@Preben Alsholm 

Thanks for the recommendations.

I will try without units, but every time I have an issue with units in Maple, I start wonndering if I shouldn't have gotten Mathcad instead. Although I clearly see how Maple is much more powerful, I don't see how engineers can work with it without a good unit management. Units are the most basic way of doing eror-checking in calculations about physics (unexpected unit = error in the equation guarantied), and the best way to avoid imperial/metric units mixup (we all know how bad that could turn out ...).

I noticed assumptions get ignored, but I thought UseAssumptions would solve the issue (as recommended here). But if there is any other way to only get positive integer solutions, I will gladly try that too !

All variables are declared with units (D__core and l__e are length, A__e is a surface area, mu__0 is a magnetic permeability), hence the use of units in the solve (g__ap is meant to be a length too, I__pk an electric current, and N__1,N__2,N__t are unitless).

@acer 

Wow, that's awesome ! I didn't know individual commands could be loaded and unloaded. That's so useful ! 

That will help me so much in the futur !

Thank you very much!

@Rouben Rostamian  

Thank you very much. It works great !

It is a bit disapointing that it can't be converted into the nice-looking 2D-math symbol anymore (or can it ?), but at least there is a formula that works and that's the most important.

Thanks for the parenthesis advice too, I worked around that because I didn't understand the error, but it makes a lot of sense now.

Thank you very much. My apologies for the double-post, this article didn't come up in my research.

this is quite disappointing. I have been working for a couple of years with Mathcad, and I found it so frustrating in many ways, I was really eager to move to Maple. Unfortunately, good unit management is absolutely essential to me (and from my interaction with other Mathcad users, I'm not the only one). I hope there are some improvements on the way on this issue in the future releases ... I might stick to Mathcad in the meantime ...