ABSTRACT
This Chapter includes some observations that may be of use to those running particulate DEM simulations or considering running them in the future. While it is probably an exaggeration to state that there is an \art" to achieving successful simulations, the user certainly needs a bit of time to develop simulation skills prior to embarking on full-scale \production analyses" that will generate the nal results. As with any numerical method, the user should also have an understanding of the numerical algorithm and details such as the behaviour of the the contact model used. A complicating factor for DEM simulations, in comparison with continuum methods, is the need to generate the initial packing conguration for the simulations. As noted in Chapter 7, the specimen generation simulation stage where the DEM model is created can be more computationally expensive than the simulation of the boundary value problem of interest. Virtually all DEM analyses simulate a non-linear system; therefore it can be dicult for the analyst to know whether the simulation results are correct or not. (Suggestions on suitable validation problems are given in Section 11.7 below.) These challenges mean that it can be di- cult initially for users to develop appropriate techniques used to achieve accurate simulation results. Documented research studies that have used DEM vary in the extent to which the details of
\dembookwcrop" | 2011/2/22 i
the simulation approach are described. The guidance provided by Kuhn (2006) and discussion by Ng (2006) are both useful. As already discussed in Section 5.3, Thornton and Antony (2000) give a particularly clear description of how to use periodic boundaries to achieve a prescribed isotropic stress state and then to shear a sample in a strain-controlled periodic cell simulation. Potyondy and Cundall (2004) discuss the use of DEM to simulate the response of a bonded material and the detailed description of the simulation approach given include comments relevant to DEM analysts simulating the response of unbonded materials. The objective of this Chapter is to provide some preliminary guidance on simulating physical systems using DEM, drawing on personal experience and documented studies. The discussion will, hopefully, be of some use to rst time users of DEM in particular.
