Lines run below ideal speed for seven usual reasons: speed turned down to avoid trouble, machine wear, poor material, hard products, drifting settings, a wrong ideal rate, or a neighbor machine setting the pace. Test each one by comparing the rate by lot, product, crew and time since cleaning. Then run a speed trial to find the rate that gives the most good units.
A slow line rarely has one cause. It usually has two or three, stacked, and each one has been normal for so long that nobody questions it.
This page is a test plan. For the definition and formula, read speed losses, then come back here to find out why your line is slow.
Step zero: confirm the line is really running slow
A low performance number does not prove slow running. Short stops nobody logged land in the same factor, and they need a different fix.
Example: a pouch line has an ideal rate of 80 units a minute. Over one shift it has 400 minutes of runtime and makes 26,400 units.
Those units are worth 26,400 / 80 = 330 minutes at the ideal rate. So 70 minutes vanished while the line was officially running, and performance is 330 / 400 = 82.5%.
A counter on the line then shows 48 short stops of about 30 seconds each. That is 24 minutes, which leaves 46 minutes of true slow running: the line made its 26,400 units in 376 moving minutes, or about 70 units a minute.
If short stops are the bigger slice, go to how to reduce micro-stops first. The rest of this page assumes the machine is moving, just slower than it could.
Also check that your system can see speed at all
Some systems report 100% performance by default, so a speed problem never appears. In a study of 884 machines in 23 Swedish companies, 702 recorded 100% performance, and the author notes that 100% was the default value in the measurement system.
If your performance number looks perfect, divide units by runtime yourself. That takes two minutes and a counter reading.
The seven causes and how to test each one
Run these in order, because the early ones cost nothing.
| Cause | What you see | Test |
|---|---|---|
| Speed turned down on purpose | Steady rate, well below ideal, on every shift | Read the set speed on the panel. Ask who set it, when, and what went wrong above it. |
| Wear or dirt | Rate slid down over weeks or months | Compare the rate in the week before and after a clean or overhaul. |
| Material | Rate changes when the lot or supplier changes | Plot rate by lot number for a month, on one product. |
| Hard product | Some products always run slower | Work out performance by product, each against its own ideal rate. |
| Settings drift | Same product, different rate by crew or day | Compare live settings against the setup sheet at the start of a run. |
| Wrong ideal rate | Nobody has ever hit it, or someone beats it | Find where the number came from. Compare it with the nameplate and your best sustained run. |
| Paced by a neighbor | Machine waits, or is held back by the next one | Stand at the line and watch the buffers on each side of the machine. |
Cause 1: speed turned down on purpose
Ask about this first, and do not assume laziness. Often someone once had a bad night at full speed, turned the dial down, and it stayed down.
That choice may have been correct at the time. The jam it prevented may also have been fixed two years ago, so test it again instead of arguing about it.
Cause 2: wear and dirt
Worn belts, tired bearings and sticky guides make a machine unstable at high speed. Crews respond by slowing it down a little at a time, so no single day looks bad.
Look at the trend, not one shift. A slow slide over months points here, and a jump back after a clean all but proves it.
Cause 3: material and environment
Film that stretches, labels that curl and cartons with loose tolerances all force a lower rate. Heat and humidity can do the same through the material.
Sort your rate data by lot. If one supplier's lots run slower on the same product, you have evidence to take to purchasing.
Cause 4: hard products
A single shift average can hide this one. A line that does fine on one size and crawls on another shows a middling number all week.
Break the data down with OEE by product. Then treat each slow product as its own problem.
Cause 5: settings drift
When crews set up from memory, each one ends up at a different point. The rate then depends on who did the changeover.
Check rate variation between runs with cycle time variation. A wide spread on one product usually means the setup is not repeatable.
Cause 6: the ideal rate is wrong
This one is not a machine problem at all. A rate set too high creates a loss that can never be closed, and one set too low hides a real loss.
See ideal cycle time for how to set it. If a line beats 100%, suspect the rate or the count before you celebrate.
Cause 7: a neighbor sets the pace
A machine can look slow when it is really waiting. It is blocked or starved, and the real limit sits elsewhere on the line.
On a packaging line, speed differences between machines should be deliberate. The V-graph shows how they are meant to step up around the key machine.
Run a speed trial
Once you suspect the dial, test it with a short trial. Do it with the crew, on a normal product, with someone counting.
One product, one lot, one crew, a clean machine.
Step up from today's set speed, one step at a time.
Stop minutes, total units and scrap in each block.
Good units divided by 30, stops included.
Keep the best rate, and name what limited the next step up.
Example: the pouch line from above runs three 30-minute blocks.
| Set speed | Stop min | Run min | Units | Scrap | Good / min |
|---|---|---|---|---|---|
| 70 / min | 1 | 29 | 2,030 | 20 | 67.0 |
| 75 / min | 2 | 28 | 2,100 | 30 | 69.0 |
| 80 / min | 6 | 24 | 1,920 | 60 | 62.0 |
The crew's fear of 80 was fair, but 70 was too cautious. Moving to 75 gives about 3% more good units (69.0 / 67.0), with no new equipment.
A short trial proves the direction, not the long-term rate. Keep the new speed for a few weeks and check that the gain holds across lots and crews.
Traps in a speed project
- Lowering the ideal rate to match reality. Performance goes up and nothing else changes. Keep the rate and name the limit.
- Testing at full speed only. One bad block proves nothing without the blocks around it.
- Measuring output and ignoring scrap. A faster line that makes more waste can lose good units.
- Fixing the wrong machine. Speeding up a machine that is not the constraint adds no output. See bottleneck analysis.
From causes to action
Put a number on each cause you confirmed, in minutes per shift, and start with the biggest. The six big losses calculator shows how much reduced speed weighs against your other losses.
For a cause that survives the tests, dig until you can name the mechanism. A vague "it just runs better slower" is where most speed projects stall.
Fabrico is a light MES for OEE that takes stops, speed and counts live from PLCs or IoT sensors, with add-on AI cameras counting output where a machine has no usable signal. Speed and stops arrive as separate data, with OEE per line and shift. See how Fabrico tracks speed and stops.
What people ask next
What is the most common cause of reduced speed?
We found no neutral source that ranks the causes, so test your own line. Start with the set speed on the panel, because it is the cheapest check. Then compare the rate by product, as shown in OEE by product.
Is reduced speed the same as minor stops?
No. Reduced speed means the machine keeps moving but too slowly, while a minor stop means it halts briefly. Both land in performance when the stops are too short to log, so split them with a counter, as in minor stops vs breakdowns.
Should we run the machine at its nameplate speed?
Only if it holds that speed with acceptable stops and scrap. Nameplate is a good start for the ideal rate, but the running rate should come from a trial. Compare both on the ideal cycle time page.
How much does running 5% slower cost?
It takes 5% off output for every minute the line runs, if that machine is the constraint. Put your own rate and margin in the hidden factory calculator to see the yearly figure.
Why does the line slow down on night shift?
Check first whether night runs the same products, lots and setup rules. A real speed gap between crews often comes from settings or support, not effort. Why OEE differs between shifts gives the order of checks.
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.
Related guides: OEE on a New Line: When to Start the Clock