problem about pressure

[Yuan Huang]

hi everyone,
      I am trying to calculate the flow of an engineer, there is no combustion, so the temperture or the Enthalpy is not considered in the calculation and the density is set to constant.
      when I set p0=1atm, rom0=1.17kg/m3, I can get the result, but when I  set p0=33atm, rom0=38.61kg/m3 and the other condition is the same,  the result is divergence.
      In the SIMPLE algorithm, only the variety of pressure is used in the calculation, so I think  the calculation is i ndependent with p0, but the result seems to be not so.
      How did this happen?
     Thanks!

Yuan Huang

[Alexandre Douce]

Hum, it is hard to say something without more details...
Just notice that  P0 must be set in Pascal, and if the pressure plays no thermodynamical role you could just set P0 = 0 because as you said only the gradient of the pressure does matter in SIMPLE.

[César Vecchio]

This is just a guess, but perhaps your fluid has now a greater inertia for such a small viscosity, the boundary layer becomes thinner and so will your y+. Can you check that your y+ has values suitable for your turbulence model? Does this happen if you run the simulation as laminar too? Have you tried changing IMRGRA and relaxation coeffs? Also, and this could be trivial but it's a source of errors, just like Alexandre suggested you for pressure, ensure all your simulation parameters and the mesh are in SI units, as many CAD software work with mm instead of m.

[Yuan Huang]

thanks very much!
    P0 was set in Pascal, I write p0=1atm just in order to facilitate.
    I am trying to calculate the flow of c ombustion chamber with s everal fuel inlet, o xidant inlet and second air inlet, it also have an outlet.
    I set relaxation coeffs to 0.05, the result is also divergence.
    But when I set p0=33atm, rom0=1.17kg/m3,  the result is not divergence.

[Alexandre Douce]

Are you steady or unsteady?
 
Could you upload your set up and perhaps your mesh if it is not confidential and not too heavy (or use files sharing web site)?

[Yuan Huang]

 It's steady. 
 I already have tried to run the simulation as laminar,  change IMRGRA, but the result is still divergence.

[César Vecchio]

Try changing the schemes to Upwind instead of centered and uncheck Flow Reconstruction, that should improve stability. Also, do you get any previous warnings such as "mesh too/not enough refined near wall"? Which is your turbulence model and which is your y+? Can you also check your mesh has elements changing size smoothly? There was another thread about a blade simulation and divergence was being caused by abrupt element size changes.
It can also be that your simulation with the bigger Rho has strong unsteady behavior not able to be captured by a steady algorithm, like local flow separation and reattachment, but that would be the last thing to check. How many steps does your calculation have and at which step does divergence occur?

[Yuan Huang]

 the schemes is Upwind, there is warning "mesh too/not enough refined near wall" . Details see attchment.

[César Vecchio]

I don't see anything wrong in your xml file setup. The warning you get is basicaly saying your first element on the wall must be smaller in order to capture the laminar sublayer. Since your simulation is all laminar acording to the xml, the y+ quantity has to be around 10.88 or less. I understand Saturne asks for a y+=2.83 but you can do with more. You can use the following y+ calculator to determine how small your mesh must be near the wall:
http://geolab.larc.nasa.gov/APPS/YPlus/
Hint: view the page source and you can see the calculation code in Javascript ;)
Can you show us a screenshot of your mesh? It's just to see how the element sizes are distributed, I suspect the problem is there. Another question, the results you get with Rho=1.17kg/m^3, do they seem physically possible?
About your listing file, it has only 19 steps and seems to be the file in your tmp directory, it doesn't show neither divergence so far nor a finished run. Can you upload the listing.[date] file you may have in your RESU directory? That one has the relevant information.

[Yuan Huang]

thanks  

I can't show the mesh as it is confidential.  The result is divergence at about 25 steps.

Even if at 19th step, the value of pressure and velocity are very large.

--------------------------------------------------------------
Variable      Min. value    Max. value   Min clip   Max clip ---------------------------------------------------------------
v  Pressure    -0.40930E+18   0.23895E+18         --         --
v  VelocityX   -0.46744E+11   0.59522E+14         --         --
v  VelocityY   -0.46210E+13   0.19242E+14         --         --
v  VelocityZ   -0.16260E+13   0.75625E+13         --         --
v  total_pressu-0.40930E+18   0.23895E+18          0          0