Equipment downtime calculations are where maintenance business cases go wrong, because there are two numbers in them and only one is defensible. The hours a machine was unavailable can be measured. The money that unavailability cost depends on whether the line was starved anyway, whether the order was late, and whether somebody worked a Saturday to catch up, and any single rate per hour is a guess that a finance director will find in about a minute. This page sets out how to calculate downtime so that the number survives scrutiny, which mostly means being disciplined about which half you claim.
Measure availability in hours, from the work orders
Downtime is the time between the machine stopping and the machine running again, including the wait for a decision and the wait for a part. If your work orders record start and finish honestly, you already have it; if they record 'morning', you do not, and that is the first thing to fix.
Price the labour, because you can prove it
On the worked example this site publishes, 84 breakdowns a month at 2.3 hours more than the planned version is 193.2 hours and $9,273.60 at a $48 loaded rate. Every input there is one you control and can evidence. This is the number to put in a business case.
State the production loss separately, and let finance value it
Hand over the hours the machine was down and let the people who know the margin, the order book and the catch-up capacity put a rate on it. A maintenance manager who invents $600 an hour loses the argument; one who supplies audited hours and asks for the rate wins it.
Watch the trend, not the absolute
Downtime hours per asset per quarter tells you whether the schedule is working. The total across a site tells you almost nothing, because the mix changes. The per-asset series is also what identifies the three machines causing most of the problem, which is usually where the money is.
Questions people ask about equipment downtime calculation
Should planned stoppages count as downtime?
Track them separately. Planned downtime is a cost you chose and can schedule around; unplanned downtime is the one that predicts trouble. Adding them together hides the useful signal.
What about slow running rather than stopped?
That is a performance loss rather than downtime, and it belongs in an OEE calculation. Mixing it in makes the downtime figure impossible to reconcile with anything.
How far back should we measure?
A year, if the records exist, because seasonal duty changes what fails. Anything shorter tells you about the last quarter's weather as much as the plant.