When you size a metal part you mostly chase yield and ultimate, with fatigue and stability alongside, and those stay fairly separable. A laminate is different in a way that takes a while to sink in. It fails as a stack of plies, the ply that goes first is rarely the one that fails the part, and the failure that actually loses the structure is often interlaminar, out of the plane your nice in-plane criterion can even look at.

First-ply failure is the first lamina to reach its limit, usually matrix cracking transverse to the fibres in an off-axis ply. Real event, conservative as a design limit, sometimes wildly so, because the laminate keeps carrying load happily afterward. Last-ply is where plies fail progressively and load sheds to the survivors until the thing lets go; closer to test, but defending it means modelling progressive damage and stiffness degradation, harder to justify on cert than a clean first-ply cutoff. Stability matters more than people expect, thin laminates buckle long before they break in-plane, so skin panels are usually stability-critical, and buckling is stiffness-driven (the D-matrix, boundary conditions, aspect ratio) so it cares about layup order and not just ply percentages. Then the interlaminar modes, which are the ones that quietly bite: free-edge delamination wherever the laminate is cut, ply-drop delamination in tapers, impact delamination from BVID. Those run on interlaminar tension and shear and on G_Ic / G_IIc, not the in-plane allowables, so a laminate that looks perfectly comfortable on Tsai-Wu can still let go at a free edge nobody capped.

The in-plane criteria, by the way, disagree, and they’re meant to. Max-stress and max-strain treat the modes as independent, simple and transparent and what a lot of cert bases default to. Tsai-Wu is one smooth interactive quadratic, convenient for a single number but it smears fibre and matrix into one surface and won’t tell you which mode you hit, which matters when a popped matrix crack and a severed fibre mean very different things. Hashin and LaRC split the two and chase the physics, more faithful, more to justify. Pick the one your cert basis accepts, understand what it throws away, and stay conservative where the test evidence is thin.

So the order I size in runs roughly stability first, then joints for bearing-bypass, then free edges and ply drops and cutouts for delamination, then impact/CAI where the basis wants it, and in-plane strength dead last, because on a decent layup raw strength is rarely the thing that gets you. The failure shows up at the detail, the free edge, the ply drop, the joint carrying bearing it wasn’t sized for, more than at the high-stress region where the fibres are pointed the right way.