Derg:
using the finite element method scientist/engineers can predict mechanical beahviour of any part in known conditions. Actually they can model it according to known and proven cases. Problem starts when we try to predict/calculate mechanical behaviour in the unknown/unpredictable/non-linear situations like worn bearing where actual 'worn' may have very different meanings including radial and/or axial play with different dampening properties. This calculation comes even worse if we get closer to loose spline coupling - i would bet a pint of guiness that there are no actual models for calculating behavoiur of loose spline couplings - simply because these couplings are designed to STAY and not to get LOOSE.
Once i heard a story of an experienced machine-tool designer who used to say: ok, so we have built it. Let's start it and see why it doesn't work.
