Surrogate models for power output per unit length, wall and outlet g-function predictions of single thermo-active piles

This repository provides open-access surrogate models and datasets for the rapid thermal performance prediction of single thermo-active piles. The resources enable estimation of power output per unit length (W/m) and dimensionless G-functions for both pile wall and fluid outlet temperatures over time. The surrogate models are based on machine learning and trained on finite-element simulation data. They are intended to replace computationally expensive numerical analyses during preliminary design and parametric studies. The repository includes trained models, associated scalers, pile numerical datasets, and Python scripts for straightforward public deployment.