Adjoint methods for the efficient computation of aerodynamic derivates with high-fidelity CFD

Abstract

This paper presents the development of a discrete adjoint method by means of automatic differentiation in the framework of the Helicopter Multiblock CFD solver. The method is suitable for applications in flight mechanics as well as shape optimisation and is demonstrated in this paper for cases reported in the literature. The application of automatic differentiation is first presented for a simple flight mechanics software with indicative results for ADS-33 maneouvres. Subsequently adjoint CFD computations were undertaken for aerofoil, wing and rotor blade cases. The obtained results were found to agree well with other published solutions or with data obtained using finite differences for computing the flow derivatives. The method has so far been demonstrated for inviscid flow cases and suggests that the current implementation is robust and efficient. The cost of the adjoint computations is relatively low due to the employed source code differentiation and most of the times it is no more than the cost for a steady-state flow solution.