Abstract

Designing a concrete mix is a process of synthesizing many components, it is not a simple process and requires extensive technical knowledge. The design process itself focuses on obtaining the required strength of concrete. Very often designing a concrete mix takes into account the need to maintain the proper water-demand and frost-resistance features. The parameters that influence the concrete class most significantly are the water-cement ratio along with the type and amount of cement and aggregate (crumb pile composition). However, in practice, the design of a concrete mix is done using additives, the so-called plasticizers, which are workability improving agents, as well as increasing frost resistance and aeration. Nowadays, the most widely used concrete mix design approaches are computational-experimental methods based on three-equations method, which allow estimating the amount of aggregate, cement and water. The results obtained in such a way requires laboratory verification. An entirely new approach is a prediction of concrete strength using extensive databases and machine learning algorithms, which can bring some potential benefits, the most important of which is the automation of the concrete mix design process. This paper aims to present preliminary work on the application of data mining in a concrete mix designing process. The primary parameter of concern is concrete compressive strength. The analysis is base on ready-made formulas of concrete mixes from personal resources, prepared for structural concretes of various classes. Furthermore, recently many researchers work on developing perfect concrete compressive strength prediction formula, which makes it an active field of research. © Federation Internationale du Beton (fib) - International Federation for Structural Concrete, 2019.

Authors (2)