Constrained, multipoint shape optimisation for complex 3D configurations

A method for performing three-dimensional, multipoint, lift-constrained drag minimisation for complex aircraft configurations is presented. Parameters representing the aircraft geometry are the design variables used in the solution of an optimisation problem. The compressible Euler equations for the flow are discretised on automatically generated unstructured meshes, and the sensitivities of the objective function and the constraints with respect to the design parameters are efficiently calculated using the discrete adjoint method. In addition, the solution algorithm has been parallelised making the approach feasible for practical applications. Several constrained minimisation strategies are discussed and some numerical tests are carried out for a lift-constrained, two-dimensional problem. The strategy found to work best is demonstrated for the three-dimensional optimisation of a wing-body configuration operating in transonic and supersonic conditions. It is concluded that Euler-based optimisation can be useful as a first step in the design process but that most applications involving transonic flows do require viscous flow modelling to avoid unrealistic pressure distributions.