Prediction of plastic viscosity and yield stress of self-compacting concrete using machine learning technics

Abstract

The use of self-compacting concrete (SCC) is increasing from day to day in many construction sites around the world. This type of concrete is made up of a combination of cementitious materials, sand, water, gravel, and chemical additives.Among the most important parameters for self-compacting concrete we find fluidity quality, generally predicted by plastic viscosity and yield stress values. For a complete characterization, these two parameters are measured by a concrete viscometer. But given its price and the difficulty of handling it on construction sites, there are empirical tests that estimate these parameters.In this sense, the objective of this work is to predict the rheological behavior (plastic viscosity and yield stress) of a SCC from the rheological parameters experimentally measured by empirical tests (slump flow diameter, V-Funnel flow time, and L-Box ratio).To do this, we used the most famous machine learning algorithms to know: Multiple Linear Regression (MLR), Random Forest (RF), Decision Tree (DT) and Support Sector Machine (SVM). Finally, the best proposed model is given on the basis of a statistical comparison between the different used algorithms.