Dynamic simulation of a geothermal reservoir

Abstract

In the context of the energy transition, rapid development of the geothermal sector in the Netherlands has to take place. The first steps have been taken by establishing the Green Deal UDG between multiple industrial consortia, which agree to share knowledge on the research and use of ultra-deep geothermal energy . The Dinantian carbonates are of interest for the deep geothermal wells, because of their high geothermal potential. This research project provides a case study of the Californië geothermal doublets in Limburg (NL), which are currently the only geothermal wells in the Netherlands producing from the Dinantian carbonates. However, the static and dynamic model prove that the Devonian Bosscheveld formation and Condroz group are also part of the reservoir formation. Due to the tight matrix of the reservoir rocks, the permeability is believed to be fracture and karst (meteoric and hydrothermal) driven. The goal of this Thesis is to create a static and dynamic model of the reservoir that confirms the current production data (history match) and that explores the development of the geothermal potential of the reservoir in space and time. The static and dynamic reservoir model are based on a limited amount of well and seismic data, which forms the main challenge in this project. A framework of assumptions has been created to construct the model and a scenario- based approach has been applied to construct a best case scenario that matches the production data. A key element in the static reservoir model is the Tegelen fault. To estimate the impact of the Tegelen fault on the permeability distribution in the reservoir, a fieldwork in an analogue carbonate quarry has been executed. The results are applied in the static model. The dynamic results in this study show that the permeability configuration applied in the best case scenario results in BHP, flow rate and temperature values that are of the correct order of magnitude and within an acceptable error margin of the production data. Multiple sensitivities have been simulated to determine the range of parameters that may cause an inaccuracy in the results. Additional data acquisition is necessary to validate and optimize the static model, which will result in a dynamic model with an improved history match and a larger predictive power for future production and reservoir management.