eChallenges e-2014 Conference Proceedings Paul Cunningham and Miriam Cunningham (Eds) IIMC International Information Management Corporation, 2014 ISBN: 978-1-905824-45-8 Copyright © 2014 The Authors www.eChallenges.org Page 1 of 9 Harnessing Heterogeneous Computational Infrastructures for Studying Metallurgical Rolling Processes Dariusz KRÓL 1 , Renata SŁOTA 1 , Łukasz RAUCH 2 , Jacek KITOWSKI 1 , Maciej PIETRZYK 2 1 AGH University of Science and Technology, Department of Computer Science and Academic Computer Centre Cyfronet AGH, Krakow, Poland, Email: dkrol@agh.edu.pl 2 AGH University of Science and Technology, Department of Applied Computer Science and Modelling, Krakow, Poland, Email: lrauch@agh.edu.pl Abstract: The paper describes the application of heterogeneous computational infrastructures to study complex metallurgical processes. This goal is achieved by integration of a domain-oriented system (VirtRoll) with a platform for massive parameter studies (Scalarm) on the basis of Service Oriented Architecture (SOA). In particular, technological and security aspects of the integration and permissions delegation are discussed. We describe: a workflow of studying metallurgical processes with the parameter study approach, the domain-oriented system for conducting metallurgical experiments, its integration with the platform for parameter studies along with its enhancements, a sample use of the developed solution and its potential business benefits. 1. Introduction In the last years the increasing demand of high and ultra high strength hot rolled steel strips, which are utilized in many branches of industry, e.g. automobile, is driving the redefinition of metallurgically based conceptual design of hot rolling strip mills. Attaining desired mechanical properties of hot rolled Advanced High Strength Steels (AHSS) and Ultra High Strength Steels (UHSS) strips requires realization of thermo-mechanical schemes, in terms of time/temperature/deformation along the hot rolling process. For many years now, computer simulation has been applied in various computational- oriented science disciplines and in the industry to verify stated hypothesis easier and more cost-effectively in comparison to physical experiments. As an example, properties prediction of material created in metal forming processes, e.g. rolling and cooling, requires expensive, long lasting experimental trials, which do not guarantee identification of the optimal technological parameters. Recent technological advances have led to significant improvements in computer simulations, reducing the time required to run a simulation and enabling refinement of simulation models with regard to their complexity [1]. Besides accelerating simulations, modern high-performance computing infrastructures are capable of processing much more data in a given time interval than ever before. As a result, many complex processes can finally be modeled during the time, which is satisfactory for industry. This justifies the development of a model-based computer system (VirtRoll) in the Research Fund for Coal and Steel (RFCS) VirtROLL project, which combines numerical simulations, multiscale modeling, meta-modelling, inverse analysis and optimization techniques to minimize costs of design of production technologies and to optimize semi- and final product properties. In a nutshell, the main objective of the project is to combine a model database and inverse solution coupled with optimization techniques in one