Comparative study of models for packed bed molten salt storage systems

To date various models for packed bed single tank molten salt storage systems have been developed. The question arises in how far certain modelling assumptions can affect the outcome of these models. This can also depend on the applied use case, for example a short validation simulation versus an annual simulation. Thus, a comparative study is performed, considering four one-dimensional models for packed beds: the Schumann model, the continuous-solid-phase model, the single-phase model and a recently developed bidisperse model. Additionally, the impact of heat losses and variable fluid properties are considered as well. The comparison is performed for different use cases and covers the single blow operation, cyclic operation and annual simulations. Common boundary conditions are derived from a 100 MWel concentrating solar thermal power plant with molten salt as heat transfer fluid. A large and a small storage volume, each with a small and a large particle packing are designed to cover a range of practical storage configurations. The aim of this work is to provide guidelines for the selection of a reasonable level of modelling detail depending on the desired accuracy and computational effort. Results indicate that the outcome can significantly differ between the use cases. Effective conductivity has visible impact in stand-by periods and small sized packings with strong thermal gradients. The single-phase model has low computing effort but is applicable only with packings in the range of millimeters. The bidisperse model has significantly improved accuracy when applied to packings with smaller particles in combination with larger particles. Heat losses showed little impact for the considered storage volumes. Temperature dependent fluid properties have an impact within short periods but average out in cyclic and annual simulations, which justifies neglecting them.

Automated summary

Various models for packed bed single tank molten salt storage systems have been compared, considering different use cases and considering the impact of heat losses and variable fluid properties. The results show that the choice of model significantly affects the outcome, and the results vary for different use cases. Therefore, selecting a reasonable level of modelling detail based on desired accuracy and computational effort is important.