Optimization and analysis of drilling-induced damage in unidirectional carbon fiber reinforced plastics

Carbon fiber reinforced plastics (CFRP) materials, widely adopted as lightweight and robust metal alternatives, face challenges in achieving satisfactory machining outcomes, particularly in drilling, leading to frequent issues such as delamination. To address this issue, an approach was undertaken to predict and optimize machining parameters to produce high-quality holes. Drilling tests were conducted, and the hole quality was evaluated using the delamination factor. The experimental results showed that higher feed rates generally increase the delamination factor, though an anomaly was observed at lower feed rates. The spindle speed also influences delamination but to a lesser extent. To obtain the optimum drilling process parameters, a multi-layer perceptron (MLP) was initially used. Subsequently, a genetic algorithm (GA) was used to determine the ideal drilling parameter-level combination for maximum performance. Although many optimization studies have been conducted on this topic to obtain high-quality holes, only a few studies have conducted optimization by combining the neural network modeling and GA. The MLP-GA optimization showed an error of 0.3% in validation and successfully found optimum feed rate and spindle speed that minimize delamination. It was confirmed that the combined MLP-GA significantly improves prediction accuracy. This method can be used as a reference to determine process parameters to obtain high-quality holes in drilling CFRP.