Optimization of Scramjet Combustor Geometries Using Genetic Algorithms

摘要

The primary focus of this effort involved the development of a methodology to integrate evolutionary optimizers with established computational fluid solvers to permit optimum design of representative scramjet combustors. Validation studies were designed to use available experimental data in order to establish a baseline for the computational solver from both the meshing as well as the solution standpoint. Methodology for partitioning the optimizer for parallel processing was introduced as a means to take advantage of current advances in computational hardware. Additionally, continuity- and momentum-based quantifications of combustor performance were presented that allowed realistic extension of three-dimensional combustor performance to a single-objective optimization environment. The resulting methodology was found to be robust in its evolutionary results around the baseline geometry during optimization runs. The optimizer was able to locate and identify theoretical and intuitive designs based on the unique physics of the ducted fluid flow inside the combustors. Improvements of mixing efficiency in excess of 42% above the experimental baseline with 0.68% losses in total pressure recovery were noted. Future work is headed toward more complex combustor geometries. It should also be possible to include other scramjet components such as the supersonic diffuser and nozzles within the optimization to allow complete engine performance modeling.