Lukasz Rauch

Digital Material Representation as an efficient tool for strain inhomogeneities analysis at the micro scale level

The summary of recent research towards development of a tool for detailed microstructure modelling is presented within the paper. The main focus is put on micro scale behaviour, where advantages of digital material representation can be taken into account. Digital Material Representation allows modelling of microstructures along with features such as crystallographic orientation, grain boundaries or phase boundaries represented in an explicit manner. Incorporation of these digital microstructures into the numerical simulation methods provides the possibility to improve the quality of numerical results. The developed method can be used to design specifically dedicated microstructures, which meet very strict requirements. The clear motivation and importance of the work is presented in the first part of the paper followed by a short description of the developed approaches for creation of the digital microstructures. Two approaches are considered that provide an exact and statistical representation of the real microstructure. The main focus is put on the application of image processing and cellular automata techniques. Afterwards, obtained digital microstructures are used as input data for the finite element analysis of the micro scale compression test. Examples of applications during multiscale simulation are also presented in the paper.

From High Accuracy to High Efficiency in Simulations of Processing of Dual-Phase Steels

Searching for a compromise between computing costs and predictive capabilities of metal processing models is the objective of this work. The justification of using multiscale and simplified models in simulations of manufacturing of DP steel products is discussed. Multiscale techniques are described and their applications to modeling annealing and stamping are shown. This approach is costly and should be used in specific applications only. Models based on the JMAK equation are an alternative. Physical simulations of the continuous annealing were conducted for validation of the models. An analysis of the computing time and predictive capabilities of the models allowed to conclude that the modified JMAK equation gives good results as far as prediction of volume fractions after annealing is needed. Contrary, a multiscale model is needed to analyze the distributions of strains in the ferritic-martensitic microstructure. The idea of simplification of multiscale models is presented, as well.

System for design of the manufacturing process of connecting parts for automotive industry

The proposition of complex hybrid system, dedicated to modelling of life cycle of materials and optimization of their in use properties, is presented in the paper. The approach is based on the conventional optimization algorithms, FE simulations of industrial production process and knowledge base, containing both theoretical and practical data in form of rules, facts and equations. Simulation and optimization of the manufacturing of the connecting part used in automotive industry was selected for the purposes of this work. The particular emphasis is put on control of selected in use properties of products by proper design of technological parameters for consecutive stages of the production chain. The concept of the life cycle modelling used in the proposed system, as well as results obtained from simulations, are also presented in the paper.

Identification of Multi-inclusion Statistically Similar Representative Volume Element for Advanced High Strength Steels by Using Data Farming Approach☆

Abstract Statistically Similar Representative Volume Element (SSRVE) is used to simplify computational domain for microstructure representation of material in multiscale modelling. The procedure of SSRVE creation is based on optimization loop which allows to find the highest similarity between SSRVE and an original material microstructure. The objective function in this optimization is built upon computationally intensive numerical methods, including simulations of virtual material deformation, which is very time consuming. To avoid such long lasting calcu- lations we propose to use the data farming approach to identification of SSRVE for Advanced High Strength Steels (AHSS) characterized by multiphase microstructure. The optimization method is based on a nature inspired approach which facilitates distribution and parallelization. The concept of SSRVE creation as well as the software architecture of the proposed solution is described in the paper in details. It is followed by examples of the results obtained for the identification of SSRVE parameters for DP steels which are widely exploited in modern automotive industry. Possible directions for further development as well as possible industrial applications are described in the conclusions.

OpenCL implementation of cellular automata finite element (CAFE) method

The implementation of multiscale numerical simulations on heterogeneous hardware architectures is presented in the paper. The simulations are composed of coupled micro and macro scale approaches, which are implemented by using cellular automata and finite element method respectively. Details of both of these methods are described in the papers as well. The simulations were performed for the problem of material heat treatment (macro scale) with simultaneous application of grain growth calculation (micro scale). Comparison of quantitative results obtained by using separated and coupled computing methods are presented in form of speedup and efficiency coefficients.

Hybrid System Supporting Flexible Design of Flat Rolling Production Processes in Collaborative Environment

The paper is devoted to advanced production processes design, based on numerical simulations of material behaviour under complex loading conditions. The computer system proposed in this work facilitates creation of the sophisticated flat rolling facilities composed of different subesequent stages e.g. heating, roughing and finishing mills, cooling, cutting, descaling. Each stage is treated as a separated module with its own features and methods that implement its functionality. However, the most demanding part of proposed system lies in reliable simulation of connection between these separated modules. To deal with this the highly fexible numerical solutions are required. Creation of this approach is the main goal of the work and is described in details including examples obtained results. Disscusion on accurate material models taking into account dynamic recrystallization or grain growth as well as on application of the optimization procedures inorder to obtain desired final properties is also presented in the paper.

Selection of the optimization method for identification of phase transformation models for steels

ABSTRACT Material models for steels, used widely in numerical simulations of manufacturing chains, require identification of their coefficients on the basis of measurements obtained from laboratory test. Precision of the identification highly influences modelling reliability. This is visible especially in the case of phase transformation models, which are crucial in predicting of the modern Advanced High Strength Steels (AHSS) properties after applied heat treatment. However, identification of phase transformation models for steels based on dilatometric tests presents serious difficulties. Two problems are investigated in the paper i.e. (i) efficiency of the inverse algorithms used for identification of phase transformation models, (ii) final reliability of the identified models in numerical simulations of manufacturing processes. In the work two phase transformation models were selected as an example. The first was a modified JMAK (Johnson-Mehl-Avrami-Kolmogorov) equation. The second was an upgrade of the Leblond equation, in which second order derivative with respect to time was introduced. The identification was performed by coupling the selected model with nature inspired optimization techniques and performing inverse analysis for the experimental data. Dilatometric tests performed for various cooling rates were used as an experiment, which supplies data for the inverse analysis. Finally, validation of identified models is presented by using industrial data.

Sensitivity analysis on HPC systems with Scalarm platform

Sensitivity analysis is widely used in numerical simulations applied in industry. The robustness of such applications is crucial, which means they have to be fast and precise at the same time. However, the conventional approach to sensitivity analysis assumes realization of multiple execution of computationally intensive simulations to discover input/output dependencies. In this paper, we present a novel computational approach for performing large‐scale sensitivity analysis integrated with an extended platform for parameter studies – Scalarm – to make use of modern e‐infrastructures for distribution and parallelization purposes, profitable for complicated industrial problems. Copyright

Model-Based Approach To Study Hot Rolling Mills With Data Farming.

The paper describes a computer system for simulating metallurgical rolling processes that consist of multiple steps, each of which is performed by a different type of devices. Both devices and processed materials are described with models, which can be dynamically reconfigured between simulation runs to study different device and environment configurations. Such an approach is especially crucial in technology design based on multi-iterative optimization procedures, for which an objective function uses computationally intensive algorithms. Due to the approach proposed in this paper, in the first stage of optimization more general and coarse models can be applied characterized by lower predictive capabilities and higher computational efficiency. Afterwards, when the optimization procedure finds a solution close to the optimal one, very detailed models can be used to obtain high quality solutions in the last few steps of calculations. To achieve such an objective a hybrid computer system able to use High Performance Computing (HPC) infrastructures was designed and implemented. The details of proposed approach are described, which is followed by presentation of a data farming platform responsible for distribution of complex numerical simulations onto various computer clusters. Finally, a concrete use case of a hot rolling mill is presented and analyzed.

Massively Parallel Approach to Sensitivity Analysis on HPC Architectures by Using Scalarm Platform

Sensitivity Analysis is widely used in numerical simulations applied in industry. The robustness of such applications is crucial, which means that they have to be fast and precise at the same. However, conventional approach to Sensitivity Analysis assumes realization of multiple execution of computationally intensive simulations to discover input/output dependencies. In this paper we present approach based on Scalarm platform, allowing to accelerate Sensitivity Analysis calculations by using modern e-infrastructures for distribution and parallelization purposes. The paper contains both description of the proposed solution and results obtained for a selected industrial case study.