Frisia Zout B.V. extracts salt from the subsurface by means of solution mining in the north-western part of The Netherlands. The caverns are situated in the Zechstein-II Halite at a depth of around 2.5 km. Because of this depth and the low operating pressure (around 60% of lithostatic pressure) the salt creep within the cavern is high. Once a cavern is at the end of its lifetime it is decommissioned and shut in. During the shut-in period the cavern blanket fluid is removed and replaced with brine and gets time to equalize in pressure and temperature with its surroundings before being permanently abandoned. At the time of abandonment there is still a small pressure deficit (cavern brine at 98% of lithostatic pressure at roof of cavern). At this pressure deficit there is, in theory, still some salt creep because of the difference in pressure between the cavern brine and surrounding salt walls induced by lithostatic pressure. Since the cavern is closed and still creeps, the brine will escape by means of permeation through the surrounding salt walls and roof. At this point an equilibrium is reached between the cavern convergence and the brine permeation around the cavern.
This research aims to get a better understanding of the cavern convergence and permeation processes after abandonment. For this, a cavern convergence- and brine permeation model is made. Next to this the potential surface subsidence due to the migration of brine to more permeable layers is investigated. In the convergence model, the cavern is modelled as a stack of cylinders and a Norton-Hoff power law squeeze model is applied to the cavern. The squeeze model consists of 2 parts, a linear and a nonlinear part. The nonlinear part is most significant during the production phase and in these high-pressure deficits the squeeze model is fitted on the available production data. Recent creep tests on salt samples under lower pressure deficits (Bérest et al., 2019) have confirmed that the linear part becomes the most significant in the low-pressure deficit region and have shown that the linear creep is smaller than the linear component of existing squeeze model used for production.
Next to this a sensitivity analysis was done on the convergence model by varying the input variables of the model. The parameters that have a large uncertainty and have a large impact on the model were the linear part of the squeeze model and the width of a slice. To give a range of outputs of the convergence model a P10, P50 and P90 scenario is created where these are percentiles from the input range of the sensitivity analysis. The outcome of the convergence model at a cavern size of 1Mm3 suggests a yearly cavern convergence of around of 5, 103 and 2313 m3/year for the P10, P50 and P90 cases respectively.
Since there is an equilibrium between the cavern convergence and the brine permeation, the output of the convergence model (convergence rate) can be used as an input for the permeation model (permeation rate). For the permeation model, different paraboloid shapes are fitted on each layer and are filled with brine from the converging cavern. Once all the salt layers are filled in, the brine reaches more permeable layers and can freely flow over a larger area. The permeation model is run with the P10, P50 and P90 convergence model scenarios as an input and predicts that the system fills after 26, 588 and 12,363 years respectively. At this point there could be some subsidence because the brine can freely flow over a larger area in the more permeable layers above the Zechstein. This subsidence is 0.016 mm/year for the P50 case after 588 years. A negligible amount compared to unrelated subsidence processes.
13
To conclude the cavern convergence rates (even the P10 at 5m3/year) are high compared to the permeability of salt according to the Darcy flow law (around 17 l/year). This could have multiple explanations. From the cavern perspective, the cavern convergence rates could be lower. This could be because of a threshold pressure for salt creep to occur (van Oosterhout et al., 2022) or because of some inaccuracies in the linear component of the squeeze model. Future research could focus on determining the creep rates of salt under low-pressure deficits. From the permeation perspective, other permeation paths next to permeability could be at play as well. In the cavern there could be permeation via anhydrite alterations or via micro fractures created during the production phase of the cavern. It would be good to look at these permeation processes in the future. Next to this the secondary porosity of the salt remains a question as well. A good understanding of this porosity is needed to assess the storage capacity of the overlying salt layers before the brine enters more permeable zones.

The goal of this study is to test out the accuracy of the upscaling approach (Jinyu Tang, 2021) for a geothermal doublet. This approach tries to simulate the physical thermal dispersion. When modeling in a geothermal reservoir it would be easy if a lot of layers can be upscaled. However, when doing upscaling one can not just take an average of the layer properties. This results in a very inaccurate representation of reality. That is where the upscaling comes in. First, some other points need to be considered. The reservoir is simulated in DARTS (Wang, Voskov, Khait, & Bruhn, 2020). This simulator uses numerical methods to model the reservoir. Before starting on modeling geothermal reservoirs first a 1D case is evaluated to check for numerical dispersion that comes into play. With that evaluation done the 2D cases were simulated. The full reservoir has 91 layers, these can be upscaled into 9 layers eventually. To start building up to that scenario first 2 other scenarios are evaluated. First, an upscaled section of the first block of upscaled layers is evaluated. This was originally 10 layers and now upscaled to one layer. This one layer was first simulated in a one grid simulation and then a single grid simulation. These results are compared to a simulation of the original layer properties and a simulation of an arithmetic average of the reservoir properties. This is also done on a second scenario with 2 upscaled groups and eventually the full 9 layer reservoir (originally 91 layers). With these scenarios evaluated the upscaling shows to give a better simulation of the reservoir compared to using an average for the layer properties. This is all compared with a simulation of the original layer properties.