The porosity value of swelling granular materials depends on the degree of swelling and the state-of-stress inside the packing. This research focusses on the swelling of a bed of Super Absorbent Polymer (SAP) particles, which are absorbent particles that can absorb 30 to 40 times their initial weight of saline fluids. Due to the complexity of swelling materials, measurements of porosity, and also hydraulic parameters, are complex and often not feasible, unless a quasi-static approach is employed. An alternative is not only to conduct measurements, but in addition to generate artificial packings using a grain-scale model from which porosity values can be derived. The combination of experiments and simulations reduce the experimental efforts whilst allowing for an understanding of particle-scale phenomena. A suitable grain-scale model is the Discrete Element Method (DEM). In DEM, particles are often represented as spheres, while SAP particles have very irregular shapes, which strongly affect their hydraulic properties. Therefore, the shapes of particles should be included in DEM simulations, which is achieved by using sets of overlapping spheres (often referred to as clumps). In this research, an algorithm was employed to make realistic clumps that are representative of measured SAP particles, using micro-CT scans of individual SAP particles. In DEM, particles were randomly generated based on 20 clumps and the particle size distribution of real SAP particles. The particles were then compacted to obtain a packing, from which the porosity values were obtained. These porosity values were close to experimental values and thus we concluded that our simulations of particle shape in DEM does improve the quality of simulations in case of granular materials. Finally, the relationship between porosity, state-of-stress and the degree of swelling is developed based on our grain-scale model. (C) 2019 The Author(s). Published by Elsevier B.V.