Seal-related repairs account for 60-70% of all centrifugal pump maintenance work, yet fewer than 10% of those seals actually wore out from normal use. The rest failed prematurely — and preventably. Operating problems drive 30-40% of failures, installation errors another 20-30%, and pump design mismatches 10-20%. The seal face damage everyone fixates on is the symptom, not the disease.
Here are the 12 failure modes I see most often, with the diagnostic evidence and prevention steps that actually work.
1. Dry Running
A seal running without liquid film between the faces generates enough friction heat to shatter carbon components in under 30 seconds. I’ve pulled seals from pumps that ran dry during a startup that lasted less than a minute — the carbon face was cracked into three pieces.
Dry running happens when pumps start without proper venting, suction lines lose prime, or process fluid vaporizes in the seal chamber. The hard mating ring shows heat checking — fine radial cracks across the entire contact surface — and a bluish-gold heat tint, while the carbon face blisters. Hardened or melted elastomers near the faces confirm it. If you see signs of dry running like these, the diagnosis is straightforward.
Install low-flow or temperature sensors on the seal chamber and enforce startup procedures that verify liquid presence before energizing the pump.
2. Misalignment and Shaft Deflection
Fretting corrosion on the shaft sleeve — reddish-brown oxide dust matching the seal’s drive pattern — is the telltale. When shaft deflection exceeds 0.002 inches, the faces can no longer track properly, and fretting accelerates failure within months.
I worked on an ammonia pump in a cold storage facility where the seal failed in under a year. The service team found 90-degree elbows in the seal piping instead of gentle slopes, plus significant shaft misalignment. Three root causes at once — and the seal took the blame for all of them.
Look for uneven face wear (one side worn through, the other barely contacted), cocked stationary seats, and uneven carbon tracking width. Laser-align the coupling and measure total indicated runout at the seal chamber face before assuming the seal was defective.
3. Contamination and Abrasive Wear
Abrasive particles between seal faces act like grinding compound. Even fine particles in the flush can score silicon carbide badly enough to increase leakage dramatically within weeks.
Circumferential scoring on both faces — parallel grooves in the direction of rotation — identifies contamination. Fine particles produce a matte, lapped appearance; coarse particles leave grooves you can catch with a fingernail. I’ve seen clean-water seals fail from rust particles in carbon steel piping that nobody considered.
Install cyclone separators or strainers upstream of the flush inlet. A clogged filter turns a Plan 11 into a no-flush condition faster than most people realize.
4. Chemical Attack on Elastomers
EPDM O-rings swell and disintegrate in petroleum products. FKM (Viton) fails in strong alkaline environments. These are the most common chemical compatibility mismatches I encounter, and they account for a disproportionate share of “unexplained” seal failures.
Swollen, softened, or dissolved O-rings are unmistakable. The elastomer may have extruded from its groove or lost all memory. If the elastomers look wrong but the faces look fine, chemical incompatibility is your primary suspect. Assembly lubricants matter too: hydrocarbon-based greases destroy EPDM on contact.
Verify compatibility for every fluid the seal contacts — including cleaning solutions and upset conditions.
5. Thermal Damage and Heat Checking
Heat checking produces fine radial cracks on the hard face (silicon carbide or tungsten carbide) when face temperature cycles rapidly, exceeding the material’s fracture toughness.
The failure most people miss: stopping and restarting an external seal flush in high-temperature service. I’ve seen a refinery seal shatter when operators stopped the external flush on a 300-plus degree Celsius service. Cold flush hitting thermally expanded components caused catastrophic thermal shock.
Size flush systems for worst-case heat load. Never interrupt flush flow to a hot seal — even briefly. For services above 150 degrees Celsius, the best flushing plan for high-temperature service is typically API Plan 23 or Plan 32 with a dedicated cooler.
6. Improper Installation
Installation errors cause 20-30% of all premature seal failures. Wrong compression setting tops the list — over-compression generates excess heat, under-compression lets faces separate under pressure. I see technicians handle lapped faces with bare hands — fingerprint oils alone prevent proper film formation. The API 682 standard specifies installation procedures for a reason, and every shortcut costs seal life.
A seal that leaks immediately or within the first few days almost always points to installation error — typically a cocked gland, contaminated faces, or incorrect compression setting, all common installation mistakes that are fully preventable with proper procedures.
Use cartridge seals wherever possible — they arrive pre-set and pre-assembled, eliminating compression measurement errors entirely.
7. Operating Off Best Efficiency Point
Pumps operating far from BEP generate radial thrust loads the seal was never designed to absorb. At low flow, recirculation creates pressure pulsations. At high flow, radial loads deflect the shaft.
An Oceaneering investigation of 40 production pumps on a North Sea platform found off-BEP operation as one of the systemic root causes behind recurring seal failures across the entire fleet — not a single seal defect in the lot. If two or three seals fail in quick succession on the same pump, check where the pump is operating on its curve before blaming the seal.
Right-size the pump for actual system conditions and install a minimum-flow recirculation line to keep operation above 70% of BEP.
8. Bearing Failure and Vibration
Heinz Bloch’s analysis of 11,000 seal failures across 148 plant sites found 13% were directly caused by bearing problems. That 13% translates to roughly $416 million per year in seal replacements globally that had nothing to do with the seal itself.
Bearings degrade gradually, increasing shaft vibration and runout until seal faces can no longer maintain contact. If you replace a seal and it fails again within the same timeframe, the bearing is the prime suspect. Add vibration monitoring — it catches bearing degradation months before the seal fails. Before replacing any seal, check the bearings first.
9. Incorrect Seal Selection
Specifying seals for normal operating conditions while ignoring startup, shutdown, and upset conditions guarantees repeated failure. A seal rated for 150 degrees Celsius steady-state will fail when the process hits 200 degrees during a transient. Rotating seal designs have a practical speed limit around 5000 feet per minute — centrifugal force at higher speeds actually opens the faces rather than closing them.
The clue: the seal meets specs on paper but no single failure mode dominates. Thermal damage one time, face separation the next, chemical attack after that. Specify for worst-case conditions. API 682 requires documenting all operating conditions for exactly this reason — follow it.
10. Flush System Failures
The flush system keeps faces cool, clean, and lubricated. When it fails, the seal follows. I’ve seen perfectly specified seals fail within days because a flush line valve was left closed after maintenance.
The North Sea study identified barrier fluid contamination from a central refilling system as a systemic cause — every top-off introduced contaminants to dozens of seals simultaneously. If the flush system shows any anomaly — low flow, high temperature, dirty fluid — that is your root cause, not the seal.
Commission flush systems with the same rigor as the seal itself. Install flow indicators on flush lines — a simple sight glass costs almost nothing compared to a seal replacement.
11. Pressure Reversals and Transients
Seal faces designed for pressure in one direction can separate momentarily during reversals, allowing fluid to flash between the faces. Water hammer, cavitation at low NPSH, and check valve slam are the usual sources.
Intermittent leakage correlating with specific operational events is the diagnostic signature. Face evidence shows localized damage — chipping on the carbon edge or a narrow scoring band — rather than the uniform wear of normal service. Mitigate transients at the source with slow-closing valves and proper NPSH margin, or specify bidirectional seal designs.
12. The 3 Most Common Misdiagnoses
Misdiagnosis drives more repeat failures than any single failure mode.
Blaming the Face When the Problem Is Upstream
Seal faces are the apparent problem in 40-70% of pump repairs — but face damage is the symptom. Shaft deflection, cavitation, or thermal distortion caused it. I’ve seen teams replace the same seal four times in 18 months before someone checked bearing clearances and found the real cause in under an hour.
Increasing Spring Load to Stop a Leak
More spring pressure, tighter gland bolts, heavier springs — every one of these “fixes” generates additional friction heat, accelerating the thermal damage that caused the leak. If spring load matches the manufacturer’s specification and the seal still leaks, the problem lies in face condition, alignment, or operating conditions.
Treating Every Failure as Independent
A pump that eats seals every six months has a system problem, not a seal problem. Seals fail for only two fundamental reasons: lapped faces opened, or a component was damaged. Both trace to conditions the system imposed on the seal.
Before replacing any failed seal, compare three things: the new seal as received, a seal that ran successfully in the same pump, and the failed seal. The differences between the successful and failed seal reveal the deviation. This three-state comparison is the fastest path to accurate root cause identification.
Taking Action on Seal Failures
Every failure mode traces back to four categories: how the seal operates, how it was installed, whether it matches the application, and whether the support systems work. The single most valuable change any plant can make is to stop treating seal replacement as the corrective action and start treating it as the trigger for root cause investigation. A $500 seal that fails every six months costs far more than the $2,000 engineering investigation that prevents the next 10 replacements.
