A multi-model incremental adaptive strategy to accelerate partitioned fluid-structure algorithms using space-mapping

Abstract

High fidelity analysis of fluid-structure interaction systems is often too timeconsuming when a large number of model evaluations are required. The choice for a solution procedure depends often on the efficiency of the method and the possibility of reusing existing field solvers. Aggressive Space-Mapping, a technique originally developed for multi-fidelity optimization, is applied to accelerate the partitioned solution procedure of a high fidelity fluid-structure interaction model. The method supports software modularity. Aggressive Space-Mapping (ASM) is applied to an academic testcase and the results are compared with the corresponding Incremental Quasi-Newton (IQN) method. An efficiency metric is defined to facilitate the comparison. The ASM method is found to be more efficient than the corresponding IQN method for the testcases considered. The efficiency of space-mapping increases with increasing fluid-to-structure mass ratio, indicating that the method is especially useful for strongly coupled problems.