Statistical Study of the Physical-Mechanical Properties of Ordinary Concrete Manufactured in Congo-Brazzaville

Abstract

This research work is an extension of the study carried out on the physico-mechanical characterisation of aggregates used in the manufacture of standard concretes. The aim is to test the relevance and impact of the properties resulting from this characterisation in relation to the quality of the concretes formulated, using statistical normality tests. This research work reports on the statistical analyses of the physico-mechanical properties of concretes formulated using the Dreux-Gorisse method. The study is based on the development of eight (8) formulations of concrete obtained from different materials taken from the localities of Brazzaville and Pointe-Noire. To do this, in order to develop an optimal concrete formulation approach taking into account its microstructural and compactness matrix, a good granular distribution was considered using two types of sand (rolled and crushed) for the correction of rolled type sands with the variable proportions (from 30% to 50%) of crushed sands. Eight (8) formulations of the concretes were studied according to the Dreux-Gorisse method, and only six (6) formulations gave the expected results. This statistical study is part of the context of understanding the crucial parameters, the distribution of values and the variabilities of the measured properties, in particular the average, which provides an overview of the central value for each property and allows the data to be summarized into a representative indicator of the average performance of the concrete. As a result, the results suggest that the resistances of the two groups of results (7 days and 28 days of age) are significantly different, so the null hypothesis is rejected. In other words, the differences observed between the means of resistance at 7 days of age (20.40 MPa) and at 28 days of age (27.80 MPa) could not be due to chance, which corresponds well to a C25/30 concrete. Thus, two tests were performed: the Shapiro-Wilk test and the Wilcoxon test. The first, just as specific, made it possible to know if the data collected followed the normal distribution, while the second, which is a non-parametric test, made it possible to test the data when they do not follow the normal distribution. For this, the use of Q-Q plots was essential and made it possible to visualize normality. On the six properties of concrete, namely strength, theoretical density, real density, gravel/sand ratio, subsidence and water/cement ratio, the Shapiro-Wilk test made it possible to verify and highlight the normality of the distributions. Indeed, only four properties whose data showed a normal distribution, thus justifying the use of parametric tests. However, two of them (subsidence and gravel/sand ratio) whose data did not follow a normal distribution, which motivated the use of nonparametric tests more appropriately.