Multi-Fidelity Aerodynamic Shape Optimisation of an Existing Fixed-Wing UAV

Unmanned aerial systems are increasingly deployed in low-infrastructure environments, where long-range beyond-visual-line-of-sight missions must be performed with variable payloads and across multiple operational roles. For aircraft already in service, aerodynamic improvement must often be achieved without changing the underlying architecture. In such contexts, optimisation is not an unconstrained configuration design problem, but a constrained performance-improvement problem in which geometric freedom is limited by structural, integration, and operational requirements.This paper examines aerodynamic performance improvement within existing UAV architectures operating under practical geometric and operational constraints. A structured multi-fidelity optimisation approach is applied in which low-order methods support broad wing-shape exploration while high-fidelity adjoint-based CFD is used for local fuselage refinement. The approach is motivated by the limitations of existing methods, in which low-order models do not capture local aerodynamic interactions required for meaningful refinement, while high-fidelity methods are computationally prohibitive when applied across broad design spaces. The proposed framework separates global exploration and local refinement according to the structure of the problem, combining low-order wing-shape exploration with targeted high-fidelity fuselage optimisation.The methodology is applied to the HANSARD UAV developed by UAVAid-Ltd. The optimised wing and fuselage are integrated into a final aircraft configuration and assessed using RANS simulations. At identical flight speed, the final configuration increased the lift-to-drag ratio from 13.503 to 16.09, corresponding to a 19.18% improvement relative to the baseline, primarily through increased lift. Under an equal-lift comparison, the optimised configuration maintained the baseline lift at a lower freestream velocity with a 13.81% reduction in total drag.