Technology / Technical guide
Cyclic Bolt Loading and Flange Fatigue
A preloaded bolt does not necessarily receive the full variation of an external separating force. A closed joint shares that increment between bolt extension and loss of clamp force.
The alternating part of bolt force is not its preload.
A simple elastic model makes that distinction explicit. It also shows why changing the mean force is different from changing the cyclic stress amplitude.
01 / Closed contact
Start with a defined mechanical boundary.
Consider a hypothetical concentric axial connection with one effective bolt spring of stiffness kb and one clamped-member spring of stiffness kc.
Consider a hypothetical concentric axial connection with one effective bolt spring of stiffness kb and one clamped-member spring of stiffness kc. Initial bolt tension F0 is balanced by clamp compression. External separating load P is introduced between the member ends.
The following relation assumes linear elastic stiffnesses and continuing compressive contact. It excludes local flange bending, load-introduction effects and changing contact area. The static joint-stiffness model establishes this increment; the question here is what happens when P cycles.
02 / Load sharing
Only part of the increment enters bolt tension.
Let C = kb/(kb +kc).
Let C = kb/(kb +kc). The bolt force is Fb = F0 +CP, while remaining clamp force is Fc = F0 −(1 −C)P. The external separating force increases bolt tension and simultaneously reduces compression; it is not simply added in full to the bolt.
Both quantities satisfy Fb −Fc = P. This equilibrium check is useful because a sign error can produce two apparently reasonable expressions that fail to balance. C is a dimensionless stiffness fraction for the stated boundary, not a material fatigue strength or a universal flange coefficient.
03 / Cycle example
Distinguish range, amplitude and mean.
Choose hypothetical C = 0.2, F0 = 600 N, and a separating-load cycle between 40 and 140 N.
Choose hypothetical C = 0.2, F0 = 600 N, and a separating-load cycle between 40 and 140 N. Bolt force runs from 608 to 628 N. Its range is 20 N, amplitude 10 N and mean 618 N. The external force range is 100 N.
At maximum P, the model’s remaining clamp force is 600 −0.8×140 = 488 N, so this example stays within its closed-contact assumption. None of these loads is a product preload, a tightening value, a load limit or a field adjustment target.
04 / Stress location
Nominal amplitude is not the thread-root stress.
For a defined tensile reference area At, a nominal axial stress amplitude is σa = Fba/At.
For a defined tensile reference area At, a nominal axial stress amplitude is σa = Fba/At. In the hypothetical cycle, using an illustrative area of 10 mm² gives 1 MPa. The corresponding nominal mean stress is 61.8 MPa, calculated with that same area.
A local thread-root stress may differ because of geometry and local load transfer. Nominal and local stress descriptions must not be mixed with fatigue data expressed for another location. A stress concentration describes a local geometric effect; it does not itself supply a fatigue curve.
05 / Preload effect
Higher mean is not automatically lower amplitude.
Within the same closed linear model and with constant C, changing F0 shifts the bolt-force mean but leaves the bolt-force amplitude unchanged at C times the external load amplitude.
Within the same closed linear model and with constant C, changing F0 shifts the bolt-force mean but leaves the bolt-force amplitude unchanged at C times the external load amplitude. This mathematical result separates preload from the alternating increment.
Changes that alter contact closure or stiffness can change that conclusion. They require a different physical model rather than the same formula used beyond its conditions. The statement is therefore an explanation of an idealised comparison, not advice to raise or lower real bolt preload.
06 / Fatigue boundary
Contact and material behaviour add further questions.
If compression is lost, the closed-contact increment law no longer describes the connection throughout the cycle.
If compression is lost, the closed-contact increment law no longer describes the connection throughout the cycle. Bending, prying and nonuniform bolt-group loading can also change which bolt receives a varying force. A positive average clamp force does not prove local contact everywhere.
Even a correctly calculated stress cycle needs a suitable fatigue description, including any justified mean-stress treatment. Actual flange geometry, bolt material, local stress and load history are not provided by this example. No lifetime or fatigue strength is assigned to a MARSTON connection.
Related technical guides: Mean Stress Correction in Wind Turbine Fatigue · Adhesive Joint Peel and Shear Mechanics · Composite Blade Failure Criteria Explained.
General background: NASA Reference Publication 1228 — Fastener Design Manual, cyclic loading discussion. These references explain industry concepts; they are separate from the applicable MARSTON product and project documents.
Key concepts and conditions.
Read each quantity with its reference, units and stated conditions. A general relation does not establish a model-specific value.
| Concept / quantity | Meaning / relation | Conditions / limits |
|---|---|---|
| Stiffness fraction | C = kb/(kb +kc) | Defined closed elastic boundary |
| Bolt increment | CP | Only part of the separating load |
| Clamp reduction | (1 −C)P | Contact must remain compressive |
| Force amplitude | (Fb,max −Fb,min)/2 | Not the initial bolt tension |
| Nominal stress amplitude | Fba/At | Area and stress location must match |
Engineering questions.
Does the bolt receive the full external force variation?
Not in the stated closed elastic model. Its force increment is CP.
Is stress amplitude the same as preload stress?
No. Amplitude is half the cycle range; preload contributes to the force level and mean.
Does more preload always reduce alternating stress?
No. With constant stiffness fraction and closed contact, it changes the mean without changing the amplitude.
Can the closed-joint relation be used after separation?
No. The physical boundary has changed and needs another description.
Does nominal tensile stress equal thread-root stress?
Not necessarily. Local geometry and load transfer can create a different stress distribution.
Does the numerical example establish a tightening specification?
No. Every force and area in it is hypothetical and illustrates cycle quantities only.
Technical enquiry
Discuss the technical question.
A preloaded bolt does not necessarily receive the full variation of an external separating force. Share the model and applicable operating or design information for an engineering discussion.