Views: 0 Author: J-VALVES Publish Time: 2026-10-31 Origin: Site
LNG is stored, pumped and transferred at approximately -162 °C. A cryogenic ball valve that serves it is machined, fitted and seat-tested in a workshop at around +20 °C, and it will be overhauled at +20 °C. Everything that makes the seat seal - the elastic interference between the seat ring and its pocket, the contact stress between seat face and ball, the preload in the springs - is established at room temperature and then asked to survive a 180-kelvin drop. When an LNG ball valve leaks internally at temperature but passed its ambient seat test, the cause is almost never the chemical compatibility of the polymer. It is a preload problem created by differential thermal contraction, and it has to be engineered out with arithmetic, not fixed with a softer material.
This article works through that arithmetic, explains the two countermeasures (cold preload and spring energization), compares the seat polymers that actually survive -162 °C and below, walks through the ISO 28921-2 and MSS SP-134 cold-test sequences with their verified leak limits, and closes with failure signatures and a procurement checklist. It deliberately does not repeat the material and bonnet-extension topics already covered in our LNG ball valve material and bonnet extension guide, and it assumes the seat leakage class definitions explained in our seat leakage classes article.
The starting point is a property table that every cryogenic valve engineer keeps within reach. Integrated from +20 °C down to liquid-nitrogen temperature (77 K, -196 °C), the total linear contraction of common materials spans nearly an order of magnitude:
Material | Mean CLTE near 20 °C (x10-6/K) | Total contraction 293 K → 77 K |
|---|---|---|
PTFE (virgin) | 100-160 (about 135 typical) | about 1.93% |
Aluminium 6061 | 23.4 | about 0.415% |
Austenitic 304/316 stainless | 16.0 | about 0.296% |
Carbon steel (mild) | 11.7 | about 0.200% |
9% nickel steel | 10.0 | about 0.185% |
Invar (Fe-36Ni) | 1.2 | about 0.040% |
Values are representative figures from published cryogenic-engineering references; exact values depend on grade and temper, and integrated contraction from 293 K to 111 K (LNG temperature) is somewhat less than the 77 K figures - of the order of 85-90% of them. The engineering point survives the imprecision: a PTFE seat ring shrinks roughly six times more than the austenitic steel around it and ten times more than carbon steel.
Now run a DN100 (NPS 4) example. The seat ring has an outside diameter of roughly 140 mm seated in a 316 body pocket, and an inside sealing face on a ball of roughly 110 mm diameter:
PTFE ring OD shrink: 140 mm x 1.6% ≈ 2.2 mm on diameter.
316 body pocket shrink: 140 mm x 0.27% ≈ 0.38 mm on diameter.
Differential: about 1.8 mm diametral, close to 0.9 mm on radius - the ring pulls away from the pocket by close to a millimetre.
Ball drop-away: 110 mm x 0.27% ≈ 0.3 mm of diametral shrink moves the sealing line inward as well.
Compare those numbers with the dimensions that actually generate seal contact stress: an engineered interference of 0.1-0.3 mm, and the elastic deflections at the sealing line of 0.05-0.15 mm. The contraction movements are one order of magnitude larger than the working deflections. Whatever load path the seat design uses, the assembly lost its preload during cooldown, and no warm workshop measurement would ever show it. A documented case illustrates the failure mode in practice: a cryogenic ball valve developed minor internal leakage after commissioning, and inspection found a 0.15-0.20 mm gap on the seat seal face - not because the body shrank more than calculated, but because a glass-filled PTFE backup ring had been substituted with shrinkage characteristics that did not match the seat polymer itself.
The first countermeasure is purely geometric: assemble the seat with deliberately oversized interference so that after differential contraction there is still positive interference left at -162 °C. This is the "cold preload" technique, and it is standard practice across LNG valve manufacturers. A widely quoted design target is that the seat preload retained at -196 °C should be no less than about 60-70% of the room-temperature value; treat the specific number as vendor practice rather than a standard requirement, but the direction is universal.
Cold preload is a trade, not a free lunch. The interference that produces useful contact stress at -162 °C produces proportionally higher contact stress and friction at +20 °C, so breakaway torque rises and the soft material creeps under a load it never carries in warm service. Three consequences follow:
Torque must be calculated at both ends of the temperature range. The actuator or gearbox is sized for the worse case, which is frequently the cold, high-preload condition immediately after cooldown. Our pneumatic actuator sizing guide covers the safety-factor arithmetic.
Thermal cycling ratchets the preload down. Each warm excursion lets the compressed PTFE creep and relax; each cooldown re-drops the parts by the full differential. After many cycles the as-assembled interference is no longer what the shop built. Spring energization (below) is the robust answer when cycling is frequent.
Every polymer in the load path must be matched. The backup-ring case above is the lesson: glass-filled PTFE shrinks differently from virgin PTFE, so the entire seal stack - seat, backup ring, retainer, bolting - has to be contracted as a system, not part by part.
Room-temperature valve practice reaches for RPTFE (typically 15-25% glass or carbon filled) to raise compressive strength and crush resistance. At cryogenic temperature that instinct is backwards. Virgin PTFE becomes brittle below roughly -50 °C, and filled grades embrittle at even higher temperatures; a glass-filled seat insert can crack outright on the first cooldown if it is not a cryogenic grade. Virgin PTFE, paradoxically, keeps the best low-temperature ductility of the PTFE family, which is why cryogenic seats are usually specified as virgin or modified PTFE rather than the filled grades that dominate ambient service.
The second problem is dimensional. Fillers cut the thermal expansion coefficient - that is their job in hot service - but in a cryogenic seat the filler changes the integrated contraction and, worse, makes it different from every adjacent part. A seat stack assembled from mixed grades contracts unevenly and tears itself loose. The rule of thumb for LNG seats: one polymer, one contraction curve, and a backup ring selected to match it. If compressive strength is genuinely insufficient, the answer is a spring-energized design or a different polymer (PCTFE, PEEK), not a hotter-service filler package.
Polymer | Typical usable range | Behaviour at -162 to -196 °C | Where it is specified |
|---|---|---|---|
Virgin PTFE | about -200 to +180 °C (with support) | Loses most resilience, needs preload help; keeps ductility better than filled grades | LNG and LN2 seats with cold preload or springs; the volume default |
PCTFE (Kel-F type) | about -240 to +120 °C | Retains roughly 8% elongation at -196 °C - it does not shatter on cooldown; stiffer and less creep-prone than PTFE | Liquid oxygen seats (PTFE is combustible under pressure in LOX); premium LNG duties |
PEEK (unfilled) | about -196 to +260 °C | High modulus maintained cold - carries load, but stiffness works against conformance on out-of-round balls | High-pressure cryo seats; often as a backed ring under a PTFE face |
UHMW-PE | about -200 to +80 °C | Excellent cold toughness, low load capacity | Low-pressure cryo duties, limited by temperature ceiling |
Metal seat + hardface | full cryo range | No polymer limits; depends on lapping, hardface integrity and load | Fire-safe ESD duty, abrasive or cleaning service, high class |
Temperature ranges are typical vendor-guide figures, not standard ratings; always check the manufacturer's pressure-temperature curve for the specific compound, because a polymer that survives the temperature may still be beyond its pressure envelope. For liquid oxygen service the choice is forced, not stylistic: PTFE can burn under pressure in an oxygen-rich environment, so PCTFE is the seat material of record.
The second countermeasure removes the dependence on interference altogether: put a metal spring behind the polymer seat and let the spring, not the thermal state of the assembly, deliver the seat load. Three families are in common use:
Helical coil springs distributed around the seat circumference - the traditional LNG ball valve arrangement, robust and easy to inspect.
Canted coil (slanted-wire) springs in a jacket - flat load-deflection over a wide travel range, which suits seats that move with ball wear.
Lip or finger springs machined as part of the seat carrier - continuous peripheral load with no loose parts.
Spring material selection is a cryogenic problem in its own right: ordinary austenitic spring steels lose spring temper and can transform metastably at cryogenic temperature, so cryogenic seat springs are typically 316-family, Elgiloy (cobalt-chromium-nickel) or Inconel X-750. One manufacturer's LNG design quote is instructive for magnitude: an Elgiloy lip spring delivering on the order of 45 pounds of thrust per inch of circumference, which for a DN100 seat face is roughly a kilonewton of peripheral preload - enough to hold the shrunken, stiffened polymer against the ball at 2,500 psi cryogenic liquid pressure. Treat the specific figure as one vendor's data point, not an industry constant; the design lesson is that the spring must be sized against the cold contact stress target, exactly as the cold preload is.
Spring-energized seats also change the low-pressure sealing story. A seat held by springs seals at the 2-bar low-pressure cold test even after full contraction, which is precisely the condition where interference-only designs are weakest - the low-pressure end of the test is where cold preload loss shows first, because line pressure cannot help push the seat home.
Soft seats are the default for LNG isolation, but three situations push a cryogenic ball valve to a metal seat: the valve is an emergency shutdown (ESD) duty where fire-safe certification to API 607 is required and the soft seat is assumed consumed in the fire scenario; the service carries particulates or hydrates that cut soft faces; or the pressure class and seat-load requirements exceed what the polymer can carry. Metal seats at cryogenic temperature do not escape the contraction problem - they manage it with geometry (the wedge or spherical-fit contact) and with hardfacing whose interface must also survive a 180 K swing without spalling, which is why the hardface-to-substrate thermal expansion match matters even more here than in ambient service. Selection of the hardface alloy itself - Stellite family versus tungsten carbide - is covered in our hardfaced trim article; the cryogenic additions are a matched-coefficient substrate choice and lapped-in spherical seating verified by cold test, not just ambient leak inspection.
Soft-seated cryogenic ball valves are usually single-piston-effect (SPE) seats so that cavity pressure relieves toward the line - the piston-effect geometry and the DBB/DIB variants are explained in our ball valve seat design article. At LNG temperature the trapped cavity volume contains liquefied gas, and the stakes change: liquid LNG flashing to atmospheric gas expands roughly 600-fold by volume. A cavity that traps liquid between two DIB seats and warms even a few kelvin generates pressure well beyond any valve rating in minutes, not hours. For cryogenic ball valves this is not an accessory decision - a full-bore cavity relief valve, sized and documented for the trapped-liquid case, is a mandatory line item, and the seat configuration (SPE, or DIB with an external relief valve per API 6D) must be declared on the datasheet rather than assumed.
Cryogenic seat performance is proven by type test. Three frameworks dominate LNG specifications, and they are complements rather than substitutes:
Framework | Nature | Test temperature | What it proves |
|---|---|---|---|
ISO 28921-2:2015 | Type test (first article + sampling) | -196 °C ±5, or -50 °C ±5, or an agreed intermediate | Full cold sequence: ambient baseline, purge, cooldown, low- and high-pressure seat tests, 200 operating cycles, external tightness |
MSS SP-134 | Practice for low-temperature valves | Down to -196 °C in liquid nitrogen | Cold seat leak limits in helium per class, plus bonnet extension dimensions that BS 6364 does not cover |
BS 6364 | Per-valve factory test (in most project uses) | -50 to -196 °C range | Every-valve cold verification before shipment; the most cited mandatory clause set in LNG project specs |
MSS SP-134 publishes the clearest quantitative seat-acceptance numbers, expressed as allowable helium leakage in standard cubic centimetres per minute per inch of nominal pipe size during the cold closure test:
Class | Gate | Globe | Butterfly | Ball |
|---|---|---|---|---|
150 | 25 | 25 | 50 | 50 |
300 | 50 | 50 | 50 | 50 |
600 | 100 | 100 | 100 | 100 |
900 | 150 | 150 | 150 | 150 |
1500 | 150 | 150 | 150 | 150 |
(Class 800, for socket- and threaded-end valves, allows 25 for gate and globe.) For an NPS 4 Class 600 cryogenic ball valve the acceptance line is 400 sccm total. Note that these cold helium numbers are one to two orders of magnitude looser than the ambient soft-seat expectations under ISO 5208 Rate A - the industry has learned what cold metal-to-polymer contact realistically delivers. A commonly quoted BS 6364 acceptance line for helium seat leakage is 1x10-5 Pa·m3/s per seal; verify the number your project actually cites, since project specs frequently tighten or relax it. Ambient leakage class definitions themselves are treated in our leakage classes article.
The ISO 28921-2 and MSS SP-134 sequences converge on the same logic; run in order:
Ambient baseline. Shell test at 1.5x cold working pressure (water) or 1.1x (gas); seat test at about 6 bar, soft seats to ISO 5208 Rate A. Then the valve is dried thoroughly - this matters more than it sounds.
Purge and mount. The cavity is purged with clean, dry gas (about 2 bar) to strip every trace of water from the warm hydrotest. Test hardware is austenitic; the valve is blinded and set in an insulated tank.
Cooldown. Liquid nitrogen is admitted until the level stands roughly 25 mm above the body-bonnet joint. Gate and globe valves go in part-open; ball valves go in fully open, with the purge maintained. Cooling is controlled (a common recommendation is no faster than about 5 °C per minute) to avoid thermal-shocking the polymer seat. Three thermocouples minimum: body, packing area, and one on the inlet line.
Stabilize and cycle. At -196 °C the purge stops and the valve is stroked - three open-close cycles under MSS SP-134, 200 cycles over the full ISO 28921-2 on-off sequence. This is where ice jams, preload collapse and binding all reveal themselves.
Low-pressure cold seat test. Helium at low pressure (5.5 bar under MSS SP-134; 2 bar under ISO 28921-2), closed and stabilized about 10 minutes, leakage collected over about 5 minutes, measured against the table above - then repeated for the second seat.
High-pressure cold seat test. Pressure raised stepwise toward the cold working pressure per the standard's pressure table, same stabilization and measurement discipline.
External tightness. Packing and body joints checked cold, by soap solution or helium sniffing - the point of the extended bonnet is that this zone stays warm enough to seal, which is the subject of our LNG bonnet extension guide linked above.
Four traps account for most failed cold tests that have nothing to do with seat design. Moisture: residual water from the warm hydrotest freezes, jams the ball and shears seats; the dry purge is not optional. Test gas quality: specifications call for high-purity helium (typically 97% or better) because contaminant gases condense and pollute the leak measurement. Cooldown rate: dunking a soft-seated valve straight into liquid nitrogen can crack the seat before any pressure is applied. Stabilization discipline: measuring before the valve mass reaches temperature produces leaks that are really warm spots, and passes that are really frozen test lines.
And the reason the ambient test lies, one more time, in one paragraph: at +20 °C the seat stack has its full designed interference, PTFE is in its tough regime, and the sealing line is pressed home exactly as the designer intended. Every mechanism that destroys cold sealing - six-to-one differential contraction, polymer embrittlement, spring temper change, ice - is absent. The ambient seat test proves the valve was built to print; only the cold test proves it seals at -162 °C. If your project applies an ambient ISO 5208 or API 598 test as the seat acceptance for LNG service, it has tested the wrong condition - the API 598 procedure itself is explained in our API 598 testing guide.
The seat does not seal alone. Body material must carry Charpy energy at the minimum design temperature - LCB and LCC grades are the -46 °C workhorses, LC3-type steels reach about -101 °C, and the -162 °C LNG and -196 °C duties belong to austenitic 304/316 or 9% nickel steel - and low-temperature bolting (A320 L7 family for intermediate cold service, austenitic B8/B8M for the deepest) because ordinary A193 B7 loses its low-temperature toughness rating below about -29 °C. The packing zone is kept above roughly 0 °C by the extended bonnet so that stem sealing stays in a regime where flexible graphite works; the length tables, the no-insulation rule on the extension and the material boundaries are covered in our LNG ball valve guide. Anti-static path and fire-safe certification round out the package for ESD duties - the fire-test sequence and its post-fire acceptance numbers are the subject of our API 607 testing article.
Signature | Most likely cause | Corrective path |
|---|---|---|
Sealed warm, leaks cold from the first cooldown | Preload collapse - interference not compensated for differential contraction | Recalculate cold preload; consider spring-energized seats; verify backup ring polymer matches seat |
Leaks at 2 bar cold test, tight at full CWP | Low-pressure path - seat load insufficient without line-pressure help | Increase spring preload or seat load; check sealing-line width and lap quality |
Seat ring cracked after first cooldown | Filled PTFE embrittlement; or thermal-shock cooldown too fast | Virgin PTFE or PCTFE; control cooldown rate; re-run cold test |
Gap of 0.1-0.2 mm found on seat face after service | Mixed-contraction seal stack (backup ring or retainer mismatch) | Contract the stack as a system; single-polymer rule |
Intermittent leak after repeated thermal cycles | Preload ratcheting - PTFE creep on warm excursions relaxes interference | Spring energization; re-torque practice; document cycle count in spec |
Torque spikes and binding right after cooldown | Ice in cavity (moisture carryover) or excessive cold interference | Verify purge/dryness practice; review cold preload margin; check ball sphericity |
External leak at packing on cold service | Bonnet too short - packing zone reached cryogenic temperature | Extend bonnet per length tables; this is a design fix, not a packing change |
Minimum design temperature and test framework named: ISO 28921-2 type test, MSS SP-134 limits, or per-valve BS 6364-style cold test, with the witness level stated.
Seat polymer named by grade, not by "PTFE": virgin PTFE or PCTFE for the cold zone, with filled grades explicitly excluded unless the vendor justifies them.
Spring type and material declared: helical, canted coil or lip; 316/Elgiloy/Inconel X-750; spring load calculated for the cold contact stress target.
Cold preload evidence: the seat load or interference calculation evaluated at the minimum design temperature, not only at assembly.
Full type test report, not a certificate: cooldown trace, stabilization records, per-seat leak values at low and high pressure, cold operating torque.
Per-valve cold test option for critical ESD and tank-side duties, priced and scheduled, since type-test coverage is statistical.
Cavity relief sized for trapped liquid: SPE seats, or DIB with an external relief valve declared per API 6D; relief valve sized against the 600-to-1 flash case.
Low-temperature bolting and body grades stated: A320 L7 or B8/B8M; body Charpy verified at minimum design temperature.
Cleanliness and dryness requirements: degreased, dry nitrogen purge provision, moisture limits - the cheapest leak prevention in this entire article.
Fire-safe and anti-static where ESD applies: API 607 certification with the report, since the fire case and the cold case are tested by different procedures and both must pass.
Why does a cryogenic ball valve that passed its ambient seat test leak at LNG temperature?
Because the ambient test proves build quality, not cold sealing. A PTFE seat ring contracts about 1.6-1.9% of its dimension on cooldown while the steel around it contracts about 0.3% - on a DN100 seat that differential opens on the order of 0.9 mm on radius, against a designed interference of 0.1-0.3 mm that generates the actual seal contact stress. The preload is gone before any pressure is applied, and every mechanism that causes it is absent at room temperature. Only a cold test to ISO 28921-2, MSS SP-134 or a BS 6364-style per-valve test exercises the seat at temperature.
Which seat material should be specified for LNG at -162°C, liquid nitrogen at -196°C and liquid oxygen?
For LNG, virgin PTFE with cold preload or spring energization is the volume default, because virgin PTFE keeps better low-temperature ductility than filled grades. PCTFE is the premium choice - it retains roughly 8% elongation at -196°C and does not shatter on cooldown. For liquid oxygen the choice is forced: PTFE is combustible under pressure in LOX, so PCTFE is the seat material of record. PEEK serves high-pressure cryogenic seats, usually as a backed ring, and metal seats with matched hardfacing take over for fire-safe ESD or abrasive duties.
Is a cryogenic type test certificate valid for every valve in the order?
No. ISO 28921-2 is a type test - first-article qualification plus periodic sampling - so it statistically represents the design, not each machined valve. For critical ESD and tank-side duties, projects add a BS 6364-style per-valve cold test before shipment. Either way, demand the full report, not the certificate: cooldown trace, stabilization records, per-seat helium leak values at low and high pressure, and cold operating torque. Confirm the tested size, class and seat configuration actually represent your items.
Why is the dry nitrogen purge before cooldown emphasized so heavily in the test procedure?
Residual water from the warm hydrotest freezes at cryogenic temperature, and ice jams the ball, shears seat faces and fakes both leaks and passes. The standard sequences purge the cavity with clean dry gas at about 2 bar throughout cooldown, and specifications typically demand 97%-pure helium or better as the test medium so contaminant gases do not condense into the measurement. Controlled cooldown - commonly no faster than about 5°C per minute - completes the protection by avoiding thermal shock that cracks soft seats before any pressure is applied.
When is cold preload alone enough, and when are spring-energized seats required?
Cold preload - assembling with oversized interference so 60-70% of the seat load survives at the cold end, a widely quoted vendor practice - is a geometric countermeasure that works well where thermal cycles are few. Each warm excursion lets the compressed PTFE creep and relax, so frequent cooldowns ratchet the interference down and only a metal spring behind the seat holds the load independently of the thermal state. Spring-energized designs also seal the low-pressure cold test more reliably, because line pressure cannot help push a shrunken seat home. Size the actuator for the cold high-preload torque either way.
