Classical reservoir simulation methods are based on first-order Finite Difference (FD) schemes applied to regular grids. Although widely used, these methods cannot resolve the sharp material interfaces and oblique faults in realistic models of reservoir geometries. When two-point flux approximations are used, finite difference solutions also exhibit strong grid orientation effects and fail to capture fingering instabilities. Using novel methods to overcome these drawbacks is inevitable. One new method that holds much potential is the Finite Element-Centered Finite Volume Method (FECFV). In this study, water-flooding simulation results obtained with a commercial FD simulator for a complexly faulted reservoir analog are compared with those from a FECFV prototype implemented on the basis of the Complex System Modeling Platform (CSMP++). The pseudo 2D cross-sectional model is based on a geological transect along the Chienberg tunnel in Switzerland. The same grid resolution was used for both simulators. The geological model was built in the SKUA- Paradigm and was output to the commercial FD simulator. Rhinoceros was used to build the model and ICEM CFD to mesh it for the CSMP++ simulation. The results reveal the influence of the different discretization schemes on the time to water breakthrough, cumulative oil production, flow velocity distributions, analytic versus discrete well model, and layer- and fault aspect ratios. The performance of the simulators in the presence of stark fault- country rock permeability contrasts have also been investigated. Our study highlights the importance of an accurate treatment of material interfaces.