A seal that passed the API 682 factory test can still leak 270 to 1,300 times the permitted rate once installed. I see it constantly — a technician spots a few drops at the gland, writes up a replacement work order, and pulls a seal that had months of life left. The problem is not the seal. The problem is that nobody measured the actual leakage rate before making the call.
Most mechanical seal replacements triggered by visible leakage are premature. Spend 20 minutes measuring the actual rate, and you will know whether that drip needs a work order or just a note in the log.
Why Every Mechanical Seal Weeps by Design
The Cubic Face Gap Relationship
Leakage rate scales with the cube of face separation — not linearly. The gap between sealing faces typically ranges from 5 to 50 microinches, and a tiny increase in that gap produces an exponential jump in leakage. Two identical seals in identical service can show wildly different drip rates simply because one face has 10 microinches more gap than the other. That cubic relationship is exactly why visual inspection alone cannot tell you whether a seal is failing.
Why Zero Visible Leakage Does Not Mean Zero Leakage
A properly operating seal requires a thin fluid film between its faces. That film is the lubrication mechanism — without it, the carbon face burns up in hours. Under standard operating parameters, the theoretical leakage rate works out to about 0.34 mL/hr. That is invisible. The fluid evaporates in the sealing gap before it ever reaches the atmosphere.
When you see zero visible leakage, that does not mean zero leakage — it means the seal is operating exactly as designed. When you see a few drops per hour, that does not automatically mean failure. The rate may still be well within acceptable mechanical seal leakage limits.
The most common mistake I see is treating any visible drip as a binary failure signal. Every seal leaks. The question is “how much, and is that rate increasing?”
How to Detect Seal Weepage in the Field
Visual and Manual Detection
Wipe the seal gland area completely dry with a clean rag, then watch. Time it. Five minutes of observation with a dry surface tells you more than a glance at a wet pump.
- Dry surface after 5 minutes — seal is operating normally, fluid film evaporating as designed
- Occasional drops forming slowly — weepage present but may be within limits; proceed to measurement
- Steady drip or stream — excessive leakage likely; measure immediately
- Discoloration or crystallized deposits — chronic weepage that has been evaporating over time; still possibly within limits
Check the drain port on the seal gland plate if one exists. Many API-plan installations route weepage to a collection point — a wet drain line means you have measurable flow.
One thing technicians miss: temperature affects everything. A seal that shows zero weepage at operating temperature may drip at startup because thermal contraction changes the face gap. Always note the pump’s operating state when you observe.
Automated Detection Systems
For critical applications, two sensor types dominate. Float-type sensors use a sealed float with an internal switch — liquid accumulation raises the float and trips a normally-closed contact. A broken wire also triggers the alarm, so you catch instrumentation failures too.
Probe-type sensors use conductive probes in an oil-filled chamber. Water entering the cavity lowers electrical resistance and triggers an alarm at 10-24 VDC. They respond faster than floats but cannot detect broken wires — a silent failure mode.
For most mechanical seal testing applications, manual detection during routine rounds is sufficient. Reserve automated systems for hazardous fluid service or unmanned installations.
How to Measure Leakage Rate
Timed Drip Collection
Place a clean container under the seal drain or gland area, start a timer, and collect for a minimum of 15 minutes. I prefer 30 minutes when production allows — longer collection times improve accuracy.
Count the drops or measure the collected volume. A useful conversion: 1 mL equals approximately 20 drops, so 60 drops per hour is roughly 3 mL/hr. For API 682 comparison, convert to grams per hour using the fluid’s specific gravity. Water is easy — 3 mL/hr equals 3 g/hr, well under the 5.6 g/h limit.
Volumetric Measurement
For dual seals with barrier or buffer fluid systems, check the reservoir level before and after a timed interval. Before you start, make sure the reservoir is isolated from the makeup supply — if the system auto-refills, you will never see the level drop. Close the makeup valve, record the starting level, and check back in 4 to 8 hours. Calculate the leakage rate by dividing the volume change by elapsed time.
When to Measure — Dynamic vs Static
Here is a trick that saves time and prevents false diagnoses: always measure under both conditions and compare.
Seals leak more during dynamic operation than at standstill. During rotation, the fluid film develops between faces and increases the gap. At standstill with no pressure, faces contact tightly with virtually zero leakage. A static-only measurement will understate the actual operating leakage rate by a wide margin.
But previously-run seals throw a curveball. Worn faces retain non-flat wear patterns that actually leak more at standstill than during operation — the opposite of what you would expect.
Measure both. Record both. If standstill leakage exceeds dynamic leakage, the faces are worn and the seal’s remaining life is limited regardless of the absolute rate.
Acceptance Criteria — Know Your Thresholds
The threshold that applies to your seal depends on your industry, location, and the fluid you are sealing.
| Standard / Regulation | Acceptance Limit | Applies To |
|---|---|---|
| API 682 | 5.6 g/h per seal face pair | Refinery and petrochemical pumps |
| California (BAAQMD) | No visible leakage; 3 drops/min triggers mandatory shutdown | VOC and hazardous fluid service in California |
| EPA Method 21 | 1,000 ppm VOC at seal boundary | Fugitive emission monitoring (~5.6 g/hr mass equivalent) |
API 682 and EPA Method 21 converge at roughly 5.6 g/hr through completely different measurement approaches. California’s standard is far stricter — it leaves zero room for the normal weepage every mechanical seal produces by design.
Seal configuration matters too. Internally pressurized seals leak roughly an order of magnitude more than externally pressurized variants under identical conditions. If your threshold assumes an external seal and you are running an internal configuration, you may be flagging normal operation as failure.
Here is why factory tests do not settle the question. The API 682 integrity test uses an air pressure decay method: 0.14 bar maximum drop over five minutes from a 28-liter reservoir at 1.7 barg. Paul Lesnau at Swagelok Northern California demonstrated that a seal with a 0.008-inch hole passed this test while being capable of leaking 270 to 1,300 times the 5.6 g/h liquid limit. The test protocol does not change based on seal arrangement, size, or face technology. Your field measurement is the only reliable data point once a seal is installed.
Replace or Monitor — The Decision Framework
Once you have a measured leakage rate under operating conditions:
- Below 50% of threshold — monitor quarterly, record the rate, trend it over time
- 50-80% of threshold — increase to monthly monitoring, investigate root cause (misalignment, incorrect flush pressure, face damage)
- 80-100% of threshold — plan replacement at next scheduled shutdown, order parts now
- Exceeds threshold — replace at earliest safe opportunity; follow your facility’s emergency procedure for hazardous fluids
- Rate increasing >20% between measurements — the seal is deteriorating regardless of absolute rate; schedule replacement
If you skip this step, you will be back in a week — either replacing a seal that did not need it, or dealing with an unplanned failure because you ignored one that did.
Next Steps
The 20 minutes you spend with a graduated cylinder and a timer will save you from pulling good seals and from ignoring bad ones. Record every measurement with the date, pump ID, operating conditions, and fluid temperature — three data points over three months tell you more about seal health than any single inspection.
Stop asking “is this seal leaking?” and start asking “what is the rate, and is it trending up?” That single change turns reactive seal maintenance into predictive maintenance. When measurement confirms the seal does need replacement, matching the correct seal type and materials to your operating conditions prevents the next seal from ending up on the same work order six months later.
