Learning Forum

[Mathias Giesbrecht]

I am currently working on a transient thermoelectric simulation for a fuse in a power distribution unit. The simulation ambient temperature is set at 60 degrees Celsius, and the range boundary condition for convection is also set at 60 degrees Celsius. However, during the simulation, I have noticed that the temperature inside the fuse drops below 60 degrees Celsius, even when the current is still flowing through it.

I have attached a photo of the problematic part of the simulation for your reference, and I would appreciate any feedback or recommendations on how to solve this issue. I am open to any insights that could help me understand and address the problem, including possible improvements to my simulation setup or approaches to analyze the simulation results.

[GLUO]

Hi Mathias,

Could you also post this on Mechanical channel? Electronics channel focuses on AEDT problems.

Thanks,

GL

[Bill Bulat]

Hello Mathias,

 

What type of analysis system are you using? Presumably coupled field transient?

Also, what version of Workbench are you using?

We sometimes see non-physical "thermal undershoot" or "spurious oscillations" in the transient thermal response when heat is applied very suddenly. Calculated temperatures in these first order systems can briefly drop below those that are physically possible.

But the behavior you are reporting does not appear to be related to this. It appears that the device temperature is dropping below the 60 C environment very gradually... some 1300 seconds into the transient. It is not immediately obvious to me why this is happening.

It might be helpful to compare the net internal energy (heat in Joules) added over time that is attributable to Joule heat to the net heat leaving the system due to your convection boundary condition(s) and/or any others that you may have. The thermal power in Watts leaving the system is easily obtained with post processing features that are natively supported in the Mechanical GUI. This would need to be integrated over time to get net heat removed from the system.

It is less straightforward to get the net heat added to the system by Joule heat. There may be an easier way to do this, but for now I can only think of using an APDL command object under the Solution branch, something along the lines of the following (untested):

finish

/clear

resume

esel,s,ename,,226,227        ! SELECT COUPLED FIELD ELEMENTS

*get,nelems,elem,,count       ! nelems = # OF SELECTED ELEMENTS

/post26

elm=0          ! INITIALIZE ELEMENT ID#

*do,i,1,nelems ! LOOP OVER SELECTED ELEMENTS

 elm=elnext(elm)   ! elm NOW EQUALS NEXT HIGHEST SELECTED ELEMENT ABOVE elm

 esol,3,elm,,jheat ! READ JOULE HEAT PER VOLUME FOR ELEMENT elm INTO VARIABLE #3

*get,volu_i,elem,elm,volu ! volu_i = ELEMENT elm's VOLUME

 add,2,2,3,,,,,,volu_i            ! ADD VARIABLE #2 AND volu_i*#3 (NET JOULE HEAT FOR ALL ELEMENTS ACCUMULATE IN VARIABLE #2)

*enddo ! END DO LOOP

int1,4,2,1,,Qin ! VARIABLE #4 = INTEGRAL OF VARIABLE #2 WITH RESPECT TO TIME

/show,png

plvar,4   ! PRODUCE AN XY PLOT OF VARIABLE #4

/show,close

 

 

This requires that you set "Save MAPDL db" = "Yes" in Details of Analysis Settings prior to solving.

 

--Bill

 

[Mathias Giesbrecht]

 

Dear Bill,

 

I experimented with different methods, but ultimately settled on thermal electric analysis and added the antype, trans command. I was using version 2022 R1, but recently downloaded the free 2023 version. Unfortunately, my model is too large for the free version, so I am waiting for the teaching version from my university, which should be released next week.

 

During my simulations, I noticed some undershoot, but as you mentioned, the magnitudes were never that large. The fuse I am testing is experiencing different currents, as shown below (time in s then current in A):

 

0.        0.

1.        130.

120.        130.

121.        -181.

240.        -181.

241.        130.

360.        130.

361.        0.

480.        0.

481.        130.

600.        130.

601.        -181.

720.        -181.

721.        8.

1320.        8.

1321.        = 8.

 

I tried setting the current to 0 for the last 3 time steps, but the same issue persisted. In a smaller simulation, I encountered fewer problems with the latest ANSYS release, but I cannot yet test it on my larger model.

 

Here are the heat flux and joule; both were given as an option in mechanical:

 

 

And for the joule heat:

 

Thanks a lot for your help and support!

 

Mathias