Concrete mix design methods are used to determine proportions of concrete ingredients needed for certain workability and strength. Each mix design method operates under certain assumptions and suggests slightly different proportions. It is of great importance that site/construction engineers know the method by which the mix was designed. However, it can be difficult to know the designing method based solely on mix proportions. Hence, in this work, a decision trees model was used to classify high strength concrete mix design methods based on their produced concrete mix proportions. It was found that the trained decision tree model is capable of classifying mix design methods with high accuracy. Further, based on dimensionality reduction methods, the amount of cement in a concrete mix was found to be the paramount predictor of the used mix design method. In this work, a novel high-accuracy model for determining a mix design method based only on mix proportion is proposed.
Keywords: compressive strength; high strength concrete; machine learning; mix design.