The Shared Paint Line
Northline Valve paints three valve families on one shared powder line running two shifts: 840 minutes of net available time per day after breaks and maintenance. Line cycle time is 30 seconds per piece. Daily demand is 800 pieces of family A, 500 of B, and 300 of C. A color change takes 90 minutes, and the current schedule runs two changes per day. Saturday overtime has become routine. Work-in-process ahead of paint sits at 4,200 pieces, order-to-cash lead time is 18 days against a target of 6, and on-time delivery is 91% against a 10-day promise. A value stream map has just been completed; the paint line is the clear constraint and the largest WIP accumulation. The plant manager has a $1.4 million quote for a dedicated line for family A and asks you to endorse it. Engineering also mentions, in passing, that families A and B parts are small enough that two could hang per hook with a fixture change costing about $30,000.
1. Do the capacity math and assess whether the $1.4 million dedicated line is the right first move.
Read the worked answer
Capacity math. At 30 seconds per piece, the line can produce 840 × 2 = 1,680 pieces per day. Demand is 1,600. Run time for demand alone is 1,600 × 0.5 min = 800 minutes. Two 90-minute color changes consume 180 minutes. Total load = 980 minutes against 840 available - 117%. That shortfall is exactly the Saturday overtime.
So the line is genuinely short of capacity, but note where the shortfall sits: 180 minutes of changeover and 800 minutes of run. A dedicated line for family A removes changeover time, not run time. A alone takes 400 minutes; the shared line would then run B and C at 400 minutes plus one change per day, 490 of 840 - comfortable. The capital would work. That does not make it the best first move.
The cheaper lever. Two-up hanging reduces effective cycle time for 60% of volume to 15 seconds: 960 pieces at 0.5 min plus 640 at 0.25 min = 480 + 160 = 640 minutes of run time. Even with three changes per day at a modestly improved 45 minutes each, load is 640 + 135 = 775 of 840, about 92%. That delivers daily mix capability and eliminates overtime for roughly $30,000 plus a SMED effort.
Recommendation: do not endorse the capital yet. Run the fixture and SMED work first - it is faster, cheaper, and reversible. Keep the dedicated line as a documented fallback with a numeric trigger.
2. Design the future-state control for the paint line. How should the shared, high-changeover resource be scheduled, and what sits before and after it?
Read the worked answer
Paint is a process with sequence-dependent setup and limited capacity, which drives two design decisions.
Schedule with a levelled mix at a defined interval. Set the planning interval to one day - every family painted every day - because the math above supports it after the cycle and setup work. Build a daily levelled sequence, for example A-B-A-C-A-B repeated, rather than running all of A then all of B. The reason is not aesthetic: batching by family is what created the 4,200 pieces of WIP and the 18-day lead time, because each family waits a week for its turn.
Use pitch to manage it. Define pitch as 100 pieces, roughly 50 minutes at current cycle, and manage the line at that interval with a simple schedule board and a completion signal per pitch. Pitch gives the supervisor an early warning inside the day rather than at the end of the shift.
Control the flow with a supermarket downstream and FIFO upstream. Downstream of paint, hold a small finished-goods supermarket by family sized to demand during the replenishment interval; that is what decouples painting from shipping and lets a pull signal trigger production. Upstream, do not build a supermarket before a high-changeover process - use FIFO lanes with explicit maximum quantities per family, and stop the upstream process when a lane is full. A supermarket upstream would simply relocate the WIP.
Set the WIP cap as design, not hope. Target 1,200 pieces ahead of paint, in three capped FIFO lanes, and treat a full lane as an andon, not as a buffer to be exceeded.
3. What would you need to see at 90 days to conclude the future state is actually working, and which metric would you deliberately refuse to use?
Read the worked answer
Confirming evidence at 90 days:
- WIP ahead of paint down from 4,200 to 1,500 or less, with the FIFO caps respected rather than raided.
- Planning interval down from about a week to one day, verified from the schedule history, not the intent.
- Changeover from 90 to 45 minutes or better, with every changeover timed and posted.
- Saturday overtime hours at or near zero, replaced by schedule adherence during the week.
- On-time delivery from 91% to 95% or better against the 10-day promise, and order-to-cash lead time trending from 18 days toward 10.
- Fixture density live on A and B, with actual cycle time measured rather than assumed.
The metric to refuse: paint line utilization. High utilization is precisely what produced this problem. At 117% load the line looked superbly busy while the plant paid Saturday overtime, carried 4,200 pieces of WIP, and missed 9% of its promises. Driving utilization back toward 100% would reward the behaviour being corrected. Track it as a constraint-capacity check only - if the line is not near capacity at the new mix, the future state is not levelled enough - but never as a performance target for the paint supervisor.
Also watch the leading indicator that predicts a relapse: whether the levelled schedule is being broken by expedites. Count expedites per week. If that number is climbing, the WIP will follow within a month.