the diffusion term of PDE

TBE_Legend

Maybe you should keep the diffusive component, even though it is of small value, actually, if the diffusive componet is zero, in this condition, the convective flux boundary condition becomes the same with insulation boundary condition. I have read some information that, some times, in order to achieve a numercial convengence, even some artificial diffusion terms are added, if so, why not consider the intrinsic diffusion? From literature, I know there is a relationship between diffusion coefficient and mobility, which is called Einstein relation, and I think it is helpful for you to set the diffusion coefficient.

I am not confident enough to convince my opinion, but I think you can try.

Best wishes.


From: "calvinkid8668@ymail.com" <calvinkid8668@ymail.com>
To: COMSOL_Users@yahoogroups.com
Sent: Wednesday, 17 June, 2009 22:02:58
Subject: [COMSOL_Users] EHD from wire to plate: charge transport boundary conditions

Hi,
I am trying to model a corona discharge from wire to plate. (2D model)

I am using the electrostatics and PDE (general mode)

I am just trying to set up the boundary conditions on the PDE eq (for charge transport). Can someone help me out here? I want a constant surface charge on the wire (a 1mm circle) and zero diffusive flux on all other surfaces.

The results I am getting are quite absurd. By physics, the diffusion term in the equation for charge transport should be negligible. But it seems to affect my solution a lot. Besides, if i leave out that term, I am getting an error that has something to do with a singular matrix (which I think is understandable considering that I give zero diffusive flux on the boundaries which would lead to a zero matrix)

Maybe you should keep the diffusive component, even though it is of small value, actually, if the diffusive componet is zero, in this condition, the convective flux boundary condition becomes the same with insulation boundary condition. I have read some information that, some times, in order to achieve a numercial convengence, even some artificial diffusion terms are added, if so, why not consider the intrinsic diffusion? From literature, I know there is a relationship between diffusion coefficient and mobility, which is called Einstein relation, and I think it is helpful for you to set the diffusion coefficient.

I am not confident enough to convince my opinion, but I think you can try.

Best wishes.


From: "calvinkid8668@ymail.com" <calvinkid8668@ymail.com>
To: COMSOL_Users@yahoogroups.com
Sent: Wednesday, 17 June, 2009 22:02:58
Subject: [COMSOL_Users] EHD from wire to plate: charge transport boundary conditions

Hi,
I am trying to model a corona discharge from wire to plate. (2D model)

I am using the electrostatics and PDE (general mode)

I am just trying to set up the boundary conditions on the PDE eq (for charge transport). Can someone help me out here? I want a constant surface charge on the wire (a 1mm circle) and zero diffusive flux on all other surfaces.

The results I am getting are quite absurd. By physics, the diffusion term in the equation for charge transport should be negligible. But it seems to affect my solution a lot. Besides, if i leave out that term, I am getting an error that has something to do with a singular matrix (which I think is understandable considering that I give zero diffusive flux on the boundaries which would lead to a zero matrix)

ruopiao1009

啊噢,我昨天发在GROUP里的,居然被版主引用了...