Production flow describes how work moves through operations from release to completion. A factory may have fast individual machines and still deliver slowly because jobs spend most of their time waiting, moving or competing for constrained resources.
Cycle time, throughput and lead time
Cycle time usually describes how often a resource completes a unit under stated conditions. Throughput is the amount of good output completed over time. Lead time includes processing plus all waiting between release and completion. These measures answer different questions and should not be used interchangeably.
The constraint
The constraint is the resource, rule or dependency that limits system output. It may be a machine, inspection step, specialist skill, supplier, tool or information approval. Improving a non-constraint can increase local output while only building a larger queue.
Variability creates queues
Even when average capacity appears sufficient, variation in arrival time, processing time, quality and downtime creates waiting. Systems operated close to maximum utilization have little room to absorb variation, so queues can grow rapidly.
Work in process
Work in process keeps resources supplied and can protect against short interruptions. Too much WIP lengthens lead time, occupies space, complicates priority and delays discovery of defects. A useful WIP limit makes the state of flow visible.
Starving and blocking
A station is starved when it has no work available. It is blocked when completed work cannot move downstream. These conditions reveal dependencies beyond the station itself. Repeated starvation may indicate supply or scheduling problems; blocking may indicate a downstream constraint.
Improvement sequence
- Confirm where good output is actually limited.
- Protect the constraint from avoidable waiting and defects.
- Align upstream release with what the constraint can handle.
- Improve or add constraint capacity where justified.
- Recheck the system because the constraint may move.
Flow improvement is a system problem. The objective is not to keep every machine busy; it is to complete the right good work predictably.
Little’s Law as a learning relationship
For a stable system, average work in process is related to throughput and average flow time. The relationship explains why releasing more work without increasing completion can lengthen lead time. It is not a scheduling formula by itself, but it is a useful check on claims that more WIP will automatically improve output.
Temporary and shifting constraints
The limiting step can change with product mix, staffing, tooling or schedule. A constraint observed during one shift may not govern another. Reviews should therefore identify the period and product family, then compare queue, utilization, downtime and output evidence.
Capacity decisions
Adding equipment is only one option. Setup reduction, improved maintenance, better material supply, revised inspection or alternate routing may recover more effective capacity. The proposed change should be tested against downstream demand so the organization does not simply move the queue.
Observe the queue, not only utilization
A heavily utilized resource is not automatically the system constraint. The stronger evidence is persistent demand waiting for that resource and final output increasing when its effective capacity improves. A resource can show high utilization because it produces work earlier than needed, while the true constraint sits downstream.