Who this is for
You have commissioned a line. Every machine passes its own test. The supplier demonstrates each unit at its rated speed and the numbers check out. But the pallets at the end of the day come to something close to half of what you budgeted, and nobody can point at the machine that is wrong — because no single machine is wrong.
This is the most common capacity dispute in imported-line projects, and it is almost always won or lost at specification time, months before anyone is standing on the floor arguing. The purpose of this article is to show you the arithmetic so you can put it in the contract.
The number on the quotation is a machine number, not a line number
The rated speed of a filler is a property of the filler: one product, ideal material, a full infeed, a clear discharge, no changeover, a technician standing next to it. Your line has none of those conditions for more than a few minutes at a time. As one plant engineer put it bluntly in discussion we collected during 2026: “machine nameplate speed or the fastest individual module is not the same as finished-bottle output.”
Another described the delivery experience precisely: “a line can look complete on paper, with every machine positioned, wired and switched on, and still fail to produce a single acceptable unit on day one.” And a third named the consequence: “a pile of fast machines spends its first quarter being debugged on your floor.”
Note what these three have in common. None of them describes a broken machine. They describe a line that was bought as a set of machines and therefore behaves as a set of machines.
Four multipliers stand between the badge and the pallet
The gap is not one effect. It is four, and they multiply rather than add — which is why the result surprises people who estimated each one separately and thought the total would be tolerable.
1. The constraint is not the machine you named the line after
A line runs at the rate of its slowest station. Buyers usually name the line after the filler or the press — the expensive machine — but the constraint is frequently something cheap and downstream: a shrink wrapper, a labeller, a cooling tunnel, a bagger. If the wrapper does 5,400 units an hour, you own a 5,400-unit line no matter what the filler is capable of.
2. Availability compounds down the line
This is the multiplier that does the real damage, and it is the one almost never discussed at quotation stage. When stations are connected in series with no accumulation between them, any station that stops stops the whole line. The line’s availability is therefore the product of the station availabilities, not the average.
Six stations that are each available 96% of the time give a line available 0.96 × 0.96 × 0.96 × 0.96 × 0.96 × 0.96 = 78% of the time. At 95% each it is 74%. At 90% each — not unusual on a line still being debugged — it is 53%. Every station you add makes it worse, which is the counter-intuitive part: a longer, more automated line is more fragile than a short one unless it is buffered.
Accumulation — conveyor buffers, accumulation tables, surge hoppers — is what breaks the multiplication. A buffer that holds ninety seconds of production lets a downstream station absorb a short upstream stop without starving. Buffers are the cheapest capacity on a line and the first thing removed from a quotation to make the price competitive.
3. Changeover eats the week, not the hour
A line quoted at a rate is quoted on one product running continuously. Your week has product changes, and each one costs cleaning, tooling, adjustment and re-qualification. Two 35-minute changeovers in an eight-hour shift remove 15% of the available time before a single unit is made.
4. Start-up rejects and yield
Every start and every changeover produces a run of out-of-spec product before the process settles — short fills, mis-registered labels, off-weight bags. On a line with frequent changeovers, start-up scrap is often the largest single quality loss, and it is systematically under-recorded because it happens during setup rather than during “production”.
The arithmetic on a real-shaped line
Below is a six-station bottling line sold as a “6,000 bottles per hour” line, because that is the rating of the filling monobloc. Nothing here is faulty. Every machine performs to specification.
| Station | Rated speed (bph) | Comment |
|---|---|---|
| Depalletiser | 8,000 | Ample |
| Rinser / filler / capper monobloc | 6,000 | The number the line is sold under |
| Labeller | 6,500 | Ample |
| Date coder | 12,000 | Ample |
| Shrink wrapper | 5,400 | The actual constraint |
| Palletiser | 7,000 | Ample |
So the line is a 5,400 bph line on paper, not 6,000. Now apply the other three multipliers:
| Step | Factor | Running figure (bph) |
|---|---|---|
| Constraint station rate | — | 5,400 |
| Series availability, six stations at 96%, minimal buffering | × 0.78 | 4,230 |
| Two 35-minute changeovers in an eight-hour shift | × 0.85 | 3,610 |
| Quality yield including start-up rejects | × 0.96 | 3,470 good bottles per hour |
Result: about 58% of the badge. The figures above are illustrative, not a benchmark — substitute your own station speeds, availabilities and changeover pattern and the shape of the answer will not change. The point is the structure: four multipliers, none of them individually alarming, compounding to roughly half.
This is also the honest way to read any supplier’s capacity claim, including ours. If a quotation gives you one number and no station table, no availability assumption and no changeover allowance, you have been given a machine specification and asked to plan a business on it. The equivalent discipline on the operating side — measuring availability, performance and quality separately rather than as one blended figure — is set out in the international KPI standard for manufacturing operations, ISO 22400-2, and it is the same decomposition used here.
What to specify instead
Everything above is fixable in the purchase specification. Six clauses do most of the work:
- Buy a line output, not machine speeds. Write the acceptance figure as good, packed units per hour on your product, measured at the end of the line. Machine ratings become supporting detail, not the contractual number.
- Make the supplier name the constraint. Require a station-by-station table with rated speeds, and require them to identify which station sets the line rate. A supplier who will not produce that table has not engineered a line.
- Specify accumulation in seconds. Ask for the buffer time between each pair of stations. “Ninety seconds between filler and wrapper” is a specification; “conveyor included” is not.
- Make changeover an acceptance item. A target changeover time, demonstrated, with your operators doing it. Otherwise it is discovered in month three.
- Require a sustained run, not a demonstration. A four-hour continuous run on your material, counting good output and rejects against the agreed threshold. This is the single clause that separates a real acceptance from a showroom visit — see our factory acceptance test checklist and FAT vs SAT for how to structure it.
- Tie the final payment to the line figure. Not to delivery, not to power-on, and not to the machines passing individually.
Reading a quotation before you sign it
Four questions, none of which requires an engineer to ask:
- Which station is the slowest, and what is its rated speed?
- What buffer time sits between the stations, in seconds?
- What availability has been assumed per station, and what does that make the line availability?
- Is the quoted rate a machine rate or good units off the end of the line?
If the answers are vague, the gap in this article is already in your project; it simply has not arrived yet. And if you are still choosing between line configurations, the capacity you actually need should be settled first — our sizing guides for maize milling and PET, glass and can beverage lines work through that step, and the total cost of ownership guide shows why annual output must be estimated at real operating rate rather than nameplate.
Where the output goes after commissioning
The four multipliers do not stop applying once the line is accepted. Availability drifts, changeovers lengthen as tooling wears, and start-up scrap creeps up. The difference is that after handover nobody is measuring, so the decline is invisible until a budget is missed. Machine-sourced measurement of the three components separately is what makes it visible, and it can be retrofitted to a line that has no modern controls — our guide to measuring OEE on an old line without a new PLC covers that path.
What CISH does in this part of the process
We specify lines as line outputs. That means producing the station table and naming the constraint before a quotation goes out, sizing accumulation deliberately rather than leaving conveyors as a cost line, writing the sustained-run acceptance protocol into the contract at purchase time, and witnessing the run in China against that protocol. Where a buyer already has quotations in hand, we will read them against the four questions above and tell you what each one is actually offering.
Frequently asked questions
Why is my production line slower than its rated capacity?
Because the rating is a machine rating under ideal conditions, and four multipliers stand between it and your pallet: the line runs at its slowest station, not its most expensive one; availability compounds as a product across stations in series; changeovers remove time from the shift; and start-up rejects remove yield. Six stations at 96% availability with no buffering already put line availability at 78% before changeover and yield are applied.
What is line balancing?
Matching the throughput of every station so no machine is starved or blocked by its neighbours. A line is balanced when the stations upstream and downstream of the constraint have enough margin and enough accumulation to keep the constraint fed and clear. An unbalanced line wastes the capacity you paid for in the fast machines, because they spend their time waiting.
How much output should I expect from a line rated at 6,000 units per hour?
As an order of magnitude, 3,000–3,800 good packed units per production hour is a realistic expectation on a typical six-station line with modest buffering and two changeovers per shift — roughly 50–65% of the badge. Your own figure depends on your constraint station, your buffer times, your changeover pattern and your yield. The way to avoid the surprise is to make the supplier state those four things before you sign.
Why does adding accumulation conveyor increase capacity?
Because it breaks the multiplication of availabilities. Without a buffer, a short stop anywhere stops everything; with a buffer that holds, say, ninety seconds of production, a downstream station keeps running through an upstream micro-stop instead of starving. Accumulation is usually the cheapest capacity on a line, which is also why it is the first item trimmed when a quotation is being made to look competitive.
What should the acceptance test measure?
Good, packed units per hour on your own product, over a sustained run of several hours rather than a short demonstration, with rejects counted against an agreed quality threshold and changeover demonstrated by your own operators. The contractual number should be the line figure at the end of the line, with the final payment tied to it — not to the individual machines passing their own tests.
A note on the numbers
The six-station example is illustrative arithmetic built to show how the four multipliers compound, not a benchmark for any particular industry or supplier. Substitute your own station speeds, availability assumptions, changeover pattern and yield. Quotes are reproduced verbatim from public engineering discussion collected during 2026 and anonymised.