One oil refinery increased its mechanical seal MTBF by 54% across its entire population in 32 months — and cut monthly seal maintenance costs by 17%. That result didn’t come from buying better seals. It came from measuring the right things and acting on what the data revealed.
MTBF gives reliability engineers a common language to quantify seal performance, justify improvement investments, and hold suppliers accountable. Yet most published guidance treats MTBF as a vague concept rather than a working metric. The challenge is defining what counts, collecting clean data, and benchmarking against conditions that actually match yours.
What MTBF Means for Mechanical Seals
MTBF — Mean Time Between Failures — is a population-level reliability metric, not a prediction for any individual seal. A fleet MTBF of 36 months does not mean every seal will run 36 months. It means your average interval between failures, across all tracked seals, is 36 months. Some will run seven years; others will fail in four months.
I’ve seen plants treat MTBF like a warranty period — blaming the manufacturer when a seal fails “early.” But MTBF is a statistical measure of your entire system: seal selection, installation quality, operating conditions, and maintenance practices rolled into one number.
MTBF vs MTBR
MTBR — Mean Time Between Repairs — counts any maintenance intervention, including preventive seal replacements during scheduled turnarounds. MTBF counts only unplanned failures. A plant that proactively replaces seals during shutdowns will have a lower MTBR than MTBF, because planned repairs inflate the intervention count.
Define which metric your plant uses before benchmarking. An MTBR of 18 months at a refinery with aggressive preventive replacement is not comparable to an MTBF of 18 months at a plant running seals to failure.
Why Film Stability Drives the Number
If the lubrication film between seal faces stays stable at all times, seal faces last virtually forever. Seals rarely fail from complete wear-through. They fail because transient conditions — dry running during startup, thermal shock, cavitation spikes — momentarily collapse the fluid film and damage the faces beyond acceptable leakage limits.
MTBF improvement is fundamentally about preventing transient disruptions, not managing steady-state wear.
How to Calculate Seal MTBF
Divide total operating time by the number of failures:
MTBF = Total Operating Time / Number of Failures
For a population of 50 seals tracked over 24 months with 8 failures:
MTBF = (50 x 24) / 8 = 150 months
That 150-month figure assumes all 50 seals were operational for the full period. For more accuracy, sum the actual running time of each individual seal rather than multiplying population by calendar time.
Calendar Time vs Running Hours
Calendar time (months) works for continuous-service plants — refineries, water treatment facilities, chemical plants running 24/7. Running hours is more accurate for intermittent service or standby pumps, where idle time inflates apparent reliability.
State which unit you use whenever reporting MTBF. A “36-month MTBF” on a pump running 50% of the time is really 18 months of actual operating time.
What Counts as a Failure
This is where most tracking programs fall apart. Define clearly:
- Does a planned replacement during a turnaround count? (No — that’s MTBR territory)
- Does a seal pulled for inspection and reinstalled count? (No — unless it was leaking beyond acceptable limits)
- Does a secondary seal failure on a dual arrangement count as one failure or two? (Define it once, apply consistently)
Pick your rules, document them, and never change mid-program. Changing definitions halfway through invalidates every comparison you’ve built.
MTBF Benchmarks by Service Severity
A single MTBF benchmark without service context is meaningless. An 18-month MTBF on a hot hydrocarbon pump with entrained solids is excellent. The same number on a clean water circulation pump signals a serious problem.
| Service Severity | Conditions | Baseline MTBF | Best Practice |
|---|---|---|---|
| Light duty | Clean water, ambient temp, low pressure | 48-60+ months | 80-100+ months |
| Medium duty | Clean hydrocarbons, moderate temp/pressure | 24-48 months | 48-80 months |
| Heavy duty | Corrosive/abrasive media, high temp, near vapor pressure | 12-24 months | 36-48 months |
API 682 sets a target of three years of reliable service with high probability. That’s a reasonable floor for medium-duty applications, not a ceiling. Plants with mature reliability programs consistently achieve MTBFs above 48 months, and best-in-class programs have used 100-month benchmarks since the 1990s.
A plant at 14-month MTBF in medium-duty service has systemic issues — installation quality, seal selection, flush systems. A plant already at 30 months needs targeted Pareto analysis of its remaining failure modes.
The expected mechanical seal lifespan ranges from two to ten years depending on design and conditions, but MTBF is the sharper metric because it accounts for your entire population, not a single unit’s best case.
From Raw Data to Actionable MTBF
The textbook says “calculate MTBF and benchmark it.” Plant reality is messier. I’ve worked with facilities that had three years of CMMS data and still couldn’t produce a reliable MTBF number because their failure records mixed seal failures with bearing replacements, coupling changes, and “found leaking” entries that didn’t specify the source.
Building Clean Data
Every seal failure record needs four fields: equipment tag, failure date, failure mode (dry running, face damage, elastomer degradation, etc.), and root cause. Without the failure mode, you have a count but no direction. Without the root cause, you fix symptoms but not systems.
I ran a reliability audit at a refinery managing over 600 seals. The first finding wasn’t about seals — it was about data. Half the work orders said “seal replaced” with no failure mode recorded. The severity-tiered inspection schedule (critical equipment every three months, semicritical every six) only delivered results after we standardized what got recorded at each inspection.
Identifying Bad Actors
A bad actor is any seal with more than one failure per year. In a fleet of 200 seals, your bottom 10% — roughly 20 positions — likely account for 60-70% of total failures. I’ve seen this pattern dozens of times.
Heinz Bloch’s analysis of 11,000 mechanical seal failures across 148 plant sites confirmed it: a small number of problematic installations drive most of the MTBF drag. Fixing those 20 positions moves your fleet MTBF more than upgrading materials on the other 180.
Most bad actors result from incorrect process conditions on the original data sheets. Before upgrading the seal, check the actual operating conditions against the spec sheet.
Improving MTBF Systematically
A pulp and paper manufacturer was running seals at 2-3 month MTBF — replacing them almost quarterly. After upgrading to double cartridge seals, MTBF jumped to three years. The seal design mattered, but the real fix was eliminating installation errors that component seals were vulnerable to.
Prioritize by Pareto Impact
Not all failure modes are equal. Dry running, fretting, broken faces, and heavy wear are the dominant four. In most plants, one or two of these account for the majority of failures. Run a Pareto analysis on your failure mode data before spending money on solutions.
If 40% of your failures are dry running events, investing in better face materials won’t move the needle. Fix the flush systems and startup procedures first. If fretting dominates, cartridge seal upgrades eliminate the shaft-to-seal movement that causes it.
The Three Levers
Every MTBF improvement program works three levers:
- Selection: Match the seal to actual operating conditions, not nameplate conditions. A seal specified for 80 C that regularly sees 120 C during process upsets will fail on the transient, not the steady state.
- Installation: Cartridge seals remove the single largest source of installation error — setting the seal face load. If your data shows repeat failures within the first 90 days, installation quality is your primary target.
- Operating discipline: Process upsets, dry starts, and thermal cycling destroy more seals than wear does. Operator awareness of seal-critical conditions — low tank level, blocked flush lines, rapid temperature swings — prevents the transient events that kill film stability.
Next Steps
Pull your last 24 months of seal work orders, define what counts as a failure, and calculate your fleet MTBF. Compare against the service-severity benchmarks above. If you’re below baseline for your service type, systemic issues are worth investigating.
Then sort failures by equipment tag and find the bad actors. Those 20 seals dragging your average down are where the first gains live — and where the budget justification for a formal reliability program writes itself.
