Aerostructural design of laminar and dry wings using multidisciplinary Bayesian optimization

Wings of conventional kerosene-powered aircraft are designed to fulfill their fuel storage requirements. However, the absence of fuel from the wing constitutes an important feature of hydrogen aircraft as opposed to their kerosene-based counterparts, potentially paving a way to unexplored design boundaries. The mid-fidelity preliminary design of such a ‘dry’ wing for a medium-range aircraft, operating at Mach 0.78, is undertaken in this work. A CFRP (carbon-fiber reinforced plastic) wing equipped with an HLFC (hybrid laminar flow control) suction system is assumed for enhanced weight and viscous drag reductions, respectively. This requires a holistic aerostructural design of a composite laminar dry wing. The aerodynamics is modeled with a conical flow 2.75D approach using Euler-boundary layer equations, while the structural sizing is performed using linear finite element formulation on a structural shell model. Considering the aerostructural design variables for such a wing, namely the planform geometry, the airfoil profiles, the lift distribution and the suction distributions led to a 116 dimensional multidisciplinary optimization problem, which had to be resolved in the absence of any gradient information. The application of Bayesian optimization in synergy with a modified IDF (individual discipline feasible) architecture for successfully overcoming a huge challenge in terms of performing a high-dimensional gradient-free wing design is the core focus of this study. The resulting laminar dry wing exhibits a fuel mass saving of 8.8% in comparison to an optimized turbulent dry wing with robust aerodynamic performance within a suitable design range obtained via multi-point optimization. These resulting dry wings are also then compared with conventional fuel storing ‘wet’ wing designs obtained via the same mid-fidelity framework for analyzing their aerostructural characteristics.