Predicting Cracking in CW617N Brass Fitting Inserts Using Logistic Regression and Expiremental Characterization

This study presents an integrative experimental and statistical approach to investigate and predict cracking in CW617N brass inserts used in PPR fittings. Four production lots, including samples from customer complaints, were characterized using scanning electron microscopy, energy-dispersive X-ray spectroscopy, optical microscopy, hardness testing, and X-ray fluorescence spectrometry to identify the metallurgical and mechanical factors driving intergranular fracture. Based on these analyses, a binary logistic regression model was developed using lead content, grain size, and hardness as explanatory variables. The model demonstrates statistically significant contributions of each factor and achieves high predictive performance, with an accuracy of 0.88, precision of 0.90, recall of 0.85, F1-score of 0.87, and a ROC-AUC of 0.90, highlighting its robustness. This framework effectively captures the complex interplay between lead segregation, grain coarsening, and hardness variations, enabling reliable quantitative prediction of crack formation. The combined methodology provides actionable insights for industrial quality control and process optimization while advancing the fundamental understanding of failure mechanisms in brass fittings.