Floatation point error
 

Hi,

I am trying to run a simulation for a centrifugal compressor with co2 (initial pressure of 65 bar and pressure ratio is 2). I am getting the finmes error, the solution is diverging. So I went through the steps mentioned in best practices guide. First I tried to get the initial condition using physical timescale and upwind condition for first order numerics. I tried a iterative process for the timescale from 1e-6 to 1e-9 but still faced the same issue. I tried using local timescale and then change to physical timescale once the solution is steadying but even that failed after a few iterations. The mesh quality is good. The boundary conditions are correct (I ran a similar compressor but for initial pressure of 33.7 bar and pressure ratio 2, the solution converged with the boundary setup). So could anyone help with the issue? (The gas is in vapor phase).

If the inlet pressure is 65 bar the outlet pressure will be ~130 bar, which will bring the CO2 above its critical pressure, correct? Will the CO2 even be compressed into the supercritical region? Do you use real gas?

Hi
Yeah I am using real gas conditions. Yeah it will compressed into the Supercritical region

The first thing to do here is to do your simulation using a simpler material model, maybe ideal gas. If the simulation runs with an ideal gas model then you know the problem is the material model - that means you know what issue you need to address.

I do am facing a table out of bounds warning. I generated the material properties table using the RGP generator in the latest 2023r2 CFx version. Maybe I'll try running it with ideal gas and see if it runs properly. Also I used the method to generate the data for my other case that's running below the critical point, it converged fine after a while.

I tried with the ideal gas conditions, and the solution converged easily. So it's not an error with boundary conditions or meshing. Now I am gonna try it using regular EOS equations like redlich kwong and see if it works(even though they are not perfectly suitable for my condition they at least give solutions rather close to my condition). But I do have another question, is there any other way to generate the RGP file? (I do not have access to refrop).