Production managementReading time 12 min
Why Production Dates Keep Slipping
A promised date is only as good as the capacity and dependencies behind it. Work through the queues, setup time and material constraints that make a simple schedule fail on the shop floor.
01Why lead time grows
A basic MRP schedule calculates the release date by subtracting a fixed lead time from the due date. But lead time changes as queues grow. If work in process increases, a plan based on yesterday's lead time can create even more lateness.
- Dates are unreliable → release material early as insurance
- WIP grows → queues lengthen → actual lead time grows
- Someone lengthens the lead time parameter in the system
- Material is released even earlier → more WIP → back to step 2
Little's Law links average work in process, throughput and lead time: WIP = throughput × lead time. It does not mean every delay can be fixed by cutting WIP alone, but it shows why releasing more jobs into a constrained system can lengthen queues.
02Find the bottleneck
- An hour lost at the bottleneck is an hour lost by the whole plant, and it never comes back.
- An hour saved anywhere else is not a saving, it just produces more WIP.
- Which means: the bottleneck must not go down, must not make scrap, must not wait for material, and must have the fastest changeovers.
| Position | Strategy |
|---|---|
| Before the bottleneck | Hold a buffer so it never starves |
| The bottleneck itself | Run around the clock, maintain it first, inspect upstream |
| After the bottleneck | No buffer needed, but never let downstream block its output |
03Three scheduling approaches
| Method | Core idea | Where it fits |
|---|---|---|
| MRP (infinite capacity) | Backs off the BOM and lead times | Material requirements planning; not for sequencing operations |
| APS (finite capacity) | Accounts for real machine, labour and tooling availability | Discrete manufacturing with many machines and operations |
| DBR (drum-buffer-rope) | The bottleneck sets the beat; release is tied to it | Lines with one clear bottleneck |
| CONWIP | Holds total WIP at a fixed level | Changing product mix, where the bottleneck moves |
| Kanban | Pull replenishment | Stable demand, repetitive production |
Drum-buffer-rope schedules releases to the bottleneck. The bottleneck sets the pace, a buffer protects it from upstream variation, and the release signal limits work entering the system. The goal is steadier flow, not simply fuller machines.
04Inputs a schedule needs
- Routings, which machine each operation runs on, in what order, with what alternates
- Standard times, run time and setup time kept as separate numbers
- Machine calendars, shifts, maintenance, holidays
- Tooling and fixtures, the second resource everyone forgets
- Operator skill matrix, not everyone can run every machine
05Where to start
- Find the real bottleneck, look at where WIP piles up, not at the utilization report
- Measure setup times and record them separately from run time
- Cap material release, set a WIP ceiling and stop releasing when it is hit
- Put a buffer ahead of the bottleneck and watch how fast it gets consumed
- Use SMED to cut changeover so small batches become viable