How a Gasket Actually Seals

Why this matters

A gasket does not seal because it is squashed. It seals because a certain amount of force is pressed onto every square inch of its face, and that force stays higher than the pressure trying to push the joint open. Once you think in force per square inch instead of "tight enough," a whole class of field problems stops being mysterious: why the same bolt load seals a narrow gasket and leaks a wide one, why a big low-pressure flange is harder to seal than a small high-pressure one, and why adding torque to a weeping joint sometimes fixes it and sometimes finishes it off.

This card is about the arithmetic of that force. A sibling article covers what physically fills the scratches in a flange face, and another covers reading a used gasket for evidence. Here the subject is the load budget: what has to be on the face, where it goes, and which number on your worksheet you can actually change.

Before any of this is useful: opening the joint

Everything below assumes a joint you are allowed to open. A bolted joint that has been in service is holding some combination of pressure, trapped liquid, spring load, the weight of the pipe or equipment hanging off it, and sometimes a bonding path for electrical fault current.

• Isolate and lock out the mechanical and stored energy sources under 29 CFR 1910.147 before the first bolt moves, then prove zero pressure by cracking a vent or drain at the joint with the isolation locked, not by reading a gauge upstream.
• If any conductor, motor lead, or heat-trace circuit lands on or near the equipment, that half is electrical work: de-energize and verify absence of voltage under 29 CFR 1910.333(b)(2), using the live-dead-live proving sequence in NFPA 70E-2021, 120.5, which binds you through your employer's electrical safety program or a contract, not on its own.
• Loosen from the side, never with your body in the split plane of the flange, and back every bolt off in the crossing pattern rather than pulling one bolt out first. A joint that still holds residual load will spring at the last few bolts even when the line is empty.
• Support the pipe or component before the last bolts come out. On a horizontal run the flange is carrying weight, and that weight arrives on your hands the moment the bolts stop carrying it.

The condition that has to hold

Two conditions, actually, at two different moments, and a joint that satisfies one and not the other leaks.

At assembly, the gasket has to be pushed hard enough to conform to the flange faces and close the leak paths across them. That is a minimum stress on the seating face, and it is a property of the gasket material and thickness. A soft sheet material needs less; a hard, high-recovery material needs a great deal more.

In service, after pressure comes up, whatever stress is left on the face has to stay above a lower but still definite threshold. Pressure inside the joint does two things at once: it pushes outward through the gasket, trying to get past the seal, and it pushes the two flanges apart, which unloads the face. So the number that matters in service is the residual stress, not the one you set at assembly.

Both thresholds belong to the gasket manufacturer for their specific material and thickness, and where the joint is inside a pressure system built to a design code, the required bolt load belongs to whoever engineered the joint. Do not carry a number from one gasket family to another.

Where the bolt load goes

Total bolt load is the whole budget. Three things spend it:

1. Seating the gasket - the load times the gasket contact area gives you the stress on the face.
2. Holding against the hydrostatic end force - pressure acting on the area inside the gasket's reaction diameter, pushing the flanges apart. That force comes straight off the load available to the face.
3. Anything else the joint is carrying - a bending moment from an unsupported pipe run, thermal load from a line that is not free to move, or the weight of a valve hanging on one side. These are real and they subtract from the same budget.

The reason the end force surprises people is that it scales with the square of diameter and only linearly with pressure. A small line at high pressure often has a trivial end force. A large line at modest pressure can have an end force that eats the entire bolt load. That is the whole reason big low-pressure flanges have a reputation for being fussy.

Area is the lever you control

Bolt load is bounded by the bolts you have, the material they are made from, and what the flange can take. Pressure is set by the system. The one term on the sheet you can often change is contact area, and stress is load divided by that area.

Halve the width of the seating band and, at the same bolt load, you double the stress on it. That is why a narrow raised face seals more easily than a wide flat face, and why a gasket cut oversized so it overhangs into the bolt circle can leak at a torque value that sealed a properly sized one. The overhang is contact area you are paying for and getting nothing back from.

The limit at the other end is just as real. Push the stress too high and a soft gasket extrudes out of the joint, a fiber sheet crushes and loses the recovery it needs to follow the flange as it heats, and a thin, hard gasket concentrates load on the high spots of an out-of-flat face instead of bridging them. Narrow is a lever, not a direction to keep going in.

The joint sheet, filled in

Here is the sheet for one real-shape joint. Every value marked illustrative is there to carry the method, not to be copied onto a work order.

The end force here costs about 6 percent of the load, which is why this joint feels forgiving. Now change one thing at a time and watch which change moves the answer.

Cut the gasket width to 0.25 in. Contact area drops to 3.93 in2 and the assembly stress doubles to roughly 12,200 psi at the same bolt load. If the material needed more seating stress than 6,100 psi, this is the fix, and it costs nothing but a different gasket. If it was already sealing, this is how you extrude it.

Move to a 12.0 in mean diameter at 600 psi. Area inside the reaction diameter is 113 in2, so the end force is about 67,900 lbf. That single number is larger than the entire 48,000 lbf budget in the sheet above. No tightening pattern rescues that joint, because the problem is not distribution, it is that the bolts cannot hold the flanges together against pressure at all. That is an engineered joint: bolt count, bolt size, and gasket selection are somebody's calculation, not a field judgment.

Leave the gasket overhanging by 0.25 in on the outside. Contact area rises to about 11.8 in2, stress falls to roughly 4,100 psi, and a joint that sealed last time now weeps at the same wrench setting with nothing else changed. The tech who trims the gasket flush is not being fussy, they are restoring a third of the seating stress.

What the sheet does not contain, on purpose

There is no torque column. Torque is an input to bolt load through a friction relationship that varies with thread condition, lubricant, and washer face, and a sibling article owns that conversion in detail. The sheet is written in load because load is what the gasket experiences. If your only tool is a torque wrench, you are estimating the first line of the sheet, and you should know that you are.

There is also no line for "how tight it felt." Feel scales with bolt diameter, thread pitch, and how greasy the stud is, none of which the gasket knows about.

How to verify you got this right

• Measure the gasket, do not assume it. Mean diameter and width off the actual part, and check that the cut gasket matches the flange face rather than the nominal pipe size. An inner diameter that intrudes into the bore restricts flow and, on a raised face, an outer diameter that runs into the bolts is unsupported area.
• Compute the end force before you decide a joint is a tightening problem. If the end force is a large fraction of the load your bolts can provide, stop and route it to whoever sized the joint.
• Read the compressed thickness at four points around the circumference after the joint is made up, using the same feeler or caliper technique at each point. A consistent compressed thickness means the load is distributed. A gasket thicker on one side is telling you the faces were not parallel when the bolts took load, and no amount of extra torque on the thin side fixes the geometry.
• Re-check after the joint has seen its operating temperature once, because gasket materials relax and bolts settle. Whether re-tightening is permitted, and at what temperature, belongs to the equipment manufacturer or the adopted piping code, not to the tech with the wrench, and it is never done on a live joint by feel.

References

• ASME B16.5 and ASME B16.20 flange and gasket dimensional standards, in the edition adopted by the piping code your installation was built to, which reach you through that code or through the project specification rather than on their own
• 29 CFR 1910.147, control of hazardous energy, for mechanical isolation and stored energy at a bolted joint
• 29 CFR 1910.333(b)(2) for electrical isolation, with NFPA 70E-2021, 120.5 for the live-dead-live proving sequence where your employer's electrical safety program adopts it
• Gasket manufacturer documentation for minimum seating stress and maximum allowable stress on the specific material and thickness
• See related: What Torque Actually Controls and What It Does Not; Why Seals Leak: The Physics; Why a Gasket Fails