Capacity is net time divided by standard time. Both halves are easy to get wrong, and most plants get the first one wrong.
Net available time, step by step
Start with the shift and take away what is not available. In area M2 a shift is 480 minutes with 30 minutes of paid break, so 450 minutes remain. Two shifts give 900 planned operating minutes per station per day.
Then apply the utilization factor. Not one from a handbook and not last year’s, but the one measured by the work sampling study: 80 percent. That gives 720 net minutes per station per day, and 165,600 net minutes per station per year on 230 working days.
That last figure is the capacity of one single workplace, and it is the number every load calculation divides by. Using 900 instead of 720 makes a capacity statement 20 percent too optimistic, which is the single most common error in the whole subject.
Load per station
Multiply each standard time by the program and divide by the net minutes. Area M2 planned 52,000 units.
| Station | Required min per year | Load |
|---|---|---|
| S1 Cartridge pre-assembly | 121,680 | 73.5 % |
| S2 Housing preparation | 99,840 | 60.3 % |
| S3 Valve block assembly | 147,160 | 88.9 % |
| S4 Cover, torque and marking | 134,160 | 81.0 % |
| S5 Leak and function test | 174,720 | 105.5 % |
| S6 Pack and label | 55,640 | 33.6 % |
Only the test bench is above 100 percent, so it is the bottleneck and it decides the area. Area capacity is the net time of the bottleneck divided by its standard time: 165,600 divided by 3.36 gives 49,285 units a year against a program of 52,000. The gap of 2,715 units is worth EUR 114,030 of contribution margin.
Two cross-checks that cost nothing
The program of 52,000 units over 230 days is 226 units a day, and 720 net minutes divided by 226 gives a takt of 3.18 minutes. The bottleneck needs 3.36. The line cannot hold the takt, which is exactly what the queue in front of the test bench had been saying for two years.
The second check is simpler still: 720 net minutes divided by 3.36 is 214 units a day, and 214 over 230 days is 49,285. If the two checks disagree with the load table, something in the arithmetic is wrong.
The decision the times changed
With the old times, the slowest station was S3 at 3.60 minutes, the stated capacity was 46,000 units, and a request for a second valve block assembly station worth EUR 95,000 was on the table. It would have bought nothing, because S3 was never the constraint.
| Option | Investment | Capacity after |
|---|---|---|
| Do nothing | none | 49,285 |
| Second assembly station on S3 | EUR 95,000 | 49,285 |
| Parallel leak test on S5 | EUR 28,000 | 56,326 |
| Third shift on S5 only | Operating cost | 58,515 |
Running two leak tests at once with a second carrier set cut the test cycle from 2.65 to 2.23 minutes, so the station time fell from 3.36 to 2.94. Capacity rose to 56,326 units, the program was covered with 8.3 percent of headroom, and payback was 2.9 months. Beyond that the next constraint is S3 at 58,515 units, which is why the third shift buys no more than the carrier set does.
Climb the ladder before you buy
Rebalance first: station S6 runs at 34 percent, so moving the marking element off S3 costs nothing. Then overtime at the loaded stations for a short peak, then a shift only where it is missing, and only then equipment.
And ask the two questions before every capacity investment: is the station we are about to buy the one that limits the area, and is its time measured. In area M2 both answers were no, and that was worth EUR 95,000.