Thermal distribution in cement-treated base: Effect of curing methods and temperature estimation using Artificial Neural Networks

Abstract

Due to the effect of curing temperature and heat absorption on the performance of cement-treated bases (CTB), thermal shrinkage cracks can happen in CTB during the early-age curing period. However, there have been non-existent studies on the relationship between temperature distribution in CTB versus various curing methods. In this paper, a field experiment was conducted to monitor temperature at different depths and to investigate fundamental mechanical properties of CTB. An Artificial Neural Network (ANN) model was also developed to estimate temperature distribution in CTB segments exposed to three different curing methods (wet burlap, wet geotextile, and asphalt emulsion). The results indicated that wet burlap and wet geotextile curing methods exhibited better quality cures than asphalt emulsion, expressed by higher 14-day compressive and splitting tensile strengths, due to less temperature absorbed into CTB structure. Shrinkage cracks were also observed on CTB surface cured with asphalt emulsion due to more heat absorption. The ANN-based model proposed using four input parameters: air temperature, measurement depths, curing methods, and times of day showed a high accuracy with R-value ≥0.96 and average MSE ≈ 0.0038. A sensitivity analysis using Connection Weight Approach has proven the importance of curing methods which contributed to the highest dominance of 30.1%, whereas air temperature, times of day, and measurement depths accounted for 26.4%, 26.6%, and 16.9%, respectively. Thus, the ANN-based model has accurately estimated temperature distribution in CTB while curing.