CFD Online Discussion Forums

CFD Online Discussion Forums (
-   OpenFOAM (
-   -   DSMC: simulating flow into vacuum (

Brinx October 15, 2010 07:43

DSMC: simulating flow into vacuum
Hello all,

I've been trying to use dsmcFoam to simulate the flow of gas from a duct into a vacuum, in order to study the free expansion plume. The situation is axisymmetric (the gas emerges from a circular opening), so I've defined a wedge-shaped region on one side of the symmetry axis to conserve computational resources (as is done in the incompressible->pimpleDyMFoam->movingCone FOAM tutorial).

After reading several threads on this forum about other people trying similar things, I've managed to make one particle emitter at the duct opening, and I've managed to make the other domain boundaries act as particle sinks by setting their temperature to a value close to zero.

However, I encountered a problem: I couldn't find any way to set the initial conditions to specify a vacuum. Regardless of how I defined the RhoN and RhoM fields (type calculated, value 0 seemed a sensible way to do it), the domain was always seeded with about 20,000 particles at the start of the simulation. I tried to resolve this by simply editing the files holding particle information at zero time (in the folder "0/lagrangian/dsmc/" ) and removing all particle entries by hand.

This indeed made the domain empty at t=0, and running the simulation shows particles being inserted from the emitter into the domain. However, the rate at which particles are introduced is very low indeed: about one particle per ten time steps. This doesn't seem to agree with the densities I have set, and it also results in particles simply following straight trajectories because they never get close enough to interact with one another.

As a free expansion plume should in principle show particles interacting and performing work in expanding the flow longitudinally, I'm wondering how I should change the settings to allow this to happen. Below I have included a screenshot of the result so far (after a short run), the top edge of the domain is the symmetry axis. The plotted quantity is the kinetic energy density (linearKEMean) divided by the density (rhoNMean). The thickness of the wedge (which is not visible in the picture) increases from zero at the top to about 1 cm at the bottom. The width of the domain is 2 metres, and the number of cells is 200 (x) by 100 (y) by 1 (z). In the picture you can see the effects of particles that were in the domain before the jet particles came through: straight lines influencing the average kinetic energy per particle in various cells.

So, any ideas on what I should change to make the emitter emit more particles per unit of time? Maybe change the number of molecules represented by one DSMC particle?

jiaoxiaolei April 18, 2013 04:22

HI,Brinx,have you succeed on this simulate?

All times are GMT -4. The time now is 07:27.