Influencing factors of mechanical properties and compressive strength prediction model of cement asphalt mortar

Cement asphalt mortar (CA mortar) is a composite cementitious material composed of cement and emulsified asphalt, and the mechanical properties of CA mortar serve as critical indicators for assessing its performance in service. However, due to the complex composition of CA mortar, the process of strength development over its entire lifespan remains poorly understood, posing challenges in accurately predicting compressive strength and optimizing the material proportion. This study addresses these challenges by testing the compressive and flexural strengths of CA mortar over 28 days and microstructural and pore distribution changes during the strength formation process. Subsequently, four machine learning algorithms were employed to predict the compressive strength based on compressive strength dataset which was built based on literatures and this study, with prediction results interpreted using Shapley additive explanations (SHAP). Last, the mix proportion is optimized based on carbon emissions and production costs as key objectives. The main conclusions are as follows: A significant number of macropores appear in CA mortar at 3 days, as hydration products gradually fill these macropores, their size decreases notably, while small and medium pores remain relatively unchanged throughout the process. The BP model demonstrated the best performance in terms of R², MAE, MSE, and RMSE. SHAP revealed that the emulsified asphalt to cement ratio (A/C) is the most significant factor influencing compressive strength, with the water to cement ratio (W/C) following as the second most influential factor. Multi-objective optimization (MOO) resulted in a reduction of carbon emissions by 18.3% and production costs by 31.3%. © 2026 Elsevier Ltd.