6.4 Just-in-Time & Lean Production Planning: Pull Systems, Kanban Sizing, Heijunka & SMED

6.4 Just-in-Time and Lean Production Planning

The NCEES specification lists JIT among production planning methods alongside capacity planning, materials planning (MRP), lot sizing, workforce planning, and aggregate planning. MRP plans by pushing time-phased orders into the shop. Just-in-time (JIT), developed as part of the Toyota Production System, controls the shop by pulling: nothing is produced until the next process needs it.

---

1. Push vs. Pull

Many plants combine them: MRP plans long-lead materials and capacity, and kanban controls daily execution on the floor.

CONWIP (constant work-in-process) is a simple pull variant: a fixed number of cards circulate for an entire line, and a new job may start only when a finished job releases a card. Little's Law ($WIP = TH \times CT$) shows why capping WIP caps cycle time.

---

2. The Seven Wastes

Taiichi Ohno's seven wastes are what JIT tries to remove:

1. Transportation: unnecessary moves of material.
2. Inventory: stock beyond immediate need, which hides problems.
3. Motion: unnecessary movement of people.
4. Waiting: idle people or machines.
5. Overproduction: making more or earlier than needed. Ohno considered this the worst, because it creates the others.
6. Overprocessing: work beyond what the customer values.
7. Defects: scrap, rework, and inspection.

(Many lean texts add an eighth waste, unused employee skill.)

---

3. Kanban Systems

A kanban ("signboard") is a signal, usually a card or an empty container, that authorizes production or withdrawal.

• Withdrawal (conveyance) kanban: authorizes moving a container from the supplying process to the using process.
• Production kanban: authorizes the supplying process to make one container to replace what was withdrawn.

Kanban rules (Toyota):

• The downstream process withdraws only what it needs.
• The upstream process produces only what was withdrawn.
• No defective parts are passed on.
• The number of kanbans is reduced over time to expose problems.

Number of Kanbans

$N = \frac{D \cdot L \cdot (1 + \alpha)}{C}$

where $D$ is the demand rate, $L$ is the replenishment lead time (waiting, processing, and transport for one container), $\alpha$ is a safety factor (often 0.10 or less in a mature system), and $C$ is the container size. Round up.

Example: A cell uses 600 brackets per day. A container holds 20 brackets. The time to replenish a container (wait, make, move) is 0.25 day, and the safety factor is 10%.

$N = \frac{600 \times 0.25 \times 1.10}{20} = \frac{165}{20} = 8.25 \Rightarrow 9 \text{ containers}$

The most brackets that can be in the loop is $9 \times 20 = 180$, which caps the WIP between the two processes. Cutting replenishment lead time to 0.15 day reduces the count to $\lceil 4.95 \rceil = 5$ containers.

---

4. Level Scheduling (Heijunka) and Mixed-Model Sequencing

Large batches of one model create surges in upstream demand. Heijunka levels both volume and mix.

Example: Daily demand is 400 of model A, 200 of model B, and 200 of model C, a ratio of 2:1:1. Instead of running all A, then all B, then all C, the line repeats the 4-unit sequence A–B–A–C 200 times per day. Every upstream process then sees steady demand for every component, and fewer kanbans are needed.

Leveling depends on short changeovers, which leads to SMED.

---

5. Setup Reduction: SMED

Shigeo Shingo's Single-Minute Exchange of Die (SMED) aims for changeovers under 10 minutes ("single-digit minutes"):

1. Observe and record every step of the current changeover.
2. Separate internal from external setup. Internal steps can only be done with the machine stopped; external steps can be done while it runs, such as staging the next die and tools.
3. Convert internal to external, for example by pre-heating dies or using standard die heights and fixture plates.
4. Streamline everything: quick-release clamps instead of bolts, eliminating adjustments, and parallel work by two people.

Example: A 60-minute changeover includes 25 minutes of work that could be done while the press runs. Moving that work to external setup cuts downtime to 35 minutes before any streamlining.

Why Setup Reduction Matters for Lot Sizing

The economic lot size is $Q^* = \sqrt{2DS/H}$, so $Q^*$ is proportional to $\sqrt{S}$. Cutting setup cost by 75% (to one quarter) reduces the economic lot size to $\sqrt{0.25} = 0.5$ of its former value. Smaller lots mean less inventory, shorter lead times, and easier leveling without higher total cost. JIT does not ignore the EOQ tradeoff; it lowers the setup cost that drives it.

---

6. Supporting Practices

• Takt-paced one-piece flow in cells sized to customer demand.
• Total productive maintenance, because pull systems have little buffer stock to cover breakdowns.
• Supplier partnerships with frequent, small, quality-certified deliveries.
• Jidoka (automation with a human touch): stop the line when a defect is detected so the problem is fixed at its source.
• Andon signals and visual management to make abnormalities obvious.