6.4 Pull Systems, Kanban, JIT & Setup/Cycle-Time Reduction
Key Takeaways
- The CQIA Body of Knowledge names set-up and cycle-time reduction, pull systems (kanban), just-in-time, and value stream mapping as specific lean tools used to reduce waste.
- A push system produces to forecast and creates inventory; a pull system produces only on a downstream signal, so work-in-process is capped by the number of kanban signals in circulation.
- A kanban is the authorization signal itself — a card, container, or electronic trigger — and no replenishment may occur without one.
- SMED (single-minute exchange of die) reduces changeover time by converting internal setup steps, which require the machine stopped, into external steps performed while it still runs.
- Shorter setups reduce economic batch size, which shortens lead time, cuts inventory, and exposes defects sooner — the quality benefit that matters most on the exam.
6.4 Pull Systems, Kanban, JIT & Setup/Cycle-Time Reduction
BoK entry III.A.2 does not simply say "understand lean." It enumerates the tools by name: set-up and cycle-time reduction, pull systems (kanban), continuous improvement (kaizen), just-in-time (JIT), 5S, value stream mapping, and error-proofing (poka-yoke). Sections 6.2 and 8.2 cover kaizen, 5S, poka-yoke, and the waste taxonomy. This section covers the remainder — the flow and changeover tools — which is where candidates who studied only the eight wastes get caught.
1. Push vs. Pull Production
This is the conceptual foundation for everything else in the section.
| Push system | Pull system | |
|---|---|---|
| Trigger to produce | A forecast or schedule issued in advance | An actual consumption signal from the next downstream step |
| Governing question | "What does the plan say we should build?" | "What did the customer just take?" |
| Inventory behaviour | Accumulates wherever the forecast exceeds real demand | Capped by design |
| Response to a forecast error | Overproduction or shortage, discovered late | Self-correcting; nothing is made without a signal |
| Defect exposure | Slow — defects sit in queues before discovery | Fast — small batches move quickly to the next step |
The connection to quality is direct and is the reason lean sits in a quality Body of Knowledge at all. Overproduction is the most severe of the eight wastes because it manufactures the other seven — it creates inventory, motion, transportation, and, critically, it delays defect detection. A push system that builds 2,000 units to a forecast may not discover a process shift until 2,000 defective units exist. A pull system with small batches surfaces the same shift after a handful.
2. Kanban: The Authorization Signal
Kanban (Japanese for signboard or card) is the physical or electronic signal that authorizes production or movement. The essential rule, and the one exam items test: nothing is produced or moved without a kanban. The card is not paperwork tracking the work — the card is the permission.
Common Kanban Forms
- Production kanban — authorizes the upstream process to make a specified quantity.
- Withdrawal (move) kanban — authorizes transporting a container from one location to the next.
- Two-bin system — the simplest implementation: when the first bin empties, it becomes the reorder signal while the second bin supplies demand.
- Electronic kanban (e-kanban) — a scan or system trigger replacing the physical card, common across supplier networks.
- Kanban square — a marked floor space; an empty square is the signal to produce.
Why the Signal Count Controls Inventory
Total work-in-process in a kanban loop is bounded by the number of kanbans in circulation multiplied by the container quantity. This makes WIP a deliberate management decision rather than an accident of scheduling. Removing kanbans from the loop tightens inventory and forces problems to the surface — the classic "lowering the water to reveal the rocks" image. Adding kanbans buffers a known problem while it is being solved.
A basic sizing relationship:
Note what drives the numerator: lead time. Shorten lead time and the required inventory falls proportionally. That is the bridge to setup reduction below.
3. Just-in-Time (JIT)
Just-in-time is the operating philosophy the pull system implements: produce and deliver the right item, in the right quantity, at the right time, and nothing more. Inventory is treated not as an asset but as a symptom concealing problems — unreliable equipment, long changeovers, unpredictable suppliers, and defects.
JIT's Non-Negotiable Prerequisites
JIT is frequently misapplied as an inventory-reduction program, and it fails predictably when its prerequisites are absent. It requires:
- Quality at the source. With no buffer stock, a single defect stops the next operation. This is why JIT and poka-yoke, jidoka, and stop-the-line authority always travel together.
- Reliable equipment. Unplanned downtime with no inventory buffer halts the line, which is why total productive maintenance accompanies JIT.
- Short, reliable changeovers. Small batches are impossible if a changeover takes four hours.
- Dependable suppliers. Frequent small deliveries require supplier quality and delivery reliability — the direct link to Section IV of the Body of Knowledge.
- Level demand (heijunka). Wildly variable demand cannot be met without buffers.
The exam trap: cutting inventory before fixing quality, uptime, and changeover time does not create JIT — it creates stoppages. Inventory reduction is the result of JIT, not the method.
4. Setup and Cycle-Time Reduction (SMED)
SMED — single-minute exchange of die — is Shigeo Shingo's method for reducing changeover time, with the target of completing a changeover in single-digit minutes (under ten). The entire method rests on one distinction:
| Setup type | Definition | Objective |
|---|---|---|
| Internal setup | Steps that can only be done while the machine is stopped | Minimize — this is the time the machine is not producing |
| External setup | Steps that can be done while the machine is still running | Move as much work here as possible |
The SMED Sequence
- Observe and document the current changeover, timing every step. Video is standard practice.
- Separate internal from external. Simply identifying which steps do not require a stopped machine — fetching tools, staging the next die, pre-heating, gathering paperwork — typically cuts changeover time substantially before anything is engineered.
- Convert internal to external. Pre-stage and pre-position materials; pre-assemble fixtures; use duplicate tooling so the next setup is built while the current job runs; pre-heat dies.
- Streamline the remaining internal steps. Replace threaded fasteners with quick-release clamps and one-turn devices; use standardized shims and stop blocks to eliminate trial adjustment; run parallel operations with two people; eliminate test runs by setting to a known position rather than adjusting to a measurement.
Why Setup Reduction Is a Quality Topic
Long changeovers force large batches, because the setup cost is amortized over the run. Large batches produce four quality consequences:
- Delayed defect detection. A process shift discovered after 5,000 units means 5,000 units at risk.
- Long lead time, which weakens the feedback loop between a process change and its observed effect.
- High inventory, concealing the problems JIT is designed to expose.
- Scrap on the trial-and-error setup itself, since adjusting to a measurement wastes parts that adjusting to a fixed position does not.
Halving changeover time roughly halves the economically justified batch size. That is why SMED is not a maintenance topic — it is a quality lever.
Cycle-Time Reduction Beyond Changeover
Cycle-time reduction also targets total elapsed process time, and applies as readily to an insurance claim or a hospital discharge as to a stamping press. The standard levers: remove non-value-added steps identified in the value stream map, eliminate waiting between steps, do steps in parallel instead of in sequence, and remove approval loops that add delay without adding control.
A plant reduces raw material and work-in-process inventory by 60% to implement just-in-time. Within three weeks the line stops repeatedly because a supplier's incoming defect rate of 3% now halts production immediately instead of being absorbed by stock. What went wrong?
During a SMED analysis of a 90-minute press changeover, a team finds that 35 minutes are spent retrieving tooling from the crib, locating the setup sheet, and staging the next die. Which SMED action applies to these 35 minutes?
In a kanban-controlled loop, what determines the maximum work-in-process inventory between two processes?
Why is reducing changeover time treated as a quality improvement rather than only an efficiency improvement?