Blog / Losses
Losses · Guide

Startup Losses in OEE: Definition, Examples, Fixes

OL By OEE Lab Editorial |Updated 14 November 2026 |7 min read
Short answer

Startup loss is everything lost between the moment a line restarts and the moment it runs at full rate with good product. That means warm-up time, slow running and scrap. Add the three together for each start, then multiply by how many starts you have.

Startup loss calculator

Minutes and units lost per start and per period, in OEE points, split by availability, performance and quality.

Open the calculator

What are startup losses?

A line does not go from stopped to perfect in one second. After a restart it warms up, settles and gets adjusted.

During that period it makes less, slower and worse product. That is the startup loss.

In Nakajima's six big losses it is called reduced yield: the loss from machine startup to stable production. See the six big losses for the full list.

Startups happen after a changeover, a long stop, a shutdown, a weekend or a material change. The fewer the starts, the smaller the loss.

Where startup loss lands in OEE

OEE has no box called "startup". The loss is split across the three factors.

What happensOEE factorLoss type
Machine is stopped while it warms up or is adjustedAvailabilityStop time
Machine runs slower than ideal until it settlesPerformanceSpeed loss
First units are scrapped or reworkedQualityWaste

This split is why startup loss is easy to miss. Each part is small, and each part is filed in a different place.

To see the real cost, tag the time as "startup" in your log, then add the three parts together. For the speed part, see speed losses.

Worked example

Example: a line has an ideal rate of 50 units per minute. After a restart it is stopped for 15 minutes of warm-up and adjustment.

It then runs for 25 minutes while it settles. In those 25 minutes it makes 750 units, and 200 of them are scrapped.

Part of the startupCalculationLost
Warm-up stoplogged stop15 min
Slow running25 - 750 / 50 = 25 - 1510 min
Scrap200 / 504 min
Total per startup29 min

Check it another way. The startup lasted 15 + 25 = 40 minutes.

The 550 good units equal 550 / 50 = 11 minutes of perfect running. 40 - 11 = 29 minutes lost.

Warm-up stop15 min
Slow running10 min
Scrap4 min
Good output equivalent11 min
Time lost in one startup of 40 minutes: 15 minutes of stop, 10 minutes of slow running, 4 minutes of scrap, and 11 minutes of good output.

Now scale it. Example: one startup a day, five days a week, on a plan of 2,400 scheduled minutes a week.

5 x 29 = 145 minutes. 145 / 2,400 = 6.0% of scheduled time, lost only to startups.

If the same line started three times a day, the loss would be three times bigger. Number of starts matters as much as the size of each one.

Startup loss calculator

Enter one typical startup and how often it happens. The defaults are the worked example above. Updates as you type.

Line speed
/min
Units per minute at full speed.
One startup
min
Stopped to warm up or adjust.
min
Running, until full rate and good product.
In the settling run, good and bad.
Of those, scrapped or reworked.
How often optional
Per period, such as a week.
min
In the same period (planned production time).
Defaults: one startup a day, five days, 2,400 scheduled minutes
Startup loss per period
-min
Loss per start x starts
0%Scheduled time
Loss per start
-
OEE cost
-
Speed loss / start
-
Scrap / start
-
Startup yield
-
Settling speed
-
Units lost / start
-
Units lost / period
-

Where it lands in OEE

OEE has no box called "startup". The loss splits across the three factors, so tag the window "startup" in your log and add the parts back together.

One startup, minute by minute

The working

Pick one rule for where the window ends, for example full rate and good quality held for 10 minutes, and use it for every start.

How to measure startup loss

  1. 1Define the startup window. It starts at the restart. It ends when the line has held full rate and good quality for a set time, for example 10 minutes. Write the rule down.
  2. 2Tag the phase. Mark every stop, count and scrap record in that window as "startup". Hansen suggests recording the process phase (startup, steady state, transition) for each event.
  3. 3Add the three parts. Stop minutes, plus slow-running minutes, plus scrap minutes. Use the example table above.
  4. 4Count starts and trend. Track minutes per start and starts per week. Both can move on their own.
Four steps to put a number on startup loss

The phase tag comes from the event data list in Overall Equipment Effectiveness (R. C. Hansen, 2001). It lets you compare startups with steady running on the same line.

Keep changeover and startup apart. Changeover is the planned stop to switch product.

Startup is what happens after it. See how to measure changeover time.

Why startups lose so much

  • Process settings are not at their set points yet (temperature, pressure, speed).
  • Tools or materials are cold, wet or not yet in position.
  • The first trial pieces are checked, adjusted and thrown away.
  • Different crews start the line in different ways.
  • Waiting for people, material or approvals right at the restart.

Which startups to attack first

Not all starts are equal. Group them by type, then multiply minutes per start by starts per week.

Example: the same line over one week.

Start typeMin per startStarts a weekMin a week
After an unplanned stop1214168
After a product changeover295145
After the weekend90190
Total20403

The weekend start is the biggest single event, at 90 minutes. But the 14 small restarts after stops cost more in total, 168 minutes.

That is why you count starts and not only minutes. A Pareto of startup minutes by start type points to the right first fix. See how to build a downtime Pareto chart.

How to cut startup losses

  • Start less often. Join small orders, and plan longer runs where the schedule allows.
  • Write down the set points. A "production ready" board of correct settings per product removes guessing. Overall Equipment Effectiveness (R. C. Hansen, 2001) recommends this in the quick changeover method.
  • Do preparation offline. Preheat tools and stage material before the stop, not during it. See SMED changeover reduction.
  • Use the same startup crew. The same book lists this among its shutdown tips.
  • Confirm set points before restarting. It stops the repeat adjust-and-scrap loop.
  • Treat startup scrap as scrap. Do not hide it. It is waste you can reduce. See how to reduce scrap and rework.

Common mistakes

  • Filing startup as excluded time. Excluded time is planned in advance and is not scheduled for production. A restart after a stop inside scheduled time is a loss.
  • Mixing it into normal waste. Then nobody sees that startups are the biggest source of scrap.
  • No end to the window. If the window is vague, the number changes with every person who reads it.
  • Only counting minutes per start. A short startup that happens ten times a day still costs more than a long one that happens once.

For the stop-time side, read planned vs unplanned downtime. For logging the stops themselves, use how to track downtime.

See it live in Fabrico

Startup loss hides in short stops after every restart. Fabrico takes stops from PLCs or IoT sensors, with add-on AI cameras counting output where a machine has no usable signal, and operators tag the cause at the line, so startup stops get their own cause and stand out in the loss list. See how Fabrico tracks stops and restarts.

Book a Fabrico demo

What people ask next

Is startup loss availability, performance or quality?

All three, depending on what happened. Warm-up stops hit availability, slow running hits performance and scrap hits quality. Tag the whole window "startup" so you can add the parts together.

Is warm-up time planned downtime?

It can be, if it is a normal and expected part of every start. Even so, it is still time you can shorten. See planned vs unplanned downtime for how to classify it.

How long should the startup window be?

Until the line has held full rate and good quality for a short set period you define, for example 10 minutes. Pick one rule for the line and use it for every start.

Does startup loss matter on a line that runs 24 hours?

Less, because there are fewer starts. The loss per start can still be large, so look at starts after changeovers and long stops.

How is startup loss different from changeover?

Changeover is the planned stop to switch from one product to the next. Startup is the settling period after it ends. See SMED changeover reduction for cutting the changeover itself.

Related guides: OEE on a New Line: When to Start the Clock · Adjustment Losses After a Changeover Explained · First Hour of Shift Losses: Why Output Drops

← More OEE Lab guides · Open the OEE Calculator →

OEE essentials: OEE calculator · What is OEE · Six big losses · Availability · TEEP · Cost of downtime · OEE benchmark · Free OEE tracker · Free OEE software · How to track downtime · How to implement OEE · Light MES vs full MES