The printed part that sagged over a summer
A load-bearing plastic part that failed slowly enough to be missed.
A load-bearing plastic part can sag visibly over a single summer with the load on it never once changing, because heat and creep, each covered separately earlier in this set, compound directly with one another, and a plastic running warmer sits closer to the temperature range where its molecular chains slip past each other easily, turning a slow, tolerable creep at room temperature into a much faster one once the same part is left loaded somewhere genuinely hot. Neither effect alone would necessarily catch anyone out, since creep at a comfortable indoor temperature can be slow enough to be negligible over a part's service life, and heat alone, with no sustained load already present, has nothing to accelerate.
The physics of temperature and creep together
Creep, as the earlier article in this set explained, is driven by a plastic's long tangled molecular chains slowly slipping and rearranging past one another under a sustained load, a process that needs a certain amount of thermal energy to proceed at all. Raising the temperature does not simply make that slipping happen a little faster in a straight-line way, it can accelerate it sharply, since a plastic running closer to its own glass transition, the temperature band where it shifts from a comparatively rigid material to a soft, easily rearranged one, has correspondingly more of its chains free to move under a given load. A part sitting in a cool room for a whole year and the same part sitting outdoors through a single hot summer can therefore accumulate wildly different amounts of permanent sag under the identical constant load, the difference owed entirely to temperature rather than to anything about how the part was loaded or how long it was loaded for. Different plastics also carry this risk to very different degrees, since each has its own glass transition sitting at its own particular temperature, so a plastic whose transition sits comfortably above anything a hot summer ever reaches behaves nothing like one whose transition sits only a little above a typical room temperature, even when the two plastics look, and print, almost identically.
The printed-shelf-bracket comparison
A shelf bracket printed from an ordinary plastic filament to hold a modest, genuinely unchanging weight can look completely fine through a cool spring, holding its shape with no visible complaint, and then visibly droop by the end of a single hot summer even though nothing about the shelf's load ever grew even slightly heavier. An identical bracket machined from aluminium, carrying the exact same shelf and the exact same weight in the exact same warm spot, keeps its original shape throughout, since aluminium sitting at an ordinary summer temperature is nowhere near any transition in its own internal behaviour the way a warmed plastic is. A person walking past both brackets every day for months, never once setting a straight edge against either one, is the person least likely to catch the difference early, since the droop develops slowly enough that each day looks almost identical to the last, and only a comparison against the very start reveals how far it has travelled.
Why the failure is easy to miss until it is far along
Creep accumulates slowly enough, especially at first, that no single day ever shows a difference worth noticing, and a part checked only occasionally, glanced at rather than measured, can drift a surprising distance from its original shape before the drift finally becomes obvious enough to register as a problem rather than simply as how the part has always looked. This is made worse by the fact that a printed part's layer-by-layer construction already introduces some directional weakness between layers even before creep enters the picture, so a bracket loaded across its printed layers can be quietly creeping along exactly the plane already identified elsewhere as its weakest, compounding two separate vulnerabilities into one slow, easily overlooked failure. A part reoriented on the print bed so that its layers run across the load instead of along it, an option covered in more detail elsewhere in this collection, removes one half of that compounding entirely, leaving heat-accelerated creep as the only mechanism left to design around rather than two working together.
The number that matters here
A rise in temperature of only a few tens of degrees, well short of a plastic's stated melting point, can be enough to multiply its creep rate several times over, since creep depends far more on how close a material is running to its own glass transition than on the absolute temperature it happens to reach, which is exactly why a part that seemed entirely adequate indoors can fail an equivalent outdoor summer test it was never actually given. A designer working from a strength figure measured briefly on a bench at room temperature has, in effect, tested the part under exactly the combination least likely to reveal this failure, comfortable temperature and short duration together hiding the mechanism a hot, sustained summer would expose within weeks.
What this changes in practice
A load-bearing plastic part intended for outdoor or otherwise warm service needs to be tested under a genuinely representative combination of load, time and temperature together, not under any one of those three checked in isolation at a comfortable indoor setting, since a part that easily passes a quick room-temperature load test can still fail slowly and quietly once summer heat and sustained loading are both allowed to act on it at once. Where that combined testing is not practical, choosing a metal part, or a plastic with a glass transition comfortably above anything the part will actually see in service, removes the risk at the design stage rather than discovering it months later as a visible droop. A simple habit costs almost nothing by comparison, photographing or measuring any plastic part carrying a sustained load against a fixed reference the day it is installed, since a single early measurement is the only thing that turns a slow, invisible drift into something that can actually be caught while it is still small.
My key error with this
The joystick was printed, and it worked perfectly when we handed it over, which is the whole difficulty with this particular failure, because the part did not break and there was never a moment where anything obviously went wrong. It was a thumb control for a child who used it daily, so it sat under a small sustained load for hours at a time, and over a warm summer the printed body crept just enough that the neutral position drifted and the travel at one edge shortened. The child adapted to it, which meant nobody reported a fault, and the calibration we had set so carefully was quietly wrong for weeks before anyone connected the growing difficulty back to the hardware rather than to the user. What replaced the belief is that a part which never breaks can still fail, that a small continuous load at a raised temperature is exactly the combination plastics are worst at, and that anything a person calibrates themselves against needs checking against its original geometry rather than against whether it still functions.