Global sensitivity analysis of a CaO/Ca(OH)2 thermochemical energy storage model for parametric effect analysis

Abstract

Simulation models have been widely used for thermochemical energy storage to better understand its behavior and consequently to improve operational control of the device. However, incomplete knowledge of system properties leads to a significant number of uncertain parameters in the simulation models, which in turn cause uncertainties in system predictions. In this work, we perform global sensitivity analysis to identify the effect of uncertain parameters on the outputs of a thermochemical energy storage model, so that we can better understand the predictive uncertainties, proceed with targeted data acquisition or even simplify the corresponding uncertainty quantification. To get reliable sensitivity analysis results, we use both variance-based and regional sensitivity analysis since they focus on different probabilistic features of model outputs. Since the simulation model is computationally expensive, we use model reduction via the (arbitrary) polynomial chaos expansion. Then, to further confirm the results, the regional sensitivity index is also estimated based on the original model with the same given sample set. Based on the results of both sensitivity analysis methods, we can find that there are 8 unimportant parameters among 16 analyzed parameters. Thus, we can focus resources on investigating the important uncertain parameters. Also, we can ignore the uncertainty of unimportant parameters to simplify the corresponding uncertainty quantification.