Abstract
Underground thermal energy storage (UTES) offers significant potential for large-scale energy storage, such as utilizing waste heat and balancing renewable energy sources like wind and solar power. Thermal subsurface flow exhibits distinctly different temporal scales, from rapid, advection-dominated flow in wellbores and fractures to slow, conduction-dominated flow in solid rock. The governing equations for mass and energy conservation are also strongly coupled due to pressure- and temperature-dependent fluid properties. Therefore, simulation technology for these systems requires robust nonlinear solution strategies that ideally can resolve processes at their intrinsic timescales, especially during abrupt temperature changes in the near-well region at the onset of charging and discharging.
The nonlinearities in UTES are mainly localized spatially (in faults/fractures and near wellbores) and temporally (at the onset of charging/discharging). This work demonstrates how we can utilize this by devising nonlinear domain decomposition strategies for right-preconditioning of Newton’s method in the open-source, fully differentiable JutulDarcy simulator. We demonstrate the method on real and realistic UTES scenarios, and discuss how the temporal resolution can be adapted in space and time to achieve high accuracy without compromising performance.