The potential for using discrete particle modeling as a tool to compute rock mechanical parameters based on microscopic information is being explored. In order to achieve this, the numerical model needs to be calibrated vs. simple and controlled model experiments. The most essential input parameters for the numerical model are interparticle bond strengths and stiffnesses. We present here a series of rock mechanical (UCS) and core scratch tests on artificial rock-like samples of glass beads cemented with epoxy, and discrete particle simulations of the same tests. We conclude that both numerical and experimental data show the same features, such as correlation between UCS and peak force in the force distribution obtained from the scratch tests. The force required to break one intergranular bond is the same size of order (within a factor . 4) when evaluated from physical and numerical simulations.
Due to enhanced computer speed and power, it is becoming more and more attractive to model rock mechanical behavior with discrete particle models. Cores from oil field are scarce, and only a limited number of mechanical tests can be performed. Cores may also be altered during the coring process, so that the data obtained needs to be corrected for core damage. Furthermore, certain experiments such as creep tests may be very time-consuming, and not feasible within practical time limits. It is therefore important to have numerical models which can be used to simulate the mechanical behavior of the cored material. In addition, the use of numerical models may in itself provide insight into the physical processes of rock deformation and failure.
In order to use a discrete particle model as a "numerical laboratory" as outlined above, one needs to obtain the input parameters necessary to describe classes of rock materials based on petrographical information (such as microscope images) plus simple experiments that preferably can be done with small amounts of recovered core material. In order to establish the numerical laboratory, a calibration of the numerical model is required. The work presented here is part of such a calibration.
We use the Particle Flow Code (PFC, ) (provided by HCItasca) as the sample generator and as the test machine of our numerical laboratory. A rock like material is generated numerically as an assembly of particles. The mechanical behaviour is controlled largely by the intergranular bond strength and stiffness. In this study we aim at providing guidance for choosing these numerical parameters by performing experiments on a controlled granular material.
2. EXPERIMENTAL SETUP
The numerical model is based on spherical particles which can be cemented with a bond, defined by shear and tensile strength, plus shear and normal stiffness. In its basic version, the discrete particle model (PFC) uses spherical particles as its building blocks. In order to calibrate the numerical bond parameters through controlled experiments, we therefore chose to use an artificial material of spherical glass beads, with epoxy as cement between the particles.
2.1. Sample preparation
The samples were made of glass beads with diameters in the range 1000-1180 microns with physical properties listed in Table 1.