Pressure Vessel Fatigue Analysis: Methods, Steps, Pitfalls
Pressure vessel fatigue analysis checks whether a vessel can survive the repeated pressure and temperature swings it will see in service, even when every single load is well below the material’s strength. You define the cyclic events, find the stress range at the most severe details (usually nozzles and other geometry changes), convert that range into permitted cycles using a fatigue curve, and add up the damage. If the total stays at or below 1.0, the design passes. This guide walks through each step, the code routes behind it, and the mistakes that most often make a result look safer than it is.
TL;DR
Fatigue failure comes from repeated or fluctuating stress whose peak is below the tensile strength, so a vessel that passes a static check can still crack.
Start with the load histogram: operating conditions, cyclic events and expected cycle counts. Everything downstream depends on it.
Cracks start where geometry changes: nozzles, diameter transitions and welded joints. Model those locations in detail.
Damage is summed with Miner’s rule. The sum of specified cycles divided by permitted cycles must be 1.0 or less.
Treat the result as a design life with margins, not a prediction, and plan inspection well before the end of it.
Table of Contents
What fatigue analysis actually assesses
The ASME Boiler and Pressure Vessel Code, Section VIII, Division 2, defines fatigue as a condition leading to fracture under repeated or fluctuating stresses with a maximum value below the material’s tensile strength, according to the National Board’s overview of pressure vessel fatigue. The same source describes fatigue damage as a progressive, localized, permanent structural change. It also notes that a vessel design is considered preliminary until the fatigue analysis is complete.
That last point matters in practice. Static checks (wall thickness, reinforcement, primary stress limits) tell you the vessel holds pressure once. Fatigue analysis asks whether it keeps holding pressure after thousands of start-ups, shutdowns, pressure swings and thermal transients. If you want the static side first, our pressure vessel stress analysis guide covers formulas and FEA checks that feed directly into the fatigue work.
When you need to do it
In ASME VIII-2 the fatigue evaluation is a screening-first process: you work out whether cyclic service is significant enough to need a full analysis. The key input is the expected number of full-range pressure cycles plus thermal and mechanical cycles. If the screening shows the cycle count is low enough, no detailed fatigue analysis is needed. If not, you move to one of the detailed methods below. Always check the edition of the Code your jurisdiction or customer specifies, since the screening criteria and curves are edition-specific.
Typical services that fail screening are the ones with frequent cycling: batch reactors, pressure-swing equipment, vessels with daily fill and empty cycles, and equipment with severe thermal transients. Our piece on thermal fatigue pitfalls goes deeper on the temperature-driven side.
Step 1: Define the cycles
The vessel user specifies the operating conditions and cyclic events to be considered in design, and a load histogram is prepared from that information, per the National Board. Do not skip the conversation with the process team. A design basis that says “10,000 cycles” without saying what a cycle is will produce an unreliable result.
For each event, write down:
Pressure range (minimum and maximum, including vacuum if it occurs)
Metal temperatures and rates of change, not only fluid temperatures
How many times per year it happens and over how many years
Whether it overlaps with other events (a pressure swing during a thermal transient is worse than either alone)
External loads from piping at nozzles, if they cycle
Counting irregular histories
Real service rarely repeats one tidy cycle. Where the load history is irregular, rainflow counting, or another method that complies with ASTM E1049, is used to count stress cycles and combine partial cycles into complete ones, again according to the National Board. A pressure trace with many small ripples on a big swing will count as one large cycle plus smaller ones, and the small ones are not free: they add to the damage sum.
Service levels
Nuclear practice under ASME Section III sorts events into Level A (normal), Level B (upset), Level C (emergency) and Level D (faulted) conditions, as laid out in an NRC design control document. You do not need that formal scheme for a process vessel, but the habit is useful: separate the routine cycles that drive fatigue from rare upsets, and agree with the owner how each is counted.
Step 2: Find where the vessel will crack
The most highly stressed areas occur at changes in geometry, including nozzles and transitions in diameter, according to the National Board. Those are your fatigue-critical locations, and they should get the finest mesh and the most careful stress classification.
Common hot spots:
Nozzle-shell junctions. The opening interrupts the shell, the nozzle adds stiffness and external loads enter here. Reinforcement sizing is a separate check; our nozzle reinforcement spreadsheet guide covers that part.
Head-to-shell transitions. Knuckles and cone junctions concentrate bending. See the head types guide for the ASME rules and FEA checks for each.
Welds. Welding defects, stress concentrations and high residual stresses at tube-connection welds can raise the risk of fatigue failure, as a 2022 study on tube connections describes. Weld quality and joint efficiency assumptions also matter; see our note on ASME joint efficiency.
Tubesheets and heat exchanger joints, where thermal gradients add to pressure. Our tubesheet thermal analysis article walks through the code checks and FEA.
Supports, lugs and attachments, where local bending and thermal restraint combine.
Step 3: Pick a method
The same 2022 study explains that ASME VIII-2 Section 5.5 offers three routes: an elastic equivalent-stress method, an elastic-plastic equivalent-strain method and an elastic structural-stress method. EN 13445-3 offers a simplified assessment (Chapter 17) and a detailed one (Chapter 18). The authors compared four methods applied to tube connections and described the simplest as easy to apply while the most refined was the most accurate, stable and reliable of the four, with the intermediate ones being complicated (Su et al., 2022).
Approach | Typical input | Effort | Where it fits |
Elastic equivalent stress (ASME VIII-2, 5.5) | Linear-elastic FEA or handbook stress, stress range with concentration factors | Low to moderate | Standard vessels with modest cycling and well-understood details |
Elastic-plastic equivalent strain (ASME VIII-2, 5.5) | Nonlinear material model, strain range | High | High-stress locations where elastic results are too conservative |
Structural stress for welds (ASME VIII-2, 5.5) | Fine-mesh FEA at weld toes, mesh-insensitive stress extraction | Moderate to high | Welded joints, tube-to-tubesheet connections, nozzle welds |
EN 13445-3 simplified or detailed (Chapters 17, 18) | Stress ranges, weld class or detailed curves | Low to moderate | Vessels designed to the European standard |
There is no single right answer. A simple method that is conservative may be enough for a vessel with few cycles. If it fails, switching to a more refined method is often cheaper than adding wall thickness, but only if you can justify the extra modeling effort.
The practical limits of code-based procedures are also well known. A review of fatigue design procedures from 2000 noted practical difficulties with both the American and British approaches. For geometries outside design-by-rule, such as multiple or closely spaced nozzles, researchers have built high-fidelity finite element models of nozzle-cylinder junctions to get the detailed stress information a fatigue assessment needs.
Step 4: Calculate the stress range
The equivalent stress range combines several contributions: average stress, bending stress, stress at a structural discontinuity, thermal expansion stress and notch (peak) stress, according to the National Board. In an FEA workflow this typically means:
Build the model with the details that matter (weld profile, nozzle fillets, transitions) and a mesh fine enough to resolve the gradient through the wall.
Run each load case in the cycle definition: pressure, thermal, external loads.
Take stress results at the critical locations and, for each pair of load states, compute the range between them. Do this on a tensor basis rather than subtracting scalar maxima.
Add the local effects not captured by the mesh, such as weld notch factors.
Apply any plasticity correction your chosen method requires.
The thermal contribution deserves its own transient run. Metal temperature gradients during heat-up and cool-down can dominate the range. Our pressure vessel thermal analysis workflow shows how to set up heat transfer so that the temperature field going into the stress run is realistic, and the thermal shock article covers the severe end.
An example of how this looks in practice: a published FEA study evaluated a typical vessel under thermal and pressure cycling using a 2D axisymmetric model, calculated the alternating stress, read permitted cycles from the material fatigue curve, and divided actual cycles by permitted cycles to get damage, accepting the design if the accumulated damage was below one (Mahardika and Nugroho). That is the full method in one sentence.
Step 5: Convert to life with S-N curves and Miner’s rule
Fatigue life is estimated by comparing the specified stress cycles with the permitted cycles on an S-N curve, per the National Board. Those curves come from polished test samples, and design curves typically reduce stress by a factor of 2 or cycles by a factor of 20, whichever is more conservative, to cover scatter, environmental effects and size effects.
For several load types, Miner’s rule sums the ratio of specified cycles to permitted cycles for each. The sum must be less than or equal to 1.0 (National Board).
A worked illustration with made-up numbers, to show the arithmetic only:
Event | Specified cycles | Permitted cycles (from curve) | Damage |
Full pressure swing | 2,000 | 20,000 | 0.10 |
Start-up thermal transient | 500 | 4,000 | 0.125 |
Small pressure ripple | 100,000 | 1,000,000 | 0.10 |
Total | 0.325 |
The total is well under 1.0, so this hypothetical detail passes. If the thermal transient had 1,500 cycles, its damage would be 0.375 and the total would climb to 0.575. Notice that the small ripple contributes as much as the full pressure swing. That is why counting matters.
Vacuum and external pressure
Vessels that see vacuum or external pressure are designed mainly against buckling and stability, so the governing check for external pressure is not fatigue. That does not mean fatigue disappears. If a vessel cycles between internal pressure and vacuum, the full stress range spans both, and the range at a nozzle or weld can be larger than for internal pressure alone. Count the swing from the most negative to the most positive pressure as one cycle, define it in the load histogram, and run it as a load pair in your stress range calculation. Where the vessel code treatment is unclear for your case, confirm the approach with the authorized inspector or the designer of record before relying on a result.
What the result means for design and operation
A passing damage sum is not a promise that the vessel lasts forever or fails at a given date. Design curves already carry margin for scatter and size effects. At the same time, code criteria use peak surface stresses and strains, a simplification that, as ASME Standards Technology, LLC noted, can give overly conservative design lives and does not account for the limited notch sensitivity of many construction materials. The same document says fatigue design technology has advanced rapidly and proposed work to update conservative criteria while keeping acceptable margins. Practical takeaway: a conservative screening result is not a reason to over-build, but it is a reason to examine whether a more refined method is justified.
On the operating side, the National Board recommends regular examination, including visual and appropriate non-destructive examination, at 50% of predicted life (National Board). Put that in the inspection plan, focused on the same hot spots your analysis flagged. Have the plan reviewed by your inspector or the relevant jurisdiction, since requirements differ.
Common pitfalls
Vague cycle definitions. Agree on what counts as a cycle before modeling.
Ignoring thermal cycles. Pressure-only analysis can miss the dominant stress range.
Coarse meshes at weld toes. Stress at a notch depends on mesh size unless you use a mesh-insensitive structural-stress approach.
Mixing up classification. Treat linearized membrane, bending and peak stress properly, or use a method that avoids classification.
Forgetting that welds are worse. Use weld-specific curves or methods.
Treating the sum as a prediction. It is a code margin, not a failure date.
Ignoring insulation and corrosion. Corrosion under insulation can create notches. See our insulation and CUI guide.
Perspective: why we start with temperature
We build thermal tools, so we look at fatigue from the heat side. In our experience, the stress range an engineer calculates is only as good as the temperature field behind it. Teams often spend days refining a mesh at a nozzle and minutes on the boundary conditions that set the metal temperature gradient. We think that is backwards. Before running a detailed fatigue model, get conduction, convection and radiation right, including how properties change with temperature, and check that the transient is realistic. Then refine the stress model where the thermal result says it matters.
We also think small teams and students should not be blocked from this by licensing costs. The reasoning is simple: hand checks and a clean thermal model get you most of the way, and you can reserve the expensive nonlinear work for the details that need it.
— Joel
Run the thermal side with the Jewlz toolkit
The Jewlz Engineering Toolkit is an Excel-based tool that combines conduction, convection and radiation analysis with variable material properties, a wide temperature range and a built-in property database. It also includes pressure vessel simulation and CFD capabilities, so you can build the temperature history that feeds your stress and fatigue work in one place. Some basic tools are free.
See the Jewlz Engineering Toolkit for the thermal and pressure vessel modules, or look at our heat exchanger optimization software if your cycling comes from exchanger duty.
FAQ
Is a fatigue analysis always required for a pressure vessel?
No. In ASME VIII-2, you first screen based on the expected number and size of cycles, and a detailed analysis is needed only if the screening says cyclic service is significant. Check the Code edition and any customer or jurisdiction requirements for your vessel.
What is a full-range pressure cycle?
It is a swing from the minimum to the maximum operating pressure, typically from start-up to full operation and back. Smaller swings are counted separately, often by rainflow counting, and they still add damage. Agree with the vessel user on the definition and document it.
Why do nozzles fail first?
Stress is highest where geometry changes, and a nozzle both interrupts the shell and adds stiffness and external loads. Welds there add notches and residual stress. That is why nozzle-shell junctions usually get the most detailed model.
Can I use linear-elastic FEA for fatigue?
Yes, the elastic equivalent-stress method works from linear-elastic results and is common. If it gives an unacceptable damage sum, an elastic-plastic strain method or a structural-stress approach for welds may reduce conservatism, at the cost of more modeling.
What does a damage sum of 0.8 mean?
It means the design meets the Miner’s rule limit of 1.0, with 20% of the code margin unused. It does not mean 80% of the vessel’s life is gone, and it is not a failure forecast. Inspection at half of predicted life is still recommended.
Sources
Recommended
Comments