Most flange failures are not mysterious. They are not exotic metallurgical events that arrive out of nowhere. They are the predictable consequence of a small number of decisions made on the original MTO: the wrong material, the wrong facing finish, the wrong gasket, the wrong bolt grade, or the wrong torque sequence. Thirty years of supplying pipe flanges to refineries, power plants, shipyards and chemical sites have given EZ Steel Industrial a fairly repeatable list of the failures that show up in the field, and an even more repeatable list of the specification choices that prevent them.

This article is the seven-failure version of that list. It is written for the engineer who has to defend a piping specification in front of a client, a classification society, or an internal review — and who would rather catch the issue on paper than on the platform.

The first failure mode is also the most common. The gasket sits in a thin crevice between the flange face and the bolting force. In a wet, chloride-bearing or sour service, that crevice becomes a corrosion cell that survives even on stainless grades. The result is pitting under the gasket ring, loss of seating stress, and a leak that shows up during the next thermal cycle rather than during hydrotest.

The prevention is to match the gasket type to the service, and to keep the flange face finish in the roughness band the gasket manufacturer actually requires. Spiral-wound gaskets need a serrated finish (typically 125–250 µin Ra) so they can bite into the face. Compressed non-asbestos gaskets need a smoother finish. Ring-type joint gaskets need a hard, smooth finish on the groove. Specifying a finish without telling the flange mill which gasket you intend to use is the single most common setup for a crevice leak later on.

The second failure is mechanical rather than chemical. When stainless steel steel flanges in austenitic grades (304, 316, F316L) are bolted with stainless nuts and studs of similar composition, the threads can seize under load — what the trade calls galling. Once galling starts, the only way to free the bolt is to cut it out, often with the line still in service.

The fix is to use a dissimilar, harder bolting material. B8M studs with 8M nuts work, but B8M-2/B8M-3 strain-hardened is better, and the highest reliability comes from B7 carbon steel studs zinc- or PTFE-coated, mated to stainless flanges. We have also seen PTFE-based anti-seize on the threads and a molybdenum-disulfide film on the nut face, both of which reduce the friction variance that drives the torque wrench off its target.

The third failure is one of the most under-appreciated in the field. Compressed fibre and graphite gaskets creep under sustained load and temperature — they "set down", which means the gasket gets thinner and the bolt load drops. If the original torque did not leave enough margin above the minimum seating stress, the joint relaxes enough to leak within a year, even though the hydrotest was perfect.

There are three ways to head this off. First, pick a gasket style with low creep for the actual temperature — spiral-wound with graphite, or metallic ring-type joint, both behave better than compressed fibre above 200 °C. Second, retorque after the first thermal cycle. Third, verify the seating stress calculation rather than trusting a generic torque chart. A gasket stud bolt nut kit that ships together with the gasket, the studs, the nuts, and the recommended torque value on one packing list is the easiest way to make sure these three rules get followed on a real site.

The fourth failure is the one nobody likes to admit afterwards. A Class 150 flange is rated for a specific pressure-temperature envelope. Above that envelope, the joint is no longer code-compliant, even if the metal itself happens to survive for a while. We have seen MTOs where a Class 150 slip-on was ordered for a 300 psig steam line because the bolt circle was the right size and the installer "had it on the shelf". The hydrotest passed; the running line did not.

The prevention is to put the design pressure, the design temperature, the fluid, and the corrosion allowance into the same line of the purchase order and refuse to accept a quote that has been built around a different combination. EZ Steel Industrial's quote-back sheets explicitly list the design conditions the offer is based on, so a future inspector can match the flange to the line class rather than the other way around.

The fifth failure is a cousin of Failure 3 but deserves its own entry. On hot, cyclic lines — superheater outlets, steam manifolds, refinery hot oil — the stud bolt itself relaxes as the differential thermal expansion between the bolt and the flange works the joint open. The standard fix is to use a higher-grade stud (B16 for high-temperature service) and to retorque after the first heat-up. The longer-term fix is a controlled bolt-loading procedure: torque to 50%, then 80%, then 100% in a star pattern, with a final pass after the line is at operating temperature.

A stud bolt that is too short is the silent version of the same failure. The bolt has to stretch enough under torque to maintain the seating stress as the gasket sets down. Under-length studs give under-load joints no matter how accurately the torque wrench is set. We routinely see this on retrofit jobs where a new gasket was fitted without re-checking the stud engagement length, particularly when the joint was originally assembled with a thinner gasket.

The sixth failure is administrative rather than metallurgical, and it is the most embarrassing because it shows up at hydrotest. Two pipe flanges on either side of a joint have to share the same facing — raised face with raised face, flat face with flat face, ring-type joint with ring-type joint. A Class 150 RF flange mated with a Class 150 FF flange will pass the bolt circle, fail the seating geometry, and leak on the first pressure cycle. The fix is to write the face type onto the isometric drawing itself, not just onto the BOM, and to verify it at receiving.

For copper-nickel seawater service, the issue is the same but the faces are different. Copper-nickel flanges are commonly supplied with a raised face or with a compound face designed for a spiral-wound gasket. A spiral-wound gasket on a standard RF face is the right combination; a flat non-asbestos gasket on a Cu-Ni compound face is the wrong one, and the joint will not seat.

The seventh failure crosses the boundary between the flange and the valve. A copper-nickel line with a carbon-steel butterfly or gate valve is a galvanic couple that the inspector may accept on paper and the dry-dock will catch in practice. The valve body pits at the seat, the trim corrodes, and the repair scope expands from the valve to the joint. Specifying a bronze, monel, or copper-nickel valve body on the same MTO as the flange removes the couple and removes the problem.

This is also the failure where the "bundled supply" approach pays for itself most visibly. When the industrial valves, the flanges, the gaskets, and the stud bolts all come from one supplier on one material spec, the galvanic question is answered at the quotation stage rather than at the leak. The cost of a bronze valve body over a WCB carbon steel body is small, and the saving on the next maintenance turnaround is substantial.

For a typical project, the seven-failure check fits inside a single design review. Walk down the line list, mark each flange with the service, the design pressure, the design temperature, the fluid, the face type, the gasket type, the stud grade, and the valve body material. Any line that has a question mark against any one of those fields is the one that will eventually leak. The exercise is mechanical and a little boring, which is exactly why it tends to get skipped; the leak it prevents is the kind that gets remembered.

For a single-source RFQ, send the line class, the fluid service, the flange type and facing, the pressure class, the bolt grade, the gasket spec, and the survey body. With that we can return a full technical-commercial offer within 48 hours, with the right material, the right documentation level, and a realistic shipping window from Changsha to your site. Where the project is a coordinated package — pipe, flanges, fittings, valves, gaskets and bolting — we will quote it as a single scope so the joint decisions are made once.