[raptorishungry]

Hello Foamers!

I am trying to validate OpenFOAM's Brownian motion force. For this purpose, I am following the process highlighted in this link .

https://bugs.openfoam.org/view.php?id=2153

I understand that the source code is updated by henry. But I am having trouble validating it, I tried all possible combinations (changing particle density, carrier fluid, mean free path, temperature of the fluid/particle) but unfortunately not able to pull it off. Also I do not understand what henry means in the last comment of this post

https://bugs.openfoam.org/view.php?id=2153

where he says "RMS of the magnitude of the displacement in the spherical distribution.", as of now I think solving that ambiguity might help me get closer to validating it.

Could someone please shed some light on this?

Thanks,
adithya

[raptorishungry]

Hey all, I found out the solution to this after performing quite a lot of computational runs and varying constants here and there to narrow down on the discrepancy. The results will not match if you follow this thread

https://bugs.openfoam.org/view.php?id=2153

This is because the sphereDrag in OpenFOAM needs to be corrected by the Cunningham slip correction factor. I modified OpenFOAM's sphereDrag to a new force following the formulation in ANSYS for Stokes-Cunningham drag

https://www.sharcnet.ca/Software/Ansys/16.2.3/en-us/help/flu_th/x1-62400016.7.3.html

I also modified the BrownianMotionFoce in OpenFOAM according to Omid et al.

https://www.tandfonline.com/doi/pdf/10.1080/02786820802587888

Making these changes validated Brownian motion force in OpenFOAM.

[jafarbakhshi1374]

hello dear
Adithya Gurumurthy
i have the same problem and my results have about 50 precent deviation by exact answere (sqrt(2Dt)) so would you please send the changes y have done to validate this force?

[xuegy]

Hello Adithya

Could you please share your modification? I also have problem validating the Brownian motion force. Seems like the code from OpenFOAM is always smaller by sqrt(2).

[raptorishungry]

Hello jafarbakshi and xuegy, sorry for the late reply. I have attached my files below.

[xuegy]

Originally Posted by raptorishungry

Hello jafarbakshi and xuegy, sorry for the late reply. I have attached my files below.

I just tried your code. The mathematics equation are basically the same as OpenFOAM cubic distribution(comment in the source code). It has the same problem verifying the Brownian motion. It's always diffusing sqrt(2) times slower than the continuum model.

[raptorishungry]

It worked for me. What are you validating your model with? could you post your case set-up/plots?

[xuegy]

Originally Posted by raptorishungry

It worked for me. What are you validating your model with? could you post your case set-up/plots?

I'm trying the easiest 1D case. A 2um*2um*20um box(1*1*100 mesh) is filled with water. Half of the box has 100nm spherical particles initially (dimensionless concentration "2") while another half is empty (0). The continuum model solves the diffusion equation while the discrete model simulates Brownian motion and converts the Lagrangian field to the void fraction.

My results shows that the discrete model is always slower by sqrt(2), e.g. the result of discrete model at t=11.3 s converges to the result of continuum model at t=16 s.

0.0.png

Please correct me if I make any mistake!