Memory error in thermal simulation

Iñigo Izaguirre

Hello,

 

I am simulating a three phase power transformer of 220KVA and when I want to do the thermal simulation of it. It appears me a memory error but I dont understand why this happens because I had done a correct electromagnetic simulation before and I dont think the mesh is very heavy. Maybe there is something wrong defined in the thermal simulation. I put here my model if you can prove it and see if you have the same problem and how I can fix it. 

To do this thermal simulation I have removed the infinite box. After that I have defined the power losses in the volume region and the thermal exchanges with the environment as face region in all the faces of the model. 

 

Thank you,

 

Iñigo

Farid zidat_20516

Hello Inigo;

If it is only a thermal application, yes you can decrease the mesh, however, if there are eddy current somewhere in you transformer, they are actually a hit source, so may be it is better to take them into account.

 

The losses soulde be given for the displayed part of the transformer, if you display half of the coil for instance, you should use halve of the coil losses.

 

For the boudary conditions, please have a look on the litterature, but the best way in this case, is to compute them thanks to a fluidic software like Altair AcuSolve (Introduction to AcuSolve (altair.com)).

 

Best regards 

Farid zidat_20516

Hello Inigo;

 

Can you please go through Flux supervisor and activate Dynamic memory and try again. In addition to that, is it possible to add the error message Flux returned ? You can may be attach the flile with extension .report

 

 

Regards

Iñigo Izaguirre

Hello, I have put the dynamic memory but I get the same error again. I dont understand how the simulation needs all that memory. What can be the reason of asking for so much memory? I put the report attached

Iñigo Izaguirre

I have managed to simulate the model reducing the quality of the mesh but there are not trustable results due to the bad quality of the mesh so I am looking for where I make the mistake. When I improve the mesh the simulation blocks and takes too much time.

 

Farid zidat_20516

Helli Inigo;

 

If you have a heavy model (lot of mesh nodes) and you don't have lot of memory you may have this issue. It is important to work on the mesh, to have finer mesh where needed only (airgaps, round shape area, and skin effect) so you can reduce the number of elements.

 

If in addition to that, you are considering the eddy current, the mesh in 3D will switch from 1st order to 2nd order automatically, it means that you will have more nodes, and then you need more numerical memory to solve.

 

Regards 

Iñigo Izaguirre

Hello

How can I do to not consider eddy currents? 

Farid zidat_20516

Hello;

The eddy current can be computed by defining a given region as solid conductor, the material of this region should have an electric resistivity.

 

Regards 

 

Iñigo Izaguirre

I dont think that is the problem because is a thermal simulation what I wanted to do. I have define two regions in the simulation one for the core what containds the iron losses and the material and another region for the coils with their losses and material. Should I divide the coil region into more regions or is okay defined like that? 

 

Another question is about the region face. I have defined the region face. Do you know where I can trustable find the convection h and radiation emissivity and absortion coefficient values of air? 

Farid zidat_20516

Hello Inigo;

If it is only a thermal application, yes you can decrease the mesh, however, if there are eddy current somewhere in you transformer, they are actually a hit source, so may be it is better to take them into account.

 

The losses soulde be given for the displayed part of the transformer, if you display half of the coil for instance, you should use halve of the coil losses.

 

For the boudary conditions, please have a look on the litterature, but the best way in this case, is to compute them thanks to a fluidic software like Altair AcuSolve (Introduction to AcuSolve (altair.com)).

 

Best regards 

Farid zidat_20516

Hello again Inigo;

 

To have more visibility on your questions in the forum, I encourage you to have one question for each subject. This one several questions were asked in the same thread, and actually they have no linke with memory management.

 

Regards