SimScale CAE Forum

Manifold to Vent CFD Error


I am working on a venting system for an arrangement of 12 CO2 cylinders (each pressurized to 4500psi).
I am having trouble getting meaningful/sensible results from my CFD simulation. CO2%20Vent%20system This picture illustrates the system I am working on.

However I’ve simplified the CAD to this (in order to save on meshing time): Simplified%20CO2%20vent%20system

Once the CO2 cylinders are activated, they vent off through the big pipe (vertical pipe) at the top of the CAD.

My setup for the CFD goes as follows:
-Simulation: Incompressible Flow (although CO2 is compressible, I chose incompressible as a first iteration to get an idea of the results)
-Mesh all good
-Material: CO2
-Boundary Conditions: Pressure Inlet at the 12 inlets of the CO2 cylinders (set to fixed value 4500psi)
Pressure Outlet at the top (vent) pipe with fixed value of 14.7psi (atmospheric)
-Run Simulation

This is where I end up with absurdly large values for pressure and velocity. Any help on this simulation would be appreciated. (I suspect my issue are the boundary conditions)


Hi @bseech!

Thanks for your question. In order to completely answer your question, please consider this post by our PowerUser @roy_g : Guideline to asking for help with a project - thanks a lot!


1 Like
  1. It would be very helpful to get a link to your project

  2. A pressure regulator (which you do not have) limits the amount of gas exiting the cylinder. If you set these pipes, which look to be of large diameter, to have a pressure difference of 4500 psi (less 14.7) then I would fully expect the velocity to be extremity high,

  3. Since there is a large pressure drop and a high mach number, you most certainly need to use the compressible solver

  4. why would you want to simulate these boundary conditions of a 4500 psi inlet and atmospheric outlet?

Link to my project:

As for the pressure regulator, I am unsure if there is one. But from my understanding, once the system is triggered, it is suppose to rapidly fire the CO2 to the test cell. Therefore, I would assume that there isn’t a pressure regulator. (I may be completely wrong about this but this is my initial hunch)

The configuration/CAD showed in the link I sent you illustrate the scenario where the CO2 is directed to a vent outside of the building. When the testers are in the test cell we do not want an accidental discharge of lethal amounts of CO2, this is why I am setting up a venting system. Hence, if there is an accidental discharge, all the CO2 would be directed outside of the building. Therefore, the reason of having 4500 psi as inlet conditions is to represent the pressurized cylinders of CO2 and the reason to have atmospheric outlet is to represent the vent pointing towards the atmosphere outside of the building.

Thank you for the link, I see that you have switched to compressible solver as I suggested. I’ve just looked at your mesh log and noticed you have a lot of mesh elements with errors. I’m not sure how much of a problem it is but you can try going to mesh —> refinements --> inflate boundary layer and choosing all surfaces. This enables you to make the external layer finer without making everything finer (which would make your simulation heavy). From a mesh point of view it would be better to change all the rounded edges of your model to sharp corners (you would require less mesh elements and have less errors), this would however affect your results to some extent.

As for the error message you are getting regarding “Maximum number of iterations exceeded”. It’s not really surprising. It means that your simulation is not converging fast enough. In fact, look at your convergence plot, there is no convergence! Truth is that supersonic simulations are tricky to keep stable and I am not expert on them! You can read the link below for some information that might help you.

Under simulation control you have used the default of start at 0, end at 1,000 and dt of 1. Given that the phenomenon will reach steady state much faster, I would set it to start 0, end 2 and dt of 0.002 (but use the write interval setting to get some intermediate results so you can see it has reached steady state). Write interval just means how often do you want to save the result to memory.

May I ask what you are hoping to learn from the simulation?