Optimization algorithms utilize known information about the system to identify solutions that are more efficient and meet the requirements of the user. The algorithms require an objective function, or formula (linear or nonlinear) that models what the user is looking to optimize, in order to begin the search for a more feasible solution. Because optimization problems can involve either linearor non-linear functions, various algorithms have been created that can locate optimum solutions faster depending on the type of objective function being optimized.
This research focuses on optimizing an aircrafts thermal management system by using one such algorithm. This was performed in a three step process: initial research and testing, algorithm search method implementation, and post processing and analysis. The aircraft was modeled using complex Matlab Simulink block diagrams to simulate the thermal response of the system for any given type of mission. Using the provided parametric data, areas of user control within the model were located and optimization methods for these areas were devised. The function characterizing the fuel feed temperatures was chosen as the objective function to be minimized. Baseline data proved the function to be nonlinear. Optimization software incorporating a genetic algorithm (GA) was chosen since they are known to be best suited for nonlinear objective functions.
Optimization method implementation results showed a decrease in fuel temperature and convergence times. Data pulled from the GA detailed feasible fuel drainage sequences that would reduce fuel temperatures to 132F from the baseline temperature of 143F. Currently, methods using smaller drain sequences have been unable to match these results due to the coarse control over the fuel drainage these sequences provide. Because numerous computations are ran during each test, only feasible sequences shown to decrease the temperature were validated. Results show a need for physical hardware testing to verify the computational results shown.