UC San Diego

Flapping wing air vehicles (FWAVs) achieve flight through the coupled interplay of wing kinematics, aerodynamic loading, and structural deformation. Their design is an inherently multidisciplinary problem. Gradient-based multidisciplinary design optimization (MDO) offers a principled approach to address such challenges, combining derivative information with formal coupling architectures to enable simultaneous optimization across disciplines.This work presents, to the author’s knowledge, the first coupled gradient-based aeroelastic MDO framework applied to a flapping wing vehicle. The aerodynamic discipline uses the Unsteady Vortex Lattice Method to capture the three-dimensional unsteady effects. The structural discipline employs Euler–Bernoulli beam finite element analysis with Newmark-beta time integration. Models for both disciplines are implemented in the Computational System Design Language (CSDL), which automatically generates adjoint-based sensitivities through the coupled time-marching simulation. The Individual Discipline Feasible (IDF) architecture manages two-way coupling by treating the structural displacement as an additional design variable, allowing each solver to be called once without modification. An Augmented Lagrangian outer loop enforces coupling consistency outside of the optimizer, preventing the constraint count from growing with the number of coupling variables. The optimization controls eleven kinematic parameters and six spar radii with the objective of minimizing cycle-averaged actuator power subject to lift, thrust, and stress. The optimization results reveal several non-intuitive outcomes that illustrate the capacity of gradient-based MDO to explore the coupled design space. The structural optimizer produces a non-monotonic spar profile with a large tip segment that departs from the classical cantilever beam while leaving the stress constraint far below the material failure limit. A comparison with a low-density polyethylene spar reduces total power by simultaneously reducing kinematic amplitude and exploiting passive wing deformation. These outcomes demonstrate that gradient-based aeroelastic MDO can uncover coupled designs that would be difficult to identify with sequential or intuition-driven approaches.