The article's own time study. BEFORE: 38, 48, 44, 29, 45 min = 204 min of work; 204 ÷ (5 × 45) = 90.7%. Eliminating waste (26 → 6 min) and trimming necessary NVA (30 → 23 min) leaves 177 min; moving 7 min of value-add out of Station 2 and 5 min out of Station 5 into Station 4 gives 34, 33, 37, 38, 35 min, and 177 ÷ 225 = 79%. The after bars' value-add split (28, 28, 30, 34, 28) follows from that move — the page tabulates the new totals but not their composition.
What a Yamazumi Chart Is and Why It Works
A Yamazumi chart is the simplest and most powerful visual tool in the line balancing toolkit. The Japanese word “yamazumi” means “stacking up” — and that is exactly what the chart does. It stacks each work element assigned to an operator or station into a vertical bar, making the total work content and its composition visible at a glance.
Here is what the chart shows:
| Chart Element | What It Represents | What to Look For |
|---|---|---|
| Each vertical bar | One operator or station | Compare bar heights across stations — uneven heights = imbalance |
| Green segments | Value-add work elements — activities the customer would pay for | This is the work you want to preserve and optimize |
| Yellow segments | Necessary non-value-add — required but not value-creating (walking, setup, inspection documentation) | Minimize but cannot eliminate — target for reduction |
| Red segments | Pure waste — searching, waiting, rework, unnecessary motion | Eliminate before rebalancing — never redistribute waste |
| Horizontal Takt line | The maximum allowable time per cycle | Any bar above this line = overloaded station = will miss Takt |
The chart’s power is that it makes three things simultaneously visible: (1) which stations are overloaded, (2) which stations have available capacity, and (3) where the waste is. No spreadsheet, no ERP report, and no verbal description achieves this with the same immediacy.
⚠️ The Most Important Rule of Line Balancing
Never balance around the waste — eliminate the waste first, then balance what remains. If Station 3 has 12 minutes of work content (8 minutes value-add + 4 minutes waste) against a 10-minute Takt, the fix is not to move 2 minutes of value-add to Station 4. The fix is to eliminate the 4 minutes of waste. Moving value-add elements to accommodate waste just embeds the waste permanently into the line design.
Building the Time Study That Feeds the Chart
The Yamazumi is only as good as the data behind it. That data comes from an element-by-element time study — not an ERP standard, not a “should take” estimate, and not a single observation. Here is the method:
Decompose the work into elements
An element is the smallest discrete task with a clear start and end point. “Install bracket assembly” is too coarse. “Position bracket on spar, insert 4 bolts, torque to 35 ft-lb, apply torque marking” — those are elements. Typical aerospace operations decompose into 6–15 elements per station.
Observe and time each element multiple times
Minimum 5 observations per element, ideally 10–15. Record every observation, including outliers. Use the median (not the mean) as your representative time — the median is resistant to outliers caused by one-time interruptions.
Classify each element
Value-add (green): Physically transforms the product toward completion — drilling, fastening, connecting, sealing, testing. Necessary NVA (yellow): Required by the process but doesn’t transform the product — reading work instructions, documenting, walking within the station, reaching for tools. Waste (red): Can be eliminated without affecting the product — searching for parts, waiting for crane, rework, walking to another area.
Involve operators in validation
Show the time study results to the operators who do the work. Ask: “Does this match your experience? Are there elements we missed? Are there elements that take longer on certain part variants?” Operators will catch errors that the IE missed — guaranteed.
⚠️ A Yamazumi Built Without Operator Input Is a Fiction
Standard work written in an office, time studies conducted by observation-only without operator validation, and line balances done from ERP data — all produce theoretical cycle times that nobody has ever actually achieved. The operators know which elements have hidden difficulty, which tool is unreliable, and which work instruction skips two steps that are actually required. If they are not in the room, the chart is not real.
The Rebalancing Process: From Imbalance to Flow
Once you have the Yamazumi built, the rebalancing process follows a strict sequence:
Eliminate waste elements first
Before moving any work between stations, attack the red segments. Can the parts search be eliminated by implementing point-of-use delivery (Water Spider)? Can the rework be eliminated by fixing the upstream quality escape? Can the waiting be eliminated by pre-staging tooling? Remove waste before you balance.
Reduce necessary NVA elements
Can the documentation time be reduced with better forms? Can the walking distance be shortened by repositioning tool racks? Can the inspection time be reduced with go/no-go gauges instead of measurements? Minimize yellow before you balance.
Redistribute remaining value-add elements
Now, with waste removed and NVA minimized, redistribute the remaining (mostly green) elements across stations so that every bar falls between 85% and 95% of Takt. Move elements from overloaded stations to underloaded ones. Respect element dependencies — some elements must follow others in sequence.
Calculate new line efficiency
Line Efficiency = (Sum of all station work content) ÷ (Number of stations × Takt Time) × 100%. If the result is below 85%, you may have too many stations. If above 95%, stations have insufficient buffer for variation.
Full Worked Example: Building and Rebalancing a Yamazumi
Scenario: A systems installation cell with 5 stations and a 45-minute Takt time. Time study data:
BEFORE rebalancing:
| Station | Value-Add | Necessary NVA | Waste | Total | vs. Takt (45 min) |
|---|---|---|---|---|---|
| Station 1 | 28 min | 6 min | 4 min | 38 min | –7 min (underloaded) |
| Station 2 | 35 min | 5 min | 8 min | 48 min | +3 min (OVERLOADED) |
| Station 3 | 30 min | 8 min | 6 min | 44 min | –1 min (tight) |
| Station 4 | 22 min | 4 min | 3 min | 29 min | –16 min (underloaded) |
| Station 5 | 33 min | 7 min | 5 min | 45 min | 0 min (at Takt) |
Diagnosis:
- Station 2 is overloaded at 48 min — it will miss Takt on every cycle, creating a bottleneck
- Station 4 is severely underloaded at 29 min — the operator is idle for 16 min per cycle (36% idle)
- Total waste across all stations = 4 + 8 + 6 + 3 + 5 = 26 minutes per cycle
- Line efficiency (before) = (38+48+44+29+45) ÷ (5 × 45) × 100% = 204 ÷ 225 = 90.7% — looks good on paper, but Station 2 is blocking the entire line
Step 1: Eliminate waste.
- Station 2’s 8 min waste: 5 min was parts searching (fixed by Water Spider delivery) + 3 min was rework from upstream quality escape (fixed by adding inspection step at Station 1)
- Other stations: similar root causes. Total waste reduced from 26 min to 6 min across all stations.
Step 2: Reduce NVA.
- Documentation time reduced by 30% through improved forms: saves 3 min across stations
- Walking reduced by repositioning tool shadow boards: saves 2 min across stations
AFTER waste elimination and NVA reduction (before element redistribution):
| Station | Value-Add | Nec. NVA | Waste | Total |
|---|---|---|---|---|
| Station 1 | 28 min | 5 min | 1 min | 34 min |
| Station 2 | 35 min | 4 min | 1 min | 40 min |
| Station 3 | 30 min | 6 min | 1 min | 37 min |
| Station 4 | 22 min | 3 min | 1 min | 26 min |
| Station 5 | 33 min | 5 min | 2 min | 40 min |
Total work content dropped from 204 min to 177 min. No station exceeds Takt, but Station 4 is still heavily underloaded.
Step 3: Redistribute value-add elements.
Move 7 min of value-add from Station 2 to Station 4, and 5 min from Station 5 to Station 4. New balances:
| Station | Total After Rebalance | % of Takt |
|---|---|---|
| Station 1 | 34 min | 76% |
| Station 2 | 33 min | 73% |
| Station 3 | 37 min | 82% |
| Station 4 | 38 min | 84% |
| Station 5 | 35 min | 78% |
Or — consider reducing to 4 stations:
Total work content = 177 min. Four stations × 45 min Takt = 180 min capacity. Line efficiency = 177 ÷ 180 = 98.3% — too tight, no buffer.
At 4 stations with 90% efficiency target: capacity needed = 177 ÷ 0.90 = 197 min. That requires 197 ÷ 4 = 49.2 min per station — exceeds 45-min Takt. So 5 stations is correct, and the 79% average efficiency provides ample buffer for variation.
The result: Same 5 stations, same operators, same Takt. Throughput increased from ~90% Takt compliance (Station 2 was blocking) to ~98% Takt compliance. No overtime. No new hires. The only change was eliminating waste and redistributing work elements.
Special Case: Handling Cure Times and Inspection Holds in Aerospace
Scenario: Station 3 of a fuselage assembly cell requires a sealant application followed by a 4-hour cure time. The cell operates on a 5-day Takt (40 productive hours per Takt period, single shift).
The problem: 4 hours of cure time is not work — it is mandatory wait time. No operator action occurs during cure. But the assembly cannot advance to the next operation until the cure is complete.
How to handle it on the Yamazumi:
- The cure time is classified as necessary NVA (yellow) — it cannot be eliminated, but it is not value-add
- On the Yamazumi, it appears as a large yellow block in Station 3’s bar
- The operator’s value-add work at this station might only be 2 hours (sealant application) + 1 hour (post-cure inspection and closeout) = 3 hours of work content
- The 4-hour cure occupies the station but not the operator
The design solution:
- Schedule the sealant application at the beginning of Day 3. Cure runs through mid-day.
- While cure is running, the operator moves to a “flex” assignment — supporting another station, performing preparation work for the next Takt cycle, or assisting with kitting
- After cure, the operator returns for post-cure inspection and closeout
- The station is “occupied” for the full cure period, but the operator is not — their labor is available for other use during the cure window
The key insight: Cure times, heat-treat times, and NDT scan times consume station time but not operator time. The Yamazumi must track both independently. Station occupancy determines physical flow. Operator loading determines labor efficiency. Confusing the two leads to either overloading operators (if you ignore cure time) or underloading them (if you count cure as their work content).
💡 The 85% Efficiency Target Is Not Laziness
A line balanced to 85% efficiency has 15% of each Takt period as buffer. This buffer absorbs: normal process time variation (±5–10%), occasional tool changes, brief interruptions for questions from adjacent stations, and the natural recovery from minor disruptions. Without this buffer, any disruption — however small — causes the line to miss Takt. The buffer is the system’s ability to maintain pace in a non-ideal world.
🎯 The Bottom Line
The Yamazumi chart translates abstract line balancing math into a visual that anyone — from the IE to the operator to the site director — can understand in seconds. Build it with operators. Eliminate waste before redistributing work. Respect the 85–95% efficiency target. And remember: the Yamazumi is not a one-time exercise. Every rate change, every engineering modification, and every process improvement should trigger a Yamazumi update. The chart is a living document that reflects the line’s current reality — or it is useless. Next: Theory of Constraints — finding and exploiting the one bottleneck that limits your facility.
Interactive Demo
The page's own five-station cell, before rebalancing. Move any station and watch balance efficiency and actual output disagree: efficiency averages the stations, but the line runs at the slowest one.
Stop reading, start modeling
Model your process flow, run simulations, optimize staffing with TOC math, and test your knowledge with 107 interactive checks — all in one platform.