From Empty Plant to First Schedule: An EDGEBIC Greenfield Walkthrough

Setting up scheduling software from scratch takes less data than most planners expect: one shift calendar, a handful of work centers, a routing, and one order are enough for a real finite capacity schedule. This walkthrough follows a fictional plant, Acme Bracket Co., from a completely empty database to its first committed schedule in EDGEBIC by User Solutions. Every number is real: you will see exactly why the cut step takes 4.5 hours, why milling spills into a second day, and why quality control starts at 10:15 rather than 10:00. It is one of the worked examples we publish so you can judge the scheduling engine by its arithmetic, not its adjectives.

If you already run RMDB, the concepts here will feel familiar; the upgrade path is documented separately. If you are new to the category, start with what production scheduling is and come back.

The Cast: One Plant, One Product, One Order

Acme Bracket Co. opens its doors with nothing in the system. By the end of the walkthrough it has:

Three details matter more than they look:

1. The shift yields 7 working hours, not 8. EDGEBIC subtracts the break: 16:00 minus 08:00 minus 1:00 = 7.0 hours per weekday. Weekend days carry zero hours, so the plant is simply closed on Saturday and Sunday.
2. Milling has 2 machines. Effective daily capacity is 7 x 2 = 14 hours, twice what Cutting offers. The instance count is the one field that multiplies capacity, and you will see it decide how the milling step splits.
3. The holiday is plant-wide. It is not tied to any single work center, so on Day 4 every work center has zero capacity.

Step 1: Enter Master Data in Dependency Order

Setting up scheduling software from scratch goes fastest when you follow the natural dependency chain, because each record references the one before it:

1. Unit of measure. Every product requires one; the system rejects a product without it.
2. Customer. Global Mounts Ltd, with a code and a contact email.
3. Shift. The weekly calendar is stored once and shared. Any work center flagged to use global shifts links to it automatically.
4. Holiday. Entered once with no work center attached, which makes it plant-wide.
5. Work centers. Cutting, Milling, and Quality Control, each set to use the global shift. Creating them links each one to the Day Shift behind the scenes.
6. Products. The raw steel plate first (with a manual opening receipt of 10 pieces into the inventory ledger), then Widget A as a make-to-order finished good.
7. Routing. Built in the drag-and-drop graphical routing editor: drag four steps onto the canvas, link the material step to the steel plate, assign each operation to its work center, type in the times, and draw the arrows Material to Cut to Mill to QC. Each arrow records which step feeds which.
8. The order. 20 pieces of Widget A for Global Mounts Ltd, due on Day 8.

The routing carries the numbers the whole schedule is built from:

Step 2: The Workload Arithmetic

Before the engine touches the calendar, you can compute each operation's total hours by hand. The formula is setup plus run time per piece times quantity:

Total touch time: 17.5 hours. The interesting question, and the reason finite capacity scheduling exists as a discipline, is where those hours land on real days with real shift limits.

Step 3: Run the Schedule and Watch the Cascade

The planner selects the order and clicks Schedule. The engine sorts the routing steps into dependency order (Material, then Cut, then Mill, then QC), and walks them one at a time. Each finished step hands its end time to the next step as the earliest possible start.

Material resolves instantly. Widget A is make-to-order, so the material step passes through at Day 1, 08:00 without consuming any work center time. Stock netting logic applies to make-to-stock products; this one just gates the start.

Cut: Day 1, 08:00 to 12:30. Cutting has 7 hours available on Day 1 and needs only 4.5. One machine, one block, done by 12:30.

Mill: Day 1, 12:30 to Day 2, 10:00. Here the finite capacity math earns its keep. Milling cannot start before 12:30 because Cut feeds it. From 12:30 to the 16:00 shift end there are 3.5 clock hours left, and with 2 instances that is 3.5 x 2 = 7.0 hours of work the engine can place on Day 1. It load-balances the work: 3.5 hours on each machine, both running 12:30 to 16:00. That leaves 11.0 minus 7.0 = 4.0 hours for Day 2. Both instances pick up at 08:00 and carry 2.0 hours each, finishing the step at 10:00.

Queue time pushes QC to 10:15. The routing puts a 0.25-hour queue after milling: cooling, moving, waiting for the inspector. The engine adds it after the step ends, so QC's earliest start is 10:00 plus 15 minutes = 10:15.

QC: Day 2, 10:15 to 12:15. Quality Control has 4.75 hours left in its Day 2 shift and needs 2.0. Done by lunchtime.

The Gantt chart tells the story at a glance:

           Day 1 (Mon)                  Day 2 (Tue)
           08:00        12:30     10:00 10:15   12:15
Cutting    [====== CUT 4.5h =====]
Milling              [= MILL 7.0h =][= MILL 4.0h =]
QC                                          [QC 2h]

The whole job runs Day 1, 08:00 to Day 2, 12:15: about 1.2 calendar days against a Day 8 due date, leaving 6.82 days of float. The holiday on Day 4 never comes into play because the work finishes before it, but had milling run long, the engine would have skipped Day 4 entirely and resumed on Day 5.

Step 4: What Gets Committed, and What the Screens Show

When the run completes, EDGEBIC commits the result in one transaction. The order flips to Scheduled. One production schedule record spans the job (Day 1, 08:00 to Day 2, 12:15), and three operation records sit underneath it: one each for Cut, Mill, and QC. The material step commits no operation row because it consumed no capacity. Under the milling operation sit four resource allocations (2 machines x 2 days), which is how the system remembers exactly which instance carried which hours on which date.

Three screens confirm the result:

• The orders grid shows the order with a Scheduled badge and its Day 1, 08:00 start.
• The Job Gantt shows three colored bars in the Cutting, Milling, and Quality Control rows, with the milling bar visibly spanning two days.
• The work center schedule view for Milling shows both instances loaded on the Day 1 afternoon and the Day 2 morning.

The dashboard's first KPI cards are already meaningful: 1 scheduled order, 1 on-time order, Cutting at roughly 64 percent utilization for the day (4.5 of 7 hours), Milling at roughly 79 percent (11 of 14 effective hours), and QC nearly idle at 14 percent.

What This First Run Proves

A greenfield walkthrough is a trust exercise. This one proves six specific behaviors you can verify yourself:

1. The shift calendar is arithmetic, not decoration. 7 net hours per weekday, zero on weekends, and the engine never places a minute outside them.
2. Instance counts multiply capacity. Milling's 2 machines offer 14 hours a day against Cutting's 7, and the two-day milling split follows from the 12:30 handoff rather than from any shortage of machines.
3. Dependencies cascade. Cut's 12:30 finish becomes Mill's earliest start; Mill's 10:00 finish plus queue becomes QC's 10:15 start. Nothing is hand-placed.
4. Queue time is modeled explicitly. The 15-minute gap between the milling bar and the QC bar is your routing data, visible on the timeline.
5. Holidays block capacity automatically. Plant-wide dates disappear from every work center's calendar.
6. Multi-instance work centers load-balance. The engine splits milling hours evenly across both machines each day so the step finishes as early as the capacity allows.

Variations Worth Trying on Day One

The fastest way to learn a scheduling engine is to change one number and predict the outcome before rerunning. Three variations from this same data set:

Give QC a second instance. The 2-hour inspection splits into 1 hour on each of two instances, and the job's finish moves from 12:15 to 11:15. One field changed, one hour saved.

Move the holiday onto Day 2. Milling's second-day allocation is now blocked. The engine searches forward, places the remaining 4.0 hours on Day 3, and QC follows. Total span grows by about a day, and no bar ever overlaps the holiday.

Raise the quantity to 100. Milling now needs 1.0 + 0.50 x 100 = 51 hours. At 14 effective hours per day that is roughly 3.6 working days, and the run will straddle the Day 4 holiday, which the engine skips. You can watch a single order occupy a work center for a week and see exactly why.

Each variation is a one-field edit followed by a reschedule, which is precisely the habit that makes a scheduling system stick: change, predict, verify.

From First Schedule to Full Operation

This walkthrough stops at the first committed schedule, but the same plant data carries forward into every other scenario we have documented. The next natural step is the full order lifecycle: a quote that becomes an order, gets scheduled, collects shop-floor actuals, and ships on time. When machines break, the breakdown reschedule walkthrough shows how logged work survives a replan untouched. When the next new part arrives instead, onboarding a new product from a blank routing to a first scheduled job repeats this sequence for one product rather than a whole plant.

User Solutions has been walking manufacturers through exactly this kind of setup since 1991. The company's tools have scheduled work for the US Navy, GE, BAE Systems, and Cummins, and the documented Plastilite implementation went from Fourth Shift data to a working RMDB schedule in five days. EDGEBIC is the next generation of that lineage, and if your data already lives in an ERP or a spreadsheet, the import-export masks mean you rarely type master data by hand at all.

Ready to run this walkthrough on your own products and work centers? Contact US for a demo and bring one routing: we will build your first schedule with you.