A CFD-Based Small-Scale Combustor Design Optimization Utilizing a Many-Objective Evolutionary Algorithm

Abstract

Abstract This work demonstrates the capability of an open-source autonomous computational fluid dynamics (CFD) meta-modeling environment (OpenACME) for optimizing small-scale combustor designs. OpenACME couples several object-oriented programming open-source codes for CFD-assisted design using a decomposition-based many-objective evolutionary algorithm (MaOEA). The CFD is based on steady, incompressible, three-dimensional simulations with k-ω SST RANS and flamelet/progress variable combustion model. There are five unconstrained design variables based on combustor liner dilution hole diameters. The CFD results are compared with existing experimental data in terms of combustion efficiency as a function of severity. The comparison demonstrates that the CFD methods capture combustion efficiency trends. Next, more than 500 combustor designs are evaluated with OpenACME. A Pareto Frontier is generated in terms of combustion efficiency, pattern factor, and total pressure losses. Pseudo weights are used to select a non-dominated Pareto Frontier design point for future fabrication and experimental testing. OpenACME is demonstrated to be a viable tool for small-scale combustor design optimization.