MULTI-FIDELITY DESIGN OPTIMIZATION OF AXISYMMETRIC BODIES IN INCOMPRESSIBLE FLOW

The paper discusses multi-fidelity design optimization of axisymmetric bodies in incompressible fluid flow. The algorithm uses a computationally cheap low-fidelity model to construct a surrogate of an accurate but CPU-intensive high-fidelity model. The low-fidelity model is based on the same governing equations as the high-fidelity one, but exploits coarser discretization and relaxed convergence criteria. The low-fidelity model is corrected by aligning the hull surface pressure and skin friction distributions with the corresponding distributions of the high-fidelity model using a multiplicative respimse correction. Our approach can be implemented in both direct and inverse design approaches. Results of two case studies for hull drag minimization and target pressure distribution matching show that optimized designs are obtained at substantially lower computational cost (over 94%) when compared to the direct high-fidelity model optimization.