What is the difference between failure rate and MTBF?

They are reciprocals. Failure rate (lambda) is failures per operating hour; MTBF is mean operating hours between failures. In the example, lambda = 4 / 14,000 = 0.000286 failures per hour, so MTBF = 1 / 0.000286 = 3,500 hours. Use lambda when building reliability or probability models and MTBF when communicating expected uptime to operations, since hours between failures is more intuitive on the floor.

How do I calculate the failure rate of a machine?

Count functional failures in a defined window, then divide by the machine's actual operating hours in that same window. For a motor with 4 failures over 14,000 operating hours, failure rate = 4 / 14,000 = 0.000286 failures per hour. Pull the failure count from your CMMS work orders and the run hours from runtime meters or SCADA, and exclude planned stops and minor adjustments from the count.

What is a good failure rate for industrial equipment?

It varies by asset, but express it as failures per 10,000 operating hours for comparison. The example motor sits at 2.86 per 10,000 hours, which is only fair; well-maintained industrial motors and pumps often run below 1 per 10,000 hours. Rather than chase an absolute benchmark, compare each unit against its own history and sister assets, and investigate any unit in the worst decile of your fleet.

Why should I use operating hours instead of calendar time?

Calendar time counts idle hours where the asset cannot fail, which understates true failure rate. A machine running 4,000 hours a year but sitting idle the rest looks reliable on an 8,760-hour calendar basis yet may be failing often per running hour. Always use actual operating hours so lambda reflects real stress and lets you fairly compare assets with different duty cycles.

How do I convert failures per hour into a per-10,000-hours number?

Multiply the per-hour rate by 10,000. From the example, 0.000286 failures per hour times 10,000 equals 2.86 failures per 10,000 operating hours. This scaling avoids tiny decimals that are hard to read and compare. To go the other way, divide the per-10,000 figure by 10,000. You can similarly scale to per-million-hours (FIT) for electronic components by multiplying by 1,000,000.

Does the failure rate formula assume a constant failure rate?

The simple count-over-hours formula gives an average lambda and implicitly assumes it is constant, which matches the flat middle of the bathtub curve. Early life sees higher infant-mortality failures and end of life sees wear-out, where lambda climbs. If failures cluster near the beginning or end of the window, a single averaged rate hides that trend, so plot failures over time or fit a Weibull model before trusting one lambda.

Maintenance and Reliability

Failure Rate Formula

Failure rate expresses how often a machine or component fails per unit of operating time. Use it to compare asset reliability, build reliability models, and calculate the probability that equipment survives a given period.

Formula

Failure Rate = Number of Failures / Total Operating Hours

Variables

Number of Failures: Count of functional failures in the measurement period
Total Operating Hours: Sum of actual operating hours in the same period

Understanding the Failure Rate Formula

Failure rate, often written as lambda, tells you how frequently an asset stops performing its function per hour it actually runs. Dividing the number of failures by total operating hours gives failures per hour, which is the backbone of MTBF (its reciprocal) and every reliability survival model. On the shop floor it lets you rank a fleet of pumps or motors by risk, spot the bad actors eating your maintenance budget, and decide where preventive effort actually pays back.

The inputs come from two systems that rarely agree: your CMMS work order history for the failure count, and your machine runtime or SCADA logs for operating hours. Count only functional failures, not planned stops or minor adjustments, or lambda balloons. In the example, 4 failures over 14,000 hours gives 0.000286 failures per hour. Because that number is awkward, most teams scale it to 2.86 failures per 10,000 hours or express it in MTBF terms as 3,500 hours between failures.

Interpretation depends on asset class, but lower is always better and the trend matters more than the absolute value. A rising lambda over successive quarters signals wear-out and pulls forward your rebuild decision. For a critical motor, 2.86 failures per 10,000 hours is mediocre; well-maintained industrial motors often run below 1 per 10,000. Compare against the same asset's history and against sister units, then target the top decile of lambda for root cause analysis before chasing the whole fleet.

Worked Example

A motor had 4 failures over 14,000 operating hours.

Failure rate = 4 / 14,000 = 0.000286 failures per hour
Or equivalently: 2.86 failures per 10,000 operating hours

Result: 0.000286 failures per hour

Common Mistake

Using calendar time instead of operating time. A machine that runs 4,000 hours per year but is idle for the rest has a much higher failure rate per operating hour than the same machine measured on calendar time. Always use actual operating hours for reliability calculations.

Frequently Asked Questions

What is the difference between failure rate and MTBF?: They are reciprocals. Failure rate (lambda) is failures per operating hour; MTBF is mean operating hours between failures. In the example, lambda = 4 / 14,000 = 0.000286 failures per hour, so MTBF = 1 / 0.000286 = 3,500 hours. Use lambda when building reliability or probability models and MTBF when communicating expected uptime to operations, since hours between failures is more intuitive on the floor.
How do I calculate the failure rate of a machine?: Count functional failures in a defined window, then divide by the machine's actual operating hours in that same window. For a motor with 4 failures over 14,000 operating hours, failure rate = 4 / 14,000 = 0.000286 failures per hour. Pull the failure count from your CMMS work orders and the run hours from runtime meters or SCADA, and exclude planned stops and minor adjustments from the count.
What is a good failure rate for industrial equipment?: It varies by asset, but express it as failures per 10,000 operating hours for comparison. The example motor sits at 2.86 per 10,000 hours, which is only fair; well-maintained industrial motors and pumps often run below 1 per 10,000 hours. Rather than chase an absolute benchmark, compare each unit against its own history and sister assets, and investigate any unit in the worst decile of your fleet.
Why should I use operating hours instead of calendar time?: Calendar time counts idle hours where the asset cannot fail, which understates true failure rate. A machine running 4,000 hours a year but sitting idle the rest looks reliable on an 8,760-hour calendar basis yet may be failing often per running hour. Always use actual operating hours so lambda reflects real stress and lets you fairly compare assets with different duty cycles.
How do I convert failures per hour into a per-10,000-hours number?: Multiply the per-hour rate by 10,000. From the example, 0.000286 failures per hour times 10,000 equals 2.86 failures per 10,000 operating hours. This scaling avoids tiny decimals that are hard to read and compare. To go the other way, divide the per-10,000 figure by 10,000. You can similarly scale to per-million-hours (FIT) for electronic components by multiplying by 1,000,000.
Does the failure rate formula assume a constant failure rate?: The simple count-over-hours formula gives an average lambda and implicitly assumes it is constant, which matches the flat middle of the bathtub curve. Early life sees higher infant-mortality failures and end of life sees wear-out, where lambda climbs. If failures cluster near the beginning or end of the window, a single averaged rate hides that trend, so plot failures over time or fit a Weibull model before trusting one lambda.