Development of chiller-attached apparatus for accurate initial ground temperature measurement: Insights from global sensitivity analysis of thermal response tests

Abstract Local sensitivity analysis (LSA) is widely used for evaluating the uncertainty in a model output or estimated parameters. However, it does not consider the effect of changing parameter values in the parameter space and the simultaneous change of the parameter set. Global sensitivity analysis (GSA) overcomes these limitations of LSA by adequately representing the propagated uncertainty in model usage or parameter estimation. In this study, by using Sobol’s method, GSA was conducted to analyze the temporal uncertainty transition of model inputs required in the thermal property estimation via a ground thermal response test (TRT). The obtained results provide important insights; specifically, among the various input parameters, the initial ground temperature and ground volumetric heat capacity account for more than ∼27% and ∼20% of the total uncertainty in borehole resistance estimation, respectively. Further, the estimated borehole resistance is larger than intended owing to the uncertainty in the borehole diameter, which is likely to increase during the construction process. Thus, a deterministic estimation by fixing the borehole diameter leads to a lower estimate than the actual resistance. Then, the insights obtained are used to develop a chiller-attached TRT apparatus to solve the limitations of conventional apparatuses that cannot measure the initial ground temperature accurately. By using the developed apparatus that can accurately control the fluid temperature, the temperature at which the circulating fluid and surrounding soil are in thermal equilibrium can be asymptotically found with an uncertainty range of 0.1 °C. This improvement in the initial temperature accuracy results in a 24% reduction in the estimation uncertainty of the borehole resistance compared to conventional practice.