Bolted Flange Joints: Bolt Torque, Sequence and Leak Prevention covers assembling gasketed flange connections so bolt load is even, correctly sized, and applied in the right order. Most in-service flange leaks trace back to assembly, not design: uneven bolt load, the wrong sequence, or a missed re-torque after startup thermal cycling.
A flange seals because the gasket is compressed enough to fill surface irregularities and hold contact stress higher than the internal pressure pushing it apart. That compression comes entirely from bolt tension. Uneven bolt load leaves some points under-compressed and leaking, while others are over-compressed and can crush the gasket or damage the flange face. Bolt load must seat the gasket, resist internal pressure trying to separate the flanges, and retain enough residual load through thermal cycling and gasket creep to stay tight. The goal is uniform tension around the full circle, not simply "tight."
Bolts are never tightened round-the-clock, which cants the flange as one side is driven down first. Instead, use a star pattern applied in progressive passes:
Larger flanges (roughly 300 mm / 12 in and above, or high bolt counts) often need four or five passes rather than three.
The quantity that actually seals the joint is bolt tension (preload), not torque. Torque is only an indirect way to reach tension: most of it is consumed overcoming thread and nut-face friction, with only a modest fraction converting into useful bolt stretch. Friction varies with thread condition, lubricant, and technique, so torque-only control carries meaningful bolt-to-bolt scatter at an identical wrench setting. On critical joints (hydrogen, high-temperature steam, large high-pressure flanges) that scatter is unacceptable, and torque wrenching is replaced or supplemented by hydraulic bolt tensioners (which stretch the bolt directly, largely removing friction), bolt-stretch measurement with ultrasonic gauges or extensometers, or torque-turn methods, where a snugged nut is rotated a specified angle instead of driven to a torque value.
Lubrication directly controls how much applied torque converts to bolt tension: a dry, rusted thread carries much higher friction than a lubricated one, so the same torque setting produces less clamp load. Always lubricate the threads and nut-face bearing surface with the lubricant the torque value was calculated for, and never swap lubricants mid-job, since that silently changes the torque-to-tension relationship.
Gaskets also relax under sustained compression, and flanges and bolts expand at different rates during heat-up. On hot or high-pressure services, standard practice is to hot torque or re-torque the joint once it reaches normal operating conditions, since cold bolt load at assembly is not what remains after cycling and creep relaxation. Skipping re-torque is a common cause of leaks appearing weeks after a turnaround, after the joint already passed its cold hydrotest.
Bolt torque cannot be specified correctly without knowing the gasket. Gasket types (spiral wound, PTFE envelope, compressed non-asbestos fiber, metal ring joint) have different minimum and maximum allowable stress before crushing, and target bolt load must sit between those limits. Selecting the wrong gasket for the flange rating, temperature, or service, or reusing a gasket, undermines even a perfectly executed torque procedure. See gasket selection for rating guidance, and pressure relief valve sizing for how overpressure protection interacts with joint integrity.
ASME PCC-1, "Guidelines for Pressure Boundary Bolted Flange Joint Assembly," is the primary industry reference for bolt pattern, pass sequencing, lubricant documentation, and tensioner versus torque-wrench selection. It does not replace the flange or gasket design standard, such as ASME B16.5 for flange dimensions and ratings; it governs how a joint built to those standards is assembled in the field.
| Bolt size (metric) | Typical grade | Illustrative target torque, lubricated (N·m) | Governing standard |
|---|---|---|---|
| M12 | ASTM A193 B7 | 55 to 85 | ASME PCC-1 / B16.5 |
| M16 | ASTM A193 B7 | 135 to 205 | ASME PCC-1 / B16.5 |
| M20 | ASTM A193 B7 | 265 to 400 | ASME PCC-1 / B16.5 |
| M24 | ASTM A193 B7 | 460 to 690 | ASME PCC-1 / B16.5 |
| M30 | ASTM A193 B7 | 910 to 1365 | ASME PCC-1 / B16.5 |
These are illustrative ranges only, not a substitute for a joint-specific calculation from gasket seating stress, bolt stress area, and the nut factor for the lubricant used.
Because torque-to-tension accuracy depends on lubricant, sequence, and pass count all being followed correctly, flange assembly is a case where paper checklists lose detail. Logging the bolt pattern, torque achieved pass by pass, lubricant lot, and re-torque date against the specific flange asset in a maintenance system such as Fabrico lets reliability teams trace a leak back to a specific assembly event. A misaligned coupling or pipe run can also strain a flange that no torque setting will fix; see coupling alignment. Book a Fabrico demo to see how bolting records tie into asset history and inspection scheduling.
Round tightening compresses one side of the gasket before the other, canting the flange and leaving uneven contact stress. The joint can pass a cold hydrotest and still leak once pressure and temperature redistribute load onto the under-compressed side.
Torque converts to tension mostly by overcoming thread and nut-face friction, which depends on lubricant condition, thread damage, and surface finish. Bolts torqued identically can end up with different clamp load if friction conditions differ.
No. A documented, lubricated, star-pattern, multi-pass torque procedure is adequate for most standard-duty flanges. Tensioning or elongation measurement is reserved for large-diameter, high-pressure, or safety-critical joints where torque scatter cannot be tolerated.
Practice varies by service, but re-torque is typically done once the system has stabilized at normal operating temperature and pressure, after the gasket's initial creep relaxation, following the site's documented procedure rather than a fixed universal time.