[Ghost_]

Hello everyone,

I am quite new here but i have been reading you a lot. I have a problem... I would like a huge pressure ratio between two reservoirs using the DSMC method and DS2V Bird's program. I attached the geometry i want to use on the file below. My problem is that no matter how many molecules i put i always get the message "The number of molecules is insufficient for a good calculation". Even when the simulated molecules are about 80.000.000. Can you please help me out...??

Please note that you have to convert the .doc file to .dat

Thank you

[flotus1]

I have never used the program, but I read the documentation in Birds book while programming my own DSMC solver.

So here is what I think might be causing the trouble you have.
As far as I know, the DS2V program automatically sets up the collision cells according to the mean free path of the molecules. The background is that these cells must be smaller than a mean free path in order to perform a valid DSMC simulation.
On the other hand, there must be enough molecules to populate every cell with a certain amount of molecules, typically in the order of 10.
Otherwise there are not enough molecules to perform collisions.
So I can only assume that the Knudsen number in your simulation is very low, leading to a very high number of collision cells and an even higher number of molecules necessary to populate them.

[Ghost_]

Thank you for your answer flotus but the Knudsen number of the simulation is 0.01 which is supposed to be OK for a DSMC simulation. By the way does anyone know any open-source codes which can do 2d axisymetric DSMC...??

Thank you

[flotus1]

So your Knudsen number is 0.01. This is rather at the lower end for DSMC calculations, especially for low mach number flows. We still dont know if this is at the inlet or at the outlet, but lets just assume for a moment that it is constant.
Lets estimate that the cell size is 4 times smaller than the mean free path which is recommended when there are no additional measures for nearest neighbor collisions.

So we arrive at 400 collision cells over the cross-section of the channel.
I dont know the length of the channel, let me estimate this as 100 times the height. Correct me if I am wrong.
That makes for a total of 1.6e7 collision cells which should be populated by approximately 1.6e8 molecules. Thats a lot.

The way I understood it the DS2V program is capable of simulating axisymmetric flows.
That is where those radial weighting factors come in.

[Ghost_]

Actually i want to study a pressure driven flow between two reservoirs. One is at a pressure of 5000Pa and the other is at 3Pa. They are connected through an orifice which is 0.75mm in height and about 2mm in length. What i want to study is the underexpanded jet that will be created from the flow moving from the high pressure to the lower pressure. The reservoirs are not big either. The "high pressure" reservoir is 3mm in length and 12.5mm in height and the "lower pressure" reservoir is 50mm in length and 12.5 mm in height.

[flotus1]

Are you sure about your estimation for the Knudsen number?
I think you should estimate the mean free path in the high pressure reservoir again and see how many collision cells and molecules you really need to perform the simulation.
I feel like this is going to be a computationally expensive simulation.

[Ghost_]

Thank you for your replies... Does your solver solve 2d axisymetric flows..? I bet too that it is going to be a very intensive calculation..!

[flotus1]

My solver can only handle 2d planar flows.
In the beginning, I also intended to use one of the free codes available. But as I tried to use them, I realized that I did not know enough about DSMC to get reliable simulation results.
So the main reason I wrote it was because I thought it was the best way to learn how DSMC works.

[Ghost_]

That's really great...! Is it possible to send me the solver in order to study it..?

Thank you

[flotus1]

I cant do that, sorry.

What I found helpful while getting started were the publications of Garcia, especially this one: http://www.algarcia.org/Pubs/DSMC97.pdf
And of course both books from Bird. The second one published in 2013 may have its flaws, but it also comes with some detailed descriptions for the actual implementation of the method like sampling quantities that are hard to find anywhere else.

[Ghost_]

Thank you very much for your references...!

[Vigneshramaero]

Hi Flotus1,
I'm also writing my own dsmc solver, to solve for any arbitrary body (2D). I have some difficulty in specifying pressure boundaries. Do you use pressure boundaries or number densities specifying at both inlet and outlet. Can you give me some brief about this.

Thank you

Vignesh