The thermal properties of soil are vital in designing the length of buried heat exchangers. Based on a matured experimental technique commonly used for thermal response test in situ, a simulation model was built in this paper to determine the thermal properties of soil. Several sets of thermal response tests were carried out using the simulation model under different experimental conditions. The line heat source model (LHSM) and the column heat source model (CHSM) were applied to identify the thermal properties of soil. In order to analyze the variation of the identification error of the thermal properties under different experimental conditions, the identification results were then compared with the original setting values. The results showed that both identification methods could produce some errors in the identification of soil thermal and physical parameters under different experimental parameters. However, the accuracy of identification could be improved by selecting an appropriate test time. The best test time was duration to determine the thermal properties of soil was defined as an optimum length of time when the identification error was minimized. The best test time duration varied with different experimental parameters and different identification models. It is found that the best test time duration of the column heat source model was shorter than the one of the line source model. Giving longer lead time before counting the test time duration, increasing the thermal conductivity of the backfill material and heat injection, or reducing the circulating fluid flow rate, could effectively shorten the best test time duration. (C) 2018 Elsevier B.V. All rights reserved.