Speed losses and SMED: recover the pace after every changeover

The speed loss: OEE’s costliest blind spot

When people talk about production losses, they think first of stops: the machine runs or it does not. But between the two lies a grey zone that few plants monitor: the machine runs, but below its nominal pace. No stop is declared, no alarm goes off, and yet capacity evaporates minute after minute. That is the speed loss, and it is one of OEE’s costliest blind spots.

Performance, the second component of OEE after availability, measures exactly this gap between the real pace and the theoretical pace. Yet the manual log does not know that real pace. The operator knows the line is running; they do not know it is running well below its target speed for a long while after each changeover. This silent under-pace leaves no trace on the shift sheets, and so it never enters the performance discussion at all.

Why pace does not come back all at once

After a stop or a changeover, a line does not instantly recover its full speed. There is a ramp-up phase: fine adjustments, first parts rejected, process stabilisation, gradual rise in throughput. During this phase, sometimes a long one, the machine produces, but well below its potential. These minutes below target speed add up with every restart.

The phenomenon matters all the more when runs are short and frequent. A high-mix workshop that changes over several times a day multiplies the ramp-up phases. The capacity lost to under-speed can then exceed the capacity lost to outright stops, without any classic indicator flagging it. The line looks busy, the operators are working, and yet the output stays well below what the equipment could actually deliver.

The direct link with changeovers

Every changeover combines two losses: the downtime of the changeover itself, and the under-pace phase that follows the restart. The first is sometimes measured, the second almost never. Yet both feed the same problem: the capacity lost around changeovers. As long as you only see the downtime, you optimise half the problem.

This is exactly where the SMED approach comes in. SMED, for Single-Minute Exchange of Die, aims to shorten changeover duration by separating the operations you can prepare while the machine runs from those that require it stopped. But a SMED run without real measurement stays a theoretical exercise: you optimise the motion without knowing what you really recover, and the under-pace phase that follows the restart stays entirely outside the picture.

Measuring to objectify SMED

Real-time measurement turns SMED from a promising method into a quantified action plan. The sensor times the true changeover duration, the exact moment of restart, and above all the pace actually delivered minute by minute after the resumption. You then see how long the line stays below its target speed, and how much capacity that represents.

With this data, the SMED project changes nature. You no longer argue about estimated durations; you work on measured ones, and the conversation moves from opinion to evidence. You identify the longest changeovers, the ones whose ramp-up drags the most, and you concentrate the effort where the gain is highest. SMED thus becomes a data-driven approach, not an intuition-driven one, and every change to the procedure can be tied back to a measured number of minutes recovered.

Visualising the restart curve

One of the most telling contributions of continuous measurement is the restart curve: the representation, over time, of the real pace after a changeover. This curve shows in black and white how many minutes the line takes to reach its nominal regime, and whether it really reaches it. Many teams discover on this occasion that some lines never quite return to the target pace.

Comparing these curves from one changeover to the next, from one team to another, from one product to another, reveals practice gaps that can be considerable. One team masters its ramp-up; another endures it. These are so many best practices to identify and then spread, investing nothing more than well-directed attention. Often the single best ramp-up on the floor already shows what the whole site could achieve.

From diagnosis to concrete action

Once speed losses are measured and tied to changeovers, action becomes concrete. You standardise the restart settings, you prepare the tooling upstream, you train teams on the sequences that win back the most pace. Each improvement is then verified on the next restart curve, in real time.

This continuous improvement loop, fed by data, is what sets a lasting approach apart from a one-off flash. You do not just shorten a changeover once; you install a mechanism where every restart becomes an opportunity to measure, learn and progress.

The gain: capacity without capex

By reducing both the changeover duration and the under-pace phases, you recover capacity without touching the machines. This is an essential point: many plants consider investing in faster equipment when the capacity already exists, locked inside restarts that are too slow. Freeing it costs only visibility and method.

This gain is measured directly in OEE points and in extra parts produced with the same resources. More than 450 plants across 30+ countries already monitor their OEE to the second with TeepTrak. In workshops with frequent changeovers, it is often the most profitable lever, because it acts on a loss that nobody could see.

Why this loss stays invisible so long

If speed losses go unnoticed, it is not for lack of skill on the teams’ part, but because nothing in conventional tracking makes them appear. A machine running slowly triggers no alarm, fills no downtime box, generates no report. It simply produces less, quietly.

This invisibility carries a cost all the more insidious because it is permanent. Unlike a breakdown, which is one-off and striking, under-pace is diffuse and continuous. It does not cause a crisis; it erodes. There is no dramatic moment to react to, no breakdown to log, just a steady leak that conventional tracking is structurally unable to register. That is why only an objective, continuous measurement can bring it to light and make action possible.

An approach installed in under an hour

Putting this measurement in place does not require a heavy project. The sensor is fitted on the machine in under an hour, with no production stop, without touching the PLC or the existing IT. The first pace curves are usable within 48 hours, on old machines as well as new ones.

This simplicity of deployment is what makes the approach accessible. You can start with a representative pilot line, measure the real speed losses, run a first quantified SMED project, compare the restart curves before and after, and decide to extend once the value is proven on real numbers. A free 60-day pilot is enough to objectify the gain before any rollout.

Key points to remember

The speed loss is the share of OEE that hides between the stop and full pace: the machine runs, but below nominal, declaring nothing. It weighs above all around changeovers, and the manual log does not see it. Real-time measurement times the true restart durations and the pace delivered, turning SMED into a quantified action plan. You thus recover capacity with no capex, on a loss that was until now invisible. Hutchinson improved its OEE from 42% to 75% with the same headcount and machines, sensor installed in under an hour.

FAQ

What is a speed loss in production?
Production below the nominal pace, with no declared stop. The machine runs, but slower than its target speed, often after a changeover, and this loss appears in no manual log.

Why does the pace drop after a changeover?
Because the ramp-up is not instant: adjustments, process stabilisation and first parts rejected keep the line below its target pace for a period that is often underestimated.

How does SMED help reduce these losses?
SMED shortens changeover duration by preparing as many operations as possible while the machine runs. Coupled with real-time measurement, it targets the costliest changeovers and verifies the gain actually achieved.

Do you need new machines to recover the pace?
No. The capacity lost to under-pace already exists in the current machines. You recover it through visibility and method, with no capex, on old equipment as well as new.

How long does it take to measure these losses?
The sensor is fitted in under an hour, with no production stop, and the first pace curves are usable within 48 hours. A 60-day pilot is enough to objectify the gain.