Development of a fast, dynamic tank thermal energy storage model for system simulations.

Due to the enlarging share of renewable energy sources and their intermittent nature, the need for energy storage is increasing. This also includes thermal energy storage systems, as heat accounts for almost half of the global energy use. The most commonly used thermal energy storage system is the water tank, in which the temperature is elevated to charge the system. This is a well-known principle with already existing applications, such as district heating networks, industrial sites, and residential buildings. A dynamic model of a water tank is crucial for the design and sizing of these systems, as well as the development of a proper control strategy for charging and discharging. Most popular for system simulations are the 1D stratified tank multi-node models, in which the tank is split in multiple horizontal layers or nodes. These models provide good accuracy and reasonable calculation time – compared to 3D CFD models – and provide an estimate of the stratification in the tank. However, for the simulation and optimization of larger energy systems comprising multiple storage tanks, the calculation of the temperature in each layer often appears to be too computationally heavy.

A 0D model based on an empirical, lumped-parameter correlation would improve the implementation of thermal energy storage in system simulations. The charging time energy fraction method is a method to develop such a correlation, which relates the dynamic outlet state with the inlet temperature and mass flow rate of a storage heat exchanger. The proposed correlation was previously experimentally validated for a water tank. In this paper, the correlation is implemented in Dymola (Modelica) to assess its accuracy and computational time in comparison with the stratified tank model available in the IDEAS and Buildings library. Identical inlet conditions and geometric parameters are applied to both models and the outlet temperature of the heat exchanger and CPU time are compared.

Results of the inter-model comparison show that the outlet temperature in the newly developed model shows good correspondence with the stratified tank model with a maximum deviation of 8%. Computational time reduced by 25%, which indicates potential for the use of this model in system simulations. Another advantage of the model is that the used charging time method was originally developed for latent thermal energy storage, and the corresponding correlation could be implemented as well. Future research could focus on the extension of the model to latent thermal energy storage.