How to Set the Priority of Alternate Work Centers in EDGEBIC

Priority on an alternate work center row in EDGEBIC by User Solutions is a tie-breaker, not a command. The alternate that can finish the operation soonest is selected first, and priority only decides between options that would finish at the same time. Understanding that order of precedence is the difference between priority working as expected and priority appearing to be ignored.

Every task in this library is mapped on the EDGEBIC how-to hub. For why the schedule reasons about completion time at all, see finite versus infinite capacity scheduling.

Before You Start

• The routing step already has two or more alternate work centers configured. Priority is meaningless with one option.
• You know which machine you actually prefer and why. "Newer" is not a scheduling reason on its own.
• You have edit rights on routings.

Step 1: Open the Alternative Work Centers Grid

Open the routing for the product, select the step, and open its Alternative Work Centers grid. Each configured machine appears as a row carrying its type, speed factor, per-machine setup, priority, and active flag.

Step 2: Number the Rows

Set Priority on each row. Lower is preferred.

Save the routing, then re-run the schedule for the affected jobs. Like every routing change, priority is read when the engine builds the candidate list, so the existing plan does not move until a run happens.

Step 3: Know What Beats Priority

This is the part worth reading twice. The resolver sorts candidates by projected completion time first. Priority is the secondary sort.

So with Mill B at priority 1 and Mill C at priority 2:

• Both free, both finish Tuesday. Mill B is chosen. Priority decided.
• Mill B booked until day 5, Mill C free on day 3. Mill C is chosen. Priority was overruled by capacity, and that is correct: the job would otherwise sit for two extra days to satisfy a preference.

If you find yourself annoyed by that, the annoyance is usually pointing at a different problem. A preferred machine that never wins is a loaded machine, and the fix is capacity or shift coverage, not a lower number.

Step 4: Choose the Right Tool for a Hard Requirement

Priority expresses a preference. Three other tools express a requirement.

1. Clear the active flag on the machines you do not want. The candidate list then contains only the machine you do. Method in how to temporarily disable a parallel work center.
2. Pin the job. Pinning holds work on a chosen station across runs, covered in how to pin a job so the optimizer cannot move it.
3. Let started work hold itself. Once a job has logged actuals on a machine, that station is locked for the remaining hours automatically. Nothing to configure.

Step 5: Leave Room to Insert

Number in tens rather than ones if the machine list is likely to grow: 10, 20, 30. Adding a fourth option that should sit between the first and second is then a single edit instead of a renumber of every row. The values themselves mean nothing, only their order.

How to Check It Worked

• Re-run and read which station the step landed on. With everything free, it should be your priority 1 machine, and it should be the same machine on the next run too.
• Deliberately load the preferred machine and re-run. The step should move to the next option, which proves completion time is beating priority as designed.
• Compare against the resource replacement audit. If work is switching stations more than you expect, the resource replacement audit shows how often and where.

Common Mistakes

• Using priority to pin a machine. It cannot do that. A faster option always wins.
• Setting every row to priority 1. With no ordering, the tie-breaker has nothing to break and selection becomes arbitrary between equals.
• Confusing priority with speed factor. The factor changes how long the operation takes on that machine and therefore changes its completion time. Priority only sorts equals. A wrong factor is far more likely to explain a surprising choice than a wrong priority.
• Expecting priority to spread work. Alternates give the whole operation to one machine. If you want the hours split across several, configure the step as an independent parallel group instead.

See how EDGEBIC chooses between machines when it builds a plan on the EDGEBIC product page.