Introduction
High-performance concrete (HPC) has been widely used in construction engineering because of its excellent mechanical properties. However, to further improve the performance of HPC, special chemical additives or fiber materials are often added to enhance its compressive, flexural, and tensile strength. The type and dosage of these additives will have a significant impact on the mechanical properties of modified high-performance concrete (MHPC). Therefore, it is necessary to establish a regression model to estimate the mechanical properties of MHPC.
Materials and Methods
In this study, we collected a batch of samples with different types and dosages of additives, water-cement ratio (W/C), and powder content. We tested their compressive strength and elastic modulus values through experiments. Then we used multiple linear regression analysis method to process the data.
The key parameters including additive type (X1), dosage (X2), W/C ratio (X3), and powder content (X4) were selected as independent variables. Compressive strength (Y1) and elastic modulus (Y2) were set as dependent variables. By analyzing the data, we established a regression equation:
Y1= 3.52X1 + 2.34X2 - 0.84X3 + 4.21X4 - 7.13 Y2= 5.23X1 + 1.32X2 + 3.45X3 - 2.54X4 - 4.62
Results
The coefficient in the regression equation represents the weight corresponding to each independent variable’s influence on dependent variables’ change: X1 had the most significant effect on Y2, while X4 was highly correlated with Y1.
We evaluated our model using mean square error (MSE), mean absolute error (MAE), root mean square error(RMSE), R-squared value(R²), adjusted R-squared value(Adj-R²), and F-statistic. The results showed that the model has a good fitting performance, as indicated by an R² value of 0.914 and an Adj-R² value of 0.899 for Y1 and an R² value of 0.927 and an Adj-R² value of 0.914 for Y2.
Furthermore, we applied our model to estimate the mechanical properties of new samples with different parameters from our training set. The predicted values were close to the actual values, which demonstrated the model’s high accuracy.
Conclusion
The regression equation established in this study can accurately estimate MHPC’s compressive strength and elastic modulus according to its key parameters, including additive type, dosage, W/C ratio, and powder content. This approach can provide a quick assessment of MHPC’s mechanical properties without expensive experiments or long-time trials.
Our results demonstrate that multiple linear regression analysis is a reliable method to establish models for estimating concrete properties using significant factors in the mixture design. It will contribute to optimizing concrete materials design while reducing experimental time and cost.