Views: 0 Author: J-VALVES Publish Time: 2026-10-31 Origin: Site
A fire-safe valve does not seal perfectly in a fire. That sentence is the single most misunderstood point in valve procurement. A fire-safe certificate promises something more useful: that after a 30-minute enveloping flame, the valve still limits leakage to a small, defined, tabulated rate instead of feeding the fire with fuel. This article walks through what the API 607 fire test actually does to a valve, what the 7th edition acceptance numbers are, how a soft-seated ball valve survives the burn, how qualification extends from one tested valve to a whole product line, and which specification mistakes quietly void the coverage you thought you had.
The purpose of fire-safe qualification is incident containment, not shutoff. During a hydrocarbon fire on a platform deck or in a refinery unit, an emergency shutdown valve has two jobs: keep hydrocarbon from escaping through the stem and body joints to feed the flame from outside, and keep the closed seat from leaking enough line fluid downstream to re-supply the fire. A conventional soft-seated valve fails both jobs quickly, because the seat and packing are polymers that decompose in the first minutes of flame exposure.
API 607, Fire Test for Quarter-Turn Valves and Valves Equipped with Nonmetallic Seats (current edition: 7th, June 2016), defines the test that proves a valve still works after that exposure. The acceptance philosophy is deliberate: the standard tolerates modest through-seat leakage during the fire and tightens the requirement after cooldown, when the fire has been knocked down and residual inventory is still escaping. Knowing which limit applies in which phase is what makes a test report readable.
Run the timeline in your head before reading the test sequence. PTFE and RPTFE seats begin to lose mechanical integrity around 260 deg C and decompose well below the flame temperature the test imposes. Within roughly the first four minutes of a 750-1000 deg C burn, the soft seat insert retreats and the line fluid begins passing the seat - one test laboratory has published data showing a leakage peak at about the fourth minute, while the ball is migrating (on the order of a millimetre, under line pressure) toward the metal secondary seat that has been machined behind the soft seat. Once that metal-to-metal contact band establishes - on the order of 1.5 to 2.5 mm wide in typical floating and trunnion ball valve designs - the leakage curve flattens at a low, bounded rate. The soft seat is gone; the secondary seat is carrying the fire.
The same story plays out at the stem. Standard PTFE packing is finished above 260 deg C. Behind it, fire-safe designs carry flexible graphite rings, which survive flame temperature and keep the gland sealing. The body joint follows the same redundancy: a spiral-wound or kammprofile gasket with graphite facing, not a PTFE-filled type. Every fire-safe feature is a second line of defence behind a component you already know will not survive.
This is also why fire-safe design is inseparable from the metal seat geometry discussion covered in our article on hardfaced trim and metal-seated wear: the secondary seat is a real metal-to-metal seal with reduced sealing force, and its leakage is tolerated by design, not eliminated.
Strip away the catalogue language and a fire-safe ball valve differs from a standard one in four places:
A metal secondary seat behind the soft seat. A stainless or hardfaced metal ring positioned so that the ball lands on it when the polymer insert is consumed. The sealing force is lower and the leakage is finite - the standard accepts that.
Graphite stem packing. Usually graphite rings with any polymeric components placed below the graphite, so graphite seals after the polymer burns off. Laboratory data from one fire-test facility suggests the packing contributes roughly 70% of total external leakage during a burn, which is why gland design, not the body joint, is the usual external-leakage battleground.
A graphite-faced body gasket. Spiral-wound with graphite filler or kammprofile with graphite facing. A PTFE envelope gasket is a fire-test failure waiting to happen.
Anti-static continuity between ball, stem and body. The static charge generated by flowing fluid must not spark outside the valve after the polymer insulators are destroyed. The anti-static device must remain functional after the test - it is a checkable item, not a brochure phrase.
Butterfly valves follow the same logic with a metal backup seat behind the elastomer or polymer seat, which is why both of our butterfly valve and ball valve product pages call out fire-safe options explicitly for hydrocarbon service.
The test is destructive, witnessed, and performed on a production-representative valve with the stem horizontal. In outline:
Pre-test verification. The valve must first pass its normal production hydrostatic and seat tests, and the rig itself is proven tight. Instrumentation includes flame thermocouples, calorimeter blocks, and - since the 7th edition - a cavity pressure tap fitted by the manufacturer.
Burn period, 30 minutes. The valve is closed and internally pressurised with water. Burners envelop the body: the average flame temperature must be between 750 and 1000 deg C, reaching at least 750 deg C within two minutes and not falling below 700 deg C. Calorimeter blocks average 650 deg C at 15 minutes and must not fall below 560 deg C. In one published 7th-edition report, the calorimeter blocks exceeded 650 deg C for 22.5 of the 30 minutes - the standard checks the fire, not just the clock.
Test pressure. For most process valves the water pressure during the burn is 75% of the valve's rated cold working pressure. In the published 4-inch Class 1500 example, that meant an average of about 2800 psig against a 3750 psig rating.
Leakage collection during burn and cooldown. Through-seat leakage and external leakage are collected and measured continuously - through-seat at the outlet via condenser and measuring vessel, externally from stem and body joints.
Cooldown below 100 deg C. Fuel off, water spray or air blast, with leakage still being collected. The quench simulates firefighting water and its thermal shock - parts that survived the fire can crack during cooldown.
Post-fire low-pressure seat test. At about 2 bar (30 psig) with water, held for 5 minutes. This is the step that surprises buyers: after a fire, low differential pressure is harder to seal than high, because the metal secondary seat relies on line pressure and contact-band geometry rather than high loading. Our article on ball valve seat preload and low-pressure sealing explains why the same valve that passes a high-pressure water test can weep at low pressure.
Operational test. With high test pressure held on one side, the valve must cycle open using its standard operator. It has to actually work after the fire - a valve that seized shut is a failed valve.
Post-operation external leakage test. Five minutes at high pressure with the valve open, measuring stem and body joint leakage again.
All leakage limits scale with nominal size. The current-edition allowables, expressed per inch of NPS:
Phase | Leakage path | Allowable (7th ed.) | NPS 4 example |
|---|---|---|---|
During burn and cooldown | Through-seat (internal) | 400 mL/min per NPS inch | 1600 mL/min |
During burn and cooldown | External (stem, body joints) | 100 mL/min per NPS inch | 400 mL/min |
After cooldown, low pressure (2 bar, 5 min) | Through-seat (internal) | 80 mL/min per NPS inch | 320 mL/min |
After operational cycle, high pressure (5 min) | External | 25 mL/min per NPS inch | 100 mL/min |
All phases | Operability | Must cycle under full rated differential with the standard operator | - |
Two things to notice. First, the during-fire limits are roughly four times looser than the post-fire ones - the standard expects the valve to leak modestly while the seats burn out and to pull tight once cooled. Second, earlier editions tabulated tighter during-burn numbers (on the order of 100 mL/min per NPS through-seat and 25 mL/min per NPS external). If you see a certificate quoting 100/25/40 style figures, it is an older edition, not an error - and as a buyer it is arguably a stronger paper. The edition is printed on the test report; the acceptance numbers are not portable between editions.
For contrast with the regime you test against in normal service: the leakage classes of ISO 5208 and API 598 discussed in our seat leakage classes article are one to two orders of magnitude tighter than any post-fire limit. A fire-tested valve is not a bubble-tight valve, and the two claims never substitute for each other.
Double-seated valves trap fluid in the body cavity between the upstream and downstream seats. In a fire, that trapped water or hydrocarbon expands and the cavity can become a pressure vessel nobody designed. Test laboratories have recorded cavity pressures exceeding ten times the valve rating before the requirement existed. The 7th edition therefore requires the manufacturer to fit cavity pressure taps before the test and limits allowable cavity pressure during the burn to 1.5 times the valve's rated pressure, or a value derived from the manufacturer's hydrostatic test data.
For buyers this connects directly to the seat-configuration logic (SPE/DPE and cavity relief philosophy) covered in our seat-design article: a fire-safe valve with self-relieving seats manages fire-driven cavity pressure by design; a double-piston-effect configuration without an external relief path relies on the cavity fitting itself to pass the same test. If your line fluid is prone to thermal expansion or you are in cryogenic or hydrogen service - where our hydrogen valve material guide also flags fire and ignition topics - state the cavity relief expectation explicitly instead of assuming the fire test covers it.
Fire testing is expensive, destructive, and increasingly impractical above moderate sizes - the tested valve is destroyed. No manufacturer fire-tests every catalogue line, so API 607 defines how one successful test qualifies others of the same basic design:
Dimension | Extension rule | Worked example |
|---|---|---|
Size, upward | Test valve qualifies valves up to twice its nominal size | NPS 2 test qualifies NPS 2 to 4 |
Size, large | An NPS 8 test qualifies all larger sizes; if the smallest size in a range exceeds NPS 8, test the smallest | NPS 8 test covers NPS 8-24 and up |
Size, downward | An NPS 2 test qualifies all smaller sizes of the same type; if the largest is under NPS 2, test the largest | NPS 2 covers NPS 2 to NPS 1/4 |
Pressure class | Test valve qualifies classes up to twice its own (within standard's tables) | Class 300 test covers Class 300 to 600 |
Bore | Reduced-bore test can qualify a smaller full-bore valve if closure member, seat seals and stem are identical in design and size | NPS 4 reduced bore covers NPS 3 full bore |
End connections | Different ends qualify if the alternate valve's mass is at least 75% of the test valve's mass | Butt-weld test covers flanged if mass rule holds |
Body materials | Generic classifications: ferritic, austenitic, duplex, nickel alloys. A ferritic test extends to others with one additional test per generic material and class range | WCB test + one 316 test covers the CF8M line |
Bolting | Alloy steel bolting (B7/L7) qualifies austenitic bolting, but not the reverse | B7 test covers B8 bolting, not vice versa |
Nonmetallic materials are stricter, and this is where certificates most often fail to cover what buyers assume. Any change in the nominal composition of the seat-to-closure seal, seat-to-body seal, stem seal, or body joint seal requires requalification. Filled PTFE may qualify unfilled PTFE and vice versa - that much is forgiving. But changing the elastomer family (FKM to HNBR, say) requires a retest of a valve at least equal to the median size of the previously tested range, and a change of packing manufacturer also triggers a retest at or above median size. If a supplier changed seat or packing suppliers since the certificate was issued, ask the awkward question.
Four standards cover fire testing, and quoting the wrong one is a project-delay classic:
Standard | Scope | Valve position in burn | Sequence nuance | Typical sector |
|---|---|---|---|---|
API 607 (7th ed., 2016) | Quarter-turn valves and valves with nonmetallic seats (ball, butterfly, plug) | Closed, pressurised | Low-pressure seat test mandated before operation; leak measured before cycling | Downstream process, tank farms, LNG terminals, offshore |
API 6FA | Valves built to API 6A and API 6D, including gate, globe, and soft- or metal-seated designs (check valves excluded - see API 6FD) | Fully open, pressurised | Operational cycle before final leak evaluations; external leakage limits on the order of 200 mL/in./min after burn | Upstream wellheads, pipelines |
ISO 10497:2022 | Soft- and metal-seated isolation valves, quarter-turn and multi-turn | Closed, pressurised | Includes cavity overpressure relief evaluation for double-seated valves and formal extension rules | Global/European projects, referenced by PED |
API 6FD | Check valves | N/A (check) | Fire test specific to non-return function | Pipelines, upstream |
Practical rules of thumb. API 607 and ISO 10497 are technically aligned for soft-seated quarter-turn valves and are widely accepted interchangeably, which lets one test serve both US-style and European-style tenders. API 6FA is not a substitute for API 607: it tests the open position and cycles the valve before final leak evaluation, a genuinely different protocol. For a monogrammed API 6D pipeline valve, the fire test specification is normally API 6FA. If your datasheet simply says "fire safe per API", you have not specified anything - name the standard, the edition, and the valve types.
Historical note for brownfield work: older North Sea specifications may reference BS 6755 Part 2, whose during-fire internal allowable (400 mL/in./min) and post-fire external allowable (20 mL/in./min) differ from both API documents. It has been superseded in practice by ISO 10497, but the numbers on a 1990s test report will not map one-to-one onto API 607 tables.
The test reports that end in "fail" cluster around a small number of mechanisms:
Failure signature | Mechanism | Where it appears |
|---|---|---|
External leakage ramps during burn | Gland loses contact as graphite packing sinters and contracts (a few percent volume change at flame temperature) while the stem shifts with body distortion | Stem seal, during burn |
Body joint weeping after cooldown | Bolt preload relaxation at high temperature - fastener load loss of half or more at 500-600 deg C is reported in fire-test practice - followed by thermal-shock distortion | Middle flange / bonnet joint |
Valve will not open after cooldown | Ball and metal secondary seat micro-weld or gall during the burn; opening torque spikes to several times normal and the stem twists or breaks. One test facility attributes on the order of 15% of failures to this opening moment alone | Operational test |
Low-pressure seat leakage just over limit | Metal secondary seat contact band insufficient or distorted by quench; leakage is bounded at high pressure but unbounded at 2 bar | Post-fire low-pressure test |
Cavity overpressure | Trapped cavity fluid expands faster than the seats relieve; tapped pressure exceeds the 1.5-times-rating cap | Burn period, double-seated valves |
Anti-static circuit dead after test | Spring or ball-type anti-static device destroyed by heat or debris | Post-test inspection |
The last failure mode - the valve surviving the fire but seizing - deserves emphasis in actuator sizing too: the standard requires operation with the standard manual operator, so an actuated valve must still be hand-wheel operable or its actuator interface must survive, and note that ISO 10497 addresses the valve's pressure envelope only; actuated assemblies need the interface covered separately.
Fire-safe certification is specified wherever flammable hydrocarbon can meet ignition energy: most offshore platform process and utility areas, refinery and petrochemical process units, tank farms and jetties, fuel gas and fuel oil systems, and gas pipeline stations. National and operator frameworks differ - Indian OISD-118 requires fire-safe qualification for hydrocarbon-duty isolation valves, and most international EPC specifications carry a fire-zone valve schedule that maps exactly this requirement. The underlying logic is ESD philosophy: the shutdown valve must remain capable of isolating fuel after the fire has started, which is precisely the post-fire portion of the API 607 sequence.
Non-flammable services generally do not need it - a fire-safe valve on a water line buys you nothing except cost. The exception is firewater-adjacent and flare systems, where operators frequently extend the requirement to any valve that must function during an incident.
"Fire-safe design" is not "fire-safe certified". A manufacturer can design to the standard's principles and skip the test. If your spec requires certified valves, require the test report, not a brochure claim. The design-only version is cheaper for a reason.
A certificate is not a range. Fire certification is a type test with defined extension rules. The certificate names a tested size and class; the qualified range is derived from the extension rules. A Class 600, NPS 6 requirement is not covered by a Class 300, NPS 4 certificate (class more than doubled, size more than doubled). Verify the supplied valve falls inside the tested envelope.
Edition matters. The 7th edition changed the acceptance numbers and added the cavity pressure rule. A datasheet citing "API 607" without an edition leaves the acceptance basis undefined; if your project copies a 1990s spec, it may be holding a supplier to tighter numbers than the current standard requires - decide deliberately, do not drift.
Seat and packing material changes requalify. The nonmetallic composition is part of the qualification. A seat supplier change after the certificate can technically void coverage until a median-size retest. Ask when the certificate was last supported by a valid test of the current material configuration.
The low-pressure seat test is the pass/fail step, not the burn. The tight post-fire requirement is at 2 bar, and that is where marginal secondary-seat geometry shows up. When witnessing or reviewing reports, look at the low-pressure numbers first - that is the line the valve will live near.
Ten lines carry most of the value:
Name the standard and edition: "Fire tested per API 607, 7th edition, 2016" (or ISO 10497:2022 / API 6FA as applicable).
State the valve types covered - do not let a ball valve certificate be quoted for a gate valve.
Require the full fire test report, not the certificate: test date, facility, flame and calorimeter temperature traces, cavity pressure trace, measured leak rates per phase, operation torque after cooldown.
Require the tested valve's size, class, bore, body material, and the derived qualified range, and state your ordered sizes/classes within it.
Require the nonmetallic materials of the qualified configuration (seat, seat-to-body, stem packing, gasket) to be listed, and the certificate to remain valid for those actual suppliers.
State that the secondary metal seat, graphite packing, graphite-faced gasket, and anti-static device are required design features, verified in the test report.
For double-seated valves, require the cavity pressure record against the 1.5-times-rating cap.
For actuated valves, state whether the actuator interface and stem extension are covered, or require them separately (the fire test standard itself only covers the valve envelope).
Name the witness level: third-party witnessed test and accredited facility, or manufacturer self-declaration with report review - self-certification is permitted by the standards, so the witness requirement is your decision, not theirs.
Require the certificate to be traceable to a named facility (for example an independent laboratory) with the report signed by a responsible engineer, as in the published reports where a licensed professional engineer certifies the result.
A fire-safe valve certificate is one of the most valuable papers in a hydrocarbon procurement file - and one of the most frequently accepted without being read. The report itself answers every question that matters: which edition was the test run to, what did the flame and calorimeter actually reach, how much did the valve leak in each phase, what cavity pressure did it see, and what torque did it take to open afterwards. Those five answers, checked against the extension rules and your ordered configuration, are the difference between coverage and a filing-cabinet decoration. J-VALVES supplies fire-safe qualified ball and butterfly valves with complete test documentation for hydrocarbon, offshore, and LNG service - bring us your valve schedule and we will match each line item to a valid tested envelope.
Does API 607 certification mean the valve is leak-tight after a fire?
No, and the standard never claims it. The 7th edition allows up to 400 mL/min per inch of NPS through the seat during the burn and still 80 mL/min per inch at the post-fire low-pressure test - for an NPS 4 valve that is 320 mL/min allowed after the fire. What the certificate proves is that leakage is bounded and the valve remains operable, so the shutdown system can still isolate fuel. Bubble-tight shutdown in normal service is a separate requirement under ISO 5208 or API 598, and the two claims are tested by entirely different procedures.
Is an API 607 certificate valid for every size and class of the same model?
No - it is a type test with defined extension rules. A tested valve qualifies sizes up to twice its own nominal size (an NPS 8 test covers all larger sizes, an NPS 2 test covers all smaller), pressure classes up to twice its own rating, and alternate end connections provided the valve mass is at least 75% of the tested valve. Class 600 NPS 6 is not covered by a Class 300 NPS 4 test. Ask for the test report, confirm the tested size, class, bore and material configuration, and check your ordered items fall inside the derived range.
Can API 6FA be substituted for API 607, or vice versa?
Not automatically. API 607 covers quarter-turn valves and valves with nonmetallic seats, tested closed, with the low-pressure seat test before any operation. API 6FA serves valves built to API 6A and API 6D - including gate and globe valves - tested fully open, and it cycles the valve before the final leak evaluation, so the protocols genuinely differ. Check valves have their own standard, API 6FD. Name the standard and edition your datasheet intends; many EPC specs require API 607 for process valves and API 6FA for pipeline valves in the same project.
Why did a valve pass the fire and leak during the low-pressure test afterwards?
Because after the soft seat is consumed, sealing relies on the metal secondary seat, whose contact band is narrow and depends on line pressure and geometry rather than soft-material compliance. At high differential the band seals; at the 2 bar post-fire test a distorted or too-narrow band weeps, and the 80 mL/min-per-inch limit is exactly where marginal geometry shows. Quench distortion and ball-to-seat micro-welding during the burn make it worse - one facility attributes around 15% of failures to the opening moment after cooldown, when torque can spike to several times normal.
What should a fire-safe purchase specification contain?
Ten lines carry most of it: named standard and edition (API 607 7th edition 2016, or ISO 10497:2022, or API 6FA as applicable); the valve types covered; the full test report, not just the certificate - flame and calorimeter traces, cavity pressure record, per-phase leak rates, post-fire operating torque; the tested size, class, bore and configuration with the derived qualified range; the nonmetallic materials as tested and a commitment that seat or packing supplier changes trigger requalification; the four design features (metal secondary seat, graphite packing, graphite-faced gasket, anti-static device) verified in the report; cavity pressure records for double-seated valves; separate coverage for actuator interfaces; and the witness level with an accredited facility and engineer-signed report.
