Each row of your log becomes one stop with a duration, a reason, a machine and a planned or unplanned label. These are the rules.
Reading the log
- Comma-, tab- or semicolon-separated text, quoted fields included. Cells copied from Excel or Google Sheets arrive tab-separated.
- The first row is a header when two or more column names are recognized (Date, Line, Start, End, Minutes, Reason and similar) and it holds no times or dates. Change any match under “Check the columns”.
- Dates: 2026-10-05, 10/5/2026, 10/5/26 or Oct 5, 2026 (slash dates month first unless the first number is over 12); impossible dates such as 2/30/2026 count as unreadable. Times: 14:30, 14:30:15, 2:30 PM, 7 pm, or date and time in one cell.
Duration of each stop
- With start and end times: duration = end − start in minutes, plus 1,440 if the end is earlier (the stop crossed midnight), so 23:50 to 00:15 is 25 minutes.
- If both cells also hold dates, the full difference counts: 2026-10-05 22:00 to 2026-10-07 01:00 is 1,620 minutes.
- Otherwise the duration column is used: minutes (45 or “45 min”) or h:mm (1:30 = 90). When a row has both times, the times win.
- Rows with no readable duration are skipped and listed; zero or negative durations are flagged and left out of every total.
Planned vs unplanned downtime
A value in your planned/unplanned column wins: Unplanned, Unscheduled, U, No, N or 0 is unplanned; Planned, Scheduled, P, Yes, Y or 1 is planned. If the column is headed just “Unplanned” or “Unscheduled” (for example “Unplanned?”), Yes, Y or 1 means unplanned and No, N or 0 means planned. Other rows take their reason’s tag. Tags start as a keyword guess: planned if the reason contains planned, changeover, setup, break, lunch, meal, meeting, huddle, training, sanitation, cleaning, CIP, PM, preventive, scheduled, no orders, calibration or trial, unless it also says unplanned, unscheduled, overrun or re-clean. Check every tag; our guide to planned vs unplanned downtime covers the gray areas.
The Pareto chart and the vital few
- Reasons are grouped ignoring capitals and extra spaces: “Conveyor jam” and “conveyor JAM” are one reason; “Jam” and “Jammed” are two.
- Minutes = total duration; Stops = rows; Avg stop = minutes ÷ stops; Share = minutes ÷ all minutes in the chart.
- Reasons are sorted by minutes, largest first (ties: more stops, then A–Z). Cumulative share = the running total of shares.
- Vital few = every reason through the first whose cumulative share reaches 80%. If the top reason alone reaches 80%, it’s the only one.
“Pareto of” switches between all downtime (the default), unplanned only and planned only. The chart shows the top 15 reasons on one 0–100% scale; the table lists them all.
Machine downtime analysis: availability, MTBF, trend and cost
- Per machine or line: unplanned minutes, unplanned stops, planned minutes, and mean stop length = unplanned minutes ÷ unplanned stops (a rough repair-time proxy).
- Scheduled time = hours per day × 60 × days × machines (per machine, without × machines). Days is your number, or the count of distinct dates in the log.
- Planned production time (PPT) = scheduled time − planned downtime. Availability = (PPT − unplanned) ÷ PPT. MTBF ≈ (PPT − unplanned) ÷ unplanned stops, in hours. If PPT is zero or less, or unplanned downtime is larger than PPT, you get a “check schedule” note instead of a number.
- Weekly trend: unplanned minutes per ISO week (Monday to Sunday); empty weeks show 0.
- Unplanned cost = unplanned minutes ÷ 60 × your cost per unplanned hour, shown only when you enter one. Our downtime cost calculator, one of our free tools, helps you build that number.
The data-quality check
Each line reads OK or Check: rows without a readable duration (row numbers are lines in your paste, header included); blank reasons; the “Other” or blank share of all downtime against your threshold (default 20%); stops that start before an earlier stop on the same machine ended (needs dates and start times); zero or negative durations; past-midnight stops; and rows with no readable date.
Worked example (hypothetical)
Example (hypothetical), not client data, with rounder numbers than the example log in the tool: two lines, 10 dates, 54 stops, 1,000 minutes. Changeover is tagged planned; the rest are unplanned.
Sensor fault is the first reason to reach 80%, so the vital few is three of seven reasons, 83% of downtime. “Other” is 3%, under the threshold. On “Unplanned only”, Changeover drops out and Conveyor jam (60%), Sensor fault (77.3% cumulative) and Waiting on material (86.7% cumulative) make the cut.
With 16 scheduled hours per day: scheduled time = 16 × 60 × 10 × 2 = 19,200 minutes; PPT = 19,200 − 250 = 18,950; availability = (18,950 − 750) ÷ 18,950 = 96.0%; MTBF ≈ 18,200 ÷ 46 unplanned stops = 396 minutes (6.6 h). At a hypothetical $200 per unplanned hour, cost = 750 ÷ 60 × $200 = $2,500.
Assumptions and limits
- Hours (“1.5 h”) and comma decimals (“1,5”) are misread. Use minutes or h:mm.
- An ambiguous date such as 3/4/2026 is read as March 4.
- A stop over 24 hours needs dates in both time cells, or a duration.
- Availability assumes every machine was scheduled on every counted day; a blank days field counts only dates with a recorded stop.
- Overall MTBF pools all machines.