Optimization of power losses in poly-V belt transmissions via genetic algorithm and dynamic programming

Abstract

Poly-V belt transmissions are often used in the front-end of engines to transmit power from the crankshaft to the several accessories. Optimizing the power losses coming from this transmission help reducing the engine fuel consumption. To this aim, an optimization method based on a genetic algorithm has been developed; principles of real life (natural selection, cross-over and mutation) applied to computations have been used to minimize power losses. Genetic algorithms are commonly used in optimization because of their adaptability to handle complex engineering problems. For comparison purposes, the dynamic programming which is a technique based on the principle of sub-problematization has also been implemented. In this study, the optimization-problem consists in minimizing the power loss function representing the losses taking place in a front engine accessory belt drive. Results have shown that the total power losses can be reduced by tens of percent by adjusting both the positions of the engine accessories and the belt setting tension. Some experimental data confirm these numerical predictions.