This work introduces a model-agnostic deconvolution framework that combines fast Fourier transform (FFT)-based deconvolution with total variation denoising (TVD) for the interpretation of thermal response test (TRT) data on borehole heat exchangers. The method enables direct recovery of the impulse response function from raw observations, avoiding restrictive assumptions of conventional inverse modeling approaches such as the Infinite Line Source (ILS) model. Validation with both synthetic and field TRT datasets demonstrated that the proposed method consistently provides accurate impulse response estimation across different functional forms, outperforming ILS-based inversion particularly in capturing short-term transients and non-stationary thermal behavior. In a field TRT case, the reconstructed response achieved a mean absolute error of 0.009 °C, compared to 0.14 °C with the ILS model. Computationally, the framework operates with a complexity of O ( n log n ) and completed processing of 11,674 data points in less than 0.005 s, making it orders of magnitude faster than conventional nonlinear inversion schemes. The combination of accuracy, robustness, and efficiency makes the proposed framework a promising alternative for TRT interpretation, enabling real-time or near-real-time analysis and integration into automated monitoring workflows. • A deconvolution framework is proposed for deriving impulse response functions from TRT data. • The method relies on FFT and straightforward mathematical operations for efficient implementation. • Robust recovery of the impulse response is achieved through total variation denoising. • Recovered responses are validated against known reference signals with consistently high accuracy. • The method enables deeper insights into transient processes in subsurface systems.
Nguyen et al. (Fri,) studied this question.