Faster computation of g-functions used for modeling of ground heat exchangers with reduced memory consumption

Abstract

Temperature response functions, known as g-functions, are a computationally efficient method for simulating ground heat exchangers (GHEs), used with ground-source heat pump (GSHP) systems or direct ground cooling systems as part of a whole-building energy simulation. In fact, at present, there are no other methods that have sufficient accuracy and are fast enough to simulate a ground-source heat pump system in a whole-building energy simulation.

The concept, mathematical derivation and an implementation of a g-function calculation program were originally developed by Claesson and Eskilson (1985). More recently (Cimmino 2018a, Cimmino 2018b, Cimmino 2019) developed an open-source g-function calculation tool known as pygfunction. This tool offers great flexibility for the user to compute g-functions for specific configurations of boreholes. However, for large borehole configurations (with ~1000 boreholes), the required time to compute a single g-function can take several hours, and the required RAM can be on the order of 100 GB, greatly exceeding most desktop PCs. In order to develop libraries of g-functions and training sets for machine learning approaches, we are computing hundreds of thousands of g-functions. This paper describes further development of Cimmino's methodology to speed the computation and reduce the memory requirements.