> Similarly I don't see why thermal calculations would require a particular geometry

I imagine a curve could have more error, or require more time, with an FTDT simulation, compared to something grid like.