Neuro-Algebraic Digital Twin for Critical State Analysis of Mine-Water Geothermal Systems

Abandoned mine workings are a proven low-enthalpy resource for geothermal production and thermal storage. Mine water flow is governed by a complex interaction of the residual coal-seam geometry, backfill volume and rheology, and the heterogenous surrounding rock mass. The controlling subsurface properties – petrophysical, thermal and mechanical (pillar strength, fracture and fault stability) – and the mine topology that couples stress, flow, connectivity and heat transfer are all highly uncertain. Quantifying this uncertainty, and its impact on system development and safe operating margins, demands an ensemble of model realisations spanning plausible subsurface scenarios [1]. Current modelling capabilities and computational cost limit the number of full-field scenarios that can be simulated to de-risk operational strategy.

The UK Geoenergy Observatories (UKGEOS) Glasgow site provides extensive geological characterisation and dynamic monitoring well data [2], enabling a data-driven system that replicates subsurface behaviour while assimilating live monitoring. Building on digital twins (DTs), proven for complex engineered systems, we propose a DT for mine-flow geothermal systems to study, monitor, predict and prevent critical states.

The DT is a hybrid of sensor data, an AI component, and a physics-based algebraic model of the coupled rock–fracture–fluid–thermal–mechanical system. The physics model generates normal, transitional and critical scenarios that train the AI; during monitoring, the AI analyses real-time sensor streams to predict potentially hazardous states, while observations continuously refine the model. Predicted states are evaluated in the algebraic model, which then proposes mitigation strategies such as adjusting pumping rates, limiting pressure, or suspending operations to avert events approaching critical conditions.

[1] Aderemi T, Elma Charalampidou E. Demyanov V., 2026, Uncertainty in Shallow Mine Geothermal Flow Systems - A Scenario-based Approach, IEA Mine Energy Symposium, April 2026.
[2] MONAGHAN, A. A., DAMASCHKE, M., STARCHER, V., FELLGETT, M. W., KINGDON, A., KEARSEY, T., HANNIS, S., GILLESPIE, M., SHORTER, K., ELSOME, J. & BARNETT, M. 2021. UK Geoenergy Observatories Glasgow: GGC01 cored, seismic monitoring borehole – final data release. In: SPENCE, M. (ed.). Nottingham, UK.