Fatigue analysis for product design

A part passes every check in the structural review. Static stress well under yield, safety factor comfortable, sign-off granted. Eighteen months later it cracks in the field, and nobody can explain why, because the load never came close to breaking it. We've seen this play out enough times to know it's rarely a mystery. The cause is almost always the same, and it's mundane: a modest load, well within what the part can take, that came back a few million times until something cracked. That's fatigue, and it's the failure mode a static analysis is structurally incapable of catching. Fatigue analysis for product design exists to answer the question static FEA can't: not "will this part survive the worst load," but "how long will it survive the normal ones." For anything that cycles or vibrates, that second question is the one that decides your warranty exposure. ## The Failure Mode Static Stress Checks Miss Static FEA answers one question well. Apply the worst-case load, check whether the part yields or breaks, confirm there's margin. That's the right check for a lifting eye that has to hold a defined load, or a bracket that must survive a one-time shock. It tells you almost nothing about a part that sees the same load thousands or millions of times. Metal fails under repeated loading at stress levels well below its static strength. For many steels the fatigue limit sits at roughly half the ultimate tensile strength, a rule of thumb that runs through standard fatigue references like Shigley's Mechanical Engineering Design. A structure can be nowhere near yield on any single cycle and still crack, because fatigue damage accumulates. Each cycle does a tiny, invisible amount of harm. Add up enough of them and a crack starts, usually at a weld toe, a bolt hole, a fillet, or a sharp corner where stress concentrates. The uncomfortable part is that the static report looks clean the whole way through. There's no warning in it, because it was never asking the question. Take a pressure vessel on an oil and gas skid that cycles with every process batch, or a defense vehicle mount that carries road and firing loads across its service life. Both have margin to spare against their peak load. Both can still crack, because the cycles pile up and nobody was counting them. ## What Fatigue and Crack-Propagation Analysis Actually Predicts Fatigue analysis reframes the question from "how strong" to "how long." Instead of a single worst-case load, it works from the loading history: the range of loads the part actually sees in service and how often each one repeats, broken into cycles the way rainflow counting (standardised in ASTM E1049) does it. Combine that with the material's fatigue behaviour — how many cycles it survives at a given stress range, read off the S-N or Wöhler curve — and a cumulative-damage rule like Palmgren–Miner turns it into an estimate of how many cycles the design lasts before a crack initiates. Crack-propagation analysis picks up where that leaves off. Once a crack exists, whether from fatigue, a manufacturing flaw, or a weld defect, the question becomes how fast it grows and how long until it reaches a critical size. That's the difference between a crack you catch at the next scheduled inspection and one that fails between inspections. For a safety-critical structure it's also what lets you set a defensible inspection interval instead of guessing one. Neither analysis needs exotic inputs. It needs an honest loading spectrum and the right material data. In our experience the first of those is where teams underinvest. They'll model the geometry in exhaustive detail and then feed it a loading assumption somebody guessed in a meeting. ## Why This Is a Business Problem, Not Just an Engineering One Fatigue failures are expensive in a particular way: they show up after the product ships. A static design error tends to surface early, in testing, when it's cheap to fix. A fatigue problem surfaces in the field, months or years in, across a whole population of units already sold. That's the shape of a warranty campaign or a recall, not a design revision. For an SME that difference can be existential. A large OEM can absorb a field-failure campaign. A 50-person product company often can't — not the direct cost, and not the reputational damage with the two or three major customers its revenue depends on. Defense and oil & gas buyers in particular don't forget a part that cracked in service, and they carry that memory into their next tender. The commercial argument for fatigue analysis is simple: it moves that discovery earlier in time. Catching a fatigue-limited detail in simulation, while the geometry is still soft, costs a few days of engineering. Catching it after launch costs a redesign, a retrofit programme, and a conversation with your customer that no one wants to have. ## Where Fatigue Analysis Belongs in the Product Design Cycle The instinct is to treat fatigue as a final check. Run it near the end, confirm the design is fine, tick the box. We think that's backwards, and it's one of the more common patterns we see. By the time a design is frozen, the details that drive fatigue life are already locked in. The fillet radius, the weld type, the bolt pattern, the section change from thick to thin — these decide whether a part lasts, and they get set early. Fatigue analysis earns its keep when it informs those decisions rather than just auditing them afterwards. That doesn't mean a full analysis on every concept sketch. It means knowing which parts are fatigue-driven — anything cyclically loaded or vibrating, anything with a service life it has to hit — and getting a fatigue view on them while the geometry can still move. A rough fatigue estimate early is worth more than a precise one after the tooling is cut. This is also where it connects to work a team may already be doing. If you're running structural FEA for CE marking or another standards-based sign-off (a Eurocode fatigue check, or a DNV or ASME assessment on a pressure-bearing part), the stress results that feed a fatigue assessment are often already sitting in the model. The marginal cost of adding the fatigue view is smaller than most teams assume. ## The Check That Decides Whether a Warranty Holds Fatigue is the failure that doesn't show up in the review that matters, on the timeline that matters. A part can clear every static check and still be running down a clock nobody started. Fatigue analysis for product design starts that clock while there's still time to change the answer: it tells you which details are quietly limiting a part's life, and it moves the discovery out of the field and back into the design phase where fixing it is cheap. For anything that cycles or carries a defined service life, that check is what stands between a warranty period you can honour and one that turns into a liability. If you already run FEA on your load-bearing parts, the honest next question is whether you've checked how long they last, not just whether they hold. --- Thinking about which of your parts are actually fatigue-driven, and whether they've ever been checked for it? Get in touch.