A plant for hydraulic valves bought an assembly cell three years ago. The specification had six pages. The supplier quoted a cycle time of 27 seconds, the cell ran at 31 seconds on the day of handover, and the planned output was reached 14 weeks after the start. Retrofits of EUR 86,000 were paid by the plant, because nobody could show a written criterion the supplier had missed.
The cell was not badly built. It was badly ordered. This series follows the next purchase of the same plant, the assembly and test cell AC7, and starts where every purchase should start: with the need.
The need comes before the machine
Before anyone talks about stations and robots, one page must answer six questions. Which product in which variants? How many units per year over the planned years of use? In which working time? With which cycle time, availability and yield? Where in the layout, with which budget? And on which day must the equipment produce?
The page is signed by the production manager. Every later requirement can be traced back to it.
Four numbers from the demand
The performance data are calculated, not taken from a catalog. For AC7 the calculation looks like this:
| Step | Calculation | Result |
|---|---|---|
| Demand per day | 432,000 units / 240 days | 1,800 units |
| Available time per day | 2 shifts x 450 minutes x 60 | 54,000 s |
| Customer takt | 54,000 / 1,800 | 30.0 s |
| Required cycle time | 30.0 x 85 percent planned effectiveness | 25.5 s |
| Specified cycle time | Rounded down, small reserve | 25.0 s |
The check is simple. At 25.0 seconds the cell makes 2,160 units per day at 100 percent. At a planned overall effectiveness of 85 percent it delivers 1,836 units, 2 percent above the demand.
Who carries which loss
The planned effectiveness of 85 percent hides three different losses, and the specification must separate them. Otherwise every stop becomes a discussion.
- Technical stops are caused by the equipment. The supplier is responsible. AC7: technical availability of at least 92 percent.
- Rejects caused by the equipment are the supplier’s as well, as long as the parts are inside the drawing tolerance. AC7: first pass yield of at least 98.5 percent.
- Organizational stops such as missing material or missing people belong to the plant. They are logged separately and do not count against the supplier.
Multiplied, 92 percent and 98.5 percent give 90.6 percent. With organizational losses of about 6 percent, the plan of 85 percent is realistic.
Plan the volume over the years of use
Ask sales for the lowest and the highest volume of the next eight years, not only for the first year. A cell that fits the launch volume exactly is too small in year three. A cell built for the best case costs money from the first day.
Lessons from the last purchase
The last input for the need costs nothing: the lessons from the previous project. For AC7 the team listed what the old cell had made harder than necessary. Feeders that could only be filled when the cell stood. A screw station nobody could reach. No list of wear parts.
Each lesson later became a requirement with a number. That is the cheapest way to avoid paying for the same mistake twice.
What you have at the end
One signed page: product and variants, 432,000 units per year, two shifts, a specified cycle time of 25.0 seconds, technical availability of 92 percent, first pass yield of 98.5 percent, 1,800 good units per day, the place in the layout, a budget of EUR 1,200,000 and the date of the first production day.
With this page, the team can start to collect requirements. Without it, every requirement is an opinion.