Machine UtilizationRovaryn Digital9 min read

CNC Setup Time Reduction: A SMED Walkthrough for Job Shops

In a high-mix shop, setups eat the capacity the schedule thinks it has. This walkthrough applies SMED to CNC changeovers and shows how recovered setup hours change what you can promise.

CNC Setup Time Reduction: A SMED Walkthrough for Job Shops

Where the Hour Went

It is 6:45 on a Tuesday and the horizontal mill is idle. The job on it finished at 6:10. The next job's program is loaded, the material is on the cart, and the operator has been working ever since. He has walked to the tool crib for a boring bar, found the fixture plate on a shelf behind the 5-axis, hunted for the right torque wrench, and run the first part twice because the work offset was a few thousandths off. The schedule gave this changeover forty-five minutes. By 7:30 the machine is still not cutting good parts.

Nobody did anything wrong. The hour went into setup, a loss a high-mix shop pays several times a day on every machine, in small pieces nobody writes down.

This is the capacity the schedule thinks it has and the floor does not deliver. This walkthrough applies SMED, single-minute exchange of die, to CNC setup time reduction: how to time changeovers, split the steps into internal and external, estimate what the first pass recovers, and put those hours back on the schedule.

What a Changeover Actually Contains

Define the clock before you start measuring. A changeover runs from the last good part of the previous job to the first good part of the next one. It is not "wheels-off to wheels-on," and it is not the time the operator spent at the machine. It includes everything between: removing the old fixture and tools, fetching the new ones, loading the program, setting offsets, running the first piece, inspecting it, and adjusting until it is good.

That last stretch is where shops lose the most honest accounting. The proving run, the first-article inspection and the offset adjustments rarely appear on a routing, yet they are part of the machine standing without producing sellable parts.

SMED divides every activity in that window into two kinds:

  • Internal steps can only be done while the machine is stopped: unclamping the old fixture, mounting the new one, touching off a tool in the spindle.
  • External steps can be done while the previous job is still running: gathering tools, staging the fixture, printing the setup sheet, pre-setting tools off the machine, loading the program into the control's memory.

The method is simple to state. Move every step you can from internal to external, then shorten whatever internal work remains. Everything in this article is a way of doing that for a CNC machine.

Step One: Time Every Changeover Before You Change Anything

Any CNC setup time reduction effort starts with a clipboard, a phone camera or a spreadsheet, not with solutions. Record real changeovers on real machines.

For each one, capture:

  1. The machine, the outgoing job and the incoming job.
  2. The clock times for last good part and first good part.
  3. Each step in between, with a start and stop time. Video is the easiest way: a ten-minute clip played back at the desk lets you time steps the operator would never remember.
  4. Waiting. Time spent looking for a tool, waiting on a forklift or waiting for the inspector is a step like any other. It goes on the sheet.

Record at least a few changeovers on each machine you plan to improve, and include different job families. A changeover between two jobs on the same fixture family is a different animal from a full fixture and tooling swap, and averaging them hides both. If you want a structured sheet, the CNC Setup & Changeover Time Study Workbook is built for this: time every CNC changeover from last-good-part to first-good-part, then break each setup into timed steps and tag them internal or external.

The goal of this stage is a baseline. A changeover improvement you cannot measure against a baseline is an opinion.

Internal vs External Setup: Sorting the Steps

With a timed list of steps, tag each one internal or external, and do it with a skeptical eye. The habit on most floors is that anything an operator does at the machine is assumed to require the machine to be stopped.

The data says that assumption is usually wrong. In most unoptimized changeovers, 30–50% of the activities classified as "internal" (machine must be stopped) are actually external and can be pre-staged (Oxmaint, 2025: https://oxmaint.com/industries/manufacturing-plant/smed-quick-changeover-techniques-reduce-setup-time).

Run each internal step through two questions:

  • Does the machine need to be stopped for this step, or is it just where the operator happens to be standing? Fetching a gauge is not a spindle-stopped activity.
  • Could this step be done before the previous job ends? Tool pre-setting, program verification and fixture staging usually can.

Mark the steps that survive as true internal work. These are the ones you will try to shorten, not eliminate.

If the machine does not have to be stopped for a step, that step does not belong in the changeover.

CNC Setup Time Reduction: The First-Pass Moves

Once the steps are sorted, the first-pass actions on a CNC machine tend to fall into a handful of families. None needs capital spending, which is why SMED is the usual first move in a shop that cannot justify new equipment.

Stage everything the next job needs before the current job ends. Build a kit for each job: fixture, tools, gauges, the setup sheet and the material. Put it on a cart at the machine. The operator's walk to the crib, the shelf and the toolbox is an external activity that should happen while the spindle is cutting.

Pre-set tools off the machine. Tool length and diameter measured at a presetter, or at a bench, means the machine does not stop for touch-off on every new tool. Tools that repeat across jobs can stay assembled and tagged.

Load and verify the program early. Getting the program into the control and checking it against the setup sheet is an external step while the previous job is still running. Finding a missing or wrong revision at 6:45 is an internal step with a long tail.

Standardize how work is held. Common base plates, repeatable locating features, a consistent datum position and quick-release clamping turn fixture changes from a fitting exercise into a swap. Where a job family allows it, design fixtures to locate the same way on the machine every time.

Write the setup sheet once, correctly. Orientation, tool list, offsets, clamping notes and the first-article check belong on one page that travels with the kit.

How much can this recover? A structured SMED implementation typically cuts changeover time 30–50% on the first pass with no capital investment, by converting internal setup steps to external ones, and mature programs can reach 50–90% (iFactory; Oxmaint, 2025: https://ifactoryapp.com/blog/changeover-time-reduction-smed-ai-analytics). A peer-reviewed SMED case study on a turning line reduced setup times by more than 45% (Gazi University Journal of Science, 2022: https://dergipark.org.tr/en/download/article-file/1099297). Those are ranges drawn from published work, not promises about your machines. Your baseline from the time study decides where in the range you land.

What the First Pass Is Worth: A Worked Illustration

Round numbers make the mechanism visible. This is an illustration, not a measured result from any shop.

Suppose one machine runs two shifts, 16 hours a day, and averages 6 changeovers a day at 60 minutes each.

  • Setup today: 6 × 60 = 360 minutes = 6 hours a day.
  • First pass at the low end of the 30–50% range: 60 × 0.30 = 18 minutes saved per changeover, so a 42-minute changeover. 6 × 18 = 108 minutes = 1.8 hours a day recovered.
  • First pass at the high end: 60 × 0.50 = 30 minutes saved per changeover, so a 30-minute changeover. 6 × 30 = 180 minutes = 3.0 hours a day recovered.

Across a five-day week, that is 9 to 15 hours on one machine in this illustration. Multiply by the number of machines that change over this often and the hours become a real planning quantity, enough to absorb a rush job or pull a late order back inside its date.

Replace the inputs with yours. The method is the same: baseline, percentage, count, hours.

Put the Recovered Hours on the Schedule

Recovered setup time only matters if the schedule knows about it. A faster changeover that is still scheduled at the old forty-five or sixty minutes just creates idle gaps nobody uses.

Three things connect SMED to scheduling:

Update the standard. When a changeover consistently takes less time, change the setup allowance in the routing or on the board. Capacity math runs on the durations you enter, so a schedule built on stale setup times is built on stale capacity.

Treat setup as a scheduling lever, not a fixed cost. In a high-mix shop the order of jobs on a machine changes how much setup the machine absorbs. Sequencing similar jobs together, same fixture family, same material, overlapping tools, shortens the changeovers between them. This is where mixed-portfolio scheduling and setup reduction meet: one reduces the cost of each changeover, the other reduces how many expensive ones you schedule.

Measure it where it shows up. Setup is one of the dominant losses in machine shops. Typical OEE for a machine shop or metal fabrication operation runs 45–65%, with a sector world-class figure of about 78%, and setup and idling are the dominant losses (OEE-Benchmark.org, 2026: https://oee-benchmark.org/en/oee-benchmarks-by-industry/). If you track availability, setup time is where a SMED program should move the number. The OEE calculation guide shows how to decompose OEE into availability, performance and quality so the effect lands in the right place, and the machine utilization rate calculator turns the recovered hours into a utilization figure you can compare before and after.

From there the recovered hours feed capacity planning. The machine capacity planning guide covers how available hours become what you can promise a customer, and the OEE and utilization hub collects the related measurements in one place.

A word of caution on honesty. A SMED effort does not stay won on its own. Kits drift, presetters go unused, sheets go stale and a new part family brings new fixtures. Re-time a sample of changeovers every few months. If the number creeps back up, the problem is usually a step that quietly moved back from external to internal.

Start With One Machine

CNC setup time reduction does not need a plant-wide program. Pick the machine with the most changeovers, time ten of them, tag every step, and move the external ones off the clock. Then put the new durations on the schedule and watch what it does to the week.

If you want a ready sheet for the timing and tagging, the CNC Setup & Changeover Time Study Workbook is a template download for exactly that first step.

Ready to go beyond the guide?

Most shops are on a live Gantt board within 60 minutes of sign-up, with their existing job list imported from Excel.

Browse scheduling templates for job shops →

Get shop floor scheduling guides in your inbox

Practical articles for production managers — no spam, unsubscribe anytime.

Related articles