Choose a bolt preload, estimate tightening torque, and check a tension joint for bolt load and separation.
Why it matters
Machines are held together by bolts, screws, pins, rivets, welds and adhesives. Bolted joints are everywhere because they can be taken apart for maintenance. They are also a common cause of failures: joints that loosen, bolts that snap in fatigue, flanges that leak. Most of those failures trace back to one idea people skip: preload.
How a bolted joint really works
When you tighten a nut, the bolt stretches like a stiff spring and the clamped parts (the members) squeeze together. The tension in the bolt after tightening is the preload, Fi. A good joint keeps the members clamped under every load it sees.
When an external load P tries to pull the joint apart, it does not all go into the bolt. The bolt and the members share it according to their stiffness. The joint constant C is the bolt's share:
C = kb / (kb + km) kb = bolt stiffness, km = member stiffness
bolt tension Fb = Fi + C * P
member clamp Fm = Fi - (1 - C) * P
Members are usually much stiffer than the bolt, so C is often around 0.2 to 0.3. That is the key insight: in a well-preloaded joint, the bolt only feels a small part of the external load. The members simply unload. If P grows until the clamp force reaches zero, the joint separates, and from then on the bolt carries the whole load and every load cycle hits it fully. That is how bolts fail in fatigue.
Choosing a preload
A common textbook recommendation for reusable joints is a preload of about 75 % of the proof load:
Fi = 0.75 * At * Sp
At = tensile stress area of the thread
Sp = proof strength of the bolt's property class
For an M10 x 1.5 bolt, At = 58.0 mm². For property class 8.8 at this size, ISO 898-1 gives a proof stress of about 580 MPa (check the current standard for your size and class):
Fi = 0.75 * 58.0 * 580 = 25,230 N = 25.2 kN
From preload to tightening torque
In the shop you rarely measure preload; you set a torque. A simple estimate is:
T = K * Fi * d d = nominal diameter, K = nut factor (about 0.2 for dry steel)
T = 0.2 * 25,230 * 0.010 = 50.5 N·m
Be careful: K depends on friction, so it scatters a lot. Lubricant, plating, thread condition and washers can change the preload from the same torque by roughly 25 % or more either way. Critical joints use the fastener maker's torque data, torque-angle methods, direct tension indicators or ultrasonic bolt measurement.
Worked example: a tension joint
The M10 bolt above holds a cover that sees an external separating load of P = 8 kN per bolt, and C = 0.25.
bolt tension Fb = 25.23 + 0.25 * 8 = 27.23 kN
bolt stress = 27,230 / 58.0 = 469 MPa (below the 580 MPa proof stress)
yield factor nL = (Sp*At - Fi) / (C*P) = (33,640 - 25,230) / 2,000 = 4.2
separation n0 = Fi / (P * (1 - C)) = 25,230 / 6,000 = 4.2
Both factors are comfortable here. Now imagine the bolt was only snugged to 5 kN preload: n0 = 5,000 / 6,000 = 0.83, so the joint separates, and the bolt sees the full 8 kN swing on every cycle. Same bolt, very different life.
Shear joints
Many joints carry load sideways, such as a bracket bolted to a frame. A preloaded joint resists slip by friction: the slip resistance is roughly mu x Fi x (number of bolts) x (number of friction surfaces), with mu often around 0.2 to 0.5 depending on the surface condition. If it slips, the bolt shanks bear on the hole edges, which is a different (and usually worse) way to carry load. Dowel pins are often added to locate parts precisely so the bolts only clamp.
Practical details that matter
- Thread engagement. In a tapped steel hole, engage roughly one diameter; softer materials such as aluminium need more (often 1.5 to 2 diameters) or threaded inserts.
- Washers spread the load under the head and protect soft surfaces.
- Loosening comes mainly from transverse slip under vibration. Adequate preload is the first defence; thread-locking adhesive and prevailing-torque nuts help. Spring (split) washers are not reliable for this.
- Torque-to-yield bolts (common in engines) are designed to be stretched past yield and must not be reused.
Common mistakes
- Using the nominal area. Bolt stress uses the tensile stress area At, not pi·d²/4.
- Trusting torque as if it were preload. Torque is an indirect, scattered measure.
- Mixing grades. Class 8.8 and 10.9 (or SAE Grade 5 and 8) look similar; always check the head marking.
- Under-tightening "to be safe". Low preload is what lets a joint separate and fail in fatigue.
A note on real design
Real joints follow fastener standards (ISO, ASME, SAE), the manufacturer's data and the design rules of the industry involved (pressure vessels, structures and aerospace each have their own). Use these calculations to understand joints and sanity-check them; safety-critical joints need review by a qualified engineer, and where public safety is involved, a licensed professional engineer.
Recap
- Preload clamps the joint; the bolt sees only C·P of an external load.
- Target about 75 % of proof load for reusable joints; T = K·Fi·d estimates torque.
- Separation factor n0 = Fi / (P(1 − C)) must stay above 1 with margin.
- Low preload is the usual root cause of loosening and bolt fatigue.
Try it
On the Do tab, finish the separation-factor function and test how preload and joint stiffness change it.