contact@nxhdmfg.com
+86 352 5330772
+86 180 3528 0246
Preload is the only load the bolt actually carries in a static joint. Everything else the assembly experiences is reacted by the clamped members, with the bolt acting as the tie that keeps the joint closed.
That single idea is where most fastener problems begin. A bolt tightened to a nominal torque and left alone will hold a joint together for years, or it will lose preload in a week — and the deciding factor is not the torque value, the property class, or the diameter. It is how much preload the joint actually needs, and how much preload the assembly process actually delivered.
This article works through that calculation: where preload comes from, how to size it against the property class, what happens when external load enters the joint, and why the safety factor for a static joint is not the safety factor for a dynamic one.
What preload actually is
Preload is the tensile force in a bolt after it has been tightened and before any external load is applied. It is generated by stretching the bolt and compressing the clamped members against each other.
That means a bolted joint holds load in two entirely different ways, and confusing the two is the source of most specification errors:
| Load type | What carries it | What the bolt sees |
|---|---|---|
| Clamp load (preload) | Friction between the clamped faces | The full preload force, permanently |
| External load applied along the bolt axis | The bolt | Preload plus the external load |
| External load applied across the bolt axis | The clamped members, by friction | A fraction of the external load, limited by the friction available |
The third row is the one that surprises people. In a transverse load — a shear load trying to slide one flange across another — the bolt does not carry the load directly. The clamped faces carry it through friction, and the friction available is limited by the clamp force. If the preload is insufficient, the joint slips and the bolt then sees the full load in shear, which it was never designed for.
Preload is not a nuisance value to be minimised. It is the working capacity of the joint.
How preload is derived: the 70% of yield rule
Bolts are specified by property class, which sets a minimum yield strength. The common practice is to target a preload at roughly 70% of that yield strength, applied over the thread's tensile stress area:
F_pre = 0.7 × R_p0.2 × A_s
Where R_p0.2 is the minimum yield strength of the class in MPa and A_s is the tensile stress area of the thread in mm².
The 70% figure is a compromise, not a maximum. Push preload higher and the bolt gains margin against yielding, but it also loses ductility reserve and becomes more sensitive to hydrogen embrittlement and to over-torque during assembly. Stay too low and the joint has no reserve to absorb anything unexpected.
| Property class | Minimum R_p0.2 (MPa) | Minimum R_m (MPa) | Yield ratio | Preload as % of yield |
|---|---|---|---|---|
| 8.8 | 640 | 800 | 0.80 | 70% |
| 10.9 | 900 | 1000 | 0.90 | 70% |
| 12.9 | 1080 | 1200 | 0.90 | 70% |
The yield ratio column is worth pausing on. Class 8.8 has a yield ratio of 0.80, meaning 20% of its tensile capacity is available as deformation before failure. Class 10.9 and 12.9 sit at 0.90, leaving only 10%. That is the reason high-strength bolts are more likely to fracture rather than bend under shock, and it is the reason 10.9 needs an embrittlement relief treatment after plating — the same strength that reduces the preload you can safely apply also reduces the margin for the hydrogen to work in.
Using the standard stress areas for common metric sizes, the resulting preload and reference tightening torques (at K = 0.20) are:
| Size | Stress area A_s (mm²) | Preload 8.8 (N) | Torque 8.8 (N·m) | Preload 10.9 (N) | Torque 10.9 (N·m) | Preload 12.9 (N) | Torque 12.9 (N·m) |
|---|---|---|---|---|---|---|---|
| M6 | 20.1 | 9,000 | 11 | 12,700 | 15 | 15,200 | 18 |
| M8 | 36.6 | 16,400 | 26 | 23,100 | 37 | 27,700 | 44 |
| M10 | 58.0 | 26,000 | 52 | 36,500 | 73 | 43,800 | 88 |
| M12 | 84.3 | 37,800 | 91 | 53,100 | 127 | 63,700 | 153 |
| M16 | 157 | 70,300 | 225 | 98,900 | 317 | 118,700 | 380 |
| M20 | 245 | 109,800 | 439 | 154,400 | 617 | 185,200 | 741 |
| M24 | 353 | 158,100 | 759 | 222,400 | 1,067 | 266,900 | 1,281 |
| M30 | 521 | 233,400 | 1,400 | 328,200 | 1,969 | 393,900 | 2,363 |
Torque follows from T = K × F_pre × d, so every number in this table inherits the K value used to produce it. A K of 0.20 assumes a dry, clean, unlubricated thread and an unlubricated bearing face. Lubricated threads routinely run at K = 0.10 to 0.12, which halves the preload for an unchanged torque setting, and coated fasteners run somewhere between 0.12 and 0.20 depending on the coating.
When external load enters the joint
Two cases have to be treated separately, because they load the bolt in completely different ways.
Axial external load. A force pulling the joint apart along the bolt axis adds directly to the preload. The bolt sees F_pre + F_ext, and the joint separates entirely when F_ext exceeds the preload. The design rule is that the total must stay below the proof load, which is why the preload target is set as high as practical in the first place.
Transverse external load. A force trying to slide the joint sideways is carried by friction between the clamped faces. The friction force available is the preload multiplied by the friction coefficient of the joint surface. Two consequences follow.
First, the preload must be sized for the friction demand, not for the bolt's strength alone. A widely used design rule for bolted joints under transverse load is to increase the preload by a factor of around 1.3 relative to the static case, because the joint's own flexibility means the friction force does not develop in step with the applied load.
Second, once sliding does occur, the bolt takes the full transverse load in shear. A bolt loaded in shear at its rated capacity is far weaker than the same bolt in tension, because the shear plane intersects the thread and the stress concentration at the thread root applies in the loaded direction too.
| Joint condition | Load on the bolt | Failure mode to check |
|---|---|---|
| Static, no external load | Preload only | Yielding at the thread root |
| Axial external load | Preload + external load | Tensile rupture |
| Transverse external load, no slip | Friction at the interface | Friction capacity, not bolt strength |
| Transverse external load, after slip | Full external load in shear | Shear at the thread |
Safety factor: static joints and dynamic joints are not the same calculation
The safety factor is where preload, property class and load case come together. For a static joint the bolt is sized against its proof load:
S = R_p0.2 × A_s / F_bolt
where F_bolt is the total force in the bolt (preload plus any axial external load).
For a dynamic joint the calculation changes completely, because the bolt is cyclically loaded between the preload and the preload plus the external load. What matters then is the stress range, and the fatigue limit rather than the yield strength becomes the governing property.
| Load case | Governing property | What drives the sizing |
|---|---|---|
| Static | Proof load / yield strength | Total bolt force including preload |
| Static with transverse load | Friction capacity and bolt shear | Preload and slip risk |
| Cyclic / dynamic | Fatigue limit at the thread root | Stress range, thread geometry, preload consistency |
| Shock or impact | Yield strength with reduced ductility reserve | Energy absorption, class selection |
The dynamic case is also where preload consistency becomes critical. If the preload varies from bolt to bolt, the joint does not share load evenly: the most highly loaded bolt sees a stress range that is larger than the design intends, and it becomes the one that fails. Consistent preload is achieved by controlling torque and the K factor, by using a tension-and-angle method, or by verifying the delivered preload directly.
The preload cannot be ignored in a fatigue calculation. A higher preload raises the mean stress, which for a bolt in tension moves the stress range away from the origin and can reduce fatigue life even though it improves the joint's resistance to separation. The optimum preload for a fatigue joint is not the maximum preload the bolt can carry.
Key takeaways
- Preload is the working capacity of a bolted joint, not a by-product of tightening. Sizing it starts from the property class yield strength and the thread stress area.
- The standard target is 70% of yield: F_pre = 0.7 × R_p0.2 × A_s. Going higher trades away ductility reserve and hydrogen embrittlement margin.
- Every torque figure inherits the K factor used to produce it. Lubricated threads at K = 0.10 deliver roughly half the preload of dry threads at K = 0.20 with the same torque setting.
- Axial external load adds to the preload; transverse external load is carried by friction first, and only loads the bolt in shear once the joint slips.
- Static and dynamic joints are sized against different properties — proof load versus fatigue limit — and the higher preload that helps a static joint can shorten the life of a cyclic one.
Frequently asked questions
What is the standard preload for a bolt?
The common target is 70% of the bolt's minimum yield strength applied over the thread's tensile stress area: F_pre = 0.7 × R_p0.2 × A_s. For an M12 class 8.8 bolt with a stress area of 84.3 mm², that gives a preload of about 37,800 N. The reference tightening torque at K = 0.20 is then approximately 91 N·m.
Why is preload target 70% of yield rather than 100%?
Because preload above roughly 70% of yield leaves too little ductility reserve. A bolt at 95% of yield has almost no capacity to absorb shock or vibration before failing, and it is more exposed to hydrogen embrittlement. Targeting 70% keeps the bolt well inside its elastic range while still delivering the clamp force the joint needs.
How much preload does a transverse load require?
More than a static joint, typically by a factor of around 1.3, because the clamped members flex before full friction develops. The preload sets the friction available to resist the sliding force, and if it is insufficient the joint slips and the bolt then carries the full load in shear, which is a far less capable condition than the design assumed.
Does tightening torque actually control preload?
Only within the tolerance of the torque coefficient K. Because K varies from 0.10 to 0.30 depending on thread condition, lubrication and coating, the same torque value can deliver preload differing by a factor of two. Where preload is critical, specify a measured K established by torque-preload testing on the actual production lot, or tighten to an angle rather than a torque.
Can I increase preload on a 10.9 bolt to get more clamp force?
You can, but each increase reduces the margin. Class 10.9 has a yield ratio of 0.90, so only 10% of its tensile capacity remains as deformation reserve, and it requires embrittlement relief after plating because of the same reduced tolerance. Many designers prefer to meet a high clamp demand with a larger bolt of a lower class rather than pushing a high-strength bolt closer to its limit.
Does NXHD supply fasteners with verified preload?
Yes. Both production bases run in-house dimensional, mechanical, coating and torque inspection, and torque-preload testing on the actual production lot can be specified where preload consistency matters. Non-standard parts are produced from a customer drawing or sample with first articles submitted for approval before the production run, so preload assumptions can be checked against the real joint before volume production.
NXHD Manufacturing (contact@nxhdmfg.com) is a precision fastener and custom metal parts manufacturer with dual production bases in Handan and Datong, China. We supply standard and non-standard fasteners to OEM customers across Europe, the Americas and Asia.
Frequently asked questions
What is the standard preload for a bolt?
The common target is 70% of the bolt's minimum yield strength applied over the thread's tensile stress area: F_pre = 0.7 × R_p0.2 × A_s. For an M12 class 8.8 bolt with a stress area of 84.3 mm², that gives a preload of about 37,800 N. The reference tightening torque at K = 0.20 is then approximately 91 N·m.
Why is preload target 70% of yield rather than 100%?
Because preload above roughly 70% of yield leaves too little ductility reserve. A bolt at 95% of yield has almost no capacity to absorb shock or vibration before failing, and it is more exposed to hydrogen embrittlement. Targeting 70% keeps the bolt well inside its elastic range while still delivering the clamp force the joint needs.
How much preload does a transverse load require?
More than a static joint, typically by a factor of around 1.3, because the clamped members flex before full friction develops. The preload sets the friction available to resist the sliding force, and if it is insufficient the joint slips and the bolt then carries the full load in shear, which is a far less capable condition than the design assumed.
Does tightening torque actually control preload?
Only within the tolerance of the torque coefficient K. Because K varies from 0.10 to 0.30 depending on thread condition, lubrication and coating, the same torque value can deliver preload differing by a factor of two. Where preload is critical, specify a measured K established by torque-preload testing on the actual production lot, or tighten to an angle rather than a torque.
Can I increase preload on a 10.9 bolt to get more clamp force?
You can, but each increase reduces the margin. Class 10.9 has a yield ratio of 0.90, so only 10% of its tensile capacity remains as deformation reserve, and it requires embrittlement relief after plating because of the same reduced tolerance. Many designers prefer to meet a high clamp demand with a larger bolt of a lower class rather than pushing a high-strength bolt closer to its limit.
Does NXHD supply fasteners with verified preload?
Yes. Both production bases run in-house dimensional, mechanical, coating and torque inspection, and torque-preload testing on the actual production lot can be specified where preload consistency matters. Non-standard parts are produced from a customer drawing or sample with first articles submitted for approval before the production run, so preload assumptions can be checked against the real joint before volume production.
