Abstract
Fractures largely control flow and heat transport in many geothermal reservoirs, but their connectivity is rarely known and they are difficult to simulate accurately. We show how Fimbul.jl, a fully differentiable geothermal reservoir simulator in Julia that represents fractures explicitly with a Discrete Fracture Model, can be used to model and calibrate fractured reservoirs in the context of fracture thermal energy storage (FTES). In a conceptual case, gradient-based optimisation calibrates fracture apertures and matrix thermal conductivity of an idealised model to match producer temperatures from a more complex reference model, achieving an almost perfect match in eight iterations and about five minutes of runtime. A field case from the Wesselkvartalet FTES plant in Asker, Norway, with 95 wells in a hydraulically fractured bedrock accumulator, demonstrates that the framework handles industry-grade models and highlights how subsurface connectivity affects system performance. Future work will address calibration against field data and control optimisation.