Grey-box model and identification procedure for domestic thermal storage vessels

This paper proposes a model and estimation algorithm, which can automatically characterize a broad range of domestic hot water cylinders and hot water storage buffers. A grey-box compartmental model takes into account the heat loss, internal heat exchange, convection and mixing dynamics associated with water storage systems. Models for these systems are often used in model-predictive controllers. The estimation algorithm is able to identify, in a robust way, the model characteristics for a diversity of storage vessels. It is based on the Markov-Chain Monte-Carlo method, which makes the procedure suited for automation since local minima in the cost function can easily be circumvented. The identification procedure is tested on four different vessels in a distributed thermal storage lab-setup. Two domestic hot water cylinders and two hot water storage buffers have been monitored in a series of charge-discharge tests. It is able to adequately reconstruct the temperature variations inside the storage vessels (errors are smaller than about 5 °C). This algorithm is suited for predicting the state-of-charge of thermal energy storage vessels in model based control applications.