New algorithm for estimating parameters from interrupted thermal response tests of ground heat exchangers

It is a long-standing challenge to determine soil thermal properties from interrupted thermal response tests (TRTs). To meet this challenge, this study develops a new parameter estimation algorithm by integrating three key elements: a short-time model, a derivative-enhanced objective function, and the zero-order Tikhonov regularization strategy. The algorithm is validated to be reliable by a reference sandbox dataset. Whereas interruption reduces the test duration and the number of available test data, it also increases the independence of sensitivity coefficients, thus resulting in high identifiability. To achieve this identifiability, a reliable short-time temperature response model is the prerequisite. Whereas temperature derivatives can boost the temperature data as a set of independent experimental data, regularization is highly necessary to stabilize the iterative solution process by suppressing the influence of the data noise amplified by the temperature derivatives.