VOLUME 123 NO. 10 OCTOBER 2023
SAIMM STUDENT EDITION
SAIMM – DEVELOPING THE MINING INDUSTRY’S FUTURE LEADERS. THE ANNUAL STUDENT COLLOQUIUM across Southern Africa to participate and share their work with peers and industry. Winners came from the University of Zimbabwe, the University of the Witwatersrand, and the University of Pretoria. The next chapter in career development for students is to start working in industry and become young professionals. The SAIMM supports young professionals 35 years old and younger by providing a platform to further share knowledge and research projects by way of the bi-annual Young Professionals Conference.
Photos: Courtesy of Minerals Council South Africa
The Southern African Institute of Mining and Metallurgy has been organizing and presenting the annual Student Colloquium since 2002, to afford the best final-year mining and metallurgical students an opportunity to present their final year projects to an audience of mining and metallurgical industry experts. These students are our future young professionals and will be fundamentally affected by how the industry operates. We must support and assist our future young professionals! As Nelson Mandela observed: ‘Education is the most powerful weapon which you can use to change the world’. The 19th Annual Student Colloquium took place on 7 November 2023 in Johannesburg. Over 150 students attended from
The 2023 conference, the 6th in the series, will take place on 16-17 November in Johannesburg. The theme this year is Empowering Young Professionals for Sustainable and Innovative Mining and Processing Practices, and the conference promises to be an engaging and informative event that will explore the latest trends, ideas, and innovations in the minerals industry. This conference offers young professionals the opportunity to network with peers and industry experts, establish professional relationships, and collaborate on future projects. A conference dedicated to young professionals emphasizes the importance of empowering young leaders to drive innovation and sustainability in the South African minerals industry. The SAIMM Limpopo Branch Graduate Colloquium was held on 29 September 2023. The objective was to create an opportunity for mining and metallurgy students to present their final year projects to an audience. The event was attended by over 70 delegates, which included industry leaders as well as academics.
Close to 20 abstracts from final year students, postgraduates, bursars, and interns were submitted. The colloquium proceedings are available online. Limpopo Branch Graduate Colloquium 2023 - Colloquium Material - Dropbox Papers from the winners will be considered for publication in the Student Edition of the Journal, which will be published in early 2024. Annual Student Debate The Johannesburg Branch has been arranging an Annual Student Debate since 2014 in collaboration with the University of Johannesburg and the University of the Witwatersrand. The aim is to provide a platform for students to hone their presentation and debating skills. Each year, the Johannesburg Branch Committee selects a topic that is relevant at the time. The debating teams are given an opportunity to rehearse beforehand, with mentoring provided by the Branch Committee members. The members of the judging panel are selected from industry, and the winning team walks away with a well-deserved prize.
The Southern African Institute of Mining and Metallurgy OFFICE BEARERS AND COUNCIL FOR THE 2023/2024 SESSION
PAST PRESIDENTS
Honorary President
* W. Bettel (1894–1895) * A.F. Crosse (1895–1896) * W.R. Feldtmann (1896–1897) * C. Butters (1897–1898) * J. Loevy (1898–1899) * J.R. Williams (1899–1903) * S.H. Pearce (1903–1904) * W.A. Caldecott (1904–1905) * W. Cullen (1905–1906) * E.H. Johnson (1906–1907) * J. Yates (1907–1908) * R.G. Bevington (1908–1909) * A. McA. Johnston (1909–1910) * J. Moir (1910–1911) * C.B. Saner (1911–1912) * W.R. Dowling (1912–1913) * A. Richardson (1913–1914) * G.H. Stanley (1914–1915) * J.E. Thomas (1915–1916) * J.A. Wilkinson (1916–1917) * G. Hildick-Smith (1917–1918) * H.S. Meyer (1918–1919) * J. Gray (1919–1920) * J. Chilton (1920–1921) * F. Wartenweiler (1921–1922) * G.A. Watermeyer (1922–1923) * F.W. Watson (1923–1924) * C.J. Gray (1924–1925) * H.A. White (1925–1926) * H.R. Adam (1926–1927) * Sir Robert Kotze (1927–1928) * J.A. Woodburn (1928–1929) * H. Pirow (1929–1930) * J. Henderson (1930–1931) * A. King (1931–1932) * V. Nimmo-Dewar (1932–1933) * P.N. Lategan (1933–1934) * E.C. Ranson (1934–1935) * R.A. Flugge-De-Smidt (1935–1936) * T.K. Prentice (1936–1937) * R.S.G. Stokes (1937–1938) * P.E. Hall (1938–1939) * E.H.A. Joseph (1939–1940) * J.H. Dobson (1940–1941) * Theo Meyer (1941–1942) * John V. Muller (1942–1943) * C. Biccard Jeppe (1943–1944) * P.J. Louis Bok (1944–1945) * J.T. McIntyre (1945–1946) * M. Falcon (1946–1947) * A. Clemens (1947–1948) * F.G. Hill (1948–1949) * O.A.E. Jackson (1949–1950) * W.E. Gooday (1950–1951) * C.J. Irving (1951–1952) * D.D. Stitt (1952–1953) * M.C.G. Meyer (1953–1954) * L.A. Bushell (1954–1955) * H. Britten (1955–1956) * Wm. Bleloch (1956–1957) * H. Simon (1957–1958) * M. Barcza (1958–1959) * R.J. Adamson (1959–1960)
Nolitha Fakude President, Minerals Council South Africa
Honorary Vice Presidents
Gwede Mantashe Minister of Mineral Resources and Energy, South Africa Ebrahim Patel Minister of Trade, Industry and Competition, South Africa Blade Nzimande Minister of Higher Education, Science and Technology, South Africa
President
W.C. Joughin
President Elect E. Matinde
Senior Vice President G.R. Lane
Junior Vice President T.M. Mmola
Incoming Junior Vice President M.H. Solomon
Immediate Past President Z. Botha
Honorary Treasurer E. Matinde
Ordinary Members on Council W. Broodryk Z. Fakhraei R.M.S. Falcon (by invitation) B. Genc K.M. Letsoalo S.B. Madolo F.T. Manyanga K. Mosebi
M.C. Munroe S. Naik G. Njowa S.J. Ntsoelengoe S.M. Rupprecht A.T. van Zyl E.J. Walls
Co-opted Council Members M.A. Mello
Past Presidents Serving on Council N.A. Barcza R.D. Beck J.R. Dixon V.G. Duke I.J. Geldenhuys R.T. Jones A.S. Macfarlane
C. Musingwini S. Ndlovu J.L. Porter M.H. Rogers D.A.J. Ross-Watt G.L. Smith W.H. van Niekerk
G.R. Lane – TP Mining Chairperson Z. Botha – TP Metallurgy Chairperson K.W. Banda – YPC Chairperson S. Nyoni – YPC Vice Chairperson
Branch Chairpersons Botswana DRC Johannesburg Limpopo Namibia Northern Cape North West Pretoria Western Cape Zambia Zimbabwe Zululand
Vacant Not active N. Rampersad S. Zulu Vacant I. Tlhapi I. Tshabalala Vacant A.B. Nesbitt J.P.C. Mutambo (Interim Chairperson) Vacant C.W. Mienie
*Deceased
* W.S. Findlay (1960–1961) * D.G. Maxwell (1961–1962) * J. de V. Lambrechts (1962–1963) * J.F. Reid (1963–1964) * D.M. Jamieson (1964–1965) * H.E. Cross (1965–1966) * D. Gordon Jones (1966–1967) * P. Lambooy (1967–1968) * R.C.J. Goode (1968–1969) * J.K.E. Douglas (1969–1970) * V.C. Robinson (1970–1971) * D.D. Howat (1971–1972) * J.P. Hugo (1972–1973) * P.W.J. van Rensburg (1973–1974) * R.P. Plewman (1974–1975) * R.E. Robinson (1975–1976) * M.D.G. Salamon (1976–1977) * P.A. Von Wielligh (1977–1978) * M.G. Atmore (1978–1979) * D.A. Viljoen (1979–1980) * P.R. Jochens (1980–1981) * G.Y. Nisbet (1981–1982) A.N. Brown (1982–1983) * R.P. King (1983–1984) J.D. Austin (1984–1985) * H.E. James (1985–1986) H. Wagner (1986–1987) * B.C. Alberts (1987–1988) * C.E. Fivaz (1988–1989) * O.K.H. Steffen (1989–1990) * H.G. Mosenthal (1990–1991) R.D. Beck (1991–1992) * J.P. Hoffman (1992–1993) * H. Scott-Russell (1993–1994) J.A. Cruise (1994–1995) D.A.J. Ross-Watt (1995–1996) N.A. Barcza (1996–1997) * R.P. Mohring (1997–1998) J.R. Dixon (1998–1999) M.H. Rogers (1999–2000) L.A. Cramer (2000–2001) * A.A.B. Douglas (2001–2002) S.J. Ramokgopa (2002-2003) T.R. Stacey (2003–2004) F.M.G. Egerton (2004–2005) W.H. van Niekerk (2005–2006) R.P.H. Willis (2006–2007) R.G.B. Pickering (2007–2008) A.M. Garbers-Craig (2008–2009) J.C. Ngoma (2009–2010) G.V.R. Landman (2010–2011) J.N. van der Merwe (2011–2012) G.L. Smith (2012–2013) M. Dworzanowski (2013–2014) J.L. Porter (2014–2015) R.T. Jones (2015–2016) C. Musingwini (2016–2017) S. Ndlovu (2017–2018) A.S. Macfarlane (2018–2019) M.I. Mthenjane (2019–2020) V.G. Duke (2020–2021) I.J. Geldenhuys (2021–2022) Z. Botha (2022-2023)
Editorial Board S.O. Bada R.D. Beck P. den Hoed I.M. Dikgwatlhe R. Dimitrakopolous* B. Genc R Hassanalizadeh R.T. Jones W.C. Joughin A.J. Kinghorn D.E.P. Klenam J. Lake H.M. Lodewijks D.F. Malan R. Mitra* H. Möller C. Musingwini S. Ndlovu P.N. Neingo S.S. Nyoni M. Phasha P. Pistorius P. Radcliffe N. Rampersad Q.G. Reynolds I. Robinson S.M. Rupprecht K.C. Sole A.J.S. Spearing* T.R. Stacey E. Topal* D. Tudor* D. Vogt* *International Advisory Board members
Editor /Chairperson of the Editorial Board R.M.S. Falcon
Typeset and Published by The Southern African Institute of Mining and Metallurgy PostNet Suite #212 Private Bag X31 Saxonwold, 2132 E-mail: journal@saimm.co.za
Printed by
Camera Press, Johannesburg
VOLUME 123 NO. 10 OCTOBER 2023
Contents Journal Comment: Student Edition by B. Genc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
iv
Presidential Address: Lessons in diversity and inclusion from the Springboks and others by W.C. Joughin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
v
NEWS OF INTEREST
SAMCODES News October 2023 . . . . . . . . . . . . . . . . . . . . . . . . . . .
vi-vii
THE INSTITUTE, AS A BODY, IS NOT RESPONSIBLE FOR THE STATEMENTS AND OPINIONS ADVANCED IN ANY OF ITS PUBLICATIONS. Copyright© 2023 by The Southern African Institute of Mining and Metallurgy. All rights reserved. Multiple copying of the contents of this publication or parts thereof without permission is in breach of copyright, but permission is hereby given for the copying of titles and abstracts of papers and names of authors. Permission to copy illustrations and short extracts from the text of individual contributions is usually given upon written application to the Institute, provided that the source (and where appropriate, the copyright) is acknowledged. Apart from any fair dealing for the purposes of review or criticism under The Copyright Act no. 98, 1978, Section 12, of the Republic of South Africa, a single copy of an article may be supplied by a library for the purposes of research or private study. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means without the prior permission of the publishers. Multiple copying of the contents of the publication without permission is always illegal. U.S. Copyright Law applicable to users In the U.S.A. The appearance of the statement of copyright at the bottom of the first page of an article appearing in this journal indicates that the copyright holder consents to the making of copies of the article for personal or internal use. This consent is given on condition that the copier pays the stated fee for each copy of a paper beyond that permitted by Section 107 or 108 of the U.S. Copyright Law. The fee is to be paid through the Copyright Clearance Center, Inc., Operations Center, P.O. Box 765, Schenectady, New York 12301, U.S.A. This consent does not extend to other kinds of copying, such as copying for general distribution, for advertising or promotional purposes, for creating new collective works, or for resale.
Honorary Legal Advisers M H Attorneys
Auditors
Genesis Chartered Accountants
Secretaries
The Southern African Institute of Mining and Metallurgy 7th Floor, Rosebank Towers, 19 Biermann Avenue, Rosebank, 2196 PostNet Suite #212, Private Bag X31, Saxonwold, 2132 E-mail: journal@saimm.co.za
Advertising Representative Barbara Spence Avenue Advertising Telephone (011) 463-7940 . E-mail: barbara@avenue.co.za ISSN 2225-6253 (print) . ISSN 2411-9717 (online)
Directory of Open Access Journals
▶ ii
OCTOBER 2023
VOLUME 123
The Journal of the Southern African Institute of Mining and Metallurgy
PROFESSIONAL TECHNICAL AND SCIENTIFIC PAPERS Contact sorption drying of chromite concentrates by C. Snyman, M. le Roux, S. Engelbrecht, and Q.P. Campbell. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Due to the ultrafine particle size required for effective processing of chromite ores, dewatering of the concentrates is challenging. Using contact sorption drying, it was possible to achieve the target moisture content of 8–10% in less than 10 minutes. The sorbent-to-chromite mass ratio used had a significant influence on the required contact time and the reusability of the sorbents. Real-time gypsum quality estimation in an industrial calciner: a neural network-based approach. Real-time gypsum quality estimation in an industrial calciner: A neural network-based approach by M. Jacobs, R-D. Taylor, F.H. Conradie, and A.F. van der Merwe. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Total bound moisture (TBM) is a typical quality indicator of industrial-grade gypsum. TBM analysis is a lengthy laboratory procedure, and an artificial neural network (ANN) TBM inference measurement is proposed as a fast online alternative. A strong correlation between TBM and the gypsum hemihydrate and anhydrite content was found. A network topology consisting of one hidden layer with logarithmic-sigmoid (logsig) and pure linear (purelin) transfer functions showed the best performance (R > 90%). Structural frame analysis of an electrically powered robotic subsea dredging crawler under static loading conditions by M.O. Ojumu and A.K. Raji . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Structural design is the most critical aspect for robotic subsea dredging crawlers. Current ocean crawlers are hydraulically powered. In this research a scaled-down model was developed to simulate and perform static loading analysis of an electricallypowered dredging crawler. The results f were used to identify specific areas to be reinforced in future crawler designs, and for considering alternative materials. Electrical resistivity of heat-treated charcoal by R.D. Cromarty, S. Bharat, and D. Odendaal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The aim of this investigation was to determine the effect of high-temperature heat treatment on the electrical resistivity of charcoal. This included considering the effect of different wood types (eucalyptus and black wattle), temperatures, and residence times on the resistivity. It was found that as the heat treatment temperature increased, the electrical resistivity of the charcoal decreased. Longer residence times decreased the resistivity, but this effect was not pronounced. Assessment of coal washability data obtained via the RhoVol analyser by D. Stone, Q.P. Campbell, M. le Roux, and M. Fofana. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The established float-and-sink method for obtaining coal washabikity data was compared with a new image-based method using the RhoVol analyser. The results showed that the RhoVol method was more rapid, safe, and precise, but tended to consistently underestimate the density of the coal sample, most likely due to the varying coal porosity.
479
483
491
501
509
Faculty of Engineering The School of Chemical and Minerals Engineering at Northwest University (Potchefstroom campus) offers programmes that are approved and accredited by the Engineering Council of South Africa - ECSA Our specialised Minerals Processing programme offers scholars specialised studies that deal with the physical and chemical processes used to extract metals and other commodities from ores. These include diamonds, coal, platinum, gold and other precious metals, as well as base metals. The production, processing and export of these commodities is the largest contributor to South Africa’s foreign earnings. This sector is also one of the largest employers in the country. Chemical Engineering in mining? The responsibilities of a Chemical Engineer in the minerals sector include the understanding and design of mineral processing systems with a focus on safety, sustainability and the community, the close monitoring and performance of the process, and addressing the
The Journal of the Southern African Institute of Mining and Metallurgy
various economic and technological challenges in mineral processing plants in a changing world. On a research level, we specialise in the following: • Sustainable bio-energy and biotechnology • Clean coal processing with emphasis on fine waste and discard coal repurposing • Dry processing and dewatering of coal and other commodities • Emissions control, especially in terms of particulates, sulphur and carbon dioxide • Mine wastewater treatment and pollution control • Advanced modelling and simulation of systems • Hydrometallurgy with emphasis on energy storage systems • Other mineral processing using gravity separation • Process safety For more about the NWU Faculty of engineering, our personnel and how to apply click on the link below https://engineering.nwu.ac.za/
VOLUME 123
OCTOBER 2023
iii ◀
Copper Cobalt
al Journ
Student Edition
nt mme
Co
W
elcome to another edition of papers for the Student Edition. Most of the papers published in this Student Edition are based on the annual Student Colloquium of 2022. The Colloquium, organised by the Southern African Institute of Mining and Metallurgy (SAIMM) since 2002, aims to identify the best final-year mining and metallurgical engineering students’ presentations. The papers presented at the Colloquium are based on the students’ final year projects. Our mining engineering students at Wits University, the School of Mining Engineering (Wits Mining) have a final-year course called Project Report, usually based on projects carried out by students on a mine during the vacation work session between the third and fourth year of study. It is a good opportunity for the students to showcase what they have learned during their summer vacation work. Wits Mining selects the top three presentations to take part in the Colloquium, similar to the other schools/departments in the country. The best performers amongst the participants were chosen by the panel of judges in the Colloquium and the winners were asked to prepare a paper to be published in the Journal of the SAIMM (JSAIMM). As with any paper submitted to the JSAIMM, the students’ papers are also subject to the Journal’s peer reviewing process. In the last few years, the quality of the student papers submitted to the JSAIMM for publication has deteriorated, as most of them did not provide anything new. Novelty in a journal paper is essential. In order to address this, an initiative has begun with the help of the mining industry to realign the communication channels between the industry and academia. This initiative has a number of aims and objectives - one of which is to relook at the project topics given to the students; the project topics have a direct impact on the quality of the student report produced. Although it is still early days, this initiative has the potential to help generate more publishable papers in the JSAIMM. Further developments in this regard will be communicated to the SAIMM community. In this Student Edition of the Journal, five papers have been selected, covering topics ranging from coal washability, contact sorption drying and real-time gypsum quality estimation in an industrial calciner. In addition, papers about electric-powered robotic subsea dredging crawler and electrical resistivity of heat-treated charcoal are some of the interesting reads. Enjoy the October edition of the Journal! B. Genc
▶ iv
OCTOBER 2023
VOLUME 123
The Journal of the Southern African Institute of Mining and Metallurgy
nt’s
Lessons in diversity and inclusion from the Springboks and others
de Presi
er
Corn
A
s I write this article, our Springboks, led by their inspirational captain Siya Kolisi, have just won the rugby world cup final for the fourth time, beating our traditional rivals, the All Blacks in the final. The journey has been tough for the talented and dedicated team, and stressful for the ardent South African fans. Onepoint winning margins and seizing the lead from extremely strong opposition in the dying seconds is not good for the hearts of the supporters, but the subsequent jubilation makes it worthwhile. Turning the game around requires team cohesion, confidence in the abilities of fellow players, and absolute commitment to each objective. When the stakes are high mistakes are made, but the team rallies together and provides support rather than criticism. Successes are hard-won and celebrated exuberantly. It is quite remarkable that this level of cohesion is achieved when the team is so diverse, with different backgrounds, races, shapes, sizes, and skills. There is obvious mutual respect among the players and recognition of their individual contributions toward a common goal. This is an incredible achievement, particularly when you look back to the apartheid era when there were three separate rugby governing bodies, each representing a different racial group. At that time only the South African Rugby Board (SARB), which represented white people, had any say in international competitions, and selected the national team. John Cruise’s 2011 paper on ‘The gender and racial transformation of mining engineering in South Africa’ summarizes the history of mining legislation (https://www.saimm.co.za/Journal/v111n04p217.pdf). Legislation in the South African minerals industry actively prevented racial and gender diversity. Women were not allowed to work underground after the promulgation of the 1911 Mines and Works Act. The objective at the time was to protect women and children from exploitation as labourers in mines. Certain racial groups were prevented from holding blasting certificates and other certificates of competency, and this was purely for job reservation. Since these certificates were necessary for more senior roles, it was not possible for these groups to progress in their career. These regulations were repealed during the 1990s to fall in line with the Constitution of South Africa. Cruise describes the subsequent transformation of mining engineering education until 2010, which was quite rapid. He shows that the racial composition of mining graduates had reached the point at which it matched the country demographics. Gender ratios had improved from zero to around 30%, where they currently remain. Zelmia Botha’s presidential address last year highlighted the relatively low gender ratios (15% to 40%) in STEM (Science, Technology, Engineering and Mathematics) graduates worldwide, and the societal and environmental factors that contribute to the under-representation. While it remains a challenge to get to 50% representation, there are many women entering the minerals industry. There is growing evidence to suggest that diverse companies are more innovative, creative, and commercially successful than non-diverse ones. Having a variety of skills, experiences, and knowledge will bring new and better ideas. However, proper inclusion is necessary to unlock the benefits to be gained from diversity. This requires empowering individuals to express their opinions and to pursue their ideas. Steve Jobs famously said ‘If you want to hire great people and have them stay working for you, you have to let them make a lot of decisions and you have to be run by ideas, not hierarchy. The best ideas have to win, otherwise good people don’t stay.’ Leaders need to create a safe, equitable environment, which is free of bullying and harassment. They also need to recognize and celebrate successes. Employees should feel a sense of belonging. ‘The greatest contribution of a leader is to make other leaders’ Simon Sinek. Successful and diverse leaders are role models, and attract diverse and talented employees. W.C. Joughin President, SAIMM
The Journal of the Southern African Institute of Mining and Metallurgy
VOLUME 123
OCTOBER 2023
v ◀
SAMCODES Environmental, Social and Governance (ESG) Working Group constituted to revise the current South African Guideline for the reporting of Implications of S-K 1300 regulations and disclosures for dual-listed companies on the JSE and NYSE – an Update ESG parameters (SAMESG Guideline). (10 October 2023) Education and Promotion
SAMCODES for Young Professionals webinar Professor Steven Rupprecht is due to hold the ‘SAMCODES for Young Professionals’ webinar on 26 October 2023.
A successful one-day webinar on ‘Implications of S-K 1300 regulations and disclosures for dual-listed companies on the JSE and NYSE – an
The SAMCODES Committee (SSC) by has constituted a multi-disciplinary group to update update’ was held on Standards 10 October 2023. This was attended some 50 delegates with presentations from 16working speakers representing dual-listedthe mining companies in South Africa and various consulting companies. The intent of the webinar was to provide a knowledge update based
existing SAMESG Guideline otherconcerning elementstheofSEC’s SAMCODES, if required) inensuing orderfrom to ensure alignment on feedback from listed entities and(and consultants views and recommendations round 1 of S-K 1300 with reporting. the rapidly evolving expectations of investors (and society) for disclosure of environmental, social and Key themes in the programme included lessons learned, opportunities to streamline the reporting process, assessment of materiality, QP
governance (ESG) considerations integral partresource of Mineral Reporting disclosures. liability, ESG/sustainability developments,as andan exclusive mineral reporting.
Mr Ben Parsons of SRK Consulting from Denver, USA, who has regular technical interaction with the SEC, delivered the keynote
An organization’s approach the management of ESG considerations isfilings rapidly becoming defining feature address in which he provided SEC to feedback and valuable guidance in compiling the various to go to the SEC. aOther presenters also in shared the invaluable lessons learned, which informedaccurate the panel discussion at the end of the day. the market. Investors continue to demand and transparent information on ESG performance to identify The panel discussion considered the key themes around what constitutes material assets and frequency of reporting material changes.
and prioritisealsofunding top tier Guideline was for anaffected industry first when it was The discussion generatedfor some ideas withinvestments. regard to aligningThe with SAMESG SEC and streamlined reporting parties. For thoseinwho were not able to attendlessons the webinar, presentations discussion will be made at a nominal R200.00 of published 2017. Since then, in the respect of itsand implementation haveavailable been learnt andcost theof world – details will be posted on the SAMCODES website https://www.samcode.co.za/
investor expectations in respect of ESG reporting has evolved. It is now an ideal time to future proof the Council for Geoscience (CGS) Meeting (6 July 2023)
SAMCODES to accommodate these lessons and the evolving ESG requirements. Dr Tania Marshall and Messrs Sifiso Siwela and Andy McDonald held a virtual meeting with Mr Mosa Mabuza of the CGS and members of
his executive on 6 July 2023. To date, the Working Group has been assimilating the outcomes of the Geological Society of South Africa’s Developments with the African Minerals & Energy Resource Classification and Management System (AMREC), which is aligned with
the United Nations Framework Classification for in Resources andto thedevelop Pan-African Resource Reporting (PARC) classification (GSSA) ESG Inquisition that was held 2021 (UNFC) in order inputs that need Code to be considered when systems, were discussed. Consideration of the SAMREC Code and its definitions highlighted the differences and similarities between
updating the and ESGSAMREC. guidance. In addition, direct engagements have withwhereas a select group of key AMREC/UNFC The fundamental difference is that AMREC considers totalbeen mineralheld inventory, SAMREC evaluates the deposit according to Reasonable Prospects for Economic Extraction. stakeholders to obtain their views on the issues that need to be addressed in the update.
Although not specifically discussed, it is important to note that the SAMCODES Standards Committee (SSC) is not established by
statute or legislation. has toprocesses remain independent of government in order for its membership of CRIRSCO to remain valid. The outcomes of The theSSC above have been distilledcontrol in a communications document which the Working The SSC has to be beholden only to its professional bodies, which is why the JSE and mining companies are not co-sponsors of the SSC.
Group is now issuing for broad public review. One of the criticisms that has been leveraged against the current Advanced and Basic SAMREC and SAMVAL Workshops
SAMESG Guideline that thereWorkshop was inadequate consultation ahead its publication. It is important tothethe The Advanced SAMRECisand SAMVAL will not be held during 2023. Dr Tania of Marshall and Mr Ken Lomberg are evaluating
merits of such a workshop for early 2024. Working Group that we therefore attempt to solicit the views of as many stakeholders as possible to inform the Recent experience suggests that the general awareness and understanding of the requirements of the SAMREC and SAMVAL Codes is
scope of work forward. not as good as wasgoing thought. The ongoing training in the Codes will need to be maintained, and perhaps intensified. The option for a more basic course as a webinar split over several weeks is being considered for 2024.
All parties who are directly or indirectly involved with Mineral Reporting and sustainability reporting are urged Compliance and Reporting in the Mineral Industry (MINN7052A) (11-15 September 2023)
toDrreview the initial consultation document from the Working Group and to provide feedback via the survey that Tania Marshall presented some intensive SAMCODES training at an MSc level during this five-day course. From the essays submitted by
the students at the end of thethis course, the level of understanding of the was quite poor. has been prepared for purpose. Persons working inSAMCODES the following disciplines are likely to be well-placed to Some of the lessons learned from these essays as they apply to ongoing SAMCODES training include investigating opportunities of
provide comments: geologists; of Competent Persons Reports; environmental, socialcourse or governance doing more for undergraduate students, authors focusing more on principles and background in the annual SAMREC/SAMVAL run by the GSSA, and addressing issues in ‘SAMCODES in GSSA/SAIMM newsletters. subject matter experts; authors ofSnippets’ corporate sustainability reports; mineral resource valuators; financial
ESG follow-up webinar –investors; 11 July 2023 managers; institutional private investors; investment analysts and stock exchange regulators as a An update on the integration of ESG aspects into the SAMCODES was held on 11 July 2023. Working drafts of the suggested ESG changes
minimum. to the SAMREC Code, SAMREC Table 1, and the SAMVAL Code were circulated to the Chairpersons of the SAMREC, SAMVAL, and
SAMOG Codes prior to the webinar, for information only. The documents were not sufficiently advanced for wider circulation to the Code
The full report is available on the SAMCODES website at https://www.samcode.co.za/. committees.
Professor Michael Solomon delivered the keynote address, followed by presentations describing the ESG landscape and stakeholder
requirements. principles and adopted incorporating thesurvey recommended in the SAMCODES were discussed. A Please createThe awareness ofapproaches this report and in the associated withinchanges your network! panel discussion considered ESG practitioner registration and criteria for defining ESG risk and materiality. The webinar was well with over 200 delegates Comments should beattended submitted via the surveywho linkregistered. by no later than 24 June 2022.
SAMCODES Committee Key Activities
The ESG Working Group looks met forward your input. The SAMCODES SAMREC, SAMVAL, and SAMESG committees in May to andreceiving August 2023 as part of normal quarterly meetings. The SAMREC Committee (Ken Lomberg, Chair) has been involved in the following activities:
Andrew McDonald OCTOBER 2023 VOLUME 123 ▶ vi Chairperson, SAMCODES Standards Committee
The Journal of the Southern African Institute of Mining and Metallurgy
(Continued)
SAMCODES Environmental, Social App, andwhich Governance (ESG) Group ➢ Considerable discussion has focused on the SAMCODES is not really being used despiteWorking a great deal of work that has been put into it. Various ways have been used to get people to use the App, with poor uptake. One of these was a monthly quiz, constituted toresponses revise the current South African Guideline for the reporting of with only nine on average. ➢ Discussion on the ESG definitions proposed by CRIRSCO. ➢ Debate whether SAMREC should RPEEE or RPEE. ESGrequire parameters (SAMESG Guideline).
➢ Recognized that a re-write of the SAMREC Code is required, but waiting for the recommendations from the SAMCODES ESG Working Group. The SAMCODES Committee (SSC) constituted a multi-disciplinary working group to update the ➢ Continued Standards involvement with CRIRSCO via Ken has Lomberg and Roger Dixon. After many years atGuideline the helm, Ken Lomberg haselements stepped down Chair of the SAMREC Committee, to be replaced by Ms Nicole existing SAMESG (and other of asSAMCODES, if required) in order to ensure alignment with Wansbury. This will be confirmed at the SSC’s meeting in November 2023. theThe rapidly evolving expectations of investors (and in society) for activities: disclosure of environmental, social and SAMVAL Committee (Vaughn Duke, Chair) has been involved the following ➢ Discussion the valuation of brines non-solidpart minerals. It was agreed that where a solid material is extracted from the governance (ESG)onconsiderations as anandintegral of Mineral Reporting disclosures. brines, it could be adequately covered under SAMVAL. ➢ There was a lengthy discussion around discount ratesofand the considerations Committee was leaning towardsbecoming a simple approach. The feature in An organization’s approach to the management ESG is rapidly a defining Committee was looking at producing a guidance note in this regard, which would go onto the website. the market. continue to demand accurate and transparent information ESG performance tothe identify ➢ ThreeInvestors new members have joined the Committee. The new member from Impala Platinum willon continue representation on Committee in future. and ➢ prioritise funding for nuances, top tiersuch investments. The SAMESG Guideline was an industry first when it was Discussion on S-K 1300 as what represents a material difference. ➢ Discussion on the G (Governance) in ESG matters and how to include governance issues in mineral valuation. published in 2017. Since then, lessons in respect of its implementation have beenasset learnt and the world of ➢ The Committee is waiting for the recommendations from the SAMCODES ESG Working Group regarding the integration of requirements into SAMVAL.of ESG reporting has evolved. It is now an ideal time to future proof the investorESG expectations in respect The SAMOG Committee (Peter Dekker, Chair) has been busy with updating the Code to make it more in line with current industry SAMCODES to accommodate these lessons and the evolving ESG requirements. practices. SAMESG Guideline Committee Chair) been involved the following activities: To The date, the Working Group has(Teresa beenSteele-Schober, assimilating thehasoutcomes ofinthe Geological Society of South Africa’s ➢ Setting up a series of portfolios, inter alia to: (GSSA) ESG− Inquisition that wasonheld in additions 2021 intoorder to develop inputs that need to be considered when Coordinate comments the ESG the SAMREC, SAMVAL, and SAMESG documents once these have been issued to the respective Code Committees. updating the ESG guidance. In addition, direct engagements have been held with a select group of key − Communicate the work the Committee is doing, via the App and website. − Liaise with other ESG matters, including professional stakeholders to obtain theirbodies viewsonon the issues that need to be registration. addressed in the update. ➢ Discussion on the ESG definitions proposed by the CRIRSCO ESG subcommittee. ➢ Discussions the ESG Division have within been GSSA and the work SAMCODES ESG Working Group.which the Working The outcomes of regarding the above processes distilled inofathe communications document ➢ Liaison with the JORC Committee examining ESG matters.
Group is now issuing for broad public review. One of the criticisms that has from beentheleveraged theand current The SAMCODES ESG Working Group (Andy McDonald, Chair), comprising representatives other Code against Committees
from variousGuideline industry groups, has met on awas monthly basis. Completed activities include: SAMESG is that there inadequate consultation ahead of its publication. It is important to the ➢ Recommended ESG text had been incorporated into the SAMREC and SAMVAL Codes, as well as SAMREC Table 1. Working Group thatcontents we therefore attempt to solicit thecompiled. views of as many stakeholders as possible to inform the ➢ The expanded of the SAMESG Guideline had been ➢ Arranged the ESG follow-up webinar held on 11 July 2023 (see above). scope ofWwork going forward. ➢ ork continues to resolve differences of opinion on certain aspects and get alignment on consistent use of terms and terminology.
All parties who are directly or indirectly involved with Mineral Reporting and sustainability reporting are urged
Sponsorship Committee to review the initial consultation document from the Working Group and to provide feedback via the survey that Sifiso Siwela, current Vice-Chair of the SSC, was asked to set up this committee, to consolidate the sponsorship drive for SSC activities, including theprepared website, SAMCODES App and training, and working to have a more focussed approach. has been for this purpose. Persons in the following disciplines are likely to be well-placed to The Committee comprises representatives from the other Code Committees.
provide comments: geologists; authors of Competent Persons Reports; environmental, social or governance
SAMCODES Website
subject matter experts; sustainability reports; mineral resource valuators; financial We have attended to the followingauthors aspects onof the corporate website: ➢ Regular changes to information andprivate notices on the home page, to retain relevance and and interest. managers; institutional investors; investors; investment analysts stock exchange regulators as a ➢ Given more prominence to the Guidance Notes for the SAMVAL Code. ➢ Archived older information. minimum. ➢ Updated pages for the Council for Geoscience, DMRE, and CRIRSCO. ➢ Details SAMCODES The full reportofistheavailable on App theadded. SAMCODES website at https://www.samcode.co.za/. ➢ Sent a request to members of the SSC for interesting photos to place on the website. Please createDevelopments awareness of this report and the associated survey within your network! International
CRIRSCO: The 2023 annual meeting will be held in Brazil from 16 to 19 October 2023. Mr Ken Lomberg is the official representative behalf of SAMREC the later CRIRSCO Comments should be submitted viaon the survey link byatno thanmeetings. 24 June 2022. Ms Teresa Steele-Schober will be attending the AGM in the place of Roger Dixon, representing SAMREC and the CRIRSCO ESG Subcommittee. The SAMCODES ESG Working Group looks forward to receiving your input. Please contact Ken Lomberg at ken@pivotmining.co.za for details. A. Mcdonald Chairperson, SAMCODES
Andrew McDonald The Journal of the Southern African Institute of Mining and Metallurgy
Chairperson, SAMCODES Standards Committee
VOLUME 123
OCTOBER 2023
vii ◀
HERRENKNECHT WORKSHOP hosted by the Southern African Institute of Mining and Metallurgy and South African National Council on Tunnelling INTRODUCING NEW DEVELOPMENTS IN MICRO MECHANIZED TUNNELING TO THE CIVIL AND MINING INDUSTRIES
1 February | Rosebank Free Registration
ECSA Validated CPD Activity, Credits = 0.1 points per hour attended.
This workshop is in response to the fact that the mining and civil engineering industries are under immense pressure to develop faster, safer and more efficient access to mines and underground structures. That said, the global call to do no harm has been taken very seriously, notwithstanding this focus on increased excavation advance rates and daily production to speed things up and shorten the overall project duration to make it financially more viable. Mechanization to remove people from hazards when tunnelling or during shaft sinking and to provide a continuous, consistently higher rate of advance have made great strides in the last five to ten years.
The purpose of this workshop is to gain an understanding from mining companies, mining contractors, EPCM organizations, civil contractors, consultants and investors on their perception of the current process. The idea is to look at current best practice, research and development that is underway and to make suggestions on how the process can be improved using the collective experience gained worldwide. At the very least, the workshop must identify areas for action to plan future improvements. Otherwise, many mining and civil engineering projects may remain out of reach for the foreseeable future (and asteroid mining may become a reality).
FOR FURTHER INFORMATION, CONTACT: Camielah Jardine, Head of Conferencing
E-mail: camielah@saimm.co.za Tel: +27 11 538-0237, Web: www.saimm.co.za
Contact sorption drying of chromite concentrates by C. Snyman1*, M. le Roux1, S. Engelbrecht2, and Q.P. Campbell1 * Paper written on project work carried out in partial fulfilment of BEng (Chemical Engineering) degree
Affiliation:
1N orth-West University, School of
Chemical and Minerals Engineering, Potchefstroom, South Africa. 2C larient Solutions Pty (Ltd), South Africa.
Correspondence to: M. le Roux
Email:
marco.leroux@nwu.ac.za
Dates:
Received: 31 Aug. 2022 Revised: 10 Feb. 2023 Accepted: 10 Jun. 2023 Published: October 2023
Synopsis
Due to the ultrafine particle size required for effective processing of chromite ores, dewatering of the concentrates presents a challenge. It is not uncommon for the ore to have elevated moisture contents even after dewatering, which must be reduced to required levels of between 8% and 10% by mass for further processing. Contact sorption drying has shown promise in test work on fine coal. This method was used to study the dewatering of chromite on a laboratory scale using 3 mm spherical activated alumina ceramic beads as a sorbent. Three different sorbent-to-chromite mass ratios, namely 0.5:1, 1:1, and 2:1, were tested with different process conditions, including dewatering in a stationary and a rotatingl bed. The experimental work showed that it was possible to achieve the target moistures in less than 10 minutes, irrespective of the sorbent-to-chromite ratio used. Ratios of 1:1 or higher, however, proved to be the best. The sorbent reusability at mass ratios of 1:1 and 2:1 were therefore tested. With a 1:1 mass ratio, the sorbents could be reused for three cycles, while with 2:1 ratio, the number of cycles increased to six. The sorbent-to-chromite mass ratio used had a significant influence on the required contact time and the reusability of the sorbents.
Keywords
chromite, contact sorption drying, dewatering.
How to cite:
Snyman, C., le Roux, M., Engelbrecht, S., and Campbell, Q.P. 2023 Contact sorption drying of chromite concentrates. Journal of the Southern African Institute of Mining and Metallurgy, vol. 123, no. 10. pp. 479–482 DOI ID: http://dx.doi.org/10.17159/24119717/2299/2023 ORCID: M. le Roux http://orcid.org/0000-0002-9330-426X Q.P. Campbell http://orcid.org/0000-0003-0510-6018
Background South Africa is currently the global leader in chromite production, having an estimated 70% of the world’s known resources, mainly located in the Bushveld Complex (Backeberg, 2021). Chromite is the only economic mineral from which chromium is obtained (King, 2016). More than 95% of global chromite consumption is for production of ferrochromium, more than 80% of which is used in the stainless steel industry (Holappa, 2013; KPMG, 2018; Murthy, Tripatjy, and Kumar, 2011). For this reason, stainless steel is the main driving factor for the demand and pricing of ferrochromium and chromite commodities (KPMG, 2018). Stainless steel prices are sensitive to changes in the costs of alloying elements since these elements are contained in large amounts (Holappa, 2013). If the costs of producing these alloying elements increase, it will inevitably lead to an increase in the production cost of stainless steel. Chromite deposits within the Bushveld Complex are present as stratified layers which differ in purity and complexity. Where chromite is mined as the primary product, it is preferable to produce a concentrate known as lumpy, chip, or pebble ore, which varies in size between 6 mm and 150 mm (Beukes et.al., 2017). However, when it is produced as a by-product during platinum group metal (PGM) production, the complexity of the ores necessitates the reduction of chromite particle size below 1 mm to produce a concentrate known as metallurgical grade (van Staden, 2018). This is achieved using primarily spiral and cyclone technology (Murthy, Tripatjy, and Kumar, 2011). Current well-established gravity techniques are, however, inefficient and become complex when treating ultrafine particles (< 75 μm) (Murthy, Tripatjy, and Kumar, 2011). Consequently, other techniques have been investigated and utilized, such as column flotation, jigging, and wet high-intensity magnetic separation (WHIMS) (Mokoena and Nheta, 2020; Murthy, Tripatjy, and Kumar, 2011). These techniques play a major role in the recovery of chromite from tailings. The abovementioned processes are all water-based and produce a concentrate in slurry form, which must be dewatered prior to ferrochromeium production. Several different industrial techniques are currently utilized for this purpose, namely filters (rotary drums and ceramic disc filters), different spray driers, high-temperature dryers (such as kilns), or natural drainage on stockpile heaps. This is done to reach a target moisture between 8% and 11% by mass, which is ideally suited for the downstream pelletizing process that precedes smelting (Rao, 1994; du Preez, 2018; McDougal, 2013). These processes are in general costly, environmentally unfriendly, inefficient, time-consuming, or require frequent maintenance which increases the operational costs.
The Journal of the Southern African Institute of Mining and Metallurgy
VOLUME 123
OCTOBER 2023
479
Contact sorption drying of chromite concentrates According to Tripathy et al., (2019), very little has been done to study and understand the best performing dewatering circuit for ultrafine chromite concentrates. To rectify this, a study was done in India using a tailing slurry from a chromite beneficiation plant containing particles with a d80 of 102 µm and d50 of 36 µm. The study investigated the dewatering of the tailings using different combinations of unit operations generally utilized in the dewatering of tailing streams in the mineral processing industry. A conventional dewatering circuit that has proven to be effective on fine coals, which mainly consisted of thickening followed by pressure filtration, was tested, as well as other circuits where hydrocyclones were included. A final chromite concentrate moisture of 20% by mass was achieved, which is above the prescribed target moisture for pelletizing. These circuits further proved to be uneconomical, even with the introduction of hydrocyclones (Tripahy et al., 2019). One technology that has proven to be very efficient for dewatering fine coals is contact sorption drying. Van Rensburg et al., (2020a) showed that an equivalent mass of ceramic spheres added to a rotary bed can reduce the moisture content of feed coal from around 20% by mass to inherent moisture content levels and lower within less than 10 minutes’ retention time. The discharge from the bed can then be screened to separate the ceramics and coal, producing a final coal suitable for use. The study highlighted the preference for using a rotational bed to aid in particle mixing, instead of a stationary bed (although both were effective), indicative of the need for particle contact to assist in the transfer of the moisture. Le Roux et al., (2018) and le Roux, Campbell, and Hoffman (2019) showed through a series of tests that the predominant transfer mechanism is via liquid adsorption of moisture into the pores of the sorbent. Contact between the ore and the sorbent is therefore required to facilitate the transfer of moisture from the ore particle to the sorbent. In addition to drying, it was further proven by van Rensburg et al., (2020a) that the sorption capacity of the ceramic can be regenerated in a packed bed using a high air flow at ambient conditions to drive off the moisture and ultrafine particles that diffused into the micro-pores of the spheres. Since this method has not yet been tested on fine chromite, the aim of this investigation was to study the possibility of using contact sorption as a suitable drying method for ultrafine chromite.
of the samples was done using a Malvern Mastersizer 2000. The samples had a d50 size of 85.5 μm, and a d90 size of 192 μm.
Experimental Materials
This study focused on the following: ➤ The effect of chromite-to-sorbent mass ratio to determine the minimum sorbent addition to the vessels that will yield a satisfactory drying performance. ➤ Stationary drying versus mixing of the chromite and sorbent to assess the effect of mixing on drying time. This would also give an indication if the addition of sorbents to a drying heap would be effective. ➤ The reusability and recyclability of the sorbents.
Chromite concentrate originating from the LG-6 (Lower Group 6) seam in the Rustenburg area in South Africa, was used in this study. The as-received material was air dried, split into 1 kg samples, and stored in sealed bags to be used as the feed to the rotary mixer. Prior to feeding to the mixer, water was added to the chromite to achieve an 18% moisture content. A particle size distribution (PSD) analysis
Sorbents Three-millimetre diameter Porocel Dryocel 848 activated alumina ceramic spheres were purchased and used as received. The spheres have a surface area of 320 m2/g, a pore volume of 0.5 cm3/g, a bulk density of 753 kg/m3, a crush strength of 20 kg, and an abrasion and attrition loss of 0.2 wt% and 0.4 wt%, respectively. The material consists of 93.5 wt% Al2O3, 0.35 wt% Na2O, and 0.15 wt% SiO2, and has a 6.00 wt% loss on ignition (1000°C). The maximum water capacity of the sorbent was tested by drenching a sample completely in water. The moisture content was measured after 24 hours, after 48 hours, and again after 12 days’ immersion. The moisture levels were measured at 29.8%, 28.7%, and 28.2%, respectively, resulting in an average maximum water capacity of 28.9% by weight.
Experimental set-up Two set-ups were used to determine the dewatering capabilities of the ceramic beads. For the first set-up, the sorbent and chromite in mass ratios of 1:1, 0.5:1, and 2:1 were placed in four stationary vessels for a duration of two hours. The concentrate was added first, and then the sorbent, without any mixing involved. Every half hour, one vessel was emptied and sampled to determine the moisture content of the chromite. This was done to simulate the possibility of adding sorbents to a heap of chromite and leaving it to dewater by itself with minimal mixing involved. For the second set-up, a series of plastic cylindrical vessels of length 5 cm and diameter 5.5 cm was place on rollers to act as the rotating bed. The rollers were set to rotate at three revolutions per minute to introduce a cascading motion of the bed inside the vessel. Material (again in mass ratios of 1:1, 0.5:1 and 2:1) was fed into the ten different vessels and allowed to dry before each was removed and emptied at predetermined time intervals. The content of each vessel was split into chromite and ceramic beads using a standard laboratory sieve. The chromite was then analysed to determine its moisture content. A schematic of the second set-up is shown in Figure 1.
Parameters tested
Figure 1—Experimental set-up for rotating bed tests 480
OCTOBER 2023
VOLUME 123
The Journal of the Southern African Institute of Mining and Metallurgy
Contact sorption drying of chromite concentrates Results and discussion Stationary tests The results of the stationary bed tests are shown in Figure 2. All results reported in this paper are averages of three tests, with errors varying between 0.5 and 2.8 percentage points. The figure shows an initial rapid decrease in chromite moisture content for the first half hour of the tests, followed by a more gradual decline. The dewatering curve is similar to those published by van Rensburg et al., (2020a). The fast initial dewatering is attributed to the excess amount of moisture that is free to move between the chromite particles. When it comes in contact with the ceramic sorbent, the free moisture attaches to the surface of the sorbent and diffuses into its pores. Contact between the sorbent and chromite is important to allow for moisture transfer. This was shown to be the primary transfer mechanism between ore particles and a sorbent (van Rensburg et al., 2020b). Therefore, within a stationary set-up like the one used here, moisture has to diffuse from between the particles to come into contact with the sorbent surface. Such diffusion becomes slower as the amount of free moisture decreases. From Figure 2 it is apparent that the target moisture of 8% by mass was reached after 23 minutes for the 2:1 sorbent to chromite ratio, and approximately 30 minutes for the other ratios. Although stationary dewatering is very slow compared with rotational vessels (Figure 4), it results in a significant decrease in total dewatering time compared to heap drying, which usually takes between two and three days. The addition of sorbents to these heaps would therefore reduce the retention time of the chromite on the heaps and thus deliver a final product much quicker.
Rotating vessels Feed mixtures with similar sorbent-to-chromite mass ratios were used to determine the drying performance of the ceramic beads when subjected to a mixing motion. The initial results of these tests are shown in Figure 3. It became evident at the start of operation that the chromite tended to agglomerate due to the ultrafine nature of the particles coupled with the moisture content. Although dewatering of the chromite was still possible and much quicker than for stationary drying, shown by the purple line in Figure 3 where the moisture target was reached in less than half the time needed for stationary drying, it was clear that thorough mixing was not achieved, and that after 30 minutes, the moisture content of the chromite was higher than for the corresponding stationary test. It was postulated that the initial quicker drying could be attributed to the ore still being free to move in the vessel prior to agglomerating to the sidewalls of the vessel. After the first two minutes of drying,
most of the chromite packed against the walls of the vessel, forming a closely packed bed which contributed to stronger interparticle capillary forces and hence weak dewatering. This resulted in the final moisture content of the chromite being higher than for the stationary tests due to the increase water retention properties. To alleviate the agglomeration of particles to the sidewalls, the vessels were each tapped against the table within the first two minutes of the test to loosen the material from the sides and ensure that better mixing would occur during operation. This would allow constant contact between the sorbent and the chromite, which promotes better dewatering. This is shown by comparison of the two data-sets in Figure 3. The non-agglomerated material reached the lower moisture target quicker than the agglomerated material took to even reach the upper moisture limit. It was able to reach the upper moisture target in just three minutes, having a final moisture content of less than 4% after 30 minutes. Figure 4 shows the effect of mass feed ratio of the sorbent and chromite. An increase in the ratio of sorbent to chromite will increase the sorbent area available for moisture transfer, which in theory will increase the initial rate of moisture diffusion from the chromite to the sorbent, as well as the final moisture content. Both these assumptions hold true when the grey curve in Figure 4 is compared to the other two curves. The increase in available sorbent surface area, either with proper mixing and/or an increase in sorbent amounts, provides more spaces for moisture to adsorb onto the surface of the sorbent. When the sorbent moisture content was measured at the end of the 30-minute cycle, the 0.5:1 ratio tests yielded sorbents with a moisture content just below 20% by mass, while the 1:1 and 2:1 ratios gave 12% and 7% respectively. This implies that the sorbents used in the 0.5:1 ratio need to be regenerated they can be recycled and reused, since the uptake of moisture closely relates to the ceramic’s maximum moisture-holding capacity of 28.9% by mass.
Figure 3—Agglomerated vs non-agglomerated drying in a tumbling vessel
Figure 2—Stationary dewatering results for different sorbent -o-chromite ratios The Journal of the Southern African Institute of Mining and Metallurgy
Figure 4—Drying performance in a mixing tumbler VOLUME 123
OCTOBER 2023
481
Contact sorption drying of chromite concentrates The same does not apply to the 1:1 and 2:1 mass ratio tests. The increase in sorbent area ensures that the amount of moisture adsorbed is not near to the maximum capacity, which allows for the recycling of the sorbents even without regeneration. This concept was tested by feeding the tumbler with either 1:1 or 2:1 sorbent-tochromite mass ratio material and allowing dewatering to take place for 15 minutes’ retention time. After the time elapsed, the sorbents were separated from the chromite and added in the same ratio with fresh moist chromite for a second dewatering test. This was repeated until the sorbent reached its maximum moisture-holding capacity or when the final moisture content of the chromite was above the upper target of 10%. For the 1:1 mass ratio feed, this was recorded to be three cycles, while the 2:1 mass ratio feed reached this point after six cycles, as shown in Figure 5 and Figure 6 respectively. For the 1:1 mass ratio feed, the total chromite moisture was reduced by 83%, from 18.5% to 3.1%, in the first cycle. In the second and third cycles, the total chromite moisture was reduced by 72% and 55%, respectively. After three cycles, the sorbents reached a moisture content of 25.6%, which is approximately 88% of their total moisture-holding capacity. During the last cycle, some chromite started to adhere to the sorbent surface due to the moisture content approaching the maximum capacity. More moisture was present on the sorbent surface, which caused some chromite to agglomerate on the sorbent surface. It was, however, mostly removed by thoroughly shaking the sorbent on the laboratory sieve during the separation step. The capillary forces of the sorbent pores were reaching equilibrium due to saturation, which caused the diffusion rate of the moisture from the sorbent surface into the pores to slow down. Reusing the sorbents for more than three cycles will result in a higher chromite moisture content than desired, and some of the chromite could be lost due to it adhering to the sorbents. It is, however, evident that the sorbent’s drying efficiency decreased with each cycle it was reused. This is due to the lower concentration gradient between the sorbent and chromite as a result of the higher saturation of the sorbent after each cycle. The same effect was seen for the 2:1 mass ratio feed, although it took six cycles to reach this point, as evident from Figure 6.
Conclusions This study demonstrated the efficacy of using contact sorption drying for dewatering ultrafine chromite. Commercially available ceramic spheres were used as sorbent and showed the potential to dewater the chromite from 20% to a target moisture content between 10% and 8% by mass within a residence time of less than 10 minutes. Since the transport mechanism is dominated by surface contact, the addition of ceramic spheres to the chromite directly influences the dewatering rate and total moisture adsorption. For
Figure 5—Sorbent reusability for a 1:1 mass ratio feed 482
OCTOBER 2023
VOLUME 123
Figure 6—Sorbent reusability for a 2:1 mass ratio feed
sorbent-to-chromite ratios of 1:1 or greater, it was possible to reuse the ceramic spheres without regeneration for three to six cycles. This research has demonstrated a promising method to help reduce the turnaround time for dewatering ultrafine chromite. It is, however, imperative to test the proposed method on a larger scale and conduct a financial viability study before a final verdict is possible.
References
Backeberg, N. 2021. Chromium: Ferrochrome prices back up, but for how long? https://roskill.com/news/chromium-ferrochrome-prices-back-up-but-for-howlong/ [accessed 13 August 2021]. Beukes, J.P., du Preez, S.P., van Zyl, P.G., Paktunc, D., Fabritius, T., Päätalo, M., and Cramer, M. 2017. Review of Cr (VI) environmental practices in the chromite mining and smelting industry – relevance to development of the Ring of Fire, Canada. Journal of Cleaner Production, vol. 165. pp. 874–889. Du Preez, S.P. 2018. Ferrochrome waste management – addressing current gaps. PhD thesis, North-West University, Potchefstroom. Holappa, L. 2013. Basics of ferroalloys. Handbook of Ferroalloys: Theory and Technology. Gasik, M. (ed.). Elsivier, UK. pp. 9–28. https://doi.org/10.1016/ B978-0-08-097753-9.00002-2 [accessed 20 July 2021]. King, H.M. 2016. Chromite. https://geology.com/minerals/chromite.shtml [accessed 20 July 2021]. KPMG. 2018. KPMG Commodity Insights Bulletin – Chromite. https://assets. kpmg/content/dam/kpmg/xx/pdf/2018/11/kpmg-commodity-insights-bulletinchromite.pdf [accessed 13 August 2021]. Le Roux, M., Campbell, Q.P., van Rensburg, M.J., and Peters, E.S. 2018. Moisture transport during contact sorption of coal fines. Proceedings of the Seventeenth Australian Coal Preparation Conference, Brisbane, Australia. Australian Coal Preparation Society. Le Roux, M., Campbell, Q.P., and Hoffman, J. 2019 Mechanism of drying coal fines by means of contact sorption. Proceedings of the XIX International Coal Preparation Congress, New Delhi, India. Woodhead, New Delhi. McDougall, I. 2013. Ferroalloys processing equipment. Handbook of Ferroalloys: Theory and Technology. Gasik, M. (ed.). Elsivier, UK. pp. 83–138. Murthy, Y.R., Tripathy, S.K., and Kumar, C.R. 2011. Chrome ore beneficiation challenges & opportunities – A review. Minerals Engineering, vol. 24, no. 5. pp. 375–380. Mokoena, T. and Nheta, W. 2020. Beneficiation of South African chromite tailings using magnetic separation. Proceedings of the 6th World Congress on Mechanical, Chemical, and Material Engineering (MCM'20). https://avestia.com/ MCM2020_Proceedings/files/paper/MMME/MMME_138.pdf Rao, P.V.T. 1994. Agglomeration and prereduction of ores. 4th Refresher Course on Ferro Alloys, Jamedepur, India.http://eprints.nmlindia.org/5783/1/3.01-3.15. PDF [accessed 22 October. 2021]. Tripathy, S.K., Murthy, Y.R., Farrokhpay, S., and Filippov, L.O. 2019. Design and analysis of dewatering circuits for a chromite processing plant tailing slurry. Mineral Processing and Extractive Metallurgy Review, vol. 42, no. 2. pp. 102–114. Van Rensburg, M.J., le Roux, M., Campbell, Q.P., and Peters, E.S. 2020a. Contact sorption: A method to reduce the moisture content of coal fines. International Journal of Coal Preparation and Utilization, vol. 40, no. 4. pp. 266–280. Van Rensburg, M.J., le Roux, M., Campbell, Q.P., and Peters, E.S. 2020b. Moisture transport during contact sorption drying of coal fines. International Journal of Coal Preparation and Utilization, vol. 40, no. 4–5. pp. 281–296 Van Staden, Y. 2018. The impact of raw material selection on damring formation and pre-reduction during ferrochrome production. PhD thesis, North-West University, Potchefstroom. u The Journal of the Southern African Institute of Mining and Metallurgy
Real-time gypsum quality estimation in an industrial calciner: A neural network-based approach by M. Jacobs1*, R-D. Taylor1, F.H. Conradie1, and A.F. van der Merwe1 * Paper written on project work carried out in partial fulfilment of BEng (Chemical Engineering with Specialisation in Minerals processing) degree Affiliation:
1 School of Chemical and Minerals
Engineering, North-West University, Potchefstroom, South Africa.
Correspondence to: M. Jacobs
Email:
marlisejacobs71@gmail.com
Dates:
Received: 29 Nov. 2022 Revised: 7 Sept. 2023 Accepted: 21 Sept. 2023 Published: October 2023
Synopsis
Total bound moisture (TBM) is a typical quality indicator of industrial-grade gypsum. This gypsum is comprised of three distinct phases, namely anhydrite, dihydrate, and hemihydrate, of which only the latter is of much industrial use. TBM analysis is a lengthy laboratory procedure, and an artificial neural network (ANN) TBM inference measurement is proposed as a fast and online alternative. An ANN inference model for gypsum TBM based on plant data was developed. The inputs to the network were primarily focused on the plant's calciner, and different network topologies, data divisions, and transfer functions were investigated. Furthermore, the applicability of the TBM value as a quality indicator was investigated based on a gypsum phase analysis. A strong correlation between TBM and the gypsum hemihydrate and anhydrite content was found, validating the plant target TBM of 5.8% as a quality indicator. A network topology consisting of one hidden layer with logarithmic-sigmoid (logsig) and pure linear (purelin) transfer functions showed the best performance (R > 90%).
Keywords
gypsum, artificial neural network, Levenberg-Marquardt algorithm.
How to cite:
Jacobs, M., Taylor, R-D., Conradie, F.H., and van der Merwe, A.F. 2023 Real-time gypsum quality estimation in an industrial calciner: A neural networkbased approach. Journal of the Southern African Institute of Mining and Metallurgy, vol. 123, no. 10. pp. 483–490 DOI ID: http://dx.doi.org/10.17159/24119717/2480/2023
Introduction Calcium sulphate and its phases Calcium sulphate hemihydrate (CaSO4·½H2O), commonly known as bassanite or Plaster of Paris, is widely used in various industries, ranging from construction to medicine and even the arts. Pure gypsum (calcium sulphate dihydrate) is found in nature as a compact rock. Therefore, to obtain the valuable bassanite, the gypsum is calcined according to Equation [1] (Dantas et al., 2007). [1] The dehydration reaction of gypsum to bassanite occurs at 100–120°C and is complete at 160°C (Dantas et al., 2007). The bassanite occurs in two forms: α-hemihydrate and β-hemihydrate (Singh and Middendorf, 2007). α-Hemihydrate is formed in wet process units such as autoclaves, and β-hemihydrate is formed in predominantly dry conditions, such as in calciners (Singh and Middendorf, 2007). Further dehydration of bassanite leads to calcium sulphate anhydrite according to Equation [2] (Dantas et al., 2007). [2] Similarly to bassanite, three types of anhydrite are formed during thermal dehydration, namely IIIanhydrite (AIII), II-anhydrite (AII), and I-anhydrite (AI) (Li and Zhang, 2021). Anhydrite III is the first dehydration phase that forms from bassanite, followed by an unusable form of gypsum anhydrite, namely anhydrite II, and lastly, anhydrite I upon further heating. However, the latter form is unstable below 1180°C (Cave, 2000; Rajković et al., 2009). AIII is metastable at ambient conditions and rehydrates in the presence of water or water vapour (Cave, 2000). When a large amount of soluble anhydrite (AIII) is blended into cement, it can result in a substantial expansion of the concrete and adversely impact the cement's strength (Tzouvalas, Dermatas, and Tsimas, 2004). Furthermore, insoluble gypsum anhydrite (AII) affects the hydration reaction rate and amount of unusable material in the gypsum calciner product (Christensen et al., 2008).
The Journal of the Southern African Institute of Mining and Metallurgy
VOLUME 123
OCTOBER 2023
483
Real-time gypsum quality estimation in an industrial calciner: A neural network-based approach Gypsum in industry Natural gypsum contains approximately 3% equilibrium moisture and 20 mass% crystal moisture (Dantas et al., 2007). The partially dehydrated form of gypsum, calcium sulphate hemihydrate (CaSO4·½H2O), is the desired product for use in the construction, ceramic, and medical industries (Singh and Middendorf, 2007). The hydration of calcium sulphate hemihydrate (bassanite) is an exothermic reaction, given by Equation [3] (Singh and Middendorf, 2007). During hydration (in paste form), the plaster sets, which develops the strength of the material (Singh and Middendorf, 2007). [3] This research study investigated synthetic gypsum produced from the phosphoric acid fertilizer manufacturing process (MechChem Africa, 2019; Jordan and van Vuuren, 2022). This phosphogypsum is formed according to the reaction given by Equation [4] (Rajković et al., 2009): [4] From Equation [4], it is clear that a substantial amount of phosphogypsum is produced with the phosphoric acid. In fact, according to Li and Zhang (2021), approximately 5 t of phosphogypsum is produced per ton of phosphoric acid, resulting in the generation of 280 Mt of phosphogypsum waste per year worldwide. Phosphogypsum must be calcined to deactivate impurities before being used as a building material, as these impurities impact the strength and settling times of the gypsum product (Li and Zhang, 2021; Tzouvalas, Dermatas, and Tsimas, 2004). According to Singh and Garg (cited by Tzouvalas, Dermatas, and Tsimas, 2004; Li and Zhang, 2021), impurities are deactivated by coatings of insoluble calcium pyrosulphate at elevated temperatures. Similarly, Liu et al., (cited by Li and Zhang, 2021) found that soluble phosphorus was converted to insoluble calcium pyrophosphate through calcination at 800°C for 1 hour. Additionally, fluoride and phosphorus pentoxide can be removed at 700°C, with phosphatic impurities also being removed at 800°C. According to Saadaoui et al., (2017), the radioactive components in phosphogypsum can be considered negligible. Calcination also produces gypsum hemihydrate by driving off crystal moisture (Koper et al., 2020). In the case of gypsum, the dihydrate phosphogypsum is calcined, and a mixture of calcium sulphate dihydrate, hemihydrate, and anhydrite is obtained (Koper et al., 2020). However, the amount of calcium sulphate anhydrite (especially the insoluble form) should be minimized (Christensen et al., 2008). This is because insoluble calcium sulphate anhydrite adds to the impurities in the gypsum, adversely affecting the quality. Furthermore, a high calcium sulphate anhydrite content i results in a significant increase in the heat of hydration of the mixture (Tydlitát, Medveď, and Černý, 2012). This would affect the properties of the gypsum, such as setting time, resulting in flash setting, which impacts the quality and strength of the gypsum (Tydlitát, Medveď, I. and Černý, 2012, p. 62; Tzouvalas, Dermatas, and Tsimas, 2004, p. 2113). Therefore, quality control of the gypsum product from an industrial plant is of utmost importance to ensure the desired specifications are met. The total bound moisture (TBM) is often used as a quality control parameter. A simple method to determine TBM is thermogravimetric analysis (TGA), where the sample is 484
OCTOBER 2023
VOLUME 123
dried and mass loss indicates the moisture in the original sample. Gürsel et al., (2021) investigated acoustic emission (AE) technology to determine the gypsum and anhydrite contents online. Seufert et al., (2009) used X-ray diffraction (XRD) analysis to determine the phase composition of dehydrated gypsum. However, Dantas et al., (2007, p. 692) found that XRD cannot be used to distinguish between the hemihydrate and anhydrite phases due to the superposition of the peaks. This was confirmed by Seufert et al., (2009, p. 940), who concluded that ‘high quality and high resolution’ XRD data is required for successful identification.
Artificial neural networks An artificial neural network (ANN) receives inputs that are multiplied by a weight (Krenker, Bešter, and Kos, 2011). A transfer function in the body then transforms the summation of the weighted inputs and bias to produce an output. However, this output is initially meaningless as the weight and bias coefficients are random. Therefore, the network is trained using feed-forward or recurrent (feedback) learning algorithms, which adjust the weights and biases to a point where the network functions independently to make decisions predictively (Abraham, 2005). The key to understanding neural network training lies in first reviewing the basics of neural network transfer functions, which connect the input, hidden, and output layers. The transfer functions are typically log-sigmoid, hyperbolic tangent, sine or cosine, and linear functions. The sigmoid function is widely used and performs well for classification problems involving learning about average behaviour (Zhang, Patuwo, and Hu, 1998). The logarithmic-sigmoid (log-sigmoid) function is mathematically described by Equation [5] and produces an output between 0 and 1. [5] The hyperbolic tangent sigmoid function is typically used for forecasting problems where the network learns about the deviations from the average (Zhang, Patuwo, and Hu, 1998). This function, which is given by Equation [6], produces an output between –1 and 1. [6] Alternatively, the sine or cosine and linear functions can be used as transfer functions. The sine and cosine function outputs also vary between 1 and –1, whereas that of the linear function is unbounded. However, the output values generated by these functions would have to be normalized (Zhang, Patuwo, and Hu, 1998).
Learning algorithms The backpropagation learning algorithm (also known as the steepest decent algorithm) is one of the most powerful ANN learning algorithms (Caocci et al., 2011, p. 223; Yu and Wilamowski, 2011). With enough hidden layers, the backpropagation learning algorithm can approximate any nonlinear function (Abraham, 2005, p. 904). This method converges asymptotically, resulting in slow convergence, especially near the solution (Yu and Wilamowski, 2011; Zhang, Patuwo, and Hu, 1998). Furthermore, according to Zhang, Patuwo, and Hu (1998), the backpropagation method is inefficient, lacks robustness, and is very sensitive to the learning rate (α). Small α-values result in very slow learning, whereas large α-values may result in oscillations (Zhang, Patuwo, and Hu, 1998). Another training algorithm that minimizes a quadratic error function is Newton's method (Abraham, 2005). This method The Journal of the Southern African Institute of Mining and Metallurgy
Real-time gypsum quality estimation in an industrial calciner: A neural network-based approach exhibits fast convergence but performs well only for almost linear systems, which is not the case in ANNs (Wilamowski, 2011). The Levenberg-Marquardt algorithm is a combination of the steepest descent (backpropagation) and the Gauss-Newton algorithms. When the solution is far from the desired one, the algorithm behaves like the steepest descent algorithm – slow but sure to converge (Lourakis, 2005; Yu and Wilamowski, 2011). Similarly, when the current solution is close to the desired one, the algorithm behaves like the Gauss-Newton algorithm (Lourakis, 2005; Yu and Wilamowski, 2011). The Levenberg-Marquardt algorithm is regarded as one of the most efficient training algorithms and has been found to apply to small and mediumsized problems (Brezak et al., 2012; Yu and Wilamowski, 2011). According to Alwosheel, van Cranenburgh, and Chorus (2018) the sizes of the training data-sets are typically between 10 and 100 times the number of weights in the neural network.
Topology The topology of a neural network is situational-dependent, meaning that one size does not fit all. The following list summarizes important characteristics to consider when designing an ANN: ➤ The training data should consist of all the characteristics of the problem. The more complex the problem is, the more data is required (Abraham, 2005, p. 904). ➤ Noise or randomness in the data will aid the creation of a robust and reliable ANN (May, Dandy, and Maier, 2011, p. 19). ➤ One hidden layer is sufficient for many practical problems (Heaton, 2008, p. 158) ➤ The network is usually trained for a specific number of epochs or until the error decreases below a specified threshold (Abraham, 2005, p. 904). However, care must be taken not to overtrain the network (Wilamowski, 2011, p. 16). This will result in a network that is too adapted to the training examples and will be unable to classify samples outside of the training set (Wilamowski, 2011, p. 16). Overtraining can be avoided by (Azadi and Karimi-Jashni, 2016, p. 19): ➤ Dividing the available data-set into training, validation, and test sets (Azadi and Karimi-Jashni, 2016, p. 19). ➤ The training set is used to train the model and is fed through the network during each epoch (Baheti, 2022). ➤ The validation set is used to validate the performance of the model during the training process (Baheti, 2022). The validation set is used to determine the stopping point – this will prevent overtraining the network (Azadi and KarimiJashni, 2016, p. 19). The model is validated after each epoch (Baheti, 2022). ➤ The test set is used after the training is completed to confirm the results (Baheti, 2022). ➤ The number of neurons in the hidden layer affects the performance of the network (Abraham, 2005, p. 904). A large number of hidden neurons will ensure proper training and forecasting; however, overtraining may occur (Sheela and Deepa, 2013, p. 1). Contrarily, too few hidden neurons will result in poor training and large errors (Sheela andDeepa, 2013, p. 1). Although many guidelines exist on the appropriate number of hidden neurons, it ultimately comes down to trial and error (Heaton, 2008, p. 159) ➤ The chosen initial weights are crucial for proper convergence; however, no recommendations exist in this regard (Abraham, 2005, p. 905). A trial-and-error approach should be followed to improve the results obtained. The Journal of the Southern African Institute of Mining and Metallurgy
➤
e learning rate influences the step size by which the weight Th is updated. A too-fast learning rate may result in overstepping the local minimum. This could result in oscillations and slow convergence. Similarly, a too-slow learning rate will result in a large number of oscillations, also resulting in slow convergence (Abraham, 2005, p. 905).
Problem statement Laboratory analysis of the quality parameter (total bound moisture TBM) associated with gypsum calcination into a hemihydrate form is time-consuming, thus prohibiting proper quality control of such equipment.
Motivation The residence time in f the calcination plant in question is very short (less than half an hour). However, conventional quality analyses, such as thermogravimetric and XRD, take significantly longer to complete. Furthermore, the laboratory methods already established on the plant were used for analyses. Therefore, this study investigates the possibility of implementing a soft sensor, which can be developed using ANNs. Using this soft sensor, the quality of the gypsum product can be estimated in real-time with the current plant conditions.
Theory Model software
The neural network toolbox of MatlabTM R2021a (using Windows 11 Pro on a machine running Intel® Core™ i5-8350U CPU @ 1.70 GHz, 8.0 GB RAM 64-bit operating system), specifically the nftool function, was initially used to obtain access to the ANN environment. Subsequently, the graphical interface of the nftool function was used to generate the relevant MatlabTM code. This code could then be adapted for different topologies, transfer functions, and training data divisions.
Model procedure Data division The training data comprises inputs and an output, where the inputs are made up of the collected data described in Table I and the output is the experimentally obtained TBM. The input dataset (comprising 813 data-sets) is divided into the training and simulation sets in a 70:30 ratio. This division ensures that the variation in the output data is captured for the training data and that enough data is available for the simulation of the network. The training data is then subdivided into the training, validation, and testing subsets in a 70:15:15 ratio.
Variables The topology and transfer function variables are summarized in Table II. Each ANN consists of an input, one hidden, and an output layer with a transfer function in the hidden and output layers. The number of neurons in the hidden layer was also varied. Lastly, the input data used to train the ANN was varied, as given in Table III and further explained in the subsection on data shuffling. The heading 'No SP' in this table describes neural networks where the set-point data (as shown in Table I) was not included as input to the ANN. Furthermore, 'Chronological' means that the input data to the ANN was in chronological order as the samples were taken and sensor data collected. The heading 'Random' refers to individual data shuffling (discussed in the VOLUME 123
OCTOBER 2023
485
Real-time gypsum quality estimation in an industrial calciner: A neural network-based approach Table I
Data collected as input to the ANN 1
The set point of the hopper discharge motor current
10
The set point of the hopper discharge rate
2
The set point of the exhaust fan motor current
11
The set point of the distribution fan motor current
3
Pressure setpoint of oil burner
12
Oil burner pressure
4
Hopper discharge rate
13
Hopper discharge motor current
5
Hammer dryer motor current
14
Exhaust fan motor current
6
Distribution fan motor current
15
Ball mill drive current
7
Hammer dryer inlet temperature
16
Hammer dryer outlet temperature
8
U-bend temperature
17
Exhaust fan temperature
9
Calciner temperature (4 sensor values)
18
Combustion furnace temperature
subsection Data shuffling). Similarly, 'Blocked random' refers to blocked data shuffling. Lastly, 'First', 'Last', and 'Daily' refer to the 70% training data being either the first 70% of the set, the last 70% of the set, or the last 70% of each sampling day, respectively. For example, attempt 1 utilizes all of the 18 inputs as described in Table I. The input data consists of the first 70% of the chronological data-set. Furthermore, an ANN was trained using 5, 10, 15, 20, 50, 80, and 100 hidden neurons. For each of these configurations, the transfer function pair, as described in Table II, was also considered. Hence, one of the networks trained for attempt 1 will comprise 18 inputs, using 5 hidden neurons and a tansig transfer function in the hidden layer with a purelin transfer function in the output layer.
Table II
Model variables (topology and transfer functions) Variables: Hidden neurons and transfer functions Hidden neurons
5, 10, 15, 20, 50, 80, 100
Transfer function in layer 1 (hidden layer)
Transfer function in layer 2 (output layer)
Tansig
Purelin
Logsig
Purelin
Logsig
Logsig
Table III
Model variables (input data) Variables: Data input Chronological Attempt
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 486
OCTOBER 2023
No SP
First
Last
Random
Daily
First
Last
Blocked random Daily
First
Last
Daily
X All data random All data random All data random X X X X X
Discrete X X X X X
X
X X
X
X X
X
X X
X
X X
X
X X
X VOLUME 123
The Journal of the Southern African Institute of Mining and Metallurgy
Real-time gypsum quality estimation in an industrial calciner: A neural network-based approach Data shuffling In the first attempt, the data to the ANN is imported chronologically using the first 70% of the data (as shown in Table III). In this division, the minimum and maximum data-points are included in the training set, which is essential due to the poor extrapolation ability of neural networks (Hasson, Nastase, and Goldstein, 2020). Furthermore, since the model does not receive dynamic data (such as a rate of change resulting from a dependency between data-points in a time sequence), data shuffling was investigated. It has been observed that data shuffling reduces overfitting, improves the testing accuracy and convergence rate, and results in better generalization of the neural network (Kaushik, Choudhury, Kumar, Dasgupta, Natarajan, Pickett, and Dutt, 2020, p. 2; Ke, Cheng, and Yang, 2018, p. 1; Nguyen, Trahay, Domke, Drozd, Vatai, Liao, and Gerofi, 2022, p. 1085). There are various methods to shuffle the data, including individual shuffling using shuffling algorithms, batch shuffling, etc. (Baheti, 2022). Shuffling can occur at different points during the training process. MatlabTM offers two main options: shuffling once before training starts (default) or after every epoch (MathWorks, 2022). All data was shuffled during attempt 2, including the simulation data. However, the question arose whether shuffling impacted the results; hence attempts 3 and 4 were also performed. Another problem with these three attempts was that the actual plant data would not enter the ANN shuffled but rather chronologically as a data stream, and random shuffling is, therefore, not an accurate representation. Thus, attempts 2–4 are considered invalid and not repeated for different transfer functions. The training data was grouped into three groups as follows (Table III): 1 Using the first 70% of the data 2 Using the last 70% of the data 3 Using the last 70% of each sample day's data (blocked shuffling). These three groups of data were shuffled as follows: 1 Chronological data (no shuffling) 2 Random shuffling of only the training data (individual shuffling) 3 Grouping the data into sample blocks (blocked shuffling). The default training parameters in MatlabTM were used as given in Table IV. However, the weights and biases were initialized at 0.5 wso that all models could be compared from the same starting point.
Experimental methodology Materials and sample preparation Phosphogypsum samples were collected at the product bagging section of the plant after the calcining and milling processes. To determine the optimum sampling rate, the residence time of the plant is required. This residence time was determined at start-up: from when the feed conveyor belt started to when the first material reached the bagging section. Using the Nyquist sampling rate criterion, it was determined that the sampling rate should be half of the residence time (Landau, 1967). The crucibles were engraved for identification purposes, cleaned, and dried in an oven. The cleaned crucibles were then left to cool in a desiccator, after which the mass of each was determined and recorded (Mc). A Mettler Toledo AB204-S/FACT analytical The Journal of the Southern African Institute of Mining and Metallurgy
balance was used for this purpose. Lastly, 95% ± 0.15% ethanol was prepared from 99.99% ethanol.
Sampling Hemihydrate samples were taken at the bagging section of the plant. The residence time of the plant was 20 minutes; so according to the Nyquist theorem, sampling should occur in 10 minute intervals (Landau, 1967). As the samples were taken and sealed in plastic containers, the time, date, and product bag number were recorded on the container. A portion of each sample was also taken and stored in a sealable bag for repeatability analysis. It is important to note that sampling could only occur once the plant was operating at a steady state, i.e. when normal operating conditions had been maintained for approximately 2 hours.
Sample analysis Total bound moisture (TBM) The TBM analyses were done by weighing and recording approximately 10 g of material in a crucible (Msc). The crucible name, sample, and mass of the sample were recorded. The samples were placed in a furnace (Carbolite Gero AAF 1100 and VMF 1000) at 450°C for 2 hours. The samples were then removed and allowed to cool in a desiccator. Finally, the mass of the dried sample (with the crucible) was recorded (MAH). The TBM was then calculated as follows: [7]
Gypsum anhydrite To determine the amount of gypsum anhydrite in the sample, 10 g of each sample was again weighed off, and the mass was recorded as Msc. Then, approximately 10 ml of 95% ethanol was added to the sample, ensuring the sample was submerged in the liquid. Care was taken to ensure no spillages occurred as the liquid was swirled in the crucible. The hydrated sample was then placed in an oven at 45°C (Scientific Series 2000) for 24 hours. At this temperature, dehydration of the gypsum does not occur, and only surface moisture is removed (Li and Zhang, 2021). Furthermore, the anhydrite samples were dried for the same time as the hemihydrate samples for consistency. The crucible was then removed from the oven and allowed to cool in a desiccator. The mass of the sample and crucible was then recorded as Msd.
Table IV
Default training parameters Training parameter
Value
Maximum number of epochs
1 000
Performance goal (MSE)
0
Minimum performance gradient
1 x 10-7
Maximum validation failures
6
Initial momentum parameter (μ)
0.001
Decrease factor for μ
0.1
Increase factor for μ
10
Maximum value for μ
1 x 1010
VOLUME 123
OCTOBER 2023
487
Real-time gypsum quality estimation in an industrial calciner: A neural network-based approach If the recorded mass of the sample (Msc) was greater than the hydrated mass (Msd – Mc), then weight loss occurred, which can be attributed to surface moisture in the sample; thus, no gypsum anhydrite was present. Alternatively, if a mass gain occurred (Msc > Msd – Mc), the water in the ethanol reacted with the soluble gypsum anhydrite and gypsum hemihydrate formed. The percentage of gypsum anhydrite (%AH) present in the sample can be calculated as follows: [8] The fraction of water gained by converting the anhydrite gypsum to hemihydrate gypsum is calculated as follows: [9]
Gypsum hemihydrate The same procedure as with the anhydrite analysis was followed to determine the amount of gypsum hemihydrate in the sample. However, the ethanol was replaced with ultrapure (Milli-Q) water. The fraction of water gained by converting the gypsum hemihydrate to gypsum dihydrate was determined using Equation [10] (note that any gypsum anhydrite will also react). [10] The percentage of gypsum hemihydrate (%HH) can then be calculated using Equation [11]:
Results and discussion TBM as a quality indicator The correlation between gypsum dihydrate and TBM (Figure 1) was poor. The correlation between gypsum hemihydrate and TBM was significantly stronger (Figure 2). This is expected as a TBM analysis is commonly used to quantify a gypsum product's plaster (hemihydrate) content (ASTM Standard C 471M-01, 2017). The correlation between gypsum anhydrite and TBM is shown in Figure 3. An even stronger correlation can be observed. However, the slope of the regression plot is negative. These correlations seem counterintuitive since one would expect the two phases with crystal water (gypsum dihydrate and gypsum hemihydrate) to correlate strongly with a bound moisture indicator. Instead, the strongest correlation with TBM is found with the gypsum anhydrite phase. These results are in accordance with an analysis done by the plant engineer. Unfortunately no explanation could be found for this observation, but additional studies are under way to understand this concept. The industry target TBM of 5.8% was investigated. A gypsum hemihydrate content of between 60% and 80% was achieved at this TBM. Furthermore, the gypsum anhydrite content was 10–30%, and the gypsum dihydrate content was 4–14%. At this TBM, a large amount of gypsum hemihydrate is present, with a small amount of gypsum anhydrite and limited amounts of gypsum dihydrate. The low gypsum dihydrate content is desired since this indicates a large conversion from feed gypsum to useful gypsum hemihydrate. Furthermore, the quantity of gypsum hemihydrate should be maximized, with a small amount of gypsum anhydrite to improve
[11]
Inert analysis After the samples for the hemihydrate analysis had been weighed, the completely hydrated samples were further heated to 450°C for 4 hours. At this temperature, all the free moisture and crystal water is removed. The percentage of pure gypsum (%Purity) can then be calculated as the ratio between the actual mass of water gained vs. the theoretical maximum water gain. [12]
Figure 1—Correlation between gypsum dihydrate and TBM
where [13] The percentage of impurities (%Inerts) in the sample can then be calculated as: [14]
Gypsum dihydrate The phase analysis was completed by a mass balance to determine the percentage of gypsum dihydrate (%DH) in the sample: [15] 488
OCTOBER 2023
VOLUME 123
Figure 2—Correlation between gypsum hemihydrate and TBM The Journal of the Southern African Institute of Mining and Metallurgy
Real-time gypsum quality estimation in an industrial calciner: A neural network-based approach al., 2012; Hagan et al., 2014; Kriesel, 2007). The performance of each transfer function depends on the application thereof, as confirmed by Dorofki et al., (2012). This result was also seen in the application to the calciner, as the logsig-purelin topology seems to outperform the tansig-purelin topology. However, Kriesel (2007) states that it is essential that the output transfer function is linear, so as not to limit the output interval. This is also seen in the results obtained – the logsig-purelin topology outperforms the logsig-logsig topology. A parity plot of the logsig-purelin configuration is given in Figure 4.
Conclusions and recommendations The industry operational target of 5.8% total bound moisture (TBM) is deemed sufficient. At this point, a gypsum dihydrate conversion of about 90% is achieved. Furthermore, gypsum hemihydrate makes up most of the product (about 70%), with a sufficiently small quantity of gypsum anhydrite (about 20%). Therefore, based on these results, the target TBM is sufficient to control the product quality. Furthermore, no offset between the calculated and experimental TBM was observed on the last day of sampling, which could be due to the stricter plant control. Lastly, no correlation was found between the purity of the gypsum product and the TBM. This indicates that impurities with crystal moisture were not present in significant amounts. Based on the simulation set of the first sampling campaign, the logsig-purelin configuration with the highest R-value provided the best performance. This model is sufficient to use as a control guideline since the coefficient of correlation is greater than 90%. The strength of the model is indicated by R2 (86%) Even though the TBM could be correlated to the gypsum phases, a more in-depth study is required. It is suggested that this study would look more at how the different phases impact product quality. For example, to what extent does gypsum anhydrite impact the setting time of the product? From this information, an optimum
Figure 3—Correlation between gypsum anhydrite and TBM
setting times (Tzouvalas, Dermatas, and Tsimas, 2004). However, the gypsum anhydrite content should also be limited, otherwise flash settling might occur (Tydlitát, Medveď, and Černý, 2012).
Artificial neural network For each attempt and each run, the training, validation, testing, and simulation performance was documented using the mean squared error (MSE) and correlation coefficient (R). The performance is based on the lowest MSE and highest R-value for each attempt. The results indicated that the best run of the logsig-logsig configuration is still worse than the other two configurations, where the logsigpurelin configuration shows the best results. The best-performing models for each of the transfer function pairs with respect to the two performance measures are shown in Table V. Of the six best networks, only two (attempt 9 of the logsigpurelin and tansig-purelin configurations) performed well when the set-point data was removed. Therefore, it can be concluded that the set-point data is valuable training information for the ANNs. A possible explanation for this is that the set-point offers a more constant value than the slightly deviated measurement. Additionally, a set-point change is immediate, whereas plant conditions have a slower response, and hence the model would ‘see’ and ‘expect’ a change. Furthermore, the chronological data also performed well, as seen in attempt 1 of the tansig-purelin and logsig-logsig configurations. Lastly, attempts 17 of the logsig-purelin and 13 of the logsig-logsig configurations also performed well. These two attempts consisted of data from the first two days of sampling with quite variable product quality and the 'stable' last day of sampling. Similarly, attempt 9 also consisted of data from all the days. This shows the importance of training the model using data that encompasses the characteristics of the entire data-set. The hyperbolic tangent and log-sigmoid transfer functions are widely used for multilayer backpropagation networks (Dorofki et
Figure 4 – Modelled vs actual TBM for logsig-purelin configuration (largest R)
Table V
DOF
Neurons
MSE x 105
R
DOF
log
R
sig-
Neurons
sig
Log
Attempt
1
15
1.23
0.38
301
17
15
1.33
0.6
301
1
5
0.90
0.58
101
9
5
17.70
0.91
81
9
5
17.70
0.93
81
13
10
2.83
0.89
201
The Journal of the Southern African Institute of Mining and Metallurgy
Attempt
DOF
elin
R
pur
MSE x 105
sig,
Logsig, logsig
Neurons
pur
Logsig, purelin
Attempt
sig,
Tan sig, purelin
MSE x 105
Log
elin
Tan
Best-performing artificial neural networks
VOLUME 123
OCTOBER 2023
489
Real-time gypsum quality estimation in an industrial calciner: A neural network-based approach compositional range for each phase should be determined. This can then be used to calculate the desired TBM. A more complex ANN topology and training structure should be investigated. Concerning the topology, different transfer function configurations should be investigated. The logsig-logsig configuration provided better results for the additional data, whereas the logsig-purelin configuration provided better results for the original data. It would be worthwhile to investigate the performance of different configurations, such as a tansig-logsig configuration. Furthermore, regarding the training structure, it is suggested that the performance of an optimized training algorithm, such as shuffling between each epoch, be investigated. In addition, the arbitrary decision of a 70:30 division between the training and simulation data could also be investigated. Lastly, a larger data-set would be beneficial to ensure sufficient data is available to obtain satisfactory training and simulation results.
Acknowledgements
We want to thank Mr R. van der Merwe and OMV Potchefstroom for answering all my questions regarding the plant and the gypsum process and allowing me to proceed with this work. Thank you also to Mrs R. Bekker for your unending help and patience in the laboratory.
Credit author statement
MJ: Conceptualisation, Methodology, Software, Investigation, Validation, Formal analysis, Writing, Visualization; FvdM: Supervision, Project administration, Funding acquisition, Reviewing; FHC: Supervision, Project administration, Funding acquisition, Reviewing; R-DT: Visualization, Reviewing.
References
Abraham, A. 2005. Artificial neural networks. Handbook of Measuring System Design. Sydenham, P.H. and Thorn, R. (eds). Wiley, London. pp. 901–908. Alwosheel, A., van Cranenburgh, S., and Chorus, C.G. 2018. Is your dataset big enough? Sample size requirements when using artificial neural networks for discrete choice analysis. Journal of Choice Modelling, vol. 28, no. 1. pp. 167–182. https://doi.org/10.1016/j.jocm.2018.07.002 ASTM Standard C 471M – 01. 2017. Standard test methods for chemical analysis of gypsum and gypsum products. ASTM International, West Conshohocken, PA. Azadi, S. and Karimi-Jashni, A. 2016. Verifying the performance of artificial neural network and multiple linear regression in predicting the mean seasonal municipal solid waste generation rate: A case study of Fars province, Iran. Waste Management, vol. 48. pp. 14–23. Baheti, P. 2022, Train test validation split: how to & best practices [Blog post]. https://www.v7labs.com/blog/train-validation-test-set [accessed 5 October 2022]. Brezak, D., Bacek, T., Majetic, D., Kasac, J., and Novakovic, B. 2012. A comparison of feed-forward and recurrent neural networks in time series forecasting. Proceedings of the 2012 IEEE Conference on Computational Intelligence for Financial Engineering & Economics (CIFEr), Croatia. https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=6327793 Cave, S. 2000. Gypsum calcination in a fluidised bed reactor. PhD thesis, Loughborough University, UK. Christensen, A.N., Olesen, M., Cerenius, Y., and Jensen, T.R. 2008. Formation and transformation of five different phases in the CaSO4−H2O system: crystal structure of the subhydrate β-CaSO4·0.5 H2O and soluble anhydrite CaSO4. Chemistry of Materials, vol. 20, no. 6. pp. 2124–2132. https://doi.org/10.1021/ cm7027542 Dantas, H., Mendes, R., Pinho, R., Soledade, L., Paskocimas, C., Lira, B., Schwartz, M., Souza, A., and Santos, I. 2007. Characterisation of gypsum using TMDSC. Journal of Thermal Analysis and Calorimetry, vol. 87, no. 3. pp. 691–695. https://doi.org/10.1007/s10973-006-7733-9 Dorofki, M., Elshafie, A.H., Jaafar, O., Karim, O.A., and Mastura, S. 2012. Comparison of artificial neural network transfer functions abilities to simulate extreme runoff data. Proceedings of the 2012 International Conference on Environment, Energy and Biotechnology. International Association of Computer Science and Information Technology Press, Singapore. pp. 39–44. Gürsel, D., Möllemann, G., Clausen, E., Nienhaus, K., and Wotruba, H. 2021. Online measurements of material flow compositions using acoustic emission: Case of gypsum and anhydrite. Minerals Engineering, vol. 172, no. 1. pp. 1–10. https://doi.org/10.1016/j.mineng.2021.107131 Hagan, M.T., Demuth, H.B., Beale, M.H., and Jesús, O.D. 2014. Neural network design. https://hagan.okstate.edu/NNDesign.pdf [accessed 12 October 2022] 490
OCTOBER 2023
VOLUME 123
Hasson, U., Nastase, S.A., and Goldstein, A. 2020. Direct fit to nature: an evolutionary perspective on biological and artificial neural networks. Neuron, vol 105, no.3. pp. 416–434. Heaton, J. 2008. Introduction to Neural Networks with Java. 2nd edn. Heaton Research, Inc., Chesterfiel, MO. Jordan, L.A. and van Vuuren, D. 2022. Heat-constrained modelling of calcium sulphate reduction. Journal of the Southern African Institute of Mining and Metallurgy, vol. 122, no. 10. pp. 607–616. http://dx.doi.org/10.17159/24119717/1530/2022 Kaushik, S., Choudhury, A., Kumar, P.K., Dasgupta, N., Natarajan, S., Pickett, L.A., and Dutt, V. 2020. AI in healthcare: time-series forecasting using statistical, neural, and ensemble architectures. Frontiers in big data. pp. 1–17. doi.org/10.3389/fdata.2020.00004 Kavzoglu, T. and Mather, P.M. 2003. The use of backpropagating artificial neural networks in land cover classification. International Journal of Remote Sensing, vol. 24, no. 23. pp. 4907–4938. https://doi.org/10.1080/0143116031000114851 Koper, A., Prałat, K., Ciemnicka, J., and Buczkowska, K. 2020. Influence of the calcination temperature of synthetic gypsum on the particle size distribution and setting time of modified building materials. Energies, vol. 13, no. 21. 5759. https://doi.org/10.3390/en13215759 Krenker, A., Bešter, J., and Kos, A. 2011. Introduction to the artificial neural networks. Artificial Neural Networks: Methodological Advances and Biomedical Applications. Suzuki, K. (ed.). InTech, Croatia. pp. 1–18. Kriesel, D. 2007. A brief introduction to neural networks. http://www.dkriesel.com/ en/science/neural_networks [accessed 12 October 2022]. Landau, H. 1967. Sampling, data transmission, and the Nyquist rate. Proceedings of the IEEE, vol. 55, no. 10. pp. 1701–1706. https://doi.org/ 10.1109/ PROC.1967.5962 Li, X. and Zhang, Q. 2021. Dehydration behaviour and impurity change of phosphogypsum during calcination. Construction and Building Materials, vol. 311, no. 13. pp. 1–10. https://doi.org/10.1016/j.conbuildmat.2021.125328 Lourakis, M.I. 2005. A brief description of the Levenberg-Marquardt algorithm implemented by levmar. Foundation of Research and Technology, vol. 4, no. 1. pp. 1–6. May, R., Dandy, G., and Maier, H. 2011. Review of input variable selection methods for artificial neural networks. Artificial Neural Networks Methodological Advances and Biomedical Applications. InTech, Rijeka, Croatia. pp. 19–44. Mechchem Africa. 2019. Gypsum reprocessing for a cleaner environment. https:// www.crown.co.za/latest-news/mechchem-africa-latest-news/10236-gypsumreprocessing-for-a-cleaner-environment [accessed 28 July 2022]. Nguyen, T.T., Trahay, F., Domke, J., Drozd, A., Vatai, E., Liao, J., and Gerofi, B. 2022. Why globally re-shuffle? Revisiting data shuffling in large scale deep learning. 2022 IEEE International Parallel and Distributed Processing Symposium (IPDPS). IEEE. pp. 1085–1096. Rajković, M., Tošković, D., Stanojević, D.D., and Lačnjevac, Č. 2009. A new procedure for obtaining calcium sulphate α-hemihydrate on the basis of waste phosphogypsum. Journal of Engineering & Processing Management, vol. 1, no. 1. pp. 99–113. Ranganathan, A. 2004. The levenberg-marquardt algorithm. https://sites.cs.ucsb. edu/~yfwang/courses/cs290i_mvg/pdf/LMA.pdf [accessed 28 Jul 2022]. Rudd-Orthner, R.N. and Mihaylova, L. 2021. Deep ConvNet: Non-random weight initialization for repeatable determinism, examined with FSGM. Sensors, vol. 21, no. 14. pp. 4772. https://doi.org/10.3390/s21144772 Saadaoui, E., Ghazel, N., Ben Romdhane, C., and Massoudi, N. 2017. Phosphogypsum: potential uses and problems – a review. International Journal of Environmental Studies, vol. 74, no. 4. pp. 558–567. https://doi.org/10.1080/00 207233.2017.1330582 Seufert, S., Hesse, F., Goetz-Neunhoeffer, G., and Neubauer, J. 2009. Quantitative determination of anhydrite III from dehydrated gypsum by XRD. Cement and Concrete Research, vol 39, no. 1. pp. 936–941. https:// doi:10.1016/j. cemconres.2009.06.018 Sheela, K.G. and Deepa, S.N. 2013. Review on methods to fix number of hidden neurons in neural networks. Mathematical Problems in Engineering. Article 425740. https://www.hindawi.com/journals/mpe/2013/425740/ Singh, N. and Middendorf, B. 2007. Calcium sulphate hemihydrate hydration leading to gypsum crystallisation. Progress in Crystal growth and Characterisation of Materials, vol. 53, no. 1. pp. 57–77. https://doi.org/10.1016/j. pcrysgrow.2007.01.002 Tydlitát, V., Medveď, I., and Černý, R. 2012. Determination of a partial phase composition in calcined gypsum by calorimetric analysis of hydration kinetics. Journal of Thermal Analysis and Calorimetry, vol. 109, no. 1. pp. 57–62. https://doi.org/10.1007/s10973-011-1334-y Tzouvalas, G., Dermatas, N., and Tsimas, S. 2004. Alternative calcium sulfate-bearing materials as cement retarders: Part I. Anhydrite. Cement and Concrete Research, vol. 34, no. 11. pp. 2113–2118. https://doi.org/10.1016/j. cemconres.2004.03.020 Wilamowski, B.M. 2011. Neural networks learning. Industrial Electronics Handbook. Wilamowksi, B.M. and Irwin, J.D. (eds). Taylor & Francis, Florida. pp 11-1—11-18. Yu, H. and Wilamowski, B.M. 2011. Levenberg-Marquardt training. Industrial Electronics Handbook: Intelligent Systems. Wilamowksi, B.M. and Irwin, J.D. (eds). 2nd edn. Taylor & Francis, Florida. pp 12–15. Zhang, G., Patuwo, B.E., and Hu, M.Y. 1998. Forecasting with artificial neural networks: The state of the art. International Journal of Forecasting, vol. 14, no. 1. pp. 35–62. https://doi.org/10.1016/S0169-2070(97)00044-7 u The Journal of the Southern African Institute of Mining and Metallurgy
Structural frame analysis of an electrically powered robotic subsea dredging crawler under static loading conditions by M.O. Ojumu1* and A.K. Raji1
* Paper written on project work carried out in partial fulfilment of the Degree Master of Engineering (Mechanical Engineering). Affiliation:
1 Faculty of Engineering and the Built
Environment at the Cape Peninsula University of Technology.
Correspondence to: M.O. Ojumu
Email:
mikeoluwaseunojumu@gmail.com
Dates:
Received: 8 Sept. 2022 Revised: 20 Sept. 2023 Accepted: 25 Sept. 2023 Published: October 2023
Synopsis
Robotic subsea dredging crawlers are dynamically and remotely controlled vehicles that are used for deep sea mining and recovery operations. These exploration machines are released from a mother ship and move around on the ocean floor using tracks. Current ocean crawlers such as the MK3 ROST are hydraulically powered. In this paper we develop a scaled-down model for simulating and performing static loading analysis of an electrically powered robotic subsea dredging crawler (EPRSDC). The modeling, simulation, and analysis were carried out using modelling software from Solidworks. The structural frame was assembled using the Tetrix max robotics kit. The kit’s structural components were produced from 1050 aircraft-grade aluminum. The results were used in optimizing and for considering other materials, and the to identify specific areas to be reinforced in future crawler designs.
Keywords
static loading, subsea crawler, NI-myRIO, hydraulic crawler, ocean crawler, electrically powered, dredging crawler.
How to cite:
Ojumu, M.O. and Raji, A.K. 2023 Structural frame analysis of an electrically powered robotic subsea dredging crawler under static loading conditions. Journal of the Southern African Institute of Mining and Metallurgy, vol. 123, no. 10. pp. 491–500 DOI ID: http://dx.doi.org/10.17159/24119717/2308/2023 ORCID: M.O. Ojumu http://orcid.org/0000-0003-4318-1115
Introduction In recent years, oceanic dredging crawlers have emerged as a rapidly expanding category of offshore excavation machinery on a global scale. The upswing in demand for marine-based natural resources, including gold, diamonds, iron ore, silver, and aluminium, has generated heightened interest in deepsea exploration for commercial mining ventures. A subsea oceanic dredging crawler serves as a pivotal apparatus for exploration and sample retrieval. As depicted in Figure 1, a crawler is designed to descend to the sea bed, where it propels itself by traction utilizing either wheels or tracks. The crawler is typically linked to a surface vessel by cables, through which it receives power and control instructions, and which provide a video feed to surface. In the sea floor mining, extraction encompasses the excavation and recovery of minerals, along with the associated waste material (undersea gravels). These undersea gravels constitute a longstanding amalgamation, primarily consisting of 96.5% water, 2.5% salts, and a small fraction of various other constituents, including dissolved inorganic and organic compounds, particulates, and a limited quantity of
Figure 1—MK3 ROST ocean dredging crawler in operation (IMDH Group, 2016)
The Journal of the Southern African Institute of Mining and Metallurgy
VOLUME 123
OCTOBER 2023
491
Structural frame analysis of an electrically powered robotic subsea dredging crawler atmospheric gases. According to IMDH Group (2016), there has been a noteworthy surge in the demand for metals, thus catalysing heightened interest in deep-sea exploration and commercial mining activities. Dredging crawlers are not only finding valuable applications within mining contexts but are also being deployed for tasks such as subsea tank cleaning operations, subsea photography, as well as providing support for pipeline installations, inspections, and facilitating bio-investigations in oceanic research. Extensive research has been carried out on underwater robotic systems and associated sensing technologies. Chutia, Kakoty, and Deka (2017) reviewed the history of underwater robotics, advances in underwater robot navigation and sensing techniques, and their applications in sea floor exploration. As illustrated in Figure 1, oceanic dredgers are driven by a hydraulic system to impart operational torque for both the suction arm and track movement. The vehicle is equipped with a primary dredging pump, linked to a riser responsible for material conveyance to the ship's processing unit. Pioneering work by researchers such as Lloyd-Smith and Immig BAppSc, (2018) introduced a conceptualization of a dredging robot featuring material transport mechanisms. This robot employs a compact vacuum head for sea bed excavation, offering a depth range of 1–50 m. Autonomous control is achieved through waterproof cables, enabling forward/backward and clockwise/counterclockwise movements. The authors emphasize the necessity of fortifying the diving capacity associated with the suction sludge recovery system, which directs sludge through a high-pressure pump to the processing unit via a robust pipeline. The researchers underscore the critical importance of ensuring the structural integrity of subsea equipment, as uneven terrain may induce deformation over time, potentially resulting in costly losses due to structural failures during recovery. Following our investigation, the electrically powered robotic subsea dredging crawler (EPRSDC) was engineered with a chassis frame serving as the foundational structural element. This frame bears the total operational payload of approximately 8.173 lg. Affixed to the frame (as depicted in Figure 3) are integral components, including a pressure pump, dredging pump, cameras, two 12 V Torquernado motors, three servo motors, motor controllers, servo motor controllers, tank track chains, and the venturi dredging pump unit. The chassis is designed for robustness for navigating diverse environmental conditions and to withstand all applied forces during experimental trials. Materials exhibiting highperformance characteristics suitable for marine applications are under consideration for the construction. The frame drive system incorporates four front wing arms to achieve an angular wheel configuration. Additionally, elements such as tank tread chains, tank tread idler wheels, tank tread sprockets, bronze bushings, motor hubs, and channels of various dimensions were utilized in the assembly of the drive system. These components collectively constitute the primary structural support for the undercarriage frame, facilitating the rolling motion of the track drive. Most other components are located on the boom arm. This paper aims at providing an in-depth analysis of the potential implications arising from the application of the EPRSDC’s structural static compressive force. The swift progression of electrical autonomous components, encompassing waterproofed electric motors, microcontrollers, and the utilization of LabVIEW for microcontroller programming, has significantly streamlined the management of electrical equipment. This integrated system has played a pivotal role in mitigating both oceanic and atmospheric pollution. In the foreseeable future, as an increasing number of 492
OCTOBER 2023
VOLUME 123
Figure 2—3D SolidWorks design model of the EPRSDC
structural platforms are orchestrated through algorithmic control for robust oceanic operations, a notable upswing in commercial production can be anticipated, contributing to the reduction of atmospheric pollutants, both on terrestrial and maritime fronts. The international market is currently witnessing heightened demand for offshore autonomous controllers and electrical robots, aligning with evolving customer preferences. In this study, we employ finite element analysis, specifically leveraging SOLIDWORKS 2019, to evaluate stress distribution across the structural framework design. This step is paramount in making cost-effective enhancements prior to the manufacturing phase. We begin by introducing the three-dimensional design of the EPRSDC's structural frame. As shown in Figure 2, the EPRSDC was modelled utilizing the SolidWorks student version 2019. This software proved instrumental in fine-tuning individual components prior to the final assembly of the design. Notably, this design exhibits scalability, rendering it suitable for large-scale industrial production of the EPRSDC. In the case of this downscaled prototype, SolidWorks mates were judiciously employed to accurately position all electronic elements, yielding a realistic 3D representation of the model. Moreover, the software facilitated the seamless integration of the three undercarriage systems of the crawler. It is imperative to utilize computer-aided design (CAD) software in the formulation of a robot's structural and hardware elements. This approach affords complete autonomy and adaptability throughout the developmental trajectory, spanning from conceptualization to the manufacturing phase. As shown in Figure 3, the crawler concept design was fabricated in strict adherence to the modelled specifications. The utilization of a three-dimensional design framework facilitated dynamic manipulation of individual components, providing enhanced insight into the visual clarity and structural coherence of the final manufactured model. The model was intentionally configured with a 45-degree angular inclination to seamlessly accommodate the integration of multiple sensors and hardware attachments, as exemplified in the illustration.
Literature review Yang et al., (2012) performed a detailed examination of hydraulic excavator design, elucidating their diverse applications encompassing mining, construction, and forestry contexts. The research underscored the pivotal role played by the structural design of the excavator undercarriage; a critical factor influenced by the performance of the backhoe front attachment. The The Journal of the Southern African Institute of Mining and Metallurgy
Structural frame analysis of an electrically powered robotic subsea dredging crawler
Figure 3—Completed hardware design concept of the EPRSDC
authors' investigation focused on the end effector attachment of the excavator, encompassing analyses in structural kinematics, dynamic behaviour, trajectory planning and control, fatigue life, and structural optimization design. The paper culminated in fundamental enhancements targeting increased reliability and efficiency, resulting in reduced production costs for excavators. Additionally, the utilization of virtual prototype technology was highlighted for its efficacy in reducing development expenses, along with the strategic application of technology and material improvements through finite element analysis (FEA) tools, ultimately leading to energy savings. The integration of microelectronic technology, microcontrollers, servo motors, and DC motors was emphasized as instrumental in augmenting excavator intelligence and mechatronics, thereby optimizing the reliability and efficiency of structural components. Fatigue failures were considered, particularly focusing on stress points within intricately designed components subjected to alternating loads, complex geometrical configurations, and the ensuing multi-axial stress patterns. The authors presented a case study centred on fatigue analysis of industrial components, particularly the structural chassis, emphasizing modes of deformation and cyclic pre-failure behaviour. Notably, the study highlighted the challenges associated with analysing models subjected to nonproportional loading situations and suggested the application of notch simulations, such as Neuber, for models featuring higher frequency elements under diverse loading conditions as stated by Medagedara and Chandra, (2012). Thavai et al., (2015) harnessed finite element analysis (FEA) to conduct a static loading analysis of a go-kart car's frame structure. Employing SolidWorks software for design and simulation, the study aimed to identify areas of maximum deflection within the structure. The results were subsequently utilized for comparative analysis against theoretical approximations, revealing a notable convergence in deflection magnitudes across various aspects. Dai, Yin, and Ma (2019) conducted an in-depth examination of a structural model using computational fluid dynamics to ascertain the static loads exerted by fluid resistance and resistance torque on a deep-sea mining vehicle. The study encompassed an assessment of the crawler's response to five stress levels at varying angles, factoring in the influence of oceanic resistance torque. The authors further employed the discrete element method to enhance riser efficiency. The Journal of the Southern African Institute of Mining and Metallurgy
The study concluded by affirming that the structure exhibited the requisite stability for seamless motion during continuous operations, aligning with operational requirements. Lian and Yao, (2010) provided an insightful exploration of fatigue characteristics through the application of finite element analysis (FEA) to analyse transverse in-plane shear direction methods. The simulation methodology involved the deliberate scattering of material properties, with individual elements affixed with distinct materials to elicit potential failure modes. The experiments were conducted across different materials using FEA, yielding results integral to the assessment of laminate fatigue, particularly in scenarios where outof-plane considerations were omitted. Yoon et al. (2012) elucidated the fundamental operational principles and components of subsea ocean crawlers. The authors expounded upon the control communication system linking the crawler to the mother ship, highlighting the utilization of RS232 on fibre optics media. The control computer system, operating on a QNX controller and MULTIPROG, was detailed for online data acquisition. The study concluded by advocating the implementation of a robust communication system to enhance the dredging capabilities at depths of 6000 m, thereby fortifying the structural stability of the crawler in marine environments.
Methodology This study's primary aim is to determine the peak deflection experienced by the structural frame of the EPRSDC chassis when subjected to static loading conditions. A comprehensive flow chart (Figure 4) is employed to provide a clear visualization of the entire research methodology. The experimental set-up encompassed the simulation of the crawler's operational conditions both under water and at its stationary position. This simulation accounted for factors such as the hydrostatic pressure exerted by the surrounding water, considering the upward-facing component of the crawler. The EPRSDC hardware design was systematically partitioned into three fundamental components prior to the final integration, which was accomplished using the Tetrix Max robotics kit. Specifically, the undercarriage system was subdivided into three distinct elements: the venturi dredging arm, the driving base, and the internal frame. The resulting design dimensions measured 450 mm × 412 mm.
Figure 4—Flow chart of the methodology VOLUME 123
OCTOBER 2023
493
Structural frame analysis of an electrically powered robotic subsea dredging crawler Modelling Utilizing SolidWorks 2019 software, a comprehensive threedimensional modelling process was conducted, as illustrated in Figure 4. The software was instrumental in performing a static loading analysis of the EPRSDC's chassis, constructed from 1050 aircraft-grade aluminium. In accordance with Romahadi, Fitri, and Buana (2023), it is emphasized that weight significantly influences the performance of any moving vehicle, whether subsea or on the ground. Addressing this concern, the authors aimed to develop a lightweight cast wheel design model capable of withstanding a load of 535 N. This necessitated a meticulous analysis involving comparative assessments of various design models and material variations, complemented by static simulations executed using SolidWorks 2018 software. The anticipated outcomes encompassed critical parameters such as von Mises stress, displacement, strain, and factor of safety, with a distinct emphasis on achieving a lightweight design. The simulation results across the three models affirm their capacity to safely endure a load of 535 N, as validated by factor of safety values consistently exceeding the threshold of unity. Moreover, the adapted wheel designs, incorporating diverse materials, demonstrate a reduction in mass compared to the original wheel configurations. Following rigorous material testing, these designs are determined to be structurally sound, as the calculated safety factors meet the prescribed minimum requirements for static loading conditions (i.e., not less than unity). Notably, the magnesium alloy ZK60A exhibits the highest safety factor owing to its superior yield strength of 303 MPa, surpassing the yield strength of aluminium alloy 6061-T6 at 276 MPa. These findings underscore the distinct advantage of the redesigned wheel configuration, yielding a lighter mass in contrast to the original wheel design. The frame, as shown in Figure 5, served as the basis for a FEA simulation to ascertain the viability of employing 1050 aircraftgrade aluminum in the construction of prospective EPRSDCs. Precise measurements of the weight of selected components were taken, with these values subsequently employed to apply specific loads to discern potential areas of deformation.
Finite element analysis FEA was employed to assess the structural integrity, a critical factor in ensuring the robustness of the EPRSDC framework.
Figure 5—Assembled structural frame made from 1050 aircraft-grade aluminum 494
OCTOBER 2023
VOLUME 123
A meticulous examination of the EPRSDC chassis was deemed imperative in meeting established standards. The choice of SolidWorks FEA as the analytical tool was driven by its demonstrated efficacy in such evaluations.
Material selection Table I presents the material properties of 1050 aircraft-grade aluminum, sourced from SolidWorks (2019). The selection of 1050 series aluminum is grounded in its adherence to the anticipated characteristic attributes of diversified application. Moreover, it demonstrates a diminished potential for environmental impact, particularly in terms of ocean pollution, in the context of marine operations.
Boundary fixed condition The chosen boundary conditions (Figure 6) involved eight specific areas with fixed points. These points were situated at the underchannels, featuring two points of support at each chanel.
Loaded force Figure 7 illustrates the downward force exerted on the structural frame model. This load was uniformly distributed across the component. The weight of each component, detailed in Table II, was calculated, initially in kilograms, then converted to grams, and ultimately to newtons. Table I
Material properties for 1050 aircraft-grade aluminum as sourced from SolidWorks (2019) Value
Units
Elastic modulus
3.7e+11
N/m2
Shear modulus
1.5e+11
N/m2
Mass density
3960
kg/m3
Tensile strength
300000000
N/m2
Compressive strength
3000000000
N/m2
Property
Yield strength
−
N/m2
7.4e-06
K-1
Thermal conductivity
30
W/(m·K)
Specific heat
850
J/(kg·K)
Material damping ratio
−
N/A
Thermal expansion coefficient
Figure 6—Boundary condition The Journal of the Southern African Institute of Mining and Metallurgy
Structural frame analysis of an electrically powered robotic subsea dredging crawler Table II
Vendor’s specification for the weight of components used for this analysis Components
Force applied
Result
The first stage of static loading force
First layer 3.773N
The force applied to this component caused the deformation as described using the alphabet A and B as seen in figure 9. The distribution of the total weight acting on the specific structural components of the frame is 3.773N
The pressure pump mount
Second layer 0.73N
The for the static load force applied to the pressure pump mount base is 0.73N as seen in figure 10.
Arm mount base
Third layer 2.808N
The weight of the components that makes up the arm was calculated and loaded to determine the static load acting on the boom arm base at its place of rest and during operation.
Four structural beam support
Fourth layer 9.74N
The four structural beams were responsible for taking the weight of the first layer, the second layer, the boom arm mount, and some other supporting structural components.
Final stage base structural support
Fifth Layer 13.52N
The base support was responsible for carrying the total weight of the first to fourth layer structural frame with other sub-components that makes up the structure of the crawler. The weights were calculated and applied to the base support frame providing us with results to see the possible component(s) that are likely to deform under this static loading force.
Figure 8—Solid mesh of the structural frame for the EPRSDC
Figure 7—Direction of forces applied on the EPRSDC
Static loaded force Meshing of the structural frame In the certification process for the preliminary design of the vehicle structure, it is imperative to take into account the rigorous and unpredictable real-world conditions it will face. This simulation is focused on identifying potential failures that may jeopardize the robot's structural integrity. The mesh model, as depicted in Figure 8, provides a comprehensive overview of the primary shell mesh. It highlights the reference nodes, which are significant in relation to the chassis surfaces subjected to compressional forces. This representation also underscores the mesh spacing within the shell model and exhibits structural variations. The nodes serve as pivotal points for coordinating spatial location where degrees of freedom (DOFs) are precisely defined within the aluminium material. Meanwhile, the elements within the material remain unified.
Results The calculated forces were applied to different components of the crawler’s frame as seen in Figure 7.
Von Mises stress distribution Figure 9 presents a crucial facet of the analysis, showcasing the von Mises stress distribution (comprising both active and residual stress) The Journal of the Southern African Institute of Mining and Metallurgy
Figure 9—Von Mises stress distribution on the EPRSDC
across the entire chassis of the crawler. Notably, the areas exhibiting the highest stress concentration, denoted as A (pressure pump base) and B (boom arm base), manifest stress values in the range of approximately 4.2 × 10⁵ N/m2 and 7.2 × 10⁵ N/m2, respectively. These points are highlighted in green on the von Mises stress data table. It is noteworthy that this stress region remains below the yield stress threshold of 1050 aircraft-grade aluminium. Consequently, the material demonstrates commendable performance under the applied loading conditions, characterized by a low risk of failure. This analysis further underscores that the majority of components fall within the safe range, exhibiting stress values ranging from 8.1 × 103 N/m2 to 2.08 × 10⁵ N/m2. VOLUME 123
OCTOBER 2023
495
Structural frame analysis of an electrically powered robotic subsea dredging crawler In evaluating the structural model in terms of FOS, it is observed that the FOS exceeds 3, aligning with acceptable standards. It is imperative to maintain the working stress on the material at least three times lower than its yield stress. The fact that the point of maximum stress concentration remains comfortably below the yield stress affirms that the model will not fail under the combined weight of its components.
Displacement As depicted in Figure 10, the points exhibiting the maximum displacement are on the boom arm (A) and pump mount base (B). This displacement is attributed to the moment of force acting on the component, which surpasses the longer perpendicular distance between the base point and the endpoints of the arm and pump mount. An increase in the distance between points A and B will result in an increase in the maximum displacement of the structural frame. We therefore propose locating the pump mount arm on the beam, with the arm being reinforced by channels from beneath the base. The point of maximum displacement registers within the red zone, measuring 4.5 × 10-6 m (0.0000455 mm). This result is deemed acceptable, as the displacement magnitude is exceedingly minute and exerts negligible influence on the crawler's performance. In conclusion, there exists potential for refining or optimizing the design to enhance performance. The range of displacement within the blue zone spans from 4.55 × 10-7 m to 1.0 × 10-33 m, affirming the vehicle's capability to withstand its own weight under anticipated operational conditions. To further fortify the design, considerations such as reinforcements or the implementation of a denser material can be explored.
Strain Figure 11 delineates the strain distribution across the EPRSDC model, portraying the material's physical deformation in response to the applied force. Strain is defined as the ratio of the change in length to the original length of a solid material model. For this analysis, we considered the Cauchy strain metric and evaluated the strain limits specific to various materials. Exceeding these limits may lead to structural failure by cracking or breakage. Our model manifests a maximum strain of 8.8 × 10-7, which falls below the established strain limit of the material. Notably, most of the material strain is concentrated within the blue region, measuring 1.4 × 10-7. This indicates that the components integrated into the model are operating well within their designated safe strain limits.
Figure 11—Strain in the structural chassis of the EPRSDC
Plotted points of the affected part of the pressure pump mount The stress plots (N/m2) at 11 characteristics points in the X, Y, and Z directions (mm) are given in Table IV
Plotted points of the affected part of the boom arm base The stress plots (N/m2) at 11 characteristics points in the X, Y, and Z directions (mm) are shown in Table V
Discussion FEA graph for 1050 aircraft-grade aluminium static loading simulation for pressure pump base mount Figure 14 illustrates the von Mises stress distribution at selected nodes. This graph offers insight into the stress profile along the 288 mm channel during the simulated process under static loading conditions for the structural frame chassis. Following the application of a load of 0.73 N to the 39068_txm-288 mm channel (3)-1, the von Mises stress underwent fluctuations at node #342713.
Table III
Summary of the static nodal stress for the pressure pump mount Description
Sum Avg Max Min RMS
Value
Unit
2.067e+06 1.879e+05 4.076e+05 1.005e+04 2.466e+05
N/m2 N/m2 N/m2 N/m2 N/m2
Table IV
Summary for the static nodal stress on the boom arm base Description
Figure 10—Displacement on the structural frame of the EPRSD 496
OCTOBER 2023
VOLUME 123
Sum Avg Max Min RMS
Value
Unit
6.732e+05 7.480e+04 1.037e+05 2.942e+04 7.856e+04
N/m2 N/m2 N/m2 N/m2 N/m2
The Journal of the Southern African Institute of Mining and Metallurgy
Structural frame analysis of an electrically powered robotic subsea dredging crawler
Figure 12—Plotted point of the deforming part affected by the applied force on the pressure pump mount
Figure 13—Deforming part affected by the forces applied to the arm base
Table V
Shows nodal stress for the pressure pump mount Node
Value (N/m2)
X (mm)
Y (mm)
Z (mm)
324273
2.604e+04
−6.48292446
426.77896118
49.41464996
342713
1.005e+04
−7.06871080
428.19320679
87.81465912
342735
9.178e+04
−6.48292446
426.77896118
116.61465454
324268
2.774e+05
−6.48292446
426.77896118
145.41465759
342756
3.905e+05
−7.06871080
428.19320679
174.21464539
342782
3.761e+05
−6.48292446
426.77896118
193.41465759
342780
4.076e+05
−7.06871080
428.19320679
212.61465454
342796
3.219e+05
−6.48292446
425.43426514
235.06993103
324262
1.363e+05
−6.48292446
426.77896118
260.61465454
342833
1.326e+04
−6.48292446
425.43426514
292.55935669
342845
1.571e+04
−7.39072990
427.17266846
335.41384888
The Journal of the Southern African Institute of Mining and Metallurgy
VOLUME 123
OCTOBER 2023
497
Structural frame analysis of an electrically powered robotic subsea dredging crawler
Figure 14—Von Mises stress distribution at selected nodes on the pressure pump base mount
Table VI
Static nodal stress on the boom arm base Node
Value (N/m2)
X (mm)
Y (mm)
Z (mm)
193995
3.716e+04
-296.48291016
364.77899170
113.41464996
193996
9.273e+04
-296.48291016
364.77899170
131.19242859
193997
1.037e+05
-296.48291016
64.77899170
148.97019958
193998
7.914e+04
-296.48291016
364.77899170
166.74798584
193999
7.855e+04
-296.48291016
364.77899170
184.52575684
194000
8.928e+04
-296.48291016
364.77899170
202.30354309
194001
9.211e+04
-296.48291016
364.77899170
220.08131409
204476
7.107e+04
-297.05371094
364.48873901
242.00335693
194004
2.942e+04
-296.48291016
364.77899170
273.41464233
It initially increased to 2.6 x 10⁴ N/m2, subsequently dropping to its lowest point at 1.0 x 10⁴ N/m2 at node #324273. The stress then exhibited a progressive surge, reaching 9.1 × 10⁴ N/m2, which further escalated to 2.77 × 10⁵ N/m2. A subsequent peak stress of 3.9 × 10⁵ N/m2 was reached, followed by a slight decrease to 3.7 × 10⁵ N/m2 at node #342782. The stress subsequently peaked at its highest value of 4.0 × 10⁵ N/m2 at node #342780. From there, it began to gradually decline, reaching its ultimate minimum at 1.5 ×10⁵ N/m2. The graph also underscores two critical points of maximum stress located at the middle of the 288 mm channel, denoted as the maximum stress points of 3.9 × 10⁵ N/m2 and 4.0 × 105 N/m2, corresponding to the base mount on which the pump is affixed.
FEA graph for 1050 aircraft-grade aluminium static loading simulation for the boom arm base Figure 15 presents an analysis of nine selected points on the boom arm base, which bears the weight of the entire arm. The combined weight of the base and arm was quantified as 2.808 N. Von Mises stress distribution was then plotted across the cross-sectional area 498
OCTOBER 2023
VOLUME 123
of the 160 mm channel beam. The analysis revealed notable stress points, initiating at a minimum of 3.7 × 10⁴ N/m2 at node #193995. The stress levels underwent a rapid escalation, culminating in a maximum stress of 1.0 × 10⁴ N/m2. There was a slight reduction in stress on the beam from 7.9 × 10⁴ N/m2 to 7.8 × 10⁴ N/m2. Furthermore, a surge in stress was observed, from 8.9 × 10⁴ N/m2 to 9.2 × 10⁴ N/m2, leading to the second recorded maximum stress. This value subsequently decreased to 7.1 × 10⁴ N/m2 and ultimately to 2.9 × 10⁴ N/m2. In conclusion, it is evident that the material experiences two distinct maximum stress points, measuring 1.0 × 10⁴ N/m2 and 9.2 × 10⁴ N/m2. Notably, the results indicate minimal deformation during the course of the experiment.
Conclusion The static loading simulation, executed through finite element analysis (FEA), yielded critical insights into the maximum deformation and von-Mises stresses exerted on the structural frame of the crawler. This assessment pinpointed specific areas of vulnerability, notably the boom arm and dredging pump base, where stress concentrations were most prominent. Iit is strongly The Journal of the Southern African Institute of Mining and Metallurgy
Structural frame analysis of an electrically powered robotic subsea dredging crawler
Figure 15—Von Mises stress distribution at selected nodes on the boom arm base
recommended that these weak points are reinforced in forthcoming iterations of the crawler. Furthermore, the simulation generated two distinct graphs, one for the pump mount base and another for the boom arm base mount. The 1050 aircraft-grade aluminium demonstrated commendable performance, underscoring its suitability for future developments and providing valuable insights for refined material selection in subsequent iterations. This substantiates its viability for continued use in the advancement of crawler designs.
IMDH Group. 2016. Team | IMDH Group International Mining & Dredging Holding Ltd. : 1–3. https://www.imdhgroup.com/team.php [accessed 28 July 2021]
Acknowledgment
Medagedara, N.T. and Chandra, P.D.S. 2012. Comparing the theoretical and Finite Element stress-strain analysis for axial and torsion loading. https://www. researchgate.net/publication/313504104_Comparing_the_theoretical_and_ Finite_Element_stress-strain_analysis_for_axial_and_torsion_loading
I wish to acknowledge Cape Peninsula University of Technology, Technology Innovation Agency (TIA) South Africa and the Ad Aptronics Advanced Manufacturing Technology Laboratory (AMTL) for their support and contribution towards the success of this research. Special thanks to Associate Professos Atanda K. Raji and Associate Professor Oscar Philander for their support.
References Chutia, S., Kakoty, N.M., and Deka, D. 2017. A review of underwater robotics, navigation, sensing techniques and applications. ACM International Conference Proceeding Series, Part F132085. pp. 1–6. https://www.researchgate.net/ publication/321237107_A_Review_of_Underwater_Robotics_Navigation_ Sensing_Techniques_and_Applications [accessed 30 July 2021] Dai, Y., Yin, W., and Ma, F. 2019. Nonlinear multi-body dynamic modeling and coordinated motion control simulation of deep-sea mining system. IEEE Access, vol. 7. pp. 86242–86251. https://www.researchgate.net/publication/334102401_ Nonlinear_Multi-Body_Dynamic_Modeling_and_Coordinated_Motion_ Control_Simulation_of_Deep-Sea_Mining_System
The Journal of the Southern African Institute of Mining and Metallurgy
Lian, W. and Yao, W. 2010. Fatigue life prediction of composite laminates by FEA simulation method. International Journal of Fatigue, vol. 32, no. 1. pp. 123–133. http://dx.doi.org/10.1016/j.ijfatigue.2009.01.015 Lloyd-Smith, M. and Immig, J. 2018. Ocean pollutants guide. Toxic threats to human health and marine life. National Toxics Network (NTN). 108 pp. https://ipen.org /sites/default/files/documents/ipen-ocean-pollutants-v2_1-enweb.pdf [accessed 31 July 2021]
Romahadi, D., Fitri, M., and Buana, U.M. 2023. Implementation of the finite element method in Solidworks to optimize the front cast wheel design for motorcycles. International Journal of Innovation in Mechanical Engineering & Advanced Materials (IJIMEAM), vol. 4, no. 3. pp. 66–73. https://publikasi. mercubuana.ac.id/index.php/ijimeam/article/download/18794/pdf Thavai, R., Shahezad, Q., Shahrukh, M., Arman, M., and Imran, K. 2015. Static analysis of go-kart chassis by analytical and Solid Works simulation. International Journal of Modern Engineering Research, vol. 5, no. 3. pp. 661–663. Yang, C., Huang, K., Li, Y., Wang, J., and Zhou, M. 2012. Review for development of hydraulic excavator attachment. Energy Science and Technology, vol. 3, no. 2. pp. 93–97. https://core.ac.uk/download/pdf/236300336.pdf Yoon, S.M., Hong, S., Park, S.J., Choi, J.S., Kim, H.W., and Yeu, T.K. 2012. Track velocity control of crawler type underwater mining robot through shallowwater test. Journal of Mechanical Science and Technology, vol. 26, no. 10. pp. 3291–3298. u
VOLUME 123
OCTOBER 2023
499
Structural frame analysis of an electrically powered robotic subsea dredging crawler
SOUTHERN AFRICAN PYROMETALLURGY 2024
INTERNATIONAL CONFERENCE
13-14 MARCH 2024
MISTY HILLS CONFERENCE CENTRE JOHANNESBURG, SOUTH AFRICA
Sustainable Pyrometallurgy - Surviving Today and Thriving Tomorrow INTRODUCTION
TOPICS
he Southern African Pyrometallurgy Conference series has covered a range of themes since the first edition in 2006. The Southern African Pyrometallurgy 2024 International Conference will structure a programme which covers crucial aspects of this continually evolving and exciting field towards sustainable pyrometallurgy. It is envisaged as an international conference on pyrometallurgy with a Southern African flavour.
•
T
Many new developments have taken place in southern African Pyrometallurgical plants since the previous conferences of this series, and it is hoped that this event will provide an update on operational developments. The conference will cover a wide spectrum of activities within pyrometallurgy. Contributions are anticipated from operating plants, universities, research organisations and engineering companies, so that the full spectrum of pyrometallurgical activities can be covered with a special focus on decarbonisation, energy and resource efficiency and cleaner processes. International participation is encouraged, with some broad overview papers covering international trends to be presented by experts in various topical technologies and commodities.
WHO SHOULD ATTEND
Sustainable, socially responsible, environmental, energy conscious smelting
• Update on plant operations and new projects • Staying competitive in an electrically constrained environment • Digital transformation in pyrometallurgy: are smelters smarter now? • Circularity in Pyrometallurgy • Education – how do we sustain and grow our talent pool?
CALL FOR PAPERS AND PRESENTATIONS Papers or presentations are invited on any topic related to the conference. Prospective authors are invited to submit titles and abstracts of their papers in English. The abstracts should be no longer than 500 words and should be submitted to: Camielah Jardine, Head of Conferencing Tel: +27 (11) 538 0237 Email: camielah@saimm.co.za
KEY DATES •
Abstract Submission: 4 September 2023
•
Pyrometallurgists
•
Abstract Acceptance: 25 September 2023
•
Process engineers
•
Paper Submission: 30 October 2023
•
Smelter managers
•
Conference: 13-14 March 2024
•
Researchers
•
Postgraduate students
• 500Academics. OCTOBER 2023
FOR FURTHER INFORMATION, CONTACT:
VOLUME 123
Jardine, E-mail: camielah@saimm.co.za The Camielah Journal of the Southern African Institute of Mining and Metallurgy
Head of Conferencing
Tel: +27 11 530 0238 Web: www.saimm.co.za
Electrical resistivity of heat-treated charcoal by R.D. Cromarty1, S. Bharat1*, and D. Odendaal1 * Paper written on project work carried out in partial fulfilment of BEng (Metallurgical Engineering) degree
Affiliation:
1D epartment of Materials Science and
Metallurgical Engineering, University of Pretoria, Pretoria, South Africa.
Correspondence to: R.D. Cromarty
Email:
robert.cromarty@up.ac.za
Dates:
Received: 29 Nov. 2022 Revised: 31 Aug. 2023 Accepted: 5 Oct. 2023 Published: October 2023
Synopsis
The aim of this investigation was to determine the effect of high-temperature heat treatment on the electrical resistivity of charcoal. Samples of two different wood types (eucalyptus and black wattle) were pyrolised in a retort at a temperature of 700°C and the resulting charcoals heat-treated in an induction furnace at temperatures from 800°C to 1800°C and residence times from 60 to 120 minutes. After cooling, the resistivities of the samples were measured at room temperature using the four-point probe technique. It was found that as the heat treatment temperature increased the electrical resistivity of the charcoal decreased, approaching an asymptotic value at higher temperatures. Longer residence times decreased the resistivity, but this effect was not pronounced.
Keywords
charcoal, electrical resistivity, high temperatures, heat treatment.
How to cite:
Cromarty, R.D., Bharat, S., and Odendaal, D. 2023 Electrical resistivity of heat-treated charcoal. Journal of the Southern African Institute of Mining and Metallurgy, vol. 123, no. 10. pp. 501–508 DOI ID: http://dx.doi.org/10.17159/24119717/2477/2023
Introduction The ferroalloy industry is responsible for a considerable portion of the world’s greenhouse gas emissions. To reduce these emissions, charcoal can be used as a replacement for coal and coke which are currently used as reductants in submerged arc furnaces. Charcoal produced from wood is considered to be a renewable source of carbon due to its short carbon cycle of 5 to 10 years (in certain climates) compared to approximately 100 million years for fossil fuels (Norgate et al., 2011). Wood is a lignocellulosic biomass and is largely composed of lignin, cellulose, hemicellulose, and extractives (He et al., 2018; Surup, Trubetskaya, and Tangstad, 2020). The Eucalypts genus in South Africa has a growth rate of 20 to 35 m3/ha·year, which means that the carbon cycle is about ten years (Sappi, 2022). Black wattle, namely the species A. mearnsii and A. dealbata, is a fast growing invasive tree species in South Africa. It is one of the most popular species used for firewood and has a density range of 550 to 850 kg/m3. The advantage of using black wattle to produce charcoal is that it has a very low ash content (0.6%), low sulphur content (0.01%), and a low nitrogen content (0.13%) (Kosowska-Golachowska et al., 2018). The low sulphur and ash content of charcoal is beneficial to the ferroalloy industry as this reduces the amount of impurities being charged to the furnace during smelting. The sulphur content of carbon reductants in the silicon and manganese industries is required to be less than 0.6%, and the ash content less than 12% (Surup, Trubetskaya, and Tangstad, 2020). Using black wattle to produce charcoal will result in a lower slag volume and reduced sulfur levels in the product alloy. The most common process used to produce charcoal from biomass is pyrolysis. Pyrolysis involves heating the wood to a high temperature in an inert atmosphere, which results in the wood carbonizing to form a solid charcoal product. Some of the factors that influence the pyrolysis process include residence time, maximum temperature, heating rate, and pressure (Dias Junior et al., 2020). In general, charcoal yield from pyrolysis can be optimized by using a low heating rate and long residence time (Dufourny et al., 2019). The available literature on the electrical resistivity of charcoal is with regard to a packed bed, and not individual charcoal particles. Charcoal that has not been subject to heat treatment generally has a high volatile content and a very high electrical resistivity. Monsen et al. (2007) found that the electrical resistivity of a packed charcoal bed before heat treating was too high to measure at room temperature. The authors found that as the heat treatment temperature increased, the electrical resistivity of the bed of charcoal particles decreased.
The Journal of the Southern African Institute of Mining and Metallurgy
VOLUME 123
OCTOBER 2023
501
Electrical resistivity of heat-treated charcoal
Figure 1—Grain-parallel and grain-perpendicular sections of (a) eucalyptus and (b) black wattle wood
Surup et al. (2020) looked at the effect of both high temperatures and particle size on the electrical resistivity of a charcoal bed. It was surmised that the high oxygen content in the charcoal coupled with the disordered carbon structures could be the reason for the high resistivity at temperatures lower than 950°C. An increase in heat treatment temperature results in thermal decomposition of organic compounds, the release of hydrogen- and oxygen-containing volatiles, and a reordering of the residual carbon. As lignin and cellulose are non-graphitizing they will form non-graphitic turbostratic carbon when carbonized at high temperature (Gimåker and Granberg, 2021). This results in a decrease in the electrical resistivity of the charcoal. Surup, Pedersen, and Tangstad (2020) also briefly considered the effect of residence time on the electrical resistivity of charcoal beds at a temperature of 1600°C. It was found that the electrical resistivity initially decreased with increasing time, but after 30 minutes the resistivity decreased at a slower rate and began to approach an asymptotic value. Although the resistivity of particle beds, including charcoal, has been measured the material resistivity of charcoal is seldom reported. To improve understanding of how charcoal behaves in a furnace it would be useful to know how the resistivity changes with increasing temperature. Ideally, it would be best to measure the resistivity at high temperatures. However, due to the difficulties involved in high-temperature resistivity measurements, in this investigation the resistivity was measured at room temperature after heat treatment. One of the main disadvantages of using charcoal as a reductant compared to coke is the cost. Charcoal produced in Australia, making use of Lambiotte retorts for pyrolysis, has an estimated production cost of US$386 per ton, while charcoal produced in Brazil costs US$255 per ton (Suopajärvi and Fabritius, 2013). The cost of producing coke is approximately US$ 237.50 per ton (Makgato, Falcon, and Chirwa, 2019).
Experimental plan Samples of the eucalyptus and black wattle wood that were used to produce the charcoal are shown in Figure 1. The wood was cut into blocks and dried in a drying oven to remove any excess moisture present. The moisture content of the wood was determined using the ASTM D442-20 standard test method (ASTM International, 2020) and the ash content was measured following the ASTM D1102-84 standard test method (ASTM International, 2021a). The charcoal was produced in a retort suspended in a resistance heated furnace. Figure 2 shows the general layout of the retort suspended in the furnace. Nitrogen gas was used to maintain an inert atmosphere in the retort. Furnace temperature was measured using a Type S thermocouple mounted in the furnace, outside of 502
OCTOBER 2023
VOLUME 123
Figure 2—Retort furnace used to produce charcoal. A: SiC resistance furnace; B: retort suspended in the furnace; C: sample on stand inside retort; D: Type S control thermocouple; E: suspension cables; F: gas inlet; G: gas outlet
Figure 3—Retort furnace temperature profile for the pyrolysis process
the retort. A programmable temperature controller was used to control the temperature of the furnace. The temperature profile of the retort used for this process is shown in Figure 3. The maximum pyrolysis temperature was 700°C. After the pyrolysis process had been completed the retort was removed from the furnace and the charcoal was allowed to cool inside the retort. Large pieces of charcoal were cut into smaller blocks for heat treatment. Proximate analysis was conducted to determine the differences in the charcoal produced from the two wood types. The proximate analysis determined the moisture, ash, fixed carbon, and volatile matter contents. The proximate analysis was conducted according to the ASTM standard test method for the chemical analysis of wood charcoal (ASTM D1762-84) (ASTM International, 2021b). The Journal of the Southern African Institute of Mining and Metallurgy
Electrical resistivity of heat-treated charcoal The charcoal samples were subjected to heat treatment cycles in an induction furnace. A schematic of the furnace is shown in Figure 4. The samples were placed in a graphite crucible which was then raised into a preheated graphite susceptor. An inert atmosphere was maintained inside the furnace using argon gas. Heat treatment temperature and residence time were varied while all other factors were fixed. After heat treatment the sample was lowered into the sample cup and allowed to cool in an argon atmosphere. A central composite experimental design was used to allow for a wide range of heat treatment temperatures and times to be investigated while minimizing the number of experiments. The heat treatment plan is shown in Figure 5. An additional heat treatment at 800°C for 90 minutes, indicated in red in Figure 5, does not form part of the central composite experimental design and was simply a further heat treatment that was conducted to obtain more data at lower temperatures. The centre point was repeated three times while all other tests were done once. The electrical resistivity of the charcoal samples was measured by the four-probe measurement technique. Resistivity was measured in two directions: with current flowing parallel to the wood grain and perpendicular to the grain. The diagram in Figure 6 shows the set-up that was used for measuring the electrical resistivities. A Velleman DC Lab power
supply (Velleman Group, 2022) provided a constant current of up to 5 A. A Keithley 2000 multimeter (Keithley Instruments, 2000) was used to measure the voltage drop across the sample. The measurement process was as follows. The probes were first connected to the charcoal sample as shown in Figure 6. The sample was placed between two pieces of aluminium foil backed by polyethylene foam. A constant load was applied to the charcoal sample using the set-up shown in Figure 7. This arrangement ensured good electrical contact across the whole area of the sample and thus a uniform current distribution, as well as reducing any variance between the measurements. The voltage drop across the sample was measured using a set of pins that were a fixed distance apart. The current was first increased from 0.5 A to 4.5 A in 1 A increments, and the corresponding voltage drop recorded. The current was then decreased from 5 A to 1 A in 1 A increments, with the corresponding voltage drop being recorded. These measurements were used to calculate the resistance using Ohm’s Law. Making use of several current and voltage readings for each measurement ensured accurate estimates of the sample resistance. The electrical resistivity was then calculated by using the resistance and the sample dimensions as shown in Equation [1]. [1] where: ρ : Electrical resistivity (Ω∙m) A : Cross-sectional area of the sample (m2) I : Current flowing through the sample (A) L : Length of the sample (m) V : Voltage drop across the sample [V]
Results Wood and charcoal proximate analysis One of the problems encountered was that it was difficult to produce charcoal at the set temperature, 700°C, without cracks
Figure 4—Schematic of the induction furnace used to heat treat charcoal samples. A: Ambrel Ekoheat 15 kW power supply; B: Type S control thermocouple; C: PID temperature controller; D: furnace shell; E: removable sample cup; F: movable sample support rod; G: sample on insulated stand; H: tubular graphite susceptor; I: gas inlet; J: gas outlet Figure 6—Set-up for measuring electrical resistivity using the four-probe method
Figure 5—Heat treatment experimental plan The Journal of the Southern African Institute of Mining and Metallurgy
Figure 7—Set-up for electrical resistivity measurements to allow for a constant load VOLUME 123
OCTOBER 2023
503
Electrical resistivity of heat-treated charcoal forming. The charcoal produced from black wattle is shown in Figure 8. It can be seen that most of the cracks formed in the direction parallel to the wood grain, along the ray cells. Due to these cracks, the charcoal had to be cut into smaller cubes before heat treating, as shown in Figure 9. The proximate analyses of the two different types of charcoal, as well as the moisture and ash contents of the two wood feedstock materials, are shown in Table I.
Figures 10 and 11 shows the response surface plots for the electrical resistivity of eucalyptus charcoal in the direction parallel to the wood grain and perpendicular to the wood grain respectively. The adjusted R2 values were above 0.8 for the two response surfaces, which indicates that the fit of the models accurately represents the measured values. Figures 12 and 13 show the response surface plots for the electrical resistivity of black wattle charcoal in the directions parallel to and perpendicular to the wood grain respectively. The adjusted R2 values were above 0.8 for the two response surfaces, which indicates that the fit of the models accurately represents the measured values. An attempt was made to measure the resistivity of the charcoal produced at 700°C prior to any heat treatment. Using a two-point technique, the resistance exceeded 200 MΩ, the maximum range of the instrument used. This indicates a resistivity of at least 2 MΩ·m. A two-point measurement method was used in this case as the current that could be passed through the sample was too low to accurately measure using the available instrumentation.
Electrical resistivity The resistivity measurements are summarized in four plots based on the wood type and the direction of measurement. The empirical equation which was used to describe the response surface plots is shown in Equation [2], and the values which were substituted in for each plot are summarized in Table II. One of the main causes of variance in the measured resistivities was the cracks present in the samples. [2] where:
SEM analysis
x1: Temperature (°C) x2 :Time (min) a, b, c : Constants
A JEOL JSM-IT300 scanning electron microscope was used in secondary electron mode to view the charcoal samples. SEM
Figure 8—Black wattle charcoal produced through pyrolysis showing cracks that formed
Figure 9—Black wattle charcoal cubes used for resistivity measurements
Table I
Proximate analysis of charcoal and wood Wood
504
OCTOBER 2023
Charcoal
Eucalyptus
Black Wattle
Eucalyptus
Black Wattle
Moisture (wt.%)
11.79
8.30
0.74
0.99
Ash (wt.%)
1.03
0.86
0.57
3.41
Volatile matter (wt.%)
7.29
5.29
Fixed carbon (wt.%)
91.40
90.31
Yield (%)
36.09
28.49
VOLUME 123
The Journal of the Southern African Institute of Mining and Metallurgy
Electrical resistivity of heat-treated charcoal Table II
Summary of values used to plot response surface curves as well as the R2 and SSE values Eucalyptus (parallel)
Eucalyptus (perpendicular)
Black wattle (parallel)
Black wattle (perpendicular)
a
8.358e+08
1.074e+09
5.22e+08
6.92e+08
b
−4.031
−5.066
−2.292
−2.548
c
1.895
3.487
−0.7132
−0.3284
R2
0.8505
0.883
0.8566
0.8463
Adjusted R2
0.8172
0.8569
0.8248
0.8121
SSE
10.78
21.93
3.153
6.692
Figure 10—Effect of temperature and time on the electrical resistivity of charcoal produced from eucalyptus wood in the direction parallel to the wood grain
Figure 11—Effect of temperature and time on the electrical resistivity of charcoal produced from eucalyptus wood in the direction perpendicular to the wood grain
Figure 12—Effect of temperature and time on the electrical resistivity of charcoal produced from black wattle wood in the direction parallel to the wood grain The Journal of the Southern African Institute of Mining and Metallurgy
VOLUME 123
OCTOBER 2023
505
Electrical resistivity of heat-treated charcoal
Figure 13— Effect of temperature and time on the electrical resistivity of charcoal produced from black wattle wood in the direction perpendicular to the wood grain
Figure 14—Secondary electron SEM images taken at 200× showing the structure of charcoal parallel to the wood grain: (a) eucalyptus original charcoal pyrolysed at 700 °C, (b) black wattle original charcoal pyrolysed at 700°C, (c) eucalyptus heat treated at 1800°C for 90 minutes, (d) black wattle heat treated at 1800°C for 90 minutes
Figure 15—SEM images taken at 200× showing the structure of charcoal perpendicular to the wood grain: (a) eucalyptus original charcoal pyrolysed at 700°C, (b) black wattle original charcoal pyrolysed at 700°C, (c) eucalyptus heat treated at 1800°C for 90 minutes, (d) black wattle heat treated at 1800 °C for 90 minutes
images were taken of both eucalyptus and black wattle charcoal to determine if there was any change in the microstructure due to the heat treatments. Figure 14 shows the structures of eucalyptus and black wattle charcoal parallel to the wood grain, and Figure 15 shows the structures perpendicular to the wood grain. 506
OCTOBER 2023
VOLUME 123
Discussion The results show that in general, the electrical resistivity of the charcoal samples decreased with an increase in heat treatment temperature and residence time, although this decrease became less The Journal of the Southern African Institute of Mining and Metallurgy
Electrical resistivity of heat-treated charcoal pronounced as the temperature increased. The greatest decrease in resistivity took place between 700°C, the temperature at which the charcoal was produced, and 800°C. The trend in electrical resistivity in the charcoal produced from eucalyptus varied slightly between the grain-parallel and grain-perpendicular directions. The resistivities measured in the perpendicular direction (Figure 11) were slightly higher than those measured in the parallel direction (Figure 10). The resistivity measured in the perpendicular direction decreased from 15.3 mΩ∙m at 800°C to 0.4 mΩ∙m at 1800°C, and in the parallel direction from 9.6 mΩ∙m at 800°C to 0.3 mΩ∙m at 1800°C. The difference between the two directions, however, was not as significant as was originally expected. It was expected that there would be a change in resistivity in the two directions due to the difference in structure. However, the SEM analysis showed that the eucalyptus charcoal has a similar structure and porosity in both directions. This can be seen when comparing Figure 14 (a) and (c), the structure parallel to the wood grain, to Figure 15 (a) and (c), the structure in the perpendicular direction. The trend in resistivity in the charcoal produced from black wattle showed a similar response. The resistivities measured in the perpendicular direction (Figure 13) were once again found to be slightly higher than those measured in the parallel direction (Figure 12). The electrical resistivity measured in the grain-perpendicular direction decreased from 7.1 mΩ∙m at 800°C to 0.3 mΩ∙m at 1800 °C, and in the grain-parallel direction from 4.7 mΩ∙m at 800°C to 0.2 mΩ∙m at 1800°C. Comparison of the charcoal structures in the two different directions showed that black wattle exhibits a more distinct structural difference than eucalyptus. The charcoal was more porous in the parallel direction, as seen in Figure 14 (b) and (d), compared to the perpendicular direction (Figure 15 b and d). The resistivity model presented in Equation [2] is a purely empirical model. An attempt was made to fit models based on the rate of recrystallization of the carbon, But none of the models provided an acceptable fit to the measured data. The data was fitted to an Arrhenius plot without taking the effect of time into account. Using the average resistivity for each temperature it was found that the activation energy for resistivity change was 59.3 kJ/mol and 51.6 kJ/mol for the eucalyptus and black wattle charcoals respectively. The microstructures of the two types of charcoal did not change significantly with temperature, as seen in Figures 14 and 15. Changes may have occurred to the structure of the charcoal at an atomic level, but these changes would not be visible in the SEM images.
Conclusions The effect of heat treatment temperature and residence time on the electrical resistivity of charcoal made from two different types of wood was investigated. The following conclusions were drawn from the results. ➤ The electrical resistivity of both eucalyptus and black wattle charcoal decreased with an increase in heat treatment temperature and residence time. ➤ In both cases the resistivities measured in the grainperpendicular direction were higher than those in the grain parallel direction. ➤ The cracks present in the charcoal blocks would have had a significant effect on the resistivity, thus affecting the results obtained. ➤ SEM analysis showed that there was no significant change The Journal of the Southern African Institute of Mining and Metallurgy
in the microstructure of the charcoal with increasing temperature. There could possibly be changes at an atomic level that would not be visible in the images.
Recommendations In future projects related to this topic, it is recommended that lower temperatures be used for the pyrolysis process to reduce the amount of cracking that occurs. Additional heat treatments could be conducted in the temperature range of 700°C to 1000°C to gain a better understanding of the trend in resistivity. It is also recommended that the ash of the charcoal produced should be analysed to get a better understanding of the composition of the material. This could be done using either ICP-OES or XRF analysis.
References ASTM International. 2020. ASTM D4442-20. Standard test methods for direct moisture content measurement of wood and wood-based materials. https://doi.org/10.1520/D4442-20 ASTM International. 2021a. ASTM D1102-84. Standard test method for ash in wood. https://doi.org/10.1520/D1102-84R21 ASTM International. 2021b. ASTM D1762-84. Standard test method for chemical analysis of wood charcoal. https://www.astm.org/d1762-84r21.html Dias Junior, A.F., Esteves, R.P., da Silva, Á.M., Sousa Júnior, A.D., Oliveira, M.P., Brito, J.O., Napoli, A., and Braga, B.M. 2020. Investigating the pyrolysis temperature to define the use of charcoal. European Journal of Wood and Wood Products, vol. 78, no. 1. pp. 193–204. https://doi.org/10.1007/s00107-019-014896 Dufourny, A., van de Steene, L., Humbert, G., Guibal, D., Martin, L., and Blin, J. 2019. Influence of pyrolysis conditions and the nature of the wood on the quality of charcoal as a reducing agent. Journal of Analytical and Applied Pyrolysis, vol 137. pp. 1–13. https://doi.org/10.1016/j.jaap.2018.10.013 Gimåker, M. and Granberg, H. 2021. Graphite materials – Production from biomass? https://urn.kb.se/resolve?urn=urn:nbn:se:ri:diva-58964 He, C., Tang, C., Li, C., Yuan, J., Tran, K.Q., Bach, Q.V., Qiu, R., and Yang, Y. 2018. Wet torrefaction of biomass for high quality solid fuel production: A review. Renewable and Sustainable Energy Reviews, vol. 91. pp. 259–271. https://doi.org/10.1016/j.rser.2018.03.097 Keithley Instruments. 2000. Model 2000 Multimeter User’s Manual. https:// download.tek.com/manual/2000-900_J-Aug2010_User.pdf Kosowska-Golachowska, M., Magdziarz, A., Wolski, K., and Luckos, A. 2018. A study into the combustion process of Acacia mearnsii (black wattle) in a circulating fluidized bed. Proceedings of the 23rd International Conference on Fluidized Bed Conversion, Seoul, South Korea. https://www.researchgate.net/ publication/325273201 Makgato, S., Falcon, R.M.S., and Chirwa, E.M.N. 2019. Reduction in coal fines and extended coke production through the addition of carbonisation tar: Environmentally clean process technology. https://repository. up.ac.za/bitstream/handle/2263/74833/Makgato_Reduction_2019. pdf?sequence=1&isAllowed=y Monsen, B., Tangstad, M., Solheim, I., Syvertsen, M., Ishak, R., and Midtgaard, H. 2007. Charcoal for manganese alloy production. Proceedings of INFACON XI, New Delhi, India, 18-21 February 2007. pp. 297–310. https:// www.pyrometallurgy.co.za/InfaconXI/297-Monsen.pdf Norgate, T., Haque, N., Somerville, M., and Jahanshahi, S. 2011. The greenhouse gas footprint of charcoal production and of some applications in steelmaking. Proceedings of the 7th Australian Conference on Life Cycle Assessment. Australian Life Cycle Assessment Society. https://publications.csiro. au/rpr/download?pid=csiro:EP104971&dsid=DS3 Sappi. 2022. Frequently asked questions about eucalypts. https://cdn-s3.sappi.com/ sfs-public/Sappi-FAQs-Eucalypts-6.pdf Suopajärvi, H. and Fabritius, T. 2013. Towards more sustainable ironmaking - An analysis of energy wood availability in Finland and the economics of charcoal production. Sustainability (Switzerland), vol. 5, no. 3. pp. 1188–1207. https:// doi.org/10.3390/su5031188 Surup, G.R., Pedersen, T.A., Chaldien, A., Beukes, J.P., and Tangstad, M. 2020. Electrical resistivity of carbonaceous bed material at high temperature. Processes, vol. 8, no. 8. https://doi.org/10.3390/PR8080933 Surup, G.R., Trubetskaya, A., and Tangstad, M. 2020. Charcoal as an alternative reductant in ferroalloy production: A review. Processes,vol. 8, no. 11. pp. 1–41. https://doi.org/10.3390/pr8111432 Velleman Group. 2022. Velleman LABPS3005N User Manual. Gavere, Belgium. u VOLUME 123
OCTOBER 2023
507
WCSB11
Electrical resistivity of heat-treated charcoal
THE 11TH WORLD CONFERENCE OF SAMPLING AND BLENDING
21-23 MAY 2024 HYBRID CONFERENCE
MISTY HILLS CONFERENCE CENTRE MULDERSDRIFT, JOHANNESBURG
SOUTH AFRICA
The World Conference on Sampling and Blending (WCSB), to be held in South Africa, 21-23 May 2024, is the eleventh such conference to promote the Theory of Sampling (TOS). The WCSB conference provide a meeting place for professionals interested in sampling theory, practice, experience, applications, and standards. The Conference will provide understanding and insights for academics, manufacturers, engineering firms and practitioners aiming to achieve representative sampling. TOS effectively identifies the source of sampling variability and provides valuable solutions for minimising each source of sampling uncertainty. The aim of WCSB11 is to invite and encourage the diverse international sampling community to adopt and disseminate the concepts and ideas for a standardized approach to sampling embodied in the TOS. The Conference will also offer a forum for fruitful discussions between statisticians committed to ‘Measurement of Uncertainty’ (MU) and proponents of the TOS by offering a unifying foundation for development of better and more general standards. While the Theory of Sampling had its historical origins in the mining industry, today it also applies to sampling of a broad range of bulk materials, minerals, agricultural raw materials and products, the food, feed, and pharmaceutical industries, as well as sampling for environmental applications. WCSB11 is an event of global significance that aims to improve sampling practices in all sectors of science, technology, and industry, for consultants, managers, sampling and quality control staff, researchers, engineers, and manufacturers operating in many industries, The opportunity to meet,
exchange ideas, and share practical experiences will be a significant benefit for attendees. The proceedings of the Conference will be published in electronic format with a strict adherence to an editorial and peer review policy that will allow academics to attract the publication subsidy for published academic research. Adherence to these standards will enable the wider dissemination of the TOS in international scientific, technological, and industrial sectors. WCSBs have helped to promote the teaching of TOS at universities, with postgraduate courses in TOS being taught in some countries. The Pierre Gy Gold Medal is awarded at each WCSB conference to individuals who have been most effective and successful around the world in disseminating and promoting TOS. This achievement will again be celebrated at WCSB11. The medallists are a unified body of champions capable of teaching, promoting, and researching aspects of sampling theory and practice, supporting the efforts of original equipment manufacturers to uphold TOS rules of sample representativeness. WCSB conferences aim to develop a unified vision for specific quality control protocols for sampling and blending activities, with participation and collaboration of industry professionals. The theme of sustainable science, technology, and industry introduced at WCSB10 is upheld, with emphasis on the UN World Development Goals number 9 and 12, addressing sustainable industry, innovation, and infrastructure, and responsible production and consumption. Topics around societal, industrial, and environmental aspects of particulate sampling in mining,
FOR FURTHER INFORMATION, CONTACT: 508
OCTOBER 2023
Camielah Jardine, VOLUME 123 Head of Conferencing
E-mail: camielah@saimm.co.za The Journal of the Southern African Institute of Mining and Metallurgy Tel: +27 11 538-0237, Web: www.saimm.co.za
Assessment of coal washability data obtained via the RhoVol analyser by D. Stone1*, Q.P. Campbell1, M. le Roux1, and M. Fofana2 * Paper written on project work carried out in partial fulfilment of BEng (Chemical Engineering) degree
Affiliation:
1 North-West University, Potchefstroom,
South Africa.
2DebTech, Johannesburg, South Africa.
Correspondence to: D. Stone
Email:
dyllanstone1995@gmail.com
Dates:
Received: 30 Nov. 2019 Revised: 31 Mar. 2021 Accepted: 8 Oct. 2023 Published: October 2023
Synopsis
Float-and-sink analysis is widely used in the coal industry to obtain washability data, yielding important information about beneficiation potential and performance. This method is associated with health and environmental problems, and research into alternative densimetric methods is important. The RhoVol is a new technology developed by De Beers Group Technology South Africa (DebTech) for the image-based generation of densimetric data from ores. The analysis is done on a per-particle basis, within a size range of +3 mm –8 mm. In this paper we compare the established methods used for the float-and-sink analysis of coal with a new method using RhoVol. The aim of the study was to determine the validity of the information obtained from the RhoVol analyser by performing comparative densimetric analysis on three different coal samples. The results showed that the RhoVol method was more rapid, safe, and precise, but tended to consistently underestimate the density of the coal sample, probably due to varying coal porosity.
Keywords
coal, float-and-sink analysis, RhoVol, washability determination.
How to cite:
Stone, D., Campbell, Q.P., le Roux, M., and Fofana, M. 2023 Assessment of coal washability data obtained via the RhoVol analyser. Journal of the Southern African Institute of Mining and Metallurgy, vol. 123, no. 10. pp. 509–512 DOI ID: http://dx.doi.org/10.17159/24119717/1047/2023 ORCID: Q.P. Campbell http://orcid.org/0000-0003-0510-6018 M. le Roux http://orcid.org/0000-0002-9330-426X
Introduction Environmental concerns and legislative restrictions on the coal industry, along with depleting high-quality coal reserves in South Africa, serve as motivation for the improvement of efficiency at coal washing facilities. Any efficient gravity-based beneficiation process requires a thorough understanding of the density profile of the material. The washability curve is frequently used to quantify the densimetric data of coal on washing plants (King, 2012). The data required for the construction of washability curves is obtained by doing float-and-sink analysis, where a sample is separated into relative density fractions by using a dense-liquid medium. Problems related to float-and-sink analysis include the high cost of dense media, adverse health and environmental effects, and long turnaround times. The RhoVol, which is a 3D image-based densimetric measurement system developed by DebTech, gives vital information on a material’s density profile. This is done by means of camera-based measurements on a per-particle basis. Published work on the RhoVol is limited to a study by Fofana and Steyn (2018), who compared densimetric analysis of kimberlite material using the float-and-sink method to results from RhoVol. They obtained by the RhoVol method. The current investigation expands on the previous study, using coal as a test material.
Literature survey Float-and-sink tests Float-and-sink analysis is a common laboratory method performed on ores to obtain densimetric data, thereby assessing the suitability of dense medium separation and determining the economic separation density of an ore (Wills and Napier-Munn, 2006). It is still considered the most effective flotation technology to separate ores based on density (Kong et al., 2018). Float-and-sink analysis delivers densimetric data by reporting particle separation based on the mass fraction of the sample. The usefulness of the results is determined by the width and number of density class intervals considered. A standard float-and-sink test is conducted as follows. Heavy liquids covering a range of desired densities in incremental steps are prepared, and the sample is introduced to the liquid of the lowest density. After sufficient time is allowed for settling, the floats product is removed, washed, dried, and weighed while the sink product is washed and placed in the next heavy liquid. These steps are repeated until the last density fraction is obtained, from which both floats and sinks are washed, dried, and weighed (SANS 7936:2010).
The Journal of the Southern African Institute of Mining and Metallurgy
VOLUME 123
OCTOBER 2023
509
Assessment of coal washability data obtained via the RhoVol analyser Coal is graded according to its ash-forming mineral content, with higher ash values resulting in lower grades (Wills and NapierMunn, 2006). Since there is a direct correlation between the ashforming minerals within the coal and the coal density (Sivrikaya, 2014), the data obtained from the float-and-sink analysis is an indication of the quality of a coal. The shape of the washability distribution also provides information about the difficulty of separation, as well as the yield and expected efficiencies and specified cut-points (Sahu, Chaurasia, and Sursh, 2018). Float-andsink analyses are routine on coal preparation plants. The results are used to evaluate and cross-check the washability characteristics of the coal and determine the process performance (Bhattacharya and Anand, 1998). Due to the high cost of heavy media and the adverse health and environmental effects of organic liquids, float-and-sink analysis is an unfavourable method for densimetric analysis (Aktas, Karacan, and Olcay, 1998). For fine particles below about 1 mm, the method is extremely time-consuming due to long settling times (Franzidis and Harris, 1986). Coal porosity and moisture content are important factors affecting float-and-sink results. Inefficient separation of floats and sinks occurs when the coal moisture content is too high (Aktas, Karacan, and Olcay, 1998). This is because during the procedure ZnCl2 solution fills the empty macropores of the coal. Any moisture present within the coal pores dilutes the ZnCl2 solution, decreasing its density. Since the solution diffuses far into the pores, a density gradient is formed, extending from the inside of the pore to the coal particle surface. This leads to a variation of both the liquid density and the apparent density of the particle (Aktas, Karacan, and Olcay, 1998). This could result in particles reporting to the incorrect density fraction, compromising the results. Permeation of ZnCl2 into coal pores remains a problem after floatand-sink analysis. The ZnCl2 solution cannot be removed, even after prolonged washing. During drying, all moisture is removed from the coal and the precipitated salt remains within the pores. All salt retained within the coal create numerous adverse effects; the additional mass compromises float-and-sink results when dried samples are weighed, and ZnCl2 increases the apparent ash value obtained from proximate analysis is performed (Campbell, le Roux, and Smith, 2015).
The RhoVol density analyser The RhoVol analyser (Figure 1) can be described as a densimetric measurement system that determines the density of an ore by weighing and estimating the volumes of individual particles (Fofana and Steyn, 2018). Along with the weight and volume parameters, the RhoVol also gives information concerning particle density, size, and shape. Shape descriptors include compactness, flatness, and elongation (DebTech, 2018). The machine operates in an automated batch process mode, processing an d 1000 particles an hour. During operation, the RhoVol is capable of sorting the sample into ten discThere are tTwo models –a fine particle analyser for the size range between 3 mm and 8 mm (the version used in this investigation), and a coarse unit for the size range between 8 mm and 25 mm. The samples are required to be correctly sized before analysis and free of fines/dust. The sample is presented to the RhoVol by means of two vibratory feeders – a primary feeder and a bowl feeder (Fofana and Steyn, 2018). The feeders both vibrate at the resonance frequency of the sample particles, allowing the particles to progress forward and ensuring that the particles are fed to the RhoVol individually. The particles are then dropped to a high-speed weighing cell to ensure accurate measurement of particle mass. The particle mass limits are 6 mg to 150 g (DebTech, 2018). The shape and volume parameters are determined from a 3D reconstruction generated by the analysis of multiple images 510
OCTOBER 2023
VOLUME 123
Figure 1—RhoVol density analyser (courtesy of DebTech)
(silhouettes) of a particle captured by seven cameras situated within the machine. These cameras are set up at different angles to capture different silhouettes of the particles while airborne, and a 3D model, called the visual hull, is generated (Mangera, Morrison, and Voight, 2016). By definition, a visual hull is the single largest object that is consistent with a finite set of available silhouettes (Forbes, Voight, and Bodika, 2003). Of the many approaches used to reconstruct 3D objects from 2D images, silhouette-based reconstruction is among the most popular options (Franco and Boyer, 2003). This is mostly due to the robustness of the algorithms and the easy implementation thereof (Franco and Boyer, 2003). Visual hulls are only approximations of an object's shape and volume and there is much room for error. An issue with the visual hull approach is its inability to discriminate between different nonconvex objects. Any physical feature of a particle that might lie in a concavity will not manifest in any possible silhouette (Laurentini, 1994). It has been shown in previous studies that visual hull-based volume reconstruction methods will always overestimate the particle volume (Forbes, Voight, and Bodika, 2003). This is also true for the RhoVol, where mean volume estimation errors of up to 20% have been found (Mangera, Morrison, and Voight, 2016). Mangera, Morrison, and Voight (2016) determined the discrepancy between the true volume (‘ground truth’ volume) and the visual hull-generated volume for three different particle shapes – round (BB1), elongated (EL3), and flat (FL1). The results can be seen in Figure 2. From Figure 2, two things are clear. The first is that the RhoVol consistently overestimates the volume of all the particles it analyses. The second is that the particle shape plays a large role in the degree of overestimation. It can be seen, by the normalized particle volume distribution, that the RhoVol is better at consistently determining the
Figure 2— Normalized volume distribution of three particles captured 10 times in the RhoVol system compared to the true volume (Mangera, Morrison, and Voight, 2016) The Journal of the Southern African Institute of Mining and Metallurgy
Assessment of coal washability data obtained via the RhoVol analyser volumes of round particles than it is for flat particles. The effect of shape on the volume estimation is mainly due to particle orientation. According to Mangera, Morrison, and Voight (2016) and Forbes, Voight, and Bodika (2003), varying orientations of particles allows different silhouettes to be fpresented, creating vastly different visual hulls. To account for the volume overestimation and orientation problems, two systems are used to increase the accuracy of the visual hulls generates RhoVol. The first is a multi-pass mode, which allows for the reprocessing of a sample multiple times (DebTech, 2018). When doing this, the RhoVol has the capability to recognize an individual particle based on its stored database of silhouettes and particle masses. When it recognizes a particle again, the silhouettes from the first pass are combined with those of the second pass to construct a more reliable 3D model of the particle, increasing accuracy. The more silhouettes the RhoVol obtains of a single particle, the more accurately a visual hull can be constructed. However, the ultimate volume of a coal particle cannot be accurately determined since RhoVol cannot determine he porosity and surface texture of the material. The second system used to improve accuracy is the dynamic correction factor. Since the error of the visual hull is largely shaperelated, RhoVol can apply a shape-dependent volume correction factor. The correction factor can be determined experimentally, comparing the RhoVol volumes to volumes calculated using other methods such as pycnometry. Although shape plays a major role in the correction factor, it should be noted that the correction factor may also differ for various materials, based on their intrinsic nature.
After each test, the remaining solution on the samples was removed by prolonged washing with water. The samples were airdried in an oven for approximately 14 hours at 110°C, in accordance with SANS 589:2009. After drying, each sample was weighed, bagged, and tagged.
Experimental
After fractionation of the samples (both by RhoVol and float-andsink) the materials were subjected to ash analysis according to SANS 131:2011. Three repeats were completed for each sample to ensure accuracy.
RhoVol analysis
The RhoVol BDD127SM density analyser at the DebTech campus in Crown Mines, South Africa, was used for the analysis. The entire population of the sample was subjected to RhoVol analysis. For each type of coal, a sample of about 3 000 particles, equivalent to about 400–500 g, was tested. A historical dynamic correction factor derived from previous tests on Leeuwpan coal,.which was assumed to be similar to the Witbank sample since both coals were from the same coalfield, was applied for the Witbank sample. This correction factor was not applicable to the Moatize and Molteno coals. The samples were analysed for size, shape, mass, and consequently density, on a particle-by-particle basis. The machine was also set to simultaneously physically separate each sample into 10 density bins. The separation criteria for each bin were determined by dividing the entire density range (1.2–2.2 g.cm-3 in this case) by 10, hence the exact values of the interval density limits for each bin were arbitrary. In the interest of time, only one RhoVol pass was used in this project, although it is acknowledged that the accuracy of the volume determinations improves significantly with each additional pass. Each pass of 3 000 particles takes about 3 hours to complete.
Ash analysis
Coal samples
Three coal samples from different sources were used in the study, as described in Table I. The samples were screened to exclude any +8 mm and –3 mm materials, which are the limits of the particular RhoVol machine that was used. Sufficient material was prepared for all the analyses to be performed.
Results and discussion
Figure 3 shows the densimetric curves for the Moatize coal. An average deviation of 8.27% is observed between the two curves. It is clear that the RhoVol densities are consistently lower than those from float-and-sink analysis. This can be explained as follows. ➤ Visual hull-based object reconstruction overestimates the volume of an object (Mangera, Morrison, and Voight, 2016). Since volume is inversely proportional to density, the density of material analysed by the RhoVol is underestimated. ➤ The RhoVol is unable to distinguish coal porosity from other surface concavities (Forbes et al., 2003). This means that the RhoVol determines the entire particle volume, which includes the volume of the pores, and not the true volume of the
Float-and-sink analysis
The float-and-sink analyses were executed according to SANS 7936:2010. Zinc chloride standard solutions were prepared, with densities ranging from 1.2 to 1.8 in increments of 0.1. For all coals, the testing was done in ascending order of relative density. To prevent coal fines and other contaminants form affecting the results, the relative densities of each of the solutions were re-measured after every float-and-sink iteration. Where any deviations were found, the appropriate corrections were made. Three repeats were done for each coal species. Table I
Origin and properties of the sample used. Sample
Size
Moisture content1
volatile matter2
Ash content2
Fixed carbon
Comments
Moatize
+3-8mm
1.6%
32.2%
17.9%
48.0%
Coal from the Moatize coalfield in the Tate province of Mozambique (Howler et al., 2012)
Molteno
+3-8mm
2.6%
8.8%
36.7%
51.0%
Coal from the Molteno-Indwe coalfield in the Eastern-Cape province of South Africa (Cobban et al., 2009)
Witbank
+3-8mm
3.5%
19.2%
38.7%
42.1%
A mixture of leftovers from the Witbank area used in previous experiments; it is not representative of any particular coal source
1 Air dried · 2 Dry bases The Journal of the Southern African Institute of Mining and Metallurgy
VOLUME 123
OCTOBER 2023
511
Assessment of coal washability data obtained via the RhoVol analyser
Figure 3—Moatize coal: RhoVol and float-and-sink densimetric curves
particle, which excludes pore volume. Due to this, the particle volume is overestimated, resulting in an underestimation of the density. ➤ ZnCl2 salt adsorption within coal pores (Campbell, le Roux, and Smith, 2015). During float-and-sink analysis, ZnCl2 permeates into the coal pores, and remains there despite vigorous washing after the analysis. The increase in the sample weight due to residual ZnCl2 increases the calculated material fraction in the specified density range. Figure 4 shows the densimetric curves for the Molteno coal. An average deviation of 11.48% is observed between the two curves. Once again, it is seen that the RhoVol underestimates the density of the sample, due to the abovementioned factors. Figure 5 shows the densimetric curves for the Witbank coal. An average deviation of 2.98% is observed between the two curves. The RhoVol data shows much better correspondence to the float-andsink data than with the other two coals. This is due to the volume correction factor used for this coal during RhoVol operation. However, a discrepancy is still observed at densities less than 1.4. This is due to the relationship between porosity and coal grade (Galvin, 2006). Porosity is more extensive in high-grade coals; i.e. the density is lower (Li et al,. 2017). This increaed porosity at lower density affects the results of the RhoVol. Explanations 2 and 3 above apply for low density classes.
Conclusions
Comparison between the float-and-sink and the RhoVol results for the three coals indicates that the RhoVol analyser is a viable technical alternative to float-and-sink analysis, provided it is calibrated correctly. This is based on the low observed experimental error for the dfferent coals, indicating that the analysis is in line with industry standards, where errors of up to 5% are common (Fofana and Steyn, 2018). The RhoVol delivered unsatisfactory results for the Moatize and Molteno coal samples, for which it was not correctly calibrated. The RhoVol underestimated the density of these samples compared to float-and-sink analysis. For the Witbank coal the RhoVol results showed very good correspondence to the float-and-sink results, for densities higher than 1.4. For lower densities a deviation was observed, with the RhoVol underestimating the density. The underestimation can be attributed to coal porosity, which leads to the particle volume being overestimated.
References
Aktas, Z., Karacan, F., and Olcay, A. 1998. Centrifugal float – sink separation of fine Turkish coals in dense media. Fuel Processing Technology, vol. 55, no. 1. pp, 235–250. Bhattacharya, S. and Anand, V. 1998. Estimation of grindability from sink-float test data for two different coals. International Journal of Mineral Processing, vol. 53. pp. 99–106. Campbell, Q.P., le Roux, M., and Smith, I.G.T. 2015. Water-only laboratory coal fractionation using the reflux classifier. Minerals Engineering, vol. 83 (January). pp. 59–63. Cobban, D.A., Rossouw, J.N., Versfeld, K., and Nel, D. 2009. Water quality considerations for opencast mining of the Molteno Coal Field, Indwe, Eastern Cape. Proceedings of the International Mine Water Conference, Pretoria, 19 – 23 512
OCTOBER 2023
VOLUME 123
Figure 4—Molteno coal: RhoVol and float-and-sink densimetric curves
Figure 5—Witbank coal: RhoVol and float-and-sink densimetric curves October 2009. International Mine water Association. https://www.imwa.info/docs/ imwa_2009/IMWA2009_Cobban.pdf DebTech. 2018. RhoVol - Densimetric measurement system. https://www.debtech.com/ product-Rhovol.html Fofana, M. and Steyn, T. 2018. Monitoring the performance of DMS circuits using RhoVol technology. Journal of the Southern African Institute of Mining and Metallurgy, vol. 119, no. 2. pp. 133–138. Forbes, K., Voigt, A., and Bodika, N. 2003. Using silhouette consistency constraints to build 3D models. Proceedings of the Fourteenth Annual Symposium of the Pattern Recognition Association of South Africa (PRASA 2003). pp. 33–38. Franco, J-S. and Boyer, E. 2012. Exact polyhedral visual hulls. Proceedings of the 2004 IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops (CVPRW’04). IEEE, New York. pp. 32.1–32.10. Franzidis, J. and Harris, M.C. 1986. A new method for the rapid float-sink analysis of coal fines. Journal of the Southern African Institute of Mining and Metallurgy, vol. 86, no. 10. pp. 409–414. Galvin, K.P. 2006. Options for washability analysis of coal A literature review. Coal Preparation, vol. 26, no. 4. pp. 209–234. Hower, J. 2016. Coal. Kirk-Othmer Encyclopaedia of Chemical Technology. Wiley. King, R.P. 2012. Modeling and Simulation of Mineral Processing Systems. 2nd edn. MSociety for Mining, Metallurgy, and Exploration, Engelwood, CO. Kong, L., Bai, J., Li, H., Chen, X., Wang, J., and Bai, Z. 2018. The mineral evolution during coal washing and its effect on ash fusion characteristics of Shanxi high ash coals. Fuel, vol. 212. pp. 268–273. Laurentini, A. 1994. The Visual Hull Concept for Silhouette-Based Image Understanding. IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 16, no. 2. pp. 150–162. Li, Y., Zhang, C., Tang, D., Gan, Q., Niu, X., Wang, K., and Shen, R. 2017. Cole pore size distributions controlled by the coali_cation process: An experimental study of the coals from the Junggar, Ordos and Qinshui basins in China. Fuel, vol. 206. pp. 352–363. Mahajan, O.P. 1982. Coal porosity. Coal Structure. Meyers, R.A. (ed.) Academic Press, New York. pp. 51–84. Mangera, R., Morrison, G., and Voigt, A. 2016. Particle volume correction using shape features. Proceedings of the 2016 Pattern Recognition Association of South Africa and Robotics and Mechatronics International Conference, Stellenbosch, South Africa. IEEE, New York. C:1–6. Rodrigues, C.F. and de Sousa, M.J.L. 2002. The measurement of coal porosity with different gases. International Journal of Coal Geology, vol. 48. pp. 245–251. Sahu, D., Chaurasia, R.C., and Sursh, N. 2018. Mineralogical characterization and washability of Indian coal from Jamadoba. Energy Sources Part A: Recovery Utilization and Environmental Effects. September 2018, pp. 517–526. SANS (South African National Standards). 2009. SANS 589. Hard coal – Determination of total moisture. SABS Standards Division, Pretoria: SANS (South African National Standards). 2010. SANS 7936. Hard coal – Determination and presentation of float and sink characteristics – General directions for apparatus and procedures. SABS Standards Division, Pretoria. Sivrikaya, O. 2014. Cleaning study of a low-rank lignite with DMS, Reichert spiral and flotation. Fuel, vol. 119. pp. 252–258. Wills, B.A. and Napier-Munn, T.J. 2006. Wills’ Mineral Processing Technology. 7th edn. Elsevier. u The Journal of the Southern African Institute of Mining and Metallurgy
NATIONAL & INTERNATIONAL ACTIVITIES 2023 7-9 November 2023 — Hydroprocess 2023 14TH International Conference on Process Hydrometallurgy Sheraton Hotel, Santiago, Chile Website: https://gecamin.com/hydroprocess/index 14-16 November 2023 — Slope Stability in Mining Conference 2023 The Western Perth, Western Australia Website: https://acg.uwa.edu.au/class/ssim2023/ 15 November 2023 — Decarbonisation of the Minerals Industry-Challenges and opportunities Online Colloquium 2023 Carbon Tax: South Africa and the context of the global mining industry Contact: Camielah Jardine Tel: 011 538-0237 E-mail: camielah@saimm.co.za Website: http://www.saimm.co.za 15-16 November 2023 — African Exploration & Technology Showcase 2023 Johannesburg Country Club, Auckland Park Website: https://www.gssa.org.za/wp-content/uploads/AESUpdated.pdf 15-16 November 2023 — MESA Year End International Summit Johannesburg, South Africa Contact: Gugu Charlie Website: https://www.mesa-africa.org/year-end-internationalsummit-successful-manufacturing-the-next-step/ 16-17 November 2023 — 6TH Young Professionals Conference 2023 Hatch Africa, Greenstone, South Africa Contact: Gugu Charlie Tel: 011 538-0238 E-mail: gugu@saimm.co.za Website: http://www.saimm.co.za 21-23 November 2023 — Critical Minerals Conference 2023 Perth, Austraia Website: https://www.ausimm.com/conferences-and-events/ critical-minerals/ 28 November 2023 — MineSafe Online Conference 2023 Contact: Camielah Jardine Tel: 011 538-0237 E-mail: camielah@saimm.co.za Website: http://www.saimm.co.za
2024 12-13 March 2024 — GMG Kiruna Forum | Tomorrow’s Mining: Innovating to Improve the Way We Mine Contact: Camielah Jardine Website: https://gmggroup.org/gmg-kiruna-forumtomorrows-mining-innovating-to-improve-the-way-we-mine/
The Journal of the Southern African Institute of Mining and Metallurgy
12-14 March 2024 — Southern African Pyrometallurgy 2024 International Conference Sustainable Pyrometallurgy - Surviving Today and Thriving Tomorrow Misty Hills Conference Centre, Johannesburg, South Africa Contact: Camielah Jardine Tel: 011 538-0237 E-mail: camielah@saimm.co.za Website: http://www.saimm.co.za 19-25 April 2024 — World Tunnel Congress 2024 Shenzhen, China Website: https://www.wtc2024.cn/ 21-23 May 2024 — The 11TH World Conference of Sampling and Blending 2024 Hybrid Conference Misty Hills Conference Centre, Johannesburg, South Africa Contact: Camielah Jardine Tel: 011 538-0237 E-mail: camielah@saimm.co.za Website: http://www.saimm.co.za 27-31 May 2024 — Nickel-Cobalt-Copper Lithium-Battery Technology-REE 2024 Conference and Exhibition Perth, Australia Website: https://www.altamet.com.au/conferences/alta-2024/ 11-13 June 2024 — 15TH International Conference on Industrial Applications of Computational Fluid Dynamics Trondhedim, Norway E-mail: Jan.E.Olsen@sintef.no Website: https://www.sintef.no/projectweb/cfd2024/ 18-20 June 2024 — Southern African Rare Earths 2ND International Conference 2024 Swakopmund Hotel and Entertainment Centre, Swakopmund, Namibia Contact: Camielah Jardine Tel: 011 538-0237 E-mail: camielah@saimm.co.za Website: http://www.saimm.co.za 3-5 July 2024 — 5TH School on Manganese Ferroalloy Production Decarbonization of the Manganese Ferroalloy Industry Boardwalk ICC, Gqeberha, Eastern Cape, South Africa Contact: Gugu Charlie Tel: 011 538-0238 E-mail: gugu@saimm.co.za Website: http://www.saimm.co.za 1-3 September 2024 — Hydrometallurgy Conference 2024 Hydrometallurgy for the Future Hazendal Wine Estate, Stellenbosch, Western Cape, South Africa Contact: Camielah Jardine Tel: 011 538-0237 E-mail: camielah@saimm.co.za Website: http://www.saimm.co.za
VOLUME 123
OCTOBER 2023
ix ◀
Company affiliates The following organizations have been admitted to the Institute as Company Affiliates 3M South Africa (Pty) Limited A and B Global Mining (Pty) Ltd acQuire Technology Solutions AECOM SA (Pty) Ltd AEL Mining Services Limited African Pegmatite (Pty) Ltd Air Liquide (Pty) Ltd Alexander Proudfoot Africa (Pty) Ltd Allied Furnace Consultants AMEC Foster Wheeler AMIRA International Africa (Pty) Ltd ANDRITZ Delkor(pty) Ltd Anglo Operations Proprietary Limited Anglogold Ashanti Ltd Anton Paar Southern Africa (Pty) Ltd Arcus Gibb (Pty) Ltd ASPASA Aurecon South Africa (Pty) Ltd Aveng Engineering Aveng Mining Shafts and Underground Axiom Chemlab Supplies (Pty) Ltd Axis House Pty Ltd Bafokeng Rasimone Platinum Mine Barloworld Equipment -Mining BASF Holdings SA (Pty) Ltd BCL Limited Becker Mining (Pty) Ltd BedRock Mining Support Pty Ltd BHP Billiton Energy Coal SA Ltd Blue Cube Systems (Pty) Ltd Bluhm Burton Engineering Pty Ltd Bond Equipment (Pty) Ltd Bouygues Travaux Publics Caledonia Mining South Africa Plc Castle Lead Works CDM Group CGG Services SA Coalmin Process Technologies CC Concor Opencast Mining Concor Technicrete Council for Geoscience Library CRONIMET Mining Processing SA Pty Ltd CSIR Natural Resources and the Environment (NRE) Data Mine SA DDP Specialty Products South Africa (Pty) Ltd Digby Wells and Associates DRA Mineral Projects (Pty) Ltd DTP Mining - Bouygues Construction Duraset ▶ x
OCTOBER 2023
EHL Consulting Engineers (Pty) Ltd Elbroc Mining Products (Pty) Ltd eThekwini Municipality Ex Mente Technologies (Pty) Ltd Expectra 2004 (Pty) Ltd Exxaro Coal (Pty) Ltd Exxaro Resources Limited Filtaquip (Pty) Ltd FLSmidth Minerals (Pty) Ltd Fluor Daniel SA ( Pty) Ltd Franki Africa (Pty) Ltd-JHB Fraser Alexander (Pty) Ltd G H H Mining Machines (Pty) Ltd Geobrugg Southern Africa (Pty) Ltd Glencore Gravitas Minerals (Pty) Ltd Hall Core Drilling (Pty) Ltd Hatch (Pty) Ltd Herrenknecht AG HPE Hydro Power Equipment (Pty) Ltd Huawei Technologies Africa (Pty) Ltd Immersive Technologies IMS Engineering (Pty) Ltd Ingwenya Mineral Processing (Pty) Ltd Ivanhoe Mines SA Kudumane Manganese Resources Leica Geosystems (Pty) Ltd Loesche South Africa (Pty) Ltd Longyear South Africa (Pty) Ltd Lull Storm Trading (Pty) Ltd Maccaferri SA (Pty) Ltd Magnetech (Pty) Ltd MAGOTTEAUX (Pty) Ltd Malvern Panalytical (Pty) Ltd Maptek (Pty) Ltd Maxam Dantex (Pty) Ltd MBE Minerals SA Pty Ltd MCC Contracts (Pty) Ltd MD Mineral Technologies SA (Pty) Ltd MDM Technical Africa (Pty) Ltd Metalock Engineering RSA (Pty)Ltd Metorex Limited Metso Minerals (South Africa) Pty Ltd Micromine Africa (Pty) Ltd MineARC South Africa (Pty) Ltd Minerals Council of South Africa Minerals Operations Executive (Pty) Ltd MineRP Holding (Pty) Ltd Mining Projections Concepts Mintek MIP Process Technologies (Pty) Limited MLB Investment CC VOLUME 123
Modular Mining Systems Africa (Pty) Ltd MSA Group (Pty) Ltd Multotec (Pty) Ltd Murray and Roberts Cementation Nalco Africa (Pty) Ltd Namakwa Sands(Pty) Ltd Ncamiso Trading (Pty) Ltd Northam Platinum Ltd - Zondereinde Opermin Operational Excellence OPTRON (Pty) Ltd Paterson & Cooke Consulting Engineers (Pty) Ltd Perkinelmer Polysius a Division of Thyssenkrupp Industrial Sol Precious Metals Refiners Rams Mining Technologies Rand Refinery Limited Redpath Mining (South Africa) (Pty) Ltd Rocbolt Technologies Rosond (Pty) Ltd Royal Bafokeng Platinum Roytec Global (Pty) Ltd RungePincockMinarco Limited Rustenburg Platinum Mines Limited Salene Mining (Pty) Ltd Sandvik Mining and Construction Delmas (Pty) Ltd Sandvik Mining and Construction RSA(Pty) Ltd SANIRE Schauenburg (Pty) Ltd Sebilo Resources (Pty) Ltd SENET (Pty) Ltd Senmin International (Pty) Ltd SISA Inspection (Pty) Ltd Smec South Africa Sound Mining Solution (Pty) Ltd SRK Consulting SA (Pty) Ltd Time Mining and Processing (Pty) Ltd Timrite Pty Ltd Tomra (Pty) Ltd Trace Element Analysis Laboratory Traka Africa (Pty) Ltd Trans-Caledon Tunnel Authority Administarator Ukwazi Mining Solutions (Pty) Ltd Umgeni Water Webber Wentzel Weir Minerals Africa Welding Alloys South Africa Worley
The Journal of the Southern African Institute of Mining and Metallurgy
Galvanize your studies The Hot Dip Galvanizers Association Southern Africa (HDGASA) will be visiting campuses across South Africa to interact with future mining engineers, mechanical engineers, and civil engineers as well as metallurgists about hot dip galvanized steel in 2024. Through the Galvanize Your Varsity events, students will learn about the hot dip galvanizing process, design, sustainability, performance, standards and specifications, and inspection. The program is starting out in 2024 with on-campus presentations and engagements by the HDGASA at events where students can comfortably interact with our specialist advisors. Educating specifiers and designers of the future is paramount to the HDGASA’s mission. The Hot Dip Galvanizers Association Southern Africa (HDGASA) is a non-profit trade organization committed to educating current and future engineers, owners, developers, fabricators, and specifiers about hot dip galvanizing for corrosion control. The HDGASA is dedicated not only to educating current specifiers of the galvanizing industry but also future members. One of the biggest market limitations to the specification of hot dip galvanized steel is ignorance. The majority of architecture and engineering students are only exposed to the fundamentals of galvanized steel while studying at varsity. However, in the real-world hot dip galvanizing is extensively used to combat corrosion of iron and steel.
Keep your eyes peeled for the ‘Galvanize Your varsitY’ event
2024!
HOT DIP GALVANIZERS ASSOCIATION SOUTHERN AFRICA
Bedfordview Office Park, Building 1, Ground Floor, 3 Riley Road, Germiston Tel: 011 456 7960 Email: hdgasa@icon.co.za Website: www.hdgasa.org.za
GALVANIZERS GAUTENG ARMCO GALVANIZERS – ISANDO +27 (0)11 974 8511 | mail@armco.co.za ARMCO GALVANIZERS – RANDFONTEIN +27 (0)11 693 5825 | mail@armco.co.za GALFERRO GALVANISERS +27 (0)11 817 3667 | leana@galferro.co.za LIANRU GALVANISERS cc +27 (0)11 814 3080 | calis@lantic.net PRO-TECH GALVANIZERS (PTY) LTD +27 (0)11 814 4292 | jonathan@protechgalvanizers.co.za TRANSVAAL GALVANISERS +27 (0)11 814 1113 | transgalv@transgalv.co.za WESTERN CAPE ADVANCED GALVANISING (PTY) LTD +27 (0)21 951 6242 | admin@advancedgalv.co.za GALVATECH (PTY) LTD +27 (0)21 951 1211 l info@galvatech.co.za SOUTH CAPE GALVANIZING (PTY) LTD +27 (0)44 884 0882 | johan@scgalv.co.za EASTERN CAPE GALVANISING TECHNIQUES cc +27 (0)41 486 1432 | galvtech@metalman.co.za MORHOT (PTY) LTD +27 (0)43 763 1143 | morhotgalv@gmail.com KWAZULU NATAL BAY GALVANIZERS cc +27 (0)35 751 1942 I jerry@baygalv.co.za DURBAN GALVANIZING (PTY) LTD – BRIAREDEEN BRANCH +27 (0)31 563 7032 | shereen@dbngalv.co.za DURBAN GALVANIZING (PTY) LTD – PHOENIX BRANCH +27 (0)31 500 1607 | reception@dbngalv.co.za PINETOWN GALVANIZING +27 (0)31 700 5599 | admin@pinetowngalvanizing.com KZN GALVANIZING +27 (0)69 335 5416 | info@kzngalvanizing.co.za
It’s a big milestone for a piece of rubber. But Linatex® is so much more than that. It’s unlike anything else, using a unique 95% natural latex formula that has made Linatex® the strongest rubber in mining for the last century – exactly as nature intended. So, here’s to 100 Years Strong, and the next 100 to come. Learn more at linatex100.weir Copyright© 2023, Linatex Limited. All rights reserved.