Note: This discussion is about an older version of the COMSOL Multiphysics® software. The information provided may be out of date.
Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.
Continuity on internal boundaries between domains with different physics
Posted Feb 19, 2012, 10:09 p.m. EST Geometry, Modeling Tools & Definitions, Parameters, Variables, & Functions Version 4.2 1 Reply
Please login with a confirmed email address before reporting spam
Hi,
I am trying to figure out how "continuity on internal boundaries" works in Comsol.
What I have is a 1D model (I reduced a complex model as much as possible while still retaining the issue of interest) with:
• Two domains, which are intervals side by side. Finalized with "Form Union".
• Two separate physics interfaces, both of type "Electric Currents". One is defined on the left domain, with the dependent variable Ve, and the other on the right domain, with dependent variable Vi. (Which stand for external and internal voltage.)
(The reason is that I am interested in modeling some nonlinear effects on the boundary, which is the point of the study. But first I want to figure out this basic example.)
I've defined the boundary conditions as follows, to couple the two:
• On the left, a potential of 9V.
• On the right, a potential of 0V.
• On the right of the Ve domain (i.e. on the boundary between the intervals), a Normal Current Density, which is defined as the negative of the normal current density of the Vi domain at that point. This is to ensure current continuity across this boundary.
• On the left of the Vi domain (i.e. on the boundary between the intervals), an Electric Potential set to equal to Ve at that point. This is to ensure the continuity of voltage across the boundary.
So what I expect is that current will be constant everywhere, and that the voltage will vary linearly from the 9V on the left to the 0V on the right.
However, this is not what happens. The computed solution has 3V on the boundary (not the expected 4.5V), and the current is 6 A/m^2 in the left domain, and 3 A/m^2 in the right domain (again, not the expected 4.5 A/m^2 everywhere).
I've plotted various values, and the basic discrepancy seems to be that the left physics' nJ (inward current density) is -3 A/m^2, while its Jx (current density, x component) is 6 A/m^2 at the same point. This I can't understand at all.
In general, I'm guessing that Comsol automatically tries to ensure continuity between the two distinct dependent variables Ve and Vi, and then treats the Normal Current Density BC *in addition* to that. For example, I can disable "Normal Current Density", so that "Electric Insulation" takes over in the left domain, and then it gives me the solution that I'm looking for. Clearly this is not what we understand by electric insulation. I've checked "Equation View" for the "Normal Current Density", and it says in "Details" that it is a "+ operation" which might be a clue. But there is no way to change that. I've looked though "Equation View" of "Current Conservation" of the left physics interface and there are no references there to the other physics interface in the other domain. As far as I can see, there is no evidence of the observed continuity in any of the equation views, so it is somehow hidden.
What I expected is that each physics will be quite independent: it would take its BCs and its PDE and solve it in its domains. But it seems that's not the case: it also interacts with its neighbouring physics and doesn't let me see the details of that.
I also thought that "continuity in the physics fields across interior boundaries" mentioned in the "Finalizing the Geometry" section in the User's Guide meant interior boundaries between the domains covered by the same physics. (The middle point between two intervals would then be the exterior boundary of each physics, because each is defined in one domain only.) But this example seems to indicate that the "interior boundaries" referred to the geometry.
There are various clues in other posts I've read, for example:
www.comsol.com/community/forums/general/thread/12992/
www.comsol.com/community/forums/general/thread/9617/
But no definitive answers.
What are your thoughts?
The model is attached. (Comsol 4.2.0.228)
Evgeni
I am trying to figure out how "continuity on internal boundaries" works in Comsol.
What I have is a 1D model (I reduced a complex model as much as possible while still retaining the issue of interest) with:
• Two domains, which are intervals side by side. Finalized with "Form Union".
• Two separate physics interfaces, both of type "Electric Currents". One is defined on the left domain, with the dependent variable Ve, and the other on the right domain, with dependent variable Vi. (Which stand for external and internal voltage.)
(The reason is that I am interested in modeling some nonlinear effects on the boundary, which is the point of the study. But first I want to figure out this basic example.)
I've defined the boundary conditions as follows, to couple the two:
• On the left, a potential of 9V.
• On the right, a potential of 0V.
• On the right of the Ve domain (i.e. on the boundary between the intervals), a Normal Current Density, which is defined as the negative of the normal current density of the Vi domain at that point. This is to ensure current continuity across this boundary.
• On the left of the Vi domain (i.e. on the boundary between the intervals), an Electric Potential set to equal to Ve at that point. This is to ensure the continuity of voltage across the boundary.
So what I expect is that current will be constant everywhere, and that the voltage will vary linearly from the 9V on the left to the 0V on the right.
However, this is not what happens. The computed solution has 3V on the boundary (not the expected 4.5V), and the current is 6 A/m^2 in the left domain, and 3 A/m^2 in the right domain (again, not the expected 4.5 A/m^2 everywhere).
I've plotted various values, and the basic discrepancy seems to be that the left physics' nJ (inward current density) is -3 A/m^2, while its Jx (current density, x component) is 6 A/m^2 at the same point. This I can't understand at all.
In general, I'm guessing that Comsol automatically tries to ensure continuity between the two distinct dependent variables Ve and Vi, and then treats the Normal Current Density BC *in addition* to that. For example, I can disable "Normal Current Density", so that "Electric Insulation" takes over in the left domain, and then it gives me the solution that I'm looking for. Clearly this is not what we understand by electric insulation. I've checked "Equation View" for the "Normal Current Density", and it says in "Details" that it is a "+ operation" which might be a clue. But there is no way to change that. I've looked though "Equation View" of "Current Conservation" of the left physics interface and there are no references there to the other physics interface in the other domain. As far as I can see, there is no evidence of the observed continuity in any of the equation views, so it is somehow hidden.
What I expected is that each physics will be quite independent: it would take its BCs and its PDE and solve it in its domains. But it seems that's not the case: it also interacts with its neighbouring physics and doesn't let me see the details of that.
I also thought that "continuity in the physics fields across interior boundaries" mentioned in the "Finalizing the Geometry" section in the User's Guide meant interior boundaries between the domains covered by the same physics. (The middle point between two intervals would then be the exterior boundary of each physics, because each is defined in one domain only.) But this example seems to indicate that the "interior boundaries" referred to the geometry.
There are various clues in other posts I've read, for example:
www.comsol.com/community/forums/general/thread/12992/
www.comsol.com/community/forums/general/thread/9617/
But no definitive answers.
What are your thoughts?
The model is attached. (Comsol 4.2.0.228)
Evgeni
Attachments:
1 Reply Last Post Feb 19, 2012, 11:55 p.m. EST
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
More experimentation. Now instead of using "Electric Currents", I've used two Weak form PDE interfaces. Everything is set up similarly to the example above. The same results are still observed...
So there is no magic related to the fact that both of the physics interfaces on the two domains were "Electric Currents" (I thought that maybe Comsol inferred the dependent variables to be voltage, and their gradients directly related to currents, so that's why it linked them.) Now Comsol just sees two PDEs with different dependent variables, so it has no way of knowing what I'm trying to model: one could be voltage, the other one could be something else.
But it still enforces continuity of the dependent variable across the boundaries between physics it seems.
EDIT: Update: The problems I mention are resolved in this thread:
www.comsol.com/community/forums/general/thread/26354/
So there is no magic related to the fact that both of the physics interfaces on the two domains were "Electric Currents" (I thought that maybe Comsol inferred the dependent variables to be voltage, and their gradients directly related to currents, so that's why it linked them.) Now Comsol just sees two PDEs with different dependent variables, so it has no way of knowing what I'm trying to model: one could be voltage, the other one could be something else.
But it still enforces continuity of the dependent variable across the boundaries between physics it seems.
EDIT: Update: The problems I mention are resolved in this thread:
www.comsol.com/community/forums/general/thread/26354/