[mliben]

We take the 80/20 approach here. We start with analytical design (can be done pen on paper, but typically happens in excel/etc) to get a high-level view on the "continent" of design possibilities. This can be done readily with magnetic circuit models and thermal resistance networks.

From there, we identify the best high-level design traits and begin some optimization work to get most of the way to a fully-optimized solution (this is more FEA-based, both mechanical, magnetic, and thermal). This gets us climbing up the right "mountain" on the continent.

At this point, we put the pens down and start building something, because we will learn more building a prototype and iterating than we will spending too much time in simulation-land.

In parallel, we have been putting together a workflow to co-optimize the design across mechanical, magnetic and thermal simultaneously. Thus far the jump across each discipline has been a bit more on the manual side, but automatic co-optimization is a long term project, and would get us to the "peak" of the mountain.