Skip to content
← All work

Manufacturing · Lean Six Sigma DMAIC · Quality Analytics

Defect Reduction & Operational Efficiency

Gold Medal · 1st Rank

The Medium Casing line ran an 8% defect rate and 12% rework rate, costing ₹76,000 a month in cost of poor quality, with no analysis isolating which causes actually drove the loss.

8% → 2%
Defect rate
₹76K → ₹22K
Monthly COPQ
−40%
Cycle time
Gold Medal
Recognition

Overview

A Lean Six Sigma DMAIC study on the Medium Casing transformer line at Atlanta Electricals Pvt. Ltd. Real defect records were cleaned in pandas/NumPy, ranked with Pareto and 6M root-cause analysis, prioritised with FMEA, and converted into SOPs, Poka-Yoke controls and a 5S programme — cutting the defect rate from 8% to 2% and monthly cost of poor quality from ₹76,000 to ₹22,000.

Why the problem matters

Atlanta Electricals (est. 1988, Vallabh Vidyanagar) manufactures electrical transformers for 200+ clients across 19 states and four export markets. On the Medium Casing line, process variability, inconsistent QC and non-value-added activity were producing an 8% defect rate, 12% rework and ₹76,000 of monthly cost of poor quality — a direct, measurable margin leak.

Data & inputs

120 recorded defect events from the plant (YASH_DATASET.xlsx), cleaned and validated with pandas and NumPy. Each record carried fault code, 6M root-cause category, production section, COPQ in ₹, rework hours, detection phase and a 1–5 severity score — so the same event could be analysed financially, operationally and by risk.

Define — voice of customer to CTQ

Customer expectations were translated into measurable Critical-to-Quality characteristics across the three failure-prone stages, with stakeholders mapped from customer and management down to operators and supply chain.

  • Winding — wire-turn accuracy ±0.5%, uniform tension and layering, no shorts or opens
  • Insulation — dielectric strength at or above BIL spec, no moisture ingress, uniform potting
  • Assembly — correct bolt torque, no loose connections, proper gasket seating
  • Targets set upfront: 8% → 2% defects, 12% → 4% rework, 10 → 6 min cycle time

Measure — baseline the process

The baseline was quantified before any intervention: 120 defects recorded, 8% average defect rate, 12% rework, ₹76,000 monthly COPQ split as ₹40,000 defect cost and ₹36,000 rework cost. Workmanship (34%) and design issues (22%) dominated the categories. A TIMWOODS waste audit rated defects 'very high' and waiting, overprocessing and unused skills 'high', while a 5S audit scored the workplace 19/50 (38%).

Analyse — Pareto, fishbone and FMEA

Pareto ranking of 120 defects showed workmanship (41) and design issues (26) alone accounted for 56% of all defects, followed by material (22), assembly (17), testing (9) and others (5). A 6M fishbone exposed the interconnected causes — skill gaps in winding, late defect detection, humidity affecting insulation, dust and poor lighting. FMEA scored each failure mode on severity × occurrence × detection, isolating testing and insulation as the highest-RPN risks.

Improve — solutions targeting verified causes

Each intervention was tied to a confirmed root cause rather than a symptom.

  • Standard Operating Procedures displayed and standardised across shifts
  • Poka-Yoke error-proofing at the failure points flagged by FMEA
  • Operator training programme closing the winding skill gap
  • Full 5S cycle — sort, set, shine, standardise, sustain — with shadow boards, labels and daily audits

Results

Defect rate fell from 8% to 2% (−75%), rework from 12% to 4% (−66%), cycle time by 40%, and monthly COPQ from ₹76,000 to ₹22,000 (−71%) — roughly ₹6.5 lakh in annualised savings against a ₹3.6 lakh target. The 5S score rose from 19/50 to a sustained high-performance standard.

Control — holding the gains

Improvement without control is temporary. Statistical process control charts monitor the line, weekly internal 5S audits are scored and reviewed monthly by management, and quarterly refresher training is driven by the latest defect patterns. Future scope includes automated inspection, scaling DMAIC to HV winding and core assembly, and pushing toward 3.4 DPMO plant-wide.

Key learnings

The value came from segmenting data the way the plant actually operates and pairing every statistic with a financial number — a ₹76,000 monthly figure moved decisions that a defect percentage alone did not. Carried out as sole contributor under Mr. Viraj M. Tur, the study was awarded a Gold Medal and 1st Rank in class.