Incompatibility of eMach evaluator with JMAG v23.0

[noguchi-takahiro]

I did eMach tutorial 5, which was working right with JMAG v22.0 but did not work with v23.0.

I got this error below after I upgraded JMAG to v23.0 and run the code:

This error is saying the csv file of ECCE2020TranPMSM_iron_loss_loss.csv is not properly generated.
If I look into the JMAG created by eMach, it looks like the material of the rotor core and stator core are set as "Air", which might cause the problem. I made sure this was not happening with v22.0.

I am expecting that the core material should be Arnon5. Maybe the code for setting up the material (for only the core) does not seem to be working with v23.0?

[elsevers]

Per 3/12/24 meeting,

@dmnewman3 and @noguchi-takahiro will try:

To Reproduce the Issue with Point and Click

Draw things in JMAG Geometry editor, set their names, and then observe whether the names are changed when you bring it into JMAG Designer.

If you can reproduce the problem this way, send it off to Dheeraj.

More Challenging: we could try to create a short self-contained script

Create the shortest possible .py script that demonstrates the region naming issue.

The script will not depend on eMach. It will open JMAG. Draw something. Try to name the region.

If a person runs the script on a computer with JMAG 22 they will get a different region name than JMAG 23.

[dmnewman3]

@noguchi-takahiro @elsevers
In going back and forth with folks over at JMAG, a solution is to use the following lines of code to set the version of the geometry editor:

geomApp.SetVersion(21.1)

This would enable us to use JMAG23 but with an older version of the editor itself, which is where the regions are being named during import to JMAG Designer. JMAG also indicated that the region naming convention of Name/Region is likely here to stay.

My thoughts are to specify the older geometry version as a fix to this issue. This would look like inserting the previously mentioned line of code in between lines 319 and 320 of the code referenced below.

eMach/mach_eval/analyzers/electromagnetic/bspm/electrical_analysis/JMAG.py

Lines 319 to 321 in 2dd1535

	geomApp = app.CreateGeometryEditor()
	# geomApp.Show()
	geomApp.NewDocument()

I have done this with Tutorial 5 with success. A PR to remedy this would involve changing the aforementioned file, as well as the JMAG file in mach_cad.

[noguchi-takahiro]

@elsevers @dmnewman3 I investigated the root cause of the JMAG naming issue in v23.

According to the email from Anvar on 5/20, it looks like both Part name and Region name are imported into JMAG Designer.

To verify this, I attempted to import a JMAG geometry editor template from the path Rotor/2D Geometries/Salient Pole/rsr_001 Round Tooth Bottom. This template has the default region name set to Core.

In JMAG v23 ...

Yes, these Part name and Region name are both imported into JMAG Designer v23 as Rotor/Core.

And then, I created a "new" .jproj file and did the same thing, but I renamed the Region name of Core to Region, then imported it into the JMAG Designer.

Again, both Part name and Region name are imported into JMAG Designer as Rotor/Region.

However, I observed a slight different behavior in v22.

In JMAG v22 ...

Here is a result of importing the Rotor core without renaming, where the Region name is Core and both Part name and Region name are imported into JMAG Designer v22, consistent with the behavior observed in the v23.

However, if I created a new .jproj file, renamed the Region name from Core to Region, and importet it into the JMAG Designer, the result were are follows:

Interestingly, the Part name looks now Rotor, without the /Region.

I guess in v22, only if the region name is Region, JMAG Designer omits the /Region when importing the parts. But, I believe that when JMAG was updated to v23, the behavior was changed to retain the /Region suffix when importing models, even if the region name is Region, which should cause this eMach naming issue.

[elsevers]

From my conversation with @noguchi-takahiro, eMach was previously allowing all of these regions to remain named as Region (this is the default name assigned when you create a region in geometry editor). JMAG 23 fixed the inconsistent part naming when a model is brought over from geometry editor. Because we happened to be using region names of Region, this resulted in our part names becoming different in JMAG23

[dmnewman3]

Not sure why this didn't get closed when #345 was merged, but the merging of PR #345 closes this issue