This paper presents a comparison of methods for aerostructural analysis and optimization. The aerostructural analysis problem is solved in parallel using a panel method coupled to a finite-element solver. The coupled nonlinear aerostructural system is solved using a nonlinear block Gauss-Seidel, nonlinear block Jacobi, Newton-Krylov or approximate Newton-Krylov approach. The approximate Newton-Krylov method is shown to be an efficient and robust solution technique. An adjoint-based sensitivity method is developed that achieves a high-level of accuracy when compared to complex-step calculations. Three levels of parallelism are exploited within the present aerostructural optimization framework: optimization-level, system-level and discipline-level parallelism. The efficient and robust solution method and accurate gradient evaluation technique provide a powerful tool for aerostructural design optimization. Aerostructural induced drag minimization results are presented for a typical subsonic turboprop aircraft wing.
展开▼
