Lean manufacturing, JIT, kaizen and TPS

Why it matters

The Toyota Production System (TPS) changed how every car plant in the world is run: build only what the next process needs, when it needs it, stop immediately when something is wrong, and improve a little every day. Indian OEMs and suppliers run on kanban, andon, 5S and kaizen, and interviewers expect graduates to explain them with numbers, not slogans. GATE asks conceptual questions on JIT and kanban and short numericals that combine takt time, Little's law and kanban sizing.

Key ideas

Lean thinking (five principles). Specify value from the customer's view; map the value stream (all steps from raw material to customer); make value flow without stops; let the customer pull production; pursue perfection by continuous improvement.

Waste. Toyota distinguishes:

• Muda — waste. The seven classic wastes: transport, inventory, motion, waiting, over-production, over-processing, defects (often remembered as TIMWOOD), plus unused employee talent. Over-production is considered the worst because it hides all the others.
• Mura — unevenness in the schedule or workload.
• Muri — overburden of people or machines. Activities are value-adding (changing the product in a way the customer pays for), necessary non-value-adding (inspection required by law) or pure waste.

The TPS "house". Foundation: heijunka (level, mixed-model production), standardised work and kaizen, built on stability (5S, TPM). Two pillars:

• Just-in-time (JIT) — produce at takt time, in continuous flow (one-piece flow in cells where possible), and use pull systems: a downstream process withdraws parts and the upstream process makes only what was taken. Inventory is deliberately reduced to expose problems ("lower the water to see the rocks").
• Jidoka (autonomation) — machines and people stop when an abnormality occurs, so defects are not passed on. Tools: andon cords/lights, automatic stop devices, poka-yoke (mistake-proofing — a jig that only accepts the correct part orientation, a sensor that counts bolts tightened). At the roof: best quality, lowest cost, shortest lead time, safety and morale.

Kanban. A card (or empty bin, or electronic signal) that authorises production or withdrawal of a fixed container quantity. Withdrawal (conveyance) kanbans move parts between processes; production kanbans authorise the upstream process to make a replacement container. Rules: no production without a kanban, only good parts are sent, the number of kanbans is reduced over time to expose problems. The number of kanbans fixes the maximum inventory in the loop.

Supporting tools.

• 5S — sort, set in order, shine, standardise, sustain.
• SMED (single-minute exchange of die) — convert internal setup (machine stopped) to external setup (done while running), then streamline both. Shigeo Shingo's work brought press-die changes from hours to under ten minutes, which made small batches and mixed-model production economic.
• Value stream mapping (VSM) — a current-state map of material and information flow with inventory triangles and process data, used to design a future state.
• Kaizen — small, continuous, worker-led improvements, often through PDCA cycles and kaizen events; contrasts with kaikaku (radical change).
• Genchi genbutsu — go and see at the actual place (gemba); 5 Whys for root cause.
• TPM — operators share basic maintenance (see the maintenance topic).

JIT prerequisites and risks. Stable demand (heijunka), short setups, reliable equipment, capable processes, nearby reliable suppliers with frequent small deliveries. With very lean buffers, disruptions (supplier fire, flood, chip shortage) stop the line quickly, so critical parts carry strategic buffers.

Formulas

T_takt = T_available / Q

• Takt time (s per unit); available production time (s) per period; Q = customer demand (units) in the same period.

N = D · T · (1 + α) / C

• N = number of kanbans (round up); D = demand rate (units/h); T = replenishment lead time — waiting plus processing plus conveyance (h); α = safety factor (–, policy value, often 0–0.2); C = container capacity (units).

WIP = TH · LT

• Little's law for a production line: WIP (units) = throughput (units/time) × flow time (time).

Lead time (days) = Inventory (units) / Daily demand (units/day)

• Used in VSM to convert inventory piles into days.

PCE = Value-adding time / Total lead time

• Process cycle efficiency (–).

Q* ∝ √S

• From the EOQ/EPQ formula: the economic batch size varies with the square root of setup cost (or setup time) S.

Worked examples

Example 1 (standard) — sizing a kanban loop. Given: a press-shop bracket is used at 960 per day over two 8 h shifts; replenishment lead time 2 h; container holds 20 brackets; safety factor 15 %.

1. D = 960 / 16 = 60 brackets/h
2. N = D·T·(1 + α) / C = 60 × 2 × 1.15 / 20 = 6.9 → 7 kanbans
3. Maximum inventory in the loop = 7 × 20 = 140 brackets.

Answer: 7 kanbans, at most 140 brackets in the loop.

Example 2 (GATE level) — value-stream lead time. Given: daily demand 800 parts; inventory found on the floor: 2,400 raw blanks, 800 parts between operations, 1,600 finished parts; four operations with cycle times totalling 180 s per part; plant works 16 h per day.

1. Lead time from inventory: 2,400/800 + 800/800 + 1,600/800 = 3 + 1 + 2 = 6 days
2. Total lead time in seconds = 6 × 16 × 3,600 = 345,600 s
3. PCE = 180 / 345,600 = 0.00052 = 0.052 %

Answer: lead time 6 days, but value is added for only 3 minutes — PCE ≈ 0.05 %. Cutting inventory is the main lever, exactly what Little's law predicts.

Example 3 — effect of SMED on batch size. Given: a die change takes 60 min and the economic batch is 1,200 pieces. SMED cuts the change to 6 min. If setup cost is proportional to setup time, find the new economic batch.

1. Q₂ = Q₁ × √(S₂/S₁) = 1,200 × √(6/60) = 1,200 / √10 = 1,200 / 3.162
2. Q₂ = 379.5 ≈ 380 pieces

Answer: about 380 pieces — less than one-third of the old batch, with average cycle stock falling in proportion.

Common mistakes

• Treating JIT as "zero inventory" or as pushing inventory onto suppliers; it means no more than what flow requires.
• Rounding the number of kanbans down; 6.9 kanbans means 7.
• Using only processing time for the kanban lead time; waiting and conveyance time count too.
• Confusing takt time (demand-driven) with cycle time (what the process achieves).
• Thinking jidoka means full automation; it means detecting abnormalities and stopping, with or without people.
• Describing kaizen as a one-off project; it is a daily habit, while large step changes are kaikaku.

For GATE ME

Expect one-mark conceptual questions on JIT, kanban, jidoka, poka-yoke, SMED, 5S and the seven wastes, and short numericals on takt time, number of kanbans and Little's law. Practise stating which tool removes which waste and keeping time units consistent in kanban calculations.

Quick check

1. Name the two pillars of TPS.
2. D = 100 units/h, lead time 1.5 h, container 25 units, α = 0. How many kanbans?
3. Which of the seven wastes is considered the worst, and why?
4. A line has throughput 30 units/h and WIP 90 units. What is the flow time?
5. What does SMED change first: internal or external setup?

Answers: 1. Just-in-time and jidoka. 2. 100 × 1.5 / 25 = 6. 3. Over-production, because it creates and hides the other wastes. 4. 3 h. 5. It converts internal setup to external setup.