I think you're right Patrick.
No objection per se. Create the locus of each tangent point, construct geometry to point loads in directions using datum points.
I wanted to attempt to include the support movements whose relative magnitude of movement is a bit more than second order.
Thanks