SUPSupply & productionStop 7 of 12
Scheduling & sequencing
The order of jobs on each line, hour by hour, and the simple rules that decide what runs next.
Read atChange any time
Planning decides how much to make this week. Scheduling decides exactly what runs on which line, in which order, starting at what time:
8:00–10:00 Shampoo A · 10:00–10:30 changeover · 10:30–13:00 Shampoo B · 13:00–14:00 Shampoo C
The order matters because jobs compete for the same machines and each one has a due date. Simple dispatching rules pick the next job, and each rule optimizes something different:
- FIFO (first in, first out): fair and simple, but blind to due dates and job length.
- EDD (earliest due date): run the most urgent job first. It minimizes the worst lateness.
- SPT (shortest processing time): run the quickest job first. It minimizes the average time jobs wait, but long jobs can sit for ages.
Real lines add complications. Changeover time often depends on the sequence: going from a light shampoo to a dark one is quick, while the reverse needs a full clean. Good schedules group similar products to cut changeovers. Then there are breakdowns, material shortages, shift patterns and maintenance windows.
That’s why large plants use finite-capacity scheduling or APS (advanced planning and scheduling) software, which respects capacity and sequence rules and recalculates when something goes wrong, instead of assuming infinite capacity the way basic MRP does.
Words to use
- Dispatching rule
- A simple rule for which job runs next, such as FIFO, EDD or SPT.
- Makespan
- Total time from the start of the first job to the end of the last.
- Sequence-dependent setup
- Changeover time that depends on which product ran before.
Four jobs, three rules
Jobs arrived in the order A, B, C, D.
| Job | Processing time | Due at hour |
|---|---|---|
| A | 6 | 8 |
| B | 2 | 6 |
| C | 8 | 18 |
| D | 3 | 15 |
| Rule | Sequence | Average completion time | Late jobs | Worst lateness | Total lateness |
|---|---|---|---|---|---|
| FIFO | A, B, C, D | 12.25 h | 2 | 4 h | 6 h |
| SPT (shortest first) | B, D, A, C | 9.25 h | 2 | 3 h | 4 h |
| EDD (earliest due first) | B, A, D, C | 10.0 h | 1 | 1 h | 1 h |
SPT gives the shortest average completion time; EDD gives the smallest worst lateness and, here, the fewest late jobs. FIFO is the easiest to explain and the worst on every measure.
These results hold in general for a single machine: SPT always minimizes the average time jobs spend in the system, and EDD always minimizes the maximum lateness.
The critical ratio rule
A dynamic rule used in many shops:
A ratio below 1 means the job will be late even if started now; around 1 it is on the edge; well above 1 it has slack. Each time a machine frees up, it takes the job with the lowest critical ratio, and the ratios are recalculated as time passes.
Beyond one machine
- Two machines in series: Johnson’s rule gives the shortest total time (makespan). Jobs whose first-machine time is shorter than their second go first, in increasing order of first-machine time; the rest go last, in decreasing order of second-machine time.
- Many machines and routes (a job shop): no simple rule is optimal; scheduling software combines rules, heuristics and optimization.
Why scheduling is hard
With 10 jobs on one machine there are 3.6 million possible sequences; with 20 jobs, more than the number of seconds since the universe began. Most realistic scheduling problems (job shops, sequence-dependent set-ups, multiple objectives) have no fast exact method, so practice relies on good rules, heuristics such as local search, and optimization solvers with time limits. A good schedule found in a minute usually beats a perfect one found tomorrow.
Sequence-dependent set-ups
When changeover times depend on the order of products, the sequence decides how much capacity is lost. Schedulers group products into families with short changeovers between them, follow a natural order (light to dark, low to high viscosity, allergen-free first), and use fixed product wheels where demand is stable enough.
Schedules break, so plan for repair
Machines break down, materials arrive late and rush orders appear. Two approaches:
- Robust schedules: leave slack in the right places, especially before the bottleneck and before key due dates, so small disruptions are absorbed.
- Rescheduling policies: decide in advance when to reschedule (periodically, or only after events above a threshold) and how much to change. Rescheduling everything at every event creates confusion on the floor; never rescheduling lets the plan drift from reality.
Keeping the next few hours frozen and repairing the schedule beyond that is a common compromise.
APS and MES
- APS (advanced planning and scheduling) creates schedules that respect finite capacity, set-up rules and material availability, and recalculates them quickly.
- MES (manufacturing execution system) runs the schedule on the floor: it dispatches jobs, records what actually happened, and feeds real start times, quantities and stops back to planning.
The loop between them, plan, execute, record, replan, is what keeps a schedule credible.
Measuring scheduling performance
Schedule adherence (did jobs run when planned?), on-time completion against due dates, average lateness, bottleneck utilization and changeover time per shift. Watch them together: a schedule can hit due dates by running the bottleneck inefficiently, or keep the bottleneck busy while orders run late.
Interview kit
Practise it out loud
Answer each question yourself before opening the model answer.
Scheduling turns the production plan into an exact sequence of jobs on each resource. Dispatching rules give quick answers: EDD minimizes maximum lateness, SPT minimizes average flow time, and FIFO is simple but ignores both. Real plants also sequence to minimize sequence-dependent changeovers and respect maintenance and shifts, often with finite-capacity APS tools that reschedule when disruptions hit.
Which rule would you use to minimize late orders, and which to minimize average waiting time?
Earliest due date minimizes the maximum lateness. Shortest processing time minimizes average flow time but can delay long jobs. In practice, plants often combine rules with changeover-aware sequencing.
Why does sequence matter so much in a paint or shampoo plant?
Because changeover time depends on the order. Light to dark colours or fragrances need little cleaning, while the reverse needs a full washout. Sequencing in a sensible order can recover hours of capacity every week without new equipment.
Connected stops
Where this idea feeds in, or gets its inputs from.
Sources and further reading
- Pinedo, M. L. Scheduling: Theory, Algorithms, and Systems. Springer.
- Hopp, W. J. & Spearman, M. L. Factory Physics. Waveland Press.
- Pinedo, M. L. (2016). Scheduling: Theory, Algorithms, and Systems (5th ed.). Springer.
- Johnson, S. M. (1954). Optimal two- and three-stage production schedules with setup times included. Naval Research Logistics Quarterly, 1(1), 61–68.