Fatigue analysis is the engineering method used to predict how many cycles of repeated load a part can survive before a crack starts, grows, and breaks it. The stresses that drive fatigue are often far below the level that would fail the part in a single pull, which is exactly what makes them dangerous. A bracket, a nozzle, or a weld can look perfectly safe on a static stress report and still fail after months of ordinary vibration or thermal cycling.
That gap between “safe once” and “safe forever” is where fatigue analysis earns its place. Get it right and a design lasts its full service life. Get it wrong and the failure shows up in the field, long after the drawings were approved.
Key Takeaways
- Fatigue causes an estimated 90% of all mechanical service failures, so cyclic loads, not one-time peaks, are usually the real threat.
- Fatigue analysis counts load cycles, matches them to a material S-N curve, and sums the damage with Miner’s rule to estimate remaining life.
- Most steels are treated as having near-infinite life once they survive about 10 million cycles below a stress threshold called the endurance limit.
- Codes such as ASME Section VIII, Division 2 make fatigue evaluation mandatory for vessels and piping that see repeated pressure or thermal cycles.
- Finite element analysis turns a fatigue check into a map of exactly where and when a crack is likely to begin.
What Is Fatigue Analysis?
Fatigue causes an estimated 90% of all mechanical service failures, according to the Failure, Fracture and Fatigue Laboratory at the University of Toledo. That single number reframes how you should think about durability. The load that finally breaks a part is rarely a record-setting spike. It is the same ordinary load, applied over and over.
Fatigue analysis studies that repetition. Instead of asking “will this stress break the part now,” it asks “how many times can this stress repeat before a crack forms and spreads.” The answer depends on the size of each stress swing, how many swings occur, the material, the surface finish, and any stress concentrations such as welds, holes, or sharp corners where damage tends to start.
The method sits on top of a static stress model, so a clean fatigue result starts with a clean stress result. Many teams use finite element analysis to get accurate stresses first, then feed those into the fatigue calculation. When the geometry, loads, or code requirements get complicated, that work is often handed to dedicated FEA consultants who run these checks every day.
Why Fatigue Analysis Matters: The Cost of Getting It Wrong
Fracture and fatigue are not a minor line item. A landmark study for the National Bureau of Standards by Battelle Columbus Laboratories estimated the cost of fracture in the United States at $119 billion per year in 1982 dollars, roughly 4 percent of the gross national product, with about $35 billion per year avoidable using technology already available at the time. Another $28 billion per year could be saved through further research, the study reported. Decades later the message holds: most of that loss is preventable with analysis done before parts are built.
The human stakes are just as real. Concern about widespread fatigue damage reached the public in April 1988, when an 18-foot section of the upper fuselage separated from a Boeing 737 in flight over Hawaii. The Federal Aviation Administration later built its aging-aircraft rules around exactly this problem, requiring manufacturers to set a limit of validity in flight cycles beyond which an airplane may not fly without further evaluation. The lesson generalizes far past aviation. Any structure that cycles, whether it is a pressure vessel, a skid frame, a crane, or a pump support, has a finite fatigue life that should be known before it is put into service, not discovered afterward.
The cost of getting it wrong
What fracture costs a national economy in a year
A landmark study for the National Bureau of Standards put a figure on it, and on how much of it was already avoidable.
$119B
annual cost of fracture in the United States, roughly 4 percent of gross national product
$35B
of that judged avoidable with technology already available
$28B
more that could be saved through further research
How Fatigue Analysis Works: From Loads to Life
Most steels are treated as having near-infinite life once they can reach roughly 10 million cycles without failing below a stress threshold called the endurance limit, per guidance from the National Board of Boiler and Pressure Vessel Inspectors. Fatigue analysis is the process that gets you to a defensible number of cycles for a specific part. It runs in four practical steps.
First, capture the load history. Real service loads are messy, so a technique called rainflow counting breaks an irregular signal into a tidy set of stress cycles, each with its own range and mean.
Second, find the stresses. A stress model, usually built with finite element analysis, converts those loads into local stresses at every critical detail, including the weld toes and geometric transitions where cracks like to start.
Third, look up the life. Each stress range is compared against the material’s S-N curve, the plot that relates stress amplitude to the number of cycles to failure. Higher stress ranges spend fatigue life quickly; smaller ones may cost almost nothing.
Fourth, add up the damage. Miner’s rule sums the fractional damage from every cycle bucket. When that running total reaches 1.0, the model predicts a crack. Keeping the sum comfortably below 1.0 across every location is the goal.
This same chain of logic applies whether the driver is mechanical vibration, pressure pulsation, or temperature swings. Thermal cycling in particular is a common and underestimated source of fatigue in piping, which is why thermal expansion stress in piping systems deserves a dedicated look. For piping specifically, a full pipe stress analysis evaluates how thermal growth, weight, and occasional loads combine into the stress ranges that fatigue depends on.
Where Fatigue Analysis Is Required by Code
Where a code requires a fatigue evaluation, it is not a judgement call. ASME Boiler and Pressure Vessel Code Section VIII, Division 2 defines fatigue as failure under repeated or fluctuating stresses whose peak value is less than the tensile strength of the material, and it lays out a formal path for checking it. Engineers screen each vessel for how many significant load cycles it will see, then run a detailed fatigue evaluation, comparing the specified cycles against the permitted cycles from the applicable S-N curve and confirming the cumulative damage stays at or below 1.0 at every point, as the National Board summary of Division 2 practice describes.
Similar expectations appear across the process and power industries. Pressure equipment that cycles, rotating equipment foundations, and piping tied to thermal or pressure transients all draw code-driven fatigue requirements. Pumps built to API 610, for instance, benefit from careful cyclic evaluation, which is why FEA for API 610 pump compliance is a recurring request. Vessels raise the same questions, and a structured pressure vessel analysis folds the fatigue check in alongside the static and buckling checks rather than treating it as an afterthought.
The practical takeaway is simple. If a component sees repeated loads and carries a code stamp, someone has to prove its fatigue life on paper. The only real choice is whether that proof is built during design, when changes are cheap, or reconstructed after a failure, when they are not.
When to Bring In a Fatigue Analysis Specialist
Fatigue work rewards experience, because the answer is sensitive to details that are easy to model wrong: how a weld is represented, which S-N curve applies, and how the load spectrum was assembled. XCEED Engineering and Consulting, P.C., founded in 2016, reports that its use of finite element analysis, delegated design, and 3D CAD modeling has reduced project costs and timelines by up to 25% for clients, a direct result of catching problems in analysis instead of in the field.
Bringing in a specialist makes sense when the geometry is complex, when a code requires a formal fatigue evaluation you have not run before, when a part is already cracking in service, or when the schedule cannot absorb a redesign later. The point of the analysis is not paperwork. It is a clear, early picture of where a crack would start and how long the part will really last, so the design can be fixed while it is still a drawing.
XCEED supports this work as a national engineering partner across mechanical engineering and structural projects. If you have a component that cycles and you want to know its true fatigue life before it ships, get in touch with the team.
Frequently Asked Questions
What is the difference between stress analysis and fatigue analysis?
Stress analysis tells you whether a part can survive a load once. Fatigue analysis tells you how many times that load can repeat before a crack forms. A part can pass a static stress check with wide margin and still fail by fatigue, because fatigue is driven by the number and size of load cycles, not by a single peak.
What is an S-N curve?
An S-N curve plots stress amplitude against the number of cycles a material can endure before failure. Higher stress ranges sit low on the curve and fail after few cycles. Lower stress ranges may last millions of cycles. Fatigue analysis uses this curve to convert each load cycle into an amount of consumed life.
How many cycles before a metal part fails?
It depends entirely on the stress range. Many steels have an endurance limit, a stress level below which they can reach roughly 10 million cycles without failing and are treated as having near-infinite life. Above that limit, life drops quickly as the stress range grows, which is why controlling peak stress ranges matters so much.
Does fatigue analysis require finite element analysis?
Not always, but FEA makes the result far more precise. Finite element analysis pinpoints local stresses at welds, holes, and transitions where fatigue cracks begin, giving the fatigue calculation accurate inputs instead of rough hand estimates. For complex parts and code-driven evaluations, FEA is the practical standard.