The planner looks at the machine schedule for next month. CNC-1 looks 60% booked. CNC-2 looks 55% booked. That feels like there is capacity. A customer wants a rush job — 60 hours of CNC time, divided between the two machines. The planner says yes, we can fit it.
He squeezes the job in. But he scheduled it based on the assumption that jobs will run at standard time. Three of the jobs booked are new jobs the planner has never run. They will be slower. Two jobs are running at a machine that just got a new spindle, and that spindle is not dialed in yet. The machine will run slower.
By week three of the month, the schedule is a wreck. Everything is delayed. The rush job got pushed. The machines are working double shifts just to keep pace. The operator is threatening to quit because he is exhausted.
Why static capacity planning fails
A schedule says a job is 10 hours on CNC-1. That is the standard time. But it assumes the operator is experienced with that job, the machine is running well, and there are no surprises. Reality is not standard.
An experienced operator runs the job in 10 hours. A new operator runs it in 13. A machine that is running well does 10. A machine with a worn spindle does 12. A job with a tricky first setup might add 2 hours of learning.
You cannot use the same schedule for a machine that is new, a machine that is old, and a machine that is average. But most planners do, because they do not have a way to track the actual times. They use standard times and they hope.
You are flying blind
You do not know if CNC-1 is really running at standard time. You do not know if the new operator is slowing down, or if the machine is slowing down, or if the job is harder than standard. You assume everything is fine until it is not.
By the time you notice a machine is running slow, four jobs are overdue. By the time you notice a new operator needs more mentoring, the week is chaos.
What real capacity planning looks like
Real capacity planning is based on actual data, not standard times. How long does CNC-1 actually take, with this operator, on a Tuesday? How much longer does a new job take versus a familiar job? How much does the spindle temperature affect run time?
With that data, you can forecast accurately. You can spot the bottleneck forming. You can tell a customer yes or no based on actual capacity, not hope.
- You know which machine is the constraint next month, not just today
- You adjust schedules before the constraint hits, not after jobs are overdue
- You can quote delivery dates based on actual capacity, not best-case times
- Your machines are loaded optimally, not just visually full
The trick is closing the loop fast enough that the data is useful. A machine that finishes a job tells you the actual time, and that feeds into the next job's scheduling. Not at the end of the month in the ERP. Right now.