Environmental stress cracking requires sustained tensile stress, and in a large share of failures that stress is not in the design load case at all. It was frozen into the part during moulding, imposed when the part was assembled, or introduced by a fit or interference nobody accounted for. Because the mechanism only needs the total to exceed a threshold, a part carrying substantial residual stress can crack under a service load that on paper is trivial — and the analysis that stops at the service load will conclude, wrongly, that the stress leg was not satisfied.
The threshold is a strain, and contributions add
Susceptibility is usefully expressed as a critical strain below which cracking does not initiate in a given resin and agent combination. What matters is the total strain at the location where cracking started, not any single contribution to it. Residual, assembly and service strains superpose, which is why apportioning them is the real work and why any one of them considered alone can look harmlessly small.
Moulded-in stress and where it concentrates
Injection moulding freezes stress into a part as it cools under packing pressure in a constrained cavity. It concentrates near gates, across weld lines, around cores and inserts, and wherever wall section changes abruptly. Cooling imbalance between cavity and core sides adds a bending component that persists after ejection. None of this is visible, and none of it appears in a drawing, but it is often the largest single contributor to the total.
Polarised light makes residual stress visible
In transparent and translucent polymers, examination between crossed polarisers renders the stress field directly as birefringence fringes, and the fringe density indicates where stress concentrates. Comparing a failed part against an unused one from the same lot, and against a part that has been annealed to relieve residual stress, separates what the moulding process contributed from what the service condition added. It is a fast, non-destructive observation that frequently reorients an investigation at the outset.
Annealing as a diagnostic rather than a fix
Heating a moulded part above its relaxation range and cooling it slowly relieves residual stress and, in doing so, quantifies it: the dimensional change on annealing is a measure of what was locked in. A part that cracks readily as moulded and resists cracking after annealing, with the same agent and applied stress, has demonstrated that residual stress was the governing contribution.
Assembly stress is the contribution most often overlooked
Over-torqued fasteners, press fits, snap fits and interference with an adjacent component all impose sustained tensile stress that persists for the life of the assembly. Unlike a service load it never relaxes and never cycles off. Boss cracking around a screw is the archetypal case: the load is entirely from installation, and the crack appears at whatever moment an agent reaches it. Reconstructing this requires the assembly, its fasteners and the actual installation torque, not the specified one.
Crack location points back at its own cause
Cracks initiate where total tensile strain is highest, so their location is evidence. Cracking that follows a weld line implicates moulding. Cracking radiating from a boss or fastener implicates assembly. Cracking at a point of maximum bending under a known service load implicates the applied load. Where cracks appear consistently at one feature across many parts, the geometry is doing most of the explaining.
Separating a design contribution from a process one
A sharp internal corner, an under-radiused boss or a rib junction that concentrates stress is a design characteristic present in every part from the tool. Inconsistent packing, an out-of-balance cooling circuit or a drifting process produces variation between parts and between lots. Comparing parts across production dates and cavities is what separates a stress concentration built into the geometry from one that varies with how the tool was running.
Why compatibility data alone will not answer this
Published chemical compatibility ratings are typically generated on unstressed specimens and describe chemical resistance rather than stress-cracking resistance. A material listed as compatible with a fluid can still crack under that fluid when stressed. Assessing a material selection against ESC requires data generated at a realistic strain, which is what the bent-strip and constant-strain protocols exist to provide.
What to preserve
The complete assembly rather than the cracked part alone, with fasteners in place if possible and installation torque recorded before anything is disturbed. Unused parts from the same lot for residual-stress comparison. The moulding process record — cavity, gate, cycle and any tooling changes — and a sample of the fluid involved, since the stress analysis only matters once the agent leg is also established.
This article is general technical orientation, not a failure analysis, an engineering opinion, or advice on any specific matter. Determining the cause of a particular incident requires hands-on examination by a credentialed expert.