Why Torque Is Not Clamp Force: How Much Torque Friction Eats

Many think "torque times a coefficient is clamp force", but when actually tightening a bolt, the torque T you apply is split into three parts: overcoming thread-pair friction, overcoming bearing-surface (head/nut washer-face) friction, and only the remainder converting into axial clamp force F_f. Rough engineering experience—only about 10% of the torque truly becomes clamp force, while about 40% is spent on thread friction and about 50% on bearing-surface friction. This means a small fluctuation in friction coefficient causes a large fluctuation in clamp force; it is the root cause of "same torque, different preload".

The three are related by the torque coefficient K: F_f = T / (K·d) (d is nominal diameter). The larger K, the smaller the clamp force produced at the same torque; the more K scatters, the less controllable the clamp force. GB/T 16823.2 (general rules for tightening threaded fasteners) is precisely built on this relationship to guide engineers in choosing the tightening method and control strategy. A practical consequence: if two apparently identical batches of zinc-plated bolts give clamp forces differing by up to 80% at the same target torque, the cause is almost never a wrong torque wrench but a changed friction state—thicker passivation, residual forming oil, or a different plating bath—and only a fresh GB/T 16823.3 measurement will reveal it.

The GB/T 16823.3 Test Method: How to Build the Calibrated Joint

The core of the torque-clamp-force test specified by GB/T 16823.3 is building a calibrated test joint that can measure axial force: screw the specimen bolt into a standard joint fitted with a load cell (axial-force sensor / load washer), apply torque with a torque wrench or automatic tightening machine while synchronously recording torque T and the resulting axial clamp force F_f, until the target preload or near-yield is reached. The test runs under the bearing-surface conditions specified in the standard, and the lubrication/coating state of the tested surfaces (thread, bearing face) must match the actual assembly. Repeating the measurement under different conditions yields the torque-clamp-force curve and friction coefficients for this batch of fasteners in this surface state.

Three Friction Coefficients: μ_th, μ_b and the K Factor

A complete torque-clamp-force test yields not just one K but breaks friction apart:

The value of breaking it apart is this: if you find K is high and clamp force won't come up, you must distinguish whether thread friction or bearing-surface friction is high. High thread friction calls first for thread lubrication/coating changes; high bearing-surface friction calls first for checking the washer-face state, coating buildup or burrs. Blanket-adding lubricating oil without discrimination often gives poor returns for the effort.

Typical Data Ranges: How Coatings and Lubrication Change K

Friction coefficients differ markedly across surface states, which is exactly where test data is most useful:

Note K is an "overall coefficient", not a physical constant—it varies with thread diameter, pitch, bearing-surface diameter and friction state, so you cannot copy someone else's K onto your own joint. This is exactly why critical joints (engine, chassis, battery pack) must run GB/T 16823.3 on their own parts. The numbers in the table are rough reference ranges only; they shift with coating thickness, bolt hardness, surface roughness of the clamped material, and even whether the test joint is lubricated dry or with the assembly oil that will actually be used on the line.

Engineering Meaning: From Test Data to Tightening Process

• Set the torque value: back-calculate the torque T = K·d·F_f for the target clamp force using measured K, rather than looking it up in an experience table.
• Assess dispersion: the standard deviation of K directly decides clamp-force scatter; if scatter is too large, upgrade to torque-angle method or yield-point control.
• Choose surface treatment: compare μ_th and μ_b across coatings/lubricants and pick the low-and-stable-friction option.
• Quality gate: bring torque-clamp-force sampling into incoming inspection, preventing coating batch variation from quietly changing preload.

For a high-volume automotive or battery-pack line, the practical workflow is: first characterize each fastener/coating/lubrication combo on a calibrated load cell, record the mean K and its standard deviation, then set the production torque T = K·d·F_f with a safety margin that accounts for scatter; if the one-sigma scatter of K is large enough that the low-preload tail risks joint separation, switch from pure torque to torque-angle or yield-point tightening. Re-run the measurement whenever the coating supplier, passivation chemistry or lubricant changes—treating K as a one-time qualification is the most common way these programs silently drift out of control.