Production Line Bottleneck WIP Surge Case Study

A production line can appear busy, fully staffed, and locally efficient while still failing customer demand. When upstream stations release work faster than the constrained station can complete good units, work in process grows, lead time rises, operators begin expediting, and delivery promises become unstable.

This case study follows an assembly line that misses daily output while inventory piles up before a functional test station. The first proposal is to add labor upstream because the upstream stations look crowded. The engineering review shows that the true constraint is the test station’s effective capacity after downtime, slow cycles, and retest loss are included.

The purpose is to connect takt time, station capacity, overall equipment effectiveness, WIP growth, Little’s Law, local-improvement traps, and validation evidence into one production decision.

Case Context

The line builds a configured electromechanical module. Work moves through five main stations: assembly preparation, mechanical assembly, functional test, final fit, and pack-out. Customer demand has increased, and planners expect the line to produce 420 good units per shift.

Item	Value
Net available production time	420\ \text{min/shift}
Customer demand	420\ \text{good units/shift}
Release rule before investigation	release one unit every takt interval
Functional test nominal cycle	58\ \text{s/unit}
Functional test downtime	35\ \text{min/shift}
Functional test performance factor	0.93
Functional test first-pass quality factor	0.96
Observed queue before test	96\ \text{units average}
Physical buffer limit before test	80\ \text{units}

The operational symptom is not only missed output. The WIP queue blocks aisles, hides priority, creates extra handling, and makes supervisors expedite jobs manually.

Takt Time

Takt time is available production time divided by customer demand:

Convert available time:

Demand is:

Therefore:

To meet demand without backlog growth, the line must produce one good unit about every 60\ \text{s} on average, after downtime, quality loss, changeover, staffing loss, and normal variability.

Station Screen

The observed station cycle times are:

Station	Function	Nominal cycle time
A	assembly preparation	48\ \text{s}
B	mechanical assembly	55\ \text{s}
C	functional test	58\ \text{s}
D	final fit	47\ \text{s}
E	pack-out	50\ \text{s}

At nominal cycle time alone, Station C looks close to takt but still acceptable:

That conclusion is wrong because it ignores downtime, speed loss, and first-pass yield. A station with a nominal cycle below takt can still be the bottleneck if its effective good-unit capacity is lower than demand.

Effective Test Capacity

Overall equipment effectiveness can be screened as:

where A_o is availability, P_o is performance, and Q_o is quality.

Functional test availability is:

Given:

the test station OEE is:

Nominal test capacity from cycle time is:

Effective good-unit capacity is:

The required output is:

The capacity gap is:

The test station is the bottleneck even though its nominal cycle time is below takt.

Effective Cycle Time

Another way to see the same constraint is effective cycle time:

Compare with takt:

The line cannot meet demand until Station C’s effective good-unit cycle time is brought below takt with margin.

WIP Growth Rate

The release rule sends work into the line at takt:

The bottleneck good-unit output rate is:

WIP before or at the bottleneck grows at approximately:

Over a 7-hour shift:

This matches the observed daily queue growth. The WIP surge is not random. It is the arithmetic result of releasing at demand rate into a station that can only produce about 356 good units per shift.

Lead-Time Consequence

Little’s Law gives:

where L is average WIP, \lambda is throughput, and W is average flow time.

For the queue before functional test:

and the bottleneck throughput is:

The implied average waiting time is:

If the target queue before test is:

then the corresponding waiting time is:

Reducing WIP without changing bottleneck capacity would reduce local waiting only temporarily. If release continues at 60\ \text{units/h} and test can only clear 50.9\ \text{units/h}, the queue returns.

False Local Improvement

One proposal is to add labor to Station B and reduce its cycle time from:

to:

Station B capacity would improve from:

to:

This does not increase line output while Station C remains limited to:

If the faster upstream station encourages release faster than takt, WIP grows even faster. At 49\ \text{s/unit}, the upstream release rate could become:

Queue growth against the unchanged bottleneck would be:

Over the shift:

The local improvement would make the visible queue worse. This is why bottleneck analysis must precede local efficiency projects.

Corrective Option Screen

Three options are reviewed.

Option	Effect	Risk
Add upstream labor	improves non-bottleneck capacity	increases WIP if release is not controlled
Add a second test stand	removes bottleneck strongly	capital, floor space, validation, maintenance burden
Improve test OEE and cycle time	addresses constraint directly	requires downtime reduction, script change, fixture discipline, and validation

The selected first action is to improve Station C directly before buying a second stand.

The corrective package is:

1. reduce test downtime from 35 to 10\ \text{min/shift} by replacing a worn fixture latch and adding pre-shift verification;
2. reduce nominal test cycle from 58 to 52\ \text{s} by removing a duplicated communication check;
3. improve performance factor from 0.93 to 0.96 by standardizing fixture loading;
4. improve first-pass quality factor from 0.96 to 0.985 by adding a connector seating check before test;
5. cap release into the line when the pre-test buffer exceeds the controlled WIP limit.

Corrected Capacity

New availability is:

New OEE is:

New nominal capacity is:

New effective capacity is:

Capacity margin against demand is:

So the corrected line has about:

good-unit capacity margin. This is acceptable as a controlled first step if product mix and downtime variation are monitored. It is not a reason to remove all buffers or stop improvement work.

Release and Validation

The line should not be released based only on one clean shift. The corrected state must prove that the bottleneck has moved or that the previous constraint is now stable above demand.

Acceptance evidence should include:

1. Station C cycle-time distribution before and after the test-script change;
2. downtime log separating fixture, software, material, operator, and retest causes;
3. first-pass yield by product variant and shift;
4. WIP before functional test measured at fixed intervals;
5. good units per shift for normal mix, not only easiest configuration;
6. release-rule compliance when the pre-test buffer reaches its WIP cap;
7. quality escape and retest records after the connector seating check;
8. operator workload review to confirm the new standard work is sustainable;
9. comparison of planned versus actual output for at least ten representative shifts;
10. escalation rule if Station C capacity falls below demand for two consecutive shifts.

The release criterion is:

with:

and no increase in customer escapes, retest loops, or manual expedites.

Engineering Lessons

The first lesson is that nominal cycle time is not enough. Effective capacity must include downtime, performance loss, first-pass quality, setup, staffing, and product mix.

The second lesson is that WIP growth is usually a capacity signal. If work is released faster than the bottleneck clears it, the queue grows no matter how busy the upstream stations look.

The third lesson is that local efficiency can be harmful. Improving a non-bottleneck station can increase inventory, handling, confusion, and lead time without improving customer output.

Good production engineering therefore treats flow as a system. Takt, bottleneck capacity, WIP, quality loss, human workload, and validation evidence must agree before the line is considered released.