Efficient models to evaluate the effect of C3S, C2S, C3A, and C4AF contents on the long-term compressive strength of cement paste

The primary chemical compositions of the cement are CaO, SiO2, Al2O3, and Fe2O3, while minor compositions are MgO, K2O, Na2O, and SO3. In this study, 142 data were collected from different research studies. The dataset was analyzed and modeled to forecast the compressive strength of cement paste (cubic sizes). The samples were cured at room temperature (25 °C). The essential input model parameters were tricalcium silicate (C3S%), which ranged from 31.5 to 73.74; dicalcium silicate (C2S%), which varied between 2.73 and 40.78; Tricalcium Aluminate (C3A%) ranged from 0.85 to 12.45, Calcium Aluminoferrite (C4AF%) varied from 4.1 to 14.9, Calcium Oxide (CaO%) varied between 60.12 and 65.74, Silicon Dioxide (SiO2%) ranged from 19.4 to 24.78, Aluminum Oxide (Al2O3%) varied between 1.72 and 6.18, Iron (III) Oxide (Fe2O3%) ranged between 1.35 and 4.9, water to cement ratio varied between 0.25 and 0.7, and curing time ranged from 1 to 90 days. The output parameter was compressive strength with 18 to 57 MPa. A linear, non-linear, multi-linear, full quadratic, and interaction regression model was used to predict the compressive strength of cement paste for the data collected from the literature. Based on statistical tool assessments such as objective function (OBJ), mean absolute error (MAE), coefficient of determination R2 functions, and scatter index, the full quadratic model with the lowest root means square error (RMSE) performed better than the other models in predicting the compressive strength of cement paste. In addition, CaO proved to be the most productive parameter of the compressive strength of cement paste. At the same time, Al2O3 was found to be the least influential variable in terms of the compressive strength of cement paste. The effect of various chemical components such as CaO, SiO2, Al2O3, Fe3O2, C3S, C2S, C3A, C4AF, water-to-cement ratio, and curing time on the long-term compressive strength of cement paste. Also, no study can be found in the literature comparing the efficiency of four models with very good accuracy in predicting the compressive strength of the cement paste. From now towards, researchers and the construction industry can use the developed models in this study with high accuracy (low RMSE and high R) to predict the compressive strength of the cement paste without any cost. They would save lots of time for the experimental lab work.