Preload is what tightening actually buys: the tension locked into the bolt, clamping the joint. The target is set as a fraction of proof load — proof load Fp = At·Sp from the tensile stress area and the property class’s proof strength, with about 75% of it the common target for reusable joints. That level develops a high, stable clamping force while leaving margin below proof and yield, keeping the joint tight under service loads without yielding the bolt; permanent single-use joints are sometimes taken to 85–90% of proof.
Torque is the imperfect messenger that delivers it. The short-form equation T = K·F·d converts a preload target into a wrench setting through the nut factor K, an empirical coefficient lumping thread friction and under-head bearing friction: about 0.20 for plain as-received steel, 0.16–0.18 lubricated, 0.10–0.12 with anti-seize — lower friction means more of the torque becomes preload. Tightening an M16 class 8.8 to 75% of proof (a preload of 68,150 N from At = 156.7 mm² and Sp = 580 MPa) at K = 0.20 takes T = 0.20 × 68,150 × 0.016 = 218.1 N·m, leaving the bolt at 435 MPa, about 68% of its 640 MPa yield.
The catch is scatter: because K is sensitive to surface finish, lubrication, and seating, torque-controlled preload typically lands within ±25–35% of the target. Critical joints therefore graduate to measured bolt elongation, turn-of-nut, or a calibrated tensioner, with VDI 2230 as the framework for a full bolted-joint analysis.