It is a challenging task to take into account the type and form of aggregate in the numerical modelling of concrete. Discrete Element Method (DEM) offers the possibility to consider these properties via its model parameters. However, to get a usable model for further purposes, these model parameters first have to be calibrated precisely, based on laboratory measurement data. The model calibration is a time consuming process that requires high computational resources. In this study, an estimation model is proposed to define the model parameters of a DE model in function of the properties (density, compressive strength, particle size distribution, etc.) of the real material. The proposed model contained a number of factors and constants which were optimized based on our measurements. The measurements were carried out on samples produced from seven different concrete mixes. The mixes differed from each other in the type of the aggregate (quartz gravel, crushed stone, expanded clay) or in their particle size distribution (A or C grading curves, no-fines). The estimation model, with the optimized parameters, was applied on a test data set. The results indicated that the estimation model was able to estimate the DE model parameter properly. By estimating the DE model parameters more precisely at the start of the calibration process, it is possible to save significant amount of time and make the modelling more productive. The accuracy of the estimation model can be further increased by considering the aggregate-to-cement (a/c) ratio of concrete which is difficult to be determined in case of a hardened concrete sample whose concrete mix composition is unknown. An image processing technique is applied in this study to determine the a/c ratio. The results showed that the thresholding method is an appropriate alternative to determine the a/c ratio of various concrete mixes.
Read full abstract