The optimization algorithm using Space Mapping has shown to be well suited for problems in crashworthiness design and sheet metal forming. All examples converged to the correct optimum and the computing time has been decreased with a maximum of 53% relative to traditional Response Surface Methodology (RSM) optimization. The square tube problem with two design variables converged using the Space Mapping optimization in 8.87 hours, which was a 53% reduction in computing time compared to the fastest traditional RSM optimization method. The coarse model was very simple to set-up when the fine model was developed. Only one small modification to the fine model to include mass scaling was introduced. The constraint in this optimization was the displacement of the rear end. The differences in the results from the fine and coarse evaluations were rather large, but still the algorithm converged. The Space Mapping algorithm also converged for the vehicle impact problem. The computing time for each evaluation was decreased from 12 h to 2 h, when the coarse model was used. If a complete vehicle model should be optimized the time saving can be much larger than in these examples, since the simulation time for a full vehicle model today can be more than 100 h. The algorithm using Space Mapping can then significantly reduce the total simulation time. Generally the optimization algorithm using Space Mapping worked well for all problems and the CPU time was often significantly reduced. This time saving must also be related to the time it takes to produce a coarse model. If the model building is time consuming, perhaps it is most time efficient to use a traditional RSM optimization. This must be taken into account when an optimization method is chosen The algorithm using Space Mapping seems to be less stable compared to the traditional RSM method, e.g. if a bad starting design is chosen Space Mapping might not converge due to a too bad mapping between the fine and coarse models. The starting design must be 'intelligently' chosen based on knowledge on how the Space Mapping technique works and how the model behaves for parameter changes. The Space Mapping method needs to be tested on larger examples with more design variables and more constraints. In addition a recommendation must be established on how much the results from the models can differ and how to choose the starting design. Other coarse models e.g. analytic solutions or other less computing time demanding solvers can also be used.


    Access

    Access via TIB

    Check availability in my library


    Export, share and cite



    Title :

    Structural optimization using space mapping and surrogate models


    Contributors:
    Redhe, M. (author) / Nilsson, L. (author)


    Publication date :

    2002


    Size :

    12 Seiten, 9 Bilder, 2 Tabellen, 15 Quellen


    Type of media :

    Conference paper


    Type of material :

    Print


    Language :

    English







    AIAA-2002-5536 Using Space Mapping and Surrogate Models to Optimize Vehicle Crashworthiness Design

    Redhe, M. / Nilsson, L. / American Institute of Aeronautics and Astronautics et al. | British Library Conference Proceedings | 2002


    Managing Surrogate Models in Optimization

    Keane, Andy J. / Nair, Prasanth B. | Wiley | 2005