Analytical results establish that a failure is consistent with environmental stress cracking. Reproducing the failure establishes something stronger: that this material, under this stress, in contact with this fluid, cracks. That shift from consistency to demonstration is why reproduction testing carries disproportionate weight in disputes over polymer failures. It is also why the conditions chosen for the test attract close scrutiny, because a reproduction run under conditions the part never experienced proves considerably less than it appears to.
The basic form of the test
Specimens are held at a fixed strain and exposed to the candidate agent, and time to cracking is recorded. The bent-strip approach in ASTM D1693 does this by clamping notched specimens into a curved holder, giving a known surface strain, and immersing the assembly at a controlled temperature. Constant-strain and constant-stress variants apply the same logic through different fixtures, with constant-stress arrangements generally better representing a part under a sustained applied load.
Strain has to correspond to the real part
The strain applied in the fixture is the parameter that most determines the result, and choosing it is where reproduction testing is won or lost. A strain far above what the failed part experienced will crack almost any susceptible material and demonstrates little. The defensible approach is to establish the actual total strain at the initiation site first — from residual stress measurement, assembly conditions and service loading — and to test at and around that value rather than at whatever the standard fixture happens to impose.
Testing a range rather than a point
Because the mechanism has a threshold, the most informative test is a series across several strain levels. That produces a critical strain for the resin and agent combination, which can then be compared against the strain the part actually carried. A single-point test tells you the material cracked under some condition. A threshold tells you whether the part was above or below the line, which is the question that usually matters.
Controls are not optional
Specimens at the same strain with no agent present establish that stress alone does not crack the material. Specimens exposed to the agent with no applied strain establish that the fluid alone does not degrade it. Without both controls, a reproduction is open to the argument that the material would have cracked regardless, or that the fluid was chemically attacking it rather than acting as a stress-cracking agent — which is a different mechanism with different implications.
The fluid has to be the real one
Testing with a generic reagent when the part contacted a specific formulated product is a common weakness. Commercial fluids contain surfactants, corrosion inhibitors and carriers that may be the active species, and concentration matters. Where the actual fluid has been preserved, it should be used; where only a product name survives, the specific formulation and its concentration in service should be identified rather than approximated.
Temperature and time both accelerate
Raising temperature shortens time to cracking, which is how these tests are made practical. It also risks changing the mechanism if the temperature approaches a transition in the polymer or alters the fluid. Elevated-temperature results are best used to rank materials or confirm a mechanism rather than to assert an equivalent service duration, and any acceleration factor claimed should be supported rather than assumed.
Testing the actual part beats testing a coupon
Standard specimens are moulded to produce consistent, low residual stress — which is precisely the condition the failed part was not in. Where exemplar parts exist, exposing complete parts under their real assembly condition captures moulded-in stress, geometry and assembly loading together, and is far more representative. The standard coupon test then serves as the controlled comparison rather than as the primary evidence.
What a reproduction does and does not establish
A successful reproduction establishes that the combination is sufficient to cause cracking and that the fracture morphology matches. It does not by itself establish that this is what happened in the field, since other sufficient explanations may also exist, and it does not date the failure or apportion responsibility between the material, the design and the fluid. Stating those limits alongside the result is what keeps the finding durable under challenge.
What to preserve so reproduction stays possible
Unused parts from the same lot, in quantity — reproduction consumes specimens, and a handful is rarely enough for a threshold series with controls. A sufficient volume of the actual fluid. The assembly hardware and torque values. And the failed part itself uncleaned, because the reproduction is only meaningful if its fracture morphology can be compared against the original.
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.