In supercritical geothermal systems in deep volcanic areas, where pressures exceed 22 MPa and temperatures exceed 374°C, supercritical carbon dioxide (CO2) has a higher density than water. In the context of CO2 injection and storage, this poses the opportunity of having a sinking plume of CO2 while utilizing in-situ geothermal fluids as a renewable energy source. Following an overview of ongoing efforts in the prospection and development of supercritical geothermal systems, this thesis aims to provide extensive finite element simulation of CO2 transport in both homogeneous and fractured supercritical reservoirs. For creating models in a 3D domain, the multi-physics simulation framework MOOSE is used. The fracture network itself is realized as a discrete network based on stochastic distributions of parameters, including fracture dimensions, strike and dip angles, and fracture density. A parallel simulation approach is applied to simplify the complicated meshing process of a single combined model. Instead of embedding the discrete fracture 2D mesh within the 3D matrix mesh, the use of two separate models communicating with each other is demonstrated and applied. Furthermore, the numerical implications of this approach are discussed. In addition to analyzing the CO2 plume phenomenologically, the dependency on varying permeability and fracture network parameters, specifically fracture dimensions and fracture aperture, are investigated. As critical factors for geothermal energy production, CO2 breakthrough analysis was performed, and temperature distributions within the models were studied. From the simulations, it was concluded that the effect of CO2 sinking in reservoirs is more pronounced in high-permeability formations compared to low-permeability ones, where viscous forces dominate. Although CO2 still tends to sink in fracture networks, fluid flow is mainly controlled by the network path towards the producer, displacing all water uniformly. Cooling of the reservoir has been observed to be limited to a zone around the injector for the investigated scenarios.