Modeling a Gas Valve desorbing into a vacuum chamber

Hello there,

I am working with Molflow v2.8.9 and am trying to model a gas valve injecting a gas puff into a vacuum test chamber. The main parts of the geometry are the gas fill chamber, the valve throat, and the test chamber. The setup is as follows.

The valve gas chamber fills up until the pressure inside reaches 11,000 mbar. Upon reaching this pressure, gas stops being injected and the ‘valve’ opens (A facet which opacity gets set to 0 at the exact moment the desired pressure is reached.). The valve stays open for 0.0015 seconds and then closes. The gas travels down the throat and is injected into the large vacuum chamber. The simulation stops at 0.1 seconds.

I have 3 ‘probe’ facets that track the pressure, one in the valve gas chamber, one the length of the valve throat, and one just inside the vacuum chamber.

The problem I am having is that the pressure in the valve gas chamber builds up to the desired pressure (the red dots), but as the valve opens and gas starts to travel down the throat (the blue dots), the pressure in the gas chamber stays constant. One would expect the pressure to start rapidly dropping during the time the valve is open but that behavior is not seen in the plots (attached below).

I feel like I might be setting up the problem incorrectly but I’m not sure. I’ll be including the .zip as well so it can be looked at.

Valve_analysis_with_gate.zip (2.9 MB)

Thanks for the help,

Skyler

Hello Skyler, very nice use of the time-dependent mode!

Before explaining the behavior you described, let’s fix together a few small issues:

  1. I recommend to set the time window from 1E-7s to 1E-5s, to correspond to the moment spacing. That way every hit is counted towards one moment, and your simulation will have 100x statistics.

  2. There’s a geometry error: the valve is overlapping the gas chamber wall (demonstrated by the two selected facets):


This is an issue, because if two facets are exactly in the same plane, Molflow’s ray tracing will choose one of the two in an undefined (random) manner. When the valve is clsed, this doesn’t matter, but when it’s open, it’s undefined whether 0 or 1 opacity is applied.

I redefine the valve a bit smaller:

Now I run the simulation. While in your screenshot it looks like there is no pressure drop at all, in linear scale we can see a tiny drop in the gas chamber’s pressure:


(The colors are the same as yours).

Simply, the throttle is too long and narrow (in vacuum terms, too conductance-limited) for the gas chamber to deplete. It loses a bit of gas (more noticeable during the emptying), then reequalizes to a medium pressure. Zooming in to the critical part:

image

You can observe what you expect if you increase the emptying duration (from 6ms to 20ms):


also visible in LOG scale:

Or better yet, decrease the throttle length (or increase the throttle diameter).

You can do a qualitative error checking on my last plot.

The gas chamber is 0.6340.76^2PI=1.15cm3
Its pressure (red line) dropped from 11000mbar to 7500mbar (it’s easier to read on the LIN scale image).
So we lost around 1.15*3500=4000 mbar.cm3 of gas.

The large test chamber is a 21x21x21cm cube (minus the throttle volume which we omit) = 9200cm3
Its pressure (green line) went from 0 to 0.37mbar.
So it gained around 9200*0.37= 3402mbar.cm3 of gas.

You can see that the gas quantities are correct: of the 4000mbar.cm3 lost from the gas chamber, the majority ended up in the test chamber, the remaining ~600mbar.cm3 is due to rounding errors and the gas still in the throttle.

I attach my simulation with the longer valve opening and the fixed gemetry.

Valve_analysis_marton_longeropening.zip (2.9 MB)

Cheers, Marton

Marton, thank you very much for the help and insight. I know this isn’t exactly a pumping problem per say, but is there some reason that the simulation in MolFlow is not behaving like I am experiencing in the experiments? In our experiments we are puffing gas at 150 psig into a volume at 1E-5 mbar. The valve stem has a travel time from full open to full closed of 2E-3 seconds roughly and depletes approximately all of the gas stored inside of the valve.

In a vacuum sense the throat is conductance limited as you said, but we are seeing a mass flow rate possible to desorb all of the gas. Maybe the problem i’m trying to solve isn’t best suited for molflow, though I might just be doing something wrong. I’m going to be going over my assumptions and setup again just to make sure but would appreciate any insight their might be.

Thank you again,

Skyler

Sorry for the slow reply.

I wanted to say earlier that your outgassing is way above high vacuum, but I assumed that they are just arbirtary units.

I don’t know what “PSIG” means, but if it’s 150PSI, it’s not molecular flow. For general geometries, molecular conditions hold below 1E-3mbar (0.014psi). Molflow doesn’t analyze your input data, but always assumes molecular flow, which I believe is not the case.

If you’re at a higher pressure, then probably a DSMC code or a fluid dynamics solver is better suited.
Sorry that I can’t give more insight,

Marton

Hi Marton,

No worries, I know life gets busy.

PSIG is an english unit for gauge pressure, thats what we use to specify the pressure in the valve for our experiments, basically the Pressure of the gas in PSI + 1 atmosphere.

The simulation not being molecular flow is kind of what I assumed, but I wasn’t sure if Molflow was able to calculate characteristics for gas regimes above Molecular. I have access to some fluid dynamic solvers, so I can use them for this particular problem. Thank you for all the help.

Skyler