skip to primary navigationskip to content
 

EECi welcomes visitors Hideki Kikumoto and Wonjun Choi from the University of Tokyo

last modified Jul 11, 2017 10:44 AM

In June Dr. Hideki Kikumoto and Dr. Wonjun Choi from the Institute of Industrial Science (IIS) at the University of Tokyo visited the EECI group to initiate collaboration on different applications of Bayesian techniques for computer model calibration.

Based on the competence of the IIS regarding the evaluation of thermal response tests (TRT), which are conducted to optimise the design of borehole heat exchangers, and the experience of the EECI group in Bayesian calibration, the existing calibration framework was adapted for application on TRT data and different thermal response models. Compared to standard approaches the newly applied technique considers various sources of uncertainty, such as measurement error and model discrepancy.

Preliminary results suggest that the approach can correctly identify the model discrepancy of simpler thermal response models, which disregard certain physical effects, such as variable heat input rates. Thus, the new approach can successfully prevent overfitting of the calibrated model parameters, and consequently provide more reliable parameter values than conventional methods. 

In addition, discussions about future work also focused on the development of a novel hierarchical approach to enable calibration of low-accuracy measurement sensor, e.g. temperature or wind sensors, against high/accuracy sensors, while addressing different forms of uncertainty in a Bayesian framework.