A Multivariate Regression Model for Predicting the 28days Flexural Strength of Lime-Cement Concrete Beam

Abstract

In this study, a multivariate regression model for predicting the 28days flexural strength of lime-cement concrete prototype beam was developed. The response function is a multivariate function of the proportions of the component materials of concrete. A total of twenty mix ratios, consisting of water, Portland cement, hydrated lime, river sand and granite chipping were used in the prediction process. The first ten mix ratios were used for model development while the remaining ten mix ratios were used as check points for model validation. The model developed was tested for adequacy at 95% level of confidence using the t-statistic. Calculated t-value was -1.3342 and this was less than the critical t-value of 2.2622. Thus, the model was found to be adequate. An average percentage difference of 14.303% was observed between the model prediction and the experimental values. A visual basic program using the Visual studio 2015 software was developed based on the regression model. It was invoked to quicken the process of selecting the mix ratios of the component materials corresponding to any desired flexural strength value that falls within the region of experimentation and vice versa Keywords— Concrete , flexural strength, multivariate regression model, response function.