The application of geothermometry has been used for the last six decades for geothermal reservoir temperature estimation. A steady evolution of conventional geothermometers to multicomponent tools as well as application of artificial intelligence are nowadays available.
The development of high-performing computers offers the possibility to use deep learning algorithm for reservoir temperature estimation. Serving a selection of geochemical input parameters to artificial neural networks, they can be used to predict temperatures in the subsurface. Therefore, the chemical composition of the geothermal fluids are required. Main cations and anions as well as the SiO₂ concentration and the pH value serve as these input parameters. Using the data of well-studied geothermal systems, the neurons within the layers of the neural network are linked and weighted. Thus, the newly developed artificial intelligence is trained and validated. As a result, the modelled reservoir temperatures match with the in-situ temperature measurements of the analysed geothermal fields. Contrary to the usage of conventional geothermometers, the application of artificial neural networks are a useful novelty. While dealing with large amounts of data, artificial neural networks are faster, more easy-to-handle, as well as higher in accuracy.

In the discussion about the future role of geothermal in the energy transition policy, the topic of underground heat storage became recently more and more prominent. High Temperature Aquifer Storage (HTAS) may make geothermal more efficient by extending it beyond its traditional usage as a base load with also covering middle and even peak load. Depleted oil reservoirs can provide this underground storage capacity and Stricker et. al (2021) have numerically described the thermal storage potential in depleted oil fields from examples of the Upper Rhine Graben.

Hydrocarbon exploration and production in the Northern Alpine Foreland Basin accelerated after 1950. It reached its peak in the 1980s, and then decreased mainly due to the low oil prices. Numerous separated reservoir units were successfully developed and exploited. The related extensive exploration campaigns provide exhaustive seismic profiles and borehole data for delineation of geometric underground features and reservoir properties. Since the outgoing 1990, parts of this data were already applied for the successful hydro-geothermal exploration of the Upper Jurassic Malm, especially in the greater Area of Munich and the Eastern part of the Molasse.

The present study focusses on the geological and hydrogeological potential of high temperature storage in the surrounding of the existing oil fields in the South German Molasse basin. Reservoir information and data as e.g. thickness, porosity and depth of the reservoir rock as well as overlying barrier properties are compiled from two meta-studies, the Geothermal Atlas of Bavaria (STMWi, 2004) and Storage Catalogue of Germany (BGR, 2011).

As a result, about one third of the area of the Bavarian Molasse shows a potential underground storage with a reservoir thickness of 10 m and more in depths between about 500 and 1700 m. In the Western part, the potential storage units are the “Bausteinschichten” of the Lower Oligocene with a porosity ranging from 5 – 31 %, and the Middle Jurassic Dogger “Eisensandstein” with an average of 15%. In the Eastern part, Chattian sandstones of the Upper Oligocene with porosities of 20% are present. In a next step, oil field information with the borehole data and its exploitation history has to be investigated, to gather more details on local reservoir characteristics as e.g. temperature, pore pressure and to develop an exploration and exploitation strategy to better determine the uncertainties and risks.

The inaccessibility of geothermal reservoirs makes the accurate determination and monitoring of reservoir properties and conditions difficult and is a major problem in reservoir engineering. We present an approach for the development of messenger nanoparticle tracers for the simultaneous determination of flow paths ("tracer") and reservoir properties ("messenger"), with a proof-of-concept example of flow-through experiments and temperature detection under controlled laboratory conditions. For this, silica particles are synthesized with a two-layer architecture, an inner closed core and an outer porous shell, each doped with a different fluorescent dye to create a dual emission system. Temperature detection is achieved by a threshold temperature-triggered irreversible release of the outer dye, which changes the fluorescence signal of the particles. The flow-through experiments were conducted in a sand packed-bed column. The breakthrough curves of the nanoparticle tracers show minor tailing and a faster breakthrough compared to conservative, conventional molecular tracers such as Uranine and Eosine. The presented particle system thus provides a direct, reliable and fast way to determine reservoir temperature and flow paths in the reservoir. The system has a sharp threshold for accurate measurement and allows detection in concentration ranges as low as a few micrograms of nanoparticles per liter.

Between 1993 and 2005, the Soultz-sous-Forˆets reservoir was stimulated through 4 different wells crossing the reservoir at two different levels R3 (about 3km deep) and R5 (about 5km deep). The figure below represents the N-S section of the reservoir with the geometry of the 4 wells. During these stimulation episodes, seismic and hydraulic data were recorded. Using hydraulic data (pressure and flow rate) and available seismic catalogs of the stimulation episodes in the Soultz-sous-Forˆets reservoir, an analysis of the evolution of the injected energy and seismic energy was made. The analysis revealed two seismic behaviors of the reservoir. First, the seismic energy grows linearly with the energy injected from a certain level of energy injected with a similar slope for wells stimulated a first time. The parts of the reservoir which are stimulated a second time (GPK1 in 1993 and 1996 and GPK4 in 2004 and 2005) show a more rapid growth of the seismic energy which can be explained by the Kaiser effect (a reservoir stimulated a first time will have to reach at least the maximum pressure level reached during the first stimulation to generate seismic activity again). Secondly, the seismic response of the deepest part of the reservoir (R5) is greater than the shallowest one (R3). Indeed, the injection efficiency, which is calculated by the ratio between the cumulated seismic energy and the cumulated injected energy shows a convergence towards 10−5 for R3 and 10−2 for R5.

The swelling of clay-sulfate rocks is a well-known phenomenon often causing threats to the success of different projects, for instance, geothermal drillings triggered swelling and ground heave with dramatic damages in Staufen, Germany. The origin of clay-sulfate swelling is usually explained by physical swelling due to clay expansion combined with chemical swelling associated with the transformation of anhydrite (CaSO₄) into gypsum (CaSO₄.2H₂O). The swelling leads to about 60% of the volume increase of the rock mass. Numerical models simulating rock swelling must consider hydraulic, mechanical, and chemical processes. The simulation of the chemical processes is performed by solving thermodynamic equations usually contributing a significant portion of the overall simulation time. This contribution presents a Gaussian process regression (GPR) model as an alternative approach to determine the solubility of mineral phases, i.e., anhydrite and gypsum, in pore water. The GPR model is developed using the experimental data collected from the literature. The GPR predicts the solubility of the sulfate minerals with a degree of accuracy needed for typical subsurface engineering applications.