Lean manufacturing and Six Sigma

In short

Lean manufacturing removes waste (anything the customer will not pay for) to shorten lead time, while Six Sigma removes variation (defects) using data and statistics; combined as Lean Six Sigma they form the standard improvement toolkit of modern industry. Lean grew out of the Toyota Production System; Six Sigma was developed at Motorola in 1986 and scaled by General Electric. The topic is examined in Industrial Engineering / Production Management papers, in GATE ME under industrial engineering, and is asked in almost every manufacturing placement interview (Maruti Suzuki, Tata Motors, Bosch, Mahindra, TVS, Ashok Leyland, Hero MotoCorp, L&T, Hindustan Unilever, Flipkart and Amazon operations). The methodology itself is standardised in ISO 13053 (Parts 1 and 2) and personnel competence in ISO 18404; quality management sits under ISO 9001, whose 2026 edition (published 16 September 2026) succeeds IS/ISO 9001:2015, the edition adopted by BIS, with a three-year transition.

Explained

Lean: the idea, the five principles and the seven wastes

• Lean = delivering exactly what the customer wants, when wanted, with the minimum resources. The term "lean production" was coined by John Krafcik (1988) and popularised by Womack, Jones and Roos in *The Machine That Changed the World* (1990), describing Toyota's system.
• Three categories of activity: *value-adding* (changes form, fit or function - customer pays), *necessary non-value-adding* (inspection, statutory testing, setup - cannot be removed today), *pure waste* (removable now).
• Five lean principles: 1) specify value from the customer's view; 2) map the value stream; 3) create flow by removing interruptions; 4) let the customer pull instead of pushing stock; 5) pursue perfection through continuous improvement.
• Seven wastes (muda - TIMWOOD): Transport (moving material), Inventory (stock waiting), Motion (operator movement), Waiting (idle time), Overproduction (making too much too early - the worst, since it creates the others), Over-processing (more work than specified), Defects (rework and scrap). An eighth waste - unused human talent/skill - is widely added.
• The 3 Ms: *muda* (waste), *mura* (unevenness in flow), *muri* (overburden on people or machines). Levelling (heijunka) attacks mura; ergonomics and capacity balance attack muri.

Core lean tools

• 5S (workplace organisation): *Seiri* sort, *Seiton* set in order, *Seiso* shine, *Seiketsu* standardise, *Shitsuke* sustain. Audited on a scored checklist; the foundation for every other tool.
• Value Stream Mapping (VSM): draws material and information flow for one product family with cycle times, inventory, uptime and lead time; current-state map → identify waste → future-state map → action plan. Key metric: process cycle efficiency = value-added time / total lead time (often under 5% before improvement).
• Just-in-Time (JIT): produce only what the next process needs - the right part, right quantity, right time; requires reliable machines, short setups and quality at source.
• Kanban: a card or signal that authorises production or movement; it is the pull mechanism of JIT. Two-bin and e-kanban variants are common in Indian auto supply chains.
• Takt time and line balancing: takt time = available time / customer demand; station cycle times must be at or just below takt.
• One-piece flow / cellular manufacturing: U-shaped cells replacing functional layouts; cuts transport, inventory and lead time.
• SMED (single-minute exchange of die): separate *internal* setup (machine stopped) from *external* (done while running), convert internal to external, then streamline - target under 10 minutes; enables small batches.
• Poka-yoke (mistake-proofing): devices that prevent an error or detect it instantly - asymmetric fixtures, limit switches, part-present sensors. *Prevention* type is better than *detection* type.
• Jidoka / andon: machines and operators stop the line automatically on an abnormality; the andon cord/board calls help. Stopping the line protects quality instead of passing defects on.
• Heijunka (levelling): smoothing the mix and volume of production so upstream demand is even; often with a heijunka box and mixed-model sequencing.
• Kaizen: small continuous improvements by the people doing the work; kaizen blitz is a 3-5 day focused event. Contrast with kaikaku (radical change).
• Standard work: documented best-known sequence, cycle time and standard WIP - the baseline without which improvement cannot be measured.
• TPM (total productive maintenance): autonomous maintenance by operators plus planned maintenance, measured by OEE = availability × performance × quality.
• Gemba walk, visual management, hoshin kanri (policy deployment), A3 problem-solving report complete the system.

Six Sigma: concept, sigma level and the belt structure

• Statistical meaning: a Six Sigma process has specification limits ±6σ from the mean, allowing a 1.5σ long-term mean shift, which gives 3.4 defects per million opportunities (DPMO). Without the shift, ±6σ would be 0.002 DPMO - exam answers must state the 1.5σ shift assumption.
• Managerial meaning: a disciplined, data-driven, project-by-project method to reduce variation and defects, with financial benefit quantified for each project.
• Y = f(X): the output (Y, the CTQ) is a function of process inputs (Xs). Six Sigma finds and controls the critical Xs instead of inspecting Y.
• CTQ (critical to quality): customer need translated into a measurable, specified characteristic - the starting point of every project.
• COPQ (cost of poor quality): scrap, rework, warranty, inspection, lost goodwill - typically 15-25% of turnover in an unimproved process; the justification for projects.
• Belts: *White/Yellow* - awareness and team member; *Green Belt* - leads small projects part-time; *Black Belt* - full-time project leader, statistics and change management; *Master Black Belt* - trains and coaches, portfolio level; *Champion/Sponsor* - business leader who removes barriers and signs off savings.

DMAIC: the five phases and the tools of each

1. Define - project charter (problem, goal, scope, team, savings), SIPOC diagram (Supplier-Input-Process-Output-Customer), Voice of Customer → CTQ tree, high-level process map, Pareto to pick the problem.
2. Measure - operational definitions, data collection plan, check sheets, measurement system analysis (MSA / Gage R&R), baseline sigma level and DPMO, process capability (Cp, Cpk, Pp, Ppk), control charts to check stability.
3. Analyse - process maps and value analysis, cause-and-effect (Ishikawa/fishbone - the 6Ms: Man, Machine, Material, Method, Measurement, Mother-nature), 5 Why, FMEA with RPN = Severity × Occurrence × Detection, hypothesis tests (t-test, ANOVA, chi-square), correlation and regression, multi-vari study → *verified root causes*.
4. Improve - generate and select solutions, Design of Experiments (DOE) - full factorial 2^k, fractional factorial, response surface, Taguchi robust design with signal-to-noise ratio; pilot run, poka-yoke, lean tools; confirm improvement statistically.
5. Control - control plan, SPC charts (X̄-R, X̄-s, I-MR for variables; p, np, c, u for attributes), standardised work and training, mistake-proofing, audit and handover to the process owner, sign-off of savings; monitor for 3-6 months.
6. DMADV / DFSS (Design for Six Sigma): Define-Measure-Analyse-Design-Verify, used for a *new* product or process rather than fixing an existing one. Related: IDOV (Identify-Design-Optimise-Validate), QFD (house of quality), tolerance design.
7. PDCA (Plan-Do-Check-Act / Deming cycle) is the simpler parent of DMAIC; 8D is the automotive customer-complaint format (IATF 16949 suppliers in India use it daily).

Lean versus Six Sigma - and Lean Six Sigma

• Lean Six Sigma uses DMAIC as the framework and inserts lean tools in Improve: waste first (quick wins, low cost), then variation (where causes are hidden in data). Rule of thumb - if the cause is known and visible, run a kaizen; if unknown, run a DMAIC project.
• Advantages: lower cost and inventory, shorter lead time, higher quality and capacity without capital spend, engaged workforce, measurable financial return.
• Limitations: JIT with zero buffer is fragile to supply or transport disruption; heavy training cost and belt infrastructure; needs stable high-volume demand for full benefit; poor fit for one-off job-shop or R&D work; projects fail without management commitment and a working measurement system; over-focus on cost can hurt innovation and morale.

Indian standards, practice and related frameworks

• ISO 13053-1/-2 - Six Sigma DMAIC methodology and tool descriptions; ISO 18404 - competence requirements for lean and Six Sigma personnel. Belt certification in India is awarded by bodies such as ASQ, IASSC, CII and QCFI rather than under an Indian Standard.
• ISO 9001:2026 quality management (process approach, risk-based thinking; published 16 September 2026, succeeding IS/ISO 9001:2015 as adopted by the Bureau of Indian Standards (BIS), whose certificates remain valid to about September 2029); IATF 16949 for automotive suppliers, which mandates SPC, MSA, FMEA, PPAP and APQP - the AIAG core tools a Six Sigma project in an Indian auto plant must work within.
• Awards and bodies: CII-EXIM Bank Award for Business Excellence (EFQM-based), Deming Prize (several Indian firms - Sundaram Clayton, TVS Motor, Rane group companies among winners), QCFI (Quality Circle Forum of India, Hyderabad) for quality circles, NPC and the Indian Statistical Institute for training, ZED (Zero Defect Zero Effect) certification for MSMEs under the Ministry of MSME.
• Related quality concepts: TQM (company-wide, culture-led), Quality Circles (small voluntary groups), Kaizen, Business Process Reengineering (radical redesign), Theory of Constraints (Goldratt - identify, exploit, subordinate, elevate the bottleneck, repeat).
• Statutory limits still bind: line-balancing and shift design must respect the working-hours limits of the OSH Code, 2020 (in force 21 November 2025, which repealed the Factories Act, 1948): 8 hours a day, 6 days a week, overtime at twice the ordinary wage (ss.25-27), with State rules setting spread-over and overtime caps.

Key points

• Lean - eliminating waste to shorten lead time; Six Sigma - reducing variation to cut defects.
• TIMWOOD - transport, inventory, motion, waiting, overproduction, over-processing, defects; overproduction is the root waste.
• 5S - sort, set in order, shine, standardise, sustain: housekeeping that makes abnormality visible.
• JIT and kanban - produce only on a downstream pull signal; kanban cards control WIP.
• Takt time - available time per unit of customer demand; the drumbeat a line must match.
• SMED - convert internal setup to external to cut changeover below 10 minutes, enabling small batches.
• Poka-yoke - design so the error cannot occur (prevention) or is caught at once (detection).
• Jidoka - automatic stop on abnormality; quality built in, not inspected in.
• Heijunka - levelling volume and mix to remove *mura* (unevenness).
• OEE = availability × performance × quality; world-class benchmark about 85%.
• Six Sigma = 3.4 DPMO, assuming a 1.5σ long-term mean shift; 3σ ≈ 66,807 DPMO.
• DMAIC - Define, Measure, Analyse, Improve, Control, for an existing process; DMADV/DFSS for a new design.
• Cp measures spread only; Cpk includes centring - Cpk ≤ Cp always, and Cpk ≥ 1.33 is the usual customer requirement.
• FMEA RPN = Severity × Occurrence × Detection (each 1-10, so RPN 1-1,000); act on high severity first.
• DPMO = defects × 1,000,000 / (units × opportunities per unit); rolled throughput yield = product of all step yields.
• Little's Law - WIP = throughput × lead time: cutting WIP cuts lead time at the same throughput.
• Belts - Yellow (member), Green (part-time leader), Black (full-time leader), Master Black Belt (coach), Champion (sponsor).
• COPQ - cost of poor quality, often 15-25% of turnover; it funds the improvement programme.

Formulas and examples

Example 1 - Takt time, stations and balance delay. A two-wheeler assembly line works one shift of 8 h with 30 min of breaks; customer demand is 450 units/shift. Total manual work content is 2,400 s.

• Available time = (8 × 60 − 30) × 60 = 450 × 60 = 27,000 s.
• Takt time = 27,000 / 450 = 60 s/unit.
• Theoretical stations N = 2,400 / 60 = 40 stations.
• If the line actually uses 44 stations with a cycle time of 60 s: line efficiency = 2,400 / (44 × 60) × 100 = 2,400/2,640 = 90.9%; balance delay = 9.1%.

Example 2 - DPMO, sigma level and rolled throughput yield. A pump-assembly cell inspects 1,200 pumps; 42 defects are found, with 5 inspection opportunities per pump. The process has four steps with yields 98%, 95%, 99% and 97%.

• DPU = 42/1,200 = 0.035 defects/unit.
• DPMO = 42 / (1,200 × 5) × 1,000,000 = 42/6,000 × 1,000,000 = 7,000.
• From the sigma table, 7,000 DPMO lies between 4σ (6,210) and 3.9σ (8,198) → about 3.95σ, i.e. just under 4σ.
• RTY = 0.98 × 0.95 × 0.99 × 0.97 = 0.8941 ≈ 89.4%, against a final-inspection yield that would look like 96-97% - the gap is the hidden factory of rework.

Example 3 - OEE on a CNC machine. Planned production time 480 min; breakdown 40 min, changeover 20 min; 300 parts produced, ideal cycle time 1.2 min/part; 12 parts rejected.

• Run time = 480 − 60 = 420 min → Availability = 420/480 = 0.875.
• Performance = (1.2 × 300)/420 = 360/420 = 0.857.
• Quality = (300 − 12)/300 = 288/300 = 0.96.
• OEE = 0.875 × 0.857 × 0.96 = 0.7199 ≈ 72.0% (good count 288 × 1.2 = 345.6 min of value time out of 480 = 72.0%, which confirms it).

Example 4 - Process capability and centring. A shaft diameter is specified 25.00 ± 0.05 mm. The process gives μ = 25.02 mm, σ = 0.012 mm.

• Cp = (25.05 − 24.95)/(6 × 0.012) = 0.10/0.072 = 1.389.
• Cpk = min[(25.05 − 25.02)/(3 × 0.012), (25.02 − 24.95)/(3 × 0.012)] = min[0.03/0.036, 0.07/0.036] = min[0.833, 1.944] = 0.833.
• Cp is acceptable but Cpk is well below 1.33: the spread is fine, the mean is off-centre. Re-centring to 25.00 mm would make Cpk = Cp = 1.389 with no reduction in variation - a tool-offset correction, not a new machine.

Example 5 - Kanban cards and Little's Law. A feeder line needs 1,200 brackets per 8 h shift, replenishment lead time 45 min, container holds 50 brackets, safety factor 20%.

• D = 1,200/480 = 2.5 brackets/min; L = 45 min; α = 0.20.
• N = 2.5 × 45 × 1.20 / 50 = 135/50 = 2.7 → 3 cards (round up).
• WIP authorised = 3 × 50 = 150 brackets. By Little's Law, lead time = WIP/throughput = 150/2.5 = 60 min of inventory in the loop; cutting to 2 cards (100 pieces) would cut it to 40 min but leaves no safety cover.

Where it is used

• Automobile and auto-component plants: Maruti Suzuki, Toyota Kirloskar, Tata Motors, Bosch, Sundaram-Clayton and Rane run TPS-style pull systems, kanban with suppliers, andon lines and IATF 16949 core tools; Deming Prize winners in the TVS group are the Indian reference cases.
• Process and FMCG industry: Hindustan Unilever, ITC, Asian Paints, Dr Reddy's and Sun Pharma use Six Sigma for yield, batch-cycle and OEE improvement; pharma must align it with Schedule M GMP and validation requirements.
• IT, BPO and banking: Infosys, Wipro, TCS, Genpact, HDFC Bank and ICICI apply Lean Six Sigma to transactional processes - turnaround time, error rate, rework in claims and loan processing; Green/Black Belt certification is a stated promotion criterion.
• E-commerce and logistics: Amazon, Flipkart, Delhivery use takt, standard work, 5S and Little's Law for pick-pack-dispatch cycle time and warehouse layout.
• Job roles: industrial engineer, production engineer, continuous-improvement/CI engineer, quality engineer, Green/Black Belt project leader, operations consultant, supply-chain analyst; graduate-trainee roles in auto plants start with a 5S or kaizen project.
• MSMEs and public sector: ZED certification and cluster lean programmes of the Ministry of MSME; BHEL, HAL and Indian Railways production units run kaizen and TPM drives.

Interview questions

• Q: What is the difference between lean and Six Sigma? A: Lean attacks waste and lead time using flow and pull; Six Sigma attacks variation and defects using statistics within a DMAIC project. They are complementary, not alternatives.
• Q: Why is Six Sigma 3.4 DPMO and not 0.002 DPMO? A: Because a 1.5σ long-term shift of the process mean is assumed, so only 4.5σ of the 6σ margin is effective on the near side.
• Q: Which of the seven wastes is most important and why? A: Overproduction, because making more or earlier than needed directly creates inventory, waiting, transport, extra motion and hides defects.
• Q: Cp is 1.8 but Cpk is 0.9 - what is wrong and what will you do? A: The spread is good but the mean is off-centre; correct the setting or tool offset first, which needs no variation reduction and no new equipment.
• Q: What is takt time and how does it differ from cycle time? A: Takt time is the rate the customer buys (available time / demand); cycle time is the rate the process actually produces. Cycle time must be at or just below takt.
• Q: What is poka-yoke, with an example you have seen? A: A mistake-proofing device - for example an asymmetric locating pin so a bracket cannot be loaded the wrong way round, or a part-present sensor that blocks the cycle if a washer is missing.
• Q: How is DMAIC different from PDCA and from DMADV? A: DMAIC is a structured, statistics-backed improvement of an existing process; PDCA is its simpler, faster parent cycle; DMADV (DFSS) designs a new product or process to meet CTQs from the start.
• Q: What would you do in the Measure phase before trusting any data? A: Run a measurement system analysis (Gage R&R) to confirm the gauge is accurate and repeatable - otherwise the baseline, capability and all later conclusions are invalid.
• Q: Give the formula for OEE and a typical world-class value. A: OEE = availability × performance × quality; around 85% is treated as world class, with availability 90%, performance 95% and quality 99%.
• Q: What is the risk of JIT with near-zero inventory? A: Any supply, transport or quality disruption stops the line immediately; the mitigation is supplier development, local sourcing, short setups and a small, deliberate buffer at the constraint.
• Q: How do you pick the root cause from a fishbone diagram with twenty causes? A: Narrow by data - Pareto, multi-vari or hypothesis tests - then verify the shortlisted causes experimentally; a cause is only a root cause when turning it on and off changes the output.
• Q: How did you quantify savings in your project? A: Convert the improvement into money - scrap and rework avoided, overtime or manpower released, capacity freed - and get the finance function and process owner to validate it before closure.

Common mistakes

• Quoting Six Sigma as "99.9997% good" without stating the 1.5σ shift assumption, or confusing DPMO with defective parts per million.
• Mixing up cycle time, takt time and lead time in line-balancing problems; forgetting to subtract breaks from available time.
• Treating 5S as cleaning only, and skipping *standardise* and *sustain* - the audit and discipline that make it last.
• Reporting Cpk > Cp, or computing capability from an out-of-control process - capability is meaningful only after the process is stable.
• Jumping to solutions in Analyse without verified root causes, and running a DOE before validating the measurement system.
• Confusing a control chart with a specification chart - control limits come from process data (±3σ of the statistic), specification limits from the customer; never plot specs on a control chart.
• Rounding kanban cards down, or ignoring the safety factor and container size, so the loop starves.
• Calling a kaizen event a Six Sigma project - no charter, no baseline, no control plan, so the gain is lost within months.

Related topics

Statistical quality control · Work study and ergonomics · Production planning and control · Inventory control and MRP · Maintenance and reliability engineering · Operations research basics