Electrical current boundary condition in ANSYS for thermal model.

[dbecker]

Hello,


I currently use ANSYS thermal to predict metal temperatures of electrical components. Some of you probably already know the extent by which I went to calculate heat generation of these electrical components.

I would like to be able to apply an electrical current and have ansys compute body wattage dissipation for me at both bulk materials AND contacts (like bolted electrical joints).

Is it as easy as just applying electrical current load to a face? I do not wish to apply a voltage because I know the current offhand. I am also doing this in AC, can ANSYS handle AC current and heating? Is there a directional input to this current BC or does ansys compute current flow automatically?

Thanks,


- D

 

[kan123]

Start with the Help documentation for LINK-68, the simplest element with thermal-electric coupling. Resistivity is one of the material properties. You can specify voltage at nodes with the D command or impose currents with the F command. I've done Joule heating problems with DC. I haven't had to investigate AC, so I'm not sure if you would just use the RMS current or have to do something else.

 

[dbecker]

I just tried applying current to nodes. I dont understand why it only allows you to apply to nodes and not areas. Anyway, it's a simple metal rod with convection cooling and current applied on one end. It did not converge, I dont know why.

I am using SOLID69 thermal electric elements.

 

[kan123]

I assume that you have set a voltage restraint somewhere -- otherwise you would have the equivalent of a rigid body motion problem. If your resistivity is temperature-dependent, then you need an iterative solution. If the applied current is too large, the resultant temperature swings might be too large for ANSYS. If that's the case, try ramping the load gradually. If your steady-state solution is still difficult to achieve, you might find that running it as a transient will allow you to achieve convergence.

 

[dbecker]

I did not set a voltage restraint somewhere. May I ask why I should set a voltage if the current is pre-specified?

I guess I am not understand what ANSYS needs.

Are SOLID69 elements sufficient if i meshed that volume and applied current to nodes? Do I have to use LINK68 elements? I read about them and they are 2 node elements with DOF and need to be applied at centerlines of geometry, but I suppose I dont need them if I am using SOLID69s and I correct?

Thanks for your help

 

[kan123]

Solid elements are fine if that is a better representation of your geometry. For the electric portion of the solution, the degrees-of-freedom are voltages and the load vector consists of the applied currents. At least one DOF has to be specified or they could all float to any value. This is analagous to requiring a displacement restraint when you apply a force to a structural model.

 

[dbecker]

Ahh I see.

So at least one voltage restraint should be specified. Well than if my customer is telling me we are running at 1500 amps, what should i put for the voltage? I mean it really could be anything (within reason), as long as the voltage difference produces the current i specify. And the voltage difference will be computed I guess, by ANSYS.

I will try this approach thank you very much for your time.

 

[dbecker]

Hello again kan123,

I just re-ran my metal rod model and it did indeed converge. Although this is a good step towards understanding thermal electric, the answer dosent really make sense to me.

Heres what I did,

I have a metal rod 26mm in diameter that is 500 mm long. A quick tet mesh with SOLID69 thermal electrics.

I applied current TO ALL NODES, 800 amps. I applied some convection value on the rod outside surfaces for heat balance (i didnt apply conv on the ends).

I applied a voltage of zero on one end and the other end with no BC (I let that potential fall out through ANSYS).

The solution converged but I expected the temperature profile of the rod to be axially uniform and radially varying (getting hot from the inside out, with lowest temp at the surface due to convection).

This was not the case, the rod showed axial temperature gradient which means there is more heat being generated at one end as opposed to the other. I do not understand why because the current applied is the same at all nodes and ref voltage is zero at one end.

Please let me know if you have any ideas to further clarify.

Thanks

 

[kan123]

Interesting result. Your reasoning seems sound -- I would expect the same thing. Other than verifying your current and heat balances at the restrained DOF, I don't have any good suggestions. (If I had more time, I'd try my own sample problem.) At this point, it might be best to involve the ANSYS Support Desk to help debug your problem.

One other suggestion to minimize end effects (but not likely to be the problem in this case because of large L/d). When you apply the current at one end, put the entire current into one node, then couple the Volt DOF of all the other nodes at the end to that Master. This should give you a uniform current density at the end even when your mesh is irregular.