The problem
Where the time was going
This maker of custom windows, doors, and vinyl porch products had grown 25 percent a year for five straight years, and that success had outgrown a single plant. The company wanted to expand to a second facility and lift production another 25 percent, while holding shifts to a 40-hour week, and it brought in a manufacturing extension partnership to do it in the most cost-effective way possible.
Supervision was sure it already knew the constraint. A cutting operation looked like the bottleneck, and the team had found the exact saw they wanted to replace it. The machine would cost 700,000 dollars, and a purchase order was effectively waiting on a green light. The only question left, in their minds, was confirmation.
The trouble with that certainty was that nobody could see the whole line at once. The plant ran a batched process, which built up large inventory and long lead times, and a hunch about one station is a risky basis for a 700,000 dollar commitment.
What we modeled
Mapping the process, then testing the fix
A model was built for each business unit and then validated to confirm it matched the real process before anyone trusted its answers. With the model standing in for the floor, the staff watched the simulation run and saw the bottlenecks emerge, and they did not emerge where the team expected.
The simulation showed plainly that the saw was not creating the constraint. The work was bogging down in the handling of the windows, a direct result of the batched process rather than any single machine. With the real problem in view, the model became a place to test changes safely.
The team ran a series of what-if scenarios on the model: unbatching the operations, building a flow line to cut movement and handling time, running one window at a time, removing sorting steps along the line, adding three people, and doubling the glass-washing capacity. The results pointed to replacing the single batch line in the glass division with two one-piece flow lines.
The result
The proof, and the payoff
Because the model proved the saw was not the constraint, the 700,000 dollar purchase was avoided before any money was committed. The company had been about to buy the wrong piece of equipment, and the simulation caught it in time.
The redesign did more than save the capital. At full-production capacity the baseline line met a weekly output of 1,945 windows, and after the lean changes the weekly output rose to 2,400 windows with no new capital and no change to the 40-hour week. The one-piece flow design also freed about 2,000 square feet of floor space for future expansion, while cutting inventory and shortening lead times.
- 2,400 / week
- output, up from 1,945 windows
- $0 new capital
- the one-piece flow needed no investment
- ~2,000 sq ft
- floor space freed for expansion
Part of our work in manufacturing.