The Daily Fourth Shift + EDGEBIC Scheduling Workflow

A Fourth Shift shop's rhythm with EDGEBIC is a weekly load and a daily twenty-minute loop: import the week's orders on Monday, then every morning bring in yesterday's actuals, run the scheduler once, and publish the dispatch lists. The ERP is never touched. The interface is a file it already produces, which is what lets a decade-old install carry a modern finite capacity schedule without any risk to the system of record.

EDGEBIC by User Solutions is the current generation of the scheduling line whose Fourth Shift integration is the most documented in the family: the ERP vendor recommended the User Solutions scheduler to its own customer at Plastilite Corporation, and the integration ran Monday to Friday. The setup is covered in the complete Fourth Shift integration guide; the fields are listed in the Fourth Shift mapping reference. This post is what happens after that, every week, forever.

Two cadences, not one

Shops running a long-established ERP usually plan on a weekly beat, and the routine should respect that rather than fight it.

Weekly alone drifts by Wednesday, because no floor executes a plan exactly. Daily-only loading adds export work without adding information, since the order file rarely changes much between Tuesday and Wednesday. The pair works.

Monday: the load

Export open orders from Fourth Shift, run the saved orders mask, read the counts. A typical Monday reads something like Created 46 · Updated 0 · Reused 180 · Failed 0: 46 genuinely new orders, 180 already known and deliberately untouched.

Reused is the healthy outcome, not a failure to act. The default behavior for a record that already exists is to leave it exactly as it is, and that is what makes re-importing the same export safe every week for years. A non-zero Failed count is a question, not a crisis: the per-run log records every row with the exact reason, and the reason is nearly always a blank required cell or a date in a format the row could not parse.

Then the rule that governs everything: imports change data, never the plan. Newly imported orders sit as unscheduled demand until you run the scheduler. Nothing lands on the Gantt on its own.

Run the scheduler, confirm the counts in the dialog that states how many jobs are being scheduled for the first time and how many are being rescheduled, and let it finish. Publish the week.

Every morning: the loop

Four moves, in this order, and the order matters.

1. Actuals in. Yesterday's hours and pieces, one row per job, work center, and day. If your shop captures at a kiosk, this step is already done by the time you arrive. If hours come through the office, import the daily entry file. The import always overwrites the days the file carries, so importing the same file twice never double-counts and a corrected file simply fixes the earlier numbers. Days the file does not mention stay as they are.

2. Capacity reality. A machine down, a crew short, an unplanned closure. This is the input that never comes from an export and always changes the answer.

3. Run the scheduler. Select the jobs or accept the offer to include everything, read the confirmation dialog, let the run finish. Cancelling mid-run is always safe: nothing partial is kept and the previous plan stands.

4. Publish. Job View exports the schedule as a workbook with colored cells and a legend sheet; work center reports export with the column layout you saved.

Two guarantees make a daily cadence tolerable rather than nerve-wracking:

• Completed work never moves. An operation with an actual start and end is historical fact. No run, no setting, no mode shifts it.
• Scheduled jobs run on the routing they were planned with. Each carries a frozen snapshot, so re-importing a corrected routing changes future jobs and leaves work in progress alone.

Reading the result before you hand it out

Six columns answer the morning's real questions: status, start date, first operation start, job end, due date, and days late. Days late is a projection rather than a complaint: it means the current plan's end lands past the promise. A positive number today is a decision you get to make while options still exist.

The review worth ten minutes:

• Every positive days-late number is a decision. Re-sequence, add hours, or move the promise. Doing it on Tuesday costs less than doing it on Friday.
• Check the constraint. If the bottleneck is over capacity on any day, everything downstream is fiction. Bottleneck identification covers the diagnosis, and anchoring the schedule around the constraint is worked through in the bottleneck anchor example.
• Ask why anything moved. Jobs move for reasons: an operation that ran long, a capacity change, or higher-priority work inserted ahead.

What legacy exports need, once

Older extracts have habits, and the mask absorbs each one permanently rather than requiring a weekly cleanup.

That table is the difference between a routine and a chore. The point of a mask is that nobody reshapes a file by hand on a Monday morning.

Who does what

A routine that needs IT every morning does not survive its second month. Keeping IT out of the loop is deliberate.

Where the modern engine shows up

None of this adds daily steps, but all of it runs underneath the loop: work center groups that re-shop a machine pool on every reschedule, alternates for machines that can only run certain tooling, the sequence-dependent setup matrix, lot streaming with transfer batches, backward scheduling per job for just-in-time work (see forward versus backward scheduling), and mathematical optimization with a proven optimality gap. The heritage schedule at Plastilite was built on alternates and finite capacity; the same shop today would get all of that on top.

The EDGEBIC product overview maps the engine, and the ERP integration architecture shows how the same file-based rhythm serves any ERP. For the first week rather than the hundredth day, read a Fourth Shift site's first week with EDGEBIC; for the questions shops ask before starting, see the Fourth Shift integration FAQ.