| Abstract |
Geological modeling is an integral part of geothermal resource estimation, exploration and reservoir modeling. A geological model typically consists of two components: a geometric representation of boundaries between major lithological units and discontinuities (faults, unconformities), and a volumetric model of relevant property distributions within each lithological unit (e.g., porosity, permeability). Both aspects contain significant uncertainties, but while multiple established methods exist to consider uncertainties in the volumetric model (e.g., conventional geostatistics, machine learning approaches), the geometric representations are still often treated as known. We motivate the consideration of uncertainties in the structural model with a simple example of the calculation of average values for bulk estimates of a specific quantity of interest, a geothermal resource estimate based on heat-in-place calculations. We consider the case that the thickness of the resource layer is subject to uncertainty. If the distribution for the bulk estimate is obtained from independent sampling, then the obtained distribution of average values has a lower variance than for the case where the variables are correlated. This means that the assumption of independence leads to an underestimation of uncertainty. A way to avoid this mistake is to explicitly consider spatial correlations of geological interfaces using probabilistic geomodelling approaches. We describe a method to enable such an approach and show the application to a geothermal resource study at the Weisweiler site in Germany. This example shows that spatial correlations in the geological model can be considered, even for structurally complex settings. With an automation of the modeling to simulation workflow, it is furthermore possible to calculate the hydrothermal state for each geological realization and to obtain maps of geothermal resource density with a corresponding estimate of uncertainty. Most of the shown aspects are not entirely novel, but they illustrate the possibility to consider uncertainties in geological models in conventional geothermal simulation workflows. |