Spend any time on manufacturing LinkedIn and you'll have seen the argument. Batch production is outdated, and one-piece flow is the way forward.
It usually goes like this. Batching causes overproduction, excess work in progress (WIP), long lead times and hidden defects. Flow moves products through each step one at a time. WIP stays minimal, quality problems show up immediately and you become more responsive to customers.
On the surface it's compelling, and in some conditions flow genuinely is the right answer. But the people making the argument almost never state the assumptions it depends on. Connect stations directly with no buffer between them, do the maths, and the picture changes quite a lot.
Coupled stations multiply their downtime
In one-piece flow each station is coupled directly to the stations either side of it. There's no buffer and no decoupling stock to absorb a problem.
So if any single station stops, every station upstream is blocked and every station downstream is starved. It doesn't matter whether it's a breakdown, a quality issue, a material shortage or someone needing the toilet. The whole line stops.
That makes the maths straightforward. If each station runs independently at 90% availability, the line doesn't run at 90%. It runs at 90% multiplied by itself once for every station.
90% × 90% × 90% × 90% = 65.6%
Four stations at 90% each give you a line that runs at 65.6%. That's 24.4 percentage points below any one station, lost to the coupling alone. The more stations you add, the worse it gets.
| Stations | Each station | Whole line | Lost to coupling |
|---|---|---|---|
| 2 | 90% | 81.0% | 9.0 points |
| 3 | 90% | 72.9% | 17.1 points |
| 4 | 90% | 65.6% | 24.4 points |
| 5 | 90% | 59.0% | 31.0 points |
| 6 | 90% | 53.1% | 36.9 points |
| 7 | 90% | 47.8% | 42.2 points |
| 8 | 90% | 43.0% | 47.0 points |
The case for flow rests on three assumptions
The one-piece flow argument depends on three assumptions. They're rarely stated, and even more rarely true in a real factory.
Perfect uptime. Flow only delivers its theoretical benefits if every station runs at or very close to 100% availability. Every station has meaningful downtime, and the table shows what that does. Even 90% per station falls apart quickly.
Identical cycle times. The model assumes every station takes exactly the same time to process a unit. In practice cycle times vary, between stations and within one station over time. Without buffers to absorb that variation, whichever station is slowest at that moment becomes the bottleneck. Everything else waits.
You can't start downstream early. The argument against batching assumes the next step waits for the whole batch to finish. Transfer batches remove that wait. You move smaller sub-batches through the line and get most of the lead-time benefit of flow without the coupling penalty. Almost nobody mentions this option.
Small, well-sized buffers beat both extremes
The answer isn't a return to massive batches with weeks of WIP sitting between stations. Nobody is arguing for that. The answer is strategic decoupling: small, carefully sized buffers between stations. They let each part of the line keep running when something goes wrong elsewhere.
This isn't a complicated idea. Size each buffer from the reliability of the stations either side and the cycle-time variation you actually observe. A buffer of 5 to 15 minutes of work between stations is often enough. It absorbs most short stops and minor breakdowns without large inventories.
The result is a line that is far more robust than one-piece flow. WIP stays low, lead times stay short and quality feedback stays fast. You get most of the benefits the flow advocates promise, without the coupling penalty.
One-piece flow is right in four conditions
None of this means one-piece flow is always wrong. There are conditions where it genuinely is the right approach.
- Very high uptime. Above 98% per station, the coupling penalty is small enough that the simplicity of flow outweighs it.
- Cycle times that already match. When every station already runs at roughly the same speed, the variation penalty is minimal.
- Perishable product or strict first in, first out. When WIP has a shelf life or a contamination risk, you minimise time between stages regardless of the throughput cost.
- No room for buffers. When there's physically no space for stock between stations, flow may be your only option.
So one-piece flow isn't bad. It's a tool with specific conditions of use, not a universal improvement. Anyone presenting it as always better than batching is ignoring the maths.
What to check on your own lines
- Coupled stations. How many stations run with nothing between them? Multiply their availabilities together. That's the ceiling for the line while they stay coupled.
- Measured availability. Do you know each station's availability from data, or is it an assumption?
- What one stop does. When a single station stops, which stations are blocked and which are starved?
- Existing buffers. Did you size them from station reliability and cycle-time variation, or did they just grow?
- Transfer batches. Could smaller sub-batches give you the lead-time benefit without coupling the line?
- The four conditions. Does any of them genuinely apply to this line?
We help manufacturers work out the right balance of flow and buffering for their own equipment, products and reliability. If you'd like to talk it through, please get in touch at makeitpossible@majaco.co. If you'd rather watch than read, we covered the coupling maths and the buffer answer in a short video on LinkedIn.