Virtual materials for the prediction of concrete mechanical properties

Abstract

Physical properties such as compressive strength and elastic moduli are of the utmost importance for the structural stability and design of cement-based materials. These properties are strictly related to the microstructure of the binder paste, which in turn varies in time, as a function of the hydration kinetics. Therefore, the development of the elastic properties and mechanical strength can in principle be controlled by affecting the microstructure and hydration kinetics. This can be achieved through an appropriate mix-design, which encompasses a careful selection of phase proportions, grain-size distribution, amount of water and aggregates, and use of additives. Changing such variables by a trial-and-error process can be extremely time consuming and has a significant impact in terms of resources employed. Moreover, a fully quantitative approach to the study of the cement microstructure and hydration kinetics requires significant efforts in terms of experimental testing, often encompassing analytical techniques such as X-ray diffraction, scanning electron microscopy and isothermal calorimetry, among others. In this contribution, an alternative quantitative characterization of the cement paste in time is illustrated, based on the numerical modeling of cement-based systems. Virtual cement pastes and mortars are generated using the software VCCTL (http://www.nist.gov/el/building_materials/inorganic/vcctl.cfm), using as input parameters the clinker phase composition, the water/cement ratio, and the size and shape distribution of the particles. The elastic moduli and compressive strength of such virtual samples is then computed from the developed microstructure by a finite element method. Extensive calibration and testing has been performed against experimental data, and the good agreement between the calculated and measured elastic and mechanical properties shows that VCCTL can be used as a truly predictive tool. Although experimental testing remains a fundamental aspect of concrete science, the coupling of experiments with computational methods provides a viable tool towards a knowledge-based mix design, with a potential reduction of costs and environmental impact.