History Matching an Operating Aquifer Thermal Energy Storage System in the Heterogeneous Chalk Aquifer under London

The Chalk aquifer in London is a highly heterogeneous dual-porosity system, characterised by high-permeability features resulting from fracturing and/or karstification within a low-permeability matrix. These high-permeability zones can be truncated by faults or marl layers of low permeability. Boreholes in London frequently reveal a prominent high-permeability flow zone at the top of the Chalk. Despite this, many predictive models for shallow geothermal systems in the region treat the aquifer as homogeneous, potentially leading to significant overestimations of heating and cooling delivery.

This study evaluates the influence of aquifer heterogeneity on the performance of Aquifer Thermal Energy Storage (ATES) systems through a model calibrated with data from an operational ATES installation in London. The system consists of four borehole doublets. Initial analysis indicates a well-balanced energy ratio of 0.09 and thermal recovery rates of approximately 40% for warm wells and 25% for cold wells. This presentation focuses on the four-step history-matching methodology employed in the study.

In Step 1, quality control was applied to the observed data, including lithological and flow logs, hydraulic head and flow rate data from borehole commissioning tests, and hourly flow rate and temperature measurements spanning five years of operation. Quality control was challenging due to the operational configuration, in which boreholes function as independent doublets, alternating between primary doublets during heating and cooling cycles and activating additional doublets as needed to meet demand.

Step 2 involved constructing multiple plausible geological scenarios informed by data from other boreholes in London, prior studies on Chalk aquifer heterogeneity, and field observations of Chalk outcrops. Permeability values were calibrated to match borehole flow log data.

Step 3 used the Nelder-Mead optimization technique to iteratively refine model inputs, achieving a match to borehole commissioning test data while maintaining consistency with flow log data. This step resulted in a set of hydrogeological models that provided comparable quality matches to the test data.

In the final step, the Nelder-Mead optimization was employed with the ensemble of models from Step 3 to match the temperature profiles recorded during system operation. This phase posed challenges due to the extended simulation times required. The outcome was a suite of coupled thermo-hydrogeological models that accurately reflected the observed data. The results highlight the critical role of heterogeneity in shaping thermal plume behaviour within the Chalk aquifer. Thin, high-permeability layers lead to "pancake-like" thermal plumes, which exacerbate conductive heat losses and increase the risk of thermal interference between laterally offset boreholes.

These findings emphasise the importance of accounting for subsurface heterogeneity in designing and operating ATES systems. The results are being used to evaluate the feasibility of scaling ATES technology across London.