Yamazumi Chart Examples: Five Worked Scenarios

How to read these examples

Each scenario follows the same structure: the process and its takt, the measured chart, what the segments reveal, the change made, and the arithmetic of the result.

One caveat worth stating plainly. These are illustrative scenarios constructed from typical manufacturing conditions — they are not reports of named customer projects, and the figures are not benchmarks to expect on your line. Their value is the engineering logic: what the engineer looked at, in what order, and why that led to a specific change rather than another. Numbers on your own line will differ; the sequence of questions will not.

Example 1 — Automotive assembly: the constraint that was not assembly

A supplier assembles steering valve units across six manual stations. Demand sets takt at 58 seconds. Output is below plan, overtime has become routine, and the initial proposal on the table is to add a seventh operator.

Measured station cycle times:

WS4 exceeds takt by 13 seconds, so the line runs at 71 seconds per unit regardless of what the other five stations achieve. Balance efficiency is 324 ÷ (6 × 71) = 76%.

What the segments show

Breaking WS4 into work elements is where the decision changes:

Only 36 of the 71 seconds are value-added. Seventeen seconds — nearly a quarter of the cycle — is walking, searching and carrying. The station is not overloaded with assembly work; it is overloaded with everything around the assembly work.

The change and its arithmetic

Moving the fastener rack within reach recovers 7 seconds. Giving the test gauge a fixed location recovers 4. A short roller section removes 2 seconds of carrying. Total: 13 seconds, WS4 drops to 58 seconds, and no work is reassigned to another operator at all.

The line now paces at 58 seconds rather than 71. Same operators, same equipment: 71 ÷ 58 = 1.22, so the same shift produces about 22% more units. Balance efficiency rises to 311 ÷ (6 × 58) = 89%.

The lesson

The proposal on the table was a seventh operator. The chart showed the constraint was layout, not labour. Adding a person to a station that is a quarter waste buys a fraction of the improvement that removing the waste does — at permanent cost.

Example 2 — Electronics: when the problem is variation, not load

A control-panel line runs six operators against a 45-second takt. Average station times look healthy: 41, 43, 39, 44, 42 and 40 seconds. On paper the line is balanced. In practice, output varies by shift and operators downstream wait unpredictably.

Averages are hiding the problem. Splitting the functional test element out by operator:

Operators C and E take roughly twice as long on the same test as A and D. Investigation shows why: there is no defined test sequence, so some operators check functions in an order that requires re-handling the unit. The variation is procedural, not mechanical — and averaging it away made the chart look fine.

The change

Nothing is rebalanced. A standard test sequence is defined from the fastest method that still covers every check, operators are trained on it, and the test element settles into a 10–12 second band. Worst-quartile station times fall below takt, which is what actually stabilises output.

The lesson

A Yamazumi chart built purely on averages will pass a line that fails several times an hour. Where an element varies noticeably between operators, chart the spread as well as the mean — the difference between them is usually missing standard work, and standardising is cheaper than rebalancing.

Example 3 — Packaging: the manual station that caps an automated line

A packaging line runs automated forming, filling and sealing equipment with a machine cycle of 12 seconds per carton — a theoretical 300 cartons an hour. Actual output sits near 225. Machine utilisation reports look excellent, which is precisely why the problem stays hidden: the equipment genuinely is running well.

The Yamazumi chart includes manual stations alongside machine time, and manual carton preparation comes in at 16 seconds:

Half the cycle is walking. The line paces at 16 seconds, not 12: 3600 ÷ 16 = 225 cartons an hour, exactly the observed figure.

The change

Blanks are staged within arm's reach and the label printer is relocated beside the bench, cutting 7 seconds of walking. Preparation drops to 9 seconds — now below the 12-second machine cycle, so the equipment becomes the constraint again and output rises to roughly 3600 ÷ 12 = 300 cartons an hour, a 33% increase with no equipment change.

The lesson

On semi-automated lines the constraint is often the manual work that feeds the machine, and it is invisible in equipment reports because the machine is not the thing that is waiting. Charts that exclude machine time cannot show this — include both, or the analysis will confidently point at the wrong station.

Example 4 — Medical devices: rebalancing without touching compliance

A surgical instrument cell assembles at low volume with a 180-second takt. Every unit requires inspection and batch documentation, both mandated and neither removable. Cycle times vary between operators by more than 20 seconds on identical work.

The chart shows the assembly elements are consistent. Documentation is not — not in duration, but in placement. Some operators complete records during the automated clean cycle, when they would otherwise be waiting. Others complete them afterwards, adding the full 25 seconds to the station:

Both approaches satisfy the same regulatory requirement. Only one fits takt.

The change

Nothing is eliminated, shortened or delegated. Standard work is updated to place documentation inside the machine cycle — the same internal-versus-external distinction that drives changeover reduction — and the whole cell moves under takt.

The lesson

In regulated production the instinct is to protect inspection and documentation from improvement work, and that instinct is correct as far as elimination goes. It does not extend to placement. Asking when an element happens, rather than whether it must, frequently recovers the time without going anywhere near the compliance question.

Example 5 — Warehouse picking: Yamazumi outside the production line

A distribution centre wants to know whether workload analysis applies to order picking. It does — the method needs repeatable, measurable work elements, not a conveyor.

A picking cycle averages 240 seconds per order:

The activity the operation exists to perform — picking and placing — is 78 of 240 seconds. Travel alone is 40% of the cycle.

The change

Fast-moving lines are re-slotted to shorten routes, picking is organised by zone, and a drop point is placed at the zone boundary rather than at the packing station. Travel falls from 96 to 58 seconds and delivery from 24 to 10.

Cycle time becomes 188 seconds. Orders per picker per hour: 3600 ÷ 240 = 15 before, 3600 ÷ 188 = 19 after — roughly 27% more, with the same headcount and no new equipment.

The lesson

Any repeated process with definable elements can be charted this way. Where the work involves movement between locations, the travel segment is usually the largest single item on the chart and almost never appears in operational reports.

What the five scenarios have in common

Different industries, different volumes, and the same handful of patterns:

The consistent finding across all five is that the constraint is rarely where it is assumed to be, and rarely the thing that looks busiest. That is the entire argument for measuring before deciding.

A checklist for reading your own chart

1. Which stations exceed takt? That set caps output — everything else is secondary until it is resolved.
2. Within the worst station, what is the value-added share? If waste exceeds roughly 15% of the cycle, remove it before reassigning anything.
3. Is the imbalance consistent across cycles, or an artefact of one observation? Check the spread, not just the mean.
4. Is the overload labour, machine, material or inspection? Moving work only helps if the constraint is labour.
5. Would layout or presentation changes recover the time without moving work at all? They usually cost less and hold better.
6. After the proposed change, where does the constraint move to? A rebalance that pushes the bottleneck one station downstream has achieved nothing.
7. How will the result be verified? Rebuild the chart after implementation and compare like for like.

Presenting the result

A chart that stays inside the engineering office changes nothing. What travels is a short before-and-after pair with the arithmetic attached: takt, the constraint station, what specifically caused the overload, the change, and the resulting pace.

The examples above are deliberately written in that form. "Station 4 was 71 seconds against a 58-second takt, of which 17 seconds was walking and searching; relocating two items removed 13 seconds; the line now paces at 58" is a case a plant manager can approve. "Station 4 was overloaded and we improved it" is not.