Uncertainty quantification of medium-term heat storage from short-term geophysical experiments using Bayesian Evidential Learning

In theory, aquifer thermal energy storage (ATES) systems can recover in winter the heat stored in the aquifer during summer to increase the energy efficiency of the system. In practice, the energy efficiency is often lower than expected from simulations due to spatial heterogeneity of hydraulic properties or non-favorable hydrogeological conditions. A proper design of ATES systems should therefore consider the uncertainty of the prediction related to those parameters. We use a novel framework called Bayesian Evidential Learning (BEL) to estimate the heat storage capacity of an alluvial aquifer using a heat tracing experiment. BEL is based on two main stages: pre- and post-field data acquisition. Before data acquisition, Monte Carlo simulations and global sensitivity analysis are used to assess the information content of the data to reduce the uncertainty of the prediction. After data acquisition, prior falsification and machine learning based on the same Monte Carlo are used to directly assess uncertainty on key prediction variables from observations. The result is a full quantification of the posterior distribution of the prediction conditioned to observed data, without any explicit full model inversion. We demonstrate the methodology in field conditions and validate the framework using independent measurements. Plain Language Summary : Geothermal energy can be extracted or stored in shallow aquifers through systems called aquifer thermal energy storage (ATES). In practice, the energy efficiency of those systems is often lower than expected because of the uncertainty related to the subsurface. To assess the uncertainty, a common method in the scientific community is to generate multiple models of the subsurface fitting the available data, a process called stochastic inversion. However this process is time consuming and difficult to apply in practice for real systems. In this contribution, we develop a novel approach to avoid the inversion process called Bayesian Evidential Learning. We are still using many models of the subsurface, but we do not try to fit the available data. Instead, we use the model to learn a direct relationship between the data and the response of interest to the user. For ATES systems, this response corresponds to the energy extracted from the system. It allows to predict the amount of energy extracted with a quantification of the uncertainty. This framework makes uncertainty assessment easier and faster, a prerequisite for robust risk analysis and decision making. We demonstrate the method in a feasibility study of ATES design