The plan that is always right, on paper
Ask most systems when a job will be finished and they do some simple arithmetic: add up the hours on the routing, count forward from today, and give you a date.
The trouble is that they do this for every job separately. Each job is planned as though it had the shop to itself. Put twenty jobs through the same five-axis mill in the same week and, on paper, they all finish on time. On the floor, nineteen of them are waiting.
This is called infinite capacity planning, because it quietly assumes each machine has no limit. It is the reason so many planners keep a whiteboard next to a system that cost a great deal of money.
What finite means
A finite capacity plan accepts what everybody on the shop floor already knows: a machine can do one job at a time, and only while somebody is there to run it.
So it places each operation on a real machine, in a real shift, after whatever is already in the queue. If the mill is full until Thursday, the next job on it starts on Thursday, and its following operations move with it.
The dates that come out are less flattering. They are also true, which makes them the only ones worth giving a customer.
Three things a good plan respects
Your calendar. Shifts, short Fridays, shutdowns, a machine booked out for maintenance. Capacity is the hours you actually work, site by site.
Your routing order. Operation 30 cannot start until operation 20 has finished, unless you have decided to move parts on in smaller batches. Work sent out to a supplier is away for calendar days, not shift hours, and comes back on a due date.
What is really left to do. This is the one most plans get wrong. They subtract the hours booked from the hours estimated. But a job that has used ten of its twelve hours and made half its parts does not have two hours left. What is left is the parts still to make, multiplied by how long each one takes. Count parts, and the plan stays honest as the job runs.
What you can do with a plan you believe
Once the plan reflects the floor, it becomes useful for decisions, not just for reporting:
- See late jobs early. You find out on Monday that a job will miss Friday, while there is still time to move work, run overtime or ring the customer.
- See why. Late because of a machine, a supplier or missing material are three different problems with three different fixes.
- Spot the overload weeks out. If one work centre is heading for 110% in three weeks' time, that is the moment to subcontract, not the week it happens.
- Batch the repeats. When the same part is due twice in a fortnight, running the jobs together saves a setup.
- Quote with confidence. A delivery date on a quote is a promise. It helps if the plan agrees with it.
Why it needs the shop floor
A plan is only as good as what it is told. If operators book work at the end of the week, the plan spends the week guessing. When good parts and scrap are entered at the machine as they happen, the plan always knows how many are left, and re-plans itself without anybody rebuilding a spreadsheet.
The short version
Infinite planning tells you when a job would finish if it were the only job you had. Finite planning tells you when it will finish, given everything else. Only one of those is a date you can put your name to.