Numerically Optimizing an Annular Diffuser Using a Genetic Algorithm With Three Objectives

A genetic algorithm was implemented to determine preferential solutions of a short annular diffuser exhaust system of length 1.5Do (outer annulus diameters). Five free variables defined the centre body shape and two variables determined the outer wall profile. Diffuser performance was evaluated using three objectives—(i) diffuser pressure recovery, (ii) outlet velocity uniformity, and (iii) total pressure loss—that were calculated from steady state solutions obtained using the computational fluid dynamics software FLUENT 13.0 with the realizable k-ε turbulence model and enhanced wall treatment. Inlet conditions were ReDh = 8.5 × 104 and M = 0.23. After thirty-five generations, a paraboloid-shaped centre body with length 0.74Do and initial annular expansion of approximately 14° produced preferential solutions. A configuration with a converging outer wall above the centre body developed greater outlet flow uniformity and lower total pressure loss whereas a straight outer wall followed by the solid diffuser generated more static pressure recovery.