A hybrid analytical-PINN model for subsurface simulation of geothermal heat exchangers in heterogeneous underground

2026-07-14 · Source: Machine Learning · Field: Science & Research — Artificial Intelligence & Machine Learning, Engineering & Applied Sciences, Mathematics & Computational Sciences · Depth: Expert, quick

Summary

A novel hybrid analytical-Physics-Informed Neural Network (PINN) model has been developed for simulating subsurface thermal behavior of borehole heat exchangers (BHEs) in heterogeneous soil. This framework addresses the complex problem by naturally removing singularities through analytical line source models. It incorporates an explicit formulation for gradient thermal conductivity, which facilitates physics-informed learning of conductivity parametrization. Furthermore, the model utilizes a learned correction as an efficient universal corrector, leveraging superposition principles. The approach involves decomposing temperature change and reformulating the problem into a governing correction diffusion or advection-diffusion equation, enabling effective neural network training by excluding delta function singularities. The PINN approximates a universal corrector for a single borehole, trained by minimizing a combined physics-informed and data-anchored loss function using sampled conductivity parameters and adaptively selected training points.

Key takeaway

For research scientists developing models for geothermal energy systems, this hybrid analytical-PINN approach offers a robust solution for simulating borehole heat exchangers in heterogeneous underground. You should consider integrating singularity removal via analytical models and leveraging physics-informed learning for material parametrization. This method provides an efficient, universal corrector, potentially reducing computational costs and improving accuracy in complex subsurface thermal simulations.

Key insights

A hybrid analytical-PINN model effectively simulates geothermal heat exchangers in heterogeneous soil by removing singularities and learning universal corrections.

Principles

• Singularity removal simplifies complex physical problems.
• Physics-informed learning can parametrize material properties.
• Superposition enables universal correctors for efficiency.

Method

Decompose temperature change, reformulate as a correction diffusion equation, then train a PINN with physics-informed and data-anchored loss using sampled conductivity parameters and a location indicator function.

In practice

• Simulate BHEs in complex geological formations.
• Model thermal conductivity variations in soil.
• Develop efficient, universal thermal correctors.

Topics

• Geothermal Heat Exchangers
• Physics-Informed Neural Networks
• Subsurface Simulation
• Thermal Conductivity
• Heterogeneous Media
• Singularity Removal

Best for: AI Scientist, Research Scientist

Related on AIssential

• Overcoming the Limits of Finite Difference Method; Physics-Informed Neural Network for Noisy High-Dimensional Heat Diffusion — PINNs overcome classical numerical method limitations in noisy, high-dimensional heat diffusion, maintaining accuracy and efficiency.
• CO$_2$ sequestration hybrid solver using isogeometric alternating-directions and collocation-based robust variational physics informed neural networks (IGA-ADS-CRVPINN) — Hybridizing IGA-ADS with CRVPINN significantly accelerates $CO_{2}$ sequestration simulations while maintaining accuracy.
• Material-Agnostic Zero-Shot Thermal Inference for Metal Additive Manufacturing via a Parametric PINN Framework — A parametric PINN framework enables zero-shot thermal modeling across diverse materials in metal AM with high efficiency.
• Physics-Informed Neural Network Modeling of Biodegradable Contaminant Transport through GCL/SL Composite Liners — Hard-constrained PINNs significantly improve accuracy and stability for modeling contaminant transport in composite liners.
• Using AI, Mathematicians Find Hidden Glitches in Fluid Equations — Bespoke PINNs can precisely identify previously unobserved unstable fluid singularity candidates across diverse fluid models.
• A Physics-Informed Neural Network Framework for Elastodynamic Wave Propagation in Bimaterial Systems — PINNs combined with explicit FEA accurately model elastodynamic wave propagation in bimaterial systems.

See Counsel's argued verdicts on the open AI decisions leaders are weighing →

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Machine Learning.