Multi-model management for time-dependent multidisciplinary design optimization problems
- PDF / 3,153,142 Bytes
- 21 Pages / 595.224 x 790.955 pts Page_size
- 75 Downloads / 179 Views
RESEARCH PAPER
Multi-model management for time-dependent multidisciplinary design optimization problems Ahmed H. Bayoumy1 · Michael Kokkolaras1 Received: 15 October 2019 / Revised: 27 February 2020 / Accepted: 6 March 2020 © Springer-Verlag GmbH Germany, part of Springer Nature 2020
Abstract In previous work, we presented a relative adequacy framework (RAF) for multi-model management in single- and multidisciplinary design optimization to select the most efficient model in different areas of the design space as the latter is being explored during the optimization process. In this work, we extend our framework to allow the consideration of timedependent multidisciplinary design optimization (MDO) problems with particular emphasis on strongly coupled transient fluid-structure interaction (FSI). The computational models used for analysis in the problems under consideration are typically not amenable to reliable and cost-effective approximation of gradients. Therefore, we use the mesh adaptive direct search optimization algorithm in conjunction with a trust-region approach for multi-model management. The algorithm is modified to automate the search for adequate modeling parameters during the time-dependent multidisciplinary analysis. We show that multidisciplinary temporal interactions may have a significant impact on multi-model management; models that could have been selected in a time-invariant MDO environment may be inadequate in certain areas of the design space in a time-dependent MDO context. At the same time, the relative adequacy framework may improve multidisciplinary temporal analysis convergence and/or MDO coordination. We demonstrate the proposed RAF for multi-model management in time-dependent MDO using a flexible plate FSI example. Keywords Multidisciplinary analysis · Multidisciplinary design optimization · Relative adequacy framework
1 Introduction Numerical design optimization of engineering systems relies on computational models to evaluate the objective and constraint functions of the formulated problem. The behavior of multidisciplinary engineering systems is predicted not only by conducting the analysis of individual disciplines separately, but by capturing and accounting for their interactions, which requires computationally intensive multidisciplinary analysis (MDA) convergence. When multiple models are available, the question becomes how to exploit them in order to attain solutions Responsible Editor: Palaniappan Ramu Ahmed H. Bayoumy
[email protected] Michael Kokkolaras [email protected] 1
Department of Mechanical Engineering, McGill University, Montreal, Quebec, H3A 0C3, Canada
efficiently. In previous work, we presented a relative adequacy framework (RAF) for multi-model management in simulation-based design optimization to select the most efficient model in different areas of the design space as the latter is being explored during the optimization process (Bayoumy and Kokkolaras 2019). In this paper, we extend this framework to consider time-dependent multidisc
Data Loading...
