
limestone flux (solid line) and after switching to continuous injection (dotted line).
Figure 7: Converting furnace draft profile comparison
The draft is more tightly controlled and the converting furnace does not go positive with continuous
limestone injection as opposed to 13 times with batch injection. The variation of the data shown in
the Figure above is summarized in Table 3.
Table 3: Draft control improvements
The range of draft swings and the standard deviation of the draft have both been cut by over 50 %.
This improvement in draft control has allowed a modest increase in furnace blowing rates and
copper production along with the hygiene improvements noted above.
6 The Clyde RotoFeed

The Clyde RotoFeed
has been used in similar applications with different materials and reported
similar improvements to system performance.
For example, Copper Concentrate Injection to the Codelco Teniente Converter where the use of
Tuyere Line solids injection has brought stability to the furnace, allowing an increase of over 50 %

in production rates and reduced carry over dust from 6 % to less than 1 %. Furthermore, high
availability levels of over 99 % have been achieved.
The Clyde RotoFeed
has been utilized within the Ausmelt Top Submerged Lance (TSL) process,
where it is used as a fuel feed, mainly pulverized coal, giving better control of the flow and
therefore ability to tightly control viscosity of the slag particularly important when separating
metals, again the high availability is an important feature.
The Clyde system can also be used to inject other materials for example electronic scrap as fuel to
TSL furnaces, ashes, dusts, reverts silica and other fluxes to P-S and anode furnaces. The system is
adaptable and delivers excellent availability stability and performance allowing the user to take
advantage of increased output, stabilized metals quality and reduced emissions.
7 Conclusions
The installation of a Clyde RotoFeed
at the Kidd smelter to continuously inject limestone into the

converting furnace has been an unqualified success. The system has proven itself to be reliable and
effective. Significant improvements to in-plant hygiene, environmental emissions and process
control have been obtained. These achievements include:
Equipment availability of greater than 99 %.
Elimination of converting furnace puffing associated with limestone injection.
Source emissions of
lead 87 %
sulphur dioxide 82 %
total particulate 89 %
More effective use of injected flux.
Improved draft control.
Higher furnace blowing rates.
Acknowledgement
The authors wish to acknowledge Xstrata Kidd Creek and Xstrata Process Solutions who have
assisted in preparing the paper and presentation and have given approval for its inclusion in Copper
2010 Hamburg.
References
[1] Weast R.C. (1980): CRC Handbook of Chemistry and Physics, 60
th
ed., CRC Press, Boca
Raton, Fla., B-93.
[2] Weast R.C. (1980): CRC Handbook of Chemistry and Physics, 60
th
ed., CRC Press, Boca
Raton, Fla., B-65.
[3] Hyde A. (2006): Continuous limestone injection system operation, Internal Kidd SOP.
[4] Salomon de Friedberg, A., (Hyde, A.), Continuous C-Furnace Limestone Injection at the Kidd
Smelter, 10
th
Mitsubishi Family Meeting, Timmins, October 2007.


Modern Flash Smelting Cooling Systems
K. Fagerlund, M. Lindgren, M. Jfs
Outotec Oyj
Riihitontuntie 7 C
FI-02201 Espoo, Finland

Keywords: Flash smelting, cooling system, campaign life, corrosion, resistance, improvement
Abstract
Flash Smelting Technology was first introduced more than 60 years ago and has now become a
leading smelting technology both for copper and nickel sulfides. Energy efficiency, environmental
friendliness, high on-line availability and effective valuable metal recovery became obvious already
in the beginning stages, when Flash Smelting was rapidly gaining acceptance around the world.
There have been lots of valuable lessons learned over the years all throughout Flash Smelting his-
tory and continuous development work has played a major role in keeping Flash Smelting Technol-
ogy in the forefront, as the state-of-art and the best available smelting method available.
An example of continuous development work has been the better understanding of cooling require-
ments in the process acquired by know-how gathered from practical industrial operations and theo-
retical evaluations. Extended campaign time requirements, well above 10 years, have gradually in-
creased the amount of cooling applied in the furnaces. In addition to enhanced cooling equipment
development, modern simulation and modeling tools with increasing fundamental background
knowledge has enabled faster and more accurate design. In combination with deep process knowl-
edge, our understanding of furnace conditions has remarkably improved over the last decades. Dur-
ing operation, increased instrumentation together with a modern process control system gives us
valuable information on the furnace integrity and process conditions. This paper will give an over-
view of some of the fundamental principles of the Flash Furnaces when considering cooling design
and description of modern tools and equipment developed to meet these challenges.
1 Introduction
Outotec introduced Flash Smelting technology more than 60 years ago for the first time in Har-
javalta, Finland. It soon became accepted globally, due to its inherent nature as an energy efficient
and environmentally sound copper and nickel production method. Over the years, lots of improve-
ments and new technology have been applied to several Flash Furnaces over the world and today
Flash Smelting has achieved a clear leading position as the worlds number one primary copper
smelting technology. A very important feature is the extended campaign lives reported up to
Fagerlund, Lindgren, Jfs
15 years, for that there is no comparison in the industry. Historically this would not have been
achievable, thus several structural and operational changes have been made into the technology dur-
ing the last decades. Our knowledge of the furnace conditions has remarkably improved together
with on-line analysis producing continuous information about the furnace integrity. Modern model-
ling and simulation tools can be utilized for better understanding of the complex high-temperature
operations, such as Flash Smelting process line with subsequent off-gas treatment.
About 10 % of the released heat in flash smelting furnaces is regarded as heat loss; most of this be-
ing recovered in cooling water, while around 10-20 % of heat loss ends up heating the surrounding
atmosphere. In addition to cooling elements in some furnaces, the hearth is cooled by forced air-
cooling. The typical energy balance is illustrated in Figure 1. The intention of Furnace cooling is to
increase the campaign time, the integrity of the furnace and safety in the working environment. As a
by-product, the cooling system can contribute in monitoring the process conditions. The more we
understand about the interaction between the process and cooling, the better process control and
safety we can achieve.
The primary cooling system for different flash smelting furnaces comprises the water tank, pumps,
heat exchangers, piping, headers and the drainage system back to the tank. The primary cooling
water is cooled by a secondary system, where heat is transferred either to cooling towers or directly
into natural waters, such as the sea or a river. In the latter case, the investment cost of a cooling
tower can be avoided. The better understanding of the cooling requirements in different parts of the
furnace has been under continuous development work. This paper summarizes some of the basic
principles, whilst choosing correct cooling design parameters and equipment.
Figure 1: An example of the energy distribution of a newly refurbished Flash Smelting Furnace.

2 Degradation mechanisms of the cooling elements
The Flash smelting furnace is a harsh environment that gives rise to several kinds of degradation
mechanisms that shorten the lifetime of the cooling elements. These mechanisms consist of various
corrosion processes, namely high-temperature corrosion, wet corrosion and erosion. They are not
independent but rather act simultaneously and reinforce each other.
In a broad sense, high-temperature corrosion can be defined as an accelerated corrosion of metal
surfaces at high temperatures that results from combined effects of oxidation and reactions with
sulfur compounds and other contaminants. These contaminants may also form a molten salt on a
metal fluxing, destroying or disrupting the normal protective oxide [1, 2]. Wet corrosion incorpo-
rates two corrosion mechanisms: wet corrosion due to condensation of water and wet corrosion due
to condensation of sulfuric acid. The first of these pertains only to the cold face of the cooling ele-
ments. The latter may occur in some cases also on the hot face. Erosion is acting when solid parti-
cles transported with the furnace gases hit the surface of a material. Due to the impacts, either the
material itself or the protective oxide layer is removed exposing the material to corrosion attack [3].
3 Flash smelting furnace as a corrosion environment
The Flash smelting furnace is an aggressive environment due to both temperature and chemical at-
tack. The main degradation-promoting factors inside the flash smelting furnace are: high heat loads,
flow of solid concentrate, flow of gases, and chemical environment (namely SO
2
).
The amount of heat load imposed on a cooling element in a particular flash smelting furnace de-
pends on the location of the element. Nowadays, the heat loads can be reliably determined with nu-
merical modelling and verified by measurements of the cooling waters. The modelling also allows
the optimisation of the cooling element geometry to ensure desirable temperature distribution in the
element. Typically, in the reaction shaft, the lowest elements suffer from the highest heat loads.
This is one reason why the elements in this location are particularly prone to degradation. Addition-
ally, if the process operation is unstable, splashing and thermal strains may occur which have a ten-
dency to remove the protective accretion, which in turn has an adverse effect on the durability of the
cooling elements. Figure 2 schematically illustrates conditions in the reaction shaft, whereas maxi-
mum and minimum absolute heat flux values facing the elements varies from one operation to
other, however, general trend is as described.
Figure 2: Typical heat flux and particle flow distribution in the Flash Smelting reaction shaft area.
Concentrate burner distributor cone spreads the mixture of air and concentrate homogeneously, but
within a confined middle region of the reaction shaft. The idea is that the downwards-moving parti-
cles do not encounter the walls of the reaction shaft. If they did, it would lead to extensive erosion
damage as hard particles with high velocity could seriously damage the protective accretion and soft
copper. The situation is completely different when the concentrate burner is dirty. Then it is possi-
ble that un-reacted particles join the upward-directed back flow and cause erosion to the reaction
shaft walls.
The cooling design of the reaction shaft in modern day furnaces is developed on basis of long-term
experience from smelters around the world and thru modelling heat loads. The design of the lowest
part of the shaft is the critical part as it is affected the most by the suspension plume and distur-
bances in process conditions. On average 40-50 % of the heat losses to cooling waters are extracted
in the reaction shaft and peak heat loads can occasionally be extremely high on this relatively small
surface area. Thus, special attention is needed in the design of the lower part of the reaction shaft.
The first modern design E-type reaction shaft transition cooling elements (Figure 3, left) were in-
stalled in 1999 and have to date not suffered any failures, due to the enhanced cooling capacity and
the improved plugging technology, which forms a corrosion-resistant metallurgical bond instead of
the older threaded design. The longest campaign time for the E-type is at present 10 years, which is
a significant improvement in relation to the older designs.
Another more recent improvement is the BIC cooling element, acronym for Brick Implanted in
Copper (Figure 3, right), which initially was designed for roof cooling. Present day design has re-
placed the horizontal ledge cooling elements, which formed a plateau for corrosion product, with
BIC-design cooling. The BIC-design has been applied in the settler and uptake shaft roofs since

2001 and trials for application on vertical walls were started in 2005. The trial took place in one of
the most rigorous Flash furnaces conditions and proved to be a real success. Through these trials the
design of the furnace cooling was once again improved.
Figure 3: E-type (left) raction shaft cooling elements pre-assembled together with BIC-type roof
cooling elements (right).
The characteristics of the gas flow together with the dimensions of the settler and the reaction shaft
determine the location of the most extensive corrosion. Computational flow dynamics are employed
to model, predict, optimise and ultimately to suppress the tendency to corrosion by gas flows. Fluid
flow and heat transfer modeling has been used in Outotec for several years. Models for Cu- and Ni-
concentrates and Cu-matte combustion have been developed, in collaboration with Universities,
which have been connected to a commercial CFD-software. Details of the development work have
been previously published and can be found in [10-12].
The main chemical load is the combined effect of elevated temperature and sulfurous gases. The
influence of oxidation is minor, unless serious overheating occurs. Some impurities, like halogens,
may have a negative effect on corrosion resistance [4].
4 Corrosion mechanisms, influencing factors and means for
improving resistance
Thermodynamical calculation methods are important in evaluating the high-temperature reactions
of copper. They show that at the temperatures relevant to hot face of copper cooling elements, cop-
per sulfate CuSO
4
is the stable phase, which is confirmed by experimental observations. Sulfide
formation requires a very low oxygen pressure. Therefore, sulfide formation can only take place

under a relatively dense accretion. This explains why copper sulfides are not found at all inside the
refractories and accretions of the flash furnace. Even though, thermodynamically metal oxide and
sulfate have their own stability regions, in practice it is common that they co-exist. The highest cor-
rosion rates are attained at the locations where temperature is the highest if the chemical environ-
ment remains similar.
High-temperature reactions of metals with sulfur are more rapid than reactions with oxygen [5]. A
major reason for the higher reaction rates in sulfur containing environment is that diffusion in sul-
fides is generally much faster than in the corresponding oxides. This, in turn, is directly related to
the considerably higher deviations from stoichiometry in the sulfides and due to their correspond-
ingly higher concentrations of lattice defects.
When material loss is evaluated based on oxidation/sulfidation rate, one has to keep in mind that
this approach is not valid if the scale is periodically detached. In other words, if spalling, thermal
cracking or erosion remove the surface layer and reveal fresh copper to the aggressive environment,
the reaction rates are increased. Erosion resistance of copper can be improved for example by steel
inserts or coatings that have higher hardness.
Thermodynamical stability diagrams state that the formation of liquid phase is possible if suitable
contaminants are present [6]. In that case, corrosion rates are drastically increased.
Copper cooling elements are in some cases exposed to slag, matte or blister. In principle, slag is
chemically inert with respect to cooling elements but aggressive to refractory that cover it. Espe-
cially refractory in blister furnaces are prone to wear induced by dissolved oxygen in the blister
copper. This is explained by wetting of oxygen and sulphur containing copper on MgO-Cr
2
O
3
re-
fractory in Figure 4 [7]. Generally <90 angles are considered to be the limit for capillary infiltra-
tion. This phenomennon is observed especially in anode and traditional tuyere blown converting
furnaces.
Figure 4: The effect on contact angle between magnesia-chrome brick and molten copper as func-
tions of dissolved oxygen and sulfur of the copper.

On the contrary, copper or nickel matte or blister copper is very reactive with copper and they can
dissolve both copper and oxygen (i.e. copper and its protective oxide layer) creating a very aggres-
sive corrosive environment. This is why, in practice, the contact of matte or blister copper with
copper cooling elements is not allowed at all (except in tapping hole elements, which are regarded
as wear parts). The contact is commonly prevented by steel inserts on the surface of the copper
cooling elements.
Outotec has modelled and conducted a number of trials in order to design a cooling element able to
withstand the contact of molten blister. Through extensive research the suitable alloys and designs
have been developed. The SIC cooling element (acronym for Steel Implanted in Copper) was the
fruit of this research and has been applied in the latest Flash Converting and Direct to Blister fur-
naces with success (Figure 5).
Figure 5: The SIC-cooling design implemented in a Slag Cleaning Furnace of the Direct to Blister
process.
Heat loads to the settler walls in general are quite moderate in comparison to the other areas of the
furnace. Typically, a very stable frozen layer of matte or blister protects the lower area of the settler
wall from fluctuating heat loads. Then again, the high heat content of matte and blister may lead to
melt leaks, due to the amount of heat that must be extracted in order for them to solidify. Other
complications arise, because direct blister contact with copper cooling elements must be avoided,
due to reasons explained in previous chapters.
In addition, the slag zone again is a different matter, as slag is very brittle and cannot form a perma-
nent protective layer on the cooling element surface. When the slag accretion collapses, the cooling
element is exposed to high heat loads due to direct radiation. In terms of brick wear, the slag zone is
most erosive as the bath level moves up and down, and through corrosion due to the dissolution of
the refractories into the liquid metal bath and through the Marangoni-effect; in the metal-slag and
slag-gas boundary layers the magnitude of the melt flow velocity is higher than normal due to the
different densities of the melts. The dissolution of MgO from magnesia-chrome refractories into
fayalite slags is particularly found in reducing conditions. This is further enhanced by cracking of

the brick surface due to thermal shocks (leading to build-ups falling) and by the motion of the melt
itself.
In order to meet the strict requirements and challenging conditions in the blister smelting, Outotec
has developed novel design with improved element integrity against blister contact by steel inserts
in the element in combination with pre-installed refractory, example shown in Figure 6.
Figure 6: The SIC-element for blister contact in the bath area of the settler (left) and a novel de-
sign of the settler wall with improved contact between the cooling element and refrac-
tory lining, with the aim of shortening the installation time and enhancing the stability
of the protective lining (right).
Wet corrosion by acid condensation occurs when the surface temperature of the cooling element is
below the dew point of the acid. Dew points of 150-250 C are typical for sulfuric acid in the rele-
vant environment indicating temperature range that is rather close to the surface temperature of the
cooling elements [8]. The dew point depends on SO
3
content and moisture content [9].
Wet corrosion may occur as a result of erosion. As erosion removes copper material, the cooling
channels become closer and closer to the surface of the element. As a result, the surface temperature
of the cooling element decreases correspondingly until it reaches the dew point for acid condensa-
tion. Then rapid damage by wet corrosion ensues. Typical example for several different corrosion
mechanisms contributing to degradation of cooling elements and steel structures is the connection
area between Furnace uptake and heat recovery boiler. Waste Heat Boiler connection between up-
take shaft and WHB is known to be a difficult area where the problems are usually associated with
very demanding operating environment:
Fugitive gases corrosion problems
High heat load at the top of the connection area
Requirements for mechanical endurance against accretion removal/blasting

Different corrosion methods in this high temperature, high erosion and chemical load area are de-
scribed below.
Figure 7: Different corrosion mechanisms contributing to degradation of elements and structures
between the Furnace uptake and heat recovery boiler.
Hot corrosion in gas phase (sulfidation and oxidation) occurs when cooling element surface tem-
peratures increase above approx. 600 C. Corrosion rate increases by degradation of oxide layer on
copper due to particles in the gas flow (erosion) and increasing/fluctuating temperatures:
Erosion by particles in the gas flow is dependent of gas velocity and angle of impingement (col-
lision angle), selection of material helps here (brittle materials with low impingement angles and
ductile material with perpendicular collision).
Erosion leads to small cavities on cooling element surfaces in which corrosion progresses.
When the thickness of the copper is reduced, surface temperatures drop so low that acid might
condensate on the surface (SO
2
-> SO
3
-> H
2
SO
4
), risk is increased by cold air infiltration.
Outotec solution for this demanding environment comes from combination of designing tight con-
nection to prevent air leakages, careful material selection and design for expansion joint, steel and
supporting structures and prevention of element degradation by improving the erosion resistance by
steel inserts or coatings (Figure 8).

Figure 8: Outotec design for uptake and waste heat boiler transition (top) and a novel application
using explosion welding for corrosion-protective coating of cooling elements (beneath).
5 Furnace integrity and monitoring through cooling
There are several different methods available to monitor, measure and collect information from
cooling circuits and/or elements. To improve safety and to optimize furnace operation, on-line data

on the furnace condition is required. Cooling system, instrumentation, DCS and data management
provide valuable information on the process status and local conditions to furnace operators.
Design options for the primary cooling system is either to implement the most commonly used open
system or the more recent closed system. Closed system provides several benefits compared to open
circuits: in closed systems, the make-up water amount is smaller, which is more economic, and pos-
sible leaks are more easily detected. Also, the water quality remains better. The boiling point of the
water is increased through higher pressure, the temperature difference (T) can be higher while
maintaining safety and less cooling water (less pumping capacity) is required. Furthermore, higher
temperatures mean higher heat content, and in the future this heat may be utilized more easily and
effectively (absorption pump, district heating/cooling).
Cooling water flow and temperature measurements in water circuits can provide information about
possible leaks and/or element wear and failures. However, the instrumentation of each circuit for
gathering extensive on-line data is expensive. For this purpose, a new header concept has been de-
veloped. This newly developed header is equipped with flow, temperature and pressure measure-
ment devices. All return water pipes are provided with three-way valves, and each flow is directed
one-by-one to the measuring loop. Only one measurement device set per header is needed and in-
vestment and maintenance costs remain at a reasonable level.
Figure 9: Example of new design for cooling headers.
Information from each circuit can be obtained approximately 1-3 times per hour depending on the
number of circuits in one header. This is frequent enough for close process follow-up. However,
when the process is stable the measuring sequence can be programmed so that one round is per-
formed, for example once per shift, significantly reducing maintenance work and expenses. It is

also possible to measure only one critical circuit continuously or, for example, every second circuit
only (typically, two circuits run through each element).
Based on these measurements, on-line heat flux maps and charts can be created to assist in follow-
up. For example, changes in the reaction shaft heat flux profile indicate changes in burning, say
build-up in the concentrate burner, and corrective actions can be taken before any damage occurs to
the reaction shaft wall.
Figure 10: Example of the utilization of the header data. Water flow and heat flux around the RS
F-elements; the water flow looks even, but heat flux is inclined, which indicates build-
up or wrong burner alignment.
6 Summary and conclusions
Increased understanding of the process conditions in copper and nickel Flash Smelting has contrib-
uted into better understanding of cooling requirements in the Furnaces. Modern calculation meth-
ods, such as thermodynamic and CFD-programs connected to Outotec developed models, provide
valuable tools for fast evaluation and problem-solving in practical process design. Solid fundamen-
tal background together with process knowledge has made Outotec cooling solutions globally ac-
cepted.
Outotec has introduced several new applications for modern copper and nickel furnace cooling sys-
tems. Novel methods include; customized overall primary and secondary cooling design taking into
consideration various environments in different processes in different locations of the furnace and
monitoring methods for the process, furnace integrity and element condition while in operation.
Modern requirements for the Flash Furnace campaign lives to extend beyond 10 years is a clear
motivation to push our development efforts even further and continue to improve the safety, integ-
rity, materials, costs and environment of the Flash Smelting process and its cooling efficiency.

References
[1] G. Y. Lai, High-temperature corrosion of engineering alloys, ASM International, Materials
Park, Ohio, 1990, 231 p.
[2] ASM Handbook, Vol 13 Corrosion, ASM International, Materials Park, Ohio, 1992, 1415 p.
[3] K. H. Zum Gahr, Microstructure and wear of materials, Elsevier 1987, 560 p.
[4] M. McNallan, High-temperature corrosion in halogen environments, Materials Performance,
Sept. 1994, 54-57.
[5] P. Kofstad, High temperature corrosion, Elsevier, London, 1988, 558 p.
[6] N. N. Evseeva, A.G. Bergmann, Izvest. Sektora Fiz. Khim. Anal. Vol 21, 1952, 217.
[7] M. Mkip and P. Taskinen, Refractory Wear in Copper Converters: Part I. Blister-Copper-
Refractory Interactions, Scand. J. Metallurgy 9, 1980, 273-281.
[8] W. M. M. Huijbregts, R.G.I, Leferink, Latest advances in the understanding of acid dewpoint
corrosion: corrosion and stress corrosion cracking in combustion gas condensates, Anti corro-
sion methods and materials, Vol 51, No. 3, 2004, 173-188.
[9] S. Sankar, Effect of SO
3
on corrosion of process equipment in copper smelters, JOM, Vol 34,
1982, 43-47.
[10] T. Ahokainen and K. Eklund, Flexibility of the Outokumpu Flash Smelting for Low and
High Grade Concentrates Evaluation by CFD-Modeling, COM2007, 25-30 August, 2007,
Toronto, Canada.
[11] T. Ahokainen, A. Jokilaakso, P. Taskinen and M. Kyt, A New Advanced CFD Model for
Flash Smelting and Converting Processes, Sohn International Symposium, F. Kongoli, Ed.,
San Diego, CA, 27-31 August, 2006, Vol. 8.
[12] T. Mkinen and T. Ahokainen, 50 years of nickel flash smelting still going strong,
COM2009, 23-26 August, 2009, Sudbury, Canada.


Introduction of a Slide Gate System for Copper
Anode Furnaces
Dipl.-Ing. Klaus Gamweger
RHI AG
Magnesitstrae 2
Leoben, Austria

Keywords: Taphole, slide gate, productivity, improvement, energy, emission, CO
2

Abstract
The anode furnace is the final vessel used for the pyro-metallurgical production of copper, whereby
anode grade copper is produced by fire refining the metal. Subsequently, high quality anodes, for the
electrolytic copper refining step, are cast in open moulds on large revolving casting wheels contain-
ing approximately 20 to 30 moulds. Typically, the copper is poured via a taphole into a launder sys-
tem by rotating the furnace into an appropriate position. Control of the casting rate is also performed
in this manner. However, this procedure has many disadvantages including splashing, oxygen pick-
up and energy losses.
This paper introduces an innovative solution to open and close the taphole with a slide gate system.
Two plates with defined drilled holes are moved against each other, which enables effective starting
and stopping of the copper flow as well as throttling. The complete system is described in detail,
including the refractory parts like the nozzles, well blocks, and plates as well as the mechanical
components for accurate movement and secure fastening of the plates themselves. Since all these
parts are in direct contact with molten copper, they have been especially developed to withstand
severe mechanical and thermal stresses as well as chemical attack.
Energy saving potentials in addition to safety and quality issues are discussed on the basis of a pilot
project run in Belgium.
Gamweger
1 Introduction
Controlling the casting rate of a copper anode furnace is usually performed by rotating the furnace
into an appropriate position. This means tapping starts with the taphole at a very high position.
Copper splashing and uncontrolled oxygen pick-up are in addition to several other disadvantages
the consequences. RHI AG and Stopinc AG have developed an innovative solution for copper ap-
plications to open and close the taphole with a slide gate system, which also enables accurate con-
trol of the flow rate. By using this system it is possible to start the casting procedure with the ta-
phole at the lowest position, directly above the launder.
This paper presents the new slide gate system in detail and demonstrates both the advantages in
production and the safety issues on the basis of a pilot project in Belgium.
2 Refractory Components
The taphole area is exposed to severe erosion attack in addition to the thermo-chemical load, which
affects the vessel refractories in general. RHI AG offers highly resistant alumina-chromium oxide
refractories, specially developed for the copper industry, to withstand the severe conditions.

well block

upper nozzle

upper plate

lower plate / sliding plate

collector nozzle
Figure 1: Refractory taphole components of the slide gate system.
Introduction of a Slide Gate System for Copper Anode Furnaces
The brick lining surrounding the taphole is installed in two flat arches above and underneath it. This
provides the best stability in case of a partial repair of the taphole arrangement. The interface be-
tween the lining and the well block comprises a low cement alumina-chromia castable.
The actual taphole shown in Figure 1 consists of a well block, an upper nozzle, the two plates
(where the lower one is sliding) and a collector nozzle. Typically, the well block has the same life-
time as the rest of the lining and stays in the furnace whereas the plates and nozzles have to be ex-
changed on a regular basis. However, the base steel plate has a special screw-fastened opening to
also provide the possibility for a well block change, if necessary.
The upper nozzle and the plates are manufactured from a high-fired alumina-chromia refractory,
which was developed especially for the requirements in the copper industry. Additionally, they are
tar impregnated for enhanced infiltration resistance. The abrasion wear is most severe during casting
in the throttling mode, when the tapping channel is not fully opened.
The lower nozzle, also termed the collector nozzle, produces a concentrated copper stream and
minimizes both oxygen pick-up and splashing. This nozzle is manufactured from a SiC material,
which should prevent droplet freezing at the end of the nozzle.
Appropriate selection of the mortars for installation has an important influence on the system life-
time. All parts have to be solidly connected to prevent any copper infiltration in between the indi-
vidual components. Two different mortars are used: (1) a phosphor bonded alumina-chromia mortar
with very high strength and adherence to connect the plates and nozzles to each other, and (2) a SiC
mortar to secure the inner nozzle into the well block. Cleaning the well block surface after removing
the worn out inner nozzle is easier with the SiC mortar and can be performed without damaging the
well block.
3 Mechanical Parts
The Interstop COP FLOW Slide Gate is a compact system, engineered with only a few components
and containing heavy-duty refractory parts. It is a high performance unit with enormous yield poten-
tial that enables rapid and simple exchange of the refractory parts. The slider can be swung out
laterally for plate exchange and the entire slide gate unit can be laterally swung away to change the
well block. Short furnace down times and consequential energy savings are a result of all the
aforementioned design attributes.
3.1 System Description
The slide gate Type CCS (Copper Compact System) is a two plate, linear, hydraulically driven
system. There are three different sizes available: CCS 60, CCS 80 and CCS 100. The number indi-
cates the maximum taphole diameter in mm.
Gamweger
The bottom and slider refractory plate are identical in shape and size and can be exchanged very
rapidly. This is performed by swiveling out the movable slider over a hinge on the housing
(Figure 2). After plate exchange, the slider is swiveled back into the normal working position. If the
well block also has to be changed, the entire ladle gate unit can be swiveled over a double hinge.
Figure 2: Opened Slide Gate Type CCS60.
The interlocked positioning of the plates is achieved by a double wedge design. Two permanent
magnets hold the plates in place.
The defined tension force of the system is generated through four preset tension elements, which are
equipped with high temperature resistant spiral disc springs and integrated into the housing. Moving
the slider hydraulically over the ramps into the working position activates tensioning of the system.
The steel encased refractory plates and exchangeable nozzle are delivered ready to use. No tools are
needed and there is no additional work in a workshop. The standard nozzle for copper has a shoul-
der and is set to a mechanical stop. Therefore, the position of the nozzle and the mortar joint thick-
ness are exactly defined.

The short build-in height of the Copper Slide Gate Type CCS does not require expensive structural
modifications to the existing installations.
3.3 System Advantages
The system consists of only two main parts: A laterally swiveling slider for easy handling of the
plates and a fixed or laterally swiveling housing, which includes the four tension elements and
two roller supports.
The self-tensioning system through the extra mounting stroke means no further hydraulic mount-
ing tool is needed.
The system can be operated in both a vertical and horizontal position and is therefore suitable for
any furnace type.
The long working stroke guarantees a higher refractory plate life.
Increased safety is achieved by high performance refractory components.
The refractory parts can be exchanged rapidly and easily without any additional mechanical fix-
ing.
Only a few spare parts are required, which minimizes maintenance and operational costs.
4 Industrial Application
The slide gate system was installed onto a 140 tonne anode furnace to upgrade a standard taphole.
The initial reason to install the slide gate was the short distance between the anode furnace and the
casting wheel with an intermediate ladle and a casting ladle. Due to the copper splashing consider-
able cleaning was required between each heat.
As a result of its particular location taphole and poling tuyeres are on the same side of the furnace
the taphole is continuously submerged during poling and casting. The total immersion time is 7 to
7.5 hours per heat and approximately 80 heats are processed per month.
4.1 Installation
The slide gate was fitted during the installation of a new brick lining. It was possible to place the
slide gate system at the original taphole position, so no significant adjustments to the steel structure
were necessary. Only the base plate of the slide gate had to be welded to the steel shell. The refrac-
tory lining was completed with two flat arches above and underneath the taphole opening. The well
block was put on place with its centreline exactly positioned. The interface between the well block
Gamweger
and the surrounding lining was cast. The slide gate was mounted on the base plate and nozzles and
plates were installed.
For safety reasons it must be possible to move the slide gate above the copper bath if an emergency
occurs. In case of power failure, a back-up power generator is available at the anode furnace. Addi-
tionally, the hydraulics are installed in such a way that the slide gate closes automatically in the case
of an emergency.
Due to the special situation in this furnace slide gate on the same side as the tuyeres the slide
gate is also immersed during reduction. To prevent copper freezing in the tapping channel during
this time, the lower plate is equipped with nitrogen purging. For this reason this moving plate has a
thin channel that is connected to a nitrogen pipe. When the slide gate is closed the gas outlet lies
exactly in the tapping channel so a constant flow of nitrogen keeps the channel open.
Due to the short distance between slide gate and casting launder, adequate heat protection was in-
stalled. This prevents overheating of the slide gate, hydraulic system and nitrogen pipes.
Figure 3: Installation of the slide gate system.
4.2 Results
In the original setup, prior to the slide gate installation, casting was started with the casting hole at a
distance from the launder and as the anode furnace was emptied during the cast the casting hole was
rotated downwards. However, a large accretion built up in the launder. As a result of the batch wise

work, this accretion caused a lot of significant copper return and damage to the launder when it was
removed. Since the slide gate enables casting to start with the casting hole at its lowest point, this
build-up in the launder was avoided. In addition, the slide gate almost completely eliminated any
splashing and the cleaning was minimal. Furthermore, because the slide gate always enabled casting
from the lowest point in the anode furnace, no slag was entrained in the anodes. As a result, an im-
provement in the surface quality of the anodes was observed.
The slide gate allowed the copper flow from the anode furnace to the intermediate ladle to be regu-
lated, thereby eliminating any repeated up and down rotation of the anode furnace. If the intermedi-
ate ladle had been on load cells, it would have been plausible to automate this stage. It was also pos-
sible to stop the copper flow for a short while and restart casting without any problems.
The biggest concern was replacement of the well block. At the beginning of the trial phase the inner
nozzle had to be changed quite frequently and the well block was partly damaged. However, by us-
ing an appropriate mortar for installation of the inner nozzle the damage was minimized and the
inner nozzle was even repaired. During the two years lifetime of the anode furnace the well block
has not had to be changed.
Figure 4: Slide gate with heat protection shield in operation.

Gamweger
4.3 Further developments
The most significant challenge at the beginning of the trial phase was the inner nozzle and plate
wear. This caused the copper to infiltrate between the plates and then subsequently the slide gate
was blocked. However, the wear was mainly caused by lancing the taphole with oxygen. Therefore,
the main requirement was to avoid burning the taphole. To address this issue, a nitrogen purging
possibility was installed in the outer plate, positioned to be in front of the taphole during filling and
poling. This kept the taphole open or during sporadic heats left only a small part to be lanced.
Thereby the wear was significantly reduced.
5 Summary
It was demonstrated that the slide gate system specifically adapted for the copper industry met all
operational and safety requirements. Increased productivity as well as decreased fuel consumption
and CO
2
emissions due to reduced rejects were also realised. The slide gate almost completely
eliminated copper splashing and the cleaning was minimal. The experiences with this first installa-
tion indicate the significant potential for slide gate applications on different vessels in the copper
industry.

New Highly Efficient Rotary Furnace for
Environmentally Friendly Refining Process
Dr. Bernhard Hanusch
KME Germany AG & Co. KG
Klosterstrae 29
D - 49074 Osnabrck, Germany

Keywords: Rotary furnace, copper, casting, refining, environment, optimization, development
Abstract
In the beginning of 2008 a modern rotary furnace has been started working at the foundry of the
KME Germany plant at Osnabrck. Subsidies from the Federal Agency for Environmental
Protection helped to make a very important step towards the substitution of old refining processes.
The rotary furnace is run efficiently with natural gas and oxygen. Thus a high melting capacity and
optimized process gas capture can be harmonized. The use of porous plugs together with the
modern refining processes result in a high degree of purity in copper recycling. Internal returns as
well as higher polluted scraps from the market can be integrated into the complex production
process at KME in an environmentally friendly way.
1 KME The Company
The KME Germany AG & Co. KG is a member of KME Group S.p.A. in Florence, one of the
leading producers of copper and copper alloy semis, and a leader in the development of innovative,
application-oriented solutions. KME is an expert partner to many diverse industries and specialised
in fulfilling individual customer needs. KME manufactures for the national and international
markets at production sites and sales places in the major countries and markets in Europe and
China. KMEs extensive product range comprises quality products and technologically leading
special solutions for the needs of the most diverse industrial fields.
Hanusch

1.1 Products and applications
The product range of KME comprises an entire portfolio of semi-finished and finished products
made of copper and copper alloys for a large number of applications. It is constantly developing
innovative and economical solutions for proven and new copper applications, in close collaboration
with customers from all fields of industry.
Rolled products for industry and building
Copper tubes for plumbing systems
Moulds for melting and casting techniques
Special and semi-special products such as tube bundles and piping systems for maritime
applications
1.2 KME: Innovations in copper
Todays world would be inconceivable without copper. Electricity supply, air-conditioning and
temperature regulation technologies, transport and telecommunication, alternative energy sources
all of these are taken for granted in modern everyday life, and they owe their high standard of
quality to some degree to the intelligent use of copper. Copper has become essential for our
technological development. Its position as a pioneering material is undisputed. As a natural
commodity, copper is environmentally friendly and 100 % recyclable. It is extremely durable, not
ageing or susceptible to corrosion, and this makes it the material of choice for numerous
applications.

New Highly Efficient Rotary Furnace for Environmentally Friendly Refining Process
2 Osnabrck Foundry Layout, Equipment, Raw Materials,
Products
2.1 Copper Refinery & Casting
Figure 1: Copper Refinery & Casting

Hanusch

2.1.1 Raw Material Input:
Asarco
o several cathode qualities
o external high grade scrap
o internal Copper rejects and new scrap
Thomas Furnaces
o external high grade scrap
o internal alloy scraps and slags
New Refining Furnace
o external low grade scrap
o internal alloy scraps and slags
2.1.2 Products Copper Casting:
Billets 115-507 mm; 17-50 t/h Casting Performance; 3-6 Strands; in several CU-Qualities,
DHP-Copper, ETP-Copper
Cakes from 720 to 1280 mm x 320 mm; 70 t/cast; 2 Strands; in DHP-/ETP-Copper
2.1.3 Internal Customers:
Billets for Tubes Plant Menden
Billets for Tubes Plant Osnabrck
Cakes for Rolling Mill Osnabrck
Billets/Cakes for Business Unit Specials Osnabrck
Total Foundry Capacity: 250,000 t/a


2.2 Alloy Plant 2 Lines

Figure 2: Alloy Plant, melting and casting
50 Alloys (CuCrZr, CuNi10/25/30, CuNiSi, CuAg, CuFe, CuMg, CuAl5Zn5Sn1) in
20 Dimensions (Billets up to 900 mm and Cakes up to 1280 mm)
> 120 Combinations
2.3 Bronze Strip Casting 3 Lines

Figure 3: Bronze Plant, melting and casting
Products: CopperTinBronce CS04 / 06 / 08
Casting Performance: 1.2-2.3 t/h per line
Width: 720-1280 mm
Melting &
Casting
Furnaces
Semi
Continuous
Casting
Melting &
Casting
Furnaces
Semi
Continuous
Casting
Melting & Casting
Furnaces
Coiler Withdrawal
Unit
Cutter
Melting & Casting
Furnaces
Coiler Withdrawal
Unit
Cutter
Hanusch


Holding Furnaces
Continuous Casting
Holding Furnace
Semi Continuous
Casting
Asarco &
Holding
Furnace
Thomas
Furnaces
1 - 3
diffusive
emissions
Holding Furnaces
Continuous Casting
Holding Furnace
Semi Continuous
Casting
Asarco &
Holding
Furnace
Thomas
Furnaces
1 - 3
diffusive
emissions
3 Change in Foundry Layout due to Diffusive Emissions
3.1 Refinery layout until 2007

Figure 4: Layout Copper Refinery until 2007

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

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


burner side
off gas side
gas
air
oxygen
burner side
off gas side
gas
air
oxygen
3.3 New Refining Furnace Concept for Copper Refinery
3.3.1 Process and Furnace Layout

Figure 6: Copper Refinery: New Refining Furnace Process
3.3.2 Features
rotating & tilting
Oxygen metallurgy
Porous plugs
Flexibility (time & raw materials)

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

2
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 pro-
ductivity and requires detailed knowledge of the slag phase equilibria. However, limited data are
available on the thermochemistry of the multi-component slag system Cu
2
O-FeO-Fe
2
O
3
-SiO
2
-
CaO-MgO-Al
2
O
3
-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 charac-
terization 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 ex-
Henao, Hayes, Jak
traction processes. In view of this, extensive experimental information regarding the sulphur solu-
bility 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 sul-
phur content of slags corresponds to gas/slag/matte equilibrium obtained both from industrial prac-
tice 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 alu-
mina 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 propor-
tions. 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 SiO
2
, MgO (or Al
2
O
3
) crucibles put significant restrictions on the range of
CaO/SiO
2
and SiO
2
/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 tech-
nique 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-Fe
2
O
3
-SiO
2

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.

2
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-Fe
2
O
3
-
CaO-SiO
2
-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 sys-
tem in support of sustainable copper smelting and converting technologies, the experimental infor-
mation 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).
Henao, Hayes, Jak
Figure 1: Example of initial mixture for phase equilibrium determination using the present tech-
nique.
All the equilibration experiments were conducted in a vertical reaction tube (impervious re-
crystallized 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 abso-
lute temperature accuracy is estimated to be within 5 K. The atmosphere within the reaction tube
was maintained at fixed oxygen and SO
2
partial pressures using CO/CO
2
/SO
2
(CO 99.5 wt. % pure,
CO
2
99.99 wt. % pure, 99.9 wt. % SO
2
) mixtures. The flow rates of gases to the furnace were con-
trolled using glass capillary flow meters with the gas flowing from the bottom to the top of the fur-
nace. 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 DS-
type 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 meas-
urements in the present study are within the accuracy of the DS-type oxygen probe, i.e. close to log
P
O2
of + 0.1 units (P
O2
in atm) [32].

2
Table 1: Gas composition and results of DS-type oxygen probe measurements of CO/CO
2
gas
mixtures.
Temperature
(
o
C)
Gas Mixture
(%Vol.)
Calculated Partial Pressure
(FactSage 5.3) P(atm)
LogP
O2
Measured
by probe P(atm.)
CO CO
2
SO
2
P
SO2
logP
O2
logP
S2

1300 12.5 76.5 11.0 0.1 -8.0 -2.4 -7.98
1250 1.9 88.0 10.1 0.1 -7.0 -5.2 -6.98
1250 6.0 83.9 10.1 0.1 -8.0 -3.2 -7.93
1250 10.5 27.3 62.2 0.6 -8.0 -1.7 -7.84
1250 19.2 57.6 23.2 0.18 -8.5 -1.6 -8.41
The starting mixtures were made from CaO, SiO
2
, Fe
2
O
3
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/Fe
2
O
3
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 ex-
periments 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 (CaSiO
3
) for Si and Ca, hematite (Fe
2
O
3
) for
Fe and CaSO
4
for S. The compositions were measured to accuracy within 1 wt. % for SiO
2
, CaO
and Fe
2
O
3
and an accuracy of 0.1 wt. % for S.
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-SiO
2
-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 SO
2
partial
pressures ranges of interest for copper smelting. Two temperatures 1300 and 1250
o
C were selected.
Oxygen partial pressure of 10
-8
atm and sulphur partial pressure of 0.1 atm were employed for the
experiments at 1300
o
C. Four sets of experiments were carried out at 1250
o
C for a range of oxygen
partial pressures from 10
-7
to 10
-8.5
atm and P
SO2
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
o
C
One set of experiments was carried out at 1300
o
C, P
O2
of 10
-8
atm and P
SO2
of 0.1 atm. The pseudo-
ternary section constructed by projection of the slag compositions onto the FeO-SiO
2
-CaO plane
is given in Figure 3. The recalculated liquid compositions were determined by normalizing the ori-
ginal compositions measured with EPMA considering only the relative FeO, CaO and SiO
2

2
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/SiO
2
or at a given SiO
2
/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 SO
2
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 tem-
perature of 1300
o
C, P
O2
of 10
-8
atm and P
SO2
of 0.1 atm.

Henao, Hayes, Jak
3.2 Experimental results at 1250
o
C
Experiments at 1250
o
C were conducted at P
O2
varied from 10
-7
to 10
-8.5
atm, and P
SO2
from 0.1 to
0.6 atm to evaluate the influence of P
O2
and SO
2
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 P
O2
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 P
SO2
partial pressure from 0.1 to 0.6 atm. Also, it was observed that spinel is unstable at
P
O2
of 10
-8
atm and P
SO2
of 0.6 atm and at P
O2
of 10
-8.5
atm, and P
SO2
of 0.18 atm.
4 Discussion
4.1 Liquidus
It was observed in Figures 3 to 6 that P
SO2
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 P
SO2
spinel is unstable. This behaviour has important implications on the copper smelt-
ing operation from the viewpoint of process control and refractory protection. The diagrams pre-
sented in Figures 3 to 6 can now be used to take into account effect of sulphur in process improve-
ment.
4.2 Sulphur capacity
4.2.1 Definition
Sulphur may dissolve in slag according to the exchange reaction suggested by Richardson and
Withers [1],
(O
2-
)
(slag)
+
2
1
S
2(gas)
= (S
2-
)
slag
+
2
1
O
2(gas)
(1)
2
perature of 1250
o
C, P
O2
of 10
-7
atm and P
SO2
of 0.1 atm.
which gives, within the limits of dilute solution, the relation:
Cs
2-
=(wt. % S
2-
)(P
O2
/P
S2
)=(Ka
O2-
)/f s
2-
(2)
where Cs
2-
is the sulphide capacity of slag, wt. % S
2-
is weight pct of sulphur in the slag, P
O2
and P
S2

are the respective partial pressures of oxygen and sulphur in the gas phase in equilibrium with the
slag and K, a
O2-
and f s
2-
represent the equilibrium constant of Equation 1, the activity of O
2-
in the
slag and activity coefficient of S
2-
in the slag respectively. The values of K, a
O2-
and f s
2-

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.

Henao, Hayes, Jak
perature of 1250
o
C, P
O2
of 10
-8
atm and P
SO2
of 0.1 and 0.6 atm.

2
perature of 1250
o
C, P
O2
of 10
-8.5
atm and P
SO2
of 0.18 atm.
4.2.2 Sulphur capacity at 1300
o
C
The values of sulphur capacity at 1300
o
C calculated using Equation 2 are represented in Figure 7
showing the logarithm of Cs
2-
vs. SiO
2
/FeO wt. % ratio at four levels of CaO/SiO
2
wt. % ratios.

Henao, Hayes, Jak
Figure 7: Relationship between sulphur capacity and SiO
2
/FeO wt. % ratio.
It is apparent that there is an approximately linear relationship. A similar linear tendency was ob-
served for the correlation between the logarithm of Cs
2-
vs. CaO/SiO
2
wt. % ratio. Thus, from the
experimental data for a range of compositions, a mathematical relationship was obtained to express
the logCs
2-
as a linear function of SiO
2
/FeO wt. % ratio and CaO/SiO
2
wt. % ratio as is indicated in
Equation 3,
log Cs
2-
(1300 C) = -1.14 (wt. % SiO
2
/wt. % FeO) + 0.21 (wt. % CaO/wt. % SiO
2
) - 2.18 (3)
R
2
= 0.97
where -1.14, 0.21 and -2.18 are parameters empirically determined for the 1300
o
C and the range of
SiO
2
/FeO wt. % ratio from 0.05 to 1.9, and the range of CaO/SiO
2
wt. % ratio from 0 to 1.3. Note
that regression coefficient R
2
= 0.97 is close to 1 which confirms that the relationship of
log Cs
2-
(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-SiO
2
-CaO at 1300
o
C. The iso-sulphur capacity (1000 Cs) lines
and the experimental results are shown in Figure 8.

2
Figure 8: Effect of slag composition on sulphur capacity: Fe-Si-Ca-S-O system at fixed tempera-
ture of 1300
o
C, P
O2
of 10
-8
atm and P
SO2
of 0.1 atm.
4.2.3 Sulphur capacity at 1250
o
C
Figure 9 shows the experimental results at 1250
o
C for the whole range of P
O2
and P
SO2
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 be-
tween the sulphur capacity and slag composition.

Henao, Hayes, Jak
log Cs
2-
(1250 C) = -1.28 (wt. % SiO
2
/wt. % FeO) + 0.015 (wt. % CaO/wt. % SiO
2
) - 2.41 (4)
R
2
= 0.95
where -1.28, 0.015 and -2.41 are parameters empirically determined for the 1250
o
C and range of
SiO
2
/FeO from 0.2 to 1.5, and the range of CaO/SiO
2
wt. % ratio from 0 to 1.3. Again, R
2
close to 1
confirms linear character of the relationship between logCs
2-
(1250 C) and selected composition ratios.
Figure 9: Relationship between sulphur capacity and SiO
2
/FeO wt. % ratio.
Fe-Si-Ca-S-O system at the following fixed conditions of P
O2
and P
SO2
: a) -P
O2
of 10
-8
and P
SO2
of
0.1 atm. b) -P
O2
of 10
-8
and P
SO2
of 0.6 atm. c) -P
O2
of 10
-8.5
and P
SO2
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-SiO
2
-CaO at 1250
o
C. The
iso-sulphur capacity lines along with the obtained experimental results are shown in Figure 10.

2
Figure 10: Effect of slag composition on sulphur capacity: Fe-Si-Ca-S-O system at the following
fixed conditions of P
O2
and P
SO2
: a) - P
O2
of 10
-8
and P
SO2
of 0.1 atm. b) - P
O2
of 10
-8
and
P
SO2
of 0.6 atm. c) - P
O2
of 10
-8.5
and P
SO2
of 0.18 atm.
4.2.4 Comparison and previous investigations and thermodynamic models
The trends of Cs
2-
indicated by Equations 3 and 4 are similar to those determined at 1500
o
C by
Fincham and Richardson [5]. The agreement is good, within experimental uncertainties, with the
experimental data reported by Simeonov for the system FeO-SiO
2
at the condition of silica satura-
tion as indicated in Figures 8 and 10 at 1300
o
C and 1250
o
C, respectively [23].

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 con-
sidered in the interpretation of the Simenovs data.
The experimental results obtained by Jalkanen at 1250
o
C 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 liq-
uidus and the sulphur capacities at similar oxygen and sulphur partial pressures [12]. Large scatter-
ing 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 logCs
2-
vs. the
reciprocal temperature are linear with a negative slope. The data represented in Figures 7 and 9 in-
dicate that at a given slag composition, Cs
2-
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-SiO
2
-CaO ternary system at 1600
o
C agree with the trends identified in the present study. In
both studies logCs
2-
is a linear function of SiO
2
/FeO and CaO/SiO
2
wt. % ratios.
5 Conclusions
The established equilibration/quenching/EPMA technique has been used to accurately experimen-
tally investigate phase equilibria of the Fe-Si-Ca-O-S slag system at conditions relevant to the cop-
per smelting operation. A range of conditions has been selected to address industrial interest re-
quired to characterise and describe copper slag chemistries. The experimental results have been
reported in the form of the FeO-CaO-SiO
2
phase diagrams. The influences of SO
2
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 P
SO2
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 composi-
tions. 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
o
C and 1300
o
C) and slag composition (silica saturation) with published experi-
mental results. The sulphur capacities calculated with the present experimental results are in a rea-
sonably good agreement with those measured using the gas/slag equilibrium and in disagreement
with the experimental results obtained using the gas/slag/matte equilibrium.
2
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 Univer-
sity of Queensland; Mr. Ron Rasch and Mr. Ying Yu from the Centre of Microscopy and Micro-
analysis (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 Interna-
tional 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.
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.
Dsseldorf.
[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 Eisenhtten (VDEh), Page 126.
[35] Slag Atlas Slag Atlas (1995): Verein Deutscher Eisenhtten (VDEh), Page 260.

TM
Slag Chemistry at
Southern Peru Ilo Smelter
Enrique Herrera, Leopoldo Mariscal
Southern Peru Copper Corporation (SPCC)
Fundicin Ilo, Punta Tablones S/N, Pacocha, Ilo
Moquegua, Peru

Keywords: Pyrometallurgy, copper, slag chemistry, ISASMELT
TM

Abstract
Ilo Copper Smelter has been operating since February 2007 with an ISASMELT
TM
furnace as a sin-
gle 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 ISASMELT
TM
furnace. The sea shell flux addi-
tion was suspended while the SiO
2
/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 ISASMELT
TM
has led to lower copper losses and get a hig-
her 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 ISASMELT
TM
furnace as a single primary
smelting unit, which is associated with two Rotary Holding Furnaces RHFs. The Figure 1 shows
the smelter flow diagram.
Herrera, Mariscal
Figure 1: Flow diagram of the Ilo Smelter
The ISASMELT
TM
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 ISASMELT
TM
furnace and the batch
Peirce Smith Converters (PSC) cycles.
The design specifications of the ISASMELT
TM
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 ISASMELT
TM
slag chemistry. The main
purpose of this change was to avoid leaving the undissolved silica from ISASMELT
TM
furnace.

TM
Slag Chemistry at Southern Peru Ilo Smelter
2 Fundamental Considerations
The slag chemistry modifications of the ISASMLET
TM
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 Al
2
O
3
of 6 wt. % with focus on the range of SiO
2

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+SiO
2
), FeO/(CaO+FeO+SiO
2
) and SiO
2
/(CaO+FeO+SiO
2
). The CaO/SiO
2
and
SiO
2
/Fe are also shown for convenience.

Figure 2: Liquidus in the system Al
2
O
3
-CaO-FeOSiO
2
at P
O2
= 10
-8.4
atm and Al
2
O
3
= 6 wt. %
The main results of the first SPCC slag chemistry modification are shown in Figure 2. The black
circle represents the design composition (SiO
2
/Fe = 0.88 and SiO
2
/CaO = 7.0) and the liquidus

Herrera, Mariscal
temperature is predicted to be around 1200 C. With the first slag chemistry modification the
SiO
2
/Fe ratio was reduced to 0.82 and it was necessary to raise the ratio SiO
2
/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 ISASMELT
TM
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 ISASMELT
TM
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-SiO
2
interaction in the liquid slag is stronger than the FeO-SiO
2
interaction. In other words,
the CaO increases the slag liquidus by enhancing the magnetite formation if there is not enough
silica.
2 [CaO] slag + [SiO
2
] slag = [Ca
2
SiO
4
] slag (1)
2 [FeO] slag + [SiO
2
] slag = [Fe
2
SiO
4
] slag (2)
3 [FeO] slag + O
2
= (Fe
3
O
4
) 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
ISASMELT
TM
furnace, increasing the efficiency of the oxygen utilization in the smelting bath.
3 Isasmelt
TM
Slag Chemistry Modifications
In order to decrease the amount of slag generated in the ISASMELT
TM
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 SiO
2
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 ISASMELT
TM

furnace until the total suspension.
TM
Table 1: Program to reduce the addition of the sea shell flux, for a concentrate feed rate of 160 tph
Sea Shell [tph] SiO
2
/ 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 SiO
2
/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 ISASMELT
TM
, which shows the interaction between them.
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 SiO
2
/Fe ratio was
reduced to 0.72 on average, while the SiO
2
/CaO ratio was higher than 20, and depended mainly of
the lime content in the concentrates smelted. The lower consumption of fluxes in the ISASMELT
TM
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.

TM
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 SiO
2
from the slag.

Herrera, Mariscal
Figure 5: The SiO
2
/Fe ratio versus magnetite contents in ISASMELT
TM
slag
In the first histogram of the Figure 6, it can note that with the interruption of the sea shell flux
added to the ISASMELT
TM
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: Histograms of copper and magnetite in ISASMELT
TM
slag

TM
The Figure 7 shows the copper content in the discarded slag from the ISASMELT
TM
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 ISASMELT
TM
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.

Herrera, Mariscal
Figure 8: ISASMELT
TM
Furnace Refractory Wear Trend II Campaign
Figure 8 shows that the rate bricks wear in the ISASMELT
TM
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 ISASMELT
TM
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 ISASMELT
TM
furnace, oxygen demand and coal were reduced
significantly.
4 Conclusions
The reduction of the CaO content in the ISASMELT
TM
slag by dropping simultaneously the SiO
2
added to the furnace has proven its effectiveness and negative effect of the CaO-SiO
2
interaction in
the liquid slag.
With this new chemistry of the ISASMELT
TM
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 %.

TM
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
ISASMELT
TM
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): Isasmelt
TM
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.


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 ex-
perimentally studied over a wide range of slag compositions, temperatures and atmospheric condi-
tions. The systems selected are of industrial interest and fill gaps in fundamental information re-
quired to characterise and describe copper slag chemistries. The experimental procedures developed
at the Pyrometallurgy Research Centre (PYROSEARCH) have been used, which involve equilibra-
tion of mixtures at high temperatures, rapid quenching of resulting phases, and accurate measure-
ment of phase compositions using electron probe X-ray microanalyses. It has been shown that fer-
rous calcium silicate (FeO-CaO-SiO
2
) slags can be characterised over a range of temperatures and
oxygen partial pressures relevant to copper production operations. The influences of Al
2
O
3
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 measure-
ments 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 non-
ferrous smelting and converting processes operate with the use of slag systems containing princi-
pally ferrous calcium silicate components (FeO-Fe
2
O
3
-CaO-SiO
2
or FeO-CaO-SiO
2
). Other com-
ponents can also present in the slags, such as impurities that are introduced in by the process feeds
(Al
2
O
3
, MgO) or minor elements and valuable metal that are oxidized during the operations
(Cu
2
O).
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 infor-
mation 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 op-
erations, 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 sys-
tems 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 con-
tamination 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 Py-
rometallurgy 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 py-
rometallurgical 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 descrip-
tion 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
o
C
from 99.0 wt. % pure CaCO
3
powder), SiO
2
(99.99 wt. % pure 1-3 mm fused lump that had been
ground with an agate mortar and pestle), Fe
2
O
3
powder (99.99 wt. % pure), MgO powder
(99.95 wt. % pure), Al
2
O
3
(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 ox-
ide/metal powders and mixing them thoroughly using an agate mortar and pestle. The initial compo-
sitions of the mixtures were selected such that at equilibrium there would be liquid phase in equilib-
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 tech-
nique [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 compo-
sition of the mixture (point b, in Figure 1), the compositions of phases at equilibrium in the multi-
phase equilibrium conditions (compositions of liquid, point a, and solid phases, point c, in Figure 1)
will be unchanged.


Figure 2: Crucible designs and suspension:
(a) platinum crucible; (b) tridymite/CaSiO
3
crucible; (c) spinel crucible
If investigation of slag in equilibrium with molten copper metal is to be carried out, the use of plati-
num 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, CaSiO
3
and spinel. The tridymite crucible was made from SiO
2

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
o
C for 24 hours. The CaSiO
3
crucible was
made by mixing an equimolar proportion of CaO and SiO
2
powders. This mixture was pressed into
the same shape as that for tridymite crucible. The CaSiO
3
crucible was then heated at around
1450
o
C for 1 day. The spinel (Fe
3
O
4
) 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
o
C in a CO/CO
2
stream resulting in an
oxygen partial pressure of 10
-6
atm.

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
o
C of
the target temperature. It is estimated that the overall absolute temperature accuracy of the experi-
ment is + 3
o
C.

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/CO
2
ratio. In this study, equilibrations were carried out at oxygen partial pressures be-
tween 10
-5
and 10
-8
atm. These oxygen partial pressures required very low CO/CO
2
ratios, thus pre-
mixed CO and Ar gases were used to achieve the target CO/CO
2
ratios. 5 % and 20 % CO diluted in
high purity Argon (Beta standards, 0.02 % uncertainty) mixtures were supplied by BOC and high
purity CO
2
(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

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 DS-
type 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 accu-
racy 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 Cu
Metal
+ O
slag, Gas
= Cu
2
O
Slag
(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 atmos-
phere. 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/CO
2
gas mixtures
Temperature
(
o
C)
Gas Mixture
Target
Log ( )
Log ( )
Measured
by Probe
1200 Stream of CO
2
-3.9 -3.8
1200 CO/CO
2
= 0.00316 -6.0 -5.9
1200 CO/CO
2
= 0.03164 -8.0 -8.0
1300 Stream of CO
2
-3.5 -3.4
1300 CO/CO
2
= 0.00429 -5.0 -4.9
1300 CO/CO
2
= 0.01356 -6.0 -5.9
1300 CO/CO
2
= 0.04288 -7.0 -7.0
1300 CO/CO
2
= 0.1356 -8.0 -8.0
2.3 Equilibration technique
Equilibration experiments were carried out using a vertical tube furnace with design shown in Fig-
ure 3. The specimen with suitable container material was introduced from the bottom of a vertical
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
o
C 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 car-
ried 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 sam-
ple 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 m, and carbon coated for micro-analysis.
2.4 Analysis technique

Figure 4: Microstructures of the quenched ferrous calcium silicate slags in equilibrium with vari-
ous condensed phases: (a) spinel (FeO), (b) tridymite (SiO
2
), (c) pseudo-wollastonite
(CaSiO
3
) and (d) rankinite (Ca
3
Si
2
O
7
)

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 under-
taken. 15-kV accelerating voltage and 15 nA probe current were selected for the micro-analyzer
operation. The Fe
2
O
3
, MgO, Al
2
O
3
, CaSiO
3
(from the Charles M. Taylor Co., Stanford, CA), and
Cu
2
O (prepared in-house using Cu
2
O, 99.99 wt. % purity) standards were used for calibrations. Cor-
rections based on the DuncumbPhilibert atomic number, absorption, and fluorescence were ap-
plied. 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 differ-
ence valence states can be obtained using this technique. Consequently, all iron and copper oxide
concentrations in the present study are reported as FeO and Cu
2
O, 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, avoid-
ing 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-SiO
2
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-SiO
2
system at fixed oxygen partial pressures and temperatures
Significant work on FeO-CaO-SiO
2
at temperatures between 1200
o
C-1350
o
C, at = 10
-5
-10
-6
atm
was carried out by Nikolic et al. [10]. Using the established equilibration/quenching/EPMA tech-
nique, liquidus isotherms at fixed oxygen partial pressure for spinel, tridymite, and CaSiO3 (wollas-
tonite/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 CaSiO
3
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 experi-
ments 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
isotherm at 1300
o
C in the rankinite and dicalcium silicate phase fields, and at 1350
o
C in the silica
phase field.

Figure 5: Liquidus isotherms in the FeO-CaO-SiO
2
system at 1200
o
C, 1250
o
C, 1300
o
C, and
1350
o
C, at a =10
-6
atm (Data from the present study indicated by open circle and
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-SiO
2
system at temperature 1300
o
C 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
phase to phase during cooling and leads to disintegration of sample into small particles (the so-
called dusting phenomenon) [11]. The ability of EPMA to carry out analysis on small area up to
5 m facilitates the examination of these small particles. Figure 6 shows backscattered scanning

electron microscope (SEM) micrographs of quenched sample, with liquid in equilibrium with spinel
and dicalcium silicate (Ca
2
SiO
4
). 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 accu-
rately. 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,
and liquid phases (Equilibration carried out for 24 hours at 1300
o
C and =10
-6
atm)
The effect of oxygen partial pressure on the liquidus isotherm of FeO-CaO-SiO
2
system at
1300
o
C is shown in Figure 7 for in the range 10
-5
-10
-8
atm. Data of Nikolic at al. [10] are indi-
cated 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
stabilities. At =10
-8
atm wustite is stable; with the increase of oxygen partial pressure spinel
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-SiO
2
phase diagram forms the basis of the equilibria involving those
elements and knowing the nature of this basic diagram facilitates understanding on the more com-
plex multi-component systems.


Figure 7: FeO-CaO-SiO
2
system at 1300
o
C and =10
-5
atm, 10
-6
atm, 10
-7
atm, and 10
-8
atm
(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 Al
2
O
3
and MgO
Experimental results of phase equilibria of FeO-CaO-SiO
2
-Al
2
O
3
system at 1300
o
C and =10
-8
atm
are provided in Figure 8. The results are presented as projections onto the pseudo-ternary diagram of
FeO-CaO-SiO
2
. Using this representation, the influence of Al
2
O
3
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 Al
2
O
3
concentration in the liquid increases. Conversely, the
stability of solid iron oxide increases with the increase of Al
2
O
3
concentration in the liquid. Figure 9
shows the experimental results of phase equilibria of FeO-CaO-SiO
2
-MgO system at 1300
o
C and
=10
-8
atm. Again, the results are presented in pseudo-ternary diagram of FeO-CaO-SiO
2
. The

minor addition of MgO clearly moves the liquidus isotherms of FeO-CaO-SiO
2
slag. The liquidus
of tridymite and pseudo-wollastonite primary phases appear on MgO addition to behave in a similar
way to Al
2
O
3
. The information of the influence of Al
2
O
3
and MgO on the phase equilibria of
FeO-CaO-SiO
2
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-SiO
2
-Al
2
O
3
system at 1300
o
C and =10
-8
atm
projected into FeO-CaO-SiO
2
plane


Figure 9: Liquidus isotherm in the FeO-CaO-SiO
2
-MgO system at 1300
o
C and =10
-8
atm
projected into FeO-CaO-SiO
2
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 Cu
2
O-SiO
2
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. Cu
2
O-SiO
2
. The pseudo-binary

phase diagrams of the Cu
2
O-SiO
2
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 Cu
2
O-SiO
2
system:
(a) equilibrated in air ( =0.21 atm), (b) equilibrated with molten copper metal

The phase diagram for Cu
2
O-SiO
2
equilibrated with air is provided in Figure 10(a). In the sub-
solidus region, SiO
2
may present in equilibrium with CuO or Cu
2
O. The transformation of CuO to
Cu
2
O was measured using differential scanning calorimetry technique and was found to be
1030.6
o
C. The phase assemblage between Cu
2
O and SiO
2
is stable up to the eutectic point of
1054
o
C, determined by using quenching and microscopy examination technique. Slightly different
measurements of the eutectic temperature of 1060 + 10
o
C 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 pre-
vious investigators are significantly higher than the present result. For example, Berezhnoi et al. [1]
reported eutectic composition of around 92 wt. % Cu
2
O-8 wt. % SiO
2
, while the present result
shows eutectic located at around 87 wt. % Cu
2
O-13 wt. % SiO
2
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

out some measurements at this primary phase. He reported melting point of pure cuprite in air to be
1230
o
C; 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
o
C.
Figure 10(b) shows the Cu
2
O-SiO
2
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 Cu
2
O-SiO
2.
The stability area of fully liquid slag
equilibrated with molten copper metal is smaller than that in air. For example at 1300
o
C in the
presence of molten copper, tridymite starts to precipitate at 13.8 wt. % SiO
2
. In air at the same tem-
perature, the tridymite starts to precipitate after slag contains silica higher than 16.9 wt. %. Phase
assemblage between Cu
2
O-SiO
2
at copper saturation stabilizes to higher eutectic temperature of
1186
o
C [6]. The increase of eutectic point was confirmed by the present experiment at 1150
o
C
showing the presence of only Cu
2
O and SiO
2
solids. Measurement of the liquidus at cuprite primary
phase field was not carried out by the present and previous studies. Kuxmann and Kurre [6], how-
ever, predicted that the liquidus line would connect the eutectic point and the melting point of pure
Cu
2
O at copper saturation, i.e. 1236
o
C. 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 Cu
2
O-Fe
2
O
3
-SiO
2
systems
The phase diagram of Cu
2
O-Fe
2
O
3
-SiO
2
system in equilibrium with air is shown in Figure 11.
The primary phases present in the system include tridymite, cristobalite, cuprite, spinel and delafos-
site. The spinel and delafossite phase boundaries for the diagram were constructed using data from
the Cu
2
O-Fe
2
O
3
join [13]. The ternary eutectic temperature was determined to be 1040 +10
o
C,
using the quenching and microscopy examination technique.
Extensive measurements were carried out in the tridymite primary phase fields. The boundaries be-
tween 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 Cu
2
O-Fe
2
O
3
join.
Further work is now being undertaken to characterise these complex Cu
2
O-FeO-Fe
2
O
3
-SiO
2
and
Cu
2
O-FeO-Fe
2
O
3
-CaO-SiO
2
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.


Figure 11: Liquidus isotherms in the Cu
2
O-Fe
2
O
3
-SiO
2
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 ac-
curately study phase equilibria of silicate slag systems relevant to the copper smelting/refining op-
erations. 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 de-
scribe copper slag chemistries. The important influences of gas atmosphere, temperature, and addi-
tional 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-SiO
2
,
Cu
2
O-SiO
2
, and Cu
2
O-Fe
2
O
3
-SiO
2
. Moreover, the influences of Al
2
O
3
and MgO on the fer-
rous calcium silicate slag system have been shown in the form of projections onto the

FeO-CaO-SiO
2
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 Op-
erations, 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 CuO-
Cu
2
O-SiO
2
, 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 pre-
diction of complex phase equilibria in industrial non-ferrous slag systems, European metal-
lurgy conference 2007; Dsseldorf.
[6] KUXMANN U., KURRE K. (1968): Miscibility gap in the system copper-oxygen and its al-
teration by the oxides CaO, SiO
2
, Al
2
O
3
, MgO.Al
2
O
3
, and ZrO
2
, 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 de-
rived from equilibrium measurements in the system Cu-Pt-O at 1000 and 1200
o
C, Transac-
tions of the metallurgical society of AIME, 245: 791-796.
[9] MENDYBAEV R.A., BECKET J.R., STOPLER E., GROSSMAN L. (1998): Measurement of
oxygen fugacities under reducing conditions: Non-nernstian behavior of Y
2
O
3
-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
o
C to 1350
o
C,
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 Cu
2
O-FeO-Fe
2
O
3
-CaO-
SiO
2
slags at 1250
o
C and 1300
o
C 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.

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].
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 CO
2
gases.
This paper introduces outline of the improvements made for FSF by the SPI project and some modi-
fication 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 con-
ducted at the maximum feed rate of 215 dry-t/h. Recent matte grade is 6368 %. 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 sul-
furic acid production plant.
Figure 1: Schematic flow sheet of Saganoseki Smelter & Refinery

Conc. Flux Coke Powder Conc. Flux Coke Powder Conc. Flux Coke Powder Conc. Flux Coke Powder
Blending Yard Blending Yard Blending Yard Blending Yard
Dryer Dryer Dryer Dryer
Slag Flotation Slag Flotation Slag Flotation Slag Flotation
Concentrate Concentrate Concentrate Concentrate
Flotati on Flotati on Flotati on Flotati on
Tailing Tailing Tailing Tailing
Air+Oxygen Air+Oxygen Air+Oxygen Air+Oxygen
Converter Converter Converter Converter
Slag Slag Slag Slag
Blister Blister Blister Blister
Copper Copper Copper Copper
Refined Refined Refined Refined
Copper Copper Copper Copper
Converter Converter Converter Converter Anode Furnace Anode Furnace Anode Furnace Anode Furnace Casti ng Wheel Casti ng Wheel Casti ng Wheel Casti ng Wheel
Anode Anode Anode Anode
Acid Pl ant Acid Pl ant Acid Pl ant Acid Pl ant
PSA
8,000Nm 8,000Nm 8,000Nm 8,000Nm
3 33 3
/h /h /h /h 24,000Nm 24,000Nm 24,000Nm 24,000Nm
3 33 3
/h /h /h /h
Oxygen Plant Oxygen Plant Oxygen Plant Oxygen Plant
Waste Heat Waste Heat Waste Heat Waste Heat
Boiler Boiler Boiler Boiler
Fl ash Smel ti ng Furnace Fl ash Smel ti ng Furnace Fl ash Smel ti ng Furnace Fl ash Smel ti ng Furnace
200t/h 200t/h 200t/h 200t/h
Cyclone Cyclone Cyclone Cyclone
El ectri c Furnace El ectri c Furnace El ectri c Furnace El ectri c Furnace
68%-Cu 68%-Cu 68%-Cu 68%-Cu
Matte Matte Matte Matte
Oxygen Oxygen Oxygen Oxygen
Waste Heat Boi l er Waste Heat Boi l er Waste Heat Boi l er Waste Heat Boi l er
Slag Slag Slag Slag
Slag Slag Slag Slag
Electrolytic Copper Electrolytic Copper Electrolytic Copper Electrolytic Copper
Tank House Tank House Tank House Tank House

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 de-
cline (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.
Figure 2: Annual production of copper at Saganoseki Smelter
Figure 3: Changes in the copper content of concentrates and S/Cu ratio

0
50
100
150
200
250
300
350
400
450
500
1994 96 98 2000 02 04 06 08
C
o
p
p
e
r

P
r
o
d
u
c
t
i
o
n

[
1
0
3

t
/
y
e
a
r

]

32.6
32.1
28.5
29.7
35.2
31.4
32.6 0.995
1.073
1.150
0.848
0.900
0.952
0.920
26.0
27.0
28.0
29.0
30.0
31.0
32.0
33.0
34.0
35.0
36.0
37.0
38.0
2
0
0
1
2
0
0
2
2
0
0
3
2
0
0
4
2
0
0
5
2
0
0
6
2
0
0
7
2
0
0
8
C
u
0.76
0.80
0.84
0.88
0.92
0.96
1.00
1.04
1.08
1.12
1.16
1.20
1.24
S
/
C
u
Cu Content in Concentrates
S/Cu ratio in Concentrates

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 sul-
furic 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
furnace
Installation of a new flux mill
Increase the feed rate
Increase the cooling capacity of FSF
Increase matte grade (65 % 68 %)
Increase the waste heat boiler(WHB) capacity
Converter
furnace
Reduce the number of converters (6 4 furnaces)
Cost saving
Change operating cycle of converters
(4 hot-2 blow 3 hot-2 blow)
Simplify layout of off gas line
Modification to converter boiler Energy saving
Sulfuric acid Installation of SO
3
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
(PSA x 2 units installed)
Increase the feed rate
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.
Table 2: Comparison of operation conditions
Unit
1995 1996 1999 2003 2007
Two FSF
operation
One FSF
operation
High In-
tensive
operation
I
High
Intensive
operation
II
After
SPI
project
Feed rate of FSF Dry-mt/h 70 140 160 170 215
S/Cu in Feed Conc. 0.90 0.91 0.87 0.93 1.03
Matte grade % 60 61 65 67 68
Total flow rate Nm
3
/min 730 350 503 585 684
Oxygen enrichment % 24 81 72 69 76
Process air temperature C 953 25 25 25 25
Smelting capacity mt/a 330 330 450 450 480
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 fur-
nace, 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 C) to reduce heavy oil consumption. In recent years, the
unit consumption of heavy oil has been further reduced using waste heat recovered from the SO
3

cooler at the sulfuric acid plant and steam from the waste heat boiler of the FSF.
Figure 4: Schematic drawings of FSF dryer

Flow conveyor
Combustion air fan
Feed rate : max 213 wet-mt/h
Water content in feed 8 10%
Temperature of Dr 75
Moisture in Concentrates 0.5%
Cage mill
Rotary kiln
Heavy oil
8001100 l/h
Anode furnace off gas fan
500 200
Receiving bin
Bag filter off gas fan
Bag
filter
Dr outlet fan
Big stack No.1
off gas
Concentrates
feeder
Concentrates
burner
FSF
200300 Nm3/min
Steam
4.06.0 t/h
Combustion
Chamber
Diluted air fan
Heat
exchanger
Preheat air
from SO3 cooler
Pneumatic
transportation tube

4 Recovery of waste heat from the SO
3
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, SO
2
concentration of inlet gas during op-
eration should be increased for the energy saving and the increasing production capacity. Sa-
ganoseki owned three lines of acid production system. These three systems are called B, D, and E
respectively. Largest system E had to set the SO
2
concentration of gases to low level, compared to
the other two systems. If the SO
2
concentration of the gas is increased, temperature in the 3
rd
and 4
th

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, SO
2
concentrations of inlet
gas had to keep low level, which considerably restricted the production capacity in system E. In ad-
dition, the unit consumption of electricity of the SO
2
blower was high due to plant operation at low
SO
2
concentrations. Under these circumstances, the SO
3
cooler was installed the inlet of the third
catalyst layer in December 2004 (see Figure 5 below). Feeding air preheated by the SO
3
cooler into
the FSF dryer to reduce heavy oil consumption.
Figure 5: Flow diagram of E system at sulfuric acid plant
5 Installed a steam heat exchanger for dryer diluted air
In February 2008, its 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.

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 generat-
ing turbine. The horizontal fin-tube type heat exchanger has a heat transfer area of 2184 m
2
and its
maximum heat exchanger duty is 18,440 MJ/h. That heat exchanger making 200 C diluted air and
supply to FSF dryer for reduce the unit consumption of heavy oil.
Figure 6: Flow diagram of the steam heat exchanger for preheating diluted air in the dryer
6 Increasing the size of the rotary kiln
To increase the feed rate of the FSF, by the dryer rotary kiln expanded in November 2008. Increas-
ing 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 manufac-
tured this kiln.

Waste Heat Waste Heat Waste Heat Waste Heat
Boiler Boiler Boiler Boiler
Flash Smelting furnace
Off Off Off Off gas
Steam
4.0MPa 50~52 t/h
Power generating
turbine
Steam heat exchanger
Waste Heat
Boiler
Atmosphere air
6001000 Nm3/min
Diluted air fan
Flash furnace dryer
Preheated air
200
Drain
Condenser
Steam
3.0MPa
4.06.0 t/h
Table 3: Comparison of specifications of rotary kiln between before and after increasing kiln size
Before increasing kiln size
( 2400 mm)
After increasing kiln size
( 3100 mm)
Internal diameter 2400 mm 3100 mm
Length 11,000 mm 10,995 mm
Volume 50 m
3
83 m
3

Average gas flow velocity
inside the kiln
21 m/s 12.6 m/s
Evaporated water content
<Evaporating capacity>
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 con-
centrates 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 in-
creases, 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 in-
creased, 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 con-
tent 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.
Figure 7: Image view of concentrates movement in the kiln
Figure 8: Comparison of movements of concentrates in the kiln before and after increased
An image drawing of
concentrates moving
Lifter bar

840
1,850
Before
After
840 mm
1,850 mm

8 Calculation of kiln volume
When determining kiln volume, the following formula was used for calculations. The caloric con-
tent required for evaporating the water content was calculated on the supposition that the kilns con-
tribution 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 kilns 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 mills contribution to drying was
assumed to be 36 % and that of the pneumatic transportation tube was assumed to be 26 %.
(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 SO
3
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 Feb-
ruary 2008, it was also reduced to 5.00 L/t (semiannual average for the first half of 2008). Further-
more, 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).
q [kcalm
3
hrC] T [C]
V[m
3
] =
Q [kcal/hr]
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 approxi-
mately 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 SO
3
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).

4.00
4.50
5.00
5.50
6.00
6.50
7.00
7.50
8.00
8.50
9.00
9.50
0
2
/
4
0
4
/
6
0
4
/
1
2
0
5
/
6
0
5
/
1
2
0
6
/
6
0
6
/
1
2
0
7
/
6
0
7
/
1
2
0
8
/
6
0
8
/
1
2
0
9
/
6
S
p
e
c
i
f
i
c

c
o
n
s
u
m
p
t
i
o
n

o
f

h
e
a
v
y

o
i
l

(
L
/
t

T
o
t
a
l

c
h
a
r
g
e
s
)
Unit consumption: 8.09
L/t
(annual average for
2002 and 2003)
Recovery of waste heat from the SO3 cooler (system
E) at the sulfuric acid plant:
Started in December 2004
Unit consumption: 6.60 L/t (annual
average for 2005)
Steam heat exchanger started operation: Reduction of specific consumption
5.00 L/t (semiannual average for the first half of 2008)
Steam heat exchanger started operation:
February 2008
Regular repair in November 2008: Increasing kiln size and
increasing steam consumption in heat exchanger (4.1 5.1 t/h)

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 SO
3
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 con-
sumption of heavy oil for drying ores from 8.09 L/mt to 4.57 L/mt.

Specific consumption of
8.09 L/t
(annual average for 2002)
Caloric content required for drying: 17,800 Mcal/h
Total charges: 210 t/h (in conversion)
Heavy oil volume: 1,600 L/h
Combustion heat of heavy oil: 16,000
Caloric content required
for drying: 17,800 Mcal/h
Recovery of waste heat
from refining furnace: 1,800
SO3 cooler (system E) at sulfuric
acid plant
Heat recovery using heat exchanger
Conventionally, the combustion heat
of heavy oil accounted for
approximately 90% of total caloric
content required for drying.
Increasing size of dryer kiln
Increasing steam volume
at steam heat exchanger
Accounted for 90% of the total calorific content Accounted for 90% of the total calorific content
Accounted for 59% the total calorific content Accounted for 59% the total calorific content
Specific consumption
of 4.46 L/t
(during May 2009)
2, 2, 2, 2,7 77 700 00 00 00
2, 2, 2, 2,1 11 100 00 00 00 2,200 2,200 2,200 2,200
2, 2, 2, 2,1 11 100 00 00 00
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.
Reduction of 720 Mcal
(accounted for 4% of total calorific content)
Drying conditions:
Hot blast temperature: 1000C
Hot airflow: 1800 Nm3/min
Water content before drying: 9.3%
Water content after drying: 0.5%
Heavy oil volume: 978 L/h
Accounted for 5 Accounted for 57 7% the total calorific content % the total calorific content
Specific consumption of
8.09 L/t
(annual average for 2002)
Caloric content required
for drying: 17,800 Mcal/h
Recovery of waste heat
from refining furnace: 1,800
SO3 cooler (system E) at sulfuric
acid plant
Heat recovery using heat exchanger
Conventionally, the combustion heat
of heavy oil accounted for
approximately 90% of total caloric
content required for drying.
Increasing size of dryer kiln
Increasing steam volume
at steam heat exchanger
Accounted for 90% of the total calorific content Accounted for 90% of the total calorific content
Accounted for 59% the total calorific content Accounted for 59% the total calorific content
Specific consumption
of 4.46 L/t
(during May 2009)
2, 2, 2, 2,7 77 700 00 00 00
2, 2, 2, 2,1 11 100 00 00 00 2,200 2,200 2,200 2,200
2, 2, 2, 2,1 11 100 00 00 00
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.
Reduction of 720 Mcal
(accounted for 4% of total calorific content)
Drying conditions:
Hot blast temperature: 1000C
Hot airflow: 1800 Nm3/min
Water content before drying: 9.3%
Water content after drying: 0.5%
Heavy oil volume: 978 L/h
Accounted for 5 Accounted for 57 7% the total calorific content % the total calorific content

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 Interna-
tional Conference on PMP (2000), 879-884
[3] Katsuya Toda, Toshihiro Kamegai, Masaharu Takahashi, Mitsumasa Hoshi, Takayoshi Fujii:
12
th
International Flash Smelting Congress (2008)


Development of the Continuous Copper Converting
Using Two-Zone Vaniukov Converter
E. Jak L. Tsymbulov
The University of Queensland Norilsk Nickel RJS
Pyrometallurgy Research Centre Gipronickel Institute JS
(PYROSEARCH) 11 Grazhdansky prospect
Brisbane, Australia 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, distri-
bution 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 produc-
tion 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 us-
ing a two-zone Vaniukov furnace. Intensive development program has been carried out to date start-
ing from laboratory investigations and progressing to the pilot plant tests of the two-zone Vaniukov
furnace with area of the base of 11.4 m
2
(results of these tests are presented in previous publica-
tion [1]).

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 bal-
ance 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 interac-
tion 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. SiO
2
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 oxidis-

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 cop-
per, 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
SO
2
-containing off-gas may be used for elemental sulphur, sulphuric acid or liquid SO
2
production.
The reduction zone CO- and H
2
-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 effi-
cient 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 SiO
2
. Small
Al
2
O
3
, MgO and Cr
2
O
3
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 tem-
perature, 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/SiO
2
ratio of the fluxes on copper and nickel partitioning between phases, on
the Fe
3+
and Fe
2+
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 experi-
mental 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 ad-
justed to achieve best agreement between predictions and the results of the pilot plant trials. Agree-
ment between predicted and experimental copper and nickel partitioning data is demonstrated in
Figure 2.
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, wol-
lastonite, 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 sche-
matic of this model of the two-zone process. In brief, equilibrium between input material was as-
sumed 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.
Figure 2: Nickel and copper partitioning (FactSage predictions).
0
5
10
15
0 5 10 15 20 25 30 35
Cu
2
O in slag [wt %]
Ox
Ox
Ox
Ox
Ox
Ox
Ox
R
R
R
R
R
R
R
Exp Ox zone
Exp Red zone
Calc Ox zone
Calc Red Zone
Results of Optimisation of t/d of Cu alloy
N
i

i
n

C
u

b
l
i
s
t
e
r

[
w
t
%
]
Oxidising Zone
R
e
d
u
c
i
n
g

Z
o
n
e
Base case
Base case

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 reduc-
tion 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.

Jak, Tsymbulov
Table 1: Input of the model.
Oxidation zone Cu concentrate SiO
2
flux CaO flux Coal 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
Cu
2
S % 88.1 0.0 0.0 0.0 0.0 0.0
Ni
3
S
2
% 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
SiO
2
% 0.0 75.0 0.0 6.8 0.0 0.0
Al
2
O
3
% 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
Fe
2
O
3
% 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
CaCO
3
% 0.0 0.0 95.7 0.0 0.0 0.0
H
2
O % 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
O
2
% 0.0 0.0 0.0 0.0 92.5 0.0
N
2
% 0.0 0.0 0.0 0.0 7.5 0.0
C
3
H
8
% 0.0 0.0 0.0 0.0 0.0 38.3
C
4
H
10
% 0.0 0.0 0.0 0.0 0.0 61.7

Reduction zone Slag-ox-zone Copper-ox-zone 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
O
2
% 0.0 0.0 92.5 0.0
N
2
% 0.0 0.0 7.5 0.0
C
3
H
8
% 0.0 0.0 0.0 38.3
C
4
H
10
% 0.0 0.0 0.0 61.7
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 con-
densed 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. % Cu
2
O concentration range for the oxidation zone is in the range
of much lower P
O2
than is practically achieved during copper blister production. Reasonable agree-
ment between experimental and predicted values is observed in the range of conditions of practical
interest (see Figure 2).
3.1 Effect of SiO
2
and CaO fluxing with the SiO
2
/CaO ratio of 2.2
Effect of SiO
2
and CaO fluxing with the SiO
2
/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 Fe
2+
form (60-70 %) at this SiO
2
/CaO ratio
of 2.2. The increase of the flux amount reduces Fe
3+
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 Fe
3+
is significantly lower than in the oxi-
dation 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 partition-
ing 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 SiO
2
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 crys-
talline phases the fraction of flux additions for the investigated SiO
2
/CaO ratio of 2.2 is predicted
to be not less than 0.35.
Jak, Tsymbulov
Figure 4: Concentration of Ni in blister as a function of total flux.
Figure 5: Concentration of Cu
2
O in slag as a function of total flux.
0
1
2
3
4
5
6
0.15 0.2 0.25 0.3 0.35 0.4 0.45
CaO&SiO flux to concentrate ratio[wt]
2
N
i

i
n

C
u

b
l
i
s
t
e
r

[
w
t
%
]
Oxidising Zone
R
e
d
u
c
in
g
Z
o
n
e

5
10
15
20
25
30
0.15 0.2 0.25 0.3 0.35 0.4 0.45
2
C
u
2
O

i
n

s
l
a
g

[
w
t
%
]
O
xidising
Z
o
ne
Reducing Zone

Figure 6: Concentration of Ni in slag as a function of total flux.
Figure 7: Fe
3+
/Fe
2+
ratio in slag as a function of total flux.
7
7.5
8
8.5
9
9.5
10
0.15 0.2 0.25 0.3 0.35 0.4 0.45
2
N
i

i
n

s
l
a
g

[
w
t
%
]
O
x
i
d
i
s
i
n
g

Z
o
n
e
Reducing Zone

0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.15 0.2 0.25 0.3 0.35 0.4 0.45
2
F
e
/
F
e
2

i
n

s
l
a
g
3
+
+
O
x
i
d
i
s
i
n
g

Z
o
n
e
R
ed
u
cin
g
Z
o
n
e

Jak, Tsymbulov
Figure 8: Ni partitioning to blister as a function of total flux.
Figure 9: Cu partitioning to blister as a function of total flux.
6
7
8
9
10
11
12
13
0.15 0.2 0.25 0.3 0.35 0.4 0.45
2
N
i

i
n

C
u

b
l
i
s
t
e
r

/

t
o
t
a
l

N
i

[
w
t
%
]
O
x
id
isin
g
Z
o
n
e
R
e
d
u
c
in
g
Z
o
n
e

60
65
70
75
80
85
90
0.15 0.2 0.25 0.3 0.35 0.4 0.45
2
C
u

i
n

C
u

b
l
i
s
t
e
r

/

t
o
t
a
l

C
u

[
w
t
%
]
Oxidising Zone
R
e
d
u
c
i
n
g
Z
o
n
e

Figure 10: Proportion of solids as a function of total flux.
Figure 11: Concentration of Ni in blister as a function of SiO
2
/CaO ratio.
0
2
4
6
8
10
0.15 0.2 0.25 0.3 0.35 0.4 0.45
2
P
r
o
p
o
r
t
i
o
n

o
f

s
o
l
i
d

p
h
a
s
e
s

[
w
t
%
]
Oxidising Zone
R
e
d
u
c
i
n
g

Z
o
n
e

0
1
2
3
4
5
6
0 1 2 3 4 5 6 7
SiO /CaO in oxidising slag [wt]
2
N
i

i
n

C
u

b
l
i
s
t
e
r

[
w
t
%
]
Oxidising Zone
Reducing Zone

Jak, Tsymbulov
3.2 Effect of SiO
2
/CaO ratio in the flux
Further analysis is focused on the investigation of the effect of the SiO
2
/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 SiO
2
/CaO ratio in the
slag.
For the oxidising zone, it was predicted that the SiO
2
/CaO ratio in the slag practically does not
effect the nickel concentration in blister (see Figure 11). Significant effect has the SiO
2
/CaO ratio
on the ferric/ferrous slag ratio (see Figure 12). As the Fe
3+
fraction is significantly higher than Fe
2+

for the SiO
2
/CaO ratio of 0.8, the ferric iron Fe
3+
fraction decreases to 0.2 in the high-silica slag. It
was predicted that the Cu
2
O concentration in the slag practically does not change as a function of
the SiO
2
/CaO ratio (see Figure 13). The exception is the range of SiO
2
/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
SiO
2
/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 SiO
2
/CaO ratio over 2 (see Figure 15).
For the reduction zone, it was predicted that the ferric iron fraction is relatively higher even at re-
ducing conditions for the SiO
2
/CaO ratios between 0.8 and 2 (see Figure 12). The increase of the
SiO
2
/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 in-
crease of the SiO
2
fraction in the slag (see Figure 17), which is the negative factor of the high-SiO
2

slags. The nickel recovery into blister, however, decreases with the increase of the SiO
2
fraction (see
Figure 15), which is an advantage of the high-silica slags since lower concentration of nickel in blis-
ter 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 Cu
2
O concentration and the Fe
3+
/Fe
2+
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 com-
pared to copper at low P
O2
resulting in the increase of the nickel concentration in the slag. Signifi-
cant partitioning of the copper oxide into slag however is predicted to occur as the concentration of
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 P
O2
. For example, at 92.5 % enrichment the heat bal-
ance is achieved at approximately 0.4 wt. % nickel concentration in the blister.
Figure 12: Fe
3+
/Fe
2+
in slag as a function of SiO
2
/CaO ratio.
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
0 1 2 3 4 5 6 7
2
F
e
/
F
e
2

i
n

s
l
a
g
3
+
+
O
x
i
d
i
s
i
n
g

Z
o
n
e
R
e
d
u
c
i
n
g

Z
o
n
e

Jak, Tsymbulov
Figure 13: Concentration of Cu
2
O in slag as a function of SiO
2
/CaO ratio.
Figure 14: Proportion of solids as a function of SiO
2
/CaO ratio.
10
15
20
25
30
0 1 2 3 4 5 6 7
2
C
u
2
O

i
n

s
l
a
g

[
w
t
%
]
Oxidising Zone
Reducing Zone

0
2
4
6
8
10
12
0 1 2 3 4 5 6 7
2
P
r
o
p
o
r
t
i
o
n

o
f

s
o
l
i
d

p
h
a
s
e
s

[
w
t
%
]
O
x
i
d
i
s
i
n
g

Z
o
n
e
R
e
d
u
c
i
n
g

Z
o
n
e

Figure 15: Partitioning of Ni to blister as a function of SiO
2
/CaO ratio.
Figure 16: Concentration of Ni in slag as a function of SiO
2
/CaO ratio.
4
6
8
10
12
14
16
18
20
0 1 2 3 4 5 6 7
2
N
i

i
n

C
u

b
l
i
s
t
e
r

/

t
o
t
a
l

N
i

[
w
t
%
]
Oxidising Zone
Reducing Zone

5
6
7
8
9
10
0 1 2 3 4 5 6 7
2
N
i

i
n

s
l
a
g

[
w
t
%
]
O
x
id
isin
g
Z
o
n
e
R
e
d
u
c
i
n
g

Z
o
n
e

Jak, Tsymbulov
Figure 17: Partitioning of Cu to blister as a function of SiO
2
/CaO ratio.
Figure 18: Concentration of Ni in slag as a function of oxygen blast rate.
60
65
70
75
80
85
90
0 1 2 3 4 5 6 7
2
Oxidising Zone
R
e
d
u
c
in
g
Z
o
n
e
C
u

i
n

C
u

b
l
i
s
t
e
r

/

t
o
t
a
l

C
u

[
w
t
%
]

4
5
6
7
8
9
10
0.94 0.96 0.98 1 1.02 1.04 1.06 1.08 1.1
O +N wt frac.of base case(labls-O/ON)
2 2
85
90
92.5
95
N
i

i
n

s
l
a
g

[
w
t
%
]
Oxidising Zone
B
a
s
e

c
a
s
e

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 SiO
2
/CaO ratios above 2 that enable
to obtain slags with low concentration of the ferric iron Fe
3+
and lower probability of the spinel pre-
cipitation. 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 SiO
2
and CaO fluxes do not strongly influence the partition-
ing of the copper and nickel between blister and slag, however this partitioning is strongly effected
by the SiO
2
/CaO ratio. It is recommended to add 35-40 % flux relative to the copper sulphide con-
centrate to avoid precipitation of solid phases.

-1.98E+07
-1.96E+07
-1.94E+07
-1.92E+07
-1.90E+07
-1.88E+07
-1.86E+07
-1.84E+07
-1.82E+07
0.94 0.96 0.98 1 1.02 1.04 1.06 1.08 1.1
O +N wt frac.of base case(labls-O/ON)
2 2
85
90
92.5
95
Oxidising Zone
H

l
o
s
s

[
K
J
]
B
a
s
e

c
a
s
e

Jak, Tsymbulov
References
[1] KNYAZEV M.V., RYABKO A.G., TSYMBULOV L.B. ET AL., Two-Zone Vaniukov Fur-
nace: 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 high-
grade 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. BLISLE; P. CHARTRAND; S. A. DECTEROV; G. ERIKSSON; K. HACK;
I. H. JUNG; Y. B. KANG; J. MELANON; 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, Montral. http://www.factsage.com/, 2009.
[5] DECTEROV S., JUNG I. H., JAK E., HAYES P. AND PELTON A.D., Thermodynamic
Modeling of the Al
2
O
3
-CaO-CrO-Cr
2
O
3
-FeO-Fe
2
O
3
-MgO-MnO-SiO
2
-S System and Applica-
tions in Ferrous Process Metallurgy, VII Int.Conf. on Molten Slags, Fluxes and Salts, Cape-
town, publ. SAIMM, Johannesburg, South Africa, 2004, pp. 839-850.

Liquidus Temperature in Calcium Ferrite Slags in
Ca
2
Fe
2
O
5
and Ca
2
SiO
4
Primary Phase Fields
with Cu and Fixed Po
2

E. Jak, B. Zhao, P. C. Hayes C. Nexhip, D. P. George-Kennedy
The University of Queensland Rio Tinto
Pyrometallurgy Research Centre (PYROSEARCH) Kennecott Utah Copper LLC
Brisbane, Australia Magna, Utah, USA

Keywords: Liquidus temperature, calcium ferrite slag, copper converting, EPMA
Abstract
Calcium ferrite slags, with the major components Cu
2
O, "Fe
2
O
3
" and CaO are used in continuous
copper-converting processes. A number of other impurities including SiO
2
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 includ-
ing the SiO
2
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 "Cu
2
O"-Fe
2
O
3
-FeO-CaO-SiO
2
system. Most of the previ-
ous 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 equilibra-
tion/quenching/Electron Probe X-Ray Microanalysis (EPMA) techniques using oxide substrates.
Dicalcium ferrite (Ca
2
Fe
2
O
5
) and dicalcium silicate (Ca
2
SiO
4
) were found to be primary phases at
high-CaO concentrations. The effects of SiO
2
on the liquidus temperatures and phase equilibria of
calcium ferrite slags in primary phase fields of Ca
2
Fe
2
O
5
and Ca
2
SiO
4
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 Ca
2
Fe
2
O
5
primary phase field decrease with increasing SiO
2
con-
centration in the liquid. Further increase of SiO
2
results in the appearance of Ca
2
SiO
4
as a primary
phase. The liquidus temperatures in the Ca
2
SiO
4
primary phase field first increase and then decrease
Jak, Zhao, Nexhip, George-Kennedy, Hayes
with increasing SiO
2
concentration in liquid. The addition of Al
2
O
3
can decrease the liquidus tem-
perature in Ca
2
SiO
4
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 Cu
2
O,
CaO and iron oxides. CaO is added to flux Fe, to reduce liquidus temperatures in the spinel (mag-
netite) 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 SiO
2
concentration that is the focus of this study. The stability of the high-
calcium phases strongly depends on the SiO
2
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 "Cu
2
O"-Fe
2
O
3
-FeO-CaO-SiO
2
system. Most of the previ-
ous studies have mostly been focused on the spinel primary phase field with relatively low concen-
tration 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).
SiO
2
is present in copper converting slags in addition to the major components Cu
2
O, 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 SiO
2
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 Ca
2
Fe
2
O
5
and Ca
2
SiO
4
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 Ca
2
Fe
2
O
5
and Ca
2
SiO
4
.
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 Ca
2
Fe
2
O
5
and
Ca
2
SiO
4
. The procedures for preparation of Ca
2
Fe
2
O
5
and Ca
2
SiO
4
substrates are different from that
for spinel.
Liquidus in Calcium Ferrite Slags in Ca
2
Fe
2
O
5
and Ca
2
SiO
4
Ca
2
Fe
2
O
5
substrate was prepared by mixing pure CaO and Fe
2
O
3
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
o
C in air.
Ca
2
SiO
4
substrate cannot be prepared by sintering the mixture of CaO and SiO
2
at high temperature.
The reason is that Ca
2
SiO
4
formed at high temperature disintegrates to powder on cooling as a result
of transformation of '-Ca
2
SiO
4
to -Ca
2
SiO
4
. After a series of different tests and attempts a strong
Ca
2
SiO
4
substrate that could further be used in experiments has been successfully prepared at rela-
tively low temperature with chemical additives. Typical microstructure of the Ca
2
SiO
4
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 Ca
2
SiO
4
phase is actually not formed. At equilibration temperature
(1250
o
C in this project) the Ca
2
SiO
4
phase is formed and surrounded by liquid phase.

Figure 1: Typical microstructure of the substrate material for the Ca
2
SiO
4
primary phase field
experiments (also contains 1 wt. % Fe
2
O
3
and 0.5 wt. % Cu
2
O).
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 Po
2
= 10
5.0
atm in
the Ca
2
Fe
2
O
5
primary phase field and Figure 3 shows a typical microstructure of a sample quenched
from 1250 C and Po
2
= 10
5.0
atm in the Ca
2
SiO
4
primary phase field. It can be seen from the fig-
ures that small dendrites were formed on cooling in the samples. In the free surface of the sample
SiO
2

CaO
Cu
2
O
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 Po
2
of 10
5
atm. That
is directly related to the copper converting operations at KUC. A number of experiments were also
carried out at Po
2
of 10
4.8
and 10
5.5
atm to examine the effect of oxygen partial pressure on the
1250 C isotherms.

Figure 2: Typical microstructure of a sample quenched from 1250 C and Po
2
= 10
5.0
atm in the
Ca
2
Fe
2
O
5
primary phase field at Cu metal saturation.

glass
Ca
2
Fe
2
O
5

Cu
metal
2
Fe
2
O
5
and Ca
2
SiO
4

Figure 3: Typical microstructure of a sample quenched from 1250 C and Po
2
= 10
5.0
atm in
the Ca
2
SiO
4
It has been previously reported that the liquidus isotherms of high-CaO phase Ca
2
Fe
2
O
5
are ap-
proximately parallel to the join Cu
2
O-Fe
2
O
3
, 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 SiO
2
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 SiO
2
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 SiO
2
concentrations in the liquid slag at
1250 C and Po
2
=10
5
atm in the Ca
2
Fe
2
O
5
and Ca
2
SiO
4
primary phase fields at Cu saturation. The
solid squares and circles represent experimental results in the Ca
2
Fe
2
O
5
and Ca
2
SiO
4
primary phase
fields respectively from the present study. The open square represents experimental result reported
by Nikolic on Ca
2
Fe
2
O
5
liquidus without SiO
2
[11]. It can be seen that when SiO
2
concentration is
below 2.2 wt. % the primary phase of the slag is Ca
2
Fe
2
O
5
and the liquidus temperatures slightly
decrease with increasing SiO
2
concentration in the liquid phase. Above 2.2 wt. % SiO
2
in the liquid
phase, the primary phase of the slag is Ca
2
SiO
4
and the liquidus temperatures increase with increas-
ing SiO
2
concentration up to 5 wt. %.

glass
Ca
2
SiO
4

Cu
metal

Figure 4: Observed relations between CaO and SiO
2
concentrations in the liquid at 1250 C and
Po
2
= 10
5
atm in Ca
2
Fe
2
O
5
and Ca
2
SiO
4
primary phase fields at Cu saturation
present liquidus in Ca
2
Fe
2
O
5
primary phase field,
present liquidus in Ca
2
SiO
4
primary phase field,
Nikolic result without SiO
2
[11].
Figure 5 shows correlations between CaO and SiO
2
concentrations in the liquid slag at 1250 C and
Po
2
=10
5
atm in Ca
2
SiO
4
primary phase field at Cu saturation. The solid squares and circles repre-
sent experimental results in the Ca
2
Fe
2
O
5
and Ca
2
SiO
4
primary phase fields respectively in the sys-
tem Cu
2
O-CaO-Fe
2
O
3
-SiO
2
from the present study. The open square represents experimental
result reported by Nikolic on Ca
2
Fe
2
O
5
liquidus without SiO
2
[11]. The dashed line represent the
predictions by the FactSage computer package [10]. Two points with Al
2
O
3
-containing slags from
the present study are also shown in the figure. It can be seen from the figure that in the Ca
2
SiO
4

primary phase field the CaO concentrations in the liquid phase decrease with increasing SiO
2
con-
centration in liquid when SiO
2
concentration is below 5 wt. %. This means that the liquidus tem-
peratures in the Ca
2
SiO
4
primary phase field increase with increasing SiO
2
concentration in liquid at
SiO
2
concentration below 5 wt. %. When SiO
2
concentration in liquid is above 5 wt. %, the liquidus
temperatures in the Ca
2
SiO
4
primary phase field decrease with increasing SiO
2
concentration. Fact-
Sage 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).
liquid + 2CaO.F
2
O
3

liquid +
Ca
2
SiO
4

liquid

2
Fe
2
O
5
and Ca
2
SiO
4
When Al
2
O
3
is present in the slag, it can be seen from Figure 5 that liquidus at 1250
o
C and
P
O2
= 10
-5
atm moves towards the higher CaO concentrations, i.e., the liquidus temperatures are
decreased by addition of Al
2
O
3
.

Figure 5: Observed relationships between CaO and SiO
2
concentrations in the liquid at 1250 C
and Po
2
=10
5
atm in Ca
2
SiO
4
primary phase field at Cu saturation in the
Cu
2
O-CaO-Fe
2
O
3
-SiO
2
-Al
2
O
3
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 Ca
2
Fe
2
O
5
and Ca
2
SiO
4
primary
phase fields at Cu saturation. The solid lines represent isotherms in the Ca
2
Fe
2
O
5
and Ca
2
SiO
4
pri-
mary 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 Cu
2
O concentra-
tion is fixed at a given temperature and oxygen partial pressure. It can be seen from the figure that
the 1250 C isotherm in the Ca
2
SiO
4
primary phase field moves towards lower CaO concentrations
(the Ca
2
SiO
4
liquidus temperature increases) with increasing SiO
2
concentration in liquid from 2.2
to 2.5 wt. %. In contrast, with increasing SiO
2
concentration in liquid from 1.5 to 2.2 wt. %, the
isotherm of 1250 C in the Ca
2
Fe
2
O
5
primary phase field moves towards high CaO direction (the
Ca
2
Fe
2
O
5
liquidus temperature decreases). The effect of SiO
2
concentration on liquidus temperature
is more significant in the Ca
2
SiO
4
primary phase field than that in the Ca
2
Fe
2
O
5
At 1250 C and Cu saturation, increase of Po
2
from10
6.0
to 10
5.0
atm results in the increase of over
~12 wt. % Cu
2
O in slag. Similarly, concentration of Cu
2
O increases as Po
2
increases further from
10
5.0
to 10
4.8
atm.
liquid + Ca
2
SiO
4

liquid

Figure 6a: Experimentally determined 1250 C isotherms (solid lines) in Ca
2
Fe
2
O
5
and Ca
2
SiO
4

primary phase fields at Cu saturation.
Dashed lines: isotherms and isobars by Nikolic [11]
Ca
2
Fe
2
O
5
liquidus at Po
2
= 10
5
atm, Ca
2
SiO
4
liquidus at Po
2
= 10
5
atm,
Ca
2
Fe
2
O
5
liquidus at Po
2
= 10
5.5
atm , Ca
2
SiO
4
liquidus at Po
2
= 10
4.8
atm.
Figure 6b: Enlarged 1250 C isotherms (solid lines) in Ca
2
Fe
2
O
5
and Ca
2
SiO
4
primary phase fields
at Cu saturation. Dashed lines: isotherms and isobars by Nikolic [11]
Ca
2
Fe
2
O
5
liquidus at Po
2
= 10
5
atm, Ca
2
SiO
4
liquidus at Po
2
= 10
5
atm,
Ca
2
Fe
2
O
5
liquidus at Po
2
= 10
5.5
atm , Ca
2
SiO
4
liquidus at Po
2
= 10
4.8
atm.

2
Fe
2
O
5
and Ca
2
SiO
4
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 SiO
2
concentration in liquid phase at 1250 C and Po
2
= 10
5
atm in Ca
2
Fe
2
O
5
and
Ca
2
SiO
4
primary phase fields at Cu saturation. It can be seen from Figure 7 that at SiO
2
concentra-
tion below 2.2 wt. % (Ca
2
Fe
2
O
5
primary phase field) the CaO/Fe weight ratios increase with in-
creasing SiO
2
concentration. This means that liquidus temperatures in the Ca
2
Fe
2
O
5
primary phase
field are decreased by increasing SiO
2
concentration. However, if SiO
2
concentration is between 2.2
and 5 wt. % in liquid, the CaO/Fe weight ratio in liquid significantly decreases with increasing SiO
2

concentration. To attain a liquidus temperature below 1250 C for a slag with high SiO
2
concentra-
tion, CaO/Fe weight ratio has to be increased.

Figure 7: Correlation between CaO/Fe ratio and SiO
2
concentration in liquid phase at 1250 C
and Po
2
= 10
5
atm in Ca
2
Fe
2
O
5
and Ca
2
SiO
4
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 SiO
2
concentration in slag, the effect being dependent on the primary
phase formed.
3 Summary
Experiments have been carried out in the system CaO-Fe
2
O
3
-Cu
2
O-SiO
2
-Al
2
O
3
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 Ca
2
Fe
2
O
5
and Ca
2
SiO
4
are primary phases of the calcium ferrite slags at
high-CaO region. The liquidus temperatures in Ca
2
SiO
4
primary phase field first increase and then
decrease with increasing SiO
2
concentration in liquid. The liquidus temperatures in Ca
2
Fe
2
O
5
pri-
mary phase field decrease with increasing SiO
2
concentration in liquid. Addition of Al
2
O
3
can de-
crease the liquidus temperature in Ca
2
SiO
4
Acknowledgments
The authors wish to thank Rio Tinto - Kennecott Utah Copper LLC for providing the financial sup-
port 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: 3-
13, 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-Cu
2
O and CaO-
FeOn-Cu
2
O-SiO
2
systems at 1200-1300 C - Shigen to Sozai, 105: 315-320.
[6] Takeda Y. (2003): Phase diagram of CaO-FeO-Cu
2
O slag under copper saturation Yazawa
Intl 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.
2
Fe
2
O
5
and Ca
2
SiO
4
[8] Nikolic S, Hayes P.C. and Jak E. (2007): Liquidus temperatures in calcium ferrite slags equili-
brated 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 SiO
2
,
Al
2
O
3
, MgO and Na
2
O on liquidus temperature of calcium ferrite slags at Cu saturation and
fixed Po
2
Copper 2010, Hamburg, Germany.
[10] Bale C.W., Pelton A.D. and Thompson W.T.: Facility for the analysis of chemical thermody-
namics (FactSage), Ecole Polytechnique, Montreal, Canada, available at:
www.crct.polymtl.ca.
[11] Nikolic S (2008): Investigation into the high temperature phase equilibria in the sys-
temCu
2
O-FeO-Fe
2
O
3
-CaO-SiO
2
at metallic copper saturation: an investigation of copper
converting slags PhD Thesis, The University of Queensland.


Numerical Simulation of Fluid Flow and Melt
Temperature in Settler
Zhou Jun Zhou Ping, Chen Zhuo, Meichi
Tongling Non-ferrous Metals Central South University
Group Cooperation Ltd. School of Energy Science and Engineering
Tongling, P.R. China, 244000 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 appeal-
ing 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 tempera-
ture 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 tem-
perature 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.
Zhou J., Zhou P., Chen, Liu, Meichi
1 Introduction
The pyrometallurgical flash smelting process is the main production method employed for extract-
ing 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 oxygen-
enriched 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 sepa-
rately. Off-gas from the furnace is cleaned in a waste heat boiler and electrostatic precipitators be-
fore 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 in-
creasing 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 prob-
lems 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 for-
mation of Fe
3
O
4
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.
Numerical Simulation of Fluid Flow and Melt Temperature in Settler
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. How-
ever, 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 corres-
pond flow belongs to turbulence. Therefore, the standard k- turbulent model is chosen in the simu-
lation.
The velocity and temperature distribution can be described by the following general control equa-
tion.

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

M1,M2,M3,M4- Matte Outlet; SS,SM-Slag Outlet
Inlet
Figure 2: Computational mesh of settler
4 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 / kgm
-3

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

4964
Viscosity / Pas
) exp( 10 36 . 3
5000
4
M
T

0.009
Melting point / K Reference [7] 1353
Surface tensor / Nm
-1
Reference [7] 0.236
Specific heat / Wm
-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 de-

M1, M2, M3,M4- Matte exits; SN,SS-Slag exits
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,
n n
k
n
T
n
p
n
w
n
v
n
u

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 is calculated with wall function, and the tem-
perature 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 veloc-
ity 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.
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 Fig-
ure 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: Velocity contour at multi-section

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 tempera-
ture 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 tem-
perature 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 pro-
duction.
The results indicate that the molten flow is chaotic at the inlet of settler, and it would to some ex-
tend 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 mol-

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 tomor-
row. 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 smelt-
ing 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 smelt-
ing 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 Non-
ferrous Metal Extraction Metallurgy (Copper and Nickel Volume). Beijing: Metallurgical In-
dustrial Press, 2000

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 CO
2
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.
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 en-
ergy. 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 dif-
ficulties 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 contrib-
uted 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).

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 experi-
ences 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

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/cm
2
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 prob-
lems 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/cm
2
)
3.3 Minimization of heat loss for the long steam pipe (3 km)
Minimization of pressure drop (over 42 kg/cm
2
)
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 SO
3
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 tem-
perature absorption, most of this energy is now recoverable at useful levels of temperature. To gen-
erate 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

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 mainte-
nance 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 SO
3
Converter in
acid plant. The process gas stream leaving the 1
st
bed converter is approximately 630 C and cooled
to approximately 420 C at 2
nd
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 su-
perheat 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/cm
2
and minimization of radiation heat.

Considering the length of new steam piping from Onsan plant to oil refinery nearby, it was impossi-
ble to keep 42 kg/cm
2
steam pressure as required. Accordingly, special piping design for long dis-
tance 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 customers
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 com-
pany using slip-joint on the condition of high pressure steam in Korea.
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).
Figure 7: Thermal Conduction Rate

Thermal Conduction Rate
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
0.45
24 60 100 200 300 400 500 600 700 800
U- Brid
Perlite
Glass
wool
(W/m.K)
()

Jung, Byun, Shin
Finally, we satisfied customers 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. CO
2
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 con-
tributes to the environmental policy of Korean government by way of decreasing CO
2
emission of
65,000 tonnes per year.

Gas Injection Phenomena in Converters
An Update on Buoyancy Power and Bath Slopping
Dr. Jol P. T. Kapusta
Air Liquide Canada Inc.
90, boul. Marie-Victorin
Boucherville, Qubec, 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 non-
ferrous 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 opera-
tions, 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 inhe-
rent 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
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 sub-
mergence 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, essential-
ly 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 mathe-
matical 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 grate-
fully 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 pub-
lished metallurgical literature using a simplified expression of the form:
(
+
=
a
bath a
bath
a
b
P
h g P
Ln
M
P Q
2 (1)
where
b
is the buoyancy power (W/kg, which is equivalent to kW/t), Q is the volumetric gas flow
rate at bath temperature (m
3
/s), P
a
is the atmospheric pressure (Pa), M bath
is the mass of the bath
(kg),
bath
is the density of the bath (kg/m
3
), g is the gravitational acceleration (9.81 m/s
2
) 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:
(
+
|
|
\
|
=
a
bath a
bath
gas
3 -
*
o
b
P
h g P
Ln
M
T R

10 4 . 22
Q
2 (2)
where
*
o
Q is the normal volumetric gas flow rate (Nm
3
/s at 0
o
C and 1 atm.), R is the universal gas
constant (J/mol. K) and T
gas
is the gas temperature (K). T
gas
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 tempera-
ture reaches the bath temperature immediately at the gas discharge point, which in itself is a debata-
ble 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 T
gas
is problematic and the
assumption that the gas temperature is equal to the bath temperature is most likely flawed. Wraiths
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 tempera-
ture 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 (bub-
ble rise or buoyancy power).
The overall power expression they developed was as follows:
(
+
(
|
|
\
|
+ +
|
|
\
|
=
(
=
+ =
a
bath a
bath
a
I
S
I
S
o
I
bath
a o
I
S
o
I
bath
a o
rise
I
o
o
I
bath
a o
growth
rise growth total
P
h g P
Ln
h g
P
1
T
T

T
T

T
T

M
P Q
2 1
T
T

T
T

M
P Q

T
T
1
T
T

M
P Q

where

(3)
where
total
is the overall input power (W/kg),
growth
is the input power during bubble growth
(W/kg),
rise
is the buoyancy power during bubble rise (W/kg), Q o
is the standard volumetric gas
Kapusta
flow rate (Sm
3
/s at 15 C and 1 atm.), P
a
is the atmospheric pressure (Pa), M bath
is the mass of the
bath (kg),
bath
is the density of the bath (kg/m
3
), g is the gravitational acceleration (9.81 m/s
2
), h is
the tuyere submergence below the bath surface (m), T
o
is the standard gas temperature (288.15 K),
T
I
is the gas temperature in the bubble at detachment (K), and T
S
is the gas temperature in the bub-
ble 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. T
I
= T
S
= T
bath
), then the buoyancy portion,
rise
, in Equation (3) reduces to:
(
+
= =
a
bath a
o
I
bath
a o
b rise
P
h g P
Ln
T
T

M
P Q
2 (4)
Since ) T (at Q
T
T
Q
bath
o
I
o
= , then Equation (4) simplifies to:
(
+
= =
a
bath a
bath
a
b rise
P
h g P
Ln
M
P Q
2 (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
rise growth kinetic Total
+ + = (6)
The kinetic power input,
kinetic
, is defined by the following expression:
bath
2
tuyeres
gas
bath
2
gas gas
kinetic
M
A
Q
Q

2
1

M
u Q

2
1

|
|
\
|
= = (7)
where
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
kinetic
,
buoyancy
and
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/m
3
, a tuyere submergence of 0.5 m, an air flow rate of 600 Nm
3
/min
injected through 50 tuyeres (5 cm inner diameter). The gas temperature is assumed to be 15 C in
the tuyere (T
o
= 15 C) and to reach the bath temperature of 1200 C at detachment (T
I
= 1200 C).
Table 1: Comparison of relative proportion of kinetic, bubble growth and buoyancy powers for a
typical copper converting operation
kinetic
rise buoyancy
=

growth

bath
2
tuyeres
gas
M
A
Q
Q

2
1
|
|
\
|

(
+
a
bath a
bath
a
P
h g P
Ln
M
P Q
2

(
I
o
o
I
bath
a o
T
T
1
T
T

M
P Q

0.61 W/kg 21.5 W/kg 36.6 W/kg
As illustrated by the values in Table 1 for an average copper converter slag blow, the kinetic pow-
er 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 dur-
ing bubble growth up until detachment,
growth
, is of the same order of magnitude as the buoyancy
power. In fact,
growth
attains its maximum value under the assumption that, at detachment, the gas
inside the bubble has reached the bath temperature. The value of
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 validi-
ty of the assumptions.
For buoyancy calculations from survey data, in addition to the T
gas
concern mentioned above and
which will be discussed later, tuyere submergence and mass of bath are the two other critical para-
meters 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/m
3
to just
below 8000 kg/m
3
. Lets 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/m
3
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. T
gas
= T
bath
). 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 func-
tion of tuyere submergence, are presented in Figure 1.
Kapusta
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. T
gas
= T
bath
), 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 for-
ward 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 pow-
er 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 Equa-
tion 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 m
3
/s (= average flowrate in the copper blow
for all smelters). The data indicates that copper converters in the copper making blow tend to oper-
ate 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].
y = 30.3 x + 2.83
R
2
= 0.76
0
10
20
30
40
50
60
70
80
0.00 0.20 0.40 0.60 0.80 1.00 1.20
Tuyere Submergence, m
B
u
o
y
a
n
c
y

P
o
w
e
r
,

W

/

k
g
Bath mass from constant 35% fill

Figure 2: Buoyancy power calculated from data from Johnson et al. [13] with Equation (1) on the
gas
= T
bath
) 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 pe-
riodically 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 oc-
cur 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.
y = 54.2 x + 9.04
R
2
= 0.56
0
10
20
30
40
50
60
70
80
0.00 0.20 0.40 0.60 0.80 1.00 1.20
B
u
o
y
a
n
c
y

P
o
w
e
r
,

W

/

k
g
Bath mass = mass of blister produced

Kapusta
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.
y = 54.2 x + 9.04
R
2
= 0.56
y = 55.3 x + 6.71
R
2
= 0.99
0
10
20
30
40
50
60
70
80
0.00 0.20 0.40 0.60 0.80 1.00 1.20
B
u
o
y
a
n
c
y

P
o
w
e
r
,

W

/

k
g

Figure 4: Buoyancy power calculated from data from Lehner et al. [14] with Equation (1) on the
gas
= T
bath
) 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 conver-
ter. 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.
y = 51.9 x + 2.61
R
2
= 0.52
0
10
20
30
40
50
60
70
80
90
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40
B
u
o
y
a
n
c
y

P
o
w
e
r
,

W

/

k
g

Kapusta
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, Fig-
ure 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.
y = 34.1 x + 3.26
R
2
= 0.68
0
10
20
30
40
50
60
0.00 0.20 0.40 0.60 0.80 1.00 1.20
B
u
o
y
a
n
c
y

P
o
w
e
r
,

W

/

k
g

Figure 6: Buoyancy power versus tuyere submergence for the screened out 1979 and 1993 sur-
vey 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:
b
= 39.1 h + 6.11 (R
2
= 0.58) (8)
The slope of the line is about 39 W/kg/m indicating that the buoyant energy input to the bath in-
creases 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:
( ) h g P M
g P Q 2
Slope
bath a bath
bath a
+
= (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 (M
bath
):
( )
bath
bath a
bath a
M
g P 2
g 0.53 P
39.1 Q
+
= (10)
y = 39.1 x + 6.11
R
2
= 0.58
0
10
20
30
40
50
60
0.00 0.20 0.40 0.60 0.80 1.00 1.20
B
u
o
y
a
n
c
y

P
o
w
e
r
,

W

/

k
g

Kapusta
For the copper blow this equation becomes:
Q (m
3
/s) = 0.36 M
bath
(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 (Nm
3
/min) = 4.0 M bath
(t) (12)
The slope of this line suggests that a blowing rate increase of 4 Nm
3
/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).
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 thre-
shold 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
y = 4.0 x
0
100
200
300
400
500
0 20 40 60 80 100 120 140
Mass of Bath, tonnes
B
l
a
s
t

F
l
o
w

R
a
t
e
,

N
m
3
/
m
i
n
Non Slopping
Slopping

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 al-
ready 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 cor-
relation between the overall power input and tuyere submergence.
Kapusta
Figure 8: Total power input calculated from the revised data set using Equation (3) with a de-
tachment 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 (
b
=
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 (R
2
= 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 (R
2
= 0.48) as shown in Figure 10.
y = 10.0 x + 18.28
R
2
= 0.09
0
10
20
30
40
50
60
70
80
0.00 0.20 0.40 0.60 0.80 1.00 1.20
T
o
t
a
l

P
o
w
e
r

I
n
p
u
t
,

W

/

k
g

y = 12.2 x + 7.62
R
2
= 0.30
0
10
20
30
40
0.00 0.20 0.40 0.60 0.80 1.00 1.20
T
o
t
a
l

P
o
w
e
r

I
n
p
u
t
,

W

/

k
g
T
surface
= 35% T
bath

y = 23.0 x + 5.79
R
2
= 0.48
0
10
20
30
40
0.00 0.20 0.40 0.60 0.80 1.00 1.20
T
o
t
a
l

P
o
w
e
r

I
n
p
u
t
,

W

/

k
g
T
surface
= 50% T
bath

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 opti-
mum 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 injec-
tion 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 so-
phisticated model of the process.
5 Conclusions
Gas injection phenomena in non-ferrous converting are complex phenomena with a range of chemi-
cal 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 gen-
eration 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 tempera-
ture 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 rela-
tionship for converter operation, based on a simple calculation of gas injection power, is seen as:
Gas flow rate (Nm
3
/min) = 4.0 Bath mass (t).
If the blast exceeds the level indicated by this equation, then the bath is expected to enter a slop-
ping 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 in-
fluenced 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 au-
thor acknowledges helpful and insightful discussions with Greg Richards on various issues ad-
dressed 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.
Kapusta
List of Symbols
g
Gravitational acceleration (= 9.81 m/s
2
)
h Tuyere submergence below bath surface, m
bath
M
Mass of bath, kg
a
P
Atmospheric pressure, Pa (= 101,325 Pa)
Q
Gas flowrate at gas temperature (T
gas
) and 1 atm, m
3
/s
o
Q
Gas flowrate at standard conditions (15 C & 1 atm), Sm
3
/s
*
o
Q
Gas flowrate at normal conditions (0 C & 1 atm), Nm
3
/s
R Universal gas constant (8.3143 J/mol K)
T Temperature, K
bath
T
Bath temperature, K
gas
T
Gas temperature, K
o
T
Standard gas temperature (288.15 K)
I
T Gas temperature in the gas bubble at detachment, K
S
T
Gas temperature in the gas bubble at the bath surface, K
V Gas volume, m
3

o
V
Molar volume of gas at standard temperature, Sm
3

growth
W
Work done by the gas during bubble growth, J
rise
W
Work done by the gas during bubble rise, J
total
W
Total work done by injected gas, J

Greek letters
b

Buoyancy power, W/kg
growth
Input power during bubble growth, W/kg

rise

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

Total input power imparted by injected gas, W/kg
bath
Bath density, kg/m

3

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 Metal-
lurgy: 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 Phe-
nomena 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 Ex-
traction 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 Con-
verters 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 Representa-
tion of Gas Dispersion in Upwardly Injected Submerged Gas Jets Canadian Metallurgical
Quarterly, Vol. 28, No. 4: 323-330
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 Conver-
ters, 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 Dis-
sipation and Agitation in Steel Processing Operations Ironmaking & Steelmaking (Quarter-
ly), 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 Proceed-
ings 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 UM-
Hoboken 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 Fundamen-
tals: 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, Van-
couver, Canada
[21] KAPUSTA J.P., STICKLING H., TAI W. (2005): High Oxygen Shrouded Injection at Fal-
conbridge: 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 Li-
quide Shrouded Injector (ALSI) Technology at the Thai Copper Industries Smelter Proceed-
ings 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 Eu-
ropean Metallurgical Conference 1994 (EMC94), Vol. I: 50-78, Germany
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
S
I S
T y
h
T T
T +
|
\
|

= or
I
S
I
S
I
T
T

h
y

T
T
1
T
T
+
|
|
\
|
= (A1)
The temperature dependence of the gas volume is defined as follows:
(
+
|
|
\
|
=
= =
I
S
I
S
o
I
o
I o
I
o
o
o
T
T

h
y

T
T
1
T
T
V
T
T

T
T
V
T
T
V V
(A2)
Once the pressure effect on the volume is included, it follows:
(
+
|
|
\
|
+
=
I
S
I
S
bath a
a
o
I
o
T
T

h
y

T
T
1
y g P
P

T
T
V V (A3)
The rate of change of volume with height is given by
( )
( )
(
(
(
(
(
+
(
+
|
|
\
|

|
|
\
|
+
=
2
bath a
bath
I
S
I
S
I
S
bath a
o
I
a o
y g P
dy g
T
T

h
y

T
T
1 dy
h
1

T
T
1 y g P

T
T
P V dV (A4)
The work done by the gas in rising a height dy is given by:
dy V g dV P dW
bath rise
= (A5)
where y g P P
bath a
+ = (A6)
Substituting equations (A3), (A4) and (A6) into equation (A5) yields:
( )
dy
T
T

h
y

T
T
1
y g P
P

T
T
V g
y g P
dy g
T
T

h
y

T
T
1 dy
h
1

T
T
1 y g P

T
T
P V dW
I
S
I
S
bath a
a
o
I
o bath
bath a
bath
I
S
I
S
I
S
bath a
o
I
a o rise
(
+
|
|
\
|
(
(
(
(
(
+
(
+
|
|
\
|

|
|
\
|
+
=
(A7)
Kapusta
The work done by the gas when rising to the liquid metal surface is obtained as:

+
(
+
|
|
\
|

|
|
\
|
=
o
h bath a
I
S
I
S
o
h o
I
a o bath
I
S
o
I
a o
o
h
rise
y g P
dy
T
T

h
y

T
T
1

T
T
P V g 2 dy
h
1

T
T
1
T
T
P V dW (A8)
Integration of Equation (A8) yields:
(
+
(
|
|
\
|
+
|
|
\
|

|
|
\
|
=
(
(
(
(
+
|
|
\
|

+
|
|
\
|

|
|
\
|
=

a
bath a
I
S
bath
a
I
S
o
I
a o
I
S
o
I
a o
I
S
o
I
a o
o
h
o
h
bath
a
I
S
bath
a
I
S
I
S
o
I
a o
I
S
o
I
a o rise
P
h g P
Ln
h
1

T
T
1
g
P

T
T

T
T
P V 2
1
T
T

T
T
P V 2 1
T
T

T
T
P V
dy
g
P
y
h
1

T
T
1
g
P

T
T
dy
h
1

T
T
1
T
T
P V 2
T
T
1
T
T
P V W
(A9)

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

(
+
(
|
|
\
|
+ +
|
|
\
|
=
a
bath a
bath
a
I
S
I
S
o
I
a o
I
S
o
I
a o rise
P
h g P
Ln
h g
P
1
T
T

T
T

T
T
P V 2 1
T
T

T
T
P V W (A10)

2) Derivation of bubble growth component
dy V g dV P dW
bath growth
= (A11)
Since the bubble is stationary (i.e. the work done by the bubble is independent of depth), then
dV P dW
growth
= (A12)
Applying the gas law,
T
V P
is constant, and for a fixed bubble P is also constant, then
0
T
dT V dV T
P
T
V
d P
T
V P
d
2
=
=
|
\
|
=
|
\
|
(A13)
Equation (A13) implies that dT V dV T = or by rearranging that
dT
T
V
dV = (A14)
Substituting Equation (A14) into Equation (A12) yields:
dT
T
V P
dT
T
V P
dW
o
o a
growth
= = (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:
[ ]
o I
o
o a
T
T o
o a
T
T
growth
T T
T
V P
dT
T
V P
dW
I
o
I
o
= =

(A16)

Finally, upon rearranging, the work of the bubble prior to detachment is defined as:
(
=
I
o
o
I
a o growth
T
T
1
T
T
P V W (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:
(
+
(
|
|
\
|
+ +
|
|
\
|

(
=
+ =
a
bath a
bath
a
I
S
I
S
o
I
a o
I
S
o
I
a o
I
o
o
I
a o
rise growth total
P
h g P
Ln
h g
P
1
T
T

T
T

T
T
P V 2 1
T
T

T
T
P V
T
T
1
T
T
P V
W W W
(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 (V
o
) by the gas
flow rate (Q o
). The overall input power per kilogram of bath,
total
, is then given by:
(
+
(
|
|
\
|
+ +
|
|
\
|

(
=
+ =
a
bath a
bath
a
I
S
I
S
o
I
bath
a o
I
S
o
I
bath
a o
I
o
o
I
bath
a o
rise growth total
P
h g P
Ln
h g
P
1
T
T

T
T

T
T

M
P Q
2 1
T
T

T
T

M
P Q

T
T
1
T
T

M
P Q

(A19)


Recovery of Valuable Metals from
Converter Slags by Reduction with Iron
Dipl.-Ing. Stefan Konetschnik, Dipl.-Ing. Helmut Paulitsch Dipl.-Ing. Dr.mont. Josef Pesl

Ao.Univ.Prof. Dipl.-Ing. Dr.mont. Helmut Antrekowitsch
University of Leoben Montanwerke Brixlegg AG
Nonferrous Metallurgy
Franz-Josef-Strae 18 Werkstrae 1-3
8700 Leoben, Austria 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 con-
verter 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 dif-
ferent 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 devel-
oped. 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 composi-
tion, 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.

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 Montan-
werke Brixlegg AG should be reduced with an iron-copper scrap to lower the content of the valu-
able 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/SiO
2
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].

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. Fol-
lowing, 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 addi-
tion, 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
SiO
2
+ xFe (FeO)
x
SiO
2
+ xMe (2)
xFeO + SiO
2
(FeO)
x
SiO
2
(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. There-
fore, 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 methane-
air 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 be-
cause 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 coef-
ficients of the equilibrium for a non-ideal dissolving behaviour of the individual phases.
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/SiO
2
ratio are also implemented in the
model.
3.2 Results
The modelling was done with varying amount of reduction agent and temperature. In the calcula-
tions, 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 tem-
perature 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 thermody-
namic model. There is also an impact on the equilibrium by the basicity. It has to be kept in mind
that the Fe/SiO
2
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 ac-
tivity of the valuable metals in the slag is less and metal losses are higher.
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 com-
position of the metal phase is shown in Figure 3.

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/SiO
2
.
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 success-
ful 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 stoichiome-
try. Crucibles made of alumina provided the necessary temperature durability. For a part of the tri-
als, iron crucibles simulated an excess of reductant.

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 tri-
als 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/SiO
2
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/SiO
2
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 ac-
cording to thermodynamic calculations. The chronology of the experiments was as follows:
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/SiO
2
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 re-
duction degree of the elements copper, lead and tin is calculated as follows:
Cu
reduction degree
= (Cu
metal
- Cu
reducing agent
- Cu
master alloy
)/Cu
converter slag
(5)
Pb|Sn
reduction degree
= Pb|Sn
metal
/Pb|Sn
converter slag
(6)
5.1 Muffle furnace
Figure 4 shows the reduction degree of copper, lead and tin in dependence of the amount of reduc-
tion agent and the stoichiometry respectively.
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 cruci-
ble, 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 calcu-
lated 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 proce-
dural 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 con-
trary.
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.

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 sepa-
ration 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 meas-
urement 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 be-
cause of the clay graphite crucible is probable. The amounts of copper, lead and tin during the proc-
ess are shown in Figure 5.
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 vola-
tilisation 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 tem-
perature and the better possibility for nodulising at the significant larger sample size. As a result of
the equilibrium of [Me]/(MeO
x
), 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.

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:

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 consid-
erable solution of iron in the black copper.
Fe/SiO
2
ratio: Because of a low slag viscosity, the Fe/SiO
2
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 pro-
cess, a ratio of 1.0 was determined.
At the experimental investigations in the induction furnace, the reduction degree is also an impor-
tant parameter for evaluating the success of the process. Table 1 shows the comparison of the reduc-
tion 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.
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 inves-
tigations 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 cop-
per in the slag are high, especially at lower temperatures. Iron crucibles simulate an excess of reduc-
tant, 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 mas-
ter 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 graph-
ite 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/SiO
2
ratio and an increasing slag viscosity took place. Fluxes
and an exact Fe/SiO
2
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 cop-
per 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 ade-
quate 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.
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


Waste heat boilers for
Copper Smelting Applications
Dipl.-Ing. Stefan Kster
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 metallurgi-
cal 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 as-
pects. 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 ex-
ample of the Aurubis AG (earlier Norddeutsche Affinerie AG) Flash Smelting plant which was ex-
tended 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
technol-
ogy 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 secon-
dary 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.
Kster
1 Introduction
The Oschatz GmbH was founded in 1849 and is one of the last private owned WHB companies in
the world. Oschatzs 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 indi-
vidual 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 sub-
supplier 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: Typical WHB geometries
2 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 elec-
trostatic 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
SO
3
has to be minimized. As an effective measure to protect the WHB from the corrosion, the heat-
ing surface walls and elements have to be kept above the dew point. Thus the water/ steam operat-
ing 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 dry-
ing) to reduce the process costs and/ or for profitable network electricity supply.

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 sur-
faces to the natural circulation system leads to smaller pump capacities and electrical power con-
sumption 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 heat-
ing 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 ori-
fices or nozzles to get uniform water flow for each tube which is a necessary condition for a suffi-
cient 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 Nm/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.
Kster
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 pro-
duct 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.

Figure 3: WHB downstream Flash Smelting process for the Aurubis AG in Hamburg, Germany
5 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.

Kster
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 de-
vices and
bundle hammering cleaning devices with one motor drive and one shaft with up to 20 devices
next to each other.

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 hammer-
ing 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 po-
sition of the fitted bolts determine the impact force of the hammer on the anvil or pestle. More bolts

Kster
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. De-
pending 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 pneu-
matic piston is pushed against one or two springs by compressed air. The hammering effect is in-
duced 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.

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
Design complexity complex simple
Modification possibility limited wide range
Upgrade capability no yes
Operation costs high low
6 CONTOP cyclone technology
The Klckner-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 concen-
trate. 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
cy-
clone 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 steelmak-
ing 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
Kster
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 resi-

dues.
Figure 7: Cyclone Reactor
Table 5: Technical data for the WHB downstream Cyclone Reactor (Figure 7)
Off gas rate 5000-50,000 Nm/h
Fume and dust inlet 0.7-7 t/h
Temperature WHB inlet 1200-2000 C
Dust WHB outlet 0.8-8 t/h
Temperature WHB outlet 240-380 C
Steam pressure 30-70 bar
Steam rate 6-60 t/h

Table 6: Technical data for the Cyclone Reactor (Figure 7)
Cyclone diameter 1-3 m
Feed rate of solids 10-100 t/h
Off gas temperature at outlet 1200-2000 C
Steam pressure 30-70 bar
Steam rate 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 con-
tinuous 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.
Kster
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 proc-
ess. 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.

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 modifi-
cations 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. Kster: Modern waste heat recovery systems for the non-ferrous industry. GDMB - Global
Growth of Nonferrous Metals Production, Volume 1, S. 103115. 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 Automotive-

Scrap Industry. Siemens VAI Metals Technologies GmbH & Co Linz / Austria
[3] Ulrich Kerney: The Harzer Zink Smelting Reactor Process for Zn Bearing Secondaries. Intensi-
vierung metallurgischer Prozesse, Heft 87 der Schriftenreihe der GDMB Gesellschaft fr Berg-
bau, Metallurgie, Rohstoff- und Umwelttechnik, Clausthal-Zellerfeld 2000, S. 155169.


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 Ka-
zakhstan. 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 above-
mentioned 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 increas-
ing 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 desul-
furation 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 con-
centrates 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 pos-
sibility in providing source materials for autogenous smelting with minimum fluxes consumption to
producing mattes of different compositions.
Kozhakhmetov, Kvyatkovskiy, Ospanov, Abisheva, Zagorodnyaya
1 Introduction
Copper industry of the Republic of Kazakhstan features two types of deposits with the ore composi-
tion differing considerably. Concentrates produced from the ore mines at Zhezkazgan group of de-
posits 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 concen-
trates are processed in electric furnaces using calcium-containing flux. For Eastern Kazakhstan con-
centrates 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 pyrometallur-
gical 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 tech-
nology for processing of high-copper concentrates electric smelting cannot be used for combina-
tion 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 concen-
trate 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 S
tot
; 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 rhe-
nium 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 disadvan-
tages such as generating secondary fumes, requirements for additional treatment, generating consi-
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 min-
eral 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 al-
lowing 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 con-
firmed the estimates on oxidative smelting of concentrates mixture given below. Note that the oxy-
gen utilization factor was not less than 90 %.
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 alun-
dum pipe; 6 millivoltmeter; 7 rubber hose; 8 rheometer; 9 gasometer)

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 =1.
Cake and filtrate were analyzed for Re, Pb, Cd, Zn, Cu, S. Grades of these elements in the mid-
dlings 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.
3.1 Combined processing of Zhezkazgan and Eastern Kazakhstan
concentrates
In order to simulate chemical and heat transformations during autogenous smelting the feed para-
meters were calculated based on high-copper Zhezkazgan concentrate an high-sulfur East Ka-
zakhstan 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 [%] SiO
2
[%] 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)
Cu [%] Fe [%] S [%] SiO
2
[%] CaO [%] Pb [%] Zn [%]
Concentrate
mixture
23.21 18.12 31.0 13.11 0.33 1.82 1.72
Required
composition
25 23 31 15 2 2 2
Deviation 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
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 cal-
culations are given in Tables 3 and 4.
Smelting temperature was 1300 . 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 Pb SiO
2
CaO Others
Matte [%] 58.08 14.64 22.55 0.90 1.13 2.38 0.23
Slag [%] 0.58 29.91 0.44 9.19 2.44 0.59 31.75 0.80 24.16

Table 4: Material balance for 58 % Cu matte process
FEED [t/h] 000 [m
3
/h]
Concentrate 100.000
Coal 0.500
Oxygen 23.454 16.500
Air 2.575 2.00
Total: 126.529
OUTCOME:
Matte 36.619
Slag 38.162
Gas 50.529 24.373
Dust 1.219
Total: 126.529
The calculations of smelting at 55 % Cu matte are given in Tables 5 and 6 below. Smelting temper-
ature is 1300 .
Cu Fe S O Zn Pb As SiO
2
CaO Others
Matte [%] 55.09 16.68 22.79 1.03 1.42 2.64 0.09 0.26
Slag [%] 0.55 28.20 0.51 8.60 2.42 0.53 0.18 33.02 0.83 25.17
FEED [t/h] 000 [m
3
/h]
Concentrate 100.000
Coal 0.500
Oxygen 22.459 15.800
Air 2.575 2.000
Total: 125.534
OUTCOME:
Matte 38.645
Slag 36,639
Gas 49.031 23.896
Dust 1.219
Total: 125.534
Matte and slag composition for 62 % Cu matte process is given in Tables 7 and 8. Smelting tempe-
rature is 1330 .
Cu Fe S O Zn Pb SiO
2
CaO Others
Matte [%] 62.01 11.97 22.24 0.72 0.79 2.00 0.19
Slag [%] 0.62 31.78 0.36 9.82 2.36 0.67 30.42 0.77 23.11

FEED [t/h] 000 [m
3
/h]
Concentrate 100.000
Coal 0.500
Oxygen 24.690 17.370
Air 2.575 2.000
Total: 127.765
OUTCOME:
Matte 34.258
Slag 39.891
Gas 52.397 24.959
Dust 1.219
Total: 127.765
The results of the trial smelting and calculations confirmed possibility of smelting of the feed com-
posed of two concentrates mentioned above in Vanyukov furnace with no flux added. The heat bal-
ance 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 () 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 S
total
; 1.05
As; 0.6 SiO
2
; 0.8 Cd; 0.04 Bi; 0.02 Ag; 0.03 Re. According to the chemical and X-ray phase ana-
lyses 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 sul-
fates while the remaining ~10 % in sulfides.
Table 9: Phase composition of lead, zinc, copper and sulfur in the dust
Grades [%] Elements
Pb Zn Cu Cd S
Sulfates 81.43 81.54 58.18 82.72 89.6
Oxides 6.10 6.20 27.70 5.43
Sulfides 5.85 10.77 12.73 11.45 10.4
Metal 6.36 - 1.82 -
Silicates - 1.54 - 0.52
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 tempera-
ture (20-70 ) on behaviour of lead, zinc, copper, cadmium and rhenium was investigated at car-
bonation stage. The quantity of soda theoretically required to convert metal sulfates into carbonates
was calculated based on the following reaction:

MeSO
4
+ Na
2
CO
3
= MeCO
3
+Na
2
SO
4
. (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-to-
liquid ratio=1:4, temperature=50 , 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 in 10 intervals (solid-to-liquid ratio 1:4, soda consump-
tion=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 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
account content of the metals in form of sulfates), solid-to-liquid ratio=1:4, contact time=60 mi-
nutes, temperature=60 .
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(NO
3
)
2
52.2 g vs. PbCl
2
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 . 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 (4 to 0.8) of the solutions mixture adjusted by adding of
sulfuric acid solution, temperature (20-70 ) 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 sedimen-
tation 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 contain-
ing 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 .
Rough ammonium perrhenate precipitated in such conditions. The rhenium recovery at re-extraction
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

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 reduc-
tion 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 ex-
tracted 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 5
th
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 Lule, 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 Di-
verse Technological landscape: Cobre 2003 the 5 the international conference, 2003, Santia-
go, 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 Cu
2
O-FeO-Fe
2
O
3
-CaO-SiO
2
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.
[6] KARELOV, S.V., MAMYACHENKOV, S.V., NABOICHENKO, S.S., YAKORNOV, S.A. & USOV .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 Petro-
leum Engineers, New York, 1976, pp, 520 533.
[11] BENEASH, 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 Non-
ferrous Metals, Publishing house Metallurgy, Moscow, 1962. pp. 45-50. (in Russian)
[12] KOKUSHEVA, .., DAIRABAEVA, G.., USABEKOVA, .SH., PERFILJEV, N..: 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 technol-
ogy in Kazakhstan, Hydrometallurgy, 2002, 63, pp.55 - 63.
[14] KUNAEV, .M., NEREZOV, V.M., DADABAEV, A. YU.: New hydrometallurgical processes for
the production of molybdenum, tungsten, and rhenium, Publishing house Science, Alma-Ata,
-: . 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.


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, ISASMELT
TM

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 ge-
ometry, 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 feasi-
ble option to be considered in the quest to improve plant operation and investment payback.
Lankinen, Peippo
1 Introduction
Foster Wheeler Energia Oy is an independent Finnish subsidiary of Foster Wheeler A.G. (formerly
Foster Wheeler Ltd). Together its 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 1950s over 100 WHB units have been supplied.
WHBs 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 WHBs but also WHBs 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 cir-
culation, a robust structure with flexible geometry and compact layout has been achieved.
Design of vertical type Foster Wheeler WHB had been developed in deliveries, such as Samim, Por-
tovesme, 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 WHBs 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 WHBs for two converting furnaces ( Ausmelt process )
in Anglo Platinums 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).

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/m
2
up to 80 kW/m
2

throughout the hood, but may locally have essentially higher spot heat flux figures. Due to possibil-
ity 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/m
2
. 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 fur-
nace 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 de-
signed 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 cir-
culation 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

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 util-
ized 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 tem-
perature. 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 cir-

Lankinen, Peippo
cuits. This enables fast continuity of operation even if the lance port collar is damaged by e.g. man-
ual cleaning of accretions around lance.
In addition to lance, burner and feed port arrangement to roof layout, a maintenance hatch for fur-
nace 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).

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 sce-
nario.
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).

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 rec-
tangular 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 require-
ments 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: Hood slope section under fabrication.

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 bal-
anced with the requirement of cooling the anchors. V-type anchor have been successfully used in-
stead 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 transporta-
tion supports can be used for site pre-assembly on ground level and for further lifting into the fur-
nace building (Figure 9).

Lankinen, Peippo
Figure 9: Hood pre-assembly arrangement at shop (left). Hood front section at site; pre-assembly
on ground level (top right) and ready for lifting.
4 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.

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 fea-
sible option to be considered in the quest to improve plant operation.

Lankinen, Peippo
References
[1] PEIPPO R., HOLOPAINEN H., NOKELAINEN J.: Copper smelter waste heat boiler technol-
ogy 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
21
st
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, Ad-
vanced 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) Proc-
ess, 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

Experimental Estimation of the Residence Time
Distribution in a P-S Converter
C. Lpez, A. Almaraz, R. Cuenca, G. Plascencia B. Hernndez, F. Reyes
CIITEC IPN Facultad de Qumica, Ed. D, UNAM
Cerrada Cecati s/n Depto. de Ing. Qumica Metalrgica
Circuito de la Investigacin Cientfica s/n
Mxico, D.F. C.P. 02250 Mxico Mxico, D.F. C.P. 04510 Mxico

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 con-
structed 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. Addi-
tionally, the results were compared with some numerical calculations of the blowing operation pre-
viously reported by this research group. It was found that nearly 50 % of the reactor works as mix-
ing 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.
Lpez, Almaraz, Cuenca, Hernndez, 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 pa-
per 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 distri-
bution (C curve) are time
-1
.
It is convenient to represent the RTD in a normalized manner, thus the C curve is defines as:
=
0
1 dt C (1)
And the mean residence time of the fluid in the vessel is defined by:
=
0
dt C t (2)
3 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
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:
( )
gas liquid
gas
D g
u
Fr

=
2
'
(3)
Where u is the velocity of the injecting gas (m/s), g is the acceleration due to gravity (9.81 m/s
2
), D
is the diameter of the tuyere from which air is injected into the liquid phase (m),
gas
is the density
of the injecting gas (kg/m
3
) and
liquid
is the density of the liquid in the vessel (kg/m
3
).
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/s
2

matte
4800 kg/m
3

air
1.2 kg/m
3

Fr = 8
With this value of the modified Froude number, the air flow rate needed for our tests was back cal-
culated. 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 con-
ditions.
Table 2: Experimental flow conditions
Variable Value Units
Gas flow rate
8.0
12.5
25.0
L/min
Gas velocity
10.6
16.6
33.2
m/s
Height from injection point
0.14
0.07
m
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.

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.
A CB
H
J
I
D
E
L
F
K
G
4
4

c
m
4
0

c
m
I
n
j
e
c
t
i
o
n
p
o
i
n
t
L
O
C
A
T
I
O
N

O
F

S
A
M
P
L
I
N
G

P
O
I
N
T
S
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).

Figure 2: 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.

Figure 3: 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.
Time (s)
0 20 40 60 80 100 120
C

(
1
/
s
)
0.004
0.008
0.012
0.016
0.020
point A
point B
point C
Time (s)
0 20 40 60 80 100 120
C

(
1
/
s
)
0.004
0.008
0.012
0.016
0.020
0.024
point G
point I
point L
Time (s)
0 20 40 60 80 100 120
C

(
1
/
s
)
0.004
0.008
0.012
0.016
0.020
0.024
point A
point B
point C
Time (s)
0 20 40 60 80 100 120
C

(
1
/
s
)
0.004
0.008
0.012
0.016
0.020
0.024
0.028
point G
point I
point L

Figure 4: 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 incor-
rect 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 veloc-
ity 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 resi-
dence 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.

Time (s)
0 20 40 60 80 100 120
C

(
1
/
s
)
0.004
0.008
0.012
0.016
0.020
0.024
point A
point B
point C
Time (s)
0 20 40 60 80 100 120
C

(
1
/
s
)
0.004
0.008
0.012
0.016
0.020
0.024
0.028
point G
point I
point L

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 af-
fected as location A is when decreasing the gas flow rate.

0.0
10.0
20.0
30.0
40.0
50.0
60.0
A B C G I L
M
e
a
n

r
e
s
i
d
e
n
c
e

t
i
m
e

(
s
-
1
)
Measuring point
25 L/min 12.5 L/min 8 L/min

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 de-
velop 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 recircula-
tion 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.
Time (s)
0 20 40 60 80 100 120
C

(
1
/
s
)
0.004
0.008
0.012
0.016
0.020
Point A, 25 L/min
Point A, 8 L/min
Point L, 25 L/min
Point L, 8 L/min

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 reali-
zation 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 evi-
dent that the nature of the blowing operation along with the geometry of the P-S converter com-
mands for the creation of recirculation zones within the vessel. In order to minimize such recircula-
tion zones, it is imperative to continue research on the gas injection into the copper matte.

Figure 8: Results from colour tracer tests. Screen shots at different times after injecting the color-
ant into the vessel, indicating the high recirculation zone at the centre of the vessel.
5 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.
(A)
(B)
(C)
Acknowledgements
The authors wish to thank SIP IPN grants # 20080019 & 20090503 to conduct this research.
References
[1] LEVENSPIEL O. (1999): Chemical Reaction Engineering 3
rd
Ed; New York (John Wiley and
sons), 257 269.
[2] BEEK W.J., MUTTZALL K.M.K., van HEUVEN J.W. (1999): Transport phenomena 2
nd
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; Trans-
actions AIME, 603 617.
[6] VAARNO J., PITKL 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., Lpez C., Barrn M.A. and Gonzlez J. (2007): Computer si-
mulation of the early stages of blown in a Peirce - Smith converter; 6
th
Intl. Conference
Cobre/Copper 2007, Vol. III Book I, 457 469.

Numerical Simulation of Air Blowing
into a Copper Matte in a P-S Converter
Using a Convergent / Divergent Nozzle
C. Lpez, A. Almaraz, I. Arellano, E. Martnez, G. Plascencia M. A. Barrn
CIITEC IPN Depto. de Materiales, UAM-Azcapotzalco
Cerrada Cecati s/n Av. San Pablo 180
Mxico, D.F. C.P. 02250 Mxico Mxico, D.F. C.P. 02200 Mxico
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 oxy-
gen 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 opera-
tional 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 conver-
gent/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 prob-
lem thus opening the door to explore in further detail the implementation of this type of devices.
Lpez, Almaraz, Arellano, Martnez, Barrn, 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 result-
ing 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 productiv-
ity, 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 converters 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 modi-
fication 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: Schematics of a P-S converter, after Davenport et al. [1].

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 P
b
/ P
0
, where P
b
is the back pressure or pres-
sure at the outlet and P
0
is the inlet pressure.
The convergent/divergent nozzle as shown in Figure 2 can be divided in three planes: Plane 1 corre-
sponds 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]:
0
2
= +
d
u du u
s
(1)
Where u is the mean velocity of the flow (m/s), u
s
is the sonic velocity (m/s), defined by Ma=1 and
is the density of the fluid (kg/m
3
).
On the other hand, the continuity equation applied to a duct with variable cross sectional area, has
the general form [3]:
( ) 0 , , = A u
dL
d
(2)
After some manipulation, equation (2) yields:
A
dA
u
u
u
du
s
=
|
|
\
|
1
2
2
(3)
And the ratio u/u
s
defines the Mach number (Ma), then equation (3) can be expressed as:
A
dA
Ma u
du
|
\
|
=
1
1
2
(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<1), an increase in the cross sectional area of the nozzle will re-
sult in a decrease in the velocity and vice versa. However, when the velocity is supersonic, the ve-
locity of the flow will increase as the cross sectional area does; this is due to the reduction in density
for a supersonic flow [3].
(a) (b)
Thus if a fluid is to be accelerated, a convergent/divergent nozzle must be designed so the sonic
velocity must be achieved at the throat; if the sonic condition is not fulfilled at the throat, the outlet
velocity will then decrease rather than increase in the divergent section of the nozzle.
To achieve such sonic condition at the throat, it is necessary to perform an energy balance between
locations 1 and 2. Such balance must assume adiabatic conditions [2-4].
Figure 2: Hydrodynamic features of a convergent-divergent nozzle. (a) velocity profiles,
(b) velocity and density profile for supersonic flow through the nozzle, after Szekely [3].
The result of such balance can be expressed in terms of the Mach number as follows [3]:
|
|
|
\
|

= =
1
1
2
1
2
1
1 2
P
P
Ma
(4)
3 Numerical model
Fluid flow is described by the Navier-Stokes equations, which in the X direction can be expressed
as [4]:
( )
( )
|
|
\
|
i
j
j
i
eff
j i j
j i
i
x
u
x
u
x x
P
x
u u
t
u
(5)
Where is the fluid density, u
i
is the velocity component in the i direction (m/s), t is the time (s), x
j

is the j spatial coordinate (m), P is pressure (Pa), and
eff
is the effective fluid viscosity (Pa s). To
maintain the mass balance in the system, the equation of continuity must be solved along with the
Navier-Stokes equations:
0 =
j
j
x
u
(6)
For gas injection a huge amount of turbulence is expected. Turbulent flows inherently bear a wide
range of time and space scales, thus, turbulent flows properties instantly change both in time and
space. Therefore, turbulent flows exhibit the presence of eddies whose sizes may be distributed in a
wide range. Larger eddies have more kinetic energy compared with the smaller ones. As the size of
eddies increase, the more effective they become at transferring momentum.
To account for this energy and to be able to depict as accurate as possible the turbulent flow, several
turbulence models have developed over the years [5]. Among these models, the - model is the
most popular one, due to its simplicity. This - model yields reliable results for flow characteriza-
tion while it does not require as much computational resources as other turbulence models do [6].
The purpose of this turbulence model is to predict eddy viscosity by means of deriving an exact
equation for (rate of turbulent kinetic energy dissipation) and to find suitable closure approxima-
tions for such equations [5, 6]. However, the derivation of such equation may become a very diffi-
cult task so it has been proposed the use of a simpler relationship [5, 7]:
5 . 1
2
=
D
(7)
On the other hand, represents the turbulence kinetic energy of the flow; this kinetic energy term
can also be expressed in terms of a transport equation; however, it has been found that a good repre-
sentation of can be obtained from the following expression [5-7]:
2
01 . 0
in
u = (8)
Where is the turbulence kinetic energy (m/s)
2
, is the rate of turbulent kinetic energy dissipation
(m
2
/s
3
), D is the diameter of the nozzle inlet (m) and u
in
is the velocity of the injecting gas at the
inlet (m/s).
The numerical model requires that the set of equations (5, 6, 7 and 8) are solved simultaneously.
4 Model conditions
In order to model the hydrodynamics of the blowing operation with a convergent/divergent nozzle,
the geometry and the mesh shown in Figure 3 were used.
To compare the flow behaviour of this tuyere, similar computations were conducted asumming a
normal tube as the injection nozzle. In every case, an injection gas velocity of 120 m/s was assumed.
The mesh consisted of 183,548 cells, and 37,242 nodal points. The commercial Fluent software
was used to carry out all the numerical calculations. All the computations were conducted in 3-D. It
is worth mentioning that injection from only one nozzle was considered in the calculations shown in
the text; considering more nozzles inyecting air into the matte simultaneously as it actually happens
in the converting operation would complicate the flow analysis. Results from the calculations are
shown below.

Figure 3: Definition of the domain for CFD computations. (A) System to analyze. (B) Meshed
domain.
5 Numerical model results and discussion
Figure 4, shows the development of the plume for both, the conventional injecting tube and the con-
vergent divergent nozzle at different blowing times. As mentioned, the velocity of the injecting gas
was kept as 120 m/s.
As seen in this figure, two injection times are presented. The first results correspond to 0.1 s of
blowing. It is clear that the formation of the first bubble is almost identical for each injecting nozzle.
However, it is apparent that in the case of the converging/diverging nozzle, the bubble detaches
from the nozzle a distance deeper into the matte rather than next to the nozzle as is the case of the
current injection tube.
As time elapses, it is evident from the calculated flows that the bubbling regime exhibit marked
differences once the plume has been developed. In the case of the current operation, the plume lies
(A) (B)
matte
air
Injection tuyere
closer to the converter walls, thus matte recirculation next to the vessel wall is expected, and the
well known wear and erosion of the refractory lining occurs.
On the other hand, once the plume with the converging/diverging nozzle develops, it establishes
well ahead of the refractory lining so refractory erosion and wear problems should significantly de-
crease. However, this flow regime shows that in spite of avoiding the refractory problems associated
with the recirculating flow, more matte projections out of the converter are expected. This can cause
a safety problem.
Another aspect to look at is the velocity distribution at the exit of the nozzle. In principle, a conver-
gent/divergent nozzle is a device that allows for a stream velocity increment. Such velocity incre-
ment is driven by the pressure drop across the nozzle. However, not always the fluid passing
through a converging/diverging nozzle accelerates. Even more, the fluid may decelerate in the di-
verging section of the nozzle if the pressure at the nozzle outlet is not the proper one. Then, flow
through one of these flow devices is determined by the outlet pressure.
Figure 5, compares the velocity vectors at the exit of both, the current nozzle and the proposed one
at different injection times. From Figure 5 is evident how the gas velocity changes at the nozzle exit.
In the case of the current nozzle, the velocity vectors remain fairly constant across the injection
point, so they transfer all the momentum from the gas to the molten matte. When the matte clogs the
nozzle due to recirculation, it is evident from Figure 5 that the gas flow still has the capacity to
penetrate the molten matte while agitating the bath. During the partial clogging of the nozzle, the
flow accommodates in such a way that the stream lines keep moving into the molten material, en-
suring the continuous agitation of the bath. Furthermore, the velocity vectors tend to keep its magni-
tude so no momentum is lost.
In the case of the convergent/divergent nozzle, the flow behaviour is quite different than that shown
by the conventional injecting tube.

Figure 4: Calculated gas stream lines during gas blowing into the copper matte after 0.1 and 3.8 s.
(A) Conventional tube. (B) Convergent divergent nozzle.
As seen in Figure 5, at the convergent/divergent nozzle exit, the velocity vectors exhibit different
trajectories and magnitudes with respect of those at the nozzle entrance. This behaviour is also op-
posite to that shown by the current injection tube.
In some instances, the velocity vectors show the gas entering into the diverging section of the nozzle
rather than leaving it. This causes a siphon effect that promotes the entrance of molten matte into
the nozzle, so clogging phenomena is likely to happen.
The obstruction of the nozzle has a significant effect on the injecting gas flow pattern. It is evident
from the images in Figure 5 that the gas plume is strongly disturbed by the matte blockage.

(A)
(B)
matte
matte
matte
matte
matte
matte
matte
matte
Figure 5: Comparison of flow patterns between the conventional injecting tube and the
converging/divergin nozzle at different injection times.


Figure 6: Comparison of flow between the conventional injecting tube and the converg-
ing/diverging nozzle at different injection times.
Current
tube
Conv. Div.
nozzle
As seen in Figure 6, the obstruction of the nozzle outlet (convergent/divergent) provokes the plume
to move away from the vessel walls; the effect of moving the plume away from the vessel walls has
been already described.
The clogging effects may be avoided if instead of decelerating the flow at the nozzle outlet, it gets
accelerated. To do so, different convergent/divergent configurations should be tested.
The results shown in this paper suggest that the gas increases its velocity towards its maximum at
the throat and when entering into the diverging section of the nozzle, the velocity is lost. This in
turn means that the pressure decreases through the converging section, reaching its minimum at the
throat, subsequntly increasing in the diverging section of the nozzle, thus compromising the velocity
of the gas at the outlet.
To overcome these flow restrictions, it is necessary to design a nozzle able to provide a supersonic
flow at the outlet. This means that it is necessary to avoid any shock within the nozzle [4]. If this
condition can be achieved, and the gas is accelerated in the divergent section of the nozzle, no matte
clogging will occur and the gas will be able to transfer its momentum to the molten phase.
Therefore, this type of nozzles may improve the productivity of the P-S converter.
6 Future work
Tests with a plexiglass model will be conducted. Different convergent/divergent configurations will
be tested, pressure drop along the test nozzles will be measured and recorded.
In addition to the tests described above, a series of experiments both numerical and with a
plexiglass model will be conducted in order to study the interation of different nozzles operating
simultaneously on the liquid phase.
7 Conclusions
CFD calculations have been conducted to predict flow conditions for gas injection using a conver-
gent/divergent type of nozzle in the P-S converter. These first results show that in principle the use
of the convergent/divergent nozzles may help to decrease the well known wear and erosion suffered
by the converter lining. However, the nozzle chosen to carry out our calculations has the problem of
promote clogging due to a siphon effect. This is caused by a decrease in the gas pressure at the noz-
zle inlet followed by a pressure increase in the diverging zone of the nozzle. This problem can be
overcome by proper design of the injection nozzles.

Acknowledgements
The authors wish to thank SIP IPN grants # 20080019 & 20090503 to conduct this research.
References
[1] DAVENPORT W.G., KING M., SCHLESINGER M. and BISWAS A.K. (2002): Extractive metallurgy
of copper 4th Ed; Oxford (Pergamon), 137 - 149.
[2] GREITZER E.M., TAN C.S. and GRAF M.B. (2004): Internal flow concepts and applications;
Cambridge, (Cambridge), 72 - 76.
[3] SZEKELY J. (1979): Fluid flow phenomena in metals processing, New York (Academic Press),
49 - 55.
[4] ENGEL Y.A. and CIMBALA J.M. (2006): Fluid mechanics fundamentals and applications, New
York (McGraw-Hill), 629 - 632.
[5] PIQUET J. (2003): Turbulent flows: models and physics; New York (Springer), 143 - 160.
[6] FERZIGER J.H. and PERIC M. (2001): Computational methods for fluid dynamics 3
rd
Ed.; New
York (Springer), 265 - 300.
[7] ALMARAZ A, ARELLANO I.A., LPEZ C., PLASCENCIA G., BARRN M.A. and JARAMILLO D.
(2009): Anlisis de flujo en un sensor de placa de orificio, Ing Hidraul Mex, 107 - 120.

Copper Volume 2 PDF

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Copper Volume 2 PDF

Hochgeladen von

Copyright:

Verfügbare Formate

Proceedings

Version 14.04 for

was installed to provide continuous

was installed to provide continuous injection of

and the impact of this project on converting

was selected as the superior concept. The operating

are two pressure vessels and a horizontal rotary

at the Kidd smelter. Granular limestone

has performed extremely well.

at the Kidd smelter to continuously inject limestone into the

Input power during bubble growth, W/kg

Bath density, kg/m

cyclones. Engineering has been done for a fourth

cyclone is an alternative to the incineration or dumping of these resi-

technology are as follows:

cyclone for the treatment of copper, zinc and

- A Cost-Effective Recycling Technology for the Steel and Automotive-

Das könnte Ihnen auch gefallen