Torque and bolting

Why a Flange Leaks After Tightening: 6 Causes and How to Prevent Them

Flange leak after tightening? The gasket is rarely the real culprit. According to the Fluid Sealing Association (2008), gasket makers estimate that “perhaps” 75 to 85% of leaking bolted flanges have a cause other than the gasket, mostly installation and assembly. Uneven preload, variable friction, a gasket that relaxes or a damaged face is enough. Here are the 6 most common causes and how to prevent each one.

The stakes are real. The Energy Institute (2021) notes that in the oil and gas industry, leaking joints are a leading cause of hydrocarbon releases. And according to a master’s thesis published at Montreal’s ÉTS in July 2026, a refinery can have more than 100,000 bolted joints.

Flange leak after tightening: the 6 causes at a glance

When a flange leaks even though it was tightened “to the right torque,” the cause usually falls into one of these six families. Several often add up on the same joint.

  1. Uneven preload from one bolt to the next (wrong sequence or too few passes).
  2. Variable friction, mostly due to lubrication.
  3. The gasket settling over time (creep relaxation).
  4. Damaged or misaligned flange faces.
  5. The wrong gasket, or a reused gasket and studs.
  6. Thermal cycles and vibration in service.

A useful reminder before getting into the details: what seals a flange is preload, the force with which the tensioned studs crush the gasket. Torque is only the turning effort applied to the nut to get there.

1. Uneven preload: the tightening sequence matters

Tightening bolts one after the other, in a single pass, crushes the gasket unevenly. Each bolt you tighten compresses the flange a little more, and its already-tightened neighbours lose some tension. This is called elastic interaction. You need several passes, in a specific order.

Quebec research is very active on this topic. Professor Hakim Bouzid’s team at ÉTS has studied elastic interaction for years (Zhu, Bouzid and Hong, Journal of Pressure Vessel Technology, 2017). In his 2026 thesis, Thomas Joly notes that the sequential method, bolt after bolt around the flange, is simple to apply but gives unsatisfactory results. A study he cites reports that bolt-to-bolt scatter drops by up to five times when going from a single pass to several.

How to prevent it:

  • Follow a star pattern or one of the ASME PCC-1 patterns, in several progressive passes.
  • Finish with at least one circular pass at final torque.
  • Work from a written procedure with trained assemblers. PCC-1 Appendix A describes their qualification.

2. Variable friction: lubrication makes the difference

At the same torque, preload can vary by more than three times depending on friction. NASA’s Fastener Design Manual (1990) shows that the nut factor K, which links torque and preload, ranges from 0.074 to 0.250 depending on the friction coefficient. A dry, rusty or poorly lubricated thread “eats” torque before it stretches the stud.

The Fluid Sealing Association adds that a lack of lubrication can reduce the achieved load by as much as 50%. The same NASA manual gives the typical accuracy of each method:

Method Preload accuracy
By feel ± 35%
Torque wrench ± 25%
Turn of nut ± 15%
Bolt elongation measurement ± 3 to 5%

How to prevent it:

  • Specify a lubricant in the procedure and apply it the same way everywhere: on the threads and under the nut, never on the gasket.
  • Use the nut factor that matches that lubricant to calculate torque. Our guide to reading a HYTORC torque chart explains the pressure-to-torque conversion.
  • Have your tools calibrated. See how often, and what the certificate must show.
  • For critical joints, choose a method that depends less on friction, such as hydraulic tensioning.

3. The gasket settles: creep relaxation

Right after tightening, the gasket starts to compress slowly and the studs lose tension. This is creep relaxation. Researchers at Polytechnique Montréal (Bouzid, Chaaban and Bazergui, 1995) showed that it significantly reduces the load remaining on the gasket, and that it is one of the main causes of lost tightness.

In his 1994 PhD thesis at Polytechnique, Hakim Bouzid reports that a 1/8 in. PTFE gasket lost about 30% of its load in a five-hour test, and that most relaxation happens in the first 15 minutes. Tests presented at the 2021 ASME PVP conference on a 4 in. flange confirm the effect of retightening:

Gasket type Loss without retorque Loss with 24 h retorque
Expanded PTFE 12.6% 3.0%
Restructured PTFE 12.5% 4.4%
Skived PTFE 20.0% 8.0%

For expanded PTFE, the same tests show that an under-loaded gasket relaxes 2.4 times more. A timid tightening makes the problem worse.

How to prevent it:

  • Choose a low-creep gasket for the service (spiral wound and kammprofile gaskets relax much less than PTFE).
  • Reach the right seating stress: the minimum pressure that lets the gasket seal.
  • Do a relaxation pass after a dwell time, as your procedure requires.

4. Damaged or misaligned flange faces

A deep radial scratch on the flange face can create a leak path, even with good tightening. According to the “Sealing Sense” column in Pumps & Systems (2013), the most common damage comes from scraping off the old gasket. Use a brass brush, never a cold chisel or a screwdriver.

Alignment matters just as much. When two flanges are misaligned, the bolts end up pulling the piping into place instead of crushing the gasket. Studs should pass freely through the holes. ASME PCC-1 provides tolerances for defects, flatness and alignment.

How to prevent it:

  • Inspect and measure the faces before every assembly: scratches, pits, dents, flatness.
  • Correct alignment with the piping and supports, not with the bolts.
  • Remachine or replace an out-of-tolerance face.

5. The wrong gasket, or a reused gasket and studs

A gasket is not reusable. Energy Safety Canada said so plainly in 2026 after a gasket blew out on a high-pressure produced water system, spilling about 4 m³ of sour water. Its first lesson is clear: “Do not reuse gaskets.” The gasket must also suit the fluid, the temperature and the pressure.

A Belgian case published by the Federal Public Service Employment (2019, in French) shows how causes add up. At the startup of a tower, a manhole gasket gave way and released 3 to 4 tonnes of hot flammable liquid. The old gasket had been reused on a larger bearing surface. No torque calculation had been done, the nuts were not lubricated, and the tool used could not deliver the required torque.

Studs wear out too. The Fluid Sealing Association (2008) notes that reused studs may be corroded or stretched past their yield point by overtightening: they then stay permanently elongated.

How to prevent it:

  • A new gasket every time the joint is opened, of the right type for the service.
  • A well-centred gasket (centering ring), with no adhesive tape on the seating area.
  • Inspected studs and nuts: clean threads, a nut that turns freely by hand.
  • A tool that can deliver the required torque or tension. If you don’t have one, renting a bolting tool avoids improvising.

6. Thermal cycles and vibration

A flange that heats and cools does not keep the same preload. If the flange and studs are different metals, they don’t expand by the same amount. Engineering firm Becht (Antaki and Lundquist, 2018) gives the example of a stainless flange with B7 studs at 600 °F (about 315 °C): the stress in the studs exceeds their yield strength. When the line cools, they stay stretched and the gasket is less compressed.

Bouzid’s thesis (1994) adds that load cycles make the gasket creep more than a constant load, especially in the first 25 cycles. Vibration, for its part, can gradually loosen a nut, a phenomenon described as early as 1969 by engineer Gerhard Junker.

How to prevent it:

  • Choose studs that are compatible with the flange at temperature.
  • Aim for a preload that leaves margin below yield at operating temperature.
  • Plan a startup retorque if your procedure allows it.
  • Against vibration: high, uniform preload, proper supports and regular inspection.

Should flange bolts be retorqued after startup?

Often, yes, but according to the procedure. ASME PCC-1 provides for a relaxation pass after a dwell time, and now uses the term “startup retorque” rather than “hot torque.” The 2021 tests show that retightening can cut load loss by more than half, depending on the gasket.

Don’t confuse this retorque with replacing a stud under pressure, which is a separate, specifically controlled operation. And always walk the lines at startup, as Energy Safety Canada recommends for turnarounds.

What is the difference between torquing and tensioning?

Torquing turns the nut against friction; tensioning stretches the stud directly. With torque, a large part of the effort is lost to friction, and preload varies with the lubricant and thread condition. Hydraulic tensioning pulls on the stud, then the nut is run down: preload depends much less on friction.

Torque remains the most widely used method, as long as it is done right. For critical joints or large diameters, our team can help you choose the right HYTORC approach. For examples of what happens when bolting goes wrong, read our real flange leak stories.

Checklist before closing a flange

  • Faces cleaned without steel tools, inspected and measured.
  • Flanges aligned, studs passing freely.
  • New gasket, of the right type, well centred.
  • Studs and nuts inspected, clean threads.
  • Lubricant specified, applied to threads and under the nut.
  • Torque calculated with the nut factor of the lubricant used.
  • Calibrated tool, able to deliver the required value.
  • Multi-pass sequence, then a final circular pass.
  • Relaxation pass and startup walkdown scheduled.

Sources

Talk to an expert now!

Give us a call. We are friendly, too. 1-844-266-2632 or send us a note.