Future Steel Forum Brochure September 2019

Page 1

FUTURE STEEL FORUM 2019

� INDUSTRY 4.0 ARTICLES � SPEAKER BIOGRAPHIES � EXHIBITOR PROFILES � FULL PROGRAMME � FLOOR PLAN � A GOOD VIBE

25-26 SEPTEMBER • BUDAPEST • HUNGARY

www.futuresteelforum.com

Future Steel Forum Supplement September 2019

STEEL TIMES INTERNATIONAL – FUTURE STEEL FORUM - SEPTEMBER 2019 cover.indd 1

29/08/2019 10:27:42


A NEW STANDARD OF PLANT AUTOMATION

PROCESS-ORIENTED OPERATOR GUIDANCE Intelligent, process-oriented and situation-specific

individual context of activities. As a result, X-Pact®

operator guidance for high flexible production with

Process Guidance is the first platform, which takes

ever smaller batch sizes and increasingly wider pro-

the increasing degree of automation into account,

duction mixes without breaks in operation - these

it guides the operator through the process, inde-

are your goals.

pendent of his previous experience, and ensures

And we are your partner!

universal interfaces for the whole digital environ-

… X-Pact® Process guidance creates the necessary

ment across all products.

transparency for your plant and links knowledge

Even more: Seamless merging of automation levels

and data intelligently and systematically. Specific

and a open modular system platform allow extend-

dialogs for specific categories of users, for example

ibility by adding further modules and applications.

line operators, technologists, maintenance personnel, production managers, etc., are focused on their

SMS group GmbH Electrics / Automation Ivo-Beucker-Strasse 43 40237 Düsseldorf, Germany Phone: +49 211 881-5895 Fax: +49 211 881-775895 www.sms-group.com/x-pact-process-guidance

X-Pact Process Guidance_A4_e.indd 1

Leading partner in the world of metals

www.sms-group.com

09.07.19 08:15


FUTURE STEEL FORUM 2019

25-26 SEPTEMBER • BUDAPEST • HUNGARY

www.futuresteelforum.com

FUTURE STEEL FORUM 2019

� INDUSTRY 4.0 ARTICLES

Contents

� SPEAKER BIOGRAPHIES � EXHIBITOR PROFILES � FULL PROGRAMME � FLOOR PLAN � A GOOD VIBE

25-26 SEPTEMBER • BUDAPEST • HUNGARY

www.futuresteelforum.com

Future Steel Forum Supplement September 2019

STEEL TIMES INTERNATIONAL – FUTURE STEEL FORUM - SEPTEMBER 2019

2 Welcome by Matthew Moggridge 4 Future Steel Forum Conference Programme

KOCKS 100th RSB® currently operating at China.

EDITORIAL/PRODUCTION Editor / Programme Director Matthew Moggridge +44 1737 855151 matthewmoggridge@quartzltd.com Production Editor / Design Guru Annie Baker Advertisement Production Martin Lawrence

SALES International Sales Manager Paul Rossage +44 1737 855116 paulrossage@quartzltd.com Sales Director Ken Clark +44 1737 855117 kenclark@quartzltd.com

CORPORATE Managing Director Steve Diprose CEO Paul Michael

10 Speaker Biographies 22 Exhibitor Profiles 30 The Alpha Dog in the Human-AI Team 33 20 Dilemmas in Digital Transformation 36 Developing Trends in Metal Production Surface Inspection 40 Fog Computing, the Cloud and Cybersecurity 43 Data Analytics and Tata Steel Europe 46 Managing Change in the Digital Workplace 49 Automation Security in the Age of Industry 4.0 52 Designing Smart Networked Manufacturing Systems 54 Getting more out of Drones 60 A Data-Driven Galvanising Line

Published by: Quartz Business Media Ltd Quart House, 20 Clarendon Road Redhill, Surry RH1 1QX, UK +44 1737 855000 www.steeltimesint.com © Quartz Business Media, 2019

contents fsf.indd 3

64 Prediction of Coke Strength by Machine Learning 68 A New Standard in Plant Automation 72 Cybersecurity

FUTURE STEEL FORUM

1 29/08/2019 10:50:29


FUTURE

STEEL FORUM

25-26 SEPTEMBER • BUDAPEST • HUNGARY

2019

www.futuresteelforum.com

Welcome

Matthew Moggridge, Programme Director, Future Steel Forum

Does life imitate art or does art imitate life? It’s a question asked by many, but it’s a bit like the chicken and the egg: answers are thin on the ground. When I was a kid I had plenty of hard-backed science books with smooth, glossy pages about ‘the future’ jam-packed with artists’ impressions of what the world might look like. Weirdly, the worlds they depicted have been and gone, and now, in a sense, I’m living beyond the future with only distant memories of those images and how right or wrong they might have been. There were, of course, depictions of Concorde, which, sadly, is no more, although I remember watching it circle over London. Today I wonder why such a fantastic aircraft was terminated. The ‘car of the future’ is still a pipe dream, although many of those artists’ impressions certainly hold water. There are electric cars on the street, but the autonomous automobile, like those in the original Total Recall movie, has not quite materialised and I don’t think anybody relies upon the services of Doctor Smile, the suitcase psychiatrist, from Philip K Dick’s novel, UBIK. When, I wonder, will the drug squad don Bob Arctor’s ‘scramble suit’ from another of Dick’s great sci-fi novels, A Scanner Darkly? Life moves fast, said Alvin Toffler in Future Shock. HAL from Kubrick’s 2001: A Space Odyssey is now a reality, and the so-called ‘factory of the future’ has already landed, just visit Big River Steel in Osceola, Arkansas, for proof. Future Steel Forum 2019, here in Budapest, Hungary, is all about the future of steel manufacturing and the application of Industry 4.0 to the steelmaking process. It’s futuristic stuff and this year features, among other things, Fog computing, machine learning, the Internet of Things, drones, cybersecurity and other forms of extreme industrial hi-tech. I hope you enjoy the conference and this very special magazine.

2 welcome.indd 2

FUTURE STEEL FORUM

Steel Times International

29/08/2019 10:47:42


info@quinlogic.de ¡ quinlogic.com

Less internal downgrades Less customer claims Faster throughput

Offering The Most Suitable Dimensional Measuring Solutions Regardless of Shape, Size, Speed or Temperature, We Measure It All ! Static, Oscillating, Rotating, Full Profile Capture, We Do It All !

www.zumbach.com • sales@zumbach.ch


CONFERENCE PROGRAMME

FUTURE STEEL FORUM 2019

25-26 SEPTEMBER • BUDAPEST • HUNGARY

www.futuresteelforum.com

DAY ONE – WEDNESDAY 25 SEPTEMBER REGISTRATION AND WELCOME 0830

Welcome to Future Steel Forum

by Matthew Moggridge, Editor, Steel Times International.

0835

Opening Address: Chinese Steelmaking and Digitalisation under Industry 4.0 – Rockcheck Steel Group’s Perspective on Intelligent Manufacturing and Green Steel

by Catherine Zhang, CEO of Rockcheck Steel Group, China.

0855

Keynote: Case Study: Innovation Ecosystems – the Steel Block by Carlos Alba, Digital and TechnoEconomic global research programs Worldwide within ArcelorMIttal Global R&D.

SMART FACTORIES – SESSION CHAIRED BY JEAN-PAUL NAUZIN, VICE PRESIDENT – MARKETING & TECHNOLOGY, FIVES’ STEEL BUSINESS UNIT. 0905

0935

Building a Foundation for Smart Factories in the Steel Industry with Fog Computing, the Cloud, and Cybersecurity, by Dr. Lane Thames, Senior Security Researcher with Tripwire Inc’s Vulnerability and Exposure Research Team (VERT).

Designing Highly Dynamic Reconfigurable Manufacturing Systems by Dr. Jelena Milisavljevic, Lecturer in Industrial Design, University of Liverpool.

1005

Coffee Break and Exhibition Time

1035

The ‘Smartisation’ of the Ironmaking Process by Hee-Geun Lee, senior vice president, POSCO, Pohang Works, South Korea.

1105

4

Closing the Loop Beyond the Smart Factory by Dr. Andrew Zoryk, Global Metals Practice Lead, Accenture. (Could be an end session paper).

FUTURE STEEL FORUM

conference programme.indd 2

Steel Times International

28/08/2019 12:02:08


FUTURE

STEEL FORUM

25-26 SEPTEMBER • BUDAPEST • HUNGARY

2019

www.futuresteelforum.com

SUSTAINABILITY – SESSION CHAIRED BY CARLOS ALBA, GLOBAL RESEARCH PROGRAMME LEADER, ARCELORMITTAL, DIGITALISATION AND TECHNOECONOMICS 1135

Keynote: Future Green Steel by Clemens Schneider, Project Manager, Wuppertal Institute, Germany.

1205

Lunch Break and Exhibition Time

1335

Keynote – The World Steel Association’s Smart Steel Initiative by Henk Reimink, Director of Industry Excellence, World Steel Association.

SUPPLY CHAIN LOGISTICS – SESSION CHAIRED BY RAFFAEL BINDER, PSI METALS 1405

1435

1505

Strategic and Tactical Logistics Planning Optimisation by Diego Diaz Fidalgo, R&D Engineer, ArcelorMittal. Data Analytics at Work in Tata Steel by Dr. Svend Lassen, Head of Reporting and Analytics, Sales & Marketing and Supply Chain, Tata Steel Europe.

Holistic Logistics Solutions as a Major Driver to Optimise Productivity in a Steelworks by Mr. Markus Ringhofer, Sales Manager, Industry 4.0, Primetals Technologies.

1535

Tea Break and Exhibition Time

1605

Keynote: Opportunities and Tools in Steel Industry Logistics Chain Development by Tony Leikas, CEO, Pesmel.

1635

Opportunities and Tools in the Steel Industry Logistics Chain by Jose Favilla, Worldwide Executive Director & Partner, Industry Solutions, IBM

BUSINESS MODELS – PART ONE – SESSION CHAIRED BY KURT HERZOG, PRIMETALS 1705

1735

The Learning Steel Plant from SMS group by Dr.-Ing Markus Reifferscheid, SMS group.

From automated device collection to predictive maintenance by Steffen Ochsenreither and Jens Hundrieser, Endress & Hauser.

1805

Conference closes

1930

Networking Dinner

Steel Times International

conference programme.indd 3

FUTURE STEEL FORUM

5 28/08/2019 12:02:09


CONFERENCE PROGRAMME

FUTURE

STEEL FORUM

25-26 SEPTEMBER • BUDAPEST • HUNGARY

2019

www.futuresteelforum.com

DAY TWO – THURSDAY 26 SEPTEMBER 0830

Keynote: Artificial Intelligence: Enabling a Leap to Radical Efficiency in the Steel Industry.

By Stephen Pratt, Founder and CEO, Noodle AI.

ARTIFICIAL INTELLIGENCE – SESSION CHAIRED BY ROGER ANDERSSON, PHD. HEAD OF RESEARCH, HEAT TREATMENT AND METAL PROCESSING, SWERIM AB 0905

The Alpha Wolf in the Human AI Team

by Patrick Henz, Primetals USA, PTUS Head of Governance & Compliance.

0935 Prediction of Coke Strength After Reaction (CSR) Using Machine Learning in a Coke Plant

by Satish Agarwal, Analytics & Insights, Jamshedpur, Tata Steel India.

1005

Machine Learning Application for Furnace Process Control and Optimisation in Steelmaking

by Giovanni Bavestrelli, Digital Engineering Director, Tenova SpA.

1035

Coffee Break and Exhibition Time11

35 1105

Combined Physical Models and AI to optimise mechanical properties on a continuous galvanising line.

1135

CASE STUDY – Industry 4.0 – A Key Driver for Business Transformation, by Michael Walter, Luc Bongaerts and Paul Vanvuchelen, OM Partners

by Cyrill Peillon, Senior Data Scientist, Fives CortX, Fives Group, France

PERSPECTIVE ON DIGITALISATION AND STEEL MANUFACTURING 1205

6

Agility in Metals: How Digital Champions are Out-performing the Competition by Nils Naujok, Partner, Utilities & Resources (EUR) Leader EMA, PwC Strategy&, Germany, and Holger Stamm, Director PwC Strategy&, Germany

FUTURE STEEL FORUM

conference programme.indd 4

Steel Times International

28/08/2019 12:02:09


FUTURE

STEEL FORUM

25-26 SEPTEMBER • BUDAPEST • HUNGARY

2019

www.futuresteelforum.com

1235 Lunch Break and Exhibition Time

CYBERSECURITY – CHAIRED BY JELENA MILISAVLJEVIC-SYED, DIVISION OF INDUSTRIAL DESIGN, SCHOOL OF ENGINEERING, UNIVERSITY OF LIVERPOOL 1405

Keynote: Automation Security in the Era of Industry 4.0

by Marcus J Neuer, Head of Department, Downstream Automation, BFI.

1435

Industrial Cybersecurity – Closing the Gap between the needs of the European Steel Industry and the direction of research and development into the implementation of Industry 4.0

by Costanzo Pietrosanti, Chairman of the Smart Factory Focus Group at ESTEP

1505

Tea Break and Exhibition Time

BUSINESS MODELS – PART TWO – CHAIRED BY LUC BONGAERTS, BUSINESS DEVELOPMENT MANAGER, OM PARTNERS 1535

The Future of Production Management as the Fundament of the Digital Transformation

by Luc Van Nerom, PSI Metals.

1605

Forge Next Generation Enterprise Asset Management for the Steel Industry

by Carlos Lemos, Solution Manager, Asset Intelligence Network, SAP

1635

The Automation of the Future is now the reality in steelmaking

by Marco Ometto, Managing Director, Danieli Automation

INDUSTRY 4.0 AND THE WORKFORCE – CHAIRED BY DR. NILS NAUJOK, PARTNER, ENERGY, UTILITIES & RESOURCES (EUR) LEADER, EMEA, PWC STRATEGY& GMBH. 1705

Frontier Issues for Succeeding in a Digitised World

by Farrokh Mistree, LA Comp Chair and Professor, School of Aerospace and Mechanical Engineering, University of Oklahoma, USA.

1735 PANEL DISCUSSION: INDUSTRY 4.0 AND THE WORKFORCE OF TOMORROW CHAIRED, DR. NILS NAUJOK, PARTNER, ENERGY, UTILITIES & RESOURCES (EUR) � Kriistian Van Teutum, Vice President Marketing & Sales, Fives’ Steel Business unit and Sales Director of Fives Stein, Fives � Farrokh Mistree, LA Comp Chair and Professor, School of Aerospace and Mechanical Engineering, University of Oklahoma, USA.

� Eric Vitse, Chief Technology Officer, Liberty House Group � Jose Favilla, Worldwide Executive Director & Partner, Industry Solutions, IBM � Three more panellists needed.

1805

Closing remarks

*Please note: the conference programme is subject to change at the organiser’s discretion

Steel Times International

conference programme.indd 5

FUTURE STEEL FORUM

7 28/08/2019 12:02:09


EXHIBITOR LIST FLOOR PLAN

EXHIBITOR LIST A10

A.L.B.A. S.R.L.

A03

AMETEK LAND (LAND INSTRUMENTS INTERNATIONAL)

A03

AMETEK Surface Vision

A20

BM SPA

A08 ENDRESS+HAUSER A07 FIVES A16

ISRA PARSYTEC GMBH

A17

KLAVENESS DIGITAL AS

A11

MATERIALS PROCESSING INSTITUTE

A18

NT LIFTEC

A04

OM PARTNERS

A09

PRIMETALS TECHNOLOGIES AUSTRIA GMBH

A06

PSI METALS GMBH

A14

QUINLOGIC GMBH

A12 SAP A01/A02 SMS GROUP GMBH A19

SPRAYING SYSTEMS CO.

A15

SWERIM AB (SWEDISH RESEARCH INSTITUTE

FOR MINING, METALLURGY AND MATERIALS)

A05 TMEIC A13

8

XOM MATERIALS GMBH

FUTURE STEEL FORUM

FLOOR PLAN + EXHIBITORS LIST.indd 2

Steel Times International

28/08/2019 12:04:14


FUTURE

STEEL FORUM

25-26 SEPTEMBER • BUDAPEST • HUNGARY

2019

www.futuresteelforum.com

16

CONFERENCE ROOM

ENTRANCE

COFFEE

COFFEE

15

17

14

18

13

19

12

11

20

8

9

10

BUFFET

GUEST ELEVATOR BUFFET

7

1

REGISTRATION STAIRCASE

6

RESTROOMS

2 CLOAKROOM

Steel Times International

FLOOR PLAN + EXHIBITORS LIST.indd 3

5 3

FUTURE STEEL FORUM

4

9 28/08/2019 12:04:15


SPEAKER PROFILES

FUTURE STEEL FORUM 2019

25-26 SEPTEMBER • BUDAPEST • HUNGARY

www.futuresteelforum.com

Matthew Moggridge, Editor, Steel Times International Matthew Moggridge has been editor of Steel Times International since January 2014 having previously edited Aluminium International Today, both published by the UK-based Quartz Business Media. During his time on both titles he has travelled extensively around the world interviewing and writing about leading figures in the metals industry and covering international steel and aluminium conferences. In addition to working as a journalist in many different industrial sectors, he is also the creator and driving force behind the development of the Future Steel Forum. Matthew’s career as a business journalist has spanned many leading titles covering other industrial sectors including food processing, foodservice, foreign direct investment, bulk handling and transportation and computers.

Catherine Zhang, CEO at Rockcheck Steel, China Catherine graduated from the University of British Columbia with dual degree in economics and political science. In 2017,she started work in the family business as executive director,Tianjin ROCKCHECK Group and then CEO, Rockcheck Steel Group and director of Tianjin ROCKCHECK PUJI Foundation.

Carlos Alba, Chief Digital Officer, ArcelorMittal Carlos Alba is chief digital officer of ArcelorMittal Research. He leads the company’s digital strategy and its links with the business units. In the recent past he used to lead the digital and techno-economic global research programmes worldwide within Global R&D at ArcelorMittal and across divisions, regions and functions. Digital core technologies, such as artificial intelligence and mathematical optimisation, are systematically merged with the ArcelorMittal value chain and is advanced in other areas, such as the industrial internet of things (IIoT) and big data analytics, covering manufacturing (and mining), procurement, commercial, supply chain, logistics, finance, strategy and product development. He joined the ArcelorMittal Global Research & Development corporate team in 2007 and was focused on AI models applied to business optimisation company wide (Europe, Americas and ACIS).

Jean-Paul Nauzin, Vice President Marketing & Technology, Fives Steel Business Unit 2013 - present CEO, Fives KEODS (France) and automotive expert 2002-2013 Metallic material expert, PSA Peugeot Citroën (France) 1996-2001 Metallic material expert, ArcelorMittal (France)

10 speakers bio.indd 2

FUTURE STEEL FORUM

Steel Times International

29/08/2019 12:56:04


Dr. Lane Thames, Senior Security Researcher, Tripwire Inc’s Vulnerability & Exposure Research Team (VERT) Lane Thames is a senior security researcher with Tripwire’s Vulnerability and Exposure Research Team (VERT) where he develops vulnerability detection and management software. He also conducts cybersecurity research, investigates newly discovered vulnerabilities, and contributes to the Tripwire State of Security blog. Lane recently coedited a Springer book on cybersecurity for Industry 4.0, has numerous publications on topics such as cybersecurity, cloud-based design and manufacturing, and machine learning, and is a frequent speaker at cybersecurity conferences such as RSA, BSides, SecTor, and the Industrial Control Systems (ICS) Cyber Security Conference.

Dr. Jelena Milisavljevice-Syed, Division of Industrial Design at the School of Engineering, University of Liverpool Jelena Milisavljevic-Syed is a lecturer in industrial design at the University of Liverpool in the United Kingdom. Her research focus is on integrating design thinking, strategy and innovation management in the realisation of cyber-physical product-service systems that are adaptable to ambitious market demands as a support of further digitalisation (smart manufacturing). Jelena has co-authored one monograph, four journal publications and 20 technical publications dealing with dynamic management in the realisation of manufacturing systems, design, and analysis of complex systems, product development, CVT in wind turbines, and FEM analysis. She is a member of ASME, ASEE, and GWI.

Hee Geun, Lee, Senior Vice President Posco, Pohang Works, South Korea - Senior Vice President in POSCO Pohang Works (2017~) - General Manager of Ironmaking (2013~2017) - Technical Section Leader, Blast Furnace; Plant Manager, Blast Furnace; Plant Manager of Ironmaking (2009~2013)

Dr. Andrew Zoryk, Managing Director, Metals Practice, Deloitte Consulting Dr. Andrew Zoryk is part of Deloitte Consulting GmbH, working in the Strategy & Operations practice based in Munich, Germany. He started his career more than 30 years ago with British Steel and Corus in the UK and since then has worked with many leading steel, metals and mill products companies globally in the areas of supply chain, manufacturing, digitalisation and enterprise processes. Deloitte Touche Tohmatsu Limited (DTTL) is one of the largest private professional services networks in the world, recording revenues of $43.2 billion in fiscal year 2018. With approximately 286,000 people worldwide, DTTL member firms deliver services in audit, advisory, tax, and consulting in nearly 900 offices in more than 150 countries and territories.

Steel Times International

speakers bio.indd 3

FUTURE STEEL FORUM

11 29/08/2019 12:56:06


SPEAKER PROFILES

FUTURE

STEEL FORUM

25-26 SEPTEMBER • BUDAPEST • HUNGARY

2019

www.futuresteelforum.com

Clemens Schneider, Project Manager, Wuppertal Institute Clemens Schneider is a project co-ordinator at the Wuppertal Institute. His field of work revolves around system analysis and modelling in the context of energy and environmental research. He develops mid- and long term energy and greenhouse gas emission scenarios as well as roadmaps on different regional levels and for individual sectors of heavy industry. Another field of his work is the assessment of low carbon technologies in heavy industry. He has developed several energy system simulation models, e.g. the bottom-up multi-level energy system model WISEE, which has been used at the level of an industrial cluster (Port of Rotterdam), at regional, national and EU level. He studied political science, economic policy and public law at the University of MĂźnster.

Henk Reimink, Director, Industry Excellence, World Steel Association Henk joined the World Steel Association (worldsteel) in 2008 and was accountable for activities on safety and health, manufacturing processes and systems in the iron and steel industry value chain as well as climate change mitigation techniques and global regulatory overview. Prior to joining worldsteel he worked as coating manager for New Zealand Steel (part of BlueScope Steel) for six years. Over the past 30 years, his roles have covered many engineering projects and operations in hot and cold rolling, metal and organic coating. He joined the steel industry as a mechanical engineer in 1978, and his career has embraced some challenging roles in New Zealand, South East Asia, Australia and the USA. Between 1999 and 2002 he undertook a contract opportunity in the field of nickel and cobalt production and a telecoms service contract in Australia. He rejoined the steel industry in 2002.

Raffael Binder, Director, Marketing at PSI Metals Raffael Binder was born on 11 July 1980 in Gmunden, Austria, and finished his innovation and product management studies with his diploma thesis about lead user methodology in the field of applied research. The software developer worked for the Austrian Institute of Technology in Vienna before joining PSI Metals as a sales manager in 2009. After taking over management responsibility in 2012 as sales director for the division in Austria, he took over the marketing department in 2015. One of his first marketing tasks was to position PSI Metals within the new arena of Industry 4.0 and digitalisation and he did this by bringing together all the relevant experts from PSI and from outside the company.

Diego Diaz Fidalgo, Global R&D Senior Specialist at ArcelorMittal Diego has been a researcher at the Business & TechnoEconomic Department (KiN) of ArcelorMittal Global R&D since its inception in 2004. This is a corporate division that provides service to ArcelorMittal globally, including corporate teams. It is a multi-disciplinary team that brings advanced analytics and artificial intelligence to the business side of the company. Within this team and in collaboration with domain specialists in each field, Diego has developed solutions across the value chain of the steel industry, including line scheduling, internal and external logistics, yard management, strategy, purchasing and sales. His main focus is on mathematical optimisation, metaheuristics, and machine learning; and to a lesser degree on simulation, algorithmic game theory, and other fields of artificial intelligence.

12 speakers bio.indd 4

FUTURE STEEL FORUM

Steel Times International

29/08/2019 12:56:08


Dr. Svend Lassen, Head of Reporting and Analytics, Sales & Marketing and Supply Chain at Tata Steel Europe Svend Lassen is a business leader in the steel industry in the areas of global supply chain & digitalisation. He is a member of the group senior management of Tata Steel in Europe, responsible for reporting & analytics in the commercial and supply chain. Tata Steel is a diversified steel producer with global operations throughout the carbon steel and electrical steel value chains and is one of the leading European producers of flat steel products. Svend is managing several teams and initiatives on big data and data management, advanced analytics and connected planning. Until 2017, Svend was chief purchasing officer and managing director at KlĂśckner & Co SE, one of the largest producer-independent steel and metals distributors in the world.

Dr. Markus Ringhofer, Sales Manager, Industry 4.0, Primetals Technology Dr. Markus Ringhofer conducted his studies of industrial engineering in Austria, Sweden and The Netherlands. In 2011 he started his career at Primetals Technologies in Austria as sales manager for the market in South Korea, where he was also serving his company as an expatriate between 2014 and 2016. Since his return to Austria in 2016, he has been responsible as sales manager for the digitalisation portfolio of Primetals Technologies. Dr. Ringhofer is author of several publications in the fields of digitalisation and R&D management.

Tony Leikas, CEO, Pesmel With an engineering degree in machine automation, Tony Leikas has held various positions in the technology industry areas of engineering, sales and management. He has 20 years of working history in the Finnish company Pesmel and, since 2011, has held the position of CEO. His experience in sales combined with a firm technical background has given him a strong basis for leadership of the company. Mr. Leikas has been a member of the management group of Pesmel for over a decade. He is also chairman of the board of Pesmel Taiwan and other subsidiary companies. He is active in the Finnish business and technology fields as member of the board in the regional chamber of commerce and in the Federation of Finnish Technology Industries.

Jose Favilla, Worldwide Executive Director & Partner, Industry Solutions, IBM Jose Favilla is a world-wide executive director/partner at IBM. He has global responsibility for Industrial Solutions. Mr Favilla has over 30 years of experience in applying analytics and artificial intelligence to solve complex business problems for his clients in the metals industry. He is a member of IBM’s Industry Academy, a selected group recognised by its expertise and contribution to the industries it serves. He has a bachelorand a masters in electronics engineering and is concluding studies for doctorate in electronics engineering specialising in optimisation and artificial intelligence.

Steel Times International

speakers bio.indd 5

FUTURE STEEL FORUM

13 29/08/2019 12:56:13


SPEAKER PROFILES

FUTURE

STEEL FORUM

25-26 SEPTEMBER • BUDAPEST • HUNGARY

2019

www.futuresteelforum.com

Kurt Herzog, Head of Industry 4.0, Primetals Technologies Kurt Herzog studied control engineering at the Technical School Hollabrun and industrial automation at the Technical University in Vienna. He recently attended the Linz Management Academy (LIMAK) where he studied engineering management. He joined Siemens VAI in 1997 as an E&A project engineer and has progressed through the company, becoming head of process control systems for ironmaking, steelmaking and continuous casting and later becoming head of E&A product development, engineering and project execution for ironmaking, steelmaking and continuous casting. His present position is head of Industry 4.0 (electrics and automation).

Dr.-Ing Markus Reifferscheid, Senior Vice President, R&D at SMS Group Dr.-Ing. Markus Reifferscheid was born in Frankfurt/Main, Germany, 1967 and finished his studies of metallurgy at the Technical University of Clausthal. He completed his PhD at the Max-Planck Institute of Ferrous Metallurgy in Duesseldorf, Germany. Since 1996 he has worked for the SMS group. He has been senior vice president of research and development at the SMS group in Duesseldorf for more than four years. His main R&D topics are related to steelmaking, continuous casting and CSP technology.

Jens Hundrieser, Regional Industry Manager Europe (Metal), Endress + Hauser Jens Hundrieser is the Regional Industry Manager Europe (Metal) forEndress+Hauser Messtechnik GmbH+Co. KG He graduated in electrical engineering from the Bergische Universität Wuppertal in Germany and started working for Endress + Hauser in 1990 as a product manager.Since 1996 he has worked as an industry manager for various industries, including mining, and is currently responsible for energy (power and heat generation) and the metal (production) industries. Different segments of electricity generation (central and de-central) and nonferrous and ferrous segments, as well as recycling, play a major role. The main purpose of his position is to support Endress+Hauser’s international sales network with his industrial processes know-how and marketing expertise.

Stephen Pratt, Founder & CEO at Noodle AI Stephen is an instigator, agitator, and pioneer in creating world-class technology services organisations. He has spent his career building innovative ways to create value for the world's most important organisations. Prior to Noodle, he was responsible for all Watson implementations worldwide for IBM Global Business Services. He was also the founder and CEO of Infosys Consulting, a senior partner at Deloitte Consulting, and a technology and strategy consultant at Booz, Allen & Hamilton. He has been selected twice as one of the top 25 consultants in the world by Consulting Magazine and has a Bachelors and Masters degree in electrical engineering from Northwestern University and The George Washington University focused on satellite communications.

14 speakers bio.indd 6

FUTURE STEEL FORUM

Steel Times International

29/08/2019 12:56:17


SUPERIOR 4D EAGLE

® MASTERS OF PERFECT VISION The superior profile measurement system with rolling defect detection for long product rolling mills – powered by Automation W & R. www.kocks.de

KNOW-HOW FOR TOMORROW


SPEAKER PROFILES

FUTURE

STEEL FORUM

25-26 SEPTEMBER • BUDAPEST • HUNGARY

2019

www.futuresteelforum.com

Dr. Roger Andersson, Head of Research at Swerim AB, Sweden Roger Andersson, born on 27 April 1966 in Boden, Sweden, finished his PhD at the Department of Production Technology in 2005 following his MSc degree in material science at Luleå University of Technology. He joined the steel research institute Swerea MEFOS (now Swerim AB) in June 2016 and is now head of research (heat treatment and metal processing). Prior to joining what was Swerea MEFOS, Andersson was CEO at Duroc Special Steel, a rerolling company for flat products of special grades.

Patrick Henz, Head of Governance & Compliance, Primetals Technologies Patrick Henz started his career in the corporate information office and compliance at the end of 2007, when he was responsible for the implementation of the Siemens Anti-Corruption programme in Mexico and several Central American & Caribbean countries. Together with these tasks, he gained valuable insights into global compliance programs, with a focus on Latin America. Since 2009 in his role as compliance officer, he has been responsible for an effective compliance programme based on identification, protection, detection, response and recovery and combined with integrity, respect, passion and sustainability. With these means, Henz defines governance and compliance as pro-active functions establishing an equilibrium between law, processes and behaviourial science – including artificial intelligence.

Satish Agarwal, Analytics Solution Manager, Tata Steel Satish Agarwal has been analytics solution manager at Tata Steel’s Analytics and Insight, Centre of Excellence in India since its inception in 2016. The Analytics and Insight Centre of Excellence is a corporate division of Tata Steel and caters to all disciplines of advanced analytics across the value chain of TATA STEEL. Within this team and in consultation with domain specialists across the globe in each area, Satish has developed solutions for steelmaking and ironmaking as well as network optimisation, sales and cash forecasting and so on. Satish manages data science projects from end-to-end delivery embracing context setting, problem understanding, conceptual solution design, solution development and deployment of models. His main focus in on machine learning, mathematical optimisation and metaheuristics and, to some extent, in the fields of simulation and AI.

Giovanni Bavestrelli, Digital Engineering Director at Tenova Giovanni Bavestrelli graduated from the Politecnico di Milano, Italy, with a Master’s Degree in software engineering, after attending high school in Johannesburg, South Africa, for five years. He joined Pomini Tenova in 1994, working on roll grinder automation, developing and commissioning software systems for running roll shops in steel plants. He later joined Unisys Corporation in the year 2000, leading the development team for the Hermes editorial system. He went back to Pomini Tenova in 2004 as software engineering director and has since worked for Pomini Tenova, leading the software development team. In 2016 he contributed to Tenova’s Digital Transformation initiative and later joined the new Tenova Digital Team as Director.

16 speakers bio.indd 8

FUTURE STEEL FORUM

Steel Times International

29/08/2019 12:56:19



SPEAKER PROFILES

FUTURE

STEEL FORUM

25-26 SEPTEMBER • BUDAPEST • HUNGARY

2019

www.futuresteelforum.com

Cyril Peillon, Senior Data Scientist, FivesCortX, Fives Group Cyril Peillon is a senior data scientist within Fives CortX. He is a business-oriented data professional who specialises in asset-intensive industries. Within Fives CortX he oversees the development of predictive quality products. Before joining Fives CortX, Cyril worked for Veolia Water in China for six years developing asset management models to optimise capital investments and operational excellence for drinking water networks. He then worked in the Netherlands on change management projects as a certified project manager in various asset-intensive industries such as utilities, oil and gas and chemical in addition to working as a data science manager, building a team of specialists and a sustainability-related product for a start-up.

Dr. Luc Bongaerts, Business Development Manager, OM Partners, Belgium Luc has been business development manager at OM Partners since 2009. He has a PhD in mechanical engineering, specialising in the integration of scheduling and shop floor control of holonic manufacturing systems, which is particularly relevant for Industry 4.0. Holonic manufacturing was part of the Intelligent Manufacturing Initiative that was focused on autonomous and co-operating agents organising themselves to form agile, adaptive and high-performance production systems for the 21st century. He is active in supply chain management and his experience includes several SCM projects, focusing on delivering true value through the integrated supply chain.

Dr Nils Naujok, Partner, Consulting Leader, Metals Industries Europe at PwC Strategy& Germany Nils Naujok is EMEA metals consulting leader and partner with PwC Strategy&, which reinvents strategy consulting as the world’s leading ‘strategy-through-execution’ firm. Based in Berlin, he specialises in strategy development, operating model development, operations and innovation strategies for the metals and process industries. He is the leader of PwC’s EMEA steel and metals consulting practice and of Strategy&’s Innovation and Development Excellence practice and is in close contact with European steel and metals industry leaders, industry associations, media representatives and technology partners. A key focus for Nils is Industry 4.0 and the metals industry.

Holger Stamm, Director at PwC Strategy& Germany Director at PwC Strategy& Germany, Holger is co-leading the EMEA metals team of PwC. Based in Düsseldorf, he specialises in operating model development, digital and M&A strategies in the metals and process industries. Holger is responsible for digital and Industry 4.0 solutions and services and is in close contact with leaders of the European steel and metals industry, companies along the steel value chain and technology partners. He supports his clients in the needed business and organisational transformation to digital operations and digital business models for more agility, efficiency, digital capabilities and forward integration. Holger has more than 20 years of experience enabling his clients to deliver substantial benefit by improving their business performance.

18 speakers bio.indd 10

FUTURE STEEL FORUM

Steel Times International

29/08/2019 12:56:22


Marcus J Neuer, Head of Department, Downstream Automation, BFI ESTEP Marcus J. Neuer was born in Germany in 1977. He received the Dipl.-Phys. degree, with distinction, from the Heinrich-Heine-University Düsseldorf in 2003. From 2003 to 2006 he was researcher at the Heinrich-HeineUniversity in different educational and scientific roles, where he received the degree Dr. rer. nat. in 2006, following a Ph.D. thesis on a stochastic theory about anomalous transport in plasmas. Since 2012, he has worked for VDeh Betriebsforschungsinstitut GmbH (BFI), initially as project manager for algorithms and artificial intelligence. Marcus has more than 15 years experience in software architecture and development. His interests include multi-agent systems, holonic manufacturing and Monte-Carlo simulations.

Costanzo Pietrosanti, Chairman at Smart Factory Focus Group, ESTEP Costanzo Pietrosanti graduated with distinction in mechanical engineering in 1978 from the University of Rome “La Sapienza,” remaining for one year as assistant professor in machine design. In June 1979 he joined Centro Sviluppo Materiali SpA (CSM), the Italian R&D centre for the steel sector. From 1983 to 1987 he moved to the aerospace industry, coming back to CSM (now RINA-CSM) in 1987, spending 28 more years in steel processing R&D, technology transfer and application engineering. In 1992 he become department director for process modelling and simulation, automation and process control, ICT architectures and system integration for steel manufacturing, acquiring deep experience of Industry 4.0. He retired in 2015 and is now contract professor at the University of Rome “La Sapienza” leading the Laboratory of Digital Technology for Mechanical Engineering.

Luc Van Nerom, Deputy Managing Director, PSI Metals Luc Van Nerom has a master's degree in mathematics and computer science from the University of Brussels (VUB). After his studies, he continued at VUB as researcher at the laboratory of artificial intelligence focusing on knowledge-based decision systems. In 1986 he co-founded AIS to bring artificial intelligence and mathematical modelling to the industrial world. Soon the first solutions (caster and HSM scheduling) became a reality. The product was extended to a complete planning and scheduling offering called SteelPlanner. Today AIS and its products are fully embedded into PSI Metals. After AIS merged into PSI Metals, Luc became responsible for product architecture, product modelling and product innovation. come together.”

Carlos Lemos, Solution Manager, Asset Intelligence Network, SAP Carlos has been working with SAP Asset Management for 19 years with repeated international consulting experience on the entire asset management lifecycle, from planning and building to commissioning, operating, maintaining and decommissioning assets within asset-intensive industries more focused on utilities and oil & gas, playing leading roles in some of world largest SAP EAM programmes worldwide. Over recent years, Carlos has been working closely with asset managers worldwide in helping to position the best solutions, design their entire asset lifecycle management processes and planning the transformation roadmap towards operation excellence. He is currently responsible for the definition, strategy and roll-out of SAP Asset Intelligence Network solution (SAP AIN).

Steel Times International

speakers bio.indd 11

FUTURE STEEL FORUM

19 29/08/2019 12:56:26


We understand how it is important to e�ciently extract and process precious metals and minerals.

PROCESS + PROTECT You save valuable resources while keeping your employees and the environment safe.

Endress+Hauser helps you to improve your processes: • With process experts who recommend the best-fit products, services and solutions according to industry requirements • With solutions that mitigate risk and reduce your environmental impact • With access to the right data at the right time

Do you want to learn more? www.endress.com/primaries-metal


FUTURE

STEEL FORUM

25-26 SEPTEMBER • BUDAPEST • HUNGARY

2019

www.futuresteelforum.com

Marco Ometto, Managing Director, Danieli Automation Marco Ometto DANIELI Automation in 1994 as project leader for the development of manufacturing execution systems (MES) in the metals industry. As manager, he had 13 engineers working with on analysis, development and commissioning of MET systems. He was appointed manager of process control systems in 1998 and by 2001 he had been appointed manager of automation systems design and development for electric meltshops and CCMs for long and flat steel products. Two years later he was appointed executive manager with the task of studying the expansion of the company abroad. He is currently executive vice president of Danieli Automation.

Farrokh Mistree, LA Comp Chair & Professor at School of Aerospace and Mechanical Engineering, University of Oklahoma Farrokh Mistree holds the L A Comp Chair in the School of Aerospace and Mechanical Engineering at the University of Oklahoma in Norman, Oklahoma, USA. Professor Janet K. Allen and Farrokh co-direct the Systems Realisation Laboratory at OU. Their research focus is on defining the emerging frontier for the “intelligent” decisionbased realisation of complex (cyber-physical-social) systems when the computational models are incomplete and inaccurate. Farrokh has co-authored two textbooks, two monographs and more than 400 technical papers dealing with the design of materials, mechanical, thermal and structural systems; ships and aircraft; engineered supply networks. Farrokh is a Fellow of ASME and an Associate Fellow of AIAA.

Kristiian Van Teutum, Vice President, Sales & Marketing, Fives Group Kristiaan is responsible for sales and marketing of Fives’ Steel business line, which covers engineering, process expertise, strip processing line and reheating furnace design and supply, including mechanical equipment, thermal technologies and induction heating solutions. Graduating as a mechanical engineer in 1988 from Portsmouth University (UK), he spent five years in the UK in technical sales then moved to Italy where over 20 years he held executive commercial roles in two multi-national engineering companies related to the steel sector for both long and flat products before joining Fives in Paris.

Eric Vitse, Chief Technology Officer, Liberty House Group As chief technology officer of the Liberty House Group, Eric is leading the company’s move towards clean, lowcarbon methods across all of its industrial operations, from mining and energy through to high-value engineered products. He was previously chief technology officer for Erdemir, Europe's third largest steel producer, after a threedecade career with ArcelorMittal. As chief technology officer for ArcelorMittal in the Americas he was responsible for capital investment strategy and technical oversight across more than 40 steel-making facilities with a combined capacity of over 40Mt/yr. At Erdemir, Eric was responsible for their engineering transformation programme.

Steel Times International

speakers bio.indd 13

FUTURE STEEL FORUM

21 29/08/2019 12:56:29


EXHIBITOR PROFILES

A.L.B.A S.R.L Cutting Technology Stand A10

AMETEK Land

Phone: +39 3346883938 Email: e.dottavi@albacut.com Website: http://www.albacut.com/

Phone: +44 1246 417691 Email: land.enquiry@ametek.com Website: https://www.ametek-land.com/

ALBA GROUP is an organised group of four companies able to satisfy any customer requests in the steelmaking industry, in the field of torch cutting, gas process and special equipment. Design, purchase of raw materials, completely in-house manufacturing, assembly, testing, distribution and service operations are directly managed. ALBA. is the family-owned head company of the group. Based in Genova, Italy, and established in 1956, it is now the worldwide leading manufacturer of torch cutting systems and special equipment for the steelmaking industry. ALBA Meccanica specialises in on-drawing steel constructions as well as complete machinery assemblies. ALBACUT Korea is the service centre and distributorship for east Asia, and ATES is the internal unit responsible for automation and service. The range of products and services available includes but is not limited to: • complete torch cutting machines for continuous casting of billets, blooms and slabs • deburring systems for billets, blooms and slabs • special gas cutting machines • gas control stations • heating/drying equipment • oxygen lances • gas burners • torches and nozzles • safety valves, flash back arrestors and regulators • on-drawing steel constructions and machinery assemblies ALBACUT4.0 is the technology that brings automatic torch cutting into a new era, responding to the newest demands of smart factories. Thanks to our HW&SW package the Torch Cutting systems will have a solution for predictive maintenance and machine management that will allow your plant to reduce downtimes, increase productivity and minimise costs.

Stand A03

AMETEK Land (Land Instruments) is a world-leading manufacturer of monitors and analysers for industrial infrared non-contact temperature measurement, combustion efficiency and environmental pollutant emissions. Through its trusted range of leading-edge technologies, the company is chosen the world over to deliver highly accurate measurement solutions that precisely meet every customer’s process needs. With unrivalled applications knowledge, choosing AMETEK Land ensures that the highest standards of process safety, process control and product quality are reached. Successful steel production requires accurate measurements across a wide range of temperatures and under a variety of different conditions. AMETEK Land provides comprehensive temperature measurement solutions supported by more than 70 years’ experience serving the steel industry. The company offers dedicated solutions for key applications and flexible instrumentation that can be customised for specific processes. The products support higher quality, lower costs, and greater safety. Built to operate in the harsh environments found in steel production, these instruments are designed to the highest performance standards, optimised for making temperature measurements at every important stage of the process. AMETEK Land’s bestselling SPOT pyrometer is a fully featured, high performance solution for fixed non-contact infrared spot temperature monitoring in steel production. SPOT offers powerful processing, communications and control functions that deliver accurate single spot measurements which users depend on to optimise their application processes; helping them maintain high product quality and protect against costly process inefficiencies. AMETEK Land is part of the Process & Analytical Instruments Division of AMETEK, Inc., a global supplier of high-end analytical instrumentation.

AMETEK Surface Vision

Stand A03

Phone: +1 (510) 431-6767 Email: surfacevision.info@ametek.com Website: https://www.ameteksurfacevision.com

AMETEK Surface Vision systems have become vital to increasing

22 profiles exhibitors.indd 2

FUTURE STEEL FORUM

Steel Times International

29/08/2019 12:58:26


efficiency, streamlining operations, improving product quality and reducing costs and waste in industrial processes. Manufacturers in the metals, paper, plastics and non-woven industries rely on the company’s solutions to detect surface flaws or defects at their production facilities across the globe. The company continues to innovate, providing cutting-edge technologies and world-class technical support that delivers highly accurate defect data, high-definition video, intelligent grading, archiving and detailed reporting. Customers who use AMETEK Surface Vision’s services get the benefits of: • Reduced operational costs • Process optimisation • Improved product quality • Maximised yield • More thorough and objective grading of material • Detection, classification and visualisation of defects • Minimised need for manual inspections • Inspection reports Headquartered in Hayward, California, the company supports customers worldwide through its network of regional centres offering expert services including application laboratories, technical support, service teams and training. The company is part of the Process and Analytical Instruments Division of AMETEK Inc., a leading global manufacturer of electronic instruments and electromechanical devices.

BM SPA

Stand A20

Email: sales@bmgroup.com Website: http://www.bmgroup.com/

BM Group, integrated solutions provider towards STEEL MILLS 4.0 BM Group is an Italian industrial concern, operating worldwide as a supplier of process automation equipment and customised robotic solutions for industry. PolytecRobotics is a BM Group brand started in 2012. Thanks to know-how acquired in the steel sector as a supplier of advanced automation systems, Polytec-Robotics is focused on the production of highly technological robotic cells for the steel and tube & pipe sectors. Deep knowledge of the steelmaking process together with a constant investment in R&D have been key to understanding the increasing needs of steelmaker. In few years, Polytec has developed a range of more than15 robotic cells that integrate the steelmaking process, from the furnace to the finishing mills, long and flat as well as tube and pipe. Each solution is manufactured according to customer requirements, tested in Polytec’s workshop and rapidly installed in new steel plants, as well as in existing ones. Advanced machine vision systems collect data and verify product quality and process performances. Safety, quality and productivity are the goals for the steelmakers who want to upgrade to 4.0 and invest in robotics. Reliability is what Polytec can offer, through a process designed beside the customer, from feasibility to assistance.

Steel Times International

profiles exhibitors.indd 3

Solutions: Furnace temperature and sample robots, obstruction opening robots, multi-tool robot for continuous casting machines, marking and tagging robots, Pre-screwing and screwing robots, Caps application robots, cleaning robots, technical assistance and spare parts. Process Automation – Power systems and drives – plant modernisation and digitisation – product tracking systems – antisway technology for cranes – scrap yard management systems consumables

Endress+Hauser

Stand A08

Phone: +49 7621 975 935 Email: jens.hundrieser@de.endress.com Website: http://www.de.endress.com/de

Endress+Hauser is a leading supplier of products, solutions and services for industrial process measurement and automation. It offers comprehensive process solutions for flow, level, pressure, analysis, temperature, recording and digital communications across a wide range of industries, optimising processes with regards to economic efficiency, safety and environmental protection. The main focus of Endress + Hauser is to focus on seven strategic industries, one of them is the metals industry. With more than 60 years of experience, the company helps its customers to improve process efficiency, cost savings, plant safety and sustainability. Based on the company’s expertise and its complete portfolio, its aim is to find the best solution for its customers. As one of the most innovative companies in the field of measurement instrumentation – Endress + Hauser owns more than 6,500 patents and spends approximately 7% of sales on R&D – it is also at the forefront of the digitalisation trend. To enable its customers to take the first step towards Industry 4.0, it supports them along three axes: • Along the value creation chain • From the field to the control level • From planning to maintenance Take digital communication, for example. It enables advanced measurement sensor diagnostics which can form the basis of effective process condition monitoring and preventative maintenance measures or calibration requests which can be triggered in the ERP system. Many of Endress+Hauser’s smart measurement sensors can be used to monitor process condition and verify measurement integrity. But that’s only one example of how an industry 4.0 approach can improve everyday business. To find out more about Endress + Hauser, just ask our colleagues on Booth A10.

FUTURE STEEL FORUM

23 29/08/2019 12:58:27


EXHIBITOR PROFILES

Fives Group

A07

Phone: +33 1 45 18 65 35 Email: steel@fivesgroup.com Website: http://www.fivesgroup.com/

Fives is an international industrial engineering group whose activities in the steel industry date back to 1867. Today, Fives’ steel business line provides steelmakers with turnkey solutions: process expertise and technologies for the carbon, stainless and silicon steel sectors in long and flat products, as well as tube and pipe, including strip processing, reheating and rolling. As a provider of turnkey solutions and a full range of equipment, Fives offers: • Metallurgical intelligence and competitive strategy development • Design and supply of strip processing lines and reheating furnaces • Thermal and mechanical technologies and proprietary equipment • Automation systems • Smart maintenance Full range of services include: cost optimisation, upgrade, repairs, training, assistance and feasibility studies. Since its foundation, Fives has always put innovation at the core of its development strategy, by investing in R&D to design and create pioneering technologies that meet the performance requirements of industrial companies in a broad range of sectors. Fives has also adopted a collaborative approach, built on partnerships and joint ventures with public and private players (start-ups, major groups, laboratories and research bodies and universities) to contribute to building the factories of the future. Today’s plants are becoming ‘smart’ and more agile. Digital technologies are powerful tools to improve operational efficiency. Fives combines its process expertise with digital tools to offer industrial companies solutions that facilitate production system management and maintenance: data and flow management, modelling and simulation of production line equipment, digital control and robotisation, and smart maintenance.

Isra Vision Parsytec AG

Stand A16

Tel.: +49 (2408) 92 700 - 0 Fax: +49 (2408) 92 700 - 500 info@isra-parsytec.com www.parsytec.co

ISRA VISION Parsytec is a leading global supplier of visual quality inspection technologies for non-ferrous strip and aluminium production. Based in Aachen, Germany, the company is a subsidiary of ISRA VISION AG, a market leader in surface inspection and machine vision. Industry leaders and globally active producers rely on ISRA VISION Parsytec’s solutions. ISRA VISION Parsytec enables comprehensive quality and process monitoring by combining highperformance inspection systems with a higher-level database and software architecture. The company’s solutions cover each and every step of the processing chain – from the casting of slabs to hot rolling, cold rolling as well as coating and cutting. Every innovative solution from ISRA VISION Parsytec is fuelled by the experience gained from over 750 installations in the metal industry. The system solutions are suitable for all metals and are already being put to successful use in the inspection of steel, aluminium, titanium or copper all around the world. Thanks to sophisticated algorithms, the systems allow for 2D and 3D inspection of slabs and heavy plates, resulting in valuable object and quality data. This data can be used in individual data reports to monitor and optimise production processes and facilitate decision-making from the line to the top management level. ISRA VISION Parsytec is a leading supplier globally of surface inspection systems in the metals industry; today, 18 of the top 20 steel producers employ the company’s products. The ISRA VISION Group is the world’s largest vendor of surface inspection systems. The systems are used in the glass, paper, banknote paper, plastics, non-woven fabrics, printing, photovoltaic and metalworking industries. The company operates globally and is headquartered in Germany (Darmstadt, Herten and Aachen) and has 25 additional subsidiaries worldwide

KLAVENESS DIGITAL AS STAND A17

Email: daniel@klavenessdigital.com Web: www.cargovalue.com Web: www.klavenessdigital.com LinkedIn: https://www.linkedin.com/company/klavenessdigital/

Klaveness Digital is a Norwegian technology company on a mission to bring shipping and logistics into the digital age. With a team of software engineers, data scientists, shipping and logistics specialists in Oslo and Singapore, the company is facilitating digital transformation and empowering companies to make informed decisions. On the back of its 70-year old heritage from Torvald Klaveness, Klaveness Digital is uniquely positioned to help organisations achieve

24 profiles exhibitors.indd 4

FUTURE STEEL FORUM

Steel Times International

29/08/2019 12:58:28


real-time and predictive logistics using the latest advances in machine learning and artificial intelligence. The company’s team is made up of distinct profiles with experience from the global shipping industry, tech community and start-up sphere, all working together to change a traditional industry driven by email and spreadsheets to one driven by real-time data, actionable insight and increased collaboration. The inspiration for CargoValue, a Klaveness Digital offering, came from the company’s long-term service to its industrial customers. Torvald Klaveness has delivered efficient and reliable transportation services to industrial cargo customers for decades, helping them reduce cost and risk related to both shipping and inventory. Today, many industrial companies rely heavily on manual collection of information and data entry in spreadsheets to stay updated on their logistics, and collaboration between stakeholders is done via e-mail and phone. These manual processes have made it challenging to quickly realign plans to manage unexpected supply disruptions, often leading to costly events such as demurrage, higher freight cost and production downtime. Since CargoValue’s development started, Klaveness Digital has worked closely with leading industrial companies in the metals and mining, aluminium, agricultural and energy industries. Together with its customers, the company has identified significant value creation opportunities in digitising and automating manual processes related to sourcing, shipping and inventory management, ultimately increasing supply chain transparency and collaboration. Today, the company is responsible for a portfolio of products and services. With valuable customer feedback, the company is able to continually improve its products to make sure they stay ahead in a constantly changing marketplace.

Materials Processing Institute Stand A11

Phone: +44 (0)1642 382000 Email: enquiries@mpiuk.com Website: https://www.mpiuk.com/

The Materials Processing Institute is a research and innovation centre serving organisations that work in advanced materials, low carbon energy and the circular economy. The Institute provides technology and R&D based services and consultancy. Scientists and engineers apply their expertise to progress innovation, develop materials and improve products and processes. The Institute is equipped with state-of-the-art steel making and refining equipment together with ingot and continuous casting facilities. The Institute has continued to be at the centre of innovation and new product development for over 70 years with many modern high performance steels and their process routes being developed in the laboratories and the Normanton Plant before mass production at the larger steel manufacturing sites.

Steel Times International

profiles exhibitors.indd 5

The Materials Processing Institute possesses a 7T pilot facility capable of producing plain carbon and alloy steel types by continuous casting and ingot casting. The plant is used primarily for development of new steel alloys, but also supplies highly specialist steels. This facility is being reconfigured as a ‘Future Steel’ plant, providing a test bed for industry 4.0 technologies to be developed. At the Institute a four-stage model has been developed and successfully applied for full application of Industry 4.0 technologies. The four stages are: measurement, monitoring, expert system and closed loop control. At the conference the approach being taken at the Institute for development and integration of Industry 4.0 will be outlined, with emphasis on the routes through the pilot and demonstration scale, for safe and reliable integration into steel plants.

NT Liftec

Stand A18

Phone: +358 3 3140 1400 Email: sales@ntliftec.com Website: https://www.ntliftec.com/

Your logistics solution provider NT Liftec Oy, previously TTS Liftec Oy, was established in 1991 and is located in Pirkkala, Finland. NT Liftec Oy designs and delivers cost-efficient horizontal transportation systems for ports and heavy industry. The company’s portfolio includes systems for containers and cargo cassette handling along with extensive service and after sales capabilities. The core of productivity of Liftec products is the Liftec Cassette system: Liftec translifters are transporting cargo cassettes with improved safety, cost-effectiveness and transportation efficiency. NT Liftec Oy is a market leader and is widely known as an innovative and reliable supplier of high quality products and solutions. As a development-oriented company, it is driven by customer needs, offering systems and services that are designed to maximise the profitability of customers’ and cargo owners’ businesses.

OM Partners Stand A04

Phone: +32 3 650 2211 Email: LBongaerts@ompartners.com Website: https://ompartners.com/en/solutions/omp-for-metals

OM Partners is a software and consulting company delivering Supply Chain Planning Solutions for Mill Products (metals, paper and packaging, floor covering...) and Semi Process industries (chemicals, pharmaceuticals, consumer products...). With more than 250 customers and 650 implementations, OM Partners has established solid partnerships with customers all over

FUTURE STEEL FORUM

25 29/08/2019 12:58:29


EXHIBITOR PROFILES

the world. With annual group sales revenues over 46,1 million EUR and a workforce of more than 400 people in our offices in Antwerp, Atlanta, Shanghai, Dubai, Sao Paulo, Paris, Rotterdam, Cologne and London, the company has become a top player in the supply chain planning market. OM Partners’ product OMP Plus is an integrated solution for all planning related issues, from the strategic down to the operational level. It is aimed at reducing logistic costs and throughput times and at increasing reliability of delivery dates and customer satisfaction. The revolutionary technology of OMP Plus makes integrated demand planning, supply planning and scheduling a reality.

Primetals Technologies

Stand A09

Email: contact@primetals.com Website: http://primetals.com/en/Pages/Home.aspx

Primetals Technologies, Limited, headquartered in London, UK, is a worldwide leading engineering, plant-building and lifecycle partner for the metals industry. The company offers a complete technology, product and service portfolio that includes the integrated electrics, automation and environmental solutions. This covers every step of the iron and steel production chain that extends from the raw materials to the finished product – in addition to the latest rolling solutions for the nonferrous metals sector. As Primetals Technologies has been a provider of automation solutions of all levels to steel producers for decades, the digitalization of the metals industry has been one of the company’s main focus areas for a long time. Primetals Technologies is a joint venture of Siemens, Mitsubishi Heavy Industries (MHI) and partners.

PSI METALS Stand A06

Phone: +49 211 60219-271 Email: info@psimetals.com Website: www.psimetals.com

PSI is the leading partner for digital production in the metals industry combining SCM, APS and MES within one software platform – PSImetals. Our software solutions enable producers of aluminium and steel products to ensure their competitive edge by delivering products as agreed in quantity, quality and time whilst considering inventory, productivity and performance targets. The PSImetals software line is an end-to-end approach for the overall supply chain caring for all the needs of the primary metals industry. From supplier to customer, PSImetals offers powerful and

26 profiles exhibitors.indd 6

FUTURE STEEL FORUM

highly configurable standard products to support all processes from planning to execution while respecting the complexity of metal production: • Planning level to support all planning processes from Business Planning via Production Planning to Detailed Scheduling, • Execution level to monitor and control production activities as well as to assure quality, • Level of material- and transport logistic to optimise all transports requested to keep production running, • Energy management level, • Cross-application KPI and production monitoring functions. All information is based on PSImetals Factory Model - a Digital Twin of the whole supply chain providing consistent real time plant status information. As market leader PSI claims technology leadership as well. Therefore PSImetals FutureLab investigates and develops the solutions of tomorrow taking into consideration • latest developments around Industry 4.0 • a collaborative approach with customers, partners and experts • leading edge IT technology based on PSI Java Framework. Combining 45 years of experience in implementing production management software with innovativeness, PSI supports numerous metals producers around the globe in achieving their competitive edge.

Quinlogic GmbH

Stand A14

Phone: +49 (2405) 47 999 40 Email: info@quinlogic.de Website: https://www.quinlogic.com

Quinlogic was established in Aachen in 2008 with the aim of harnessing large quantities of complex measurement data for quality management in the steel and aluminium industries so as to facilitate significant increases in production efficiency. QuinLogic’s customers include a large number of well-known steel rolling mills throughout the world. The SMS group has been a majority shareholder in the company since 2016. The outstanding and easy-to-use technology of the QES – Quality Assurance System – is increasingly applied in premium steel rolling mills as well as developing into the standard for aluminium flat rolling plants. Industry 4.0 QES – A pragmatic step towards Industry 4.0 The discussion about the future transformation of industrial production processes is ongoing. This “4th industrial revolution” does not have Steel Times International

29/08/2019 12:58:30


one standardised definition, but there are major criteria to describe it: • Better use of data • Merge production processes with information technology • Real-time availability of relevant information • Mass customisation • Technical decision assistance • ...finally it is an application that reveals know-how, preserves it and makes it available to anybody in the plant 24/7 • Link the real production world with the virtual world QuinLogic’s QES is a 4.0-compliant application that: • makes production data transparent • supports the consideration of individual customer specifications in a mass production environment • provides relevant information to support the human being in making a decision in case of quality deviations or process deviations – throughout the entire value chain.

SAP

Stand A12

Email: info@sap.com Website: https://www.sap.com/industries/mill-products.html

A leader in enterprise application software, SAP helps companies of all sizes and industries run at their best. From back office to boardroom, warehouse to storefront, desktop to mobile device – SAP empowers people and organisations to work together more efficiently and use business insight to stay ahead of the competition. SAP applications and services enable more than 378,000 business to operate profitably, adapt continuously, and grow sustainably. Massive shifts in the steel industry are forcing dramatic changes in global and local markets. Steel companies need to find solutions to be profitable in oversupplied, competitive, and constantly changing markets. SAP has 40 years’ experience working closely with hundreds of steel manufacturers across the globe to use digital innovation to anticipate real-time demand and supply, enhance process excellence for operational efficiency, operate resilient supply chains, and innovate the customer experience. SAP is in tune with how to apply the newest technologies to challenges and opportunities facing your steel business. One of our newest innovations solves challenges in the area of reducing the high costs of asset maintenance. Organisations manage thousands of assets to keep plants operational. Better pooling and sharing of all up-to-date asset-related content is one of the key problems faced by organisations. Maintenance can be a difficult challenge, due to heterogeneous systems, or incomplete operators’ manuals. SAP’s Asset Intelligence Network (AIN), solves these issues. AIN is a cloud-based collaborative network that allows companies to collect, track, and trace equipment information in a central repository. Operators can access up-to-date maintenance strategies, manuals, and more from manufacturers – and manufacturers can automatically receive asset usage and failure data from operators. Steel Times International

profiles exhibitors.indd 7

SMS digital

Stand A01

Phone: +49 211 881-5332 Email: hello@sms-digital.com Website: https://sms-digital.com/#home

SMS digital GmbH is a start-up of the SMS group, the market leading constructor of metallurgical plants and machinery for the industrial processing of steel, aluminium, and non-ferrous metals. With more than 13,000 employees worldwide, SMS group generates a revenue of about 3.3 billion EUR a year. SMS digital takes up the challenges of “digitisation“ and “Industry 4.0” within the SMS group. With the use of state-of-the-art innovation methods, know-how of metallurgical processes and technological expertise, we create new digital products, that, from the very beginning, are developed in close collaboration with our customers and end-users as well as with SMS group experts. This partnership results in the best possible solutions, which are perfectly tailored to the customers‘ needs, with immediate added value. Founded in May 2016 as an independent business unit, SMS digital is still under development – creating ideas, interviewing customers, recruiting further staff. Domiciled in their office in Düsseldorf’s Schwanenhöfe, the continuously growing, young and dynamic team is tackling the challenge of developing digital products for the steel industry. The aim of SMS digital is to become a leading provider of industrial IT services and digital solutions. All our products are first steps of the vision of an intelligent steel plant, that takes full advantage of the technology provided by the 21st century, to increase its productivity and user friendliness.

SMS group

Stand A02

Website: https://www.sms-group.com/ Email: communications@sms-group.com

The SMS group unites global players in plant and machinery construction for processing steel and non-ferrous metals, operating under the roof of SMS Holding GmbH. The family-owned company – now run by the fourth generation – stands out with its strong market position and corporate culture of responsibility, together with high-performance products and services tailored to individual customer requirements. The SMS group combines the flexibility of medium-sized company units with the vast resources of a global group. We build on the continuous training and comprehensive expertise of our employees to develop ground-breaking technologies. What’s more, as a systems supplier, the SMS group also covers electrical systems, automation and service. The SMS group is your partner for new plant and machinery as well as modernisations and upgrades. FUTURE STEEL FORUM

27 29/08/2019 12:58:30


EXHIBITOR PROFILES

Spraying Systems Co

Stand A19

Phone: +48 / 32 / 238 81 11 Website: http://www.spraying.pl/ Email: spraying@spraying.pl

TMEIC

Stand A05

Website: https://www.tmeic.com/industry/metals

You’ll find spray technology solutions for every area of your mill in Spraying Systems Co’s extensive product line. Its CasterJet®, DescaleJet® Pro, VeeJet® and FullJet® nozzles are industry standards. But that’s just the beginning – we have nozzles and systems for precision oil application, selective roll cooling and heating, dust suppression, descaling and more. Our steel industry experts work with mills around the world optimising spray operations. Services include on-site evaluations, impact and wear testing in our spray labs, descale header design using proprietary software, gas cooling calculations and process modelling for pollution control operations.

TMEIC drives industry around the world through a comprehensive offering of unique systems solutions including variable frequency drives, motors, photovoltaic inverters and advanced automation systems for a wide range of industrial applications. Established in 2003, Toshiba Mitsubishi-Electric Industrial Systems Corporation (TMEIC) resulted from the integration of Toshiba and Mitsubishi Electric Corporation’s industrial systems divisions. The company’s committed approach to collaborative solutions development ensures every industry throughout the globe can benefit from the world’s brightest minds. As a result of this powerful combination of resources, TMEIC is positioned to develop innovative technologies, quickly respond to industry trends and apply solutions to a wide variety of industrial market segments around the world.

SWERIM AB

XOM Materials

Stand A15

Website: https://www.swerim.se Email: info@swerim.se

Swerim is a leading industrial research institute within mining engineering, process metallurgy, materials, manufacturing engineering and applications. Swerim’s strength is applied research for resource-efficient and sustainable industry, and it has solid knowledge and experience when it comes to applying research results in practical industrial applications. Swerim customers are mainly within the mining, steel and metal industries, but we also work closely with suppliers to these sectors. As a strategic R&D partner, Swerim strengthen its clients’ competitive advantage and contribute to the development of new solutions for processes, materials and products. Swerim develops and optimises metallurgical processes and total solutions in materials and manufacturing engineering, mainly with a focus on energy efficiency, recovery and recycling of residual materials, reduced CO2 emissions, materials development, joining, additive manufacturing and component properties. Our longstanding experience from research in large pilot scale for extreme environments is a success factor for our customers’ competitive advantage and sustainable development.

28 profiles exhibitors.indd 8

FUTURE STEEL FORUM

Stand A13

Website: https://xom-materials.com/marketplace/en/

XOM Materials is the international go-to online marketplace and procurement platform for all products and services related to the manufacturing industry. As an interface between vendors and buyers, the platform digitises the operative ordering process of materials and supports its customers with flexible financing services and streamlined logistics. In 2017, the independent platform was founded with its headquarters in Berlin and has since expanded its presence to offices in Duisburg (DE), Prague (CZ), Valencia (ES) and Atlanta (USA) with around 50 employees.

Steel Times International

29/08/2019 12:58:31


Creating the Mill of the Future As a global leader in system solutions, TMEIC specializes in automation through innovative control hardware and software, power electronics equipment, and large motors. TMEIC’s unique solutions serve industry and social infrastructures worldwide. We grow with our customers and consistently exceed expectations by providing solutions underpinned by high-quality, advanced products and exceptional engineering. At TMEIC, we drive industry. It is what we do. It is our passion, our singular focus and our never-ending pursuit.

WWW.TMEIC.COM +1-540-283-2000 2060 Cook Drive Salem, Virginia 24153 U.S.A.

JAPAN | NORTH AMERICA | SOUTH AMERICA | EUROPE | SOUTHEAST ASIA | INDIA | CHINA | MIDDLE EAST | AUSTRALIA


ARTIFICIAL INTELLIGENCE

The Alpha Dog in the Human-AI Team On 11 May 1997, a computer’s first-ever tournament defeat of a reigning world chess champion marked a milestone in the history of artificial intelligence. Twenty-three years later, intelligent algorithms may soon replace human employees. By Patrick Henz*

T

he AI chess win marked the first of many that later included the Chinese board game Go (2016) and Poker (2017). One year later, the massive online multiplayer game StarCraft II joined the list. According to a 2015 Deloitte study, the replacement of human employees by intelligent algorithms and machines is expected to begin around 2020. At risk are all jobs that are highly repetitive and/or include a decision-making process based on a large amount of data. This includes middlemanagement positions in which responsible algorithms could maximise the operation of workshops along with co-ordinating the

30 patrik henk.indd 2

FUTURE STEEL FORUM

human employees. Intelligent algorithms, combined with machine learning, will reduce the number of today’s jobs. On the other hand, humans are freed from check-the-box processes to concentrate on tasks that require human ingenuity and creativity. Therefore, new jobs can be created. W. Edwards Deming, an expert on systems and quality, understood the human employee as an integrated part of the system. So, it is not surprising that his approach to understand and improve the system included the philosophy of continuous learning. Accordingly, Alibaba founder Jack Ma demanded at the World Economic Forum: “Everything we teach

should be different from machines.” Explicitly he mentioned values, beliefs, independent thinking, teamwork and care for others. This would be ensured by an holistic educational approach such as STREAM (Science, Technology, Reading, Engineering, Arts and Mathematics). To process his personal defeat against IBM’s “Deep Blue,” chess Grand Master Garry Kasparov organised a “freestyle chess tournament” where groups of humans, chess programmes and mixed human-AI teams could join and play against each other. To his surprise, a team of average players, using average chess programmes, won the Steel Times International

28/08/2019 12:26:43


tournament based on a superior process. He concluded: “Weak human + machine + better process was superior to a strong computer alone and, more remarkably, superior to a strong human + machine + inferior process.” According to Deming and Kasparov, we can conclude that a team with a superior system is the perfect combination of the individual players. Speaking of systems, the ideal process is not limited to fixed bureaucratic structures, but it has to be “liquid.” This means that the process should be capable of adapting to changing economic and technological realities, empowering individual employees to respond based on values and attitudes. The liquid condition can be strong and at the same time flexible (comparable to James Cameron’s shapeshifting T-1000 robot from the 1991 movie “Terminator 2: Judgement Day”). Erik Brynjofsson, director of the Massachusetts Institute of Technology (MIT) Initiative for Digital Economy, adds: “AI won’t be able to replace most jobs anytime soon. But in almost every industry, people using AI are starting to replace people who don’t use AI, and that trend will only accelerate.” Paul Ryan, IBM Watson’s UK director of Artificial Intelligence, goes one step further with his prediction that in a few years “every major decision, business and personal, will be made with the assistance of cognitive technologies.” A successful team requires a cooperative mindset of its integrands, not only by the human ones, but also the machines. Responsible algorithms can control machines or be part of the corporate software. Apart from the term “Artificial Intelligence,” the algorithms had been programmed and approved by humans. Accordingly, they represent human vision, experience, attitudes and values. If responsible algorithms get embedded into machines, we speak about the so-called “cobots,” short for “collaborative robots.” This philosophy includes the premise that the machines take good care of their human colleagues. Thanks to their robust metal bodies, cobots can replace human workers in such dangerous environments, like near an electric arc furnace, where scrap steel is melted. They can also be used for emergency events, like Mitsubishi Heavy Industries’ Water Cannon Robot, which is designed for firefighting with a water discharge capability Steel Times International

patrik henk.indd 3

of 4,000 litres per minute. As industrial robots are connected via the Internet of Things or the cloud, it is imperative that cybersecurity ensures that no intruders gain access to the machines. Regular tests should ensure the AI still acts as required, because Machine Learning could have altered its behaviour. Robots are machines, and responsible algorithms and sensors convert them into cobots. The concept is not limited to them, but can also include exoskeletons, Augmented Reality glasses or autonomous vehicles. Responsible AI can even act without a physical body, as part of corporate software to support an adequate decision-making process for human employees. Intelligent nudging can provide moments of personal disruption.

Responsible algorithms, for example, can identify potential phishing emails and flag them accordingly. This notifies the user, who can then start a more extended decisionmaking process to analyse the suspicious email – an important task since due to the daily workload, many humans work rigidly “like robots.” The algorithms can act as a second set of eyes, a colleague who asks critical questions, so that we start to challenge our own point of view and try to gather more information before we make a final decision on a topic. As Apple CEO Tim Cook said at MIT in 2018: “I’m not worried about artificial intelligence giving computers the ability to think like humans, I’m more concerned about people thinking like computers, without values or compassion, without concern for consequence.” Perception is based on sensors and companies such as Deloitte play with the

idea of using Augmented Reality glasses not only in workshops and factories, but also in office spaces. The integrated camera (possibly including an integrated microphone) is able to detect potential red flags for the employee. This may be overstepping security boundaries on a factory floor, or detecting a certain paragraph inside a new contract. The machine intelligence acts as the first level of support. The human, alerted or not, remains the final decision-maker. Today, Industry 4.0 and artificial intelligence enter all kinds of business environments. As they mostly aim at smart data and decision-making (for example for predictive maintenance and system optimisation), white collar employees have the greatest opportunity of working with this technology and getting enhanced by it. Nevertheless, as the boundaries between white and blue collars vanish and evolve into “new collar,” Industry 4.0 is growing in factories and workshops. Science fiction movies, such as Alien Resurrection and Avatar, present exoskeletons. By definition, such equipment provides both, support and protection. Smaller exoskeletons are not designed to provide general additional force, but “it redistributes the load over your shoulders and into your core muscles, so you’re using the right muscles to perform the lifts, thus reducing risk,” StrongArm Technologies chief marketing officer Matt Norcia explained. Instead of just muscle, exoskeletons provide an individual with additional brain power. The equipment’s task is to avoid costly accidents inside the workshop. The processes aim to reduce physical risks – a benefit for the individual and the organisation, as fewer work accidents in most countries result in lower insurance fees and taxes. As a self-fulfilling prophecy, the Pygmalion Effect (developed by two social psychologists, Robert Rosenthal and Leonore F. Jacobson) explains that: if the robot’s behaviour is perceived by the employee as respectful (according to lawyer and robot ethicist Kate Darling, humans have the tendency to humanise robots), it confirms the co-worker’s positive self-esteem that he/she is a respectful person. By imposing this value, the individual treats the machine respectfully in return. A cycle is created, and a positive corporate culture reduces the risk of sabotage and inadequate machine handling. FUTURE STEEL FORUM

31 28/08/2019 12:26:46


ARTIFICIAL INTELLIGENCE

The implementation of such machines is not a pure technical initiative, but requires holistic corporate change management, including adequate support from top management. The cobot philosophy supports the acceptance by the employees, as it respects the existing corporate culture. We may have to change our romantic view of robots, they are not independent as we perceive them based on their appearance. Rather, they are an integrated part of the network. Primetals Technologies’ Trimrob is such a robotic system, designed for coil trimming and sample taking, placed within the coil handling area of a wire rod mill. Trimrob operates within a system of physical safety fences, laser light curtains, and laser area scanners to ensure that human operators do not enter the work area while the machine is operating. Additional safety measures include collision detection and torque monitoring. “He who is cruel to animals becomes hard also in his dealings with men. We can judge the heart of a man by his treatment of animals.” (Immanuel Kant) The human employee is the “alpha dog” regarding relations with intelligent algorithms: the determining part. Not only limited to strategy and decision-making, it also takes care of the selected information and

code.” This definition gives AI certain human characteristics, as the behaviour of the theoretical model (Digital Twin) represents the thoughts, knowledge and experience of human engineers and other subject matter experts. For example, engineers at Industrial Technology Research (ITR), one of the world’s original predictive maintenance companies, manually collect, process, and analyse rich data sets with every analysis they perform. With experience averaging more than 20 years, their analysts look at steel machinery (complex asset analytics) coupled with a thorough peer review process to eliminate false alarms and missed problems. ITR digitises this knowledge with every cycle, to use it as a basis for training. This collaboration between the human alpha (engineer) and AI beta (ITR’s proprietary platform) removes bias and error to the greatest possible extent. Despite ongoing advancements in technology, complex asset data analysis still requires human intervention to ensure effective condition monitoring and predictive maintenance. Patrick Henz is head of governance and compliance at Primetals USA and regional compliance officer, Americas.

Sources:

Between Humans and Robots: Experimental Results”

female? Gender bias in AI must be remedied”

•Cameron, James (1991): “Terminator 2: Judgment Day”

•Li, Borui / Lauden, Andrew / Davis, Jonathan / Stohl,

•Shirota, Jun (2019): “New collar: a new fit for a dynamic

•Cameron, James (2009): “Avatar”

Klaus (2019): “High Quality Predictive Maintenance Through

manufacturing hub”:

•Conbolly, Byron (2018): “Are you a middle manager? AI

Advanced Analytics”

•Siniscalchi, Marcello / Stipo, Carlo / Quaranta, Angelo

may take your job,”

•Ransbotham, Sam / Kiron, David, Gerbert, Philipp /

(2013): “ ‘Like Owner, Like Dog’: Correlation between the

•Cuteness Team (2018) on Pet Care Blog: “Are Dogs Good

Reeves, Martin (2017): “Reshaping Business with Artificial

Owner’s Attachment Profile and the Owner-Dog Bond”:

Judges of Character?”,

Intelligence.”

•Teegavarapu, Sid / Palfreman, Matthew / Shen, William

•Finney, Sherry / Corbett, Martin (2007): “ERP

•Roe, David (2018): “7 Jobs That Artificial Intelligence (AI)

/ Zelle Jason (2019): “Intelligent Robotic Coil Trimming for

implementation: a compilation and analysis of critical success

Will Soon Overtake,”

Wire Rod Mills”

factors”

•Rosenthal, Jacobson (1968): “Pygmalion in the Class Room”

•Violino, Bob (2017): “Can Artificial Intelligence Help with

•Jeunnet, Jean-Pierra (1997): “Alien Resurrection”

•Schuetz, Anja (2019): “His name is Yochai”

Heavy Lifting?”

•Kshirsagar, Alap / Dreyfuss, Bnaya / Ishai, Guy / Heffetz,

•Scott, Ridley (1982): “Blade Runner”

•Whitwam, Ryan (2019): “DeepMind AI Challenges Pro

Ori / Hoffman (2019): “Monetary-Incentive Competition

•Sheppard, Emma (2019): “Why are virtual assistants always

StarCraft II Players, Wins Almost Every Match”

32 patrik henk.indd 4

algorithms. On the other side, the “beta,” the cobot, has to be empathic (predicting risks) not only to avoid accidents, but also to work efficiently as a member of the pack. It is a common misbelief that because they are “descendants of the mighty wolf,” dogs can perceive the character of a person. But unlike wolves, dogs are no longer pack animals. For example, they share their food with others only in exceptional cases. Nevertheless, they still accept the role of the “alpha,” which would be the human owner. They not only respect the alpha, but also learn from it, as they interpret their owner’s behaviour as correct. This affects their subsequent perception, emotions, thoughts and expectations. In other words, “people who act in a manner the dog is used to may make the dog feel calm, while people who act different might make the dog feel uneasy.” (Pet Care Blog) The animal confirms its owner’s opinion and perception. By no means is Machine Learning only based on mathematical truth. Instead, through observed behaviour (sensors) and processing (code), algorithms learn from humans and, therefore, embrace their perceptions. Ivana Bartoletti, founder of the “Women Leading in AI”network, proclaimed, “An algorithm is an opinion expressed in

FUTURE STEEL FORUM

Steel Times International

28/08/2019 12:26:47


DIGITAL TRANSFORMATION

20 Dilemmas in digital transformation Italian plant builder Tenova shares some of the dilemmas, or trade-offs, encountered when the company embarked upon its digital transformation. By Giovanni Bavestrelli*

I

recently shared my experience with Tenova’s digital transformation at a couple of industry conferences in Germany. The final few slides in my presentation listed some of the dilemmas, or trade- offs, we had encountered on our journey. My thinking was that if we had found these decisions critical, so might others. Indeed, there was a lot of interest in these particular slides. Since several people have asked me to share them, I thought writing an article might be appreciated. Of course, when undergoing a digital transformation, you will probably face many more decisions than are listed here, but this small set includes some I identified on our journey. Keep in mind that these are dilemmas,

Steel Times International

Tenova done.indd 3

or decision points, between two seemingly opposite concepts, and yet there is no absolute right or wrong positioning. More often, a blend of the two concepts will prove to be the right combination to bring the most value to your organisation and clients. I cannot know where the right compromise is for you, I can only speak about our struggle to find the best position for ourselves. However, I do believe that wherever you position yourselves between the two concepts will have a significant impact on your company. Products versus services This is obvious. We all know that today’s economy requires moving from products to services. In our case, our plants are big, heavy and expensive, built at customer sites and not

mobile, hence not so easy to rethink of them as services. We will probably still sell products for a long time. Nevertheless, defining an assortment of new services to offer alongside the physical products is the right direction to take. Ownership versus access Similar to the issue described above, we can all name companies that experienced exponential growth using business models based on access rather than ownership (think of Airbnb or Uber). This is not so easy in our industry, and yet every time we think of a big investment that results in ownership, we should consider how to capitalise on new or existing assets and sell access to them rather than focus solely on ownership. FUTURE STEEL FORUM

33 29/08/2019 08:59:00


DIGITAL TRANSFORMATION

IoT and sensors versus digital strategy When we started on our digital journey, we focused on individual technologies, mostly on IoT and sensors, while also starting pilot projects. We soon realised that the transformation underway was bigger than any sum of individual technologies, and that we really needed an overall digital strategy that would affect the whole company. We then developed a strategy and built a team. Having a strategy was key to bringing awareness and focus. Exciting technologies versus customer value People like me get excited about technology, but focusing solely on the technologies themselves can be distracting. New technologies redefine what is possible, but what is possible is not necessarily useful. Maintaining focus on customer value will indicate the right direction to take. Nevertheless, keeping up with new technologies can give technicians an idea of tomorrow’s needs even when today’s customers cannot yet perceive their value. Return on investment versus leap of faith In the initial phase of a big project, it is natural to set about predicting the potential return on investment. On our digital transformation journey, we found this very challenging. We quickly became aware of how difficult it is to determine the return on investment with any reasonable accuracy. For instance, we do not yet know what price our customers will be willing to pay for new services. Despite this, we took a leap of faith and invested anyway. Cost of investing versus cost of not investing A leap of faith may feel uncomfortable. It is easier to face, though, when you weigh the potential costs of not investing. By neglecting to explore the possible returns of a digital transformation, a company risks waking up late to discover its market has evaporated

34 Tenova done.indd 4

FUTURE STEEL FORUM

while it clung to old business models. In short, the winning strategy is to balance the cost of investing with that of not investing. Hard-metal culture versus software culture Our company has a traditional engineering culture focused on the tangibles of mechanics, steel, machinery, and plants. A digital transformation requires a flexible and novel way of thinking about solutions, a mindset that is more similar to what is common practice in the software world, think for instance of agile methodologies. These new approaches might feel uncomfortable at first, but a change in culture is necessary. Predict and control versus learn and adapt One way in which the culture must evolve is by shifting from the previous management style of “predict and control” to that of “learn and adapt.” In today’s digital reality, we learn as we go, which requires us to adapt and change direction often as we apply new knowledge and correct mistakes. Agile iterative processes are useful for establishing discipline while also allowing flexibility. Expectation of success versus openness to failure It is normal to strive for success when we undertake projects. However, we need to tolerate and even welcome mistakes along the way since mistakes are at the very heart of the human learning process. Corporate culture must allow for mistakes and see them as opportunities for growth, not as something to punish, or people will not take necessary risks or think outside of the box. True innovators are ready to fail fast, learn, and change direction. Focus inside versus focus outside In the corporate world, we often overestimate the value of internal contributions while demonstrating skepticism towards what is developed elsewhere. However, in this digital transformation an optimal use of external resources is mandatory. Look outside. Attend

conferences. Learn from customers, suppliers, and even competitors. Investigate hackathons. Consider crowd sourcing, remote working, and open innovation. Internal digital team versus outsourcing digital projects One way to strike an internal/external balance is by deciding whether to build an internal digital team or outsource digital projects. If you have enough resources to build an internal team to develop competencies in-house, it is an excellent option. After all, the digital evolution will continue and a dedicated team could facilitate its progress as needs arise. Even so, some projects will still likely need to be outsourced. Finding the right mix is key. Alone versus partnership Even if you build a strong digital team, that team will still require support. Today it is fundamental to select good partners for key areas. One benefit of having an internal digital team is that the team members can effectively evaluate candidate partners, select them, manage the interaction with them, learn from them, and bring competencies in-house for future projects. Build versus buy Do not expect to be able to build the entire digital infrastructure on your own. Technology is moving too fast. What you decide to build and what you decide to buy will have big implications. Again, there is no right or wrong balance. We chose to control key parts of our solutions by building them with our partners’ help, while focusing on open source software and standard market solutions in other areas. Copy versus invent Copying has a bad reputation, but in these rapidly changing times, it is often a valid option. There is no shame in copying ideas. We looked to industries that are further ahead in the game, like aerospace and automotive, to “copy” ideas and solutions. Assume that what worked for others may work for you. You Steel Times International

29/08/2019 08:59:03


can still be innovative in your industry, while saving time and lowering risks by borrowing valuable ideas already tested by others. Hire versus train We all need data scientists and we know that they are expensive and difficult to find and retain. Hiring young talent is important, of course, but if you already have curious, disciplined people with a background in software development, they can be trained in data science rather efficiently using excellent online resources. This way you get data scientists with domain knowledge. Hire talent, but facilitate training and continuous learning too.

join the best: 30 March – 03 April 2020 Düsseldorf, Germany www.wire.de I www.tube.de

Backlog versus innovation Data scientists are useful, but you also need domain experts to come up with valuable solutions. Unfortunately, domain experts are already engaged in backlog projects, which are usually more urgent, so these technicians struggle to find time for innovation. This can be a big obstacle where compromise is necessary. We have found that domain experts and data scientists enjoy working together; the key is enabling them to commit time to the effort. Share versus hide This is tricky in our industry; we do not usually want to share data and information as we assume it gives us a competitive edge. This mindset can slow down innovation and become a handicap. When the benefits outweigh the risks, as is often the case, we must be open to sharing. Learn from companies like Microsoft and IBM, who after initial opposition decided to invest heavily in open source software. Cloud versus on-premise Many of our customers are not ready to allow their plant data to be stored on the cloud, preferring an on premise solution instead. We do provide on premise solutions, but we encourage clients to consider the many benefits of cloud services while reassuring them of their security. Closing the door to the cloud might become too costly for clients in time, as their competitors start reaping benefits of cloud services while they do not.

International Wire and Cable Trade Fair Internationale Fachmesse Draht und Kabel

International Tube and Pipe Trade Fair Internationale Rohr-Fachmesse

Simplicity versus security It is very easy to ignore security for the sake of simplicity at the beginning of an IoT or cloud project. This is a mistake. Security is fundamental in an IoT and cloud world, although not all applications require the same level of security. Keep simplicity in mind, but start thinking about security from the very beginning. Involve company IT early and rely on market leaders in the field. China versus rest of the world Many regulations, especially those regarding cloud technologies, are fundamentally different in China. What works elsewhere might not work in China and vice versa. If you have a cloud platform supporting your customers, it will most probably need to be redefined for China. Keep this in mind as you select where to invest. Legal and intellectual property issues are not trivial.

* Digital engineering director at Tenova Messe Düsseldorf GmbH P.O. Box 10 10 06 _ 40001 Düsseldorf _ Germany Tel. +49 211 4560 01 _ Fax +49 211 4560 668

www.messe-duesseldorf.de Tenova done.indd 5

29/08/2019 08:59:04


TESTING & ANALYSIS

Developing trends in metal production surface inspection

H

istorically, surface inspection systems have been built on the adaptation of traditional camera technologies. This has consisted of looking for anomalies in a black and white image in a 2D landscape, and then using software technologies to automatically sort and classify them. This has been the basis for quality inspection, particularly in the metals industry, for the past two decades. Within this landscape, the fundamental issues were relatively straightforward, with only one question to answer: how surface inspection system manufacturers optimise both the optical set-up and the software filtering technologies required to process hundreds of thousands of false detections, or non-critical pseudo-defects with the vision system. These pseudo-defects could be detections that are just standard anomalies within the product, or within the variability of the process. As a consequence, the surface inspection industry has spent a long time developing

36 AMETEK done.indd 2

FUTURE STEEL FORUM

better ways to filter the massive amount of visual data being received. Once that data was filtered and the critical defects extracted, the debate for vendors was the best way to classify it. For AMETEK Surface Vision, the answer has been machine learning, where techniques such as decision tree classifiers and support vector machine are used to classify defects. In the past five years, the most hotly contested battleground in surface inspection has been to deliver the highest camera resolutions possible. Over this period, camera resolution has gone from inspection areas measuring around 1mm in diameter, to current systems of about 100 microns resolution – which is approaching the limits of what the human eye can see. The expectation is that resolution will reach the 50-micron level, a point at which the process begins to move beyond surface inspection and into the realms of materials analysis. However, that ‘arms race’ continues

to this day, and all the leading inspection manufacturers have arrived at virtually the same point, in terms of resolution. The next evolution will be centered around increasing the amount of information produced, which will be a transition to additional sensor technologies, including utilising 3D inspection sensors. The demand for 3D What customers are requesting now is 3D measurement – either complementary to a system, or as a standalone solution separate from what 2D black and white cameras can achieve. In particular, they need to know more about the shape, severity, and origin of a specific defect. A defect might be above the surface, on the surface, or sub-surface. Understanding that depth of a defect would dramatically improve the accuracy of the defect classification, and potentially provide critical information about pre-inspection production issues. Steel Times International

28/08/2019 14:17:43


Jason Zyglis* looks at the future of surface inspection systems for the metals industry

However, as this is a new frontier, what hasn’t yet been defined is exactly what that 3D measurement means. Critically, we need to understand what it is that customers really want and need. So, while it is a desirable talking point, implementing 3D technology techniques currently requires a significant increase in cost of any surface inspection system. This means there’s a balance to be reached between system cost and customer requirements and expectation. What is the satisfactory level of 3D performance required to meet and improve inspection requirements, while still falling within a price-point that the customer is willing to pay? There’s an assumption from customers that if they know and understand the 3D topography of a product, it will dramatically increase the value proposition of the vision system information. Of course, producers would love to measure everything, but this currently means installing a dramatically more expensive system. Nonetheless, we should Steel Times International

AMETEK done.indd 3

anticipate seeing this cost-performance balance being reached somewhere within the next five years. 3D measurement technologies Without determining where the balance between performance and cost lies, it’s difficult to know which kinds of 3D technology is likely to dominate the field. There are plenty of candidates in development and deployed to beta sites. For example, two cameras can be used to create a stereoscopic image, creating an image with 3D characteristics that is similar to the human eye. This doesn’t give the most accurate measurement, but it is cost-effective, and usually good enough to determine if a defect is on the product or penetrating below the surface. A handful of other methods exists. For instance, a more accurate method projects a fine laser line across the surface of the material, with a matrix array camera looking for the displacement of the laser, but this technology comes at a premium cost. But without common agreement on the level of specification required, it will be difficult to understand which technology will become the most likely candidate for adoption. Do customers need to measure the depth of a 100-micron pit, or is it simply enough to know that the pit exists? Academia and customer research teams have ways to measure this in a lab, but no-one has scaled it up to a production environment yet. Once this is done, it could be transformational. In terms of adopting 3D technology, the order is likely to follow a similar pattern as the original early adopters of surface inspection systems. We should expect to see the early drivers being aerospace, automotive and electronic handheld device manufacturers. Benefits for hot materials and oxidation A particular driver for 3D – particularly in the steel and aluminum production – is the inspection of hot materials: caster, slab, ingot and hot mill. 3D is particularly for these hot process for two reasons. Firstly, it provides additional detailed information for small and

subtle cracks; secondly, it enables us to distinguish scale from underlying defects With a 2D black & white camera, the surface just looks like it has a lot of optical noise. With 3D, the operator would quickly understand the gouges and scratches underneath that. 3D will essentially detect issues that are invisible, or at least undefinable, in a 2D visual inspection technique. Artificial intelligence Just as 3D inspection is about delivering an incremental increase in detection capabilities above that available in visual inspection, the industry is looking at artificial intelligence (AI) for the next step forward in data handling and analytics. Machine learning systems for classifying surface defects have led the way over the last 15 years. This is predicated on assembling information – sometimes hundreds or thousands of images – and running them through decision tree classifiers to create purpose-built classification for specific applications, such as the hot mill, cold mill, galvanising line or pickling line. These systems are trained and can “selflearn”, which happens by customers sorting defect images into groups, and then feeding them into self-learning classifiers that can learn about each group. Changing production grades means building a whole new classifier, and if you’re on an automotive galvanizing line you may have between five and 10 discrete recipes, each created from the ground up. Since material and machine can both change over time, it’s a never-ending cycle of testing and optimising the classifiers.

FUTURE STEEL FORUM

37 28/08/2019 14:17:50


TESTING & ANALYSIS

Artificial Intelligence (AI) could help minimise the number of iterative classifications that a customer has to set up. This would make it much easier for less experienced users to get the most out of their surface inspection data, and dramatically reduce the time spent fully optimising an inspection system. Analytics and big data Surface inspection companies are essentially data businesses. With thousands of cameras in operation, each recording terabytes of data per day, an incredible amount of information is created and archived. However, because of the need to filter the data and provide only the information needed to identify the relevant defects, almost all of it is thrown away. This means a lot of contextual information is ignored. This unused data may be useful in understanding more than just defects: it could inform the customer about developing issues with the product or equipment. The potential is exciting; for instance, if you could take hundreds or thousands of individual inspections, and virtually stack them one on top of the other, what patterns might you see, and what could you learn about your process? The data could help you optimise where you allocate your production, let you schedule your material more effectively, make mechanical changes to the process, or metallurgical changes to the recipe.

It could also support a more predictivebased approach to maintenance, for instance informing you your machine may have a problem in three weeks, because it can spot early indications of a problem that hasn’t visibly manifested itself elsewhere in the process. Currently this level of data is just too much for systems to manage. And while surface inspection vendors are not going to be the players in “big data” aggregation, because our focus is more on the sensor technology, there may well be opportunities in the future to make our data more valuable to the customer. Industry 4.0 Industry 4.0 and the Internet of Things (IoT) are now frequently used buzzwords in the marketplace. At a corporate level it’s very appealing, especially in an industry like steel where they are transitioning to a hightech, data-driven approach to a traditional industry. However, talking to those who actually have to make sense of the information, and use it to run a steel mill, the case needs to be made more clearly. Their primary concern is: ‘can I use this?’ The opportunity lies in getting the industry and the vendors to agree on a common meaning beyond buzzwords. Customers may ask “What’s your Industry 4.0 solution for surface inspection?”, but without any common agreement, it’s an impossible question to answer. Feedback At base level, Industry 4.0 is much more about feedback. It’s about predictive maintenance on a machine, allocating resources and standardising equipment. Customers are still making the same decisions, except that where you once had a spreadsheet with 50 parameters, now you have millions of data points. That’s where AI scales and make sense of that data stream coming at you. Job-specific training Many heavy industries, including metals, have become less directly skilled, with many experienced engineers now entering retirement. Accordingly, facilities are training people for specific job functions, which means these roles are turning into officebased positions. AI will compensate for this

38 AMETEK done.indd 4

FUTURE STEEL FORUM

skills gap, because the systems themselves will diagnose problems and tell users what’s trending. Instead of independent systems, a community of systems will share cloud-based information, which will inform better decisionmaking. As surface inspection vendors, our role will be to provide more meaningful information at the sensor level. Conclusion Right now, the lack of certainty around specifications and requirements is the biggest problem facing surface inspection in taking the next big leap forward in data analysis. When customers ask, “What do you have that’s Industry 4.0?”, they are looking for solutions, but the lack of standardisation means – even with the best of intentions – vendors run significant risks of selling solutions that don’t address customer needs. The foundation of “what are we attempting to do here?” hasn’t been defined. So rather than have a wasted decade with vendors throwing ideas at the wall and seeing what sticks, it will be better for all parties if individual producers, industry organisations and web inspection suppliers can sit down together and define the specifications, requirements and expectations. The industry is going to have to coalesce around a defined set of industry expectations and standards, and that will inform the path forward. If leading suppliers – and there are only a few of them – can work together, it will be for the benefit of all. They can fight over market share, as they always have, afterwards. * Divisional VP of project and product management, AMETEK Surface Vision Steel Times International

28/08/2019 14:17:56


Rolling out a stronger future. Together.

Two world leaders in industrial process fluids, Quaker Chemical and Houghton International, have come together as one company to keep our steel customers ahead in a changing world. As Quaker Houghton, we are proud to partner with the world’s most innovative steel companies, whose products are building a sustainable, more durable—and more prosperous—future. With our customized industrial process fluid solutions, we are able to help you optimize operations and performance.

quakerhoughton.com ©2018 Quaker Houghton. All rights reserved.


FOG COMPUTING

How fog is good for smart factories So-called Fog Computing moves cloud-type resources closer ‘the edge’ where ‘things’ and the Network of Things can use more advanced computation resources with much smaller network delays. By Lane Thames*

D

igital technologies have been transforming our world for the past few decades. Within the industrial realm, we are only beginning to see the impacts of this transformation. The Internet of Things (IoT) and cloud computing have induced an evolution in the way we as society live our everyday lives, as well as how many enterprises conduct business. This evolution has started to enter the industrial realm, most notably seen by those involved with ideas such as the Industrial Internet of Things (IIoT) and Industry 4.0. The IIoT and the cloud are major driving forces behind other innovative ideas such as smart factories. Smart factories can achieve significant advancements with IIoT and cloud technologies. For example, predictive analytics using data from the IIoT and processed in the cloud enable optimizations of various processes for smart factories. However, many industrial organizations, including those in the steel industry, have systems with more stringent requirements such as real-time computational and communication constraints that cannot be offered by the cloud. To address these limitations, fog computing has emerged. Fog Computing is a new paradigm of computing that will provide significant benefits to industry. This paper will discuss these interconnected ideas and the relationships between the IIoT, cloud computing, and fog computing, review

40 Thames.indd 2

FUTURE STEEL FORUM

some of the benefits and problems being solved, describe how the steel industry will be able to take advantage of this new paradigm in the near future, and discuss the importance of cybersecurity and how it must be viewed as a fundamental component of this entire ecosystem of technology. Concepts In this section, we will seek to understand the concepts related to smart factories so that we can understand its connection to the fog and the cloud. Let us start by asking: What is a smart factory? A smart factory can be viewed as a highly digitized and networked production environment containing numerous devices (things) with computing and communication capabilities. Some of these devices only communicate locally with other devices or legacy backend Information Technology systems. Some, however, such as cyber-physical systems, can even communicate with other systems via the global Internet. Smart factories enable smart manufacturing. Smart manufacturing generally seeks the utilization of smart factory resources to provide benefits such as robust, reliable, and safe manufacturing operations, to enable manufacturing automation, to optimize processes by reducing waste or reducing downtime, and much more. The next question to consider is: How can

we achieve the vision of smart factories? There are numerous technological paradigms being studied and developed to achieve the vision of smart factories as well as other smart entities such as smart cities and smart homes. These technological paradigms include the Internet of Everything (IoE), the Internet of Things (IoT), the Industrial Internet of Things, and Industry 4.0. Essentially, all of the paradigms are highly related. For example, the IIoT is a subset of the of the IoT, and the IoT is a subset of the IoE. Let us explore these related paradigms further. The IoE was introduced by Cisco Systems, Inc. [Cisco, 2013] as a system that “brings together people, process, data, and things to make networked connections more relevant and valuable than ever before — turning information into actions that create new capabilities, richer experiences, and unprecedented economic opportunity for businesses, individuals, and countries.” The IoT, however, is a technological paradigm based on things – objects with computing and communication capabilities embedded within it. These objects are networked together to form Networks of Things (NoT) potentially using numerous different communication protocols. These objects can be accessed via the Internet, currently enabled via connectivity protocols executed in cloud computing infrastructure. These objects often provide Steel Times International

28/08/2019 14:28:01


Artificial intelligence (Advanced)

Cloud

Big data

Artificial intelligence (Advanced)

Cloud

Big data Artificial intelligence (Intermediate)

Fog

Edge

Edge

Network of things Network of things

value-added features to things, such as being able to view a surveillance camera video feed over the Internet or turning a light off in a kitchen while on vacation. The primary difference between the IoE and the IoT is that of people and process. The IIoT is similar to the IoT, the main difference is that, by definition, the IIoT utilizes technologies and things that are based on industry needs and resources whereas the IoT goes beyond industry and contains things of any nature. Indeed, using the term IIoT is really just to emphasize that one is focused on an industrial perspective. Finally, let’s consider Industry 4.0 and how it fits into this picture. German Chancellor Angela Merkel [European Parliament Briefing, 2015] states that Industry 4.0 is “the comprehensive transformation of the whole sphere of industrial production through the merging of digital technology and the internet with conventional industry”. This implies that with Industry 4.0, all entities associated with an industrial system such as manufacturing are digitally connected. Forbes and Schaefer [Forbes and Schaefer, 2017] describe Industry 4.0 as a tight interconnection between the Internet of Things and the Internet of Services. All of these ideas reveal a type of equivalence between Industry 4.0 and the Internet of Everything, but with a main difference being that the IoE can, by definition, encompass all things, whereas Industry 4.0 has an industrial focus. Foundations and State of the Art The goal of the previous section was to provide an understanding of the various Steel Times International

Thames.indd 3

concepts, technologies, and paradigms that are being used as foundations for smart factory technology. One particular technology that is at the core of all the aforementioned ideas is the Internet of Things. The IoT is a paradigm, but it is also a reality—a highly complex system that exists worldwide. IoT research and development is advancing at a very rapid pace, and its adoption is growing exponentially. Numerous problems have been solved via the development of IoT technologies. For example, 10 years ago it would take a very crafty technical person to install a video surveillance camera in one’s home that was accessible via the Internet. Nowadays, this is a trivial task, which is a result of IoT technological advancements. Cloud computing plays a very important role for the IoT – without the cloud, we would not have an IoT as we know it today. Figure 1 is a simplistic view of the current state of the art for IoT architecture. To be brief, the IoT’s current architecture is composed of things that live at the edge of the Internet. These things are often referred to as edge devices. These things form Networks of Things (NoT) and they have the ability, generally speaking, to communicate with other devices or humans either directly using various types of specialized IoT communication protocols or via the Internet. As seen in the figure, IoT devices communicate to the cloud via the Internet. State of the art IoT technologies use the cloud for data processing, intelligence, and communication fabrics. For example, when one uses a smartphone to turn on a connected light in one’s home, the application on the smart phone communicates with an

endpoint in the cloud, and the cloud then uses a pre-established, always-on connection back to the device with the given command. This is a type of command-and-control architecture very commonly found with IoT products. Future Advancements with Fog Computing The IoT’s current architecture works well for many application domains, especially within the consumer market space. However, there are various shortcomings of the architecture, especially for smart factories and the IIoT. One of the most crucial shortcomings is that of time. Smart factories as well as other IIoT systems have many components (things) that are sensitive to time delays, meaning there are real-time constraints for many of the components in a smart factory (the same realtime constraints that exist in legacy, non-smart factories). Industrial systems utilize what is known as operational technology. Operational Technology (OT) is “the hardware and software dedicated to detecting or causing changes in physical processes through direct monitoring and/or control of physical devices such as valves, pumps, etc.” [Wikipedia, Operations Technology]. OT is comprised of systems such as programmable logic controllers (PLC), supervisory control and data acquisition (SCADA), distributed control systems (DCS), and computer numerical control systems (CNC) and, generally speaking, industrial control systems (ICS). Many of these technologies along with the many sensors and actuators that are used by FUTURE STEEL FORUM

41 28/08/2019 14:28:05


FOG COMPUTING

them have real-time constraints. For example, a PLC in certain environments might fail if a signal is not received and processed within an order of milliseconds. This real-time constraint poses a significant challenge for smart factories that need to send data to the cloud for processing – the amount of time between sending and receiving data from the cloud is too large for real-time control systems. A solution to this problem, as well as various others, is Fog Computing. Figure 2 illustrates how fog computing fits into the IoT/ IIoT architecture. In essence, the fog moves cloud-type resources closer to the edge where things and NoTs can use more advanced computation resources with much smaller network delays. With this architecture, smart factories can be achieved even for control system operations.

centers, and even dedicated fog computing nodes living at the edge with the networks of things.

There are several definitions of fog computing in the literature. According to the OpenFog Consortium [OpenFog, 2018], fog computing “is a system-level horizontal architecture that distributes resources and services of computing, storage, control and networking anywhere along the continuum from Cloud to Things. It is a: • Horizontal architecture: Support multiple industry verticals and application domains, delivering intelligence and services to users and business • Cloud-to-Thing continuum of services: Enable services and applications to be distributed closer to Things, and anywhere along the continuum between Cloud and Things System-level: Extend from the Things, over the network edges, through the Cloud, and across multiple protocol layers – not just radio systems, not just a specific protocol layer, not just at one part of an end-to-end system, but a system spanning between the Things and the Cloud”. Figure 2 illustrates this definition where the “fog computing layer” of the figure encompasses Internet connected systems between the edge and the cloud that can be used as a fog computing resource. Examples of these resources include switches and routers near the edge, on premise data

Challenges Ahead Although many benefits are achievable with the IIoT and smart factories, there are some notable challenges that lie ahead. Cybersecurity will be one of the most – if not the most – critical obstacles to overcome, if we are to see the full potential and benefits of the IIoT. System complexity and countless new cyberattack vectors will arise as more advanced IIoT systems come online. This brave new world will require some evolution within all parties involved – including those who build IIoT technology, those who use IIoT technology, and those who secure IIoT technology. Cybersecurity for the IIoT will require strong interdisciplinary collaboration between all of these associated parties. In other words, cybersecurity will no longer be siloed, and everyone will need to play some role. We will also need to move towards the core of the problem – education. Our educational ecosystem has failed miserably with cybersecurity. Every single Science, Technology, Engineering, and Mathematics (STEM) based student should be required to learn the fundamentals of cybersecurity. Nowadays, all of the technological things we design and build must be viewed with a cyber perspective, as the goal is to “connect all the things”. As such, every person who studies the design and implementation of technological

42 Thames.indd 4

FUTURE STEEL FORUM

Benefits and Problems Solved Fog computing will be a key technology to enable the full benefits that can be achieved with the IIoT and smart factories. These benefits are galore such as increased productivity, increased product quality, and increased safety. The IIoT will provide a technological path for clean and green manufacturing. Manufacturing industries will achieve customer level collaboration like we’ve never seen where will be able to implement the ideas of mass customization and individual customization at scale. The potential opportunities to optimize every aspect of the smart factory are endless.

things should be well versed in cybersecurity fundamentals, with an importance akin to the requirement of having to study calculus, i.e., no engineering student on this planet can get by without knowing calculus. In summary, a key to success for securing the IIoT is collaboration and appropriate cybersecurity education highly coupled with cybersecurity technologies. Closing Remarks This paper discussed the technological foundations required for smart factory technologies and other paradigms such as the Internet of Things, the Industrial Internet of Things, and Industry 4.0. Technologies such as cloud and fog computing where discussed to show how all of these ideas and technologies are interrelated. The Fog is a newer technology that is evolving and will be a critical component to fully achieve the potential for the IIoT and smart factories. One of the main obstacles for the IIoT and smart factories is cybersecurity. Collaboration and appropriate cybersecurity education highly coupled with cybersecurity technologies will be necessary for achieving our goals and visions for a future world where everything is smart. References Cisco, 2013: https://www.cisco.com/c/ dam/en_us/about/business-insights/docs/ ioe-economy-faq.pdf European Parliament Briefing, 2015: http:// www.europarl.europa.eu/RegData/ etudes/BRIE/2015/568337/EPRS_ BRI(2015)568337_EN.pdf Forbes and Schaefer, 2017: “Social Product Development: The Democratization of Design, Manufacture and Innovation,” Procedia CIRP, vol. 60, pp. 404-409. Wikipedia, Operations Technology: https:// en.wikipedia.org/wiki/Operational_ Technology

* Senior Security Researcher, Tripwire Inc’s Vulnerability and Exposure Research Team (VERT) Steel Times International

28/08/2019 14:28:09


DATA ANALYTICS

Data analytics and Tata Steel Europe

Tata Steel in Europe is a diversified steel producer with global operations throughout the carbon steel and electrical steel value chains and one of the leading European producers of flat steel products. The company is serving demanding markets worldwide, including construction and infrastructure, automotive, packaging and engineering, and is striving to understand the different market and customer needs. Customer focus is part of Tata Steel’s strategy, and it is regularly collecting customer feedback to guide its activities. By Dr. Svend Lassen1 and Matthé de Vent2

O

ver years, customer feedback indicated that challenges in our supply chain had a significant impact on the customer experience and thus should get the highest priority. There was a lack of transparency on ‘where the steel was’, long lead times and late deliveries, low predictabilities of orders, and low responsiveness to supply chain events. On this background, we launched a multi-year transformation programme called Future Value Chain (FVC). FVC is addressing the challenges of our supply chain as well tapping in sales and marketing-related opportunities, thus covering the complete value chain (see chart 1): • Aligned system design: Capacities and stock buffers in the production system must be aligned for the most efficient set-up to reliably deliver the plan. • Optimal demand-supply balance: The

Steel Times International

Svend lassen done.indd 3

balance between demand and supply must be optimised, requiring to forecast and plan future demand and capacity more accurately. • Agile deviation management: Fast interventions are needed for deviations in the supply chain in order to retain and maximise the value. • Effective fulfilment and delivery: The supply chain must run more smoothly with an effective fulfilment and delivery. • Integrated portfolio steering: Sales must optimise the product mix and create long term value for our customers from differentiated products. Big data and advanced analytics are promising new technologies to grasp the opportunities and create an excellent value chain spanning across operations, logistics, sales and marketing. Data has become the main lever for additional business value,

and we aim to exploit its value in the FVC programme. We are developing new use cases continuously to apply data and advanced analytics for our value chain challenges, and these include: •Demand forecasting: Forecasting customer demand to optimise strategic priorities and profitability •Capacity forecasting: Forecasting available capacities, helping to make more realistic commitments to customers •Logistics: Predicting order completion to plan the best transport modality and avoid rush transports •Product value management (PVM) Optimising the product portfolio regarding strategic targets, profitability and complexity

FUTURE STEEL FORUM

43 29/08/2019 13:59:48


DATA ANALYTICS

Fig 1. The FVC control tower – comprehensive, interconnected system

•Order fulfilment: Analysing lead times and yields of different products to better steer orders through the supply chain •Virtual mill: Simulating the supply chain (creating a “digital twin”) to understand risks and bottlenecks in our plans and better react to deviations Applying data and analytics requires new data and digital capabilities, that we initially did not have and that had to be developed in parallel to the FVC programme. These comprise of technological capabilities like new tools, data and IT architecture, as well as people capabilities like data engineering, data science, and data governance. We set up a Data & Analytics (DnA) Platform in the Microsoft cloud (see Fig. 2), which consists of a Big Data environment in Hadoop (the “Data Lake”), Informatica data management tooling, advanced analytics and machine learning in Python and R as well as interactive Spotfire dashboards. The Platform connects to our local systems and pulls source data, which is then checked, refined and further transformed to meet the use case needs at company-level and create business insights. DnA supports our ambition to combine company-wide data management with analytics capabilities. It allows to develop a full range of decision making tools and extract business value with trustworthy data as the basis. It is “one source of the truth” for

44 Svend lassen done.indd 4

FUTURE STEEL FORUM

reporting, business planning and advanced analytics. By managing the data, we also break the compromise of supporting use cases quickly and building on consistent data. In addition, a new Reporting & Analytics (R&A) function was set up to support the FVC programme and drive the data-driven transformation (see Fig. 3). Existing teams dealing with data, reporting and master data management were transferred into R&A. These were complemented with new data and digital talents for data engineering, data science, digital planning, and data governance, that were recruited externally. Both groups are making up roughly half of the team.

R&A has been established as a support function on the business side, as it requires business expertise to understand the data and extract its value. R&A members are part of the FVC use case teams and are working on user stories in their specific fields of expertise. R&A is also driving its own initiatives to build data and digital platforms for the business, manage master data, and apply data governance, which require a good planning of the interdependencies and strong IT support. The interaction with IT is thus critical, when it comes to underlying IT systems and architectures. All digital initiatives are following agile methodology for managing the speed and complexity of the digital change, tailoring solutions to customers' needs and creating the user experience, applying digital and data for business value, and improving the collaboration across functions. Agile working supports digital by aiming for quick results and value in developments (“minimum viable product”), developing tools in iterations with frequent feedback from users, empowering the teams to take decisions and adapt the direction, and working in cross-functional teams closely together (co-location). Summarising, FVC has been a significant investment in building new capabilities, navigating Tata Steel Europe in the digital journey and making customer focus a reality. FVC has been driving differentiation and customer satisfaction improvement with higher service and lower arrears, while capturing higher double-digit million Euro EBITDA effects and medium double-digit million Euro cash impact over three years.

Fig 2. Data and Analytics (DnA) Platform in the cloud

Steel Times International

29/08/2019 13:59:50


The supply chain and sales and marketing teams are now equipped with dynamic and interconnected tools enabling: • Demand shaping on strategic priorities • ‘Real time’ agility and flexibility in capacity allocation • Focus on business decisions and interventions • Improved delivery performance to customers The World Economic Forum (WEF) inducted Tata Steel Europe in 2018 into its prestigious community of ‘Lighthouses’, a distinction awarded to manufacturing facilities which are seen as leaders in the technologies of the Fourth Industrial Revolution. Assessing more than 1,000 factories, the WEF recognised Tata Steel state-of-the-art, in having successfully adopted and integrated cutting-edge technologies of the future and driving financial and operational impact. To aid the learning and adoption of technologies by other companies, the

Fig 3. R&A competences and capabilities

Lighthouses companies are committed to share their knowledge with other manufacturing businesses.

1. Head of Reporting and Analytics in Commercial, Tata Steel in Europe. 2. Head of Future Value Chain, Tata Steel in Europe.

Durable Labels & Tags for all Applications

Self-Adhesive Labels High Temperature Tags

Rip-Proof Tags Ambient Tags

DISCUSS YOUR REQUIREMENTS NOW +44 (0) 114 242 2111

Steel Times International

Svend lassen done.indd 5

metals@dura-id.com

www.dura-id.com

FUTURE STEEL FORUM

45 29/08/2019 13:59:51


DIGITAL WORKFORCE

Managing change in the digital workplace Digitisation is disrupting our world and shaping the challenges that engineers in the workforce and newly graduated engineers will need to address in their professional careers. These disruptions are further compounded by the tensions that arise among people, environment, and profit. To successfully and sustainably manage change in the digitised workplace, we suggest that people in the workforce need to continually hone non-technical, career-sustaining competencies. By Farrokh Mistree 1,2 Janet K Allen 2 and Dinsha F A Mistree3

I

n this article, we share our thoughts on the digitisation of the manufacturing enterprise, ramifications of digital transformation on the workforce, upskilling talent in a digitally transforming enterprise, and non-technical, career-sustaining competencies to foster generative learning that is foundational to innovation. Based upon these premises, we suggest seed questions to foster a dialog aimed at identifying a way forward for people to acquire and hone non-technical, career-sustaining competencies that are necessary for success in an enterprise that continues to transform itself. Much has been written about Industry 4.0 and its attendant ramifications on industry, society and the workforce. Without question there has been “hype” foreshadowing significant changes and prospects for social and economic development. Therefore, we could wonder whether the current writings are perhaps just new wine being poured into old bottles. Be that as it may, Darwin’s

46 Farrokh done.indd 2

FUTURE STEEL FORUM

statement rings as true today for industry and society as it does for evolution: “It is not the strongest of the species that survive, nor the most intelligent but ones most responsive to change.” We recognise that people in the workforce need both domain dependent competencies (for example, technical, business, and management) and nontechnical, career-sustaining competencies. In this communication, we suggest five nontechnical, career-sustaining competencies for success in the digitised workplace. It is in this context that we pose the question: What are the non-technical, career-sustaining competencies for success of people in enterprises characterised by the forces released by the adoption of Industry 4.0? Digital transformation of a manufacturing enterprise. Digitised manufacturing is characterised by a digital model of an end-to-end manufacturing network thereby making it possible to transfer autonomy from the physical realm to the cyber-physical

realm wherein the role of the human is critical for the cost-effective and safe operation of the cyber-physical-human (social) system. Digitisation has made it possible: • For an IoP to communicate with cyberphysical systems (machines, devices, sensors) and each other via the IoT and innovate. • For cyber-physical systems to perform tasks as autonomously as possible with a human in the IoP getting involved in resolving exceptions, interferences, or conflicting goals. • For an IoP to be supported by assistance systems anchored in a virtual copy of the physical world (digital facility models enriched with sensor data). Assistance systems are designed to provide appropriate information for humans in the IoP to make informed decisions to resolve critical problems on short notice in an IoP where decisions are decentralised. The leadership of any manufacturing enterprise must not only worry about productivity, efficiency, and cost-effectiveness, Steel Times International

28/08/2019 14:33:47


but it must also focus on innovation. It must respond to new and different societal pressures, usually while internalising new environmental disruptions and addressing social justice concerns. To survive in such a difficult business climate, a contemporary manufacturing enterprise must embrace digital transformation for it to innovate and survive. What does digital transformation mean for the future workforce? Clearly, with more automation, there is going to be a shift in manufacturing from operational/ reactive problems (firefighting) to proactive, system-focused approaches where potential problems are predicted and addressed before they become actual problems (policing). Undoubtedly some firefighters will always be required, but being able to solve a problem will not be as valuable a skill as being able to identify a systemic flaw at a stage before it becomes an actual problem. Not only do members of the IoP have to have the skills to do firefighting, but they also must be trained to do policing. Members need to become experts at sandboxing, piloting, trialing, debugging code, etc. If one takes the Boeing 737 Max disaster, for instance, the failure was not just with the plane crashes themselves. It had more to do with the fact that the code had not been properly tested with a representative set of human pilots; it did not help that Boeing executives initially refused to acknowledge that they had a systemic problem. Boeing has spent a lot to fix the code over the past few months (firefighting), but that cost will pale in comparison to the resources that the company will spend in the coming years and decades to make its software testing processes more robust (policing). There will be organisational resistance to change as well, which means that in a policing world conditioned by increasing automation, non-technical (soft skills) competencies (for example, ability to continue learning coupled with systems thinking) will be even more essential for members of the IoP. For Peter Senge, real learning gets to the heart of what it is to be human. Through learning we can re-create ourselves . This applies to both individuals and organisations. Thus, for a learning organisation it is not enough to survive. Survival learning or what is more often termed adaptive learning is Steel Times International

Farrokh done.indd 3

TO PROVIDE CONTEXT FOR THE ENSUING DIALOG, WE OFFER THE FOLLOWING DEFINITIONS OF SOME KEYWORDS AND PHRASES WE HAVE USED IN THIS COMMUNICATION: Autonomy is a state in which a robot or piece of equipment operates independently, without explicit instructions from a human. https://www.oqton.com/zh/from-automation-to-autonomy/ Automation is a set of human-defined functions performed by a robot or piece of equipment. https://www.oqton.com/zh/from-automation-to-autonomy/ Competence is the ability to perform a specific task, action or function successfully. Incompetence is its opposite. http://en.wikipedia.org/wiki/Competency Competencies are the result of integrative learning experiences in which skills, abilities, and knowledge interact to form bundles that have currency in relation to the task for which they are assembled. http://nces.ed.gov/pubs2002/2002159.pdf Digitisation is the transformation from analog to digital or digital representation of a physical item with the goal to digitise and automate processes or workflows. Digital Business is the creation of new business designs by blurring the digital and physical worlds. It promises to usher in an unprecedented convergence of people, business and things that disrupts existing business models - even those born of the Internet and e-business eras https://www.forbes.com/sites/gartnergroup/2014/05/07/digital-business-iseveryones-business/#660921477f82 Digital Transformation is the novel use of digital technology to solve traditional problems. These digital solutions enable - other than efficiency via automation - new types of innovation and creativity, rather than simply enhance and support traditional methods. Success is anchored in both digital business and digitization. https://cn.bing.com/search?q=https:// en.wikipedia.org/wiki/Digital_ transformation&PC=MENEPB Internet of Things is the interconnection via the internet of computing devices embedded in everyday objects, enabling them to send and receive data. Internet of People (IoP) is associated digitally with a cyber-physical system. This network is responsible for resolving exceptions, interferences, or conflicting issues in a cyberphysical system that is vested with autonomy.

FUTURE STEEL FORUM

47 28/08/2019 14:33:51


DIGITAL WORKFORCE

“It is not the strongest of the species that survive, nor the most intelligent but ones most responsive to

change.

CHARLES DARWIN important – indeed it is necessary. But for a learning organisation, adaptive learning must be joined by generative learning, learning that enhances our capacity to create. Equally important is that only those in the workforce who can recognise and respond to change brought about by digitisation will succeed. We suggest that foundational to being adaptable is generative learning. Upskilling talent in a digitally transforming enterprise. Enhancing talent involves improving the skills of a person, usually through training, so that they will be better at firefighting and policing. Typically, in a manufacturing enterprise of yore, Talent could be expressed as follows: Talent = Experience + Skills + Training However, for a digitalised manufacturing enterprise, we suggest that the preceding expression for Talent is inadequate. A person in a digitally transforming enterprise, in addition to being able to adapt to advances in technology, needs to be able to communicate and relate to people (from different disciplines, cultures, values) who may not be co-located. We suggest that Talent in a digitally transforming enterprise (Talentdt) can loosely be expressed as follows: Talentdt = Talent + Generative Learning Although people can be trained to use new technologies and how to communicate and relate to other people, we cannot teach people how to learn, unlearn what is no longer relevant and relearn that which is needed. We can, however, provide an opportunity for people to learn by reflecting on doing. Through learning, unlearning and relearning people can recreate themselves.

48 Farrokh done.indd 4

FUTURE STEEL FORUM

Generative learning is learning that enhances our capacity to innovate and create. Non-technical, career-sustaining competencies for generative learning. We suggest that the following non-technical competencies are necessary for generative learning that is foundational to innovation: 1. The ability to learn, unlearn and relearn through reflection on doing and the associated creation and articulation of knowledge. 2. The ability to speculate about the future and to identify gaps in state-of-theart or state of practice in ways that may lead to innovation. One cannot innovate without being able to identify gaps that are foundational to generative learning. 3. The ability to ask questions, actively listen, reflect and identify opportunities worthy of further investigation. Questions embody gaps in a person’s knowledge and the answers may lead to generative learning. 4. The ability to make decisions to move forward using incomplete information. Innovation involves risk. Generative learning has utility only when a decision is made to move forward. 5. The ability to assess and think critically (deductive reasoning and inductive speculation) and identify a way forward. The ability to think critically is foundational to moving forward on a technical project and in identifying what one needs to learn next. For learning to start, a person must know what he/she does not know and is motivated to find out. There are two foci, one that is external to the individual and the other that is internal. When focusing internally the individual, through critical introspection, identifies what competency he/she needs to develop further and why, and then identifies a way forward. The external focus takes the form of diagnosing what needs to be fixed in a system (product), the root cause, and then a way forward. Closing Remarks In this communication, we offer some thoughts on enhancing the non-technical, careersustaining competencies that are essential for generative learning of individuals in an IoP in a digitally transforming learning organisation. In keeping with our desire to foster discussion we offer the following questions to critically evaluate what is presented in this communication:

1. How should the description of a digitally transforming enterprise in this communication be enhanced? 2. How should the proffered nontechnical, career-sustaining competencies to foster generative learning in people constituting the IoP, be modified to cover what is missing in our description of the digitally transforming enterprise? 3. How can we ensure that people constituting the IoP, through generative learning, continue to improve productivity, efficiency and cost-effectiveness in a learning organisation through innovation and under societal pressure to integrate sustainability and social justice with technical efficiency? 4. What is needed to motivate the people that constitute the IoP to join/form a generative learning organisation? 5. How should a digitally transforming enterprise facilitate people constituting the IoP to develop the non-technical, careersustaining competencies? We repeat Charles Darwin’s prescient statement “It is not the strongest of the species that survive, nor the most intelligent but ones most responsive to change.” Recognising that there is a transformation taking place in the workplace and a need to adapt, let us ponder the question What are the nontechnical, career-sustaining competencies for the success of people in enterprises characterised by the forces released by the adoption of Industry 4.0? Acknowledgments The authors are grateful to Reza Alizadeh, John K. Antonio, Xun Ge, John Hall, Sherri Irvin, Kristen Lurie, Jelena MilisavljevicSyed, Behram Mistree, Xing Qingqing, Beheruz Sethna and Randa L. Shehab for their encouragement and comments as we developed this communication. JKA and FM acknowledge the financial support from the John and Mary Moore Chair and the L.A. Comp Chair, respectively.

1. Corresponding author. Email farrokh.mistree@ou.edu 2. Systems Realisation Laboratory, University of Oklahoma, Norman, Oklahoma 73019, USA 3. Rule of Law Programme, Stanford Law School, Stanford, CA 94305 USA Steel Times International

28/08/2019 14:33:54


CYBERSECURITY

Automation security in the era of Industry 4.0 Cyber-attacks on European steel producers have become a real threat, where massive production breakdowns are rare, but very expensive events [1]. Moreover, a new threat has emerged that was not a key focus of cybersecurity considerations in the past: the slow and latent degradation of production processes by intentional sabotaging software on the automation systems. By Marcus J Neuer[1], Martin Kretschmer [2] and Andreas Wolff [3]

C

ompanies prepare for these problems, often only focusing on the information technology (IT) layer, where firewalls, internet restrictions or staff compliance rules are enforced to secure the on-premise networks. In contrast, the so-called operational technology (OT) comprises all relevant computer technology, hardware and software related to production machinery. While the transition between both is sometimes smooth and floating, IT and OT can be divided by some key requirements: For IT components, hardware and software are frequently updated, both are easy to patch and can be obtained from any online or local electronics store. IT is fast living, with rapid ageing. There is rarely a human safety threat, if IT components fail. They can be rebooted at any time. OT components, however, control complex production processes, including the safety systems of machinery. They are not allowed to fail and cannot be rebooted during the operation time. Moreover, these components grow old, are rarely replaced and form complex brown-fields with heterogeneous mixtures. Both hardware and software are rarely patched. Breaches in IT security have been reported with increasing frequency. The virus of choice in these incidents is mostly ransomware. It encrypts vital parts of the computer hard disks and then asks the company victim to pay money in order to regain access Steel Times International

BFI.indd - query 1st column page one....indd 3

to their systems again. Spreading across a whole company intranet, wide parts of the infrastructure are practically disabled for a long time. Process downtimes and costly maintenance to the network become necessary. Economically, the halted production leads to dramatic financial deficits. While these breaches are quite different to attacks on the OT layer, they show the relevance of the threat and the lack of IT security to prevent such attacks. If ransomware can reach the IT of larger process industry companies, so can attack software reach the OT systems of modern process plants. Security dilemma of the OT systems For the IT sector, several tools exist that can be used for analysing attack vectors, performing penetration tests (pen-tests) and hardening the IT infrastructure. Among these tools, Kali Linux and the Parrot Security OS are only two examples. Pen-testing and the rising field of ethical hacking are offered as a service to customers. The latter helps to explore vulnerabilities and exploits that might endanger the computational systems for industrial companies. Yet, deep knowledge and skills in this domain do not belong to the common portfolio of an IT department. But as such tools exist, the principal risk for IT systems can be reduced by an educated pen-testing strategy on a regular basis together with a strict policy for external data drives for the personnel.

A much more important and urgent problem is that there are currently no tools for reliable OT security analysis. The reason for this is simple: the main way to detect attacks within the OT layer is by carefully checking the process. Thus, detection requires a solid understanding of the underlying production processes and this type of expertise is not found with software companies selling security programs for the IT layer. Beyond the actual attack scenarios, a more elaborated and politically involved problem is also currently emerging: hardware chipset designed for industrial espionage. Such systems may be able to interact with any production processes and the number of experts, able to identify dangerous hardware parts is quite limited. Why does Industry 4.0 accelerate the threat? The described threats are, of course, fundamentally linked to Industry 4.0 and its ongoing endeavour for integration. For years, new integration concepts have been propagated. The core of these activities is to foster communication between the automation layers and from the OT to the IT components. Potentially harmful code, once breaking through the IT barriers, is, therefore, capable of spreading across the OT systems very effectively. The well-known automation pyramid hierarchy is de-facto broken by any FUTURE STEEL FORUM

49 29/08/2019 09:11:11


CYBERSECURITY

These are just two examples to illustrate, how easy such attacks interfere with the production processes.

integrative approaches: sensors from the lower levels can directly exchange data with higher level manufacturing execution systems, often also software for the enterprise level [2], [3] . Production deterioration instead of destruction A key strategy of intentional cyberattacks is to disturb production rather than stop it or destroy machinery. The reasons for this are simple: If machinery is destroyed, the cyber attack is exposed and the inflicted damage is high, but locally restricted. However, if small but permanent damage is inflicted over time, the production of a specific company deteriorates without the attack being detected. In this case, the product quality is reduced, the processes require more maintenance and the whole plant requires more money to sustain operation All these factors will lead to lower customer acceptance, bad reputation for the products and finally a loss of market share. Attacks can be as simple as putting an arbitrary (negative or positive) offset value on a temperature sensor value, either to render products too warm or too cold when they leave a furnace. Too cold products at the roughing mill lead to a higher rate of products that fail to reach their target thickness. Excessive heat in the furnace distorts the steel recipe and leads to higher energy consumption than is needed for the process.

50

Who performs such attacks? While modern IT viruses are mostly due to misled programming talents trying to get money through the use of ransomware, OT attacks require a profound understanding of the process and its automation. Potential OT attackers are consequently different to IT virus developers – explicitly they are not people who do this for fun, they are experts in their field, equipped with the necessary knowledge on how to drive, for instance, a rolling mill slightly out of the normal operation. Detection strategies Detecting attacks on the automation is a difficult task. Often, attacks are hidden in the signal noise, exploiting the zero-dynamics of the control system, driving the optimal production set point to a less optimal, perturbed value. Using replay strategies, the operators are not aware, that such an attack is ongoing, because they only see the process variable faked by the attacker, mimicking a completely normal state of operation. For those replays, the original process variables have been recorded for some time by the attack software. At a certain point in time, the attack starts by exchanging the true sensor and actuator values of a control loop with the recorded historical data. In parallel, hidden from the operator, the attack alters the input to the controller so that a damaging impact is achieved in the process. Either this leads to a straight destruction of machines or a crawling and latent erosion of the process quality. Detection of fault states and anomalies[4] in process data streams have made quite some progress over the past few decades[4], [5] . Anomaly detection exploits methods from machine learning [7], especially concentrating on unsupervised learning methods like Autoencoders [6] or Restricted Boltzmann Machines (RBM) [7], [8]. They learn the “normal state� as thoroughly as possible.

Both techniques reduce the dimensionality of input data streams to a much smaller set of numbers, key quantities to decide whether the system behaves normally or not. Such approaches can also help to detect attacks, assuming that a sufficient amount of data is available that corresponds to the normal state of operation. Automation systems can also be actively probed. Imagine that the design of the controller is not a black box and the mathematical functionality is completely known. If any attack exploits the dynamics of the controller, a dedicated scanning of the system may include a rapid change of the controller for a very short duration of time. This change would not occur in the replayed signal stream. If the replay approach is more complex, the system answer of this sudden change may then reveal the attack, as the replayed signal dynamics and the predicted dynamics would no longer match. Active probing can also utilise an artificial signal stream that is added to the original data. This artificial signal is designed in a specific way to identify the presence of attacks. Unfortunately, all these approaches need intense application testing, to make sure that the detection method itself does not harm the process. In almost all described attack schemes, it is wise to drive the process into some emergency state, once the attack is detected and verified. AutoSurveillance In a recently started research project called AutoSurveillance, funded by the Research Fund for Coal and Steel (RFCS), an international consortium led by the Betriebsforschungsinstitut (BFI) in Germany offers a structured approach to improve the cybersecurity of automation systems. First, it reviews and designs attack strategies on

FUTURE STEEL FORUM

BFI.indd - query 1st column page one....indd 4

29/08/2019 09:11:17


the steel process chain. Then, based on a proper understanding of the attack vectors and longstanding expertise in advanced process control, the project will work out novel detection algorithms to find stealth attacks. The central objectives of this research project are to fill the identified gap for security surveillance of OT systems and to provide reliable means of detecting attacks on automation systems. Bibliography [1] "Hack attack causes 'massive damage' at steel works," BBC, 2014. [2] M. J. Neuer, A. Ebel, F. Marchiori, M. Rößiger, N. Matskanis, G. Mathis and A. Wolff, "Raising economic efficiency of steel products by a smart re-allocation respecting different process routes," in Proceedings of the European Steel Technology and Application Days, Düsseldorf, 2015. [3] M. J. Neuer, A. Ebel, F. Marchiori, N. Holzknecht and J. Brandenburger, "How-tooptimize Steel Quality by Applying Industry 4.0 Techniques in Real-World Examples," in Proceedings of the European Steel Technology

and Application Days, Vienna, 2017. [4] D. K. Gaud, P. Agrawal and P. Jayaswal, "Fault diagnosis of rolling element bearing based on vibration and current signatures: an optimal network parameter selection," in Intl. Conf. on Electrical, Electronics and Optimization Techniques (ICEEOT), Chennai, India, 2016. [5] Z. Gao, C. Cecati and S. Ding, "A survey of fault diagnosis and fault-toleratn techniques part I: Fault diagnosis with model-based and signal-based approaches," IEEE Transactions on Industrial Electronics, vol. 62, pp. 3757-3767, 2015. [6] Z. Gao, C. Cecati and S. Ding, "A survey of fault diagnosis and fault-tolerant techniques part II: Fault diagnosis with knowledgebased and hybrid/active approaches," IEEE Transactions on Industrial Electronics, vol. 62, pp. 3768-3774, 2015. [7] T. Matsuura, "An application of neural network for selecting feature parameters in machinery diagnosis," Journal of Materials Processing Technology, pp. 203-207, 2004. [8] G. E. Hinton and R. R. Slskhutdinov, "Reducing the Dimensionality of Data with Neural Networks," Science 313, 28 July 2006.

[9] G. E. Hinton, "Learning multiple layers of representation," Trends in Cognitive Sciences Vol. 11(10), Elsevier, pp. 428--434, 2007. [10] G. E. Hinton, S. Osindero and Y.-W. Teh, "A fast learning algorithm for deep belief nets," Neural Computation 18, pp. 1527-1554, 2006.

1. Head of Department, Automation Downstream, 2. IT manager, 3. Senior expert, Betriebsforschungsinstitut (BFI)

We create benefit for industry!

Swerim conducts needs-based industrial research and development concerning metals and their route from raw material to finished product. We wish to strengthen industrial competitiveness by enabling improved product quality, greater resource efficiency and more sustainable manufacturing processes. Our vision is a fossil-free and circular industry. www.swerim.se

Steel Times International

BFI.indd - query 1st column page one....indd 5

FUTURE STEEL FORUM

51 29/08/2019 09:11:21


SMART MANUFACTURING

Designing smart networked manufacturing systems With a view to fostering discussion on designing smart networked manufacturing systems, the authors examine the following: the latest trends in manufacturing, bridging physical and cyber worlds, the benefits of Industry 4.0, an architecture for the realisation of the Fourth Industrial Revolution as a digital thread and a digital twin, addressing the real industrial problem via the DFDM Framework, and ways forward and plans for the digital transition of industry. By Jelena Milisavljevic-Syed1 , Janet K. Allen2 , Sesh Commuri3 , and Farrokh Mistree2

I

n global markets characterised by dynamic changes and necessitated by changes in customer preferences, timely adjustments are required in manufacturing to meet fluctuating demands. Conventional manufacturing processes are designed for mass manufacturing and are not suited for agile, flexible and highly reconfigurable smart manufacturing. Foundational to the Industry 4.0 construct is digitisation that should be harnessed for smart manufacturing. Hence, we propose designing smart Networked Manufacturing Systems that embody Design for Dynamic Management (DFDM). The DFDM framework is a service-oriented product/system development computational framework that facilitates the harmonisation of conflicting needs associated with the design and manufacture of engineered systems. The trend Bringing together technologies such as Internet of Things (IoTs), Big Data Analysis, Machine Intelligence with conventional technologies such as Smart Automation, Supply Chain, Logistics, and Cloud Computing has resulted in a new wave of advances in Manufacturing Technology collectively called Industry 4.0[1]. Industry

52

FUTURE STEEL FORUM

UNI LIVERPOOL.indd – Read MM, seems alright....indd 2

4.0 represents the Fourth Industrial Revolution and provides a framework for digitization of manufacturing. Connecting physical and cyber worlds Industry 4.0 is characterised by a digital model of an end-to-end supply network and all the manufacturing processes. It provides a mechanism to transfer autonomy from the physical realm to the cyber-physical realm [2] . Central to the adoption of Industry 4.0 is the concept of the ‘Digital Twin’. The Digital Twin is a virtual replica of every entity in the manufacturing process and allows the bridging of the physical and cyber worlds. Such a bridge allows for seamless data exchange in real-time between the physical world and the virtual representations. Since the digital twin represents a high-fidelity model of the physical processes, it can be used to analyse the overall performance of the physical system, diagnose performance, and invoke data analytics and learning strategies to improve system performance. The benefits To appreciate the benefits of Industry 4.0 it is essential to realise the full value chain that

includes the supply networks, customer and markets. Digitisation of processes across the value chain enable new capabilities such as personalisation, real-time alerts and interventions, innovative service models, dynamic product improvement, increased productivity, higher up-time and, ultimately, new business models. Architecture for realisation of digitised manufacturing in Industry 4.0. Digitisation of manufacturing systems that conform to the Industry 4.0 construct is not a trivial task and can only be realised through the adoption of an appropriate computational architecture; see Fig. 1. The architecture must facilitate the integration of digital threads in two ways. The vertical integration of a digital thread facilitates the flow of information between the enterprise and resource planning entities to manufacturing services and planning and control entities within a manufacturing facility, see Fig. 1. In the vertical integration low-level components such as IoT sensors, machine controllers and their digital twins across the manufacturing enterprise are networked with big data analytics to monitor the manufacturing Steel Times International

28/08/2019 14:42:33


process and improve system performance and product quality. Vertical integration also allows for the creation of interoperable ‘systems of systems’ that can integrate a seemingly diverse set of machines, sensors and controllers to produce a resilient manufacturing process that is capable of withstanding disruptions in the supply network, environment and in the market. Horizontal integration deals with data and material flows across the end-to-end value chain, see Fig. 1. The horizontal integration is represented by several digital twins from the supply chain and the manufacturing processes, data flows and IT systems in the product development, to logistics, distribution and ultimately the customer. Elements of the architecture are expanded on in the context of digitised manufacturing processes in the steel industry and are discussed next.

Scenario 1 If we have an ideal situation, a process without errors where models used to represent the process are complete and accurate, and all data are known, then we have an ideal manufacturing process. The outcome is a product of acceptable quality. Scenario 2 A lack of thermal and chemical homogeneity results in an error associated with the reduced cleanness of the steel [4]. Such an error may appear during continuous casting when the molten metal is in the tundish, the second column in Fig. 2, due to the inability to maintain superheat. Further, due to the characteristics of a networked manufacturing system (NMS) these errors will propagate through the remainder of the process and Steel Times International

UNI LIVERPOOL.indd – Read MM, seems alright....indd 3

Fig 1. Architecture for realisation of digitised manufacturing in Industry 4.0

affect the final product quality[5]. The outcome is a product of unacceptable quality. Scenario 3 If a processing error that appears in Stage 1 cannot be fixed, then we can make improvements in subsequent stages to ensure a product of acceptable quality. The way to do it is to design a system to make changes by identifying the location and cause of

the error, and then take steps to correct the error. As a result, the process rebounds and continues within specified tolerances. The outcome is a product of acceptable quality. Scenario 4 Since mathematical models that are used to simulate the process are incomplete and inaccurate, we must manage the uncertainties embodied as a result in the

I

II Scenarios

Digitised manufacturing A steel manufacturing process consists of number of unit operations, such as ladle refining, tundish processing, continuous casting, rolling, heat treatment, etc., where each requires proper attention and control [3]. A representation of the different operations involved in the production of sheets (product) from slabs (semi-product after casting) is shown in Fig. 2. Each of the operations in the process is performed within specified tolerances, outlined in five scenarios in Fig. 2. We describe different scenarios, (see Fig. 2) that may occur in a steel manufacturing process.

III

IV

V PI: Processing issue UI: Uncertainty issue DM: Dynamic management EQ: Excellent quality UQ: Unacceptable quality

Fig 2. Problem description - design and analysis of steel manufacturing process

FUTURE STEEL FORUM

53 28/08/2019 14:42:34


SMART MANUFACTURING

Key: SoV Stream of Variations cDSP Compromise Decision Support Problem OA Operability Analysis MTC Minimum Time Control DT Decision Tree GT Game Theory

Adaptability

• SoV • cDSP

• Flexibility in selection and determination of design parameters

Fig 3. Computational framework, design for dynamic management

• Concurrent design • Solutions as cost-quality tradeoff • Diagnosable and controllable

Design Computational • Design framework • skills

Reconfigurability

• DT • cDSP • GT

Strategy to reconfigure system through configuration design of physical system followed by configuration of control system

Decision-based design where design thinking, strategy and innovation management is integrated to design system adaptable to dynamic changes in the market

mathematical control algorithms. Uncertainties may appear due to inherent randomness or unpredictability of a system, model parameters uncertainty, and model structure uncertainty[6]. Hence, we need to design the system to be operable, to manage uncertainty, transit and stabilise in the presence of change and get back on track within specified tolerances. Scenario 5 If critical errors appear that cannot be fixed, we can design a system to observe, detect but not to mitigate errors nor manage uncertainty. In this case the decision is to reconfigure the system and not proceed with this process [7, 8]. In the context of the preceding we identify key requirements for a computational framework for a smart networked manufacturing system. Computational framework A smart networked manufacturing system (NMS) must be adaptable to dynamic changes and respond to unexpected disturbances, and uncertainty. Additionally, fault-tolerance and robustness to disruptions are critical requirements that must also be addressed in the design. Maintaining connectivity among elements of the process is one of the key characteristics necessary for these systems to be integrated into the architecture. Ultimately, the long-term viability of the system depends on the capability to glean information from the various data

54

FUTURE STEEL FORUM

UNI LIVERPOOL.indd – Read MM, seems alright....indd 4

Operability

• OA • cDSP • MTC

• Analyze system operability undergoing dynamic changes

• Analyze dynamic performance of system undergoing dynamic changes

streams in the process and learn in realtime to optimise and manage the process. This information is also necessary for the long-term health and resilience of the system. Current manufacturing systems have been designed for large-scale manufacturing of a product in the most efficient manner. Hence, we advocate the design of smart NMS that can adapt to rapid changes in product requirements, a system failure occurs because of a breakdown of interconnectivity between elements of the NMS that existing approaches cannot easily handle. Foundational to a smart NMS is a computational framework, Design for Dynamic Management (DFDM), see Fig. 3. DFDM has three critical components:

adaptable and concurrent design, operability analysis and reconfiguration strategies [9]. Adaptable and concurrent design methods offer flexibility in selection of design parameters and the concurrent design of the mechanical and control systems [10]. Operability analysis is used to determine the functionality of the system undergoing dynamic change [11]. Reconfiguration strategies allow multiple configurations of elements in the system. The limitation of the DFDM is that we cannot design processes to mitigate errors when the processing takes place at a station, but the effect of the error in the process is only seen in subsequent stations. Way forward We contemplate expanding DFDM to address feed-forward dynamic management, Fig. 4, where tools and sensors communicate with each other via the Internet of Things (IoT), and sensors data is used to create enriched digital system models to achieve process control, advanced measurements, perfect product, and fast process development. With feedforward dynamic management we are able to use information picked up by sensors (Step 1 in Fig. 4), to explore the solution space (Step 2 in Fig. 4), manage the process through smart design through decision-making process (Step 3 in Fig. 4), capture and reuse the knowledge with off-line model (Step 4 in Fig. 4) for future design or reconfiguration of the process (feed forward control of the process, Step 5 in Fig. 4). In the next iteration we will obtain

Fig 4. Feed-forward Dynamic Management

Steel Times International

28/08/2019 14:42:35


new information (feedback learning, Step 6 in Fig. 4) that will be further captured and reused. This is an iterative cycle of exponential learning that is foundational to incorporating intelligence in feed-forward dynamic management of NMS. The resulting outcome is a next generation of smart networked manufacturing systems adaptable to rapidly changing market requirements, resulting in higher quality products and at a lower rework cost. To transition this framework to industry we suggest that DFDM be extended to a Smart Platform, Fig. 1. Some possible extension of DFDM in various industries are illustrated in Fig. 5: • Cyber-physical design. Model-based intelligent decision-based design systems. We speculate that there are applications in steel making processes. • Cyber-physical-social system design. Model cyber-social design decision network in the realisation of intelligent cyber-physicalsocial systems. We speculate that there are applications in smart healthcare systems. • Cyber-physical product and material design. Integration of materials, products, and manufacturing processes. We speculate that there are applications in the design of smart sports equipment. • Cyber-physical manufacturing. Design of reconfigurable intelligent manufacturing systems. We speculate that there are applications in additive manufacturing, and semiconductor lithography processes. Closure In furtherance of Industry 4.0 we advocate the realisation of smart Networked Manufacturing Systems. We suggest that Design for Dynamic Manufacturing is foundational to the realisation of smart NMS. In the International System Realisation Partnership (ISRP) we are creating a Cloud-based Design and Manufacturing (CBDM) platform with potential applications in cyber-physical design, cyber-physical manufacturing, cyber-physical product and material, and cyber-physical-social system design. Further, we speculate that CBDM will be transitioned to industry and be applied in the steel making processes, additive manufacturing, design of smart healthcare systems and so on. For more information please visit: www.liverpool.ac.uk/ engineering/staff/jelenamilisavljevic-syed. Steel Times International

UNI LIVERPOOL.indd – Read MM, seems alright....indd 5

CURRENT STATE

WAY FORWARD Cyber-physical design

Cyber-physical-social systems design

DFDM

SMART PLATFORM

Fig 5. Extension of DFDM

References [1] Xu, L.D., Xu, E.L and Li., L., 2018, “Industry 4.0: State of the Art and Future Trends,” International Journal of Manufacturing Research, vol. 56, no. 8, pp. 2941-2962. [2] Monostori, L. “Cyber-Physical Manufacturing Systems: Roots, Expectations and R&D Challenges, in Variety Management in Manufacturing,” 47th CIRP Conference on Manufacturing Systems, Procedia CIRP 17, 2014, pp. 9 – 13. [3] Shukla, R., Anapagaddi, R., Singh, A.K., Panchal, J.H., Allen, J.K. and Mistree, F., 2015, August. Exploring the Design Set Points of Refining Operation in Ladle for Cost Effective Desulfurization and Inclusion Removal, ASME Design Automation Conference, Boston, MA, August 2-5. Paper Number DETC2015-46265. [4] Anapagaddi, R., Shukla, R., Goyal, S., Singh, A.K., Allen, J.K., Panchal, J.H. and Mistree, F., 2014, August. Exploration of the Design Space in Continuous Casting Tundish, ASME Design Automation Conference, Buffalo, New York. Paper Number DETC2014-34254. [5] Ding, Y., Ceglarek, D., and Shi, J., 2002, “Design Evaluation of Multi-Station Assembly Processes by Using State Space Approach”, ASME Journal of Mechanical Design, vol. 124, no. 3, pp. 408-418 [6] Allen, J.K., Seepersad, C.C. and Mistree, F., 2006, “A Survey of Robust Design with Applications to Multidisciplinary and Multiscale Systems,” ASME Journal of Mechanical Design, special issue on Risk-based and Robust Design, Editors S. Azarm and Z. Mourelatos, vol. 128, no. 4, pp. 832-843. [7] Wang, G., Shang, X., Yan, Y., Allen, J.K., and Mistree, F., 2018, “A Tree-Based Decision Method for the Configuration Design of Reconfigurable

Cyber-physical product and material

Cyber-physical manufacturing

Machine Tools”, Journal of Manufacturing Systems, vol. 149, pp. 143-169. [8] Shang, X., Milisavljevic-Syed, J., Wang, G., Allen, J.K., and Mistree, F., 2018, “A Key Feature-Based Method for Configuration Design of Reconfigurable Inspection System”, Journal of Intelligent Manufacturing, JIMS-D-18-00505, under review. [9] Milisavljevic-Syed, J., Allen, J. K., Commuri, S., and Mistree, F., 2019, “Design of Networked Manufacturing Systems for Industry 4.0”, 52nd CIRP Conference on Manufacturing Systems, Ljubljana, Slovenia, June 12-14, 2019. Paper Number PROCIR-D-18-00262. [10] Milisavljevic-Syed, J., Allen, J.K., Commuri, S., and Mistree, F., 2018, “A Method for the Concurrent Design and Analysis of Networked Manufacturing Systems under Uncertainty,” Engineering Optimization, vol. 51, no. 4, pp. 699717. [11] Milisavljevic, J., Commuri, S., Allen, J.K. and Mistree, F., 2018, “Steady-State Operability in Design for Dynamic Management in Realization of Networked Engineering Systems,” ASME Design for Manufacturing and Assembly Conference, Quebec City, Quebec, Canada. Paper Number DETC2018-85864.

1. Corresponding author. j.milisavljevic-syed@liverpool.ac.uk Division of Industrial Design, University of Liverpool, Liverpool, L69 3GH, UK. 2. Systems Realisation Laboratory, University of Oklahoma, Norman, Oklahoma 73019, USA. 3. Department of Electrical and Biomedical Engineering, University of Nevada, Nevada 89557, USA. FUTURE STEEL FORUM

55 28/08/2019 14:42:36


DRONE TECHNOLOGY

Getting more out of drones

The use of drones in an industrial environment is becoming widespread for numerous applications like structural inspection, quantification of raw material stocks and hazardous area monitoring. By Gabriel FRICOUT1, Philippe GOURDAIN2, Victor MONCADA3.

T

he Unmanned Aerial Vehicles (UAV) technology has been evolving quite fast in recent years leading to more robust and stable carriers, with more autonomy and heavier payloads, which, in parallel to the progress of sensor technology (camera, localisation devices…) and more generally embedded electronics, opens the door to acquiring much more data during a drone flight. Following this trend, the use of thermal cameras on drones have started to spread heavily with many possible applications in industrial environments. In the last years, ArcelorMittal’s Global Research and Development division (Global R&D) coordinated a research project to explore those opportunities. The target of the project was twofold: • Investigate the potential of the drone technology combined with thermal imaging at ArcelorMittal factories. • Specify and develop the missing tools for interpreting easily the results of a drone measurement campaign.

Global R&D has been able to value its strong know-how in optical pyrometry coming from more than 20 years of hot-steel temperature monitoring in many harsh conditions (varying product emissivity, noisy, polluted environment, multi-reflections…), as well as its experience in industrial measurement in a wide sense to optimise the use of the data acquired during the flight (using automatic processing in particular). This work has been carried out in strong partnership with StudioFly Technologies on the one hand, and ThermaDiag on the other hand. StudioFly Technologies is a drone company with many flight references in industrial environments and a strong expertise in embedding dedicated sensors, including thermal cameras, on drones. ThermaDiag is a spin-off start-up from CEA/Cadarache that provides expertise for the monitoring of high temperature processes and associated software that have been used as a basis for the new software developed according to the specifications mentioned above.

In the first section of this paper, some typical industrial use-cases will be described and used to derive specifications for an automatic interpretation software, that will be discussed in the second section. Through those specifications, ArcelorMittal

Industrial use-cases for UAV based thermal imaging As a first step for investigating the potential of infrared inspection in industrial areas, several use-cases have been listed, investigated and totally or partially addressed.

56

FUTURE STEEL FORUM

CARLOS ARCELOR done.indd 2

Hardware provided by StudioFly Technologies All the inspection discussed in the following sections have been carried out by StudioFly Technologies using their own carriers equipped with FLIR cameras for temperature monitoring. The detailed specifications of the carriers and camera can change from one example to another, since some acquisitions have been conducted with several years of intervals, but typical examples are summarised in Fig.1. Pipe network monitoring: The first use of thermal cameras came from the maintenance department of one of our plants where the corking of gas pipe was an important issue. We are considering a transportation network that provides coke gas to various facilities. Dirt from the gas is progressively depositing into the pipes, leading to a progressive decrease of the pipe section available for gas flow. If the corking phenomenon becomes too important, the charge loss becomes too high and gas transportation yield decreases. Because the gas is hot, the surface temperature of the pipe will reflect the thermal resistance of the pipe wall. Over a temperature profile (typically a vertical profile) around the pipe, it is possible to quantify the temperature difference due to the Steel Times International

29/08/2019 14:01:38


Fig1. Typical quadcopter or octocopter from StudioFly Technologies (speed >60km/h, payload: ~1kg, maximum wind speed around 30km/h), FLIR camera PRO R 640 (thermal sensitivity < 50mK)

variation of thermal resistance, that comes from dirt accumulation (see Fig. 2). Flare and hot-stoves monitoring In many situations, the thermal inspection of hot structures can provide relevant information on isolation and refractory states and there has been raised interest in industrial flare monitoring (Fig. 5a). One of the targets was to monitor the good isolation and good functioning of flaps that can switch one flare on and off and the delay for this flap to act when requested. The monitoring of the flare cooling phase can also enable to validate some thermocouples information and internal state of refractory degradation. Inspection of cowpers and pipes to extract gas from blast furnaces have also been identified as valuable use-cases for thermal inspections. Hot points identified in thermal imaging are indicators of refractory damage. They can be identified much more easily using drones than by the use of endoscopic cameras within the pipes which involves stopping and cooling down the installations under scrutiny. Several successive monitoring operations using drones (typically after several years of no drone activity) enable the steelmaker to establish the root causes of the damage and potentially plan maintenance operations going forward (Fig. 3). Product temperature follow-up Apart from installation monitoring, the steel-making process also requires, in many situations, the follow-up of product temperature, typically during cooling phases. This can be a very important issue for mechanical properties and more generally for product quality. The challenge in that situation is the range of temperatures than can be several hundreds of degrees wide. Typical values, for example, could range from Steel Times International

CARLOS ARCELOR done.indd 3

Fig 2. Infrared and visible image of the same area. Differences of temperature between the top and bottom of the pipe is an indication of dust deposit

50-80°C to more than 600°C. In many situations, like the cooling of sheet piles, the products are slowly moving on a table, and it is, therefore, difficult to have fixed sensors to monitor the temperature. Drones can help as they have an aerial view of the cooling table (Fig. 4) and can monitor movements in temperature as they happen without the need for fixed systems (sensors), which are difficult and costly to implement and maintain. In such a scenario, it is very important to consider the emissivity problematics to obtain an accurate temperature value without being impacted by surface artefacts. Common use-case characteristics By addressing all those examples, it has been

possible for ArcelorMittal Global R&D to identify the key functionalities that would be needed to significantly improve the value creation potential of drone flights for the industry. Some functionalities are very general and can be applied to any kind of data obtained during a drone flight, while others are very specific to thermal imaging. The first common characteristic of all use-cases is the very high level of manual processing required to extract quantitative temperature measurements at interesting locations. It is first necessary to review the video stream until the object of interest is visible, and then to use dedicated software where the emissivity of the material can be set to extract a temperature value. This particularly the case when the output of two drone flights have to be compared after an interval of several years: it is necessary to check in each video stream which frames can see the same object. If the same point is visible at several points in the video stream, some different temperature values can also be obtained. This can occur because of a real change in temperature values within the time of the flight, but it can also be the result of multi-reflection artefacts in complicated hot structures (a hot surface emitting a strong radiative energy reflecting on another surface and reaching the camera). This problematic was at the heart of Thermadiag expertise in the monitoring of tokamak reactors. Thermadiag development To decrease the manual interaction with the video stream acquired during the flight, and to progressively reach more automatic processing of those data, it was decided to develop a software implementing the following functionalities: • A 3D model of a scene (corresponding to the flight area) can be loaded and manipulated inside the software FUTURE STEEL FORUM

57 29/08/2019 14:01:41


DRONE TECHNOLOGY

Fig 3. Same location monitoring in 2014 and 2016, revealing a new hot point in the gas pipe

which implements functionalities such as mesh manipulation and pruning. • All flights data, like infrared or visible video stream or any sensor information from the flight, can be loaded into the system. • The trajectory of the drone and its orientation related to the 3D model during the flight is then computed (Fig. 5c). • All pixels of all images of the video stream (visible camera or infrared camera) can be associated to a vertex of the 3D model (Fig. 5b). • Doing so, information can be associated to the same physical position in the 3D models in the same way the information corresponding to the different positions (potentially different viewpoints) of the drone at different moments. All this information is stored (temperature information, light intensity from the visible camera, position of the drone when acquiring these data…) and can be used to estimate more precisely the characteristics of one point of the 3D model. Very simple aggregation functions have been implemented so far, like getting the average value (to reduce noise), or the minimum one (to correct multi-reflections or sun reflection). • The software then displays the resulting data structure as a static 3D model, that can be compared easily to the output of a prior flight mapped onto the same 3D model. To obtain the reference 3D model used to register the video stream, it is possible to use prior drawings of a mechanical, or to conduct an acquisition campaign with a 3D laser scanner (Fig. 5a). At the beginning of the flight, the position of the drone within the

58

FUTURE STEEL FORUM

CARLOS ARCELOR done.indd 4

Fig 4: Sheet piles cooling monitoring

3D model must be assessed, but once this is done, the flight data (including GPS, inertia sensors, etc…) are used to locate precisely the drone in the 3D model at each timestamp. Perspectives and conclusion In this work, ArcelorMittal R&D has brought its expertise in contactless temperature measurement as well as its knowledge of industrial environments to improve both accuracy and usability of UAV temperature measurement, thanks to light reflection correction and 3D mapping. In many industrial situations, getting

thermal information using a thermal camera or a pyrometer is only a very first step toward temperature measurements. Emissivity variations or perturbing environment (like water droplets for instance) as well as multi-reflections from various hot surfaces can strongly perturb the link between the measured signal and the object temperature. The correction of these effects can require advanced data processing, and sometimes even dedicated complementary measurements (like imaging the surface at different wavelengths for instance). The ability to embed the right sensors on dedicated UAV and to reconcile all the information from a 3D spatial point of view instead of a temporal sequence of measurements will open the door to many complex measurements in industrial conditions. In addition, the measured temperature will usually be used for more complex purposes: corking detection, maintenance planning and product quality alerts. The developed software will enable the steelmaker to fully automate a second level of data processing after the 3D mapping for decision making, with limited or without manual manipulation. This will bring the added value of a drone flight at a much higher level compared to a direct and often manual interpretation of a video stream. Acknowledgement and reference Thanks to ArcelorMittal plant’s at Fos sur Mer, Dunkerque and Esch Belval for their assistance.

Gabriel Fricout. 2. StudioFly Technologies. Thermadiag.

1 3

Steel Times International

29/08/2019 14:01:42


Fig 5. Overview of Thermadiag developments a. Global view of the are area with the laser scanner used for building the area 3D model b. Final mapping of infrared data over the 3D model c. Parallel review of the video stream and associated drone trajectory in the 3D model (intermediate information used for the 3D mapping)

a)

b)

c)

performance for high productivity

GSM 200 to capacity and GFM 100 to capacity in operation at NAF, New Castle / PA

GLAMA Maschinenbau GmbH

Headquarters: HornstraĂ&#x;e 19 D-45964 Gladbeck / Germany Fon: +49 (0) 2043 9738 0 Fax: +49 (0) 2043 9738 50 email: info@glama.de

GLAMA USA Inc.

768 W Bagley Road Berea, Ohio 44017 Fon: +1 877 452 6266 Fax: +1 440 201 6900 Email: sales@glama-us.com

glama.de

Steel Times International

CARLOS ARCELOR done.indd 5

GFM 150 to capacity

GIR-P 1 to capacity

in operation at Scot Forge, Spring Grove / IL

in operation at Standard Steel, Burnham / PA

FUTURE STEEL FORUM

59 29/08/2019 14:01:46


GALVANISING LINES

A data-driven galvanising line In 2017, Marcegaglia proposed an original idea to develop a master model to steer a galvanising line, using data coming from pickling, cold rolling and hot dip galvanising (HDG) lines; and asked Fives to co-develop this project. By C Peillon1, P Rocabois2, A Ferraiuolo3, A Fiorini4 and S Pantarotto5.

I

n response to Marcegaglia’s request, Fives proposed a NeoKoil® Smart line, with a view to achieving the following objectives: 1. Improve productivity without degrading quality; 2. Improve quality levels compared to a manual steering; 3. Decrease energy and raw material consumption. NeoKoil Smart line achieves this by: 1.Searching for the optimal maximum speed; 2. Applying automatically quality rules translated from the operators’ practices; 3. Predicting the results, in terms of mechanical properties, coating thickness, surface defects, folds in the furnace and so on, using predictive models. In this context, Fives was asked to develop a machine learning (ML) model to predict

60

FUTURE STEEL FORUM

FIVES.indd - needs job titles on last page....indd 2

the mechanical properties of dual-phase (DP) steel galvanised coils, given its input properties and its production parameters. This prediction model will be incorporated within the master model, orchestrating the whole economical optimisation. This article describes: • the steps taken to develop this machine learning model; • the findings made during this study; • the overall architecture of the solution. Those machine learning models have been built on top of the metallurgic models which have already been developed by Fives in co-operation with Marcegaglia, by using its parameters as input features. It leads to a significant improvement in accuracy compared to the approach solely based on physical models.

Data preparation The data used for this study comes from various sources: • The chemical composition of input coils • The tensile test results of input coils • Operational data from the pickling cold rolling and galvanising line, aggregated at the coil level • Results and intermediate results of the physical model developed by Fives and Marcegaglia • Mechanical properties of incoming material out of the cold rolling mill, performed by TensilPro software (developed by Marcegaglia) • The tensile test results of input coils, which will be our target variable. Marcegaglia received funding from the Italian Ministry of Industrial Development for the specific concept of including data from Steel Times International

29/08/2019 14:04:10


pickling, cold rolling and HDG lines into this decision process (ref. F/150012/00/X40). The data represents 1,000 coils of recently produced DP. Our three target variables are yield strength, the ultimate tensile strength and total elongation. We kept the following variables as potential inputs for our models, as they are either: • Information available before a coil is galvanised • A production parameter which can be directly tuned. An exploratory analysis has been performed at the beginning of the project. During this step, we found out that a significant number of variables present some bias. Those bias represent a challenge to calibrate the physical models properly. Distributions Most machine learning models are sensitive to the input and target variables distribution. To prevent learning issues, those distributions should be: • Centred • Reasonably spread The original probability distribution of some variables shows an important ‘skewness’. For instance, the distribution of the variables below represents a positive skew (the right tail is longer) on ‘Var2’ and a negative skew (the left tail is longer) on ‘Var3’. (Fig. 1) To solve this problem, we transformed the distribution of skewed variables with various transformation methods, improved the symmetry and made the data better fit a centred distribution. Time series analysis We studied each variable as a time series to determine if general trends could be found in the time domain. A priori, we consider that the variables are not auto-correlated, i.e. that the knowledge of the previous coil doesn’t influence the state of the current coil. Also, we want to make sure that no shift can be identified in the data. Most variables don’t present any deviation throughout the observation period, as shown on the following sample variable. (Fig. 2) Looking at all variables throughout this mean enables the analyst to detect any shift Steel Times International

FIVES.indd - needs job titles on last page....indd 3

Distribution of the variable ‘var2’

Distribution of the variable ‘var3’

Fig 1. Distribution of some variables with significant skewness

in the data that could be caused by sensor recalibration, moving, or other maintenance activity. Link between variables Some machine learning models are sensitive to ‘multi-collinearity’ in the input dataset: models without a regularisation mechanism will lead to some instability in the solution space, and to overemphasise the multicollinear features. To prevent such an issue, we treated multi-collinear input features by removing the perfectly correlated ones and performing various features generation on the strongly correlated once. Outliers Outliers are data points identified as abnormal and should, therefore, be removed from the dataset. Outliers can have a strong negative influence on certain types of machine learning algorithm. To prevent this effect, we implemented several outlier detection techniques, both univariate and multi-variate. As an illustration of a multi-variate outlier detection method, we describe next the Mahalanobis distance technique. Using a univariate approach, the red dot on the following figure couldn’t be identified. Mahalanobis distance uses a combination of

the most relevant dimensions, called principal components, to calculate a Euclidian distance to the data centre in this hyperspace. All the points on the green lines on the following figures are equidistant to the centre in the Mahalanobis definition. (Fig. 3) Fig. 4 is the distribution of the Mahalanobis distance across the training dataset. Machine learning models are easily impacted by the outliers in the training data. To avoid this kind of problem, we remove outliers present on the entire train dataset, using multi-variate outlier detection methods. In the training set case, we excluded 1.2% of the whole dataset using this method. The production model includes this outlier treatment, which results in some cases to the model providing a warning that the provided input parameters are outside the confidence space of the model. Therefore, in some cases the input parameters will be considered as outliers and there will be no prediction for them. Categorical variables The input dataset contains few categorical variables, such as supplier’s id, tensile test samples position and orientation. The behaviour of each of those categorical variables should be studied compared to

Fig 2. Sample output data represented as a time series

FUTURE STEEL FORUM

61 29/08/2019 14:04:11


GALVANISING LINES

Multivariate outlier example 4 3

1 # of coils

Variable 2

2

0

-1 -2 -3 -4 -3

-2

-1

0

1

2 2

3 3

Variable 1 Fig 3. Mahalanobis distance principal illustration

Mahalanobis distance Fig 4. Mahalanobis distance distribution on the training set

Fig 5. Fives process control solution

Coil information NeoKoil® smart line

Subsystems models

Max prod instructions

Metallurgic model

Optimisation

ML model

Coil prod instructions

Furnace level 2 Line steering

continuous features, as they can hint at strong behaviour differences in the data. Depending on the cases, it can lead to the development of separate models per categorical value or to the exclusion of certain categorical values. Models During our study, we implemented several models of machine learning from the main model families. As each family has a different resolution approach, trying various model types usually yields good results. The regression models used rely on the following algorithms: •Lasso •Random forest •Support Vector Machines (SVM) •Extreme Gradient Boosting (XGBoost) •Neural networks The best prediction of YS and El% are reached with the regressor of XGBOOST. Regarding the variable UTS, the prediction is based on a Support Vector Regression (SVM) model. Neokoil® Smart line, automatic line management

Furnace

62

FUTURE STEEL FORUM

FIVES.indd - needs job titles on last page....indd 4

Zinc pot Wiping

Skin-pass

Additional implemented techniques On top of the already quoted methods, we used the following techniques: •Cascading models: using the best prediction of a variable to predict another one. •Aggregating: performs the weighted sum of different models. The optimal weights are determined through an optimisation problem. This technique usually improves the predictions, as some models perform better in certain areas of the solution. It also improves the solution robustness, as outstanding predictions will be evened out by other models. •Cross validation: this technique is used to validate the stability and quality of the models. After splitting the original data between test set and training set, the training set is again split into a training subset to train the model and a validation set to evaluate it. We use a “k-fold” cross-validation, where the original dataset is randomly partitioned into k equal size subsamples, one of them is retained as a validation set, and the k-1 subsamples are used as a training data. This process is then repeated k times (each subsample is used once as the validation data). The validation dataset provides an unbiased evaluation of a model fit on the training set. The Cross Validation will limit problems like overfitting (the model is too closely fit to the training dataset), underfitting (the model isn’t well adapted to the training Steel Times International

29/08/2019 14:04:13


YS (MPa)

UTS (MPa)

El (%)

Physical model

14,83

28,6

6,87

ML/Physical model

7,36

9,6

1,05

XGBoost

SVM

SGBoost

ML model type

Improvement 50% 66% 85%

Fig 6. Median error per model and improvement summary

system developed by Fives will then pilot the line in order to minimise, among other things, the transition states from one coil to another and the energy consumption.

dataset) and get an insight on how the model will generalise to data sets used in production. Architecture The combined physical/machine learning model described in the present paper is integrated in the NeoKoil Smart line system developed by Fives. The process control system, as shown on the following figure, uses this model to try a series of production scenarios and evaluate them against quality targets. It takes as an input parameter the information of the coils ready to be galvanised. It provides as output the static optimal production parameters to produce a specific coil. The furnace level 2 automation 20190823_XOM-Ad_184x128mm_ZW.pdf

Machine learning architecture The interface between the NeoKoil Smart line and the ML machine is provided by a REST API, which provides important flexibility in terms of integration, and scalability for future uses. The data pre-processing features generation and machine learning models are encapsulated in highly standardised modules and integrated into a data pipeline. This ensures the reproducibility and traceability of the predictions, as well as an ease of deployment and maintainability. 1

23.08.19

Results The combination of physical and machine learning models yields an improvement compared to a physical model only approach. The following table summarises the improvement made thanks to machine learning, but displaying the improvement achieved on the median error. The solution for DP steel will be commissioned during Q4 2019. Machine learning models for other steel types will be developed during 2019.

1. Cyril Peillon, senior data scientist, Fives CortX, Fives Group 2. Philippe Rocabois, Metallurgy Expert, Fives KEODS, Fives Group 3. Alessandro Ferraiuolo, R&D and Supplier Development Manager, Marcegaglia

4. Aldo Fiorini, Plant Manager, Marcegaglia, Ravenna plant. 5. Stefano Pantarotto, Deputy Plant Manager, Marcegaglia – Ravenna plant.

13:11

C

M

Y

CM

MY

CY

CMY

K

Steel Times International

FIVES.indd - needs job titles on last page....indd 5

FUTURE STEEL FORUM

63 29/08/2019 14:04:13


MODELLING COKE PRODUCTION

Prediction of coke strength by machine learning G

ood quality raw materials produce low cost hot metal. One of the important quality parameters for blast furnace operation is Coke Strength after Reaction (CSR), as it refers to the strength of the coke when hot - a quality test in a simulated reaction to predict performance in the blast furnace. In the research described, the effects of coal properties and process parameters on the coke CSR were studied using the XGBoost model. This ‘tree’based method uses historical data from the previous three years to estimate the CSR value. Twenty-four input parameters, moisture, volatile matter, ash, fluidity, battery temperature, and so on, were input. XGBoost proved to be optimal using eleven parameters of coal properties and coke oven process parameters with respective accuracies of 79%, 76% and 74%, during the training, testing and validation datasets. The operating range of coal properties and controllable process parameters is derived from the model developed to aim for a consistent CSR value of 65.5% and above. Coke CSR is a function of coal parameters and process parameters. A lower CSR results in higher coke rates in the blast furnace while a higher CSR value enables greater opportunity for coal blending to lower costs. Previously, no model was deployed at the Tata Steel Jamshedpur coke plant which could make appropriate recommendation to increase CSR and reduce fluctuations in its value.

64

FUTURE STEEL FORUM

Satish Tata.indd - Read MM, minor author query.indd 2

The project targeted a reduction in the standard deviation of the CSR to improve the mean value from 66.3 to 67.0 by better predicting the blend of coal mix, process parameters and coke oven conditions. Fig 1 illustrates the improvement achieved over two years in achieving a higher and more consistent CSR value averaging 66.66 with a standard deviation of 0.80. (Fig.1) Attributes driving the complexity towards improving CSR are: • Coal properties vary between different types of coal; • Coke making process highly complex. CSR depends on the coal blend and process variables during coking. Operating in the optimal zone should lead to consistency in coke quality, but due to the high complexity of the coke making process, operating Actual CSR Aspiration Mean Mean CSR

conditions are not always consistent which leads to variation in coke quality. XGBoost modelling XGBoost is an optimised distributed gradient boosting library designed to be highly efficient, flexible and portable. It implements machine learning algorithms under the Gradient Boosting framework. XGBoost provides parallel tree boosting (also known as GBDT, GBM) to solve many data science problems in a fast and accurate way. The same code runs on major distributed environment (Hadoop, SGE, MPI) and can solve problems beyond billions of examples. XGBoost is used for supervised learning problems, where we use the training data (with multiple features) xi input to predict a target variable yi.

Actual CSR (02-Jan 2016 to 28 Feb 2018)

Good CSR range Avg. SD 66.66

0.80

Mean CSR Aspirational cutoff 66.3

Low CSR range

� 40% of times the CSR values are below 66

CP2_CSR

Avg.

SD

64.10

1.11

Data analysed: FY26-FY18 Records available: 765 days Aspired CSR range: 66.3 67 Aspired mean CSR: 67.0

Fig1. Improvement achieved over two years in achieving a higher and more consistent CSR averaging 66.66

Steel Times International

29/08/2019 09:22:22


Following the introduction of a decision support model in the Coke Plant to better monitor the operating ranges of critical parameters and alert operators and managers of deviations, the CSR value has increased 1% and its variability reduced 36%, offering an estimated annual saving of Rs 86M (US$1.25M) per million tonnes of coke produced. By S Agarwal1*, R Thakur2, A Kumar2, S Dutta2, S Shrivastava3, PK Choudhary2, Vikas4, A Kumar Kabra4, R Karn4

Fig 2.Web based interactive dashboard of predicted vs actual CSR

Methodology The approach was to develop a decision support model which could intelligently define the operating range for critical variables in coke making and warn operators of trends and alerts to highlight any process deviations. Appreciating the impact of process variability on coke quality, the project aim was to use data for available variables to predict the CSR value. Further, using the prediction model, the operating ranges for all critical variables were determined. Operators could then manage the coke oven in the defined operating regime to reduce the variability of the CSR and improve its mean value. The standard operating regime defined for critical variables would then guide the range of all such variables so that operating the ovens in that regime would result in the desired CSR.

Steel Times International

Satish Tata.indd - Read MM, minor author query.indd 3

The steps followed were: 1. Data identification and extraction for blend properties and process variables; 2. Prepare master data table; 3. Run variable selection models to establish the relative importance between the variables; 4. Develop a CSR prediction model and do validation trials of the model to estimate its accuracy; 5. Use statistical simulations to identify a high CSR value operating regime (CSR>67); 6. Develop a graphic dashboard for operations to observe the operating regime and compare the same with the coke oven running data; 7. Develop alerts and an escalation system to inform coke plant operators, then senior management, of any deviations from the optimum regime.

The previous three years of operating data for coke oven control and coal blend enabled a model to be developed using R tool – a library using ‘dplyr’ and ‘xgboost’. The model in supervised learning usually refers to the mathematical structure by which the prediction yi is made from the input xi. A common example is a linear model, where the prediction is given as: y^i=Σjθjxij This is a linear combination of weighted input features. The value predicted can have different interpretations, depending on the task, ie regression or classification. For example, it can be logistic transformed to get the probability of positive class in a logistic regression, and it can also be used as a ranking score to rank outputs. A salient characteristic of objective functions is that they consist of two parts: a training loss and a regularisation term: obj(θ)=L(θ)+Ω(θ) where L is the training loss function, and Ω is the regularisation term. The training loss measures how predictive our model is with respect to the training data. A common choice of L is the mean squared error, which is given by: L(θ)=Σi(yi−y^i)2 Another commonly used loss function is logistic loss, to be used for logistic regression: L(θ)=Σi[yiln(1+e−y^i)+(1−yi)ln(1+ey^i)] The regularisation term controls the complexity of the model, which helps avoid overfitting. Model implementation Before deploying the model, internal tests were performed using it for 15 days to evaluate its performance. The daily predicted value of CSR was then communicated by e-mail for the next 15 days to all the stakeholders to build their confidence on the model, before actually employing the system. For online deployment of the model, discussion with key stakeholders such as shop floor operators, shift managers and the head of plant was carried out to determine how the model could be used to the utmost benefit. Key solutions, identified with the help of business stakeholders were: a) Daily prediction of CSR (one day in advance); FUTURE STEEL FORUM

65 29/08/2019 09:22:27


MODELLING COKE PRODUCTION

Table 1. Dashboard display highlighting parameters out of spec

Fig 3. Automatic e-mail alert of out of range parameters

Unit

Value

Average CSR (Batt#8,9)

%

66.3

Avg Revised CSR (Batt#8,9)

%

67.0

Difference (Batt#8,9)

%

0.8

1% change in Coke CSR reduces fuel rate by Total Volume of HM (A-F BF)- ABP FY19

kg/thm

2.0

KT

2452.0

Savings in Coke

Tons

3781.5

Avg Cost of Coke FY18

Rs/t

15000.0

1. Benefits from Coke Rate reduction

Rs

Crs 5.7

Production benefit for every point CSR increase

%

2%

KT

37.8

Expected increase in Production Margin over pooled iron @ FY18

Rs/t

4190.0

Crs 1

5.8

Rs Crs

0.0

only 40% was achievable with such models.

40%

2. Benefit from HM Production Improvement Rs 3. Benefits from Coal Blend Optimization Kept out of the scope of the project for the time being Out of similar projects credit allocated to this project can be @30-40% like we have done with Tuyere Failure Analysis project with backup wherein Total Benefit

Rs Crs

8.6

Table 2. Expected cost savings by stage of raising the CSR by 1% for production of 1Mt/a of coke

66

FUTURE STEEL FORUM

Satish Tata.indd - Read MM, minor author query.indd 4

Steel Times International

29/08/2019 09:22:28


b) Operating range of controllable parameters to achieve the target CSR; c) Alert of any deviations in process parameters; d) An inventory alert; and e) Coal blend playbook for coal selection.

Finally, a web based live dashboard incorporating all the solutions developed was deployed showing daily prediction of the CSR and the deviation of the process and coal parameters from the suggested range as shown in Fig. 2. a) Daily prediction of CSR The daily predicted CSR value is shown on the trend chart with the actual CSR value in Fig 2. Operators can access this to see the difference between the actual and predicted CSR to be aware of the quality of coke expected a day in advance and take proactive action if needed. b) Controllable parameters for target CSR The operating ranges for the two coke oven batteries were incorporated in the dashboard display to show coal variables and process parameters. Table 1 displays any out of range variables highlighted in red or yellow

for factors above upper and below the lower limits respectively, to enable the operator to focus on deviations so as to keep the CSR at the desired levels. c) Process alert of deviations In case any variable is out of range for two days consecutively, an automatic e-mail message is triggered from the system to the shift supervisor (Fig 3). If the deviation continues over the next four days an alert will be triggered to the operation manager and head of plant, and, after six days continuous out of range, an auto alert e-mail will be triggered to the chief of coke plant. d) Inventory alert The coal inventory visibility has also been provided to stakeholders which can help plan the blend according to availability to avoid coke quality issues. e) Coal blend playbook The coal blend playbook shown in Fig 4 is also incorporated in the web based display so the user can track the latest blend of coal in use, the individual coal properties, individual coal cost and coal inventory (Fig 4). The user can adjust the existing coal mix to see the impact on the coke CSR (in lower left of Fig.4) and the resulting mix cost. This tool is

very helpful in optimising blend selection for CSR quality and cost. Impact on costs A cost rule given by the Business Analysis Group (BAG) states that: A 1% increase in CSR leads to a 2kg/thm reduction in fuel rate and a 2% improvement in productivity. Table 2 shows that a total saving of Rs 86M (US$ 1.25M) can be achieved by a 1% increase of CSR from 66.3 to 67 for the production of 1Mt of coke a year. With the implementation of this project the CSR value has been increased 1% and the value’s consistency improved from a standard deviation 1.63 to 1.04 in recent months resulting in a reduction in coke rate in the blast furnace and so less coal being procured with subsequent cost savings of Rs 86M (US$ 1.25M) a year per million tonnes of coke produced.

The authors are with Tata Steel Ltd, Jamshedpur, Jharkhand-831001, India. 1 Analytics & Insights, 2 Coke Plant, 3 Technology Group, 4 OneIT, *Corresponding Author: Satish.agarwal@tatasteel.com, +91 9230522510

Fig 4. Coal blend Playbook showing coal blend on cost (Rs)

Steel Times International

Satish Tata.indd - Read MM, minor author query.indd 5

FUTURE STEEL FORUM

67 29/08/2019 09:22:28


AUTOMATION

A new standard in plant automation The SMS group’s X-Pact Process Guidance is claimed to be ‘the smart connection and systematic networking of knowledge’. The company claims to have set a new standard in plant automation. According to Joerg Thomasberger* the system supports all the usual Industry 4.0 communications standards or digital products and allows new sensors to be connected. The uniform style of the system ensures that all plants from the company have the same graphical appearance. By Joerg Thomasberger*

68 sms GROUP done.indd 2

FUTURE STEEL FORUM

Steel Times International

29/08/2019 10:13:58


P

roduction processes in the steel industry have been and continue to be increasingly challenging, and the reasons for this are manifold: the growing flexibility in production, the ever smaller batch sizes that are possible as a result of this flexibility, and the steadily expanding product mixes. Besides these factors, the ongoing development of automation technology is an essential requirement in terms of remaining competitive and supporting production personnel with the latest automation solutions. As the degree of automation in steel plants increases, it is more important than ever to have systems that are capable of providing support in non-standard scenarios. This support must be designed in such a way that every operator, no matter how experienced he may be, is able to handle extraordinary situations, even if they are very rare. SMS group’s X-Pact system has been designed to cope with these kinds of challenges in the modern and data-hungry digital age. To create an automation system that is future proof, a modular architecture was required. Every part of it should be capable

Steel Times International

sms GROUP done.indd 3

of being replaced or exchanged with minimal effort. For this, it is critical that most functions are realised in modules, while the core architecture remains structurally the same. The X-Pact automation system features a modular design throughout and it is easy to integrate conventional modules, such as reporting, tracking, materials management, and process models. New benchmarks are also being set by the software architecture, thanks to the flexible run-time and servicebased networking of automation functions. What’s more, process guidance from SMS group features a wide variety of virtualisation functions. Virtualisation not only helps to reduce infrastructure costs by consolidating the available hardware, it also boasts a high level of availability, as it is already prepared for functions such as redundancy and disaster recovery, making it independent of the system hardware. Process slider One of the primary focuses of the X-Pact system is to present data and guide users through it. The previous separation between

level 2 and level 1 has been combined into one process guidance-oriented approach. As the entire process is imaged into the information hub, process guidance is able to provide all users with a complete view of the plant without the need to switch between different automation levels. However, with so much data it is important that users are able to navigate within the system without information overload. That is why X-Pact ensures that the data is presented using ergonomic screen functions, with data clarity as its prime focus. Operation and visualisation are based on the X-Pact Vision concept, where the focus of interaction is always on the operator. Specific dialogues have been developed for specific categories of users, for example line operators, technologists, maintenance personnel, production managers and so on, while focusing on their individual sphere of activity. X-Pact Vision also provides a multitude of tools, such as breadcrumbs, wizards, or context-sensitive help tools. The process slider is the plant operator’s central HMI unit; this operating screen offers intelligent, process-oriented and situation-

FUTURE STEEL FORUM

69 29/08/2019 10:13:58


AUTOMATION

New standard for the future automatic control of complex plants Video: https://www.sms-group.com/process-guidance-animation

specific operator guidance. The process is displayed chronologically from left to right. When the system is in a process phase, the steps within this phase are run through one after the other, from top to bottom. This guidance ensures that the operator receives only the information needed for a particular process step. In addition to the main sequence flow, the operator is also provided with a full list of non-standard scenarios

70 sms GROUP done.indd 4

FUTURE STEEL FORUM

commonly associated with each individual process step. RetroďŹ tting and future-proof scalability A typical steel plant features many supplier and automation systems at various levels. Data from all the systems are gathered by means of interfaces, which are complex and fraught with problems in respect of their

implementation. Modernising an archaic automation system is always a challenge. It is even more difficult if the customer leaves no scope for shutdowns to install, test, and switch over to the new system. The task here is to achieve a modernised system within an existing automation landscape, which assumes fault-free control of the steel making process, and to create a future-proof automation architecture that can be easily expanded further and further with additional systems and functions. X-Pact has already been implemented in more than 10 different applications. For one of our customers in Canada, SMS group has equipped the whole steel mill area with a new process optimisation and planning system based on X-Pact. This tailor-made solution begins with the takeover of the pig iron at the blast furnace via a GPS acquisition of the torpedo cars. The charged material for the steel mill, such as scrap, lime, and other additions, is calculated dynamically with regard to the costs and availability for each heat. Production planning for the integrated steelworks includes a total of six primary and secondary metallurgical plants (2xBOF, 2xLF, 2x CAS-OB, and scrap yard), right up to the two continuous casting plants. Dynamic models for steel and temperature calculations for the entire production chain ensure high quality production in terms of yield, energy consumption, and costs.

Steel Times International

29/08/2019 10:14:01


Other highlights of the system are: uniform operating philosophy, full integration of the laboratory equipment, integrated planning of the entire production in one system, fully automatic tracking, and recognition of all ladle movements. The X-Pact Process Guidance system renews the automatic control of complex plants. Its modular architecture and universal interfaces mean it is designed and built for the future. Cloud computing and digitalisation play a crucial role here. Operation is process-oriented, ergonomic and easy. Process guidance is highly transparent and scalable. * Contact: Joerg Thomasberger. Email: joerg.thomasberger@sms-group.com Website: www.sms-group.com/x-pactprocess-guidance

Example of process slider

Steel Times International

sms GROUP done.indd 5

FUTURE STEEL FORUM

71 29/08/2019 10:14:08


Cyber security – the Danieli route

Digitalisation is changing the world. Where manufacturing is concerned, it is changing the way things are done and our daily interaction with systems and technology and, as a consequence, the way we design, produce, deliver, install and maintain our products and services.

72 danieli.indd done..indd 2

FUTURE STEEL FORUM

CYBERSECURITY

Fig 1

Steel Times International

29/08/2019 09:34:59


Fig 2

0

ur dependence on interconnection and data is becoming more crucial if we’re going to be competitive in a fastchanging world. The steel market will change faster too because of the benefits achievable from data-driven services. That said, it exposes assets and data to malicious cyber-threats. For these reasons, Danieli, considering its established heritage as a plant builder and steel maker, has started to investigate and develop approaches and solutions in the field of cybersecurity. Danieli and cybersecurity Nowadays, many new technologies are enabling the development of more smart manufacturing processes by interconnecting the operational technology (OT) to the Steel Times International

danieli.indd done..indd 3

enterprise information technology (IT or ICT) and even to the cloud. Every connected device, as well as production plants, are possible targets for cyber attacks because of their strategic and economic relevance. Danieli is committed to protect its customer assets, data and productivity using state-ofthe-art security technology and practices. Being the custodian of the company's 100year history as a full cycle provider from raw materials to finished products to the metals industry , Danieli is aware of the importance of both efficiency and productivity. Being conscious of cyber threats too, in the last four years Danieli has started and implemented a structured programme to increase its own security, while looking at ways it can transfer its know-how to its customers.

Under the pressure of the risk of cyberattacks , the programme, known as BeSafe was organised in different project streams to cover all aspects of security. The direct experience, the impact and importance, for example, of the human factor in the achievement of the goal of “secure company�, denoted that it is not only impacting ICT, but also the organisation, training and skill improvement. The capitalised experience and the outcomes of BeSafe encouraged Danieli to share the importance of with its customers, the idea being to develop a series of reliable and competitive services and solutions for the steel sector. Among such needs, the company identified the importance given by the customer to carrying out their cybersecurity FUTURE STEEL FORUM

73 29/08/2019 09:35:02


Do you have 2020 vision? Well, we do! We’ve already been on the prowl for a venue capable of hosting next year’s Future Steel Forum and it’s looking highly likely to be in Prague, the capital of the Czech Republic. We are now looking for speakers so if you are interested in making a presentation or being part of a discussion panel, or simply have an idea you’d like to expand, contact programme director Matthew Moggridge today. The Future Steel Forum is all about the future of global steelmaking technology. BE A PART OF THE FUTURE TODAY.

FUTURE

STEEL FORUM 2020

matthewmoggridge@quartzltd.com

IBC ANNOUNCEMENT.indd 2

29/08/2019 10:00:37


CYBERSECURITY

plan with a partner who speaks their language and shares their technology. Aware of the strength of its background, Danieli launched the initiative for the metals industry because it knows the ‘art of steel’ and it has a strong knowledge of cybersecurity. When the BeSafe project started, Danieli decided to define a reference framework in order to implement a security model tailoring actions to a specific case, providing an adaptable organisational landscape and systems that could be easily updated when technological advances, such as cognitive systems and Artificial Intelligence were developed and advanced. The result is shown in Fig. 2. A flexible security model can be tailored in-house and in the production plants to act as a partner for Danieli’s customers that will enable them to reach the status of “secure company ”. For this step of the process, the pilot project and the production took place at Danieli’s sister company Acciaierie Bertoli SAFAU SpA (ABS). Exactly what happened and the achievements made are summarised in the following paragraphs. The Danieli approach The implementation of an efficient and effective security plan is a challenging journey and there are two well known considerations that drove the company’s approach to cybersecurity: 1 That security is a ‘probabilistic’ concept; 2) That cybersecurity is intrusive .

Cybersecurity Implementation Plan (CIP) defines the action sequence and Steel Times International

danieli.indd done..indd 5

its correlations of the measures to impede/ mitigate potential cyber-threats; it must also consider the impact on the organisation and the acceptance and co-operation of the workforce. The easier option is to provide security and plant design together along with personnel training as part of the supply deal on the understanding that a carefully integrated security system is and always will be a crucial aspect for both greenfield and brownfield projects. For this reason, the journey is supported with the invaluable help of standards and norms. Danieli’s approach is compliant with IEC 62443 and references to ISO/IEC 27000 (27001 and 27002). However, their application is complex and can be daunting. Such an approach to security has been designed to be compliant with the ‘new’ digital manufacturing scenario of Industry 4.0, in accordance with the pillars of the DIGIMET Strategy: • Business (business process engineering); • Technology (selection of enabling technologies); • Intelligence (advanced analytics from data to value); • People and organisation (new culture, processes and skills). The methodology is based on the principle of discarding the distinction between the IT and OT environments and integrating them into a unique vision that includes organisation,

Fig 3

people, security and technology designed to increase flexibility in order to add innovative solutions along the way. Co-operation and involvement are crucial when implementing a cybersecurity system based on an inclusive and living concept that is the responsibility of all the players in the company. It is a way of achieving resilience and resistance simultaneously by proposing a model of co-operation between human players and systems because “the threat of today is not the threat of tomorrow”. Looking ahead, Danieli intends to use its cybersecurity experience to support customers with a working model (see Fig. 3), based on an extensive assessment capability and technological solutions for new and existing plants. In both scenarios, the Cybersecurity Implementation Plan needs to be carried out under the strong commitment of senior management and must be the direct responsibility of an apical manager at the customer end. Danieli wants to establish such a partnership and define the services, actions and tools to be implemented for providing an efficient and effective level of security. To achieve this, a working group of experts FUTURE STEEL FORUM

75 29/08/2019 09:35:10


CYBERSECURITY

from both sides should be appointed in order to share and gather information, define a common language and test the solutions as soon as they are installed. New plants For new plants supplied by Danieli, the model refers to supply plants, systems and machines characterised by the outlined approach to be “secure by design” i.e. to offer mechanical, electrical, automation and technological design integrated with the principles of security. It gives a greater degree of freedom to provide efficient and effective solutions rather than starting from a brownfield site. The first test site where the approach has been developed and validated following the principle of ‘security by design’ is the manufacturing area of a new rod mill line. While an existing plant is being used, the rod mill line in question has auxiliaries, equipment, services and its own buildings that represent a self-contained manufacturing area defined by its physical perimeter and interfacing with the remaining areas of the plants and is, therefore, representative of a new plant. In practice, the main steps and principles of the approach are summarised as follows:1. The first step consists of increasing security by monitoring access control and the restriction of user capabilities. 2. After that, the design phase defines the new layout of the plant (or part of it) with the relevant manufacturing areas and the inventory of assets. Among them, critical assets are identified and prioritised taking into account their physical and logical vulnerabilities as single objects and as a whole in the resulting manufacturing chain. 3. A risks list is defined and a risk assessment is carried out in order to protect the equipment and the manufacturing chain in order to achieve the goal of optimal process continuity. Fixes and mitigations are then planned to secure these assets. 4. Once the systems are defined and built for in-house and in-field acceptance tests, specific stress and penetration trials are carried out to test the security level. These steps can be applied to each new plant. Existing plants Existing plants differ because of their physical

76 danieli.indd done..indd 6

FUTURE STEEL FORUM

constraints and the legacy systems they employ. Moreover, information must be collected carefully, and verified on site. Looking at Fig. 3, the following steps are necessary: 1. Listen and understand: the first step is the submission of questionnaires and round table meetings among the appointed working group to gather and validate information relevant to the “as is” status. 2. Simplify and fit: the Cybersecurity Implementation Plan is prepared by adapting the general security model to the specific context and adopting a step-by step and minimally intrusive procedure. 3. Rationalise: information is verified and processed to define and carry out the detailed actions of the plan. Legacy systems must be taken into account, in order to achieve the expected goals. The definition of an action plan is based on gap analysis, defining intervention priorities based on a combination of the importance and urgency of the identified gaps. 4. Execution and validation: the execution of the Cybersecurity Action Plan and then the validation phases are supported offline and in-line. The results can be constantly monitored through additional services (monitoring, intrusion detection, vulnerability assessment). Almost in parallel, personnel training is carried out too. The method shown in Fig. 3 reconciles the two scenarios in the validation procedure. Investments for the existing plants can be progressive in accordance with customer needs. Highlights • The appointment of a bi-lateral working group is a key step as it includes experts from

both sides; • A quantitative assessment of ‘Digital Maturity’ i.e. the digitalisation status of the company is highly advised; cybersecurity maturity and the security posture i.e. the cybersecurity level you want to reach, are decided in line with the state-of-the-art and, therefore, are constantly updated. • Asset analysis to define those critical in terms of risk allows the vulnerability of each item of equipment and their effects on the vulnerability of the entire security system to be assessed; • The aforementioned Gap Analysis with respect to the needed cybersecurity posture is also used to prioritise investments and build up the Cybersecurity Implementation Plan with actions for filling the gaps. This is an extensive activity where existing plants are concerned, and it is advisable to increase the security during the life of new plants. The human factor As mentioned before, experience dictates that safety comes first and foremost from the people. One of the weak points of cybersecurity programmes can be the human factor. When a rigid top-down approach is used, it limits workforce support to standard training that is more devoted to learning how to use systems rather than creating a security culture within the company. For this reason, an integral part of Daneli’s solutions is to provide easy-to-use tools that do not affect operational activities. Danieli designs and runs structured programmes of training and user awareness, to change or improve those behaviours that pose a potential risk for the company. Specific awareness workshops and simulated phishing campaigns are also offered. Conclusions Cybersecurity is a continuous journey because it is linked to the evolution of digital systems. Taking a step backwards is not an option. It is important, therefore, that steelmakers are ready for the new digital world. Danieli’s ‘three days on-site’ approach enables customers to identify the ‘major evidences’ to be addressed on their cybersecurity journey. For further information contact Danieli on BeSafe@digi-met.com

Steel Times International

29/08/2019 09:35:11


From John Cockerill to CMI. From CMI to John Cockerill. The encounter between traditional values and modern trends has helped us to offer the bestadapted response to modern clients’ aspiration.

Inspired by the visionary and entrepreneurial personality of its founder, CMI once again becomes John Cockerill. Since 1817, the strong commitment to a culture of creative thinkers helped us to provide innovative and profitable answers to the needs of our clients. Resolutely oriented towards the future, innovation forms an inherent part of our engineering. While growth is substantial, the challenge is to focus on generating sustainable progress. This is what we have done for the past 200 years.

Design | Engineering | Technical Assistance & Training | After Sales

johncockerill.com/industry

INDUSTRY METALS Reheating Furnaces Pickling Acid Regeneration Cold Rolling Strip Processing Automation


THE SMART STEEL PLANT OF THE FUTURE

At Primetals Technologies, we don’t just build steel plants. Instead, we ensure all your equipment is perfectly synchronized for maximum performance and reliability. We like to continually innovate, to facilitate progress, and to pioneer new solutions. We love steel, and we will change the way you produce it. primetals.com

Ad-297x210-FSF-magazine-Future-Steel-Forum-Budapest.indd 1

01.08.2019 14:21:29


Turn static files into dynamic content formats.

Create a flipbook
Issuu converts static files into: digital portfolios, online yearbooks, online catalogs, digital photo albums and more. Sign up and create your flipbook.