Analyzing the Impact of Design Factors on Solar Module Thermomechanical Durability Using Interpretable Machine Learning Techniques

Solar modules in utility-scale PV systems are expected to maintain decades of lifetime to effectively rival the cost of conventional energy sources. However, the long-term performance of these modules is often degraded by cyclic thermomechanical loading, emphasizing the need for a proper module design to counteract the detrimental effects of thermal expansion mismatch between module materials. Given the complex composition of solar modules, isolating the impact of individual components on overall durability remains a challenging task. In this work, we analyze a comprehensive data set capturing bill-of-materials and post-thermal-cycling power loss from over 250 distinct module designs. Using the data set, we develop a machine learning model to correlate the design factors with the degradation and apply the Shapley additive explanation to provide interpretative insights into the model's decision-making. Our analysis reveals that the type of silicon solar cell, whether monocrystalline or polycrystalline, predominantly influences the degradation, and monocrystalline cells present better durability. This finding is further substantiated by statistical testing on our raw data set. We also demonstrate that the thickness of the encapsulant, particularly the front side one, remains another important factor. While thicker encapsulants lead to reduced power loss, further increasing their thickness does not yield additional benefits. The study moreover provides a blueprint for utilizing explainable machine learning techniques in a complex material system and can potentially steer future research on optimizing solar module design.