Differential pressure decay testing charges a mechanical seal assembly and a leak-tight reference volume with air at the same time, closes them off from each other, and reads a leak as the pressure difference that builds up between them.
On a seal, it is one way to run the static air test of the assembled seal, the check API 682 sets for every seal assembly before it ships. A pass shows that the faces, secondary seals and gaskets hold air while the seal stands still; it says nothing about how the seal leaks running.
How Differential Pressure Decay Differs From Absolute Pressure Decay
An absolute or gauge pressure decay test reads the test volume’s own pressure on one sensor and treats any fall over the hold time as a leak; a differential test reads the test volume against a reference volume charged to the same pressure and treats a growing difference between the two as a leak.
Sensitivity
The differential method finds smaller leaks than an absolute or gauge test, but by no fixed ratio: how much smaller depends on the two sensors being compared, the trapped volume and how stable the temperature is.
Heat From Filling
The differential method reduces the effect of the fill heat but does not remove it, so the tester still needs a stabilise time before it measures. Published work on differential leak testers puts the minimum stabilise time at about five to six times the thermal time constant of the trapped air, and longer at higher test pressure and larger volume.
On a new fixture, the working stabilise time comes from testing a known-good part repeatedly at longer and longer stabilise times until the readings stop scattering.
How the Test Cycle Runs
A differential test cycle runs in five steps: fill, stabilise, isolate, measure and exhaust.
- Fill: the test volume and the reference are charged together from the same supply to the test pressure.
- Stabilise: the two volumes stay connected to each other while the fill heat dissipates, and the supply is then shut off.
- Isolate: the valve between the two volumes closes, so the differential sensor starts from zero with both sides at the same pressure.
- Measure: over a fixed measure time, the tester reads how far the differential pressure grows and compares it with the reject limit.
- Exhaust: both volumes are vented and the part comes off the fixture.
How the Reference Volume Is Chosen
The reference is a leak-tight volume that matches the test side as closely as possible in size, temperature and the way it takes up and loses the fill heat. The closest match is a known-good part of the same design in an identical fixture.
A reference that fills through a different restriction, holds a different volume, or has thicker or thinner walls cools at its own rate after filling and puts a differential reading on a sound part.
How the Pass/Fail Limit Is Set
The limit starts as the leak rate a part may not exceed, and is converted into the rise in differential pressure that leak produces in the test volume over the tester’s measure time. With both volumes equal and the temperature settled, the leak rate follows from that rise:
QL = V · Δ(ΔP) / (ρ · R · θ · Δt)
Here QL is the leak flow in m³/s, V the test-side volume in m³, Δ(ΔP) the change in differential pressure in Pa over the measure time Δt in s, ρ the air density in kg/m³, R the gas constant of air and θ the air temperature in K. V is everything on the test side, the part’s free volume plus the fixture, tubing and sensor port.
For a pressure decay test that certifies a maximum leak rate, ASTM E2930 asks for test equipment and a test duration whose resolution is ten times finer than the specified limit and whose accuracy is four times finer, with the total pressure change kept under 10 % of the starting pressure.
Before a limit goes into production, it is checked on the tester itself. A known leak, either a calibrated orifice or a restrictor set against a flow meter, is fitted to a known-good part, and the tester has to read that rate and reject the part.
What Causes False Results
A false result comes from anything other than a leak in the part that changes the pressure on one side of the tester and not on the other.
- Part temperature: a part warmer or colder than the reference, straight from washing, lapping or a warm store, keeps changing temperature through the measure time. A part that is cooling reads as a leak; a part that is warming hides one.
- Short stabilise time: leftover imbalance from the fill heat is still in the reading when the measure time starts, and it can read as a leak or cancel one.
- Part movement under pressure: an O-ring that settles into its groove or a ring that shifts as pressure comes on enlarges the test volume, and the pressure fall reads as a leak.
- Leaks in the fixture or circuit: a leaking fixture seal, fitting or valve on the test side fails a sound part, and a leak on the reference side hides a real leak in the part.
- Trapped volumes: a pocket connected through a restriction, such as a partly plugged port or a poorly vented cavity, keeps filling or emptying during the measure time, which can mimic or mask a leak and stretches the stabilise time.
- Liquid in the leak path: water, lapping fluid or lubricant on the faces or O-rings can hold air for the length of a test and pass a seal that leaks when dry.
How the Test Is Applied to a Mechanical Seal
The seal is tested as an assembled unit: a fixture stands in for the seal chamber, the side the process sees in service is pressurised with air, on a dual pressurised seal the volume between the two seals is pressurised, and every port that is not part of the test is plugged, so air can leave only across the faces, the secondary seals and the gaskets.
In the API 682 air test, the assembled seal is pressurised to 1.7 bar (25 psi) and may lose no more than 0.14 bar (2 psi) of pressure in five minutes.
To apply the API 682 limit on a differential tester, convert its 0.14 bar drop into a leak rate with the fixture’s test-side volume, then convert that leak rate into the differential rise it gives over the tester’s own measure time.
For the reference, a known-good assembly of the same seal in a second identical fixture matches the part’s volume and heat behaviour best. A rigid volume equal to the fixture’s free volume is the simpler alternative.
Where the Method Falls Short on Seals
Differential pressure decay is the wrong test when the question is how the seal will leak running.
The Fluid Sealing Association notes that seals typically leak more running, once a fluid film separates the faces, so a pass on the static air test does not predict running leakage. The exception is a standby seal that has already run, which can leak more standing than running.
Some seal designs also need an air-test limit of their own. Gas seals and containment seals may be designed to run on a slight leakage, and a dual pressurised seal traps so little volume between its two seals that the API 682 air test becomes very sensitive on it.