TO-22-C047 - Improvement of Estimation Accuracy of Soil Effective Thermal Conductivity and Ground Heat Exchanger Simulation by Data Assimilation using Ensemble Kalman Filter

The prediction accuracy of ground heat exchanger simulation is an important factor in the design or control of ground heat utilization systems. Uncertainty
in simulation is inevitable due to model imperfectness and estimation errors in initial/boundary conditions, and model parameters. On the other hand, there
is a method called data assimilation, which modify the results and estimates/reduces the uncertainty, assimilating the observation into the model predictions.
Here we show that the feasibility of improving the accuracy of the simulation and the estimation of soil effective thermal conductivity by applying the ensemble
Kalman filter, one of the data assimilation methods, to the ground heat exchanger simulation. In this study, we conducted data assimilating simulation by
generating virtual observations from another true simulation results. A numerical experiment called the twin experiment was conducted to compare the
results of the data assimilation with the true simulation values. Data assimilation was performed assuming that the inlet and outlet temperatures of the
ground heat exchanger were to be observed, and the observed values were generated by adding a normal random number of N(0, 0.5) to the true value. As
a result, the estimated values of the inlet and outlet temperatures of the ground heat exchanger were 0.033 and 0.007, respectively, as the time average of
the error from the true value, these values are smaller than the standard deviation of the observed values. Furthermore, the ensemble Kalman filter method
can improve the estimation accuracy for variables that are not observed. For the soil effective thermal conductivity, while the true value is 2.000 W m-1 K-1,
an initial distribution of 2.300 ± 0.500 W m-1 K-1 is given for data assimilating experiment, resulting in an estimated value of 2.001 ± 0.001 W m-1
K-1, which greatly improves the estimation accuracy. The application of the ensemble Kalman filter to ground heat exchanger simulation has shown the
feasibility of improving the accuracy of prediction calculations and the estimation of the soil effective thermal conductivity. This method can be applied to
systems for which observed data are available, such as those in actual operation, and is expected to be applied to real-time prediction calculations such as
model predictive control. Furthermore, it is suggested that the method can be used for inverse analysis of soil properties such as effective thermal conductivity
of soil from actual operation data of shallow geothermal systems.