The Gasket That Was Crushed and the One That Was Not

Why this matters

A gasket that comes out of a joint is the only unbiased record of how that joint was assembled. It carries the imprint of the flange faces, the distribution of the bolt load, the temperature it lived at, and whether anyone ever got it to the stress it needed. Almost all of that gets thrown away, because the ordinary reflex on a leaking joint is to scrape the old one off, put a new one in, and pull the bolts a little tighter. That reflex is correct for exactly one of the two failures below and it will destroy the other, and you cannot tell which one you have without looking at what you just scraped off.

Before you open a bolted joint

A bolted joint that has been holding pressure is a stored-energy device, and the energy releases at the split line where your hands are.

• Isolate, lock, tag, and prove zero at a gauge before a bolt moves. Control of hazardous energy on a pressurized or mechanical path is 29 CFR 1910.147, and a closed valve is not an isolating device until it is locked and the line is proved dead.
• A drained line is not an empty line. Liquid sits at low points and in dead legs, and hot liquid at a low point will come out of the split under its own head. Open the low-point drain and confirm flow has stopped, not just that a valve was turned.
• Break the joint on the far side from where you stand. Loosen bolts in a pattern rather than one side at a time, leave two opposite bolts loosely engaged while the seal breaks so the flange cannot swing, and keep your body out of the plane of the split line.
• Assume the contents will reach you. Eye and face protection, and gloves selected on their performance characteristics for that substance under 29 CFR 1910.138(b), with the substance's safety data sheet available as required by 29 CFR 1910.1200.

The two jobs a gasket does

A gasket has to do two different things, and they are set by two different numbers.

Conform at assembly. It has to be pressed hard enough to flow into the surface finish of both faces and close every leak path in the machining marks. That takes a minimum seating stress, and it is a property of the gasket material. It has nothing to do with the pressure the joint will see; even a joint that will run at low pressure needs the full seating stress or it never seals at all.

Stay in contact under load. Once pressurized, internal pressure tries to push the flanges apart and the residual bolt load has to keep enough stress on the gasket to hold the seal. That takes a multiple of the operating pressure, which is why the requirement scales with service conditions.

The bolted-flange-connection appendix of the ASME Boiler and Pressure Vessel Code, Section VIII, Division 1 tabulates exactly these two as the minimum design seating stress and the gasket factor, per gasket type, which is the formal statement of why a soft sheet gasket and a spiral-wound gasket in the same flange do not take the same bolt load. The practical consequence for a shop is short: the torque figure for a bolted joint comes from the gasket manufacturer for that gasket in that flange, not from a fastener torque chart. A fastener chart tells you what the bolt can take. The gasket needs whatever torque produces the right stress on it, which is often far less.

The witness mark is the record

Pull the gasket and look at the burnished band the flange faces left on it.

• A uniform, bright, full-width band on both faces, all the way round means the joint was seated evenly. This is what a correctly assembled joint leaves and it is the reference every other reading is compared against.
• A band on part of the circumference only means the faces were not parallel: a warped flange, a pipe pulled into alignment by the bolts, or bolts tightened in the wrong order so one side closed first.
• A band narrower than the raised face means the gasket was oversized for the seat or moved during assembly, so part of the load never landed on it.
• No band at all means it never reached seating stress. The material did not fail. Nobody ever asked it to seal.

Three states, three signatures

The radial groove in the under-seated row deserves a name, because it is easy to misread as damage. A joint that is weeping at low stress lets fluid track across the face, and over time that flow erodes a channel from bore to outside edge. It looks like the gasket was cut. It is the record of a leak that was already running, and it is a consequence of under-seating rather than a cause of the leak.

The scalloped gasket, and why more torque is the wrong answer

There is a fourth pattern that sits inside the over-compressed row and is worth pulling out, because the obvious response to it makes it worse.

Thinner at the bolt positions and thicker in the spans between them. The bolts are doing their job and the flange is bending between them, so the gasket is being crushed at each bolt while the midspans get less stress than they need. On a joint like this the leak, if there is one, is at the midspans, and the reflex is to tighten the bolts.

Tightening increases the crush at the bolts, where the material is already at its limit, and adds very little at the midspans, where the flange simply bends more. The stress distribution gets worse, not better. The real answers are a gasket that reaches seal at lower stress, a stiffer flange, closer bolt spacing where that is available, or a correct multi-pass cross pattern so load distributes before any bolt reaches final value. Which of those is available is a design question more often than a field one, but recognizing the pattern is what stops you from making it worse in the meantime.

Compression set, and what recovery actually means

Compression set is the portion of the squeeze a material does not recover when the load comes off, reported as a percentage in standard testing under ASTM D395 for rubber. Zero percent set is full recovery, and 100 percent is a material that stays exactly where it was crushed.

A gasket is supposed to take some set. That is how it fills the surface finish, and a material with no set would push back out of the machining marks. So a gasket that comes out looking untouched is not good news, it is the primary evidence of under-seating.

A gasket with nothing left to give is the other end of the same scale. Once the material is at full set it has no springback, so any movement of the flange from thermal cycling, vibration, or bolt relaxation opens a gap it cannot follow. This is the joint that seals perfectly at ambient and weeps every time the system heats up.

The pair: two flanges, one skid, one day

Two flanged joints on the same skid, same soft sheet gasket material, same nominal bolt size, opened the same afternoon. Nominal sheet thickness as printed on the roll: 0.062 inch. Every other figure below is a caliper reading taken off the pulled gaskets on that job.

Flange A, which was leaking. The gasket measured 0.060 inch on removal, which is about 3 percent thinner than it went in. There was no visible witness band over most of the circumference and a faint one on the remainder. A radial groove ran from the bore to the outer edge at one position. Breakaway on the bolts that were still in place was almost nothing on several of them.

Flange B, which was not leaking and was opened for unrelated work. The gasket measured 0.030 inch at the bolt positions and 0.042 inch at the midspans. Against the 0.062 inch it started at, that is a loss of about half its thickness at the bolts and about a third at the midspans, and the difference between bolt and midspan is 0.012 inch, close to 20 percent of the original thickness. The surface was hard and glazed, it had cracked radially in several places, and a lip of material had extruded into the bore.

What the pair says. Same material, same day, same skid, and two failures in opposite directions. Neither is a gasket-material problem, which is what a supplier would be asked about if only one of them had been looked at.

Flange A never reached seating stress. The material is nearly its original thickness, there is no witness band to speak of, and the groove is the record of fluid that had been tracking across an unseated face for a while. It leaked because nobody asked it to seal.

Flange B was over-compressed and was sealing on brute force with nothing in reserve. It was not leaking on the day, which is exactly why it would never have been investigated. The scalloped thickness says the flange is bending between bolts, and the extruded lip in the bore is material that has already been squeezed out of the joint and is not coming back.

The reflex that would have been applied to both. A tech who scraped both gaskets off without measuring them would have fitted new ones and pulled the bolts a little tighter than last time. On Flange A that is correct and would have fixed it. On Flange B it adds crush at the bolts where the material has nothing left, drives more extrusion into the bore, and sets up a blowout on the next thermal cycle. One reflex, two joints, opposite outcomes, and the only thing that distinguishes them is the scrap in the bin.

What each one actually needed. Flange A: clean faces, new gasket, and the gasket manufacturer's torque value applied in the specified cross pattern over the specified number of passes. Flange B: a check that the torque figure in use was ever specified for this gasket material rather than carried over from a harder one, because a value that suits a spiral-wound gasket will crush a soft sheet, plus a gasket selection that seals at lower stress if the flange stiffness cannot be changed.

Preserving the evidence so it can be read

None of the above is available if the gasket comes off in pieces with no idea which way round it sat.

• Photograph the joint before the last bolts come out, then photograph both gasket faces once it is off.
• Mark orientation before you pull it. A scribe or a marker line lined up with one bolt is enough. Without it you can say a gasket is uneven, and you cannot say whether it is thin at the bolts or thin on one side, which are two different findings with two different fixes.
• Measure at bolt positions and at midspans, and write the numbers down. Feel is not sensitive enough to separate a third from a half, and the difference between those two readings is the difference between over-compression and flange bowing.
• Keep the gasket until the joint is back in service and holding. It costs a bag and a label, and it is the only physical evidence you will get if the repair does not hold.

References

• ASME Boiler and Pressure Vessel Code, Section VIII, Division 1, bolted flange connection appendix, for gasket factor and minimum design seating stress
• ASTM D395, standard test methods for rubber property, compression set
• 29 CFR 1910.147, control of hazardous energy, for isolating and relieving a pressurized joint before opening it
• 29 CFR 1910.138(b) and 29 CFR 1910.1200, for glove selection and safety data sheet availability on the joint contents
• Gasket manufacturer documentation for the torque value, pattern, and number of passes for that gasket in that flange
• See related: How Seals and Gaskets Fail; The Torque Sequence That Matters