Prediction of Concrete Peak Load and Compressive Failure Strength Using Machine Learning

Abstract

Peak load and compressive failure strength are influent parameters regarding the mechanical properties of concretes. Experiments such as compression tests are usually performed to extract relevant values. It is well known that experimental measurements are relatively costly and energy-consuming. Therefore, it is useful to identify and apply a model prediction from available data. In this work, the influence of the initial size of cylindrical normal-weight concrete considering three different mixtures is presented. Peak loads and associated compressive failure strength of multiple sizes concretes are predicted using machine learning. Decision tree (DT) and random forest (RF) regressors are presented in this work. A comparison between the models is made. The results of the models are found to be consistent with the experimental ones on peak loads (a coefficient of determination of 0.98 is obtained with the DT algorithm and 0.99 with the RF one) and should be improved with respect to the compressive failure strength (a coefficient of determination of 0.77 is obtained).