Production Flow, Line Capacity, and WIP Exercises

These exercises practise production-flow calculations for takt, bottleneck capacity, work in process, setup loss, line balance, product mix, rework loops, learning curves, staffing reserve, local improvement traps and release evidence. The focus is the production system itself: how fast good units can move through a real line, cell, test process or mixed-model operation.

The goal is not only to compute a rate. A line can appear balanced on average and still miss customer demand because downtime concentrates at the constraint, rework consumes hidden capacity, setup time is ignored, WIP is released faster than the bottleneck can process it, or staffing assumptions are not valid on the shift being released.

Assume deterministic screening data unless an exercise states otherwise. Real release work should also check cycle-time distribution, operator qualification, tooling condition, fixture availability, material presentation, quality holds, maintenance windows, changeover discipline, line-side ergonomics, MES timestamps and the escalation rule for missed takt.

Release Evidence Notes

Production-flow evidence should be tied to a named product family, operating window and production boundary. State whether the calculation covers a manual cell, automated station, mixed-model line, inspection step, rework loop, packaging process, warehouse kitting lane or complete value stream. A capacity number without a boundary cannot support a staffing, release or investment decision.

Capacity evidence should separate nominal cycle time from effective good-unit output. Staffed time, planned breaks, downtime, speed loss, setup, first-pass yield, rework, test retest, changeover and material starvation can each govern a different constraint. Passing a nominal cycle-time screen does not prove that the production system can deliver good units at demand.

WIP evidence should identify the queue boundary. Little’s Law, Kanban caps, buffers and release rules only work when arrival rate, departure rate and inventory are measured over the same control volume. A WIP reduction that hides shortages upstream or starves the bottleneck is not an improvement.

Validation evidence should come from observed production behavior. Use time studies, station logs, downtime reason codes, first-pass yield, rework tickets, queue counts, dispatch history, maintenance logs, shift handover notes and trial-run records. Spreadsheet capacity should be treated as provisional until the constraint is observed under the released product mix.

Engineering Boundary Notes

The boundary for these exercises is production flow, not supplier replenishment policy. If the decision depends mainly on reorder point, safety stock, lead-time distribution, forecast bias, bullwhip amplification or supplier capacity commitment, use the companion supply-chain exercise set instead.

Within the production boundary, define the unit of flow before calculating. One unit may mean one finished product, one kit, one batch, one pallet, one job, one test fixture, one lot, or one order line. Mixing these units is a common source of false capacity.

The constraint should be stated explicitly. A station can be the current bottleneck by cycle time, by downtime, by labor qualification, by yield loss, by setup exposure, by inspection hold, or by information delay. The release action should match the real constraint.

Scenario Map

Scenario	Exercises	Primary check	Engineering decision
Takt and station capacity	1, 2, 4, 5, 16, 18	Takt, station load, balance efficiency, mixed-model demand and release margin	Staff, rebalance, slow release, protect capacity or reject a claimed line rate.
WIP and queue control	3, 10, 11	Little’s Law, buffer time and WIP release cap	Set queue limits, avoid over-release and protect the constraint from starvation.
Effective output and hidden losses	6, 7, 8, 9, 12, 13	Downtime, OEE, setup, yield, rework and starts required	Decide whether improvement work targets the actual loss mechanism.
Ramp, staffing and investment	14, 15, 17	Learning curve, absenteeism reserve and payback from added constraint capacity	Release a ramp plan, cross-training plan or constraint investment.

Exercise 1: Takt Time and Required Staffing

A cell has available production time:

Customer demand is:

The manual work content is:

Estimate takt time and the minimum number of operators before allowances.

Solution

Takt time is:

The operator count from work content is:

Because a fraction of an operator cannot run the standard work as stated:

Engineering Comment

This is a staffing screen, not a final line design. The engineer should still check balance loss, walking time, ergonomic load, material presentation, rework, quality checks, qualification and whether demand variation requires cross-trained support.

Plausibility Check

Demand of 130 units in 390 minutes gives exactly 3 minutes per unit. Work content of 7.2 minutes is more than two takt intervals but less than three, so three operators is the expected rounded result.

Exercise 2: Bottleneck Shift After Parallel Capacity

A three-step line has cycle times:

Step	Cycle time
A	45\ \text{s/unit}
B	60\ \text{s/unit}
C	50\ \text{s/unit}

A second identical station is added in parallel at step B. Find the line capacity before and after the change.

Solution

Before the change, the bottleneck is step B:

With two identical B stations in parallel, the effective B cycle time is:

The slowest remaining step is C:

Therefore:

Engineering Comment

The added parallel station improves capacity, but it moves the bottleneck to step C. Further investment at B would not increase output unless C is also improved or the operating policy changes.

Plausibility Check

Capacity rises from 60 to 72 units per hour, not to 120. That is credible because only one station was duplicated and another station becomes the limiting step.

Exercise 3: WIP and Lead Time with Little’s Law

A production lane completes:

The average WIP inside the controlled boundary is:

Estimate average flow time through the boundary.

Solution

Little’s Law gives:

so:

Convert to minutes:

Engineering Comment

The result is valid only if WIP and throughput were measured over the same boundary and stable period. If WIP includes a queue that throughput does not include, the lead-time estimate is not meaningful.

Plausibility Check

At 48 jobs per hour, the lane finishes 0.8 jobs per minute. A queue of 36 jobs therefore represents roughly 36/0.8=45 minutes of work, matching the calculation.

Exercise 4: Line Balance Efficiency

A four-station line has station loads:

Station	Load
1	42\ \text{s}
2	55\ \text{s}
3	48\ \text{s}
4	50\ \text{s}

Customer takt is:

Calculate line balance efficiency and idle time per cycle.

Solution

Total work content is:

Available station time per cycle is:

Balance efficiency is:

Idle time per cycle is:

Engineering Comment

Every station is below takt, so the line can meet the nominal rate. The improvement question is whether the idle time can be removed without creating ergonomic overload, precedence violations, quality escapes or material handling conflicts.

Plausibility Check

The largest station load is 55 seconds, below the 60 second takt. Efficiency near 80\% is plausible because 45 seconds of idle time are spread across four stations.

Exercise 5: Mixed-Model Product-Mix Capacity Reserve

A line builds products A and B. The daily demand mix is:

Product	Demand	Constraint time
A	180\ \text{units/day}	80\ \text{s/unit}
B	90\ \text{units/day}	130\ \text{s/unit}

The constraint has:

Find whether the mix fits with at least 8\% reserve.

Solution

Required constraint time is:

Available time is:

Reserve fraction is:

The required reserve is 8\%, so the mix is not releasable.

Engineering Comment

The problem is not total daily units alone. Product B consumes more constraint time, so a shift in mix can break a line that looked adequate under average cycle time.

Plausibility Check

The required time is only 900 seconds below available time, or 15 minutes. That is far less than an 8\% reserve on a 7.5 hour day, so failure is expected.

Exercise 6: Downtime-Adjusted Weekly Capacity

A machine has a nominal rate:

It is staffed for:

Planned changeover consumes 3.5 hours and unplanned downtime averages 7\% of the remaining time. Estimate weekly capacity.

Solution

Time after changeover is:

Effective production time after downtime is:

Weekly capacity is:

Round down for a release screen:

Engineering Comment

Nominal rate is not capacity if changeover and downtime are real. The release package should identify whether downtime is random, recurring, maintenance-driven, material-driven or operator-driven.

Plausibility Check

Without losses, 52(38)=1976 units would be possible. Losing 3.5 hours plus 7\% of the remainder should reduce capacity by about 300 units, which matches the result.

Exercise 7: OEE and Good-Unit Capacity

A packaging line is scheduled for:

It runs for 410 minutes, has a performance rate of 92\% against standard speed, and first-pass quality is 97\%. The standard speed is:

Compute OEE and good-unit output.

Solution

Availability is:

Performance is:

Quality is:

OEE is:

Good-unit output is:

Engineering Comment

OEE is useful because it keeps downtime, speed loss and quality loss visible. For release decisions, the loss categories should be backed by reason codes rather than a single blended percentage.

Plausibility Check

The theoretical output is 8640 units. An OEE near 76\% gives about 6600 good units, which is consistent with the computed value.

Exercise 8: Setup Time and Batch Utilization

A machine runs at:

Each batch requires:

Batch size is:

Find setup loss as a fraction of batch processing plus setup time.

Solution

Processing time is:

Total batch time is:

Setup fraction is:

Engineering Comment

An 11.5\% setup loss may be acceptable or unacceptable depending on demand variability, product mix and inventory exposure. Reducing setup can justify smaller batches, but only if material handling and scheduling discipline support the change.

Plausibility Check

Setup is about one eighth of the processing time. The total-time fraction should therefore be slightly below one eighth, near 12\%.

Exercise 9: Changeover Reduction and Capacity Recovery

A line completes 920 good units per week. Current changeover time is 10 hours per week. A SMED project is expected to reduce changeover by 35\%. The effective output rate during production is:

Estimate extra weekly output if demand exists.

Solution

Recovered time is:

Additional output is:

New weekly output is:

Engineering Comment

The benefit is real only if the line constraint was blocked by changeover time and there is enough demand, material, labor and downstream capacity. Otherwise the recovered time may become idle time.

Plausibility Check

Recovering 3.5 hours at about 30 units per hour should add just over 100 units, so 112 is credible.

Exercise 10: Buffer Time Before a Bottleneck

A bottleneck consumes one job every:

The release team wants a protective buffer of:

How many jobs should be queued ahead of the bottleneck?

Solution

The buffer count is:

Engineering Comment

The buffer protects the constraint from starvation, but too much buffer hides upstream instability and lengthens lead time. The release rule should define both minimum and maximum queue limits.

Plausibility Check

At one job every 4.5 minutes, 10 jobs cover 45 minutes. A 90 minute buffer therefore needs 20 jobs.

Exercise 11: Pull Release Cap from WIP and Takt

A pull lane targets:

The takt rate is:

Set the maximum WIP cap for the controlled lane.

Solution

Using Little’s Law:

Therefore:

The WIP cap is:

Engineering Comment

The cap is useful only if releases stop when WIP reaches the limit. If planners continue to release work for local utilization, the cap becomes a report instead of a control.

Plausibility Check

A rate of 24 units per hour is 1 unit every 2.5 minutes. Holding 60 units means about 150 minutes, or 2.5 hours, of flow time.

Exercise 12: Yield Loss and Starts Required

Customer demand is:

First-pass yield is:

Estimate how many units must be started if no rework is credited.

Solution

Good units are:

so starts are:

Round up:

Engineering Comment

Yield loss consumes capacity, material and inspection time. A production plan that schedules only 960 starts will miss demand even if the nominal rate appears sufficient.

Plausibility Check

A 6\% loss on about 1000 units is about 60 units. Starting about 1020 units to get 960 good units is therefore reasonable.

Exercise 13: Rework Loop Capacity Penalty

A test station can process:

Each unit requires one initial test. A fraction:

requires one retest after rework. Estimate good-unit throughput limited by the test station.

Solution

Tests required per released unit are:

Good-unit throughput limited by test capacity is:

Engineering Comment

Retest load is a hidden capacity consumer. The corrective action may be yield improvement, diagnostic automation, separate retest capacity or release throttling, not simply adding upstream production.

Plausibility Check

If every unit needed one test, throughput would be 80 units per hour. A 12\% retest burden should reduce that to the low 70s, matching the result.

Exercise 14: Learning Curve Staffing Ramp

A new assembly process starts at:

After training, the expected time is:

Demand requires:

Compare required operators before and after training.

Solution

Initial operator count is:

Round up:

After training:

Round up:

Engineering Comment

The ramp plan needs one extra operator until the trained work content is demonstrated. Releasing at the future staffing level before evidence exists risks missed takt, fatigue and quality escapes.

Plausibility Check

Reducing work content by 25\% should reduce labor need from about five people to about four, which is exactly what the rounded calculation shows.

Exercise 15: Absenteeism and Cross-Training Reserve

A line requires 18 qualified operators. Historical unplanned absence is 7\% per shift. Management wants at least one extra qualified operator beyond the expected absence load. How many qualified operators should be scheduled?

Solution

Expected absence count is:

Add one extra qualified operator:

Round up reserve:

Schedule:

Engineering Comment

The calculation is an average staffing screen. For critical stations, the reserve must be role-specific; three extra people are not useful if none are qualified on the constraint operation.

Plausibility Check

Seven percent of 18 is a little more than one person. Adding one additional reserve and rounding up gives three extras, so 21 scheduled qualified operators is plausible.

Exercise 16: Constraint Investment Payback from Throughput

A constraint improvement costs:

It increases good output by:

Contribution margin is:

Find simple payback in weeks.

Solution

Weekly contribution increase is:

Simple payback is:

Engineering Comment

The payback is valid only if the added output is at the system constraint and can be sold or used to reduce backlog. If another constraint appears immediately, the benefit is overstated.

Plausibility Check

At about \2500per week, a$50000investment should pay back in about20weeks, so19$ weeks is credible.

Exercise 17: Overtime Recovery After a Missed Shift

A line loses one shift of:

Its effective good-unit rate is:

Demand recovery requires the lost output to be made up over four days. Overtime productivity is expected to be 85\% of normal. How many overtime hours per day are needed?

Solution

Lost output is:

Overtime rate is:

Total overtime hours required are:

Per day over four days:

Engineering Comment

The recovery plan should check fatigue, overtime staffing, material availability, maintenance access and whether quality inspection can also support the added hours. Output recovery without inspection capacity can create a delayed quality backlog.

Plausibility Check

One lost shift is about seven normal hours. Because overtime is less productive, recovery should require more than seven total overtime hours, and 8.24 hours is consistent.

Exercise 18: Production Flow Release Gate

A mixed-model line has available shift time:

The released mix requires:

Expected downtime is:

Rework retest consumes:

The release rule requires at least 5\% free constraint capacity after losses. Decide whether the line can be released.

Solution

Effective available time after losses is:

Free time after required mix is:

Reserve fraction relative to effective available time is:

The required reserve is 5\%, so the line should not be released without mitigation.

Engineering Comment

The production release fails even though the raw required time is below raw available shift time. Downtime and rework consume the margin. The practical actions are reduce release quantity, add overtime, improve first-pass yield, protect the constraint or delay release until evidence improves.

Plausibility Check

The raw margin is 2400 seconds, but losses consume 1550 seconds, leaving only 850 seconds. A reserve near 3\% is therefore expected and below the 5\% gate.

Validation Package Checklist

A strong production-flow solution should check:

• whether demand, takt and product mix use the same time window;
• whether the constraint is identified by effective good-unit capacity, not nominal cycle time only;
• whether WIP, throughput and flow time use the same control boundary;
• whether line-balance changes respect precedence, ergonomics, variation and quality checks;
• whether setup, downtime, retest, rework and changeover losses are measured separately;
• whether staffing assumptions include qualification, absence, fatigue and shift handover;
• whether improvement benefits occur at the system constraint rather than a non-bottleneck station;
• whether release decisions include observed trial evidence, not only spreadsheet capacity.

Common Release Mistakes

Common mistakes include using average cycle time when product mix changes the bottleneck, counting installed equipment as effective capacity, balancing work content to takt without checking station variation, releasing WIP faster than the constraint can process it, reducing visible inventory while starving the bottleneck, ignoring setup and rework loops, treating OEE as a single unexplained loss number, improving a non-bottleneck and expecting system output to rise, applying future learning-curve performance before training evidence exists, and releasing a line without measured constraint behavior under the real shift pattern.