Body

Hello!

I am currently working on a simulation that consists of several perforated, quadratic layers stacked on top of each other. All layers exhibit outgassing properties. Around these layers, there is a cuboid which absorbs all particles which eventually make it from the inner zones of the layers through the holes to the outside. Due to the perforation (quadratic holes), the facet count is rather high, causing increased simulation time. However, it can be assumed that the whole setup is geometrically symmetrical, so the team has decided that we shall only simulate one quarter of the actual domain, using fully opaque and reflecting surfaces in the symmetry planes.

The quantity of interest is the pressure distribution in the cross section (between the layers as well as across the layers). In order to observe this, we have decided to use the reflecting planes (the ones which are installed to simulate symmetry) as counting layers with a texture resolution fine enough to depict the pressure between the layers. In addition, one fully transparent counting layer was installed diagonally.

After a few days of simulation, the diagonal layer has shown intuitively adequate results. However the values of the reflecting counting layers are lower than the ones of the diagonal layer by a factor of two (estimated by texture colour). Intuitively, the setup should be more or less radially consistent (circular lines of equal pressure). The first approach to determine which result reflects reality was to insert a counting facet parallel to the layers between two layers. The values on this counting layer have shown consistency with the diagonal layer as well as with the expected circular symmetry. Implicitly, the pressure values calculated by the reflecting facets (normal to the layers) do not comply with the pressure values of the parallel, transparent, layer-like counting facet.

My question is why this discrepancy appears. I should mention that all facets are two-sided and diffuse reflection is used for all reflecting facets.

I have an idea what might be causing the problem, but I am not sure if it is correct and I do not want to put false ideas in the readers head.

Thank you in advance for your help!

Submitted by Marton Ady 5 years ago

Hello Martin,

I admit that without a picture I have a hard time to imagine the setup (for example, I don't know what quadratic layers mean). A few general hints, if they don't solve the issue consider whether you can share the geometry so I have a look.

- A factor of 2 seems to point at a 2-sidedness issue. If your surface is getting hits from only one side, it should be 1-sided, its normal towards the simulated particles. In contrast, transparent "diagnostic" facets in the volume should be 2-sided to collect statistics from both sides.

- A surface representing a symmetry plane should have fully specular reflection (not diffuse), to act like a mirror.

- Pressure is vectorial: check if the factor of 2 also appears if you change the texture mode to impingement rate (which only takes the hit count density into account).

(- Also, based on the screenshot you uploaded, I think the mesh resolution is way too big (larger cells would get more hits and have less scattering).)

Let's see if these first rounds of hits help, if not, share more details, or if possible the geometry.

Good luck, Marton

Submitted by b793e221da1d98026279 5 years ago

Hello Marton,

thank you for your quick reply. The image I have linked to in the initial post is one of those quadratic layers, viewed from the top. More specifically, it is one quarter of such a layer. The high pressure in the bottom right corner indicates the centre of the full setup.

My thought was that the two-sidedness might be the root cause of the issue, but I do not understand the calculation well enough to know for sure. Thank you for your advice, I will implement this shortly and see if the values match up then.

From this I gather that you also think that the values in the diagonal should be trustworthy for further considerations, as they match with the ones from the large counting layer, correct? (I understand if you do not want to make any statements based on the little information I have provided.)

Just out of curiosity - should it not be irrelevant whether the symmetry plane is specular or diffuse? In both cases the number of particles flying towards the plane are equal to the number coming from the plane.

Thank you for your help.

Submitted by Marton Ady 5 years ago

I'm almost sure, based on your description, is that you have a 2-sided facet, which only receives hits from one side, therefore its texture averages the actual pressure with the 0 pressure from the other side, resulting in half the displayed pressure.

I can't say for sure whether the diagonal values are correct, but if it's a facet within the volume, transparent and two-sided, then indeed it displays correct values.

Since you encompass your geometry with a perfectly sticking outer boundary, if your counting facets are near those "perfect absorbers", then no hit will come from that direction, resulting in a physical decrease of the pressure.

It is important that in ultra high vacuum pressure is vectorial: next to a perfect pump, for example, the pressure would be zero on a surface facing the pump, and non-zero on a surface showing its back to the pump - while exactly at the same location! The density, however, is scalar: it is exactly the same regardless of facet orientation. You can check this in Molflow: checking texture mode to Density, see if the factor of 2 difference still exists.

And no, specular or diffuse reflection doesn't give the same result (unless the system is fully isotropic, which is not the case in your source/sink setup): the pressure is a mechanical quantity caused by the impulse change of the particles when hitting the surface. This is taken into account by Molflow (the relation between number of hits and pressure is linear only if the system is isotropic). Therefore if you change the direction of the particle at a rebound, depending on the original direction, the impulse change might be smaller or larger depending on the original direction.

Submitted by b793e221da1d98026279 5 years ago

The issue with double the pressure seems to be resolved. Thank you very much for your help and also for the other tips.

I do have one further question but it is not related to this topic (it is about reaching a state where the pressure no longer changes in a non time-dependent setup). Should I create a new post for that?

Submitted by Marton Ady 5 years ago

Oh, I'm sorry Martin, I've overlooked this comment.

If the issue is still actual, feel free to open a new thread, I'll reply as soon as possible.