AFR turns a pile of failure records into one number you can plan around: the share of a population of assets you should expect to fail in a single year.
A 2% AFR on 500 motors means budgeting for about 10 failures this year. This guide shows the formula, a worked example, how AFR converts to MTBF, and where the metric falls short.
What Is the Annualized Failure Rate (AFR)?
The Annualized Failure Rate (AFR) is the percentage of a population of identical assets that is expected to fail over one year of operation. It converts raw failure counts and operating hours into a single yearly rate, so you can compare the reliability of different equipment, models, or vendors on even footing, regardless of how many units you run or how long they ran.
AFR answers a planning question every maintenance and reliability team eventually asks: out of the machines we operate, how many should we budget to repair or replace this year? That number drives spare-parts stocking, labor scheduling, warranty reserves, and capital-replacement forecasting. AFR is usually expressed as a percentage per year, though it can also be stated as failures per unit per year.
Because it is normalized to a full year, AFR works well for assets that do not all run the same amount of time. Ten pumps that each ran half a year and twenty that ran the full year can be rolled into one comparable rate. That is why AFR shows up in reliability engineering, hard-drive specifications, and equipment-fleet analysis alike.
The AFR Formula (With a Worked Example)
The AFR formula divides the number of failures by the total operating time of the population, expressed in unit-years, then multiplies by 100 to get a percentage:
AFR (%) = ( Number of failures ÷ Total operating time in unit-years ) × 100
where unit-years = (number of units × hours each unit ran) ÷ 8,760 (8,760 = hours in a year).
Here is a worked case. A facility runs 200 identical pumps continuously for a full calendar year, and 9 of them fail during that year:
| Step | Value |
|---|---|
| Units in the population | 200 pumps |
| Operating hours per unit | 8,760 hours (one full year) |
| Total operating time | 200 × 8,760 = 1,752,000 pump-hours = 200 pump-years |
| Failures during the year | 9 |
| AFR | (9 ÷ 200) × 100 = 4.5% per year |
A 4.5% AFR means that, if nothing changes, about 4 to 5 of every 100 pumps will fail in a given year. The unit-years denominator handles partial-year operation cleanly. If half the fleet had run only six months, each of those pumps would count as 0.5 unit-years, the denominator would shrink, and the result would still be a fair annualized rate. Data centers use the same logic for drives: failures divided by total drive-days in service, converted to drive-years.
AFR vs MTBF: Two Views of the Same Reliability
AFR and MTBF (Mean Time Between Failures) describe the same underlying reliability from opposite directions. AFR expresses it as a yearly percentage. MTBF expresses it as an average number of operating hours between failures. The two convert directly, which is why reliability specs often quote one and expect you to derive the other.
For a population with a roughly constant (random) failure rate, the relationship is:
AFR = 1 − e(−8,760 ÷ MTBF), which for high-MTBF equipment is very close to the simpler AFR ≈ 8,760 ÷ MTBF (with MTBF in hours).
Plugging in a range of MTBF values shows how the two scale together, and why a large-sounding MTBF can still mean a failure rate you need to plan around:
| MTBF (hours) | Approx. AFR (8,760 ÷ MTBF) | What it means |
|---|---|---|
| 50,000 | ~17.5% | Roughly one failure per six unit-years |
| 100,000 | ~8.8% | Close to one failure per eleven unit-years |
| 500,000 | ~1.8% | Low rate, but still about 9 failures a year across 500 units |
| 1,200,000 | ~0.73% | The AFR Seagate publishes for a hard drive rated at 1.2 million hours MTBF1 |
MTBF is an average interval that is easy to misread as “this unit will last 1.2 million hours.” AFR restates the same data as “about 0.7% of these units fail each year,” which is a far more useful planning figure for a population. To see how MTBF, MTTR, and MTTF fit together, read our guide to MTTR, MTBF, and MTTF metrics.
How AFR Is Used in Reliability and Asset Management
AFR is used to quantify, compare, and forecast failures across any population of similar assets. Wherever you run many of the same thing, AFR turns scattered breakdown records into a rate you can act on.
Hard Drives and Electronics
The storage industry made AFR familiar. Drive manufacturers publish AFR as a headline reliability spec, and large data centers track observed AFR across tens of thousands of drives to decide which models to keep buying and when to retire a batch.2 Because every drive logs its power-on time, the drive-year denominator is easy to compute, and the resulting AFR gives a direct comparison between models.
Equipment Fleets and Facility Assets
The same math applies to pumps, motors, HVAC units, compressors, conveyors, and vehicle fleets. A maintenance team can calculate AFR per asset class or per manufacturer to answer questions a single breakdown never could. Is Brand A’s gearbox failing more often than Brand B’s? Is AFR trending up as the fleet ages? Should we standardize on one model in the next capital cycle? Rolling AFR into an asset management program ties each failure rate back to the specific equipment driving it.
AFR also feeds budgeting. Multiply AFR by the population size to get an expected failure count, then multiply that by the average cost per failure (parts, labor, and downtime) to produce an annual repair and replacement budget you can defend in a capital-planning review.
How to Reduce AFR
You lower AFR by removing the causes behind the failure count. A falling AFR year over year is one of the clearest signs that a reliability program is working.
- Preventive maintenance. Scheduled inspections, lubrication, calibration, and part replacement catch wear before it becomes a failure. A disciplined preventive maintenance program is the biggest lever on AFR for assets that wear predictably.
- Condition-based and predictive monitoring. Vibration, temperature, oil, and current sensors flag a developing fault while there is still time to plan the repair. eWorkOrders integrates with condition-monitoring partners such as AssetWatch, whose sensors can automatically generate a work order inside eWorkOrders when a reading crosses a threshold.
- Root-cause analysis. Logging each failure with a failure code and investigating its cause stops the same fault from recurring across the fleet.
- Installation and commissioning quality. Many early failures trace back to poor installation, contamination, or misapplication. Tight commissioning and operator training keep those out of the failure count.
- Vendor and model selection. Tracking AFR by manufacturer steers future purchases toward the equipment that fails least in your environment.
A CMMS makes this measurable. Every work order, failure code, and repair timestamp lives in one system, so the failure count and operating hours AFR depends on are already captured and can be trended per asset class.
Limitations of AFR
AFR is a population average over a year, so it hides when and why failures happen within that population. Keep these limits in mind before treating one AFR figure as the whole reliability story.
- It assumes a roughly constant failure rate. Real assets move through early-life, useful-life, and wear-out phases with very different failure rates. A blended AFR can mask a sharp rise as a fleet ages or a cluster of early failures on a new install.
- It is only as good as the data behind it. Undercounted failures, missing operating hours, or inconsistent failure definitions all distort the rate.
- Small populations are noisy. With only a handful of units, one or two failures swing AFR dramatically. The figure needs enough unit-years to mean something.
- It averages away severity. AFR counts a trivial failure and a catastrophic one the same way. Pair it with criticality and downtime data to understand business impact.
- Operating conditions matter. An AFR measured in a clean, climate-controlled plant will not transfer to a dusty, high-vibration, or outdoor site running the same equipment.
AFR is strongest as a comparison and forecasting tool, read alongside interval metrics like MTBF, failure-phase models, and criticality rankings.
Track Failure Rates Across Your Equipment
See how eWorkOrders helps your maintenance team schedule preventive maintenance, track inspections, and turn problems into work orders before small issues become costly downtime. See how it works in a free, personalized demo.
Frequently Asked Questions
How do you calculate Annualized Failure Rate?
Divide the number of failures by the total operating time of the population in unit-years, then multiply by 100 for a percentage. Unit-years equal the number of units times the hours each ran, divided by 8,760. Example: 9 failures across 200 pumps that each ran a full year (200 unit-years) gives (9 ÷ 200) × 100 = 4.5% AFR.
What is a good AFR?
It depends on the asset and its operating conditions, and there is no universal threshold. Lower is always better. Seagate, for example, publishes a 0.73% AFR for a hard drive rated at 1.2 million hours MTBF, while mechanical equipment in harsh duty can run much higher. The most useful benchmark is your own trend over time and comparisons between models or vendors in the same environment.
What is the difference between AFR and MTBF?
They describe the same reliability from opposite directions. AFR is the percentage of a population expected to fail in a year, while MTBF is the average number of operating hours between failures. For a constant failure rate they convert directly: AFR is approximately 8,760 divided by MTBF in hours, so an MTBF of 1,200,000 hours corresponds to an AFR of about 0.73%.
How can a CMMS help track AFR?
A CMMS captures work orders, failure codes, meter readings, and repair timestamps in one place, which supplies both numbers the AFR formula needs: the failure count and the operating time. That lets you calculate and trend AFR per asset class or manufacturer instead of reconstructing it from memory or spreadsheets.
- Seagate, “Hard disk drive reliability and MTBF / AFR,” Seagate Support Knowledge Base. seagate.com
- Backblaze, “Backblaze Drive Stats for Q2 2022” (AFR method: failures divided by drive-days, annualized). backblaze.com
- MTTR, MTBF & MTTF Metrics Explained
- Preventive Maintenance: Proactive Care
- Asset Management
- Maintenance Terms & Definitions Glossary
- Best CMMS Software Comparison
Disclaimer: Third-party figures on this page are attributed to their original publishers and were read in October 2026. Manufacturer AFR and MTBF ratings describe specific products under specified operating conditions and may not reflect results in your environment. The worked examples are illustrative. Verify figures with the original source before relying on them.
