Latent heat thermal energy storage systems exploit the solid–liquid phase transition of a material to store thermal energy almost isothermally with high energy density. Numerical methods are increasingly being used for the targeted design of these storage systems. The problem, however, is that the results obtained with these methods – despite a frequently shown near-perfect agreement between validation experiments and simulation results – are subject to large uncertainty. To illustrate the extent of this problem and to show approaches to solving it, a combined uncertainty and sensitivity analysis of the melting process of a paraffin (octadecane) in a cuboid heated from one side is carried out in this two-paper series. In part A, we present the model and validate it; in part B we use it to carry out the combined uncertainty and sensitivity analysis. The validation of the model is performed with the help of experimental results. For this purpose, we measure the liquid phase fraction, the position of the phase boundary, the velocity field in the liquid and the temperature inside the test capsule. We make sure that the experimental results are reproducible. Systematic deviations are discussed and, if possible, the results are corrected for their value. When measuring the velocity field using particle image velocimetry (PIV), the greatest systematic deviation is due to the inhomogeneous refractive index field in the liquid phase. Finally, it is shown that, in the present case, the influence is very low if constant material properties evaluated at the mean temperature are taken for each phase compared to a temperature-dependent implementation. An exception is the density. If the density change during melting is not taken into account the deviation of the melt fraction lies between 6 and 8%. In summary, the validation of the numerical model is very satisfactory and demonstrates the suitability of the model for performing the investigations of the part B paper.
Read full abstract