Bonded Particle Method (BPM) has recently been extensively used to simulate brittle materials. In BPM, brittle material gets represented as a dense assemblage of particles (grains) connected together by contacts (cement). Only particle motion law and contact constitutive model are therefore needed to perform the simulation. BPM does not require complicated plasticity constitutive laws, but it seriously depends on the contact micro-properties. Therefore a calibration process is needed to establish a unique set of these micro-parameters. In this research, discrete element code of UDEC is employed to apply BPM. This code can create random-shaped and non-uniform-sized particles interacting in frictional and cohesive contacts. Using this approach, contact micro-parameters are calibrated to fit the conventional test results (uniaxial and triaxial compression, and Brazilian tension).
In recent years, Discrete Element Method (DEM) is extensively used to model brittle materials [1,2]. Jing and Stephansson [3] have recently provided fundamentals of DEM and its application in rock mechanics. One use of DEM is to represent rock-type material as a dense packing of non-uniform-sized particles that are bonded together at their contact points. (Fig.1). This method is already known as Bonded Particle Method (BPM) [4,5]. The main advantage of this method is that a crack can be modelled as a real discontinuity between particles not just as a modification in material properties [3,4]. The major requisite of BPM is to calibrate the micro-contacts properties to fit the material macro-response [4,5]. So far most BPM models have been performed by Particle Flow Code (PFC) [8] in which the grains are simulated as rounded discs. Potyondy and Cundall [4] showed that the tensile strength obtained by PFC is approximately 0.25 of the uniaxial compressive strength. This is unrealistically high, where the ratio of tensile to compressive strength is typically reported around 0.05 to 0.10 [7]. This inconsistency is due to the fact that PFC simulates the particle as a circular disc. Therefore grain interlocking is basically neglected. Without interlocking, other contact parameters must be chosen disproportionately stronger to achieve the right material compressive strength, which itself leads to an exceedingly high tensile strength. Thus, they argued that cluster logic is required to have closer results to real behaviour. In a recent research, Cho et al. [5] have proposed a clumped logic to resolve this shortcoming. However in addition to its difficulty, this method is merely applicable to strong brittle rocks. They did not mention if weak rocks can be simulated in this way. In this paper, we are going to investigate the brittle materials by BPM and using the Universal Distinct Element Code (UDEC) [9]. We study the effects of UDEC micro-properties on the material macro-response. A series of experiments (Brazilian tension, uniaxial and triaxial compression) are used to calibrate UDEC 1078 micro-properties. It will be showed that UDEC does not show the defects and difficulties of the other methods.
