Modelling a latent heat thermal storage demonstrator and identification of the model key-parameters

Abstract

Latent Heat Thermal Energy Storage (LHTES) has a great potential to reduce CO2 emissions in urban areas, when associated to a district heating network. It facilitates the flexibility of the network by decoupling the production and consumption. Thus, consumption peaks that are often managed with fossil fuels will be managed by unloading the storages of the network. LHTES based on tube and shell is the most widely studied technology, as it combines low cost, high maturity, and high compactness factor. This paper presents a numerical model to simulate the thermohydraulic behaviour of a shell-and-tube LHTES, based on a 1D approach for the heat transfer fluid (HTF) and a 1.5D approach for the Phase Change Material (PCM). The objective is to develop a model representing a whole system, faster than a Computational Fluid Dynamic approach while keeping a sufficient degree of accuracy. The paper focuses on the presentation of the model and its validation against experimental data from a real-scale 180 kWh demonstrator, currently in operation in the substation of the urban heating network of Grenoble, France. The validation against three different charges is presented, two charges at a constant flow rate (0.6 and 1.6 kg/s-1) and one at constant power (60 kW). The model gives results with an accuracy superior to 95 % and >80 % of the points are between the experimental uncertainty curves. It has also been found that the model loses its accuracy when the inlet HTF temperature abruptly decreases and the top collector does not behave as a stratified tank anymore (the relative error on total energy accumulated reaches 35 %). In the second part of the paper, a sensitivity study to the main parameters of the model shows that the model is sensible to the discretization of the collectors, the precision of the Nusselt correlation for convective transfers in the tubes, and the fusion temperature of the PCM. These parameters can induce up to respectively 22 % (0D model for the collectors' volume), 4.5 % to 9 % (considering the annular space as a flat plane or a cylindrical tube) and 3 % to 6 % (error of 0.5 to 1 °C in the PCM fusion temperature) error.