ABSTRACT
Ultra-high performance fibre reinforced concrete (UHPFRC) is an emerging material with a high scientific and technological interest due to its outstanding mechanical behaviour. However there are aspects of its mechanical characterisation that need to be solved. In particular, this work focuses on the determination of the tensile strength, interpreted as the stress at which the crack initiates in pure tension and softening begins. In a recent study, applicability of the Brazilian test or diagonal compression splitting test to determine the tensile strength of UHPFRC was assessed from the results of numerical simulations and experiments conducted on specimens with three contents of fibres and specimens without fibres. From the numerical simulations it was demonstrated that a local maximum in stress may occur in the test for a stress very close to the tensile strength of the material, which can be detected as a pop-in in the curves of results of experiments run under load control with the adequate instrumentation, thus providing an adequate approximation of the tensile strength. In this paper, a numerical study of the Brazilian test is presented which continues the previous work and investigates the size effect and the influence of the material properties. Simulations of the tests have been carried out within the finite element framework COFE (Continuum Oriented Finite Element), which implements elements with an embedded adaptable cohesive discrete crack. To reproduce the cracking behaviour, the softening law is assumed to present a steep initial softening, predominantly due to cracking of the matrix, followed by a long tail, due to the contribution of the fibres. In this study, influence of the content of fibres is analysed by modifying the parameters of the softening curve, while the size effect is analysed by modifying the ratio between a fracture length of the matrix, which is defined in detail in the paper, and the diameter of the specimen. The results show a high effect of the size and material properties on the test results, which may guide through the proper design of the experiments in order to ensure appearance of a measurable sufficiently-accurate local maximum.
