Int. J. Simul. Multidisci. Des. Optim.
Volume 2, Number 4, December 2008
|Page(s)||273 - 282|
|Published online||27 February 2009|
MPI-enabled Shape Optimization of Panels Subjected to Air Blast Loading
Mechanical and Nuclear Engineering, Pennsylvania State University, University Park, PA 16802, USA
2 Research Computing and Cyberinfrastructure, Pennsylvania State University, University Park, PA 16802, USA
3 Civil, Environmental and Sustainable Engineering, Arizona State University, Tempe, AZ 85287, USA
Accepted: 25 September 2008
The problem of finding the optimal shape of an aluminum plate to mitigate air blast loading is considered. The goal is to minimize the dynamic displacement of the plate relative to the test fixture, while monitoring plastic strain values, mass, and envelope constraints. This is a computationally challenging problem owing to (a) difficulty with finding optimal shapes with higher dimensional shape variations, (b) non-differentiable, non-convex and computationally expensive objective and constraint functions, and (c) difficulties in controlling mesh distortion that occur during explicit finite element analysis. An approach based on coupling LS-DYNA finite element software and a differential evolution (DE) optimizer is presented. Since DE involves a population of designs which are then crossed-over and mutated to yield an improved generation, it is possible to use coarse parallelization wherein a computing cluster is used to evaluate fitness of the entire population simultaneously. However, owing to highly dissimilar computing time per analysis, a result of mesh distortion and minimum time step in explicit finite element analysis, implementation of the parallelization scheme is challenging. Sinusoidal basis shapes are used to obtain an optimized 'double-bulge' shape.
Key words: structural optimization / shape optimization / blast loads / MPI / load balancing / differential evolution / parallel computing
© ASMDO, 2008
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.