A 20 inch Class 600 ball valve on a natural gas export line developed a seat leak that measured 300 parts per million by volume on the ultrasonic meter at the downstream flange. The leak was not a safety emergency and it was not going to be fixed by anything short of a shutdown, which was fourteen months away. The maintenance team connected a hand pump to the seat injection fitting, injected 45 cubic centimetres of PTFE paste at 75 bar against a line pressure of 60, and the leak fell to 40 parts per million within the hour. It stayed there until the planned outage, when the valve was replaced and the seat was found to have a single score about 0.3 millimetres deep running radially across the contact band.
That is a good outcome and it is the outcome the fitting is designed for. It is also the exception, because most sealant injection jobs in most plants are done by people who do not know which cavity the fitting leads to, how much pressure to apply, or what the sealant does after it stops moving. The result ranges from a temporary repair that saves a shutdown to a seat that is hydraulically locked against the ball and an actuator that stalls on the next stroke. This article is about the difference.
1. What the fitting actually connects to
A sealant injection fitting is a check valve with a grease nipple on the outside and a drilled passage on the inside, and where that passage ends is the whole question. On a trunnion mounted ball valve there are three possibilities, and they behave completely differently. The seat injection port leads into the annular space behind the seat ring, between the seat and the body, so that sealant pressure pushes the seat ring towards the ball. The stem injection port leads into the packing box, usually between two rings of packing or into a lantern ring, so that sealant displaces the leakage path along the stem. The body or cavity port leads into the space between the two seats, which is the cavity, and filling it does not seal anything at all unless the seat has a relief passage that lets the sealant reach the seat face.
Getting this wrong wastes a cartridge and, worse, creates a false record. An engineer who fills the cavity through the cavity port, sees the pressure hold and closes the work order has proved that the cavity holds pressure, not that the seat seals. The leak will still be there, and the next person will assume the seat has already been treated.
The way to be certain is a drawing, not a guess. Ask the supplier for the body passage diagram, or trace it on site with the valve isolated and the cavity vented. A simple test is available: with the valve closed and the line pressurised, open the injection fitting slowly and see what comes out. Process fluid means the passage communicates with the seat face or the cavity under pressure; nothing means it leads to a dead annulus; sealant from a previous job means someone has been here before.
Injection point | What it can seal | What it cannot do | Typical fitting |
|---|---|---|---|
Seat annulus, upstream | Seat to ball leak from wear, score or particle damage | Repair a cracked seat ring or a ball with a deep groove | 1/4 inch NPT with check valve, one per seat |
Seat annulus, downstream | Seat leak on the downstream side, and cavity pressure build up | Seal if the relief passage is open to the cavity | 1/4 inch NPT, often paired with the upstream port |
Stem packing box | Stem leakage through packing, gland or O-ring | Repair a scored stem or a failed lip seal lip | 1/4 inch NPT into a lantern ring |
Body cavity | Nothing directly; used for flushing and for pressure relief | Seal a seat, unless a relief passage carries sealant to the face | 1/2 inch NPT with plug and drain |
Body shell or flange | Nothing; a through wall defect is a pressure boundary failure | Any repair at all; this is a shutdown item | Not applicable |
2. Pressure: the rule is a margin, not a maximum
The injection pressure has one job, which is to be higher than whatever is pushing back on the other side of the seat, and only just higher. A practical rule is line pressure plus 15 to 35 bar, or about 1.5 times line pressure where the line is below 40 bar. Going higher does not seal better; it deforms the seat insert, extrudes a soft seat through its clearance, and can hydraulically lift the seat ring so far that the ball is pinched between the two seats and the valve will not turn.
The failure mode is worth describing because it is common and it is misdiagnosed. A soft seated valve is over-injected until the seat ring is pushed hard against the ball. The sealant is incompressible, the annulus is closed, and the seat has nowhere to go. The actuator then tries to rotate a ball that is clamped between two seat rings at several times the normal seat load, the torque demand triples, and the valve either stalls or shears a key. The report says the actuator failed. The cause was a grease gun.
The correct procedure has three parts. Inject slowly, in small increments, and pause between them so that the pressure can equalise through the seat contact. Watch the injection pressure gauge rather than the pump handle; a sudden rise means the annulus is full and further pumping is doing damage. And after injection, wait fifteen to thirty minutes before stroking the valve, so that the sealant can migrate into the leak path rather than being squeezed straight out again.
3. The seat design decides whether injection works at all
Sealant injection assumes that the seat ring can be pushed towards the ball by pressure applied behind it, which is true for a single piston seat and for a double piston seat, but the two behave differently when the cavity is involved. A single piston seat is pressure energised from the upstream side only: line pressure enters the annulus through a small passage and pushes the seat onto the ball, while cavity pressure acts on the back of the seat in the opposite direction. A double piston seat is energised from both the line and the cavity, so it seals in either direction, which is what a double block and bleed valve needs.
Now consider what happens when sealant is injected into a seat annulus that has a relief passage to the cavity. If the relief passage is open, which it is by design in a self relieving seat, the sealant takes the easy path into the cavity instead of building pressure behind the seat. The gauge shows almost nothing, the leak does not change, and the operator concludes that the fitting is blocked. The real diagnosis is that the seat is self relieving and the injection port is on the wrong side of the relief, or that the relief is passing because the seat is not seated.
There is a simple field check. Vent the cavity to a safe location with the valve closed and the line pressurised, then inject. If sealant appears at the cavity vent, the relief passage is passing and the injection is going into the cavity, not into the leak path. Stop, and treat the valve as one that needs a seat replacement rather than an injection.
Double block and bleed adds a further complication that catches people out. With two seats and a cavity between them, a leak could be past the upstream seat, past the downstream seat, or through both. Injecting the upstream seat will not fix a downstream seat leak, and the cavity pressure will keep rising. The correct sequence is to determine which seat is passing by venting the cavity and watching which side repressurises, and only then to inject the seat that is actually leaking.
4. Stem sealing is a different problem with a different answer
The injection port on the stem leads into the packing box, and what it can achieve depends entirely on the packing arrangement. A graphite packing set with a lantern ring can be injected with a high temperature paste that fills the lantern space and displaces the leakage path, and this is a recognised temporary measure on high temperature valves where the packing has dried out. An O-ring energised stem seal cannot be injected in any useful way, because there is no space for the sealant to occupy and the O-ring is a line contact that either seals or does not. A lip seal is similar: the sealant has nowhere to go except past the lip.
This is why the low emission generation of ball valves, with live loaded graphite packing or with a dual seal arrangement and a tell tale port, is easier to maintain than the older O-ring designs. The tell tale port between two seals is a diagnostic, not a repair point: it shows which of the two seals has failed, and it allows a secondary seal to be energised while the primary is replaced at the next opportunity. A valve without a tell tale port gives no such information and the leak is only visible when it reaches the insulation.
The temperature ceiling is set by the sealant, not by the valve. PTFE paste is good to about 200 degrees C, silicone grease to 230, perfluorinated grease to about 280, and graphite based products higher but with poor injectability. Above the ceiling the sealant carbonises, the annulus blocks, and the fitting becomes a decorative item that cannot be used when it is finally needed.
Sealant type | Temperature limit | Cures? | Where it is right | Where it is wrong |
|---|---|---|---|---|
PTFE paste in oil | 200 degrees C | No, stays soft | General seat sealing, gas and liquid | High temperature, or where washout matters |
Silicone grease | 230 degrees C | No | Stem packing, low differential seat | Hydrocarbon service, swelling risk |
Perfluorinated (PFPE) grease | 280 degrees C | No | Oxygen, high temperature, chemical | Cost, and poor load carrying |
Epoxy sealant | 150 degrees C | Yes, hard | Permanent plugging of a small defect | Any valve that must be stroked or rebuilt |
Polyurethane sealant | 80 degrees C | Yes, elastic | Water service, low pressure | Hydrocarbons, high temperature |
Graphite dispersion | 400 degrees C and above | No | Steam and high temperature packing | Injection through small fittings, it blocks |
5. The volume arithmetic, and why a small cartridge goes a long way
The annulus behind a seat ring is thin. For a 20 inch valve the seat circumference is about 1.6 metres, the gap between the seat ring and the body bore is typically 0.3 to 0.8 millimetres, and the pocket depth is 3 to 6 millimetres. Multiplying those gives a volume of 3 to 8 cubic centimetres, and the drilled passage adds one or two more. A standard 400 gram cartridge holds roughly 400 cubic centimetres, which is enough for twenty or more seats, and that is why a single cartridge in a maintenance store tends to be used on five different valves by five different people with no record of any of it.
The useful calculation is the one that tells you when to stop. If the gauge does not rise after injecting two or three times the annulus volume, the sealant is going somewhere: into the cavity through a passing relief, into the line past a failed seat, or out through a fitting that is not seated. Continuing to pump in that situation pressurises the cavity or the process, neither of which is intended.
Injection should always be done with the valve in the closed position, so that the seat is against the ball and the sealant has a face to seal against. Injecting an open valve fills the annulus and then the line. After injection, the valve should be left closed for the settling period, and then stroked once to confirm that it still moves and that the torque has not risen. If the torque has risen sharply, the seat is over-loaded and the valve should be left closed until the sealant has had time to migrate.
6. Four leaks it fixes, and four it does not
Sealant injection works when the leak path is small, distributed, and accessible from the annulus. Wear of the seat insert, a score across the contact band that is shallow relative to the sealant's ability to fill it, a particle trapped between the seat and the ball that has left a dent, and a packing that has dried and shrunk are all good candidates. In each case the sealant fills a gap of tens of micrometres over a length of millimetres, and the geometry holds it in place.
It does not work when the defect is larger than the sealant can bridge, or when the geometry cannot retain it. A deep score that runs radially across the whole contact band is a channel with a low resistance path to the downstream side, and the sealant is washed out as soon as the valve is stroked. A cracked seat ring has a gap that opens and closes with temperature. A ball with coating loss has a rough surface that abrades the sealant away. A body wall defect is a pressure boundary failure and no sealant is relevant. And a seat that has lost its relief passage or its spring cannot be made to seal by filling the annulus, because the seat no longer tracks the ball.
Situation | Inject or replace | Reason | Expected duration of the repair |
|---|---|---|---|
Seat leak from wear, leak rate stable | Inject | Leak path is small and distributed | 6 to 18 months, to the next planned outage |
Seat leak after a particle passed | Inject, then monitor | Dent is local, sealant fills it | Months, but expect recurrence if the particle source remains |
Stem packing leak, graphite packing | Inject if a lantern ring exists | Space exists for the sealant | Weeks to months, re-inject as needed |
Cavity pressure rising with both seats closed | Neither | Seat or relief failure, injection will not hold | Shutdown item, add a cavity relief |
Deep radial score across the seat band | Replace | Channel too large to bridge, sealant washes out | Not repairable online |
Cracked seat ring or broken spring | Replace | Seat no longer tracks the ball | Not repairable online |
Leak at a body flange or a weld | Neither | Pressure boundary, not a seat issue | Shutdown, clamp or weld repair by procedure |
O-ring energised stem seal | Neither | No cavity for the sealant | Replace the seal at the next opportunity |
7. Verification: the number that goes in the register
An injection that is not measured is not a repair, it is an activity. Before and after, the leak rate should be quantified the same way, at the same pressure and temperature, using the same instrument. A portable ultrasonic leak detector with a quantified leak rate in cubic centimetres per minute, a downstream pressure decay test on a closed volume, or a rotameter on a vented cavity will all do, provided the same method is used both times. "It sounds better" is not a record.
The verification should also include a stroke test after the settling period, with the torque recorded if the actuator reports it, and a check that the injection fitting itself is not leaking. A fitting that is left finger tight after the job is a new leak point, and it is at the exact height where somebody will put a hand.
The register entry should carry the date, the valve tag, the injection point, the sealant product and quantity, the injection pressure, the line pressure and temperature at the time, the leak rate before and after, and the next inspection date. Thirty days is a sensible first recheck, then quarterly. A leak that returns within thirty days was never fixed, and a leak that returns at ninety days is telling you the sealant is being washed out and the valve needs to be planned for replacement rather than injected again.
8. Fire, fugitive emissions and the limits of a grease seal
Two regulatory points are usually missed. The first is fire safety. A sealant dependent seal has no fire resistance, because the sealant burns or decomposes and the metal to metal contact that would have provided a secondary seal was never lapped to do so. A valve that passes API 607 before injection may not pass it afterwards, and a valve whose seat is held only by sealant has no qualified fire performance at all. In hydrocarbon service that is a real risk, not a paperwork one.
The second is fugitive emissions. A stem injection that stops a visible leak does not necessarily bring the valve within the 100 parts per million threshold of ISO 15848-1, and it certainly does not make it a low emission valve. If the site is under an LDAR programme, an injected packing leak is still a monitored component and still subject to the repair deadlines, and recording it as repaired when it is above the threshold is a compliance problem as well as an environmental one.
The honest framing is that sealant injection buys time to plan. It converts an unplanned shutdown into a planned one, and that is worth a great deal: a planned seat replacement in an outage costs a fraction of an emergency intervention, and the production loss avoided is larger than both. It does not restore the valve to its original condition, and it should never be the last entry in a maintenance history.
9. Specifying the fittings so that they work when needed
Most injection fittings fail at the moment of use, not in service. The common causes are a fitting that has corroded shut, a check valve that is blocked with dried sealant from a previous job, a passage that was never drilled through at manufacture, and a fitting that is buried under insulation with no access hatch. All four are preventable at the specification stage and none of them are expensive.
Fittings should be 316 stainless steel, rated to the valve class, with a metal to metal seated cap rather than a plastic plug, and they should be identified on the general arrangement drawing with the seat they serve. The passage should be drilled through and proved at the works with a wire or a low pressure air test, and the record should be in the test package. Insulation should be cut back or a hatch provided, and the fitting should be accessible from a platform rather than from a ladder. On cryogenic and high temperature valves, the fitting should be extended away from the body so that the sealant in the passage is not at the service temperature.
These details sit inside a wider envelope at J-VALVES, where the production range runs from DN15 to DN1400, or 1/2 inch to 56 inch, in ASME Class 150 to Class 2500, with bodies in carbon steel, stainless, duplex and nickel alloys for service from minus 196 to plus 650 degrees C, shell proof at 1.5 times the cold working pressure and seat proof at 1.1 times to API 598 or API 6D, sealant injection ports drilled and flow tested on every valve where they are specified, and quotations issued within three working hours.
10. Procedure on one page
Confirm which cavity the fitting serves from the drawing, and confirm the valve is closed and the cavity vented. Measure and record the leak rate. Select a sealant whose temperature rating exceeds the service temperature by at least 20 kelvin. Connect a hand pump with a gauge that reads the expected injection pressure at mid scale, and inject in small increments, pausing to watch the gauge, until the pressure holds at line pressure plus 15 to 35 bar. Stop if the gauge does not rise after two annulus volumes. Wait thirty minutes, re-measure the leak, then stroke the valve once and record the torque. Cap the fitting, record everything, and set the recheck date at thirty days. If the leak is unchanged, vent the cavity and check whether sealant has reached the cavity, which tells you the relief is passing and the valve needs a seat rather than a cartridge.
None of that is difficult, and all of it is skipped in the average emergency. The fitting on the side of a ball valve is a genuine maintenance capability, and like most capabilities it rewards the person who knows what is on the other end of the hole.
11. Frequently Asked Questions
What pressure should sealant be injected at?
Line pressure plus 15 to 35 bar, or roughly 1.5 times line pressure where the line is below 40 bar. The pressure only has to overcome whatever is pushing back on the other side of the seat; more does not seal better. Over-injection is a real failure mode: the sealant is incompressible, the annulus is closed, and the seat ring is pushed so hard against the ball that the ball is clamped between two seats. The actuator then stalls or shears a key, and the report blames the actuator when the cause was the grease gun. Inject slowly, in increments, and watch the gauge rather than the pump handle.
How do I know which cavity the injection fitting leads to?
From the body passage drawing, or by tracing it on site. With the valve closed and the line pressurised, crack the fitting slowly and see what comes out: process fluid means the passage communicates with the seat face or the cavity, nothing means it leads to a dead annulus, and old sealant means somebody has injected it before. On a trunnion valve there are three possibilities, the seat annulus, the stem packing box and the body cavity, and only the first two seal anything. Filling the cavity through a cavity port proves the cavity holds pressure, not that the seat seals.
Why did the seat leak not change after injection?
Most often because the seat is self relieving and the sealant went into the cavity instead of behind the seat. Vent the cavity with the valve closed and the line pressurised, then inject: if sealant appears at the vent, the relief passage is passing and the injection is not reaching the leak path. The other common reasons are an injection passage blocked with dried sealant, a fitting that is not drilled through, or a leak that is on the other seat of a double block and bleed valve. Determine which seat is passing before injecting anything.
Can sealant injection fix a stem leak on any ball valve?
Only where there is a cavity for the sealant to occupy. A graphite packing set with a lantern ring can be injected successfully, and it is a recognised temporary measure on high temperature valves. An O-ring energised stem seal cannot, because there is no space and the O-ring is a line contact that either seals or does not; a lip seal is the same. Valves with a tell tale port between two seals are the easiest to manage, because the port shows which seal has failed and allows the secondary to be energised while the primary is replaced at the next opportunity.
Is a sealant repair acceptable in hydrocarbon service?
As a temporary measure to reach a planned outage, yes, provided it is recorded and monitored. It is not a restoration of the valve. A sealant dependent seal has no fire resistance, because the sealant burns or decomposes and the metal to metal contact that would provide a secondary seal was never lapped to do so, so a valve that passed API 607 before injection may not pass it afterwards. It also does not necessarily bring a stem leak below the 100 parts per million threshold of ISO 15848-1, which matters if the site is under an LDAR programme with repair deadlines.
