The model is not one model. A global shell model covers the tub, the deck, the track-interface frames, the recovery crossmembers and the battery protection, and separate sub-models cover every place load concentrates: the cartridge joint, the recovery eyes, the mission hard points, the lift and tie-down brackets. Global models give global answers, and a bolt hole needs its own.
Masses go in where they actually are. The packs, the power module, the traction units, the hydraulics and the deck load are explicit masses at controlled coordinates, not a smeared density. The holes are modelled, because a box with a hole in it is not the same box. Module loads pass through the real hard-point pattern and never through one convenient node.
Two clauses exist purely to close the easiest routes to a passing answer. Track support conditions may not artificially lock the chassis in torsion, and a recovery restraint may fix only the ground reactions that genuinely exist for the direction of pull. Welds are judged on fatigue against a duty spectrum rather than on static stress, which is the difference between a structure that is strong and one that lasts. And the analysis owes the parts nobody enjoys: mesh convergence, element quality, and a reconciliation of the model's mass and centre of gravity against the budget.