On groundwater flow and shallow geothermal potential: A surrogate model for regional scale analyses

Alberto Previati; Giovanni Crosta

doi:10.1016/j.scitotenv.2023.169046

On groundwater flow and shallow geothermal potential: A surrogate model for regional scale analyses

Sci Total Environ. 2024 Feb 20:912:169046. doi: 10.1016/j.scitotenv.2023.169046. Epub 2023 Dec 3.

Authors

Alberto Previati¹, Giovanni Crosta²

Affiliations

¹ DISAT - Department of Earth and Environmental Sciences, University of Milano-Bicocca, Piazza della Scienza, 4, Milan 20126, Italy. Electronic address: a.previati1@campus.unimib.it.
² DISAT - Department of Earth and Environmental Sciences, University of Milano-Bicocca, Piazza della Scienza, 4, Milan 20126, Italy.

PMID: 38052389
DOI: 10.1016/j.scitotenv.2023.169046

Abstract

Many cities worldwide lay upon alluvial aquifers which have a great potential for low temperature geothermal installations thanks to the thermal diffusive properties of saturated porous media and the constant temperature of the subsurface. In addition, aquifers with fast moving groundwater have a higher potential due to the additional energy replenishment by advection, which is often underestimated. This work aims at bridging the gap between quantitative hydro-thermal numerical analysis and regional scale assessment developing a process-based surrogate model for the estimation of the thermal exchange (geothermal) potential of ground source heat pumps (GSHP) considering groundwater advection. The proposed method is based on a synthetic 3D FEM model reproducing the infinite line source configuration and introducing groundwater advection. Conductive/advective g-functions were derived from the numerically simulated space-time thermal perturbation for a comprehensive set of hydrogeological regimes, and a surrogate model was developed by a machine learning (ML) regression of the thermal response of the system. This solution, beyond the run time of the numerical study and the ML training phase, is very fast, applicable at any scale and scalable to any desired depth. The trained model can be used to predict the geothermal potential of GSHP for almost all sedimentary basins around the world upon the availability of the required input data (aquifer thickness and saturation, aquifer porosity and groundwater flow velocity). In this study, large scale geothermal potential maps were generated from input layers implemented in a GIS, for a demonstrative area in northern Italy showing highly variable groundwater flow (Darcy velocity from 10^-3 to 10⁺³ m/y). A promising increase (up to +250 %) in the thermal exchange potential of GSHP due to the contribution of advection was highlighted discussing the benefits of groundwater flow and the amount of usable potential with implications on shallow geothermal energy management and development.

Keywords: Geothermal potential; Groundwater advection; Machine learning regression; Regional scale; Shallow geothermal energy; Surrogate model.