What a fire-safe claim actually promises
A fire-safe ball valve is not a valve that survives a fire. It is a valve that keeps sealing after the parts which normally do the sealing have been destroyed. The soft seat insert, the stem O-rings and any elastomer in the body joint are gone by the time the metal reaches fire temperature, and the design has to hand the sealing duty over to a second, all-metal or graphite path that was sitting behind them from the first day in service. That handover is what the test measures.
Two consequences follow, and both are commercial as well as technical. A fire-safe certificate says nothing about performance before the fire, so a valve can be fire-safe and still fail an ordinary seat test. And fire-safe is a qualification, quoted as a standard plus an edition, covering a defined set of sizes and pressure classes. Two valves can both be honestly described as fire-safe and still be qualified to different standards, over different ranges, with different acceptance limits.
The burn: thirty minutes, 761 to 980 degrees, valve closed and full of water
API 607 puts the valve in a test pit, fills it with water, leaves it in the closed position, and exposes it to a flame that holds the surrounding temperature between 761 °C and 980 °C for 30 minutes. Thermocouples on the body have to show that the metal itself arrived at that temperature: the standard requires the average body temperature to reach 761 °C, so a flame that burns the paint off while the body stays cool does not count. The valve then cools in air, and only after cooling is it pressure tested, external leakage first and then seat leakage in each direction.
Two conditions in that sequence carry most of the meaning. The valve is tested closed, because the scenario being modelled is a valve that was shut when the line failed. And the valve is full of water, because the water is what transmits pressure to the seals while they are being destroyed; a dry valve proves nothing about sealing. A shop that runs the burn with the valve open has not run API 607.
Zero leakage outside, a rate inside
After the fire the acceptance is deliberately asymmetric. External leakage must be zero: no weep at the body joint, the stem gland or the seat retainer, because any of those is a path that feeds the fire. Seat leakage is allowed, up to a rate tabulated in the standard, and it is measured in both directions. A soft-seated valve that was bubble-tight cold is expected to pass a measurable amount of water once its seat has burned away.
This is where purchase orders go wrong. Zero leakage after fire test is a common line, and no fire test will certify it: the requirement asks for something the standard deliberately does not require, and a supplier who signs it has either misread the test or intends to argue about it later. What belongs in the order is the standard, the edition, the size and class, and the acceptance exactly as the standard writes it.
What burns, and in what order
The sequence is predictable, and the design has to survive each step of it. PTFE melts at about 327 °C, so the seat insert and any PTFE packing rings are liquid or gone long before the body reaches 761 °C. Elastomer O-rings pass their limit earlier still. A soft body gasket goes at roughly the same moment as the seat. What remains, if the valve was designed for it, is graphite in the gland, a machined metal face on the seat carrier, and a graphite-filled body gasket.
The order matters because the handover cannot leave a gap. If the graphite packing sits above the PTFE rings instead of behind them, the gland has nothing to seal against once the PTFE has flowed out. If the metal face on the seat carrier is set further back than the burned residue can be cleared, the ball never touches it. Fire-safe geometry is mostly a question of what sits directly behind what.
The secondary seat, and why it stays out of the way
A fire-safe seat is two seats in one holder. The soft insert does the sealing in normal service because polymer conforms to the ball. Behind it, and slightly inside it, the carrier presents a machined metal face at the same radius as the ball. In normal service that face is held just clear of the ball, so it adds no torque and does not mark the surface; after the insert burns, line pressure and the seat spring push the carrier onto the ball and the metal face becomes the seal.
Three details decide whether the handover works. The carrier has to stay guided so it cannot tilt and touch on one side only. It needs travel forward greater than the space the insert occupied, because that space is now empty. And there must be a seal behind the carrier, normally graphite, so that the cavity is still separated from the downstream side once the insert is gone; without it the metal face seals the bore and the valve still leaks around the back of the seat. Where the burned material has nowhere to go, a closed pocket traps the residue and holds the carrier off the ball.
Stem sealing is the usual failure point
The stem is the hardest place to keep sealed in a fire because it is the one penetration that also has to rotate. The fire-safe answer is die-formed graphite rings, compressed hard enough to behave as a solid seal rather than as a packing that needs periodic adjustment, with a live-loading spring on the gland so that preload survives differential growth between stem and bonnet. An anti-blowout stem belongs to the same package, for a different reason: at fire temperature the retaining geometry is the only thing holding the stem inside the body.
In practice the gland is where a fire test is most likely to fail, and it is also where a field failure is hardest to see, because a gland weep at temperature evaporates and leaves no puddle. A specification that names the seat and says nothing about the gland has specified half the valve.
The body joint, a gasket asked to do two jobs
On a three-piece body there are two joints, and each has to hold both process pressure and fire temperature. A spiral wound gasket with a 316 winding and a graphite filler does both: the winding carries the compression, the graphite survives the heat. A PTFE-filled spiral wound gasket is a fire path with a metal jacket around it, and it is one of the most common reasons a valve that looks fire-safe does not pass.
Bolting is the other half of the joint. It has to keep preload while the body and the end cap sit at different temperatures, so the stud grade and the washer arrangement are chosen for the fire case rather than for ambient. A193 B7 studs with A194 2H nuts are the normal starting point on carbon and stainless bodies; the point is that they were chosen for a reason, not carried over from a water-line bill of materials.
Which standard, and what the certificate covers
Three names appear on certificates and they are not interchangeable. API 607 is the fire test written for quarter-turn valves with non-metallic seating materials, and it is the one that applies to a soft-seated ball valve. API 6FA is the fire test for API 6D pipeline valves and is what a pipeline specification normally asks for. ISO 10497 is the international fire type-test used across Europe and much of Asia; it follows the same principle of burn, cool and retest, but the burn conditions, the test pressure and the acceptance details differ from API 607, so qualification to one does not imply qualification to the other.
Two limits belong in the conversation as well. The certificate covers a range of sizes and classes derived from the size that was actually tested, and the standard sets how far a single test may be extrapolated; that coverage is printed on the report, and a test on a small valve is not blanket approval for a large one, because thermal mass and joint geometry change with size. Beyond that, a fully metal-seated valve has no polymer seat to destroy, so its fire performance is normally demonstrated on the same procedure and argued from the all-metal sealing path rather than claimed under a soft-seat certificate.
What to write into the purchase order
Six lines settle it: the fire test standard and edition; whether the test is to be witnessed and by whom; the sizes and pressure classes to be covered, stated as a range rather than a single size; the acceptance as the standard defines it, with zero external leakage named explicitly; the materials of the secondary seat, stem packing and body gasket; and the test report, including the burn chart and the post-test leakage figures, shipped with the valve.
J-VALVES supplies fire-safe ball valves from 1/2 to 56 inch (DN15 to DN1400) in ASME Class 150 to 2500, with a metal backup seat behind the soft insert, die-formed graphite stem packing, a graphite-filled spiral wound body gasket and an anti-static device bonding stem, ball and body. Valves are shell tested at 1.5 times and seat tested at 1.1 times rated pressure, fire-safe construction is available to API 607 or ISO 10497, and EN 10204 3.1 certificates are issued for body, ball and stem.
Frequently Asked Questions
Does fire-safe mean the valve is leak-tight after a fire? No. External leakage must be zero, but seat leakage is allowed up to the rate tabulated in the standard. A valve that was bubble-tight cold will normally pass some water once the soft seat has burned away, and that is the accepted outcome rather than a defect.
What is the difference between API 607, API 6FA and ISO 10497? API 607 is written for quarter-turn valves with non-metallic seats, API 6FA for API 6D pipeline valves, and ISO 10497 is the international fire type-test with its own burn conditions and acceptance. Qualification to one does not automatically qualify the valve to the others, so the certificate has to be read against the standard the project names.
Where do fire-safe ball valves most often fail the test? At the stem gland and the body joint, because the seat has an obvious all-metal backup path behind it while the gland depends on the packing actually being graphite and actually being live loaded. A weep at the gland at fire temperature evaporates and is easy to miss afterwards.
