CFD-Based Shape Optimization of a Bladeless Drone Thruster Using Gaussian Process Surrogate Modeling With Experimental Validation

Traditional exposed rotor designs for aerial propulsion systems pose significant safety risks and noise pollution challenges. Bladeless thrusters offer a promising alternative by leveraging air induction and entrainment to generate thrust. However, shape optimization of complex geometries using high-fidelity Computational Fluid Dynamics (CFD) simulations is often computationally prohibitive. This study presents a unified shape optimization framework for bladeless drone thrusters that couples ANSYS CFD simulations with a Gaussian Process (GP) surrogate model and a sequential adaptive sampling strategy. To ensure physical accuracy, a momentum-theory-based thrust equation was derived and validated against experimental data obtained from a thrust stand and a dedicated test rig equipped with a 7-hole pitot tube for detailed flow measurements. The optimization framework was applied to the thruster's angle of attack (AoA) and hydraulic radius (r_h). The optimal configuration (r_h = 0.05 m, AoA = -1.48 deg) was successfully identified, yielding a maximum thrust of 11.22 N. By balancing exploration and exploitation, the adaptive sampling strategy achieved these results in 307 minutes, reducing computational time by approximately 55% compared to brute-force methods. These findings demonstrate the efficiency of surrogate-assisted optimization in designing safer, aerodynamically efficient bladeless propulsion systems.