EnergyFactorySense

Augos Factory / Learn / Research

Research · interruptions (short stops) on South African production lines

On 50 measured lines, interruptions took a tenth of paid time.

In short

Across 50 South African production lines over three months, stops too short to be given a reason took 10.1% of paid time. The median line lost 9%, and 22 of the 50 lines lost 10% or more. How much of this a plant sees depends on its stop threshold. On one bottling line, the same week showed interruptions as 2% of downtime at a 30-second threshold and 32% at five minutes.

10.1%
of paid time in interruptions, 50 lines, three months in 2025
Measured
9%
for the median line
Measured
22 of 50
lines at 10% or more
Measured
R4,950
per line per month to measure your own, ex VAT
Read the methodMeasure one of your lines ›

Finding 1

Interruptions are a large, uneven loss

Pooled across the 50 lines, interruptions took 10.1% of paid time. The spread between lines is wide, so the median, 9%, is the better guide to a typical line. Across the wider sample, including lines with less complete data, the figure was 6.5%.

A separate registered month on one beverage line shows the same order of size: 13,260 interruptions, 54.6 hours, 7.58% of 720 paid hours.

Interruptions as a share of paid time, by line

Measured
0% 10% 20% 30% 40% Line 1 of 50: 0.0% of paid time Line 2 of 50: 0.0% of paid time Line 3 of 50: 0.0% of paid time Line 4 of 50: 0.0% of paid time Line 5 of 50: 0.0% of paid time Line 6 of 50: 0.2% of paid time Line 7 of 50: 0.5% of paid time Line 8 of 50: 0.8% of paid time Line 9 of 50: 1.4% of paid time Line 10 of 50: 2.6% of paid time Line 11 of 50: 2.7% of paid time Line 12 of 50: 3.2% of paid time Line 13 of 50: 3.6% of paid time Line 14 of 50: 3.6% of paid time Line 15 of 50: 4.5% of paid time Line 16 of 50: 5.5% of paid time Line 17 of 50: 6.0% of paid time Line 18 of 50: 6.0% of paid time Line 19 of 50: 6.4% of paid time Line 20 of 50: 7.2% of paid time Line 21 of 50: 7.4% of paid time Line 22 of 50: 7.4% of paid time Line 23 of 50: 7.6% of paid time Line 24 of 50: 8.5% of paid time Line 25 of 50: 8.6% of paid time Line 26 of 50: 9.2% of paid time Line 27 of 50: 9.8% of paid time Line 28 of 50: 9.9% of paid time Line 29 of 50: 10.3% of paid time Line 30 of 50: 10.4% of paid time Line 31 of 50: 12.8% of paid time Line 32 of 50: 13.1% of paid time Line 33 of 50: 13.9% of paid time Line 34 of 50: 13.9% of paid time Line 35 of 50: 14.2% of paid time Line 36 of 50: 14.7% of paid time Line 37 of 50: 16.2% of paid time Line 38 of 50: 17.2% of paid time Line 39 of 50: 17.7% of paid time Line 40 of 50: 18.4% of paid time Line 41 of 50: 18.6% of paid time Line 42 of 50: 20.1% of paid time Line 43 of 50: 20.8% of paid time Line 44 of 50: 23.2% of paid time Line 45 of 50: 23.2% of paid time Line 46 of 50: 23.8% of paid time Line 47 of 50: 27.6% of paid time Line 48 of 50: 29.2% of paid time Line 49 of 50: 32.3% of paid time Line 50 of 50: 40.8% of paid time 10% OF PAID TIME MEDIAN LINE 9% 22 OF 50 LINES AT 10% OR MORE 50 LINES, LOWEST TO HIGHEST SHORT STOPS AS A SHARE OF PAID TIME · POOLED 10.1%

Augos platform, Production Timesheet · 50 anonymised South African lines with complete line-level data · September to November 2025 · site, products and people removed.

Finding 2

The stop threshold decides how much a report shows

Most systems ask for a reason only after a stop passes a set length. Shorter stops are either counted as interruptions or lost inside "running". On one bottling line, the same week of stops reads very differently depending on that threshold.

Across the study, sites with longer thresholds reported a larger interruption share. On 15 monthly points the link was moderate (r = 0.59), but those months come from five sites and are not independent of each other. At site level the sample is too small to confirm it (r = 0.75, p = 0.15). Treat it as a pattern worth checking on your own line, not a law.

Share of downtime counted as interruptions, by stop threshold

Measured
Stop thresholdStops countedShare of downtime
30 seconds2812%
1 minute4246%
2 minutes55714%
3 minutes61821%
5 minutes68832%

Plant export analysed by Augos · one anonymised bottling line · one week in 2026 · site, products and people removed.

What the data does not show

Frequent stops cost time, not speed

We report this so the study is not read for more than it holds.

Interruptions sit inside performance

In standard OEE, interruption time sits inside performance. A line that stops often looks slow, even when it runs at its rated speed between stops. Augos reports interruptions apart from speed, so each shows on its own.

The fragmentation gap

In our 2025 whitepaper, the sites with the most frequent interruptions averaged 12.54 percentage points lower OEE performance than the steadiest sites. The comparison used three-month site averages, four frequent-stop sites against three steady ones. Sites with performance above 100% were left out.

With interruption time taken out, the frequent-stop sites were not slower between stops. The gap is time lost to interruptions, so we call it the fragmentation gap.

Augos 2025 whitepaper, OEE performance by site · 7 of 12 anonymised sites, three-month averages · August to October 2025.

Method

How the figures were made

  • Source: stop events timestamped by a count sensor on each line's bottleneck, not operator logs.
  • Interruption: a gap in the count: the sensor misses at least three expected pulses, but the gap is shorter than the stop threshold. The machine may have halted, or run on with nothing to make. The sensor times it without asking for a reason.
  • Paid time: the whole shift less unpaid breaks. Paid breaks and planned stops count as paid time.
  • Lines included: the headline figures (10.1% pooled, 9% median, 22 of 50, the chart) come from 50 lines with complete line-level data at the same 12 sites, September to November 2025. The 2025 whitepaper used site-level totals for all 62 lines at those 12 sites, August to October 2025. Measured as a share of paid time, both give 10.1%.
  • Statistics: Pearson r for the threshold link.
  • Anonymity: no site, customer, product or person can be identified.

Limits

Read it with these in mind

  • The lines are Augos customers, not a random sample of South African plants.
  • Stop thresholds differ between sites, which changes what counts as an interruption.
  • Three months cover one season; a year may differ.
  • Correlation is not cause: the threshold link shows what reports show, not why lines stop.
  • Paid time is as each site configured it.

Citation

Use the findings in your own work

The findings and the method on this page may be quoted with the citation beside it. The line-level records behind them are not published as a dataset. If your research needs line-level data, ask us.

Ask about research data

How to cite

Augos (2026). Interruptions on South African production lines. Durbanville: Augos. augos.co.za/factory/research/short-stops/

Our 2025 whitepaper

Our 2025 whitepaper measured interruptions as a share of all downtime: 14.28% across 12 sites for August to October 2025. On the same data, interruptions came to 10.1% of paid time, the same as this page's 50-line figure.

Questions

The research, answered

What counts as an interruption?

A gap in the count: the sensor misses at least three expected pulses, but the gap is shorter than the stop threshold. The machine may have halted, or run on with nothing to make. The sensor times it without asking for a reason.

Is 10% typical for South African plants?

It is what these 50 lines showed over three months. They are Augos customers, so treat it as a guide, not a national average.

How do I find my own figure?

Time every stop on one line for a week, at an interruption threshold, with a sensor rather than a log. The three-month paid pilot does that on one of your lines. The paid pilot ›

Can I use the data in my thesis or report?

Yes, with the citation above. For students and lecturers ›

Find your own line's interruption share.

Thirty minutes with a manufacturing specialist, then a three-month paid pilot that times every stop on one of your lines, short ones included.

  • Every stop timestamped
  • Interruptions counted, not estimated
  • Prices in rand, ex VAT
Book a demoSee the paid pilot

A manufacturing specialist replies within two business hours to set a 30-minute call.