The analysis for optimal design of an air-cooled internal combustion engine cooling fin array by using genetic algorithms (GA) is presented in this study. Genetic Algorithms are robust, stochastic search techniques which are also used for optimizing highly complex problems. In this study, the fin array is of the traditional circular fin type, which is subject to ambient convective heat transfer. The parameters (degrees of freedom) selected for the analysis include the cylinder wall thickness-to-radius ratio, fin thickness, fin length, the number of fins, and the local heat transfer coefficient. By using a single objective GA procedure, the heat transfer through the fin arrays is set as the objective function to be optimized with each parameter varied within the physical ranges. Proper population size is selected and the mutations, cross-over and selection are conducted in the GA procedure to arrive at the optimal set of parameters after a certain number of generations. The GA proves to be an effective optimization method in the thermal system component designs when the number of independent variables is large.

In this paper, we examined an imaginary underperforming prototype of a real high bypass-ratio turbofan gas turbine engine that has been assembled to specifications. The prototype was designed and assembled to generate a predetermined value of specific thrust while consuming fuel at a predefined specific fuel consumption value. To meet the required performance targets, improvements needed to be made to one or more of the engine components. In most real world scenarios, the improvement of any or all engine parameters pertaining to its performance is tied to a cost per percentages of improvements of individual component in the engine. There is always a narrow room for improvement in each or all of the components. However, the improvements come with high costs, since the engine has been designed in an efficient way to begin with. The cost of improvement of each component is indexed by a dollar cost per percent value of the component performance characteristics. It is of technical and economic importance to find a combination of performance improvements of each of the components that yields the lowest overall rework cost, thereby the total design cost, given the specified engine performance criteria. To achieve this goal, simulations for a real gas turbine turbofan cycle are performed in conjunction with the genetic algorithm (GA). A single objective (i.e. total improvement cost) GA with two constraints (i.e. desired values of specific thrust and specific fuel consumption) and eight degrees of freedom (i.e. diffuser pressure ratio, fan polytropic efficiency, compressor polytropic efficiency, burner efficiency, burner pressure drop ratio, turbine polytropic efficiency, fan duct exit nozzle pressure ratio, core duct exit nozzle pressure ratio) is employed. As a result, we have found the lowest overall cost associated with the improvements for each of the components.