Challenges in simulation of geological CO${}_2$ sequestration and supercritical geothermal reservoirs

In a next step, the simulation code for two-phase flow was coupled to a module calculating streaming potentials. Saturation dependence of the coupling coefficient and of electrical conductivity are taken into account when implementing this feature into SHEMAT-Suite. The objective of this new module was to examine the possibility of detection of CO${}_2$ movement in the overburden of a storage reservoir due to CO${}_2$ leakage through an abandoned well via low-cost self-potential (SP) measurements at the surface. In a typical leakage scenario including a leaky well with conductive metal casing, SP signals originating from injection can be identified at the surface. While SP monitoring can be applied for detecting abandoned wells, the smaller leakage signals, however, are masked by the injection signals. Three methods are proposed to detect the small leakage signals: i) simulate a non-leaky scenario and substract the simulated signals from the measured ones, ii) exploit the symmetry of the injection signal by analyzing potential differences of dipoles with a dipole center at the injection well, iii) measure SP signals when injection is interrupted. A prerequisite for method i) is a well-known reservoir structure, as, e.g., in a depleted oil or gas field. Best, use a combination of method i) and ii). In order to analyze the quality of these methods, three-dimensional numerical (SP) modeling of two-phase flow and the electrokinetic coupling between flow and streaming potential was performed. The SP signal is influenced by CO${}_2$ injection, coupling coefficient, rock conductivity and the electrical conductivity of the well casing and is approximately ten times higher for conductive metal casings as compared to the presence of no metal casing. A major advantage of SP monitoring is the fact that SP signals at the leakage location arise shortly after injection started, even if CO${}_2$ is still far away from the leaky well. The brine being displaced through the leaky well causes an SP signal. Thus, the measurement of SP signals is the only method which is able to detect a leakage before any fluid from injection actually leaks out of the reservoir. Hence, SP monitoring is recommended for subsurface CO${}_2$ or fluid storage monitoring, but will benefit from a lower detection limit of the sensors. In a next step, the two-phase flow simulator has been enhanced, such that CO${}_2$ may dissolve into brine and H${}_2$O into gas making the (dis)appearance of phases possible. From the numerous different possibilities to deal with phase (dis)appearance, I chose the method of extended saturations. This method is relatively straight-forward to implement into an existing two-phase flow simulator and avoids primary variable switching (PVS) during phase changes. Switching of variables is associated with deteriorating convergence in Newton's method during phase change. The new method has been benchmarked against existing test cases. A primary feature of the two-phase, two-component module is the appearance of gravity fingering due to CO${}_2$ dissolving into brine. To this end, I examined the three relevant trapping mechanisms, namely, stratigraphic, residual and solubility trapping for different synthetic reservoirs featuring homogeneous and heterogeneous permeability fields. In a last part, I implemented a pressure-enthalpy formulation for supercritical water/steam geothermal reservoirs. Relying on pressure and enthalpy no PVS is needed. During phase appearance a meaningful enthalpy can be calculated directly without the need to instantiate saturation with a small value (as needed for PVS methods). The implementation has been verified against existing test cases. In addition, the local geothermal reservoir around the well Venelle-2 has been simulated in a three-dimensional study. Further, Monte Carlo simulations of Venelle-2 try to explain the newly found elevated temperatures within the well.