Assessing Key Factors Influencing Fire-Induced Spalling of Concrete Using Explainable Artificial Intelligence (XAI)

This thesis adopts eXplainable Artificial Intelligence (XAI) to identify the key factors influencing the fire-induced spalling of concrete and to extract new insights into the fire-induced spalling phenomenon. In this pursuit, an XAI model was developed, validated, and then augmented with two explainability measures, namely, Shapley Additive exPlanations (SHAP) and Local Interpretable Model-Agnostic Explanations (LIME). The proposed XAI model not only can predict the fire-induced spalling with high accuracy (i.e., >92 %) but can also articulate the reasoning behind its predictions (as in, the proposed model can specify the rationale for each prediction instance); thus, providing us with valuable insights into the factors, as well as relationships between these factors, leading to spalling. This model was created and validated using a comprehensive database, which reports on 43 influencing factors spanning material, mechanical, and geometrical properties, as well as environmental and casting conditions. Finally, the validated XAI model was utilized to contrast and quantify the most important factors (found in the spalling-based knowledge domain and literature to identify concrete mixtures with a low tendency to spall under elevated temperatures.