Numerical simulations of the governing geophysical processes are crucial for geothermal applications in order to characterize the subsurface. This characterization presents us with major challenges ranging from the correct physical and geometrical characterization to the quantification of uncertainties. Quantifying rock physics uncertainties and performing other probabilistic inverse methods is, even with current state-of-the-art finite element solver and high-performance infrastructures, computationally not feasible for complex basin- and reservoir-scale geothermal applications due to the large spatial, temporal, and parametric domain of the applications. Therefore, a common approach is to construct, for instance, models with a lower degree of resolution. The consequence of this is a significant loss of the information content of the model. Hence, with these models, we fail to improve the characterization of the subsurface, as we will demonstrate in this work. As an alternative, we propose to construct a surrogate model by using the reduced basis method. The reduced basis method constructs low-dimensional models while maintaining the input-output relationship. Hence, we do not restrict our physical domain. In this presentation, we demonstrate how this concept can be used for enabling a combined workflow of global sensitivity analysis and uncertainty quantification to improve our understanding and characterization of the subsurface.

Fracture dimensions largely control rock properties like strength and permeability. Thus, knowing their statistical distributions is of great importance in many applied fields of the geosciences e.g., in geothermics, mineral exploration, and hydrology. They are also of academic interest since the statistical distribution of fracture dimensions (length, height, width) might provide inside in fracture formation mechanisms.

However, in the vast majority of cases this information is derived from observations in 2 dimensions, i.e., instead of a fractures length, not the true length, but a fractures-trace length (FTL) is measured. In conclusion information or even estimates about the length of an individual fracture and about the statistical distribution of fracture lengths of a fracture population are not possible.

We analyze the statistical distributions of FTLs mapped at 3 different scales under the application of different mapping schemes that are commonly used to account for the limitations that are unavoidable, when recording fracture lengths in 2 dimensions.

In our study we test how well powerlaw-, exponential-, Weibull-, lognormal-, and log-logistic distributions fit the FTL data. Our results show that FTLs are lognormal distributed independent of scale and mapping scheme and that the parameters of the lognormal distributions reflect outcrop quality and dimension.

In addition, we provide a comprehensive model that explains the observed lognormal distributions of FTLs. This model is based on random restrictions that control the observable FTLs and includes human error and bias in mapping.

Hydro-mechanical (HM) simulations are essential aspects of geothermal reservoir studies to assess the heat production and the associated-environmental impacts, such as seismicity. HM simulations are numerically expensive (especially for large-scale simulations) since they require a relatively fine mesh to capture the complex interplay between the fluid-flow and geomechanical processes. This aspect makes it difficult to perform detailed studies on uncertainties in HM simulations. In this work, we present a comprehensive review and numerical demonstrations about critical elements in HM simulations for geothermal applications. We then discuss potential surrogate models to reduce the computational cost in performing the simulations, specifically for uncertainty quantification and optimization purposes.

There are four important elements in HM simulations for geothermal applications: the equation of state, the porosity-permeability relationship for both the matrix and fracture, the stress-dependent porosity in the matrix and fracture, and lateral and vertical heterogeneities. We compile the discussion from more than 60 papers and numerically show the significance of these parameters using the MOOSE simulator. The incorporation of these parameters into HM simulations leads to realistic descriptions of geothermal applications. However, accommodating for these complex physics also elevates the computational cost.

We compile surrogate-modeling approaches (dedicated for HM problems) from more than 40 papers. The approaches span from reduced-basis to polynomials chaos expansion methods and machine learning approaches. We found that the combination of reduced-basis methods and machine learning approaches enables to effectively deal with non-linearity in HM simulations, to preserve the physics, and to reduce computational cost for further uncertainty quantification.

The analysis of uncertainties in the description of the subsurface is an important aspect for resource exploration and material storage. Because of the complexity of the subsurface and an often inhomogeneous data situation, models exhibit several aspects of uncertainties. These may be caused by the interpolation of locally sparse data and must be considered when constraining a structural geological model. Further, these interpolations may be subject to errors caused by psychological biases, which need to be identified to avoid error propagation during the model building.

The aim of this study is to develop structural geological models of the Cretaceous units of the Münsterland Basin on the basis of stratigraphic boundaries and orientation measurements derived from maps, boreholes and literature as a framework for future geothermal exploration. In the model construction phase, it is attempted to separate processed input data in a first model setup from additional geological assumptions required to obtain geologically meaningful representations. Potential sources for bias are evaluated during the data processing, and standard deviations of input data points are accounted for during a subsequent uncertainty analysis using probabilistic geomodelling approaches.

The resulting structural models reveal the effects and limitations of purely input data-driven models versus models with additional integrated data and the uncertainties derived from different input data types. The integration of results of the planned seismic investigation in 2021/2022 by the Geological Survey NRW and the results of seismic campaigns acquired in the 1970s and 1980s may help to close these knowledge gaps in future work.

Deep geothermal energy is a key to lower local and global CO2 emissions caused by the burning of fossil fuels. Different initiatives aim at establishing deep geothermal energy production at the Weisweiler coal-fired power plant near the city of Aachen, Germany, in order to replace district heat generated as a side product of coal burning. But how much information do we actually have about or need of the subsurface to carry out such a project?

The conducted investigations will provide a 3D geological and probabilistic subsurface model of the area between Aachen and Weisweiler created with the open-source package GemPy developed at RWTH Aachen University. This model is in contrast to established regional models and more detailed local models.

The geological structures between Aachen and Weisweiler represent a SW-NE striking syncline, the Inde Syncline, embedded in the Aachen fold-and-thrust belt. The syncline is offset by Cenozoic normal faults of the Lower Rhine Embayment. The target layers comprise of karstic Lower Carboniferous Kohlenkalk platforms and Upper/Middle Devonian Massenkalk reef carbonates outcropping along the flanks and down faulted within the Lower Rhine Embayment.

Results show that the Aachen fold-and-thrust belt and the down faulted fault blocks can be modeled integrating the available surface and sparse shallow subsurface data. The probabilistic modeling provides information about uncertainties of the target layers in the subsurface. It can be deduced that a planned exploration well for fall/winter 2021 will reduce uncertainties in the subsurface in the vicinity of the target layers enabling improved economic decisions.