Using the fluidproperties module to calculate fluid flow results appears abnormal

[shin-chan-cpu]

In the moose version released in April, the CO2FluidProperties class added e_from_p_T and cp_from_p_T two functions so that my input card can start to perform steady-state two-dimensional tube heating flow calculations, but my output temperature and pressure have not changed. In my separate tests, the output results of these two functions are basically consistent with the reference data given in the official website tutorial, but the results I get have not changed. I replaced it with Water97FluidProperties for calculation. The calculation results look normal. I hope to know how to optimize it next

[GiudGiud]

Hello

It would be good to isolate the problem.
Can you use simple a MaterialRealAux auxkernel and this material
https://mooseframework.inl.gov/source/materials/FluidPropertiesMaterialPT.html
to output these properties and make sure they are correct.

Guillaume

[GiudGiud]

If we have not reached steady state then the odd solution could just be a transient state, somewhere on the way from an unphysical initial condition to the real steady solution.

It's hard to know without running the input.
Feel free to output all the fluid properties computed for CO2 and check for inconsistencies. It may be that in the parts of the domains where T and P are unconverged, the fluid properties of CO2 are unphysical and that leads to even more problematic results

[shin-chan-cpu]

I am currently using the constant physical property model for calculation, because the variable physical property model calculation effect is very poor. The properties of the fluid take the parameters near the temperature and pressure

[GiudGiud]

ok. Can you please plot these results? And maybe also the evolution of some integral quantities though time? We would want to know if they are converged

[shin-chan-cpu]

I recalculated it, and the current residue looks like this.

[GiudGiud]

That doesnt look very converged for the energy equation.
A plot with each iteration reported would be better than a screenshot