Surrogate based shape optimization of airship envelopes

The shape of an envelope plays a critical role in design and optimization of an airship. This paper investigates the efficacy of an existing shape generation algorithm in arriving at candidate shapes for an airship envelope. The envelope shape is parametrized in terms of five parameters which can be varied to obtain various candidate shapes, meeting certain geometrical constraints. To predict the drag coefficient of the generated shape more accurately and efficiently, a surrogate model is developed and validated against a two-variable test function. The surrogate model is seen to predict drag coefficient for the generated shape with less than 1% error with respect to CFD results. Using this surrogate model, optimal envelope shapes are obtained by coupling the surrogate model to an evolutionary optimization method. The results indicate that the shape generation algorithm is able to faithfully capture the geometry of four standard airship envelope shapes. The optimum shapes obtained by this approach are seen to have 4 to 8 % lower drag coefficient as compared to a standard shape at low and high Reynolds numbers.