The setup of a digital rock model from 3D CT scans and 2D thin sections to derive petrophysical and multiphase flow properties has gained influence over the years. Primarily due to the improved computational power and scanning abilities, the replication of rocks in a 3D computational environment is widespread available. Through Digital Rock Physics (DRP), it is possible to derive petrophysical and multiphase flow properties, which are typically obtained by traditional methods such as routine core analysis or special core analysis, from a digital model. The technique proposes an alternative for unconsolidated sandstones, where the rocks are friable and often unusable for conventional methods. Unconsolidated reservoirs have often only limited information available, therefore grain size distributions obtained by sieving analysis are used as basic input. The goal of this thesis is the development of a workflow to set up a model of an unconsolidated rock sample to evaluate multiphase flow processes. The main focus of this study is put in building models on basis of grain size distribution, assuming it to be the least amount of information available. As a result, the sensitivities of grain size binning, shape and orientation are investigated. The data used was provided by OMV Petrom. The objective of this work is to obtain the pore throat size distribution as close as possible to the real rock using reconstructed 3D models and simulating MICP experiments in a computational environment (GeoDict). The work here was conducted on rocks generated randomly on grain size statistical data and using the pore morphology method (Hilpert and Miller, 2001) to simulate the phase distribution inside the rock. By varying the original grain size distribution in different bins, models were created with both spheres and ellipses. Parameters like shape, grain dimensions, anisotropy and variations in bins were studied and ranked on how they are influencing the results. In the end, the parameter that dramatically influences the improvement of the model is the choice of different shapes for the grain generation. Moreover, the simulation proved that using a model based on grain size distribution manages to match the pore throat sizes. Consequently, with better optimization of the workflow on matching the whole capillary pressure curve, crucial information for flow characterization like relative permeabilities, and capillary pressures can be extracted.