CFD Optimization of Small Gas Ejectors Used in Navy Diving Systems

Optimization of small gas ejectors is typically completed by selecting a single set of operating conditions and optimizing the geometry for the specified conditions. The U.S. Navy is interested in utilizing a small gas ejector design in multiple diving systems with varying operational conditions. This thesis is directed at developing a Quasi Newton-Raphson Multivariate Optimization method using Computational Fluid Dynamics (CFD) to evaluate finite difference approximations. These approximations are then used as inputs to the gradient vector and the Hessian matrix of the standard Newton-Raphson multivariate optimization method. This optimization method was shown to be timely enough for use in the design phase of a multiple parameter system.

CFD investigation of the level curves of the simulation cost function hypersurface verified the success of the method presented at optimizing each independent parameter. Additional CFD simulations were used to investigate the ejector performance for operational conditions deviating from the operational conditions used during optimization. A correlation was developed for selecting the optimum throat diameter, and corresponding maximum efficiency, as functions of the input conditions only. Experimental models were manufactured using fused deposition modeling and evaluated with good agreement to the CFD simulation results.