ICEM and EPNC papers

Phew, the last two weeks have been somewhat busy. Last Sunday, I finally finished and submitted my 7-page conference paper for ICEM, one of the biggest conferences on electrical machines. Immediately after that, I scribbled together a shorter two-page synopsis paper for EPNC, a conference that my research group is organizing next summer. To top things off, I stepped on some broken glass during a sauna event last evening and had to get a tetanus booster. You know, to not die and stuff. Nothing is more harmful to publishing, and general success and awesomeness than dying.

Actually, both of my papers are worth a small story on their own. In fact, I wasn’t going to attend any conference whatsoever this year; just focus on journals instead to get the two remaining publications for my thesis. So what changed my mind?

Well, I had a half-working approach about spectral stochastic modelling of random windings. It wasn’t working very well and I was quite ready to abandon it altogether, but since ICEM was coming and all I decided to give it one more change. Needless to say, it didn’t work out. I’ve now placed spectral stochastic methods on an indefinite hiatus, but we’ll see what the future brings. Most likely, I’ll publish my raw findings here on my page and as unpublished work on my Researchgate profile, in the name of scientific openness and the thin hope that somebody might be able to build something useful on them. Cite me if you do.

Luckily, I had another method in mind about reduced-basis finite element modeling of stranded windings. I was first going to simply implement that to get on with my main work, and briefly describe the approach in a wider scope in a future journal article. With my current knowledge, I’m happy to say that wouldn’t have worked out at all. Indeed, I was barely able to describe the approach and demonstrate its accuracy in seven pages in a satisfactory fashion. I will add a description of my work here once I receive a word from the review organization.

Oh and yeah, that paper I was expecting to be finished with in two weeks. It took six in total, and required a deadline extension to be able to make it at all. Just goes to show that things usually take much longer than planned. An yes, I sweated and toiled for nothing: the deadline was extended for the second time. Lots of conferences this year competing for the same papers, it seems.

The EPNC paper, on the other hand, was about time-stepping 3D reluctance network modeling of an axial flux permanent magnet machine. Basically, it was a veeeery brief documentation of a reluctance network software I wrote for an EU project. In general, I think reluctance networks have been treated far too briefly (and often incorrectly) in the literature, so my paper’s a small step to the right direction.

Anyway, for that project I programmed both a loop method based solver for general nonlinear reluctance networks (or any flow-network for that matter, now that I think of that), and a fully parametriced machine model for the prototype machine we were developing. In the end, we were able to obtain acceptable accuracy with a 200-fold speedup when compared to the commercial Comsol software. And that was in Matlab, so we could have probably done at least twice better with compiled code. I will write more about that project too in the future.

ICEM and EPNC papers