The present work proposes a new algorithm for the optimization of cutting parameters in the high speed drilling of woven composites. The cutting parameters under consideration are the feed rate and the spindle speed. Three performance parameters are to be minimized. These are the exit delamination, the surface roughness and the thrust force. These performance parameters are observed experimentally. One of the challenges that face the experimental testing of these parameters is the high cost of the drilling tools and specimen materials. Therefore, the minimization of the number of experimental tests is a necessary requirement. The algorithm presented hybridizes Kriging as a meta-modeling technique with evolutionary multi-objective optimization to optimize the cutting parameters while intelligently selecting the new set of cutting parameters in each iteration. After starting with a factorial design of the search space, and after testing the performance criteria at these points, the algorithm fits a multi-dimensional surface using Kriging. This step is followed by an evolutionary search on the fitted model. The search spreads a population of search points in the direction of better performance criteria as well as in the direction of un-sampled space. The previous two steps are conducted iteratively for a pre-defined number of iterations. In the final iteration, the population of search points is clustered to yield a small number of new points at which the new experiments will be conducted. The whole process is iterated until the maximum number of allowable experiments is achieved. The algorithm is tested using an existing set of previously published experimental data that are dense enough to predict the actual response surface of the performance criteria. Results showed that the algorithm smartly moved into the direction of higher performance criteria with a low number of experimental trials.

This content is only available via PDF.
You do not currently have access to this content.