Applying genetic algorithms to complex computational fluid dynamics simulations

Aerospace applications often require complex flow simulations that cannot be computed quickly. To determine the optimum configuration or design in these applications can require the tuning of numerous parameters, the effectiveness of which can only be tested through repeated computational fluid dynamic simulations. Depending on the degree of the flow complexity, systematic searches of the multi-dimensional parameter space is often too costly in time, materials, and computational power. A viable alternative is evolutionary search algorithms which can look for regions of high performance with many fewer configuration evaluations. However, even in these cases the computational costs can remain quite high for more complex flows.This paper applies a genetic algortihm approach to two test problems with relatively complex flows: steady multijet flow control and atomic Oxygen sensor placement on an sounding rocket. The long-term objective is to build on this research to generate a more efficient approach to design optimization in these types of applications.