Estimation of flow-channel structures with uncertainty quantification: Validation by 3D-printed fractures and field application

Reinjection is an integral part of operating enhanced geothermal systems. Since cooling of reservoirs may occur due to cold-water injection, the possible effects of injection should be assessed. Subsurface structures, especially surface areas of flow channels connecting injection and production wells, determine the onset and rate of thermal breakthrough at a production well caused by reinjection. Previously, a method to estimate the surface area using temperature data was proposed (temperature-based surface area estimation method), which has only been applied to synthetic data from numerical simulation models and field data from geothermal fields with limited structural information available. In this study, we validated the temperature-based surface area estimation method through thermal flow experiments using a 3D-printed fracture network with known structural and physical properties. Based on measured temperature data, the flow-channel surface area was estimated with an approximate Bayesian uncertainty quantification method. The estimated uncertainty bounds were in good agreement with the design of the 3D-printed sample. We also applied the estimation method to field data from a well-studied experimental field. The estimates were consistent with other geophysical observations and previous numerical modeling studies, which had been used previously to probe fluid pathways in the field. It is expected that the thermal response estimation approach validated in this study can be useful for designing reinjection strategies. Furthermore, 3D-printed flow-channel networks may be useful for validating other estimation methods. [Display omitted] •We validated a method of estimating surface area of flow channels based on temperature data with uncertainty.•Thermal flow experiments were conducted using 3D-printed fractures.•Estimated flow-channel properties were in good agreement with the 3D-printed design.•The estimation method was applied to a well-studied mesoscale field.•The field estimates were consistent with other geophysical measurements and numerical studies.