Estimation of soil and grout thermal properties through a TSPEP (two-step parameter estimation procedure) applied to TRT (thermal response test) data

In this paper, a versatile TSPEP (two-step parameter estimation procedure) based on a three-dimensional numerical model of a geothermal system is presented. The procedure is applied to both simulated and experimental TRT (thermal response test) data in order to restore the grout and soil thermal conductivities and volumetric heat capacities. The TSPEP is essentially a two-step process. The first step uses the parameter estimation procedure, in the early transient regime to restore the grout thermal conductivity and volumetric heat capacity. The values from the first step are used as the input values in the second step, in which the parameter estimation procedure is applied to the late transient regime to restore the soil thermal conductivity and volumetric heat capacity. Further iterations of these two steps can be used to improve the accuracy of the procedure and are discussed in this paper. The time separation used between the estimation of the soil properties and the estimation of the grout properties partially uncouples the two problems and makes the estimation of these four parameters feasible. A criterion to select the time separation is discussed and validated in this paper.

Research highlights
► A versatile Two Step Parameter Estimation Procedure (TSPEP) applied to Thermal Response data is suggested and validated.
► The procedure, based on a numerical three-dimensional model of a geothermal system, is aimed at restoring.
► The proper value of both the soil and the grout thermal conductivities and volumetric heat capacities.
► The procedure is first verified through its application to simulated noisy data. It accurately restores the proper value of both the soil and the grout thermal properties.
► The TSPEP is then validated through its application to a medium-scale set of experimental laboratory data.
► To the authors’ knowledge, three-dimensional models have never been coupled with parameter estimation procedures applied to experimental data under an inverse problem approach