Line Balancing & Yamazumi Chart

The one rule

The slowest station sets the rate of the line, no matter how fast the others are. If four stations take 48, 63, 41 and 52 seconds, the line produces one unit every 63 seconds. Not 51, the average. Not 41, what the quickest operator manages. Sixty-three — because every unit must pass through the station that takes 63 seconds, and nothing downstream can work on a unit that has not arrived. Speed up station 3 and you change only how long its operator waits; take four seconds off station 2 and the whole line gets four seconds faster. Everything else follows from that sentence: it is why an average cycle time misleads, why "everyone is busy" is not the same as "the line is fast", and why the only work worth moving is work sitting on the longest station.

What this tool does

CM8-287 is a working line-balancing register. You list the work elements that build the product, assign each to a station, time it, and say whether it could be done elsewhere. The tool totals each station against takt, calculates line efficiency and the theoretical minimum number of stations, draws the Yamazumi stacked-time chart, and names the elements that could come off the bottleneck and the stations with room to take them. Everything runs inside this single file: no account, no upload, no network request.

Takt — the design target

Takt time is the rhythm the customer sets: available production time divided by demand. Work 27,000 seconds in a shift for a demand of 500 units and takt is 54 seconds. It is not a measurement of your line — it is the pace the line must hit to meet demand.

Balancing to takt means loading each station as close to takt as you sensibly can and stopping there; extra speed would only produce units nobody ordered. Balancing to the fastest possible line squeezes the work into the fewest stations at the highest rate — right only when you can sell everything you make. Leave a margin either way: a station at exactly 100 % of takt has nothing left for a dropped part. Around 85–90 % is a common choice, and the target line efficiency setting reflects it.

The Yamazumi chart

Yamazumi means "stacked up". One bar per station in line order; the height is that station's work content; each bar is split by type of work — value-adding, necessary non-value-adding, and pure waste. Read both things it shows:

• The height differences are the imbalance. Every second a short bar sits below the tallest bar is an operator waiting.
• The size of the waste band is work that should not exist at all — walking, waiting, searching, double handling, rework. Delete it; do not redistribute it.

A well-balanced line is a row of bars of nearly equal height, all a little below takt, with the waste bands squeezed to almost nothing. Level bars that are half waste are not a balanced line.

Line efficiency and the formulas

Line efficiency, or balance efficiency, asks what share of the paid station time is actually spent working: real work content against the time the line consumes when every station runs at the slowest station's pace.

For the four stations above: 48 + 63 + 41 + 52 = 204 seconds of work content, longest station 63 seconds. Efficiency is 204 ÷ (4 × 63) × 100 = 81.0 %. The missing 19 % is balance loss — 48 seconds of operator time per unit that produced nothing.

The theoretical minimum is the stations the work would need if each were loaded to exactly takt: 204 ÷ 55 = 3.71, rounded up to 4. Equal to the stations you run, as here, the work fits and the answer is to level them rather than add one. Smaller, and there is a station's worth of labour to release. Larger, and no rearrangement will meet takt: remove work content or add capacity.

Remove waste before you rebalance

This is the step people skip, and skipping it wastes the exercise. Move a station's nine seconds of walking to a quieter station and you have saved nothing — you have designed nine seconds of walking permanently into the balance. Rebalancing waste spreads it evenly; it does not remove it.

So mark each element's work type honestly first. Waste is a candidate for deletion: move the rack, present the parts at the station, remove the wait. Delete what you can, shorten what you cannot, then rebalance what remains — the candidates table follows the same rule and tells you to remove a waste element rather than move it. Necessary non-value-adding work sits in between: traceability scans, torque checks, packing. The customer will not pay for it, but you cannot simply stop.

Movable, splittable and fixed

An element that cannot move is not a candidate for anything, however attractive its seconds look. Three answers are available: can move, can be split, and fixed here — equipment, sequence, services or certification pin it to this station.

The tool refuses a fixed element without a written constraint, deliberately. "We have always done it here" is a habit wearing a constraint's coat. A real constraint names something: the press is bolted to the floor, the test rig is calibrated at that position. Written down it can be challenged and costed, and is sometimes removed by a second fixture or a longer hose; left unwritten it blocks the balance forever. The movable-work chart shows the slack you actually have — a station over takt whose time is nearly all fixed needs equipment duplicated or the element shortened, not a decision about where to put things.

Splitting an element

Splitting is the last resort. Remove the waste, move whole elements, shorten the long ones — and only then split, because a split element means the part is handled twice, two people must know the same job, and the balance depends on both doing exactly their share. Split at a natural break where the part is stable, and record the two halves as two rows with their own times and stations; never record one element against two stations. If a single element is longer than takt, no arrangement can meet takt until it is split or shortened.

Current and proposed scenarios

Every row carries a scenario, so one file holds both the line as it runs today and the rebalance you propose. Record the current state first, complete and honest; read the tiles, the Yamazumi chart and the candidates table; then copy the elements, set their scenario to "Proposed rebalance", and reassign stations until the proposed ones sit level and inside takt. The tiles, the Yamazumi chart and both tables always describe the current state, so they do not drift while you experiment; the comparison chart puts the two side by side once a proposed scenario exists.

When the balance is agreed, a Standard Work Sheet records the new element sequence at each station, and a Takt & Cycle Time Analyser re-derives takt from current demand — one built on last quarter's demand is out of date the day it lands.

The spreadsheet workflow

If the element times already live in a spreadsheet, you do not have to retype them.

• Spreadsheet template saves a CSV whose headings are exactly this tool's column names, with a guidance row showing what each expects: the accepted values for the scenario, work-type and movable lists, and the unit for each number.
• Fill in one row per work element — one element, one station, one time — and delete the guidance row before saving. Keep the file as CSV.
• Import spreadsheet reads it back, matching columns by heading, so their order does not matter and extra columns are ignored. Rows missing a required column or failing a validation check are skipped and reported by row number.

The file is read by this page in your browser: nothing is uploaded, and importing adds to what is here.

What this does not model

The arithmetic assumes a simple sequential line: one operator per station, each unit passing through every station once, elements assignable whole. Where yours is not, these numbers are indicative.

• Parallel stations — two stations doing the same work halve that step's effective time; here they appear as two heavily loaded ones.
• Mixed models — variants down one line each have their own element set, and the balance must hold across the mix weighted by volume.
• Shared or walking operators — one person covering two stations breaks the assumption completely.
• Buffers and machine cycles — stock decouples stations, and an unattended cycle is not operator time.
• Variation — every time here is a single number, and a station at 98 % of takt with variable work misses takt more often than the arithmetic suggests.

FAQ

How fine should an element be? Fine enough to move, coarse enough to observe — three to thirty seconds is a useful band. "Assemble the pump, 204 seconds" cannot be rebalanced.

Should machine time be an element? Only if the operator is occupied by it. If the machine runs unattended and the operator moves on, it is not station work content; if the operator stands and watches, it is, and it belongs in the waste band.

Why is a receiving-station suggestion sometimes missing? Because no other station can take that element without going over takt. Each is worked out independently against the current loading, so after one move the rest need re-reading — that is what the proposed scenario is for.

Can one file hold several lines? Yes, but the figures, charts and tables describe the line with the most elements recorded. Use the search filter, or keep a file per line.

Saving your work

Elements, settings and the report header are written to this browser's local storage as you type. That storage belongs to one browser on one computer: another browser, a private window, a second machine or a clean-up tool that clears site data will not have it.

Treat Export .json as the real save — one file containing everything, which Import .json restores anywhere. Export CSV gives you the element list for spreadsheet work. Reset asks twice, then erases everything. There is no undo.

Accuracy & disclaimer

The tool adds the times you entered and divides them as the formulas above state. It cannot tell whether an element was timed properly, whether a constraint is real, or whether the takt time still matches demand. One observation carries all the uncertainty of one observation — time several cycles before you move work.

A balanced chart is not a balanced line. Nothing here is proved until the operators have run the new arrangement for a full shift and the output has been counted, so involve the people who do the work before you move it. Changes to equipment, fixturing, guarding or layout must be assessed by whoever is competent for that line. This is an analysis aid, not engineering advice.