PM Interval Fundamentals

Finding the Optimal PM Interval: The Total-Cost Curve Between Too Often and Too Late

By Rovaryn Digital· November 17, 2026· 9 min read

When the OEM interval and the failure log disagree

It is budget week, and a maintenance manager is staring at two documents that do not agree. The OEM manual for the line's main transfer pump says to service the bearings every 30 days. The shared spreadsheet of work history says the pump has failed twice in the past year anyway, once on a 28-day-old bearing and once on a 75-day-old one. Meanwhile the plant manager has asked whether the crew could stretch the interval to 90 days and give back some of those labor hours.

Both positions sound reasonable, and neither is a calculation. Servicing more often costs labor and parts every time. Servicing less often costs repairs and lost output when something fails between visits. The optimal preventive maintenance interval sits where the sum of those two costs is lowest, which is rarely at either extreme.

This guide shows how to price each candidate interval as PM cost plus expected failure cost, how to read the resulting total-cost curve, and how to test whether your answer survives a change in assumptions.

The two cost curves that cross at your interval

Every candidate interval produces two annual costs, and they move in opposite directions.

Annual PM cost rises as the interval shortens. It is the per-event cost of a PM (labor hours × labor rate, plus parts and consumables) multiplied by the number of PM events per year. Events per year is 365 ÷ interval in days, so halving the interval doubles the PM bill.

Annual expected failure cost rises as the interval lengthens. It is the number of failures you expect per year at that interval multiplied by the full cost of one failure. The full cost includes repair labor and parts, any collateral damage, the output or capacity lost while the asset is down, and the overtime or expedited freight used to recover.

Add the two for each interval and you get a total-cost curve. It is typically U-shaped: high on the left where you are over-maintaining, high on the right where you are waiting for failures, and lowest somewhere in between. The bottom of the U is your lowest total cost maintenance interval.

The bottom is often flat, so neighboring intervals can sit close together, and you can choose among them on operational grounds such as aligning with a shutdown. The curve can also be lopsided, as the worked example below shows.

Not every failure responds to the interval. Random failures with no wear-out pattern do not become less frequent because you inspect more often. The curve applies to failure modes where age or usage drives the failure rate. Our guide on using failure history to set PM intervals covers how to tell which modes qualify.

Finding the optimal preventive maintenance interval, step by step

The method has five steps and works for one asset, one failure mode or one asset class, provided the inputs stay consistent. Following them gives you an optimal preventive maintenance interval you can defend with your own numbers.

  1. Choose the candidate intervals. Pick four to six values that bracket your current interval: for example 30, 60, 90, 120 and 180 days. Include the OEM interval and the one you are tempted to move to.
  2. Cost one PM event. Multiply labor hours by your fully burdened labor rate and add parts and consumables. If the PM needs a planned stop, add that cost too.
  3. Cost one failure. Add repair labor, repair parts, collateral damage, and the cost of lost output or capacity for the repair duration. If you do not yet have a defensible figure for lost output, calculate it from your own line: hours down × fully loaded cost of lost output or capacity per hour. We do not publish a generic per-hour figure, because a number that is not yours will be wrong for your plant.
  4. Estimate failures per year at each interval. This is the hard input, and it must come from your own records, not from a benchmark. Use the failure history for the asset or its siblings, grouped by how long the component had run since its last PM when it failed. How to calculate MTBF explains the arithmetic for turning failure counts and operating time into a failure rate. Where history is thin, use a range rather than a single value, and run the calculation at both ends.
  5. Compute and compare. For each interval, annual PM cost = PM events per year × cost per PM event. Annual failure cost = expected failures per year × cost per failure. Add them. The interval with the smallest total is the lowest-cost candidate.

One caution on step 2. A PM that replaces a part with useful life left carries a hidden cost, because you are buying the next part sooner than you need to. The article on premature part replacement cost shows how to put a number on that and fold it into the per-event cost.

A worked example: one pump, five candidate intervals

The inputs below are round, assumed values chosen to demonstrate the method. They are not benchmarks and not measurements from any plant. Substitute your own.

Assumed inputs

  • Cost per PM event: $400 (about 4 labor hours at a burdened rate of $75 per hour, plus $100 of parts and consumables)
  • Cost per failure: $6,000 (repair labor and parts, plus lost output during the repair)
  • Expected failures per year at each interval, taken from an assumed failure history: 0.3 at 30 days, 0.5 at 60 days, 0.8 at 90 days, 1.3 at 120 days, 2.2 at 180 days
  • PM events per year rounded to whole numbers for readability: 365 ÷ 30 ≈ 12, 365 ÷ 60 ≈ 6, 365 ÷ 90 ≈ 4, 365 ÷ 120 ≈ 3, 365 ÷ 180 ≈ 2

Calculation

Interval (days) PM events/yr Annual PM cost Expected failures/yr Annual failure cost Total annual cost
30 12 $4,800 0.3 $1,800 $6,600
60 6 $2,400 0.5 $3,000 $5,400
90 4 $1,600 0.8 $4,800 $6,400
120 3 $1,200 1.3 $7,800 $9,000
180 2 $800 2.2 $13,200 $14,000

The lowest total is at 60 days, at $5,400 per year. Notice what the table shows about each of the options that started the argument. The 30-day OEM interval costs $1,200 more per year than the minimum, because the extra six PM events cost $2,400 and only avoid $1,200 of expected failures. The 90-day interval costs $1,000 more, because the PM savings of $800 are outweighed by $1,800 of additional expected failure cost.

The curve is lopsided here: moving from 60 days to 30 adds $1,200, while moving from 60 to 120 adds $3,600. In this example, waiting too long costs more than looking too often.

Total annual cost = (365 ÷ interval) × cost per PM + expected failures per year at that interval × cost per failure. Calculate it for every candidate and read off the minimum.

This is arithmetic on assumed numbers, so the 60-day answer is a property of the example and says nothing about your pump. The method is what transfers.

How sensitive is the answer to your inputs?

The expected-failure counts are estimates, so a single answer deserves a stress test before it becomes a schedule. Rerun the table with the inputs that you are least sure of changed in both directions.

Using the same example, suppose the true cost per failure is $3,000 rather than $6,000. Failure costs become $900, $1,500, $2,400, $3,900 and $6,600, so the totals become $5,700, $3,900, $4,000, $5,100 and $7,400. The minimum stays at 60 days, and 90 days is only $100 behind. In that case the 90-day interval is a legitimate choice if it lines up better with a planned outage.

A result that holds at both ends of your plausible range is robust. If the minimum jumps from one interval to another when an input moves within its uncertainty, that is useful information too: it tells you the decision hinges on that input, and that collecting better failure data is worth more than refining the arithmetic.

Where the optimum differs sharply from the OEM interval, treat that as a reason to investigate rather than a license to deviate. The OEM may be assuming harsher duty, a different environment or a warranty condition. PM interval override: when to deviate from OEM sets out how to document the reasoning so the change holds up in an audit or a warranty discussion.

Where the curve misleads, and what the calculation leaves out

A total-cost calculation is a planning aid, and its honest limits are worth stating.

It is only as good as the failure counts. If your history is short, incomplete, or mixes several failure modes, the curve will look more precise than it is. Fit it per failure mode where you can, and be skeptical of a minimum that depends on a single event.

It ignores safety and compliance. Some intervals are fixed by regulation, insurer requirements or manufacturer safety instructions, and cost minimization does not apply to them. Confirm the requirements with the relevant authority and treat them as constraints on the candidate list, not as inputs to the optimization.

It does not capture every benefit of preventive work. Preventive work can also bring benefits that are hard to price, such as steadier output and fewer emergency call-outs. Whatever general claims you read about the savings from preventive over reactive maintenance, your own total-cost table is a better basis for your schedule than a general figure, because it uses your costs and your failure history.

It does not run the program. This is a planning calculation. It does not issue work orders, dispatch technicians, track parts inventory or manage vendors, and it is not a CMMS. It produces the interval you then load into whichever system schedules your work, whether that is a planning calendar, a spreadsheet or a work-order system you already own.

Running the comparison without rebuilding it each time

The arithmetic is simple enough to build in a spreadsheet. The trouble is repetition: you will rerun it per asset class, per failure mode, and whenever a labor rate or failure count changes. Each rebuild is a chance to break a formula.

The PM Interval Cost Optimiser is an Excel workbook built for this one job. You enter your candidate intervals and your own expected failure counts, and it returns the lowest-total-cost interval. The counts are yours: the workbook does not supply failure data, and a template cannot replace the history only your plant has.

To start, pick one asset where the OEM interval and the failure log disagree and fill in the table by hand. To choose between days, hours and cycles as the interval unit, read how to set PM intervals in days, hours or cycles. Then use the same total-cost method to settle on the optimal preventive maintenance interval for the next asset, and the one after.

Get maintenance guides in your inbox

Related guides