Assessment of the usability of OpenGeoSys tools for Aquifer Thermal Energy Storage (ATES) simulations

Abstract

Aquifer thermal energy storage (ATES) is an energy efficient technology to temporarily store groundwater of different temperatures in an aquifer. The basic idea behind ATES is to store thermal energy in warm and cold wells so that the energy can be used for heating and cooling of buildings in the next season. Design and planning of ATES systems requires numerical simulation tools such as SEA- WAT, COMSOL, and FEFLOW. COMSOL and FEFLOW are expensive commercial products. SEAWAT is a free computer program based on MODFLOW and MT3DMS. The flow field is modelled with the finite difference method while there are several methods to simulate the heat transport including the finite difference (FD) method and the total variation diminishing (TVD) method. OpenGeoSys (OGS) is an open-source alternative based on the finite element method which is able to simulate groundwater flow and heat transport processes and can potentially be used for design and planning of ATES systems. There are very few applications of OGS to simulate ATES systems. The main goal of this research is to assess the usability of OpenGeoSys in the simulation of ATES sys- tems. The numerical solutions of OGS and SEAWAT (both the TVD scheme and FD scheme) are compared to the analytical solutions for a single well that injects water with a constant flow rate of water and constant temperature. For a doublet with a cold well and a warm well, the OGS solution is compared to the SEAWAT solution for both a sin- gle cycle system and a multiple cycles system with constant flow rates and temperatures. Finally, a field case of a doublet with varying flow rates and injection temperatures is studied to compare the performance of OGS to SEAWAT. The main findings of this study are as follows. Both OGS and SEAWAT can reproduce the heat transport process of the analytical solutions for a single well injecting warm or cold water. There are some deviations between the temperature distributions computed by the numerical models and the analytical solutions which are primarily caused by nu- merical dispersion. The TVD scheme results in the smallest numerical dispersion while OGS shows slightly more numerical dispersion when applying the same number of cell- s/nodes. Numerical dispersion can be reduced by application of finer spatial resolution for the SEAWAT since the time step is adjusted automatically. For OGS, the time step must be reduced manually when the grid is refined. Both OGS and SEAWAT show energy balance errors smaller than 1%. Numerical dis- persion has an effect similar to physical dispersion so that the thermal radius increases when the numerical dispersion is larger. The interaction between the warm well and cold well of a doublet increases when the thermal radius increases, which may result in an increase of the energy balance error. According to the analytical solution for the model where the heat transport between the aquifer and aquitards is included, OGS shows the smallest overestimation of thermal energy that remains in the aquifer part compared to the two solution schemes of SEA- WAT. The vertical temperature distribution in aquitards simulated with OGS is closer to that of the analytical solution compared to SEAWAT. Both numerical dispersion and the overestimation of thermal energy in the aquifer have an influence on thermal recovery efficiency of an ATES system. Larger numerical dis- persion indicates a longer heat transport distance, resulting in more time needed for thermal energy to transport back to the wells during the extraction period. Similarly, the overestimation of the thermal energy in the aquifer provides more thermal energy dur- ing the extraction period. There are some overshoots/undershoots when simulating an ATES system with varying flow rates and injection temperatures by the TVD scheme. This can be eliminated by refining the spatial resolution. It is more complicated to simulate an ATES system with OGS than SEAWAT. The com- putational time is much longer for OGS compared with SEAWAT by a factor of around 10 for the same number of cells/nodes. Despite these drawbacks, OGS is suitable for the simulation of ATES systems as the simulated temperature distribution in both the aquifer and aquitards compares well to the analytical solution, and there are no prob- lems of overshoots/undershoots in the simulation of varying flow rates and injection temperatures.