The scripts querying the NIST database in https://github.com/Simulation-Benchmarks/11thSPE-CSP/tree/main/thermodynamics were using the default reference state for determining the enthalpies. As it turns out, different reference states are used for H2O and CO2. This has been harmonised by setting the reference state to "IIR", corresponding to "setting enthalpy to 200 kJ/kg and entropy to 1.0 kJ/(kg-K) for the saturated liquid at 0 °C", see Reference State - REFPROP 10.0 documentation. Thanks to Kai Wendel for spotting this.
If you stick to the CSP description, phase internal energies and therefore enthalpies are assumed to be independent of the phase composition. Therefore, different reference states should not have any effect on the results. However, if your simulator calculates the values based on the composition, there can be an influence. Please double-check.
Kind regards
Bernd