Extended Decline Models Improve Geothermal Temperature Forecasting
Mina Khalaf et al. (submitted Jan 2, 2026) propose physics-consistent extended Arps decline curves and equation-informed surrogates to forecast temperatures in Enhanced Geothermal Systems. They validate the equilibrium-temperature-augmented decline family against Utah FORGE measurements and THM simulations, then train an equation-informed neural network, Gaussian Process surrogate, and XGBoost baseline; GP achieves RMSE 3.39°C and MAE 2.34°C across 3–60 month horizons. Implication: provides calibrated, efficient multi-horizon forecasting tools for EGS design and operations.
Key Points
- 1Generalizes Arps decline by adding an equilibrium-temperature term, enforcing finite late-time temperature limits.
- 2Validates models on Utah FORGE data and THM simulations, demonstrating near-perfect extended-decline fit (median RMSE 0.071°C).
- 3Offers practitioners equation-informed NN and Gaussian Process surrogates for multi-horizon forecasts, with calibrated uncertainty.
Scoring Rationale
Strong methodological novelty and practical surrogates for geothermal forecasting, limited by single-source preprint and narrow domain focus.
Sources
Public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems