Proceedings home preface contents authors keywords copyright reference ©2012 Civil-Comp Ltd				Paper 203 An Open Computational Framework for Reliability Based Optimization E. Patelli and M. de Angelis Centre for Engineering Sustainability, School of Engineering, University of Liverpool, United Kingdom Keywords: reliability based optimization, Matlab, open source, high performing computing. full paper (pdf) - reference In engineering practice today, optimization is almost an indispensable step of the design cycle for any product or component. Optimize means the design of a better product or system that can reach significant reductions in terms of the manufacturing and operating costs, as well as the improvement in the performance. However, these products are affected by uncertainties, caused by lack of sufficient knowledge and, or by natural unpredictable external events. In order to cope with this problem and to guarantee that the components or systems will continue to perform satisfactory despite fluctuations and changes of model (e.g. arising from production processes) and environmental conditions (e.g. arising from climate change), the design has to be "robust". Consequently, the field of optimization has been coupled with reliability analysis forming the so-called reliability based optimization analysis [1,2]. In this paper, an open computational framework for reliability-based optimization is presented. Developed in an object oriented fashion in Matlab environment, this framework provides the necessary flexibility, modularity and usability to be adopted in different contexts. As a result of the terms of the LGPL license [3] adopted, that anyone is permitted to use, verify and modify the proposed framework or derived code from it. This framework is used with the general purpose software COSSAN-X [4]. This framework allows a reliability based optimization to be performed by adopting the direct approach, global and local surrogate models [5]. It enables a combination of the state of the art in reliability analysis (e.g. adopting importance sampling, line sampling) in the direct approach as well in the construction of the meta-models (e.g. response surface, artificial neural networks, kriging models and polynomial chaos). A set of widely used gradient-based and gradient-free optimization algorithms (e.g. SQP, Cobyla, genetic algorithms, simulated annealing etc. [6]) are also available. The framework has been designed to provide the maximum flexibility in combining the different components required by the reliability based optimization and to use efficiently the high performance computing capability (grid and cloud computing). The applicability and the flexibility of the proposed framework for solving real-life problems is demonstrated with two applications considering static and dynamic loading. References 1 R. Rackwitz, "Reliability analysis - a review and some perspectives", Structural Safety, 23, 365-395, 2001. 2 G.I. Schuëller, "Efficient Monte Carlo simulation procedures in structural uncertainty and reliability analysis - recent advances", Journal of Structural Engineering and Mechanics, 32(1), 1-20, 2009. 3 Free software foundation, "GNU lesser general public license". http://www.gnu.org/licenses/lgpl.html 4 E. Patelli, H.M. Panayirci, M. Broggi, B. Goller, P. Beaurepaire, H.J. Pradlwarter, G.I. Schuëller, "General purpose software for effficient uncertainty management of large finite element models", Finite Elements in Analysis and Design, 51, 31-48, 2012. 5 A.R. Conn, K. Scheinberg, L.N. Vicente, "Review of surrogate model management", 2009. 6 J. Arora, (Editor), "Optimization of Structural and Mechanical Systems", World Scientific, 2007.