On the efficiency of coupled discrete-continuum modelling analyses of cemented materials

Abstract

Computational load of discrete element modelling (DEM) simulations is known to increase with the number of particles. To improve the computational efficiency hybrid methods using continuous elements in the far-field, have been developed to decrease the number of discrete particles required for the model. In the present work, the performance of using such coupling methods is investigated. In particular, the coupled wall method, known as the “wall-zone” method when coupling DEM and the continuum Finite Differences Method (FDM) using the Itasca commercial codes PFC and FLAC respectively, is here analysed. To determine the accuracy and the efficiency of such a coupling approach, 3-point bending tests of cemented materials are simulated numerically. To validate the coupling accuracy first the elastic response of the beam is considered. The advantage of employing such a coupling method is then investigated by loading the beam until failure. Finally, comparing the results between DEM, DEM-FDM coupled and FDM models, the advantages and disadvantages of each method are outlined.

Comparison of continuum (PFEM) and discrete (DEM) approaches for large insertion BVPs in soft rocks

“…Since then various authors improved on this method by introducing beam theory in the contact interaction and providing a physical rationale to the bond strength and stiffness [19], [20]. Following [21], [22] and who introduced the macro element framework to frame DEM contact models, [23] recently developed a bond-softening damage model able to capture the complex pressure dependent behavior of soft highly porous rocks. Such contact model, which will be used int this study, is defined within the macro-element framework characterized by a generalized force-displacement failure envelop defined as…”

“…( 3) are model parameters to control the softening rate of the bond. The effectiveness of the model to capture the ductile failure of soft rocks is shown in [23]. The DEM contact model elastic (the effective modulus 𝐸𝐸 � * , the normal-to-shear stiffness ratio 𝜅𝜅̅ * ) and strength (𝜎𝜎 0 and 𝜏𝜏 0 ) parameters are calibrated by matching experimental Young's Modulus (E) and Unconfined Compression Strength (UCS).…”

“…As shown in [28], the macroscopic modelled rock behavior depends on Particle Size Distribution (PSD) and coordination number. To reduce the computational burden, following [29] the PSD of destructured chalk has been upscaled by a factor of 2.1 near the pile (<0.6D) up to a factor of 4.8 at 5D. Considering that the damage model is framed to be scale independent the scaling used does not affect the calibrated parameters.…”

“…The effectiveness of the coupling can be evinced by the displacement contours shown in Figure 2, that represents a 2D section of the 3D model. Additional details on model the coupling are available in [23], [28]. Given the significant stiffness contrast between the pile and the rock, the pile representation is simplified by using a rigid, non-deformable wall and the installation force is obtained as the sum of the vertical components of the contact forces between the wall and the DEM particles.…”