CC BY
38S. Spuzic
The University of South Australia
HOW TO REDUCE IMPACT OF INDUSTRIAL ROLLING ON GLOBAL RESOURCE CONSUMPTION AND CLIMATE CHANGE?
Abstract. Motivation for scrutinizing rolling process is triggered by growing evidence of anthropological causes of global climate change that are amplified by worldwide acceleration in natural resource consumption. While the key industries such as steel manufacturing systems are condition sine qua non for our progress, the rationalisation of relevant technologies is of critical importance. Hot rolling systems, which are in the heart of steel industry, are ineludible examples of unavoidable man-made enterprises crying for rationalisations. The promising avenue for decisive progress in (re)designing rolling systems is by implementing Big Data strategy for mitigating the problems in untenable production and in designing new processes. This strategy stems from recognising importance of knowledge extracted from data accumulated in industrial repositories. An internationally maintained collaboration between industry and academe is the prerequisite for application of proposed strategy in practice.
Keywords: global warming, natural resources, industrial rolling, rationalisation, big data.
Introduction
The alarming evidence of global warming, key resource decline, and consequent socioeconomic disruptions indicate a need for addressing aspects of significance for future wellbeing. Rather than bluntly proposing various cutbacks in living standards (such as reductions in water and energy supply to urban settlements) a more intelligent and humanistic approach is to focus on further improvements and rationalizations in operating large-scale industrial enterprises. These gigantic man-made systems are governed by multinational corporations that are too often driven by narrow-mindedly defined profit. Typical examples are steel manufacturing plants. A conservative estimate indicates that the production of one tonne of steel product requires nearly 90 tonnes of fresh water. The complete metalworking sector and majority of complementary industries are seething with sources of environmental pollution and hazard. As another example, the gas and ore mining (that present the primary suppliers for steel industry itself) have critically affected landscape, food production, water security and communities across too many geographic regions [1].
However, awareness of this state of affairs must not overshadow mindfulness of their intrinsic and intended usefulness. The significance of metallic materials such as steel for constructing and maintaining infrastructure is well known. Abundant examples include Chaotianmen Bridge (with the world's longest arch span of 1,700 m) over the Yangtze River, Burj Dubai, the tallest skyscraper in the world (incorporating 4,000 tonnes of steel), Sydney Harbour Bridge (the world's largest steel arch bridge), trans-Pacific 9,000 km long undersea cable, (that includes steel wire along with other metallic components) connecting Oregon US and Japan, to mention just a few [1].
What makes steel so favourable as a construction material? Steel is one of the world's most-recycled materials, which makes it environmentally sustainable. This versatile, durable and affordable material has a very advantageous combination of elasticity, strength, plasticity and strain hardening, and these properties do not decline during the recycling process. When a structure such as a bridge or a power transmission tower is assembled and fastened, any individual component (column, joint) that is exposed to extreme stress, will react elastically. If load persists this component will adjust its dimension plastically to transfer the fraction of load on its neighbours. It will not lose integrity nor will undermine the integrity of the assembly. On the contrary, the deformed element will gain in strength, thus relaxing the stress in other members.
As the world's largest steel producers India and China China's iron and steel technology have stepped into post industrialization stage, the new road to industrialization must be taken and the ecological iron and steel technology system must be built, in order to solve the problems of resource, energy and environment. The entire industrialised world is faced with an imperative to realize "green production", to enforce the harmonious and sustainable development of the steel sector.
About 90% of all steel products (and about 80% of all metallics) are at some stage processed by rolling, and the foreseeable long-term global views imply growing requirements for metallic products of increasing quality, [2-7]. The challenge for contemporary rolling technology is how to reconcile this with the reality that, in addition to environmental pollution, the rolling plants consume enormous amount of energy that is generated mostly by fossil fuels (which, in turn, are nonre-newable resources).
The fossil fuel peak of discoveries has been passed and followed by a steady decline more than a half a century ago. Global environment in which these trends make their hefty impact is characterized by the disproportionate amount of energy resources consumed in technically inflated countries. Although the orientation on renewable sources of energy is gaining in progress (e.g. South Australia has closed its coal-based power plants; by 2020 Denmark will be producing half of its electricity from renewables) the growing challenge remains: how to reduce energy consumption and yet to sustain industrial development [1]?
Indisputable efforts are invested into improving the sustainability of rolling mill systems, but the global warming forecasts indicate that more must be done, more urgently. Apart from energy issues, the mill infrastructure is a matter that demands a particular focus. Steel rolling operations rely heavily on the availability of equipment such as rolls. The owners of such assets need to keep their equipment up and running as efficiently as possible. The last resort is replacing defective components with ready for use components. The defective components are usually expensive. Therefore, a defective component is repaired if possible and put back in operation. It is important to note that any instant of the process interruption (in order to perform such a replacement) causes considerable losses. One hour of delay (stoppage) in a typical rolling mill production process costs about $ 30,000 [1].
The above issues ought to be resolved along with a continuous strive to improve the quality of rolled products. Quality is not just a complementary aspects, it is in fact the principal key performance indicator. The numerous paradigms promoted over the recent decades converge to emphasising the need for implementing continuous improvements in manufacturing systems. Typical examples are Kaizen and Taguchi's philosophy.
Key features of Kaizen include.
1. Improvements are based on many small changes rather than the radical interventions.
2. As the ideas come from the workplace, they are less likely to be fundamentally different, and therefore these suggestions are easier to implement.
3. Small improvements are less likely to require major capital investment than major process changes, etc.
Taguchi's quality philosophy can be summarised as follows.
1. An important dimension of the quality of a manufactured product is the total loss generated by that product to society.
2. In a competitive economy, continuous quality improvement and cost reduction are necessary for staying in business.
3. A continuous quality improvement program includes an incessant reduction in the variation of product performance characteristic about their target values.
4. The final quality and cost of a manufactured product are determined largely by the engineering designs of the product and its manufacturing process.
The overall picture would be incomplete without realising that large multinational and national corporations operating rolling mills remain burdened by over-commercialisation and an emphasis on short-term profit. This results in too slow introduction of technological advances and, as a consequence, many current rolling systems are unsustainable. Moreover, there is a strong but narrow-minded trend of creating exclusive circles of so called inter-industry and intra-industry trades -closed within politically and economically allied countries [1].
In spite of efforts of institutions such as World Steel Association (http://www.worldsteel.org), Association of International Roll Pass Designers and Rolling Mill Engineers (http://www.aikw.org),
Kalibrovochnoe Biro (http://www.passdesign.ru) and Association for Iron & Steel Technology -AIST (https://www.aist.org), the above problems persist on a worldwide scale. Rolling system design consultants (such as WICON, Gerdau, Hatch, Morgardshammar and Danieli) and industrial technology departments (such as those incorporated in Voestalpine, Siemens AG, Sumitomo Heavy Industries, and Sandvik) are restricted by competitive environment and do not collaborate efficiently.
Academic interactions present a movement that acts to break these barriers, yet these initiatives need more institutional endorsement in order to prevail. Scientific advances in principle allow for significant technological improvements by means of introducing intelligent modifications in the material processing and product finalisation. Operations research theories developed a variety of models for analyzing complex dynamic systems, such as Markov modulated Poisson process, Non-stationary Poisson processes, Static or Dynamic Priority Rules, Decomposable Lagrangian multipliers. Multi-item repairable inventory models are abundant in literature [8].
Along with the vigorous research into advanced materials and new sources of energy, there is a whole spectrum of theoretical, analytical and intelligent methods that are explored within the research into design and optimisation of rolling systems. Theory of plasticity has been greatly elucidated by application of deterministic models such as FEM, slip-line and finite difference analyses. In spite of all this, the actual rolling practice outstrips understanding of it [9, 10]. When facing problems such as low productivity and poor quality, rolling system designers still must rely on costly trials and errors. New solutions by virtue of evolutionary algorithms are increasingly sought for by utilising rapid advances in computing technology [11-14].
In summary, there is an evident urge for complementing and combining the existing methodologies with additional strategies, might they be found within or beyond the boundaries of accustomed disciplines.
Promising avenues
The contemporary availability of growing industrial and academic databases and the capability of information technology and communication systems open a possibility to analyse this treasure and share the extracted knowledge in unprecedented volumes and at the sped of magnetic waves. Big data is a concept that embraces very large and/or complex data sets that may be analysed computationally to reveal patterns, trends, outliers and correlations. This strategy is already broadly utilized in the fields of social sciences, medicine, climatology, and increasingly in research related to global warming.
Big data analyses can make important contributions but also present unique challenges to international development, such as privacy and interoperability issues. Advancements in big data analysis offer cost-effective opportunities to improve decision-making in critical development areas such as technological infrastructure, operations and product quality [15-17].
Big data provides a methodology for unravelling uncertainties such as inconsistent component performance and availability. It facilitates real-time decisions that mitigate operation problems, such as downtime, by online transforming measurements into useful information. Vast quantity of sensory records (such as vibration, pressure, voltage, dimensions) present an information rich repository that can be correlated to the historical evidence of process key performance indicators [17, 18].
Cyber-physical implementations use information during the actual operations while analytical algorithms perform more reliably when more data throughout the machine's lifecycle is entered. There is a potential to systematically integrate, manage and analyse machinery, tool or process data during different stages of manufacturing operations. This includes the capacity to handle information such as system configuration more efficiently and further achieve better transparency of equipment and tool utilisation [18].
There is a well-accepted general understanding about the importance of analysing actual industrial records. The most valid evidence about the interplay of all possible variables in a rolling mill and the ultimate outcome is a factual manufacturing process. ''Intelligent systems that continu-
ously learn and optimise the performance of rolling mills are extremely valuable for the manufacture of consistent/repeatable product quality. Rolling mill operations generate huge quantities of records, and this database has not been put to good use in the past. Developing analytical tools to utilise the manufacturing databases would be of great benefit to the industry. It only requires a small yield improvement to make a valuable financial contribution when rolling such large quantities of steel products" [19].
In essence, the principles of multivariate statistics, nonlinear optimisation and matrix algebra are projected to increasingly more complex and voluminous sets of data by virtue of analysis of principal components, canonical correlation and MANOVA, to mention just a few techniques. Analysing data in high dimensions by including hundreds of parameters is viable as long as the sufficient number of observations is made available. In summary, the reliable information can be extracted from big data and combined with a variety of empirical and deterministic approaches. Assortment of available mathematical methods that can be employed is quite rich (starting with descriptive techniques, via multivariate distributions and inferential statistics to probabilistic optimisation). Factor analysis and fuzzy logics can be used to uncover the latent structure (dimensions) of a set of variables. It reduces attribute space from a larger number of variables to a smaller number of factors. In the absence of sufficient observations, the statistical analysis can be combined with empirical and theoretical models leading to hybrid models.
However, on one side, the competitive environment that dominates in steel manufacturing industry prevents standardisation and open sharing these records. On the other side, even when published, these records show only summative key performance indicators, and technical parameters. When it comes to a more detailed information, the available records differ in customs and conventions. For example, technical drawings are dimensioned using non-standardised codes, tool life is defined using different criteria, etc. In order to translate industrial records into structured database, there is need to define these specificities using generic and standardised format. This will allow for applying mathematical methods to analyse processes of interest.
Initial attempts to utilize big data and apply mathematical analysis are taking place, usually including limited collaboration between a single industrial organisation and one research institution. For example, authors in [20] reported successful collaboration with a specific corporation operating industrial rolling mills. The company was interested in implementing corrective actions to improve quality management, and especially to prevent recurrence of production drawbacks. The need for investigation arose when management was urged to reduce the portion of finished product of unsatisfactory quality due to the presence of defects. Deciding on specific corrective measures for such complex production process that involves multiple stages entailed identifying an unknown combination of (two or more) factors that were responsible for the defect [20, 21].
Rolling technology aspects
A number of researchers observed that dynamic and morphometric factors (deformation rate, geometry and dimensions) rank high in the hierarchy of the variables controlling the rolling process. It is known that suboptimal geometric conditions during the rolling operation can distort the shape of the finished product as well as the distribution of the local plastic flow and kinematic parameters. The use of mathematical models to determine the reasons that a given metallurgical product is of low quality can significantly shorten the time needed to devise corrective measures and reduce the costs incurred in substantiating technological innovations [9-11, 21-22].
"Normal rolling conditions" can undergo many different changes during tool exploitation. After a roll change or an operation delay, rolls need some time to return to stable thermal conditions. Every new bar entering the mill creates an impact. Sometimes, severe rolling accidents occur, due to faults by operators, internal defects in the roll or rolled material, a power cut, mechanical problems, transportation hold-up or interruptions in the water cooling system. These "abnormal" rolling conditions are more or less quite "normal" for industrial operations, as roll damage often occurs with consequences for the mill and the rolled product. Evidently, the reasons of roll failure have to be discussed and analysed to reduce further risks. It is impossible to predict all such accidents, how-
ever, reasonable emergency rolling schedules can be anticipated, provided that engineers have sufficient knowledge of possible design boundaries and variants [23, 25].
Brinza et al. have shown that relations between geometric factors in the section-rolling and theoretical indices that characterize various components of the quality of deformed material can be generalized in mathematical models. A range of such models has been constructed, tested, and put to practical use in analysing the laws which determine the quality of rolled products. These models have been constructed using experimental and theoretical solutions of three-dimensional plastic flow expressed in terms of the mechanics of deformable solids. A set of geometric factors that uniquely characterize the deformation zone during the rolling in roughing passes was defined for the purpose of modelling the rolling process in bar and wire rod mills operated by company Evraz. Geometric parameters of rolling pathways were transformed into dimensionless factors and statistically analysed. The multivariate representation of the passes yielded significant information which enabled detection of technological reserves in the rolling process. Production records were collected over the period of three years of full production in 250 mm bar and wire rod mills. A universal method for process improvements is defined allowing for rapid corrections of the groove elements and the rolling protocol [20, 21].
Another example leading to big data approach is discussed by Appleton [11] who promotes hierarchical classification system termed Durham matrix-based methodology. This strategy, which has been successfully used in several areas, such as design for manufacture and maintenance planning, gives the designers an incentive to consider all input-related parameters, issues and concerns they think are important. The whole spectrum of mechanistic, materials, operation and maintenance data is embraced and analysed by means of manifold iterations.
Other authors explored whether significant statistics can be inferred by analysing published information about morphometric and dynamic aspects of rolling series. Pilot statistics indicated that significant correlations can be inferred by using various linear combinations of morphometric variables. There is a wide range of publications that provide information about rolling mills, roll materials, and rolled products that all fit into the same category with regard their technical configurations, chemical compositions of processed materials, etc. In order to apply multivariate analysis on morphometric data there is a need to invent appropriate generic functions that assign to each observation a unique combination of coefficients. Preliminary simplified analyses of morphometric data for symmetrical products have shown statistically significant correlations [9, 10, 24, 26].
Roll pass design (RPD) is a concept that is embedded in the heart of rolling mill technology. Bearing in mind that RPD is one of the most influential aspects, an assumption has emerged that significant patterns and correlations can be observed by analysing morphometric RPD series. Since all published designs of rolling series aimed to achieve the best key performance indicators, it follows that the statistical patterns and correlations can be extracted from this database. When these patterns are combined with empirical knowledge (e.g. spread calculations), probabilistic design allows for anticipating morphometric improvements for each individual pass. For example, minute corrections can be introduced to counterbalance the inevitable wear of the roll surface, when the analysed case statistically falls on the wrong side of the common trend [9, 10, 24].
The complexity of applied RPD depends on the level of expectations; some designs are intended to enable energy or productivity effective schedules, others aim at more exact morphometry or the best thermo-mechanical processing for the product microstructure. Roll grooves can be designed in a variety of ways at the same cost, but with significant economic consequences for both maintenance and operations. A complete RPD needs to take into account the gradual change in the groove contour during the rolling campaigns due to roll wear. As discussed above, this class of optimisations can be pursued based on knowledge extraction from databases constructed from industrial records.
First step in developing Big Data RPD is to describe rolling processes by means of structured matrices such as illustrated by Eq. (1):
A =
X11 X12 X21 X22
(1)
mn
In order to apply multivariate analysis on morphometric records defining more complex grooves there is a need to invent appropriate generic functions that assign to each observation a unique combination of coefficients.
The elements in the matrix A must present morphometric parameters for the subsequent passes along the rolling sequence. These parameters are successfully defined using Chebyshev polynomials to provide a generic formulation that filters out the insignificant differences in morphometric RPD records. In this way the principal features of the rolling pass sequence are represented by di-mensionless parameters. This multivariate representation provides significantly more information which further enables making inferences about technological RPD reserves of the process [26].
Although a significant correlation can be inferred based on morphometry of rolling passes exclusively, for extracting more comprehensive knowledge about the process the matrix A needs to be further extended. This is done by including additional process parameters of interest (e.g. product and tool specifications such as chemical composition, and operation variables such as rolling temperature and velocity).
Matrix B components can be summarised as follows:
1. productivity;
2. yield;
3. reliability;
4. quality;
5. costs.
Ultimately, matrix A needs to be correlated with corresponding matrix B where the process output parameters are embraced. When knowledge about mutivariate distributions of RPD parameters is presented in a conceivable fashion, the reliability of predictions increases proportionally to the quantity of the observations and the predictors taken [26].
The quantity of published theoretical works addressing plastic forming by rolling and related solid mechanics and materials science is impressive. However, the resulting models do not offer sufficiently close approximations to real actualisations in the case of the design of rolling process, simply because they fail to include all significant variables. In addition, it appears that idealisation of experimental designs in laboratories and the resulting deterministic models undermine their service as possible realisations in practice [27].
Theoretical derivations and idealised experiments are caught in the trap of the first principles and so-called natural laws. Practical applications (empowered by today's improved detectors and sensors) indicate that the "laws" hold only under limited circumstances that often do not reflect the real situations of practical importance. Complex mathematical formalisations appear to serve only for the self-purpose of theory itself - not for the purposes of practice, i.e. for solving the problem as it has been stated. As explained by Griesemer [28]: "A theory does not count as formalized unless its presentation in a conceptual notation is actually used in these practices. In other words, formalization is a practice, not merely a state, property, or condition of a theory. Successful formalization affords increased facility or even makes possible new kinds of theoretical and empirical practice (that were) unavailable without (such) formalization".
Theories of physics of solids, continuum mechanics, elastoplasticity and thermodynamics provide the grounds on which the industrial processes of rolling are designed and constructed initially. Yet, these methods alone do not provide sufficient optimisation effects in the actual manufac-
Sequel
turing operations. This means that costly corrections and trials must be undertaken at the resource-consuming industrial scale. In this scenario the information about the multifactorial correlation between the critical dependent variables and actually independent and controllable parameters is decisive. The knowledge of how we can affect the outcomes within the scope of real controls must be ranked highly in the hierarchy of modelling of the rolling process [27].
Although the sensors, actuators and the logics of pertinent loop control are constructed based on exact scientific theories, the key idea of their operation is the so-called "black-box" principle. We do not need to know exactly why the output changes due to other factors, as long as we know that by changing a limited number of the input control factors we can control the process. The prerequisite for this is to define the system operation and maintenance statistics, which are ultimately estimated by means of a statistically significant number of repetitions. Records of such repetitions are abundant in rolling mills, however, published sources providing such databases are rare [27].
Attempts were made to share in such data bases with institutions and companies involved in similar research and fabrication. Communication with potential industrial stakeholders was met with initial interest that advanced only to a point where access to industrial data bases was requested. The industrial correspondents exhibited an interest in receiving, however, not in providing, information.
On the other side, academics are interested in collaboration related to this topic; the following are several examples of institutions from which the positive responses were received over recent 5 years.
1) The University of West Bohemia; contact: Prof. Dr. Ing. Bohuslav Masek, Director FORTECH (e-mail: [email protected] and [email protected]) The Research Centre of Forming Technology, Czech Republic (http://fortech.zcu.cz).
2) National University of Science and Technology "MISIS", http://en.misis.ru/; contact: Dr. Viacheslav V. Brinza, Director, The Scientific and Research Center of Technological Forecasting (e-mail: [email protected]) Russia (misis.ru/spglnk/b3631767).
3) The University of Adelaide; contact: A. Prof. Reza Ghomashchi, School of Mechanical Engineering (http://mecheng.adelaide.edu.au) (e-mail: [email protected]) Structures and Materials Group, Australia (www.mecheng.adelaide.edu.au/mrg/).
4) The University of South Australia, http://www.unisa.edu.au/ Contact: A. Prof. Kazem Abhary (http://people.unisa.edu.au/Kazem.Abhary) School of Engineering, (http://www.unisa.edu.au/IT-Engineering-and-the-Environment/School-of-Engineering) (e-mail: [email protected]).
5) CQ university Australia, Contact: Dr. Ramadas Narayanan, School of Engineering Technology, (https://www.cqu.edu.au/), Australia, Qld 4670 (e-mail: [email protected]).
6) Faculty of Metallurgy and Materials (http://www.famm.unze.ba/) - a member of the University of Zenica (http://www.unze.ba); contact: Prof. Faik Uzunovic e-mail: [email protected] (Head of the Department for Working-Processing of Metals.
Conclusions
Initiative for employing Big Data strategy to RPD is motivated by recognising the following.
(i) The need for sustaining industrial practice in operating rolling mills that deliver current and develop new products of continuously increasing quality and quantity (volume of manufactured steel has surpassed 1,6 billion tons per year).
(ii) The necessity for urgently reducing the natural resource consumption and carbon emission caused by the above industry.
(iii) The demand for activating multi-disciplinary research synergy by connecting academics and professional engineers at a level of coordinated international collaboration.
(iv) The urge for unlocking access to growing industrial and academic databases and take advantage of capability of information technology and communication systems to share the extracted knowledge.
(v) The need to shift the competition attitude from an emphasis on short-sighted profit and knowledge ownership towards a competition based on inventiveness in applying advanced technology, sustainable operations, and constructive distribution of profit.
Along with the well established statistical merits, the criteria for evaluating the Big Data RPD strategy include the following questions:
(i) whether or not it incites continuous improvement in product quality, and improved control with regard to morphometric, mechanical, metallurgical and other aspects as specified by standards;
(ii) does it lead to an increase in production yield (due to a decrease in scrap and losses caused by production interruptions);
(iii) how it enhances and increases productivity (due to a decrease in the frequency and duration of the delays, and due to decrease in the overall duration of the manufacturing campaigns);
(iv) whether it assists in improving the system reliability (due to extended tool life); and
(v) does it lead to a decrease in the overall sustainability of the operations, both with regard to commercial viability of the metalworking plant and with regard to environmental impacts?
Big data paradigm, which provides designers with powerful tools for transforming intricate records accumulated in manufacturing systems into useful knowledge, opens promising avenues in search for improving design of complex systems such as rolling mills.
References
1. S. Spuzic, R. Narayanan, P. Gudimetla, Big Data Model - An Application to Design of Rolling Process, keynote lecture presented at the International Conference on Innovative Material Science and Technology (IMST2016), Shenzhen, China, 19-21 August 2016.
2. S. Kalpakjian, Manufacturing engineering and technology (Pearson Education India, 2013).
3. Y. Lee, Rod and bar rolling: theory and applications (CRC Press, 2004).
4. M. Poursina, N. T. Dehkordi, A. Fattahi and H. Mirmohammadi, Application of genetic algorithms to optimization of rolling schedules based on damage mechanics, Simulation Modelling Practice and Theory, 22, 61-73 (2012).
5. R.A. Muniz, Non-Linear Finite Element Method Simulation and Modeling of the Cold and Hot Rolling Processes (Virginia Polytechnic Institute and State University, 2007).
6. A. Dubois, E. Luc, M. Dubar, L. Dubar, C. Thibaut and J.-M. Damasse, Initiation of sticking during hot rolling of stainless steel plate, Procedia Eng, 81, 1958-1963 (2014).
7. J. Schey, Introduction to manufacturing processes (McGraw-Hill, New York 1999).
8. J. Arts, A multi-item approach to repairable stocking and expediting in a fluctuating demand environment, Eur. J. Opera.l Res. (2016).
9. K. Abhary, K. Garner, Z. Kovacic, S. Spuzic, F. Uzunovic and K. Xing, A Knowledge Based Hybrid Model for Improving Manufacturing System in Rolling Mills, Key Eng. Mater., 443, 3-8 (2010).
10. K. Abhary, Z. Kovacic, S.-E. Lundberg, R. Narayanan and S. Spuzic, The application of a hybrid algorithm to roll pass design, Int. J. Adv. Manu. Tec. (2015).
11. E. Appleton and E. Summad, eds., Roll Pass Design: A 'Design for Manufacture'View, 2nd European Rolling Conference, AROS Congress Center, Vasteras, Sweden, pp. 24-26 (2000).
12. S. Riljak, Numerical simulation of shape rolling (Stockholm: Royal Institute of Technology 2006).
13. F. Lambiase and A. Langella, Automated procedure for roll pass design, J. Mater. Eng. Perform., 18, 263-272 (2009).
14. V. Oduguwa and R. Roy, A review of rolling system design optimisation, Int.J. Machi.To. Manu., 46, 912-928 (2006).
15. E. Letouze et. al., Big data for development: Challenges & opportunities, New York: UN Global Pulse (White Paper): Big Data for Development: Opportunities & Challenges (2012). Retrieved on 13 April 2016 from http://www.unglobalpulse.org/projects/BigDatafor Development.
16. M. Hilbert, Big data for development: From information-to knowledge societies, United Nations Economic Commission for Latin America and the Caribbean, Retrieved on 25 July 2016 from http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2205145.
17. "Manufacturing: Big Data Benefits and Challenges". TCS Big Data Study. Mumbai, India: Tata Consultancy Services Limited. Retrieved on 3rd Jun 2014 from http://sites.tcs.com/big-data-study/manufacturing-big-data-benefits-challenges/#.
18. J. Lee, F. Wu, W. Zhao, M. Ghaffari, L. Liao and D. Siegel, Prognostics and health management design for rotary machinery systems-Reviews, methodology and applications, Mech. Sys. sig. process. 42, 314-334 (2014).
19. O. Gregory, correspondence with the Director, ARC Research Hub for Australian Steel Manufacturing (2014).
20. V. Brinza, I. Kuznetsov and A. Yur'ev, Geometric conditions in bar rolling, Steel in Translation, 44, 517-522 (2014).
21. V. Brinza and I. Kuznetsov, Modeling the Laws that Determine the Quality of Rolled Rounds During the Rolling Operation, Metallurgist, 58, 510-515 (2014).
22. S. Spuzic, D. Hapu Arachchige, Z. Kovacic, K. Abhary and R. Narayanan, Some trends in roll design for manufacture of long products, Steel and metallurgy - a complete magazine special issue on 2nd international conference on rolling and finishing technology of Steel, 17, No.11, 32-39, (2015).
23. K. Schroder, A basic understanding of the mechanics of rolling mill rolls, Eisenwerk Sulzau-Werfen, ESW-Handbook, 54-56 (2003) Retrieved on 23 Jun 2016 from http://www.brcil.com/common/downloads/technical/Rolling_Mill_Rolls.pdf.
24. S. Spuzic and K. Abhary, A Contribution to Rolling Mill Technology - Roll Pass Design Strategy for Symmetrical Sections, Der Kalibreur, 75, 14-27 (2014).
25. S. Byon, D. Na and Y. Lee, Effect of roll gap adjustment on exit cross sectional shape in groove rolling-Experimental and FE analysis, J. Mater. Process. Technol., 209, 4465-4470 (2009).
26. S. Spuzic, R. Narayanan, Z. Kovacic, D. Hapu Arachchige, K. Abhary (2016) Roll Pass Design Optimisation, Int. J. Adv. Manuf. Tech. (accepted for published in the International Journal of Advanced Manufacturing Technology).
27. D. Mulcahy, S. Pignata, N. Rajendhiran, S. Spuzic, F. Uzunovic, R. Narayanan, N. Vaikundam and K. Fraser, eds., Some issues related to knowledge transfer in postgraduate research and education, 11th Biennial QPR Conference - Quality in Postgraduate Research, Adelaide (2014).
28. J. Griesemer, Formalization and the meaning of "theory" in the inexact biological sciences, Biological Theory, 7, 298-310 (2013).
