Machine downtime metrics: the minimum to track on the shop floor

Why recorded downtime is almost always lower than reality

On the shop floor people usually remember big breakdowns that stopped a machine for half a day. But output is more often eaten by dozens of short pauses of 3, 7 or 12 minutes. By the end of a shift no one remembers them, and they drop out of the overall picture.

On a CNC lathe this looks familiar: waiting for a blank, searching for a tool, tweaking a program, cleaning chips, calling a setup technician, restarting work. Each stop seems minor. Over a shift they comfortably add up to an hour or more.

There’s a second problem: the same reason is named differently by different people. The operator writes "the machine stopped", the foreman calls it "part correction", the setup tech considers it a setup. When labels float, you can’t compare shifts, days or cells.

If there’s no precise start and end time, arguments start immediately. One insists the machine stood "about ten minutes", another says "at least half an hour." After two days nobody remembers which was a short stop and which was a real breakdown.

Most often you lose waiting for a blank or tool, restarting after a fault, small size adjustments, cleaning the work area and waiting for a foreman or setup tech. Because of this, the shop sees only loud problems and misses what cuts output every day. The plan looks realistic, people are busy all shift, but fewer parts come out than expected.

So proper downtime accounting doesn’t begin with an expensive system, it begins with discipline in recording. If you log at least the start, end and one clear reason, the picture changes very quickly. After a few shifts it becomes clear: most time is spent not on rare breakdowns but on recurring small losses.

What minimum data you need to start

At first you don’t need a sensor on every unit or a costly system. A simple shift log is enough. Five clear entries a day are better than a complex form that nobody fills out.

The minimal set is: shift start and end, cycle time, all stops with duration and reason, breakdowns, setups, and count of good parts. That’s enough to understand where time goes and why the plan doesn’t match reality.

It’s easiest to make a single table on paper or in a plain file. Six fields are sufficient:

• shift start and end;
• cycle time per part or per batch;
• each stop with start, end and reason;
• breakdowns as a separate line;
• setups as a separate line;
• count of good parts for the shift.

Shift time gives the basis for all calculations. Without it you can’t know how many hours the machine was actually available. Even an entry like "08:00-20:00" removes confusion about lunch, late starts and early stops.

Cycle time doesn’t have to be perfect at first. If a part runs steadily, measure the cycle a few times and take the working average. If it’s easier to count by batch, record batch start, batch end and quantity. You’ll still get a meaningful measure.

It’s better to log all stops, even short ones. If the machine stood for 6 minutes because a tool was being searched for, it’s not trivial. Over a shift such pauses often consume more time than one noticeable breakdown. Write the reason in simple words: no blank, waiting for setup tech, tool change, first-part check.

Don’t mix breakdowns and setups with ordinary stops. Spindle failure, axis error and fixture change are different losses. If you collapse them into one line "downtime," the log won’t explain anything later.

The count of good parts completes the picture. It shows what the downtime turned into at the output. Sometimes there are few stops, but a long setup made the shift produce 18 parts fewer than the norm.

Even a simple record is useful: shift 08:00-20:00, 92 good parts, one setup 35 minutes, sensor-related breakdown 14 minutes and five short stops totaling 27 minutes. That’s enough to know where to start investigating.

How to measure cycle time without sensors or integrations

First you need an honest cycle time. Without it any stop looks like "not long" or "nothing serious." At the start a stopwatch, a log sheet and one part the machine makes most often are enough.

Don’t try to measure every item in the plan at once. Take a part that runs in series during a normal shift and repeats many times. With that job it’s easier to see the normal rhythm and get a real cycle time, not a random number.

One measurement is useless. It’s much better to time 10–15 cycles in a row during a regular shift, not during a demonstration run. Pick one start and end point and don’t change it. For example, from pressing "Start" to the moment the machine is ready for the next cycle.

The first cycle almost always differs. The operator checks size, listens to cutting, sometimes makes an adjustment. So remove the first trial cycle from the calculation. If you recorded 12 cycles and the first took 9 minutes while the usual is about 6, the average will be skewed.

The method is simple:

• choose one serial part;
• measure 10–15 consecutive repeats;
• exclude the first cycle;
• add the remaining times and divide by their number.

This gives a working standard for that specific part and program. It doesn’t have to match to the second. You need a clear benchmark to compare with the actual shift.

Recalculate the standard if the process changes: new tool, program change, adjusted cutting mode, or new material—update the average. Even an extra 20–30 seconds per cycle in a series quickly becomes hours of loss.

On a lathe this is especially visible. Yesterday a part ran in 6 minutes 10 seconds, after a insert change it became 6 minutes 40 seconds. Formally the machine didn’t stop, but the shift output fell. This manual accounting lets you notice the issue right away without sensors or complex software.

How to record stops so they can be analyzed later

If the record reads "machine stood about 20 minutes, waiting for something," it’s almost useless. A week later no one remembers when the stop started, what caused it and who brought the machine back. You need order, not a complex system.

First rule: record the exact start time. Not "after lunch" or "around 11," but for example 10:47. This helps link the stop to a shift, batch, operator, setup or tool fault.

Second rule: close the record immediately after the machine restarts. Not at the end of the shift when memory has faded. As soon as the spindle runs again, the operator or foreman records the end time. That gives an honest duration without the usual 5–10 minute understatement.

Reasons don’t need a paragraph. Keep 5–7 short, clear options everyone understands: waiting for material, tool change, setup, breakdown, part inspection, operator absent, unknown reason. This list works better than long descriptions where everyone writes differently.

Don’t mix stop types on one line. If the machine waited for a blank, that’s not a setup. If a setup tech changed a fixture, that’s not a repair. If a drive error occurred, don’t hide it as a "technical pause." Otherwise the log quickly becomes a set of vague entries.

A separate tag for unknown reasons is useful. It’s more honest than marking everything as "normal" or "work moment." If there are many such lines, you’ll immediately see a weakness in accounting.

In practice one line per stop is enough: date, machine, start time, end time, reason and a short 3–5 word comment. For example: "14:12-14:26, CNC lathe, waiting for blank, warehouse didn’t deliver." After a few shifts this suffices for a proper review.

How to separate a breakdown from a short stop

If you mix all stops together, the numbers quickly lose meaning. A machine might stop for 3 minutes to replace an insert, or for 40 minutes because of a drive error. In the report these are just two downtimes, though causes and actions differ greatly.

It’s better to adopt one rule per shift: call a breakdown any stop of 10 minutes or more if it was caused by a machine or component fault. The threshold can be different, but adopt it once and don’t change it by situation. Otherwise one foreman will log 8 minutes as a breakdown and another won’t.

But one time threshold doesn’t solve everything. Cause is more important than duration. If a drive error occurred, a sensor failed, lubrication stopped, the spindle seized or a hydraulic unit didn’t respond, that’s a breakdown. Even if a setup tech quickly fixed it, record it separately. Repeating short failures often hide in these episodes.

Short stops belong in another category. These usually include chip clearing, tool replacement or feed-up, first-part inspection, restarting after a minor fault not related to a component failure. If the insert or tool broke, don’t rush to write "machine breakdown." Often that’s a separate problem tied to tooling or cutting mode.

A practical classification of four categories works well:

• breakdown — machine or component fault or clear CNC/drive error;
• short stop — machine is OK, process quickly restored;
• tooling failure — problem with insert, tool holder or tool setup;
• setup — planned change of part, fixture or program.

Add a short, plain-language comment to each breakdown. Entries like "axis X, drive error", "coolant pump wouldn’t start", "hydraulic clamp losing pressure" are enough. These save a lot of time when reviewing a week’s data.

If the shop has different machine types, the rule should be common to all. Then you’ll see not just total downtime but the real picture: what equipment fails, where tooling is the issue, and where the process is stopped too often over small things.

How to log setup by steps

Setups often produce debatable numbers. They’re often recorded as one pause between batches and everything is put into it: tool change, finding a program, waiting for blanks and even talking about the job.

To make the accounting useful, record several simple time points. Then you can see where 12 minutes went and where 40 went, and avoid confusion between operator work and organizational delay.

First note the last good cycle of the previous batch. That’s where the old run truly ended. If you miss it, final actions from the old order easily slip into the setup time.

Then record when the operator actually started changing fixtures, tools or the program. That’s when setup begins. If at that moment there was no drawing, blank or job instruction, don’t include that waiting time in the setup. Those reasons are better tracked separately.

Usually four time points are enough:

• last good cycle of the old batch;
• start of changing fixture, tool or program;
• first trial run or first trial part;
• first good part of the new batch.

Don’t lose the segment between the trial run and the first good part in the total. That’s where the most frequent issues hide: adjustments, measurements, retries, tool changes, minor program errors. Even a short note like "2 trial parts" or "size correction" explains a lot.

Example: old batch ended at 10:12. Tool and program change started at 10:15. Trial run at 10:31. First good part at 10:38. If the operator waited for the job from 10:12 to 10:15, the setup took 23 minutes, not 26.

You can keep this record manually on a sheet, in a table or in a shift log. In a few days it becomes clear that timing variance comes not from the machine but from preparing tools, programs or agreeing the first part.

Example of one shift on a CNC lathe

Take a typical eight-hour shift. By memory the foreman might say: "The machine worked almost all day, there was one drive error." The log usually tells a different story.

What the shift log looks like

The shift started on time. The first two hours the machine ran without stops; the cycle was stable.

Then the operator changed an insert, checked the size and made a correction. That took 18 minutes. After lunch the machine waited for a blank for 27 minutes. Near the end of the shift the drive threw an error, and another 14 minutes were spent finding the cause and restarting.

In the log this can be recorded simply:

• 08:00-10:00 — work without stops;
• 10:00-10:18 — insert change and correction;
• after lunch 27 minutes — waiting for blank;
• end of shift 14 minutes — drive error.

These lines already say more than the vague phrase "there was a breakdown."

Total losses for the shift are 59 minutes. For a 480-minute shift the machine was idle 12.3% of the time. Of those 59 minutes only 14 were due to the drive error — less than a quarter of total losses.

The biggest loss here isn’t the breakdown but waiting for the blank. Another 18 minutes were spent on tooling work. So the main improvement opportunity lies not in repairs but in blank supply and tool preparation.

That’s the point of minimal accounting: see cycle time, setup, waiting and breakdown separately. Then the right action becomes clear. In this example, a mechanic shouldn’t be the first task; it’s better to check why blanks don’t arrive on time and whether the insert and size correction can be prepared in advance.

Even one shift gives a reason to act. If waiting for blanks repeats daily, the shop loses 135 minutes a week on a single machine. By memory it looked like the problem was only the drive error.

What you’ll see after a week

Seven shifts of records quickly break the phrase "the machine barely stops." When a foreman or operator logs the time and reason for each pause, one stop that repeats more than others usually stands out. It may be waiting for a blank, searching for a tool, a small clamp fault or a long first-part check.

By memory such pauses seem rare. On a sheet or in a table they add up to hours. So within a week you’ll have not a feeling but a clear picture of where the shift time goes.

Usually you’ll spot the gap between planned and real cycle time. The process says a part should take, say, 6 minutes, but in reality it’s 7 minutes 20 seconds. The difference seems small until you multiply it across the batch. For 80 parts that’s already more than an hour and a half.

You’ll also quickly see how much time setups take. While counted as a single pause, you can argue endlessly. After a week you’ll see where minutes are lost: the operator waits for fixtures, spends long adjusting size, changes tools too often, or runs another trial after a correction.

Then it’s easier to choose what to tackle first. If the machine often waits for blanks, chuck, inserts or measuring tools, start with supply and shift preparation. If cycle time grows and the operator frequently tweaks parameters, check tooling durability and replacement procedures. If the same drive, hydraulics, sensors or clamps fail repeatedly, bring in maintenance and keep a stricter fault log. If most time is lost between the last good part of the old batch and the first good part of the new one, break down the setup by steps.

After a week you won’t have perfect statistics, but you’ll have an honest list of losses for a specific machine. That’s enough to remove the single most frequent cause and noticeably reduce downtime the next week.

Common mistakes in this kind of accounting

Even a careful log quickly loses meaning if people record events from memory and everyone names them differently. In the end the numbers look smoother than the actual shift, and loss reasons scatter across different lines.

The most common mistake is simple: the operator fills the log at the end of the shift. After a few hours a 6-minute short stop easily becomes "about 15 minutes", and two separate pauses merge into one. If records aren’t made immediately, accuracy falls fast.

Another harm is a line "other." When everything goes into it, you no longer see where time is lost: waiting for blanks, searching for tools, program corrections or minor faults. If "other" occupies a noticeable part of the shift, the log is almost useless.

A frequent error is treating the first trial part as a normal cycle. After a setup the operator checks size, makes adjustments and sometimes changes the cutting mode. That run can’t be compared to a stable series cycle or the average will be inflated.

Confusion also comes from mixing planned pauses with breakdowns. Lunch, housekeeping, waiting for a job and a breakdown are different things. If you lump them together, the log will show more problems than there really are.

There’s a quiet mistake noticed later: every week people change the names of reasons. Today they write "no blank", tomorrow "waiting for material", then "warehouse didn’t deliver." Semantically it’s the same, but in the report it’s three different causes. Comparing weeks becomes difficult.

Simple rules help:

• record events immediately, not at the end of the shift;
• keep a short list of reasons and don’t change wording unnecessarily;
• mark the first part after setup separately;
• separate planned pauses, short stops and breakdowns.

If you start with this discipline the data gets cleaner after a few shifts. Then accounting shows not "general shift fatigue" but concrete losses you can remove.

Where to start next week

Don’t try to cover the whole shop at once. To start, one CNC lathe and one part that repeats shift to shift are enough. This lets you quickly see where time is lost without drowning in mixed records.

In the first week don’t add new forms, reports or extra fields. Take the simplest log and keep it consistent for five shifts. If everyone writes in their own way, numbers quickly become guesses.

Keep the process very simple:

• choose one machine and one repeating part;
• fix the same fields for all 5 shifts;
• appoint who opens a downtime record and who closes it;
• at week’s end count the most frequent cause of losses.

Fields are better kept basic: cycle start time, cycle end time, stop start, stop end, short reason and who recorded it. That’s enough to see the difference between normal work, short stops and real faults. If you add ten reasons and codes at the start, people will begin to skip entries.

Check responsibility within the shift. Often everyone notices a stop but nobody closes the record. Then a 7-minute pause at the end of the day looks like an hour. A simple rule helps: the operator opens the record immediately and the foreman or setup tech closes it after work resumes. You can do the opposite, but keep the order consistent for the week.

After five shifts don’t try to fix everything at once. Pick one recurring cause — for example waiting for setup, searching for a tool or a prolonged first trial after setup. Fixing one frequent cause usually gives more benefit than five small changes without follow-up.

If you’re choosing a new CNC or discussing service, it’s useful to clarify what data should be collected from day one. That’s a reasonable question for a supplier. At EAST CNC this order can be discussed during selection, commissioning and service so you don’t have to redo accounting in the shop later.

If the log was filled consistently for a week, the foundation is already there. Next you need discipline and one further step, not a complex system.