Blog / Downtime
Downtime · Guide

Tool Change Downtime: Measure It and Cut It

OL By OEE Lab Editorial |Updated 2 November 2026 |6 min read
Short answer

Tool change downtime is the time a machine tool stops cutting because a worn or broken tool is being replaced, measured from the last good part to the next good part. Measure it as changes per shift times minutes per change, then cut the walking, offset and first-part steps, because the swap itself is often the short part.

Three different things get called tool change

Most confusion comes from mixing these up. Each one lands in a different place in your numbers, so sort them before you measure anything.

KindWhat happensWhere it goes
Automatic change in the cycleThe turret or changer swaps tools under the programCycle time. See ideal cycle time.
Worn or broken tool replacedA person stops the machine to fit a new insert or toolStop, availability loss. This article.
Tooling for a new jobTools are swapped because the part changedChangeover, availability loss. See counting changeovers.
Only the middle row is the subject of this page.

A changer that faults or crashes is a breakdown, so log it apart from routine changes. If it raises alarms, try your control's code list, such as Haas alarm codes.

When the clock starts and ends

Start when the machine finishes its last good part before the change. End when the next good part comes off, and that includes any offset work and first-part checks.

Ending at cycle start is the common shortcut, and it hides most of the loss. A part scrapped while dialing in the new tool also shows up in quality, so note it in the same log line.

One honest variation: a very quick change that falls below your stop threshold lands in performance, since it is not logged as a stop. Either way it stays in OEE, so keep the threshold fixed.

How 6.5 minutes becomes 53 minutes

Example: a lathe runs one job with a 2-minute cycle time and three tools. Tool life comes from the setter's experience: roughing lasts 40 parts, finishing 90 and drilling 200.

Tool change loss = changes x minutes per change
Over 200 parts: 200/40 + 200/90 + 200/200 = 8.2 changes
8.2 x 6.5 min = 53.3 min per 400 min of cutting
Hypothetical job. Every change is a separate stop, which is what makes the total so high.

The planned run of 200 parts takes 400 minutes, and tool changes add 53.3 more. That comes to 11.8% of the total 453.3 minutes.

Timing one change

Film or stopwatch three changes on different shifts, then break each one into steps. You will usually be surprised by which steps are long.

Walk and fetch insert2.0 min
Swap insert1.0 min
Set offset1.5 min
First part check2.0 min
Hypothetical tool change, 6.5 minutes end to end. The actual swap is the shortest step.

The swap is 1 minute of the 6.5. Everything around it is the opportunity, and that is the same lesson as in how to measure changeover time.

How to cut it

Remove the walking

Keep inserts, keys and a torque driver in a marked kit at the machine. Refill it at a set time, so nobody goes to the crib mid-job.

Change tools together

Separate stops are the expensive part of the example. In the example, change finishing at every second roughing change (80 parts), which uses a little less of its life but merges stops.

That takes the job from 8.2 stops to 5, with a little more time at each. A second tool at the same stop costs the swap and the offset, 2.5 minutes, with no extra walk or first-part wait.

Merged: 5 stops x 6.5 min = 32.5 min
Extra tools: 3.5 x 2.5 min = 8.8 min
Total 41.3 min, against 53.3 min with separate stops
Hypothetical job, same 200 parts. Merging stops saves 12 minutes before any step gets faster.

Set offsets before the stop

A presetter lets you measure a tool off the machine and enter the value at the change. Where the control supports it, tool probing can do the same on the machine, so check with your builder.

Use sister tools

Many CNC controls can move to a second identical tool when the first reaches its life count. The stop disappears from that point in the run, and you replace the worn tool at a natural break.

Change on a count, not on a surprise

A worn tool found by a bad part costs a stop plus scrap. Change on a part count set from your own data, then adjust the count as you learn.

To set the first count, record the parts made at the last 20 changes of that tool, then start just under the low end of the usual range. Example: if most changes land between 42 and 52 parts, start at 40.

Be careful, since a count set too low wastes tool life, and one set too high brings broken tools. Record the reason for each change, so you can see which is happening.

What the method's author and others report

The thinking is the same as SMED: separate what can be done before the stop from what must be done during it. Shingo reported setups falling to about a fortieth of their original time in A Revolution in Manufacturing (1985), which is the author describing his own cases.

Published single-plant studies in Portugal report much smaller cuts, of 46% (Brito et al., 2017) and 43% (Sousa et al., 2018), on changeovers rather than on tool wear changes. See SMED changeover reduction for the method, and treat those figures as examples, not a forecast.

Measuring it on the floor

  • Paper first. Add "tool change" as a reason code, with the tool number as a note.
  • Use the machine's stop signal. Spindle or cycle-state data gives you the time, as explained in production monitoring for CNC.
  • Track two numbers per tool. Parts per change and minutes per change. Multiply them out by job, since a long job with frequent changes matters most.
  • Compare setters. A big gap between people shows a method worth copying, so ask how the fastest does it.

To put a cost on the time, the changeover cost calculator works for tool changes too. Enter your current and target minutes, changes per day, output rate and margin per piece.

Ways the loss gets hidden or moved

  • Hiding it in cycle time. If you add worn-tool stops to the ideal cycle, the loss vanishes from OEE and nobody chases it.
  • Slowing the feed to save the tool. Fewer stops can look good, while the longer cycle comes back as speed loss. Compare parts per shift before and after.
  • Stretching tool life to save stops. Quality falls later, and the cost moves to scrap and first pass yield.
  • Logging changes as "setup". That mixes wear with job changes, which need different fixes.
  • Skipping the first-part check to save time. A shorter stop is no gain if the next 20 parts are out of size.
See it live in Fabrico

Counting tool changes starts with a stop signal from each machine, including older ones. Fabrico takes stops, speed and counts live from PLCs or IoT sensors, and operators tag tool change as the cause at the machine. See how Fabrico connects to machines.

What people ask next

Is tool change time part of cycle time or downtime?

An automatic tool change under the program is part of cycle time. A person replacing a worn or broken tool is downtime, so it is an availability loss. OEE for CNC machines covers the rest of the setup.

How long should a tool change take?

We found no neutral benchmark, as it depends on the machine and part. Time your own changes, then compare setters and shifts on the same job. Your fastest repeatable time is the first target.

Is a tool change a changeover?

Only when the part changes. Replacing a worn tool on the same job is a stop, and changing tooling for a new job is a changeover. How to count changeovers in OEE has the rule.

How can I reduce tool change time on a CNC?

Keep a kit at the machine, merge changes where tool life allows, set offsets before the stop, and use sister tools if your control supports them. Time each step first, as the walking and checks are often longer than the swap.

How do I calculate how much tool changes cost?

Multiply changes per week by minutes per change, then by your lost margin per minute. Add any scrap from the first parts. The downtime cost calculator does the math.

See OEE per line and shift on your own machines

Fabrico is a light MES for OEE. It takes stops, speed and counts 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.

Book a Fabrico demo

← More OEE Lab guides · OEE for CNC machines →

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