What is production scheduling? Production scheduling places confirmed work orders on specific machines and time slots, within the capacity a plant actually has. A plant can run a well-configured ERP and still watch its schedule fall apart by Monday’s second shift. A due-date sort gives you a queue. A schedule decides which job runs on which machine, and when. This guide covers the definition, the planning stack around it and seven places an ERP-generated schedule breaks.
What Is Production Scheduling?
The production scheduling process gives each released work order a machine and a time window. The sequence follows the item’s routing and respects finite capacity and changeover time. So the meaning of production scheduling comes down to sequence. Scheduling is an execution decision, taken after the materials decision is already made.
Each level of the planning stack hands the next one its input:
- S&OP and the production plan set volume by product family over months.
- The master production schedule (MPS) commits build quantities for end items by period.
- MRP explodes the MPS into material requirements and planned orders.
- Capacity planning, through CRP, turns those orders into a load profile per work centre.
- Production scheduling sequences the work orders on named machines and shifts.
- Shop floor control reports what actually ran and feeds it back.
The MPS works at end-item level. Aggregate planning belongs one level higher, in S&OP and the production plan.
What Is the Difference Between Forward and Backward Scheduling?
Under forward scheduling, the plan starts from today and books each operation as early as capacity allows, which gives the earliest date a job can finish. With backward scheduling, the plan starts from the due date and works back to the latest start that still meets it. Late starts keep material and work-in-progress off the floor until they are needed.
Real plants use both. A rush order gets scheduled forward. A make-to-order job with a fixed ship date often gets scheduled backwards.
The ERP default matters more than the theory. By default, ERP backwards-schedules from the due date, using fixed lead times and infinite capacity. The start date it calculates answers one question only. When would this order have to start if every machine were free? The trouble starts with that infinite capacity assumption.
What Does an ERP Actually Hand the Floor?
In most plants we work with, the floor gets dates and a dispatch list. The backward pass sets a start date for each order. CRP then shows the overload in a load profile. It never resolves it. The planner sees the press shop at 130% for week 32 and gets no sequence that removes the peak.
Levelling that load means moving orders by hand, in a spreadsheet.
SAP PP/DS, Oracle, D365 and Infor all offer finite scheduling. In our implementation work, many plants never configure these modules. Others configure them, then stop maintaining the routings, setup matrix and shift calendars the modules depend on, so the finite schedule drifts back toward the infinite one.
So the paper on the floor is a dispatch list, sorted by due date or by a priority flag a planner set by hand. Our guide to MRP covers the infinite-capacity logic in more depth.
Where Do ERP-Generated Schedules Fall Apart?
ERP-generated schedules fall apart at seven points on a typical floor. Three of them sit in the calculation itself: machine capacity, changeover sequence and lead time. The other four show up as work moves through the plant, from order release to material shortages. Every one of them has a mechanism behind it and a symptom the floor already knows. Two rows, over-scheduling and material availability, come from our implementation work.
| What breaks | Why the ERP output misses it | Symptom on the floor |
|---|---|---|
| Machine capacity | Backward scheduling assumes every machine is free | Two orders claim one machine, and a supervisor decides |
| Sequencing and changeover | Jobs run in due-date order, whatever the setup | Changeover hours pile up across the week |
| Lead time and routing | One average time per operation, with no queue | Promise dates drift the further out you look |
| Over-scheduling non-constraints | Every work centre is scheduled to the minute, with no buffers | One breakdown ripples across the schedule, and planners re-plan daily |
| Work order release and WIP | Orders release on planned start dates | WIP floods the floor, lead times stretch, throughput stays flat |
| Material availability | MRP treats planned receipts as available | Jobs start, stall for a missing part, then get expedited |
| Shop floor control and adherence | The dispatch list is printed once. Supervisors re-sequence on the floor, and those changes never flow back into the plan | The plan and the floor drift apart, and nobody measures by how much |
Machine capacity
Machine capacity is the first assumption to fail. A backward calculation places each operation in the window the due date implies. The calculation never asks what else is booked on that machine. Say two orders land on the same press on Tuesday morning. The system has no rule for choosing between them, so the supervisor on shift picks one. The choice may be sensible. Nobody writes it down. The next planner cannot learn from it, and the late order’s promise date moves with no reason in the system.
Sequencing and changeover
Changeover hours pile up fastest in a due-date sequence. Setup time often depends on which job ran before. The survey by Allahverdi and colleagues in the European Journal of Operational Research reviewed more than 300 papers on scheduling with setup times. The review treats sequence-dependent setups as their own class of problem. A due-date sort ignores that dependency. For example, a paint line might run four colour changes in due-date order when two groupings would have covered the same orders. Each extra change costs line hours, and sequencing by setup similarity would have saved them.
Lead time and routing
Lead time carries the same defect in a quieter form. The routing holds one average time per operation, set when the item was created and rarely revisited. At a loaded work centre, most of the elapsed time is queue. John Little restated his result in Operations Research on its 50th anniversary. Work in the system equals throughput multiplied by flow time, so with the completion rate held fixed, flow time rises with every job added to the queue. A fixed routing time cannot show that. Promise dates decay as the schedule runs further out.
Over-scheduling non-constraints
Scheduling every work centre to the minute looks like discipline, but it leaves no protective time anywhere in the flow for a breakdown or a quality hold to land. The first one pushes into every operation downstream. Planners then re-plan the whole floor daily. The bottleneck sets the plant’s pace. Other resources need protection from variability, and a minute-level plan gives them none. Plan the constraint in detail, and give the resources around it buffer time and looser slots.
Work order release and WIP
Order release makes the bottleneck problem worse. Orders release on the start date the ERP calculated, whatever the constrained resource can take that week. Work-in-progress (WIP) builds in front of it. Queues lengthen and flow time stretches with them, exactly as Little’s Law predicts. Throughput stays where the bottleneck left it. Before releasing material, check what each line can absorb that week. If a line is already overloaded, holding back the extra material keeps it off a floor that cannot use it yet.
Material availability
Material shortages break the schedule from the other side. MRP netting treats every planned receipt as available stock, but a planned receipt is only a date in a table. When the supplier slips, the work order keeps its slot, starts anyway and stalls halfway through its routing. The job then gets pushed aside by whatever expediting decides is urgent that afternoon, and the machine it held sits idle or half-loaded until someone finds the part. Check components before release.
Shop floor control and schedule adherence
Schedule adherence is the last failure anyone notices. Supervisors get a dispatch list printed once, then re-sequence on the floor, swapping machines and pulling jobs forward. None of those changes flows back into the plan. Nobody reconciles the two until a customer asks where an order is. Measure adherence per shift and per work centre. Say a plant reports 85% adherence overall. Inside that figure, its constrained work centre could sit at 55%, and that is the work centre to fix first. A plant-level average hides it.
A planning layer can close this gap. At each refresh from ERP transactions, Oritiq captures what changed on the ground, including re-sequenced jobs, machine swaps and what actually ran, and adjusts the plan. The next sequence starts from the floor’s real state, and last week’s printed list stops being the reference.
What Does a Real Production Schedule Require?
A production scheduling process that holds on the floor needs four things. The bottleneck comes first, because every other work centre runs at its pace. Sequencing, lead times and the refresh follow from it.
- Schedule the constraint in detail. Find the constrained resource and give it a full sequence, shift by shift.
- Sequencing by setup similarity. Group work so changeover hours fall out of the sequence itself. The setup matrix behind that logic is master data, and it needs an owner who keeps it current.
- Lead times that include queue. Update per-order lead time from actual completions at each loaded work centre. Promise dates then reflect the queue the order will join.
- A regular refresh. The schedule refreshes from ERP transactions on a set schedule. Adherence is measured against what ran, and the next sequence starts from the floor’s real state. Our production scheduling software guide and APS constraint checklist turn these four points into buyer questions.
Where Does Oritiq Fit?
Oritiq is the decision layer above the ERP. Its execution planning builds the plan for the constraint, sequencing work to minimise setups while meeting customer priorities. Production and procurement execution carries that sequence into release.
The plan is checked against finite capacity before it is published, and planners can override it where the floor needs to. Before material goes to the floor, planners can simulate the release against the load on each line, so excess material does not reach a line that is already overloaded. Capable-to-promise gives customers dates the plant can meet, and exception prioritisation flags material shortages that put scheduled orders at risk. Oritiq also captures what changed on the ground at each refresh from ERP transactions and adjusts the plan, so the next sequence starts from what actually ran.
FAQs on Production Scheduling
1. What is production scheduling?
Put simply, production scheduling is the step that turns released work orders into a sequence per machine and per shift. Take a plant with 12 machines on two shifts. Such a plant needs 24 short sequences each day, each one respecting changeover and capacity. A due-date list sorted across the whole plant cannot give a supervisor that.
2. What is the difference between production scheduling and MRP?
MRP works out which materials are needed and when, by exploding the master production schedule into component requirements with fixed lead times and an assumption of unlimited capacity. A schedule then decides the order in which machines run the resulting work orders. That step exists because capacity is finite.
3. Why does an ERP-generated schedule often fail on the shop floor?
Standard ERP logic schedules backwards from the due date, with fixed lead times and unlimited capacity, then prints a dispatch list. Capacity requirements planning reports the overload without fixing it, so the first double-booked machine breaks the sequence, and the supervisor re-sequences by hand from that point on.
4. What is the difference between forward and backward scheduling?
Choose by what the order needs. A rush order goes from the first free slot, to finish as soon as the line allows. A make-to-order job due in 10 days gets planned back from that date, so its components arrive just before the first operation. One plant often runs both in a single week.
5. What is the difference between a production schedule and a master production schedule?
The MPS commits build quantities for end items by period and feeds MRP. A production schedule sequences the released work orders on named machines inside those periods. The MPS might say 4,000 units in week 32. The schedule says which press runs which lot on Tuesday.
6. Does production scheduling require software, or can it be done manually?
A single-line plant with stable products can schedule on a whiteboard. Manual sequencing gets harder once changeovers matter and routings share machines. From then on, every breakdown forces a full re-plan, and a planner with 300 open work orders cannot redo it by hand each shift.
7. How is schedule adherence measured?
Adherence compares what the schedule said would run with what the shop floor reported. The usual measure is the share of scheduled operations completed in the scheduled period on the scheduled resource. Track it per shift and per work centre, so a plant-level average cannot hide the constraint.
What Should You Check in Your Own Schedule First?
Take last week’s dispatch list. Mark every job that ran on a different machine or a different shift from the printed plan. Count the marks. A high count means supervisors are building the real schedule by hand. Then compare one week of changeover hours with a sequence grouped by setup similarity.
Talk to the Oritiq team about a scoped pilot on your own work order data.
