A steam header rises from 20 °C to 200 °C over four hours. The flange was torqued cold, the wrench clicked and the hydro test passed. Within a shift the spiral-wound gasket is weeping at the outer ring, and the wrench still clicks.
Cold torque was correct; operating preload was not. Bolt, flange and gasket expand differently, and the gasket does not spring back as much as it was compressed.
Three rules:
- Cold torque sets assembly preload, not operating preload.
- Hot tightening is a measured operation at a stable temperature, not a second pull on the wrench.
- If the gasket is PTFE, a bolt has yielded, or the leak is growing, the hot pass is not the next step.
Why cold torque fails on a hot flange
Three mechanisms move preload after startup.
Differential expansion comes first. In an insulated line the flange ring is usually hotter than the bolt shank. The flange grows more, the joint opens and the bolt loses stretch.
Assumptions: 200 mm grip length, E = 210,000 MPa, 20 °C to 200 °C, carbon steel α = 12×10⁻⁶/K, bolt 30 K cooler than the flange. Convert 0.072 mm to force:
26.6 kN is about 11% of the 247 kN target used later for an M24 10.9S bolt.
A stainless bolt (α ≈ 17×10⁻⁶/K) in a carbon steel flange at the same temperature reverses the sign: it grows 0.612 mm against 0.432 mm, adding about 66.6 kN, or 27% of target.
| Material | α, typical (10⁻⁶/K) | Length | Expansion at ΔT = 180 K |
|---|---|---|---|
| Carbon steel bolt | 12 | 200 mm | 0.432 mm |
| Stainless steel bolt | 17 | 200 mm | 0.612 mm |
| PTFE gasket | 120 | 3 mm | 0.065 mm |
Order-of-magnitude only; real flanges have gradients and restraint.
Bolt temperature lag is second: the bolt is a fin with one end in the flange. One probe on the ring does not give the bolt temperature; add one on the shank or nut. If the bolts are boxed in with insulation, the lag shrinks and the direction can reverse.
Gasket creep is third. At temperature the binder softens, the filler creeps and the thickness changes. A gasket that recovers returns load to the bolt; one that creeps does not. The bolt can be at the correct torque and the joint can still lose preload because the gasket got thinner.
What hot tightening actually is
Hot tightening is a controlled re-loading after the joint reaches a stable operating temperature, using a measured temperature and a recorded torque or elongation. It is not a walk around a hot flange.
Temperature is measured at defined points, not estimated from product temperature. It is stable: one field criterion is two readings at each point, 30 minutes apart, within 10 K. Torque or elongation is recorded against temperature and pass number.
Insulation matters. A flange measured with the insulation removed is not at its running temperature. Put the insulation back, let the line soak, then read.
Why not by feel? A torque wrench measures torque, not bolt force:
K changes with thread friction, lubrication, galling and contact pressure; at temperature it can move outside the cold-assembly value. A hot bolt that feels tight can be below target; one that feels loose can be at target. A hot wrench is not calibrated.
Standards treat K as a batch property. GB/T 1231-2024 §6.3.1 requires a mean torque coefficient of 0.110–0.150 with a coefficient of variation no greater than 0.077, on a calibrated axial-force device; its batch test uses eight assemblies.
Hot tightening is outside room-temperature assembly standards. GB/T 32076.10-2018 §8.3.2 sets the initial and re-tightening torque at 50% of final and expects the passes within 24 h; GB/T 32076.2-2015 §5 tests at 10–35 °C. Use them for mechanics, not as a hot-pass permit.
The temperature window and hard stops
The window has two ends; the lowest upper limit wins:
- Lower end: thermal stability, confirmed by the two-reading check.
- Upper end, gasket: the manufacturer’s maximum service temperature for that type and thickness.
- Upper end, bolt: the design-code allowable, with the hot proof load still above target preload by the design margin.
- Upper end, material: a fire or excursion can take a bolt past its tempering range. GB/T 3098.1-2010 §7 states the property-class requirements apply at ambient temperature; the 10.9 minima (Rm 1,040 MPa, Rp0.2 940 MPa) are not hot allowable stresses.
Prohibited conditions:
- Above the gasket’s maximum service temperature.
- A bolt has yielded or grown permanently.
- The leak has grown after a pass, or is a jet.
- Toxic, flammable or asphyxiant medium without isolation, depressurization, purge and gas test.
- No joint temperature reading, or only a reading on the insulation.
- Unknown bolt grade or gasket type.
- Overheated flange with no metallurgical review.
- Spring hangers, expansion joints or restrained pipework without an engineering review.
If any is true, the hot pass is a shutdown job, not a procedure.
Sequence and passes that work
Sequence follows stiffness. GB/T 32076.10-2018 §8.3.4 starts at the stiff, constrained side and works toward the freer side, or starts at the center of the bolt group and works outward; near a locating pin, start closest to the pin. On a flange, start at the restrained quadrant, then cross to the opposite.
Use a star pattern and 2–3 passes:
- Pass 1 at 50% of target, following the 50% initial-pass anchor in GB/T 32076.10-2018 §8.3.2.
- Pass 2 at 80% of target. A field convention, not a standard value; it keeps one side of the gasket from being crushed before the other bolts load.
- Pass 3 at 100% of target, same pattern.
After pass 3, re-check the first four bolts. Tightening one bolt compresses the flange locally and lets its neighbors relax; the first bolts usually drop. Record the re-check. If the drop is large, add a pass at 100%, not a higher torque.
Adjacent bolt interaction is why passes are needed; the first bolt is not at the same preload after the last. Measure the drop on the first joint and use it to set the re-check pattern.
Preload, torque and K-factor drift
The torque-preload relation is the same one used in torque, preload and the K-factor:
For an M24 10.9S assembly, the preload target from the GB/T 32076 family is 0.7 Rm As (GB/T 32076.2-2015 §4). The grade is an example; the standard covers structural bolting, not flanges.
It sits in the 237–263 kN M24 10.9S range in GB/T 1231-2024 Table 14.
The same 770 N·m gives different preloads:
| K | Torque for 246.8 kN (N·m) | Preload at 770 N·m (kN) |
|---|---|---|
| 0.11 | 652 | 292 |
| 0.13 | 770 | 247 |
| 0.15 | 889 | 214 |
GB/T 1231-2024 §6.3.1 allows a batch mean K of 0.110–0.150 with VK ≤ 0.077. The range is from a structural bolting standard, not a flange value; it alone moves preload from about 13% below to 18% above the 247 kN nominal. At temperature K can move outside it. A hot torque needs the K or the elongation that produced it.
Converting hot torque back to cold requires K at the hot condition, which you usually do not have. Record three numbers: cold initial torque, hot re-tightening torque, and the temperature at which it was applied. Record bolt elongation if possible; it is K-independent.
An over-torqued bolt can fail in the engaged threads first; see thread stripping and shear area.
Gasket behavior decides the procedure
Gaskets do not share one hot-tightening rule.
| Gasket type | Creep at temperature | Recovery | Hot re-tightening |
|---|---|---|---|
| Spiral wound, graphite filler | Moderate; filler creeps, winding stays elastic | Good | Within the manufacturer’s limit; re-check after pass 1 |
| Flexible graphite sheet | High; the sheet takes a set | Low | Limited; if it has taken a set, replace |
| PTFE / ePTFE | High cold flow; the material thins under load | Low | Not by metal-gasket logic; often no hot re-torque |
| Metal jacketed / RTJ | Low creep; the metal carries the load | High | Only per design; torque is not the seal parameter |
PTFE is the clear case. Its linear expansion coefficient is about 120×10⁻⁶/K, ten times that of carbon steel. A 3 mm PTFE gasket grows about 0.065 mm over a 180 K rise; the flange and bolt grow by a different amount, and the gasket also creeps. Apply the metal-gasket rule at temperature and you can thin the gasket, open the leak path and leave a joint that leaks again on cooldown. PTFE joints need the manufacturer’s re-torque policy, often with a spring-loaded or Belleville washer stack.
A spiral winding recovers; a graphite sheet does not.
RTJ is metal-to-metal: bolt load seats the ring, and extra torque can deform the groove.
Records and acceptance
An unauditable record is a torque log, not evidence.
| Field | Why it is there |
|---|---|
| Time and date | Links pass to temperature history |
| Line and flange tag | Identifies the joint, not just the shift |
| Flange ring temperature at each point | Input to the expansion estimate |
| Bolt shank or nut temperature | The lag term |
| Bolt position or number | The sequence and re-check pattern |
| Pass number | 1, 2, 3 or re-check |
| Torque and instrument ID | Measured value and calibration |
| Bolt elongation, if measured | K-independent evidence |
| Leak observation before and after | Stable, reduced or grown |
| Operator and reviewer | Accountability and handover |
Acceptance at temperature: no visible leak at operating conditions for the observation period; bolt elongation within the elastic range; gasket compression within the manufacturer’s range; record signed by the operator and reviewed by the responsible engineer.
At the next shutdown, do not use hot torque as the cold baseline; it includes thermal strain and gasket creep. As it cools, bolt and flange contract differently and the gasket may recover or take a set. Ambient re-torque is new. Compare it with the original cold value and compare bolt length with the pre-hot-pass length. Replace a bolt with permanent elongation; replace a gasket that has taken a set. Hot torque is history; the cold measurement is the next baseline.
Safety before the wrench moves
Hot tightening puts a person with a wrench on a pressurized, hot joint.
- Burns: flange and bolts are at operating temperature. Insulated gloves, face shield, long sleeves, platform. Do not work from a ladder with a long wrench.
- Bolt fracture: a bolt at or beyond yield can break under the next pull. Stand to the side of the bolt axis, not in line. Use a break-over wrench, not a pipe extension.
- Medium: steam flashes, hot oil ignites, toxic gas does not announce itself. Isolate, depressurize, purge, gas-test and use the permit system. A growing leak is a stop signal, not a reason to add torque.
- Thermal shock: do not water-cool the flange; the gradient can crack a casting or distort a face.
Confirm wrench calibration and ID before the first pass; a hot joint is a bad place to find a wrench reading low.
The decision, in one page
Cold torque is the assembly number. Operating preload is set by temperature, gasket and bolt-group interaction. Hot tightening is worth doing only when the joint is at a stable measured temperature, the gasket is within its rating, the bolts are elastic and the leak is not growing. The record is the product: time, temperature, bolt number, torque, elongation and who applied it. A failed pass means downtime.
Before the next hot pass, gather the line tag, flange and bolt temperatures, gasket type and limit, bolt grade, target preload and cold torque. An engineer can check the window and sequence against the design file.
References
- GB/T 32076.10-2018 — Assembly specifications — §8.3.2, §8.3.4
- GB/T 1231-2024 — High-strength bolted assemblies — §6.3.1, Table 14
- GB/T 32076.2-2015 — Suitability for preloading — §4, §5
- GB/T 3098.1-2010 — Mechanical properties — §7
- ASM Handbook, Vol. 1 — typical linear expansion coefficients for carbon and stainless steels
- Machinery's Handbook, 31st ed. — torque-preload relation and K-factor ranges
Key Standards
Standards and clauses referenced by this article.
- High-strength structural bolting assemblies for preloading — Part 10: Assembly specifications§8.3.2, §8.3.4
Initial and re-tightening torque at 50% of final torque, passes within 24 h; sequence from the constrained side to the freer side or from the center of the bolt group outward. Scope is room-temperature assembly, not hot tightening.
- High-strength large hexagon head bolted assemblies for steel structures§6.3.1, Table 14
Torque coefficient mean 0.110–0.150 and coefficient of variation ≤0.077; M24 10.9S preload test range 237–263 kN. Structural bolting standard, used here for the K-factor scatter.
- High-strength structural bolting assemblies for preloading — Part 2: Suitability for preloading§4, §5
Specified preload Fp = 0.7 Rm As; suitability test at 10–35 °C; test stops when axial force falls to Fp or the bolt fractures.
- Mechanical properties of fasteners made of carbon steel and alloy steel — Bolts, screws and studs§7
Property-class requirements apply at ambient temperature; 10.9 minimum values Rm 1,040 MPa and Rp0.2 940 MPa are not hot allowable stresses.
References
Sources used for fact checking and background context.
- GB/T 32076.10-2018 — Assembly specifications — §8.3.2, §8.3.4standard
- GB/T 1231-2024 — High-strength bolted assemblies — §6.3.1, Table 14standard
- GB/T 32076.2-2015 — Suitability for preloading — §4, §5standard
- GB/T 3098.1-2010 — Mechanical properties — §7standard
- ASM Handbook, Vol. 1 — typical linear expansion coefficients for carbon and stainless steelsbook
- Machinery's Handbook, 31st ed. — torque-preload relation and K-factor rangesbook
Deep Reading
More systematic selection, procurement, or inspection guides.
- ASME PCC-1-2022 — Guidelines for Pressure Boundary Bolted Flange Joint Assembly
A systematic flange-joint assembly guideline; use it with the project specification and the gasket manufacturer's data.
- EN 13555:2014 — Flange gaskets and gasket material parameters
Defines the gasket parameters used to calculate required bolt load; read before setting a hot re-torque value.
This page is one entry in the fastener engineering reference. The library covers standards, dimensions, materials, and failure analysis; every figure is traced back to its source.
Continue reading in the library →