Automatic Optimization of Preswirl Nozzle Design

The objective of the research described here is to develop and demonstrate use of automatic design methods for preswirl nozzles. Performance of preswirled cooling air delivery systems depends critically on the design of these nozzles which is subject to manufacturing and stress constraints. The best solution may be a compromise between cost and performance. Here it is shown that automatic optimization using computational fluid dynamics (CFD) to evaluate nozzle performance can be useful in design. A parametric geometric model of a nozzle with appropriate constraints is first defined and the CFD meshing and solution are then automated. The mesh generation is found to be the most delicate task in the whole process. Direct hill climbing (DHC) and response surface model (RSM) optimization methods have been evaluated. For the test case considered, significant nozzle performance improvements were obtained using both methods, but the RSM model was preferred.