Executive Summary

Best scale manufacturing business, This case study examines the operational transformation of an Engineer-to-Order (ETO) pharmaceutical packaging machinery manufacturer operating in a classic High-Mix Low-Volume (HMLV) jumbled flow environment. Through application of Value Stream Mapping, Group Technology classification, Quality Function Deployment, and constraint-based scheduling methodologies, the organization achieved a 91.4% reduction in manufacturing throughput time (from 722 days to 62 days) while transitioning from negative profitability to 12% EBITDA margin.

The transformation centered on migrating from pure customization to a mass customization paradigm using product family-based platform architecture, thereby creating operational predictability within an inherently unpredictable HMLV environment where traditional capacity planning and line balancing methodologies were inapplicable.

Initial State Analysis: The HMLV Complexity Paradigm

Process Architecture and Flow Characteristics

Manufacturing System Classification:

• Layout Type: Functional/Process Layout (Machine Shop, Fabrication, Sub-assembly, Final Assembly cells)
• Production Control: Push-based MRP system with chronic nervousness
• Material Flow: Spaghetti flow with routing complexity index of 8.7 (industry benchmark: 3-5)
• Process Variability: Coefficient of variation (CV) > 2.5 for process times across work centers
• Queue Time Ratio: Non-value-added time constituted 88.6% of total throughput time

HMLV Characteristics Quantified:

• Product variants: 180+ unique configurations over 3-year analysis period
• Average batch size: 1 unit (pure make-to-order)
• Product mix Shannon Entropy Index: 4.8 (indicating extreme variety)
• Setup time to run time ratio: 3.2:1 (highly unfavorable)
• Routing variability: 47 different process routings identified

Capacity Planning Impossibility in Jumbled Flow:

Traditional capacity planning methodologies (capacity bills, resource profiles, RCCP) failed due to:

1. Stochastic Arrival Patterns: Order arrival followed non-Poisson distribution with high variance (CV = 2.8)
2. Heterogeneous Processing Requirements: Processing time variance σ²/μ² > 6 across critical work centers
3. Dynamic Bottleneck Migration: Constraint resources shifted based on product mix, violating steady-state assumptions required for capacity modeling
4. Infinite SKU Problem: Theoretical product configurations exceeded 10,000 combinations
5. Path Dependency: Process sequence dependent on design evolution, creating non-deterministic routing

Performance Baseline Metrics

Throughput Time Analysis:

• Total Manufacturing Lead Time (MLT): 722 days (mean), σ = 147 days
• Theoretical throughput time (all operations): 82 days
• Manufacturing Critical Path Ratio (MCPR): 0.114 (industry target: >0.6)
• Touch time percentage: 11.4% (industry target: >40%)

Detailed Time Breakdown (Little’s Law Application):

Using Little’s Law (WIP = Throughput Rate × Lead Time):

• Design Engineering WIP time: 128 days (average 3.2 design iterations in queue)
• Procurement queue time: 285 days (material unavailability, vendor lead time variability)
• Shop floor queue time: 247 days (average WIP: 18-22 machines)
• Rework and rectification cycles: 62 days

Constraint Analysis (Theory of Constraints – TOC):

Bottleneck identification revealed dynamic constraint migration:

• Week 1-8: CNC Machining Center (utilization: 127% theoretical – overloaded)
• Week 9-15: Sheet Metal Fabrication (design changes triggering rework)
• Week 16-24: Electrical Panel Assembly (component obsolescence)
• Week 25-32: Final Assembly (cascading delays, skill shortage)

Bottleneck Migration Causation Matrix:

Root Cause	Frequency	Impact on Throughput Time	Constraint Shift Probability
Engineering Change Orders (ECOs)	15-20 per machine	+45-68 days	0.35
Material Stockouts	8-12 instances	+35-52 days	0.28
Quality Escapes/Rework	6-9 instances	+22-38 days	0.22
Labor Unavailability	Variable	+12-28 days	0.15

Financial Performance Indicators:

• Annual Throughput: 5-6 machines (theoretical capacity: 24-28 machines given infrastructure)
• Throughput Accounting Metrics:
  • Throughput (T): ₹60 Crores revenue – ₹42 Crores TVC = ₹18 Crores
  • Operating Expense (OE): ₹22 Crores
  • Net Profit (NP): -₹4 Crores (losses)
  • Return on Investment (ROI): Negative
  • Inventory (I): ₹28 Crores (obsolescence: 40-45%)
• Cash Conversion Cycle (CCC):
  • Days Inventory Outstanding (DIO): 485 days
  • Days Sales Outstanding (DSO): 125 days
  • Days Payable Outstanding (DPO): 68 days
  • CCC: 542 days (unsustainable working capital requirement)

---

Root Cause Analysis: Systemic Pathologies In HMLV  Environment

Engineering Change Order (ECO) Proliferation

Quantitative Analysis:

• Average ECOs per machine: 17.3 (σ = 4.2)
• ECO classification:
  • Class I (Critical – functionality): 38% of total ECOs
  • Class II (Major – fit/form): 44% of total ECOs
  • Class III (Minor – cosmetic): 18% of total ECOs

Stage-Gate Violations:

• Design freeze protocol: Non-existent
• Customer requirement documentation completeness: 42% average (Kano model not applied)
• Design for Manufacturing (DFM) review compliance: 18%
• Design Failure Mode Effects Analysis (DFMEA): Conducted on only 12% of machines

Propagation Impact: Each ECO triggered cascading effects quantified through:

• Engineering Change Propagation Index (ECPI): 3.7
  • (Average number of components affected per change)
• Bill of Material (BOM) churn rate: 22% of components became obsolete per machine
• Procurement disruption: 68% of ECOs occurred post-material procurement initiation

Supply Chain Bullwhip in HMLV Context

Forrester Effect Amplification:

Demand signal variability amplification factor: 4.8 (Ratio of supplier demand variance to customer demand variance)

Contributing Factors:

1. Lead Time Demand Variability: σ_LTD = σ_demand × √(Lead Time) exacerbated by design changes
2. Batch Ordering: Economic Order Quantity (EOQ) models invalid for unique components
3. Shortage Gaming: Internal departmental demand inflation during scarcity
4. Price Promotions: Vendor minimum order quantities (MOQs) creating excess inventory

Inventory Pathologies:

• Obsolete Inventory: ₹11.2 Crores (purchased but unutilized due to design changes)
• Slow-Moving Inventory: ₹6.8 Crores (turnover ratio <1)
• Dead Stock: ₹2.4 Crores (vendor return refusal on customized components)
• Inventory Holding Cost: 28% annually (opportunity cost + storage + obsolescence)

Quality Cost Economics (PAF Model Analysis)

Prevention-Appraisal-Failure (PAF) Framework:

Category	Annual Cost (₹ Crores)	% of Sales	Industry Benchmark
Prevention Costs	0.4	0.67%	2-3%
Appraisal Costs	1.2	2.0%	3-4%
Internal Failure	3.6	6.0%	2-3%
External Failure	2.8	4.67%	1-2%
Total Quality Cost	8.0	13.3%	8-12%

Dominant Failure Modes:

• Dimensional non-conformance: 34% of rework instances (tolerance stack-up issues)
• Functional performance deviation: 28% (inadequate simulation/validation)
• Fitment issues during assembly: 22% (interface management failures)
• Material specification errors: 16% (supplier quality variability)

Workforce Utilization and Skill Matrix Deficiency

Labor Productivity Metrics:

• Direct Labor Efficiency: 54% (industry target: 75-80%)
• Value-Added Ratio (VAR): 0.31 (time in value-adding activities/total time)
• Overall Labor Effectiveness (OLE): 41%
  • OLE = Availability × Performance × Quality × Utilization

Skill Matrix Analysis:

Polyvalence Index (cross-training effectiveness):

• PI = (Actual skills)/(Required skills × Number of workers) = 0.38
• Target PI for HMLV: >0.75

Constraint Migration Due to Human Factors:

• Absenteeism coefficient of variation: 0.42 (unpredictable availability)
• Attrition rate: 18% annually in critical skills (welding, electrical integration)
• Knowledge tacitness: 73% of assembly knowledge undocumented (tribal knowledge)
• Training cycle time: 6-9 months for competency development. Best scale manufacturing business.

---

Transformation Methodology: Lean In HMLV Context

Phase 1: Extended Value Stream Mapping (Months 1-3)

Current State VSM – HMLV Adaptation:

Traditional VSM assumes product families with similar process flows. We adapted methodology for jumbled flow:

Product Family Clustering Using Group Technology:

Applied Modified Production Flow Analysis (MPFA):

1. Component-Machine Incidence Matrix: Created 850 components × 48 machines binary matrix
2. Similarity Coefficient Calculation: Jaccard Index for component commonality
   • J(A,B) = |A ∩ B| / |A ∪ B|
3. Hierarchical Cluster Analysis: Dendrogram construction using Ward’s linkage
4. Optimal Clustering: Determined using Silhouette Coefficient (s = 0.68, indicating good clustering)

Cluster Results:

• Family A (2-Fold Machines): 42% of total volume, 68% component commonality
• Family B (3-Fold Machines): 35% of total volume, 71% component commonality
• Family C (4-Fold Machines): 23% of total volume, 66% component commonality
• Cross-family commonality: 72% (revelation: significant platform opportunity)

Process Activity Mapping (PAM):

Classified all activities using ASME symbols with HMLV granularity:

Activity Type	Time (days)	% of Total	Value Classification
Operation (○)	82	11.4%	Value-Adding
Transportation (→)	19	2.6%	Non-Value-Adding
Inspection (□)	47	6.5%	Business Value-Adding
Delay (D)	523	72.4%	Non-Value-Adding
Storage (∇)	51	7.1%	Non-Value-Adding

Critical Path Method (CPM) Application:

Constructed network diagram for critical manufacturing path:

• Critical Path Duration: 82 days (operations only)
• Float/Slack in non-critical paths: 628 days average
• Schedule Performance Index (SPI): 0.114 (severe schedule inefficiency)

Constraint Identification Using Drum-Buffer-Rope (DBR) Precursor Analysis:

Work Center	Utilization	Queue Length	Queue Time	Constraint Frequency
CNC Machining	142%*	18 jobs	67 days	38%
Sheet Metal	118%*	12 jobs	43 days	28%
Assembly	96%*	8 jobs	38 days	22%
Electrical	87%*	6 jobs	28 days	12%

*Utilization >100% indicates overload and chronic backlog

Phase 2: Design for Mass Customization (Months 4-9)

Quality Function Deployment (QFD) – House of Quality:

Implemented four-phase QFD methodology:

Phase I: Customer Requirements → Engineering Characteristics

Deployed Kano Model analysis across 127 customer requirements:

• Must-be (Basic) Requirements: 42% – hygiene factors (GMP compliance, safety)
• Performance Requirements: 38% – linear satisfaction (throughput speed, changeover time)
• Excitement (Delighter) Requirements: 20% – innovation (remote monitoring, predictive maintenance)

Importance-Performance Matrix:

• Prioritized 23 critical engineering characteristics (CEs)
• Established target values with tolerance specifications
• Competitive benchmarking against 3 key competitors

Correlation Matrix Analysis:

• Identified 14 strong positive correlations (synergistic design opportunities)
• Identified 6 strong negative correlations (trade-off management required)
• Engineering complexity index: 3.2 (reduced from 7.8 through rationalization)

Phase II-IV: Product Planning → Process Planning → Production Planning

• Cascaded CEs to component specifications
• Defined critical-to-quality (CTQ) parameters
• Established process capability requirements (Cpk targets)

Modular Function Deployment (MFD):

Applied Axiomatic Design principles:

• Functional Requirements (FR): 34 primary functions identified
• Design Parameters (DP): Mapped to FRs maintaining Independence Axiom
• Modularity Index: Improved from 0.23 to 0.78 using Module Indication Matrix

Product Platform Architecture Development:

Configuration Management Framework:

Established three-tier architecture:

1. Platform Layer (70% commonality):
   • Base frame structure (welded construction, stress-analyzed for max loading)
   • Power transmission assembly (motor, gearbox, shaft, bearings)
   • Main control system (PLC, HMI, I/O architecture)
   • Safety system (guards, interlocks, emergency stops)
   • Pneumatic/hydraulic power unit
   • Material handling input/output conveyors
2. Module Layer (20% semi-customized):
   • Folding mechanism modules (2/3/4-fold cassettes – interchangeable)
   • Box size adjustment systems (servo-controlled positioning)
   • Feeding system variants (continuous, intermittent, precision)
   • Product detection and rejection modules
3. Peripheral Layer (10% fully customized):
   • Customer-specific automation interfaces
   • Special material handling fixtures
   • Unique process monitoring sensors
   • Cosmetic/aesthetic customization

Commonality Index Calculation:

CI = (Total common components) / (Total components in product family)

• Achieved CI = 0.72 (target: >0.65 for mass customization viability)

Design Structure Matrix (DSM) Optimization:

Created N×N matrix of component interactions:

• Resequenced design tasks to minimize iteration loops
• Identified highly coupled clusters requiring concurrent engineering
• Reduced design iteration cycle time by 68% through proper sequencing

Design Freeze Protocol – Stage-Gate Process:

Implemented rigid phase-gate methodology:

Gate 0: Opportunity Identification

• Market requirement document (MRD)
• Preliminary feasibility (technical, commercial)
• Go/No-go decision

Gate 1: Conceptual Design Review (CDR)

• QFD Phase I completion (Customer → Engineering requirements)
• Concept sketches, 3D CAD visualization
• Preliminary Design FMEA (pDFMEA)
• Deliverable: Customer approval on functionality concept
• Tolerance: Maximum 2 iterations allowed

Gate 2: Preliminary Design Review (PDR)

• Detailed 3D CAD models with complete BOM
• Finite Element Analysis (FEA) for structural validation
• Kinematic/dynamic simulation for motion systems
• Design for Manufacturing and Assembly (DFMA) review
• Detailed DFMEA with RPN prioritization
• Deliverable: Customer approval on detailed design
• Critical requirement: Design freeze declaration

Gate 3: Critical Design Review

• Final manufacturing drawings with GD&T specifications
• Process FMEA completion
• Prototype validation (where feasible)
• Supplier readiness assessment
• Deliverable: Release to manufacturing

Engineering Change Control:

Post-design freeze, implemented formal ECO process:

• Classification: Impact assessment (cost, schedule, performance)
• Approval authority: Tiered based on impact (minor: team lead, major: customer approval required)
• Cost recovery: Changes post-PDR billable to customer with 15-20% margin loading
• Schedule impact: Transparent communication of lead time extension, Best scale manufacturing business

Impact on ECO Proliferation:

• ECOs reduced from 17.3 to 2.8 per machine (83.8% reduction)
• Average ECO cost impact: ₹2.8 lakhs (now recoverable from customer)
• Schedule variance due to ECOs: 92% reduction

Phase 3: Mixed-Model Production System Design (Months 7-14)

Challenge: Implementing Flow in HMLV Jumbled Environment

Traditional lean principle of “continuous flow” seemingly contradicts HMLV reality. Solution: Hybrid production system combining:

• Cellular manufacturing for platform assembly
• Job shop flexibility for customization modules
• Pull-based replenishment for standard components

Group Technology (GT) Based Cellular Manufacturing:

Cell Formation Methodology:

Used Part-Machine Grouping Algorithm:

1. Created component-process incidence matrix (850 × 12 processes)
2. Applied Rank Order Clustering (ROC) algorithm
3. Optimized for block diagonalization minimizing exceptional elements

Resulting Cell Configuration:

Cell 1: Platform Manufacturing Cell

• Dedicated to base platform production (70% of machine)
• U-shaped layout optimizing material flow
• Cross-trained operators (polyvalence = 0.82)
• Takt time: Not applicable (batch of 1, cycle time focus: 4.2 hours per station)
• Process sequence: Fabrication → Machining → Sub-assembly → Platform integration

Cell 2: Module Manufacturing Cell

• Folding mechanism modules (2/3/4-fold variants)
• Smaller footprint, modular tooling/fixtures
• Quick changeover design (SMED principles applied)
• Cycle time: 2.6 hours per module

Cell 3: Electrical/Control Systems Cell

• Panel assembly, wiring, PLC programming
• Standard panel design with parametric PLC code
• Testing jigs for functional validation
• Cycle time: 3.8 hours (80% standard, 20% custom programming)

Cell 4: Final Integration Cell

• Platform + Modules + Electrical assembly
• Functional testing, calibration, validation
• Customer-specific configuration
• Cycle time: 6.2 hours

Intercell Logistics:

• FIFO lanes between cells
• Kanban triggering for standard components
• Visual management boards for job status
• Electronic Job Traveler system (barcode tracking)

Heijunka (Load Leveling) Adaptation for HMLV:

Traditional Heijunka Box assumes repeatable product mix. Our adaptation:

Capacity Smoothing Strategy:

Instead of mix leveling (impossible in ETO), implemented capacity leveling:

1. Aggregated Capacity Planning:
   • Converted all variants to “Standard Machine Equivalents (SME)”
   • 2-fold machine = 0.92 SME
   • 3-fold machine = 1.0 SME (baseline)
   • 4-fold machine = 1.15 SME
   • Planned weekly capacity in SME units, not individual machines
2. Time-Buffered Scheduling:
   • Established planned lead time: 62 days
   • Divided into time-phased capacity buckets
   • Load smoothing across buckets to avoid peaks/valleys
3. Constraint-Based Scheduling:
   • Identified primary constraint (CNC machining)
   • Applied Drum-Buffer-Rope (DBR) scheduling
   • Drum: Constraint schedule (optimized for setup reduction)
   • Buffer: Time buffer protecting constraint (5.2 days)
   • Rope: Release schedule tied to drum consumption

Heijunka Scheduling Board:

Modified for ETO environment:

• Horizontal axis: Time periods (weekly buckets)
• Vertical axis: Capacity slots (in SME units)
• Color coding: Product families (visual load balancing)
• Daily management: Capacity adherence review

Results:

• Capacity utilization variance: Reduced from CV = 0.74 to CV = 0.22
• Overtime incidence: Reduced 68%
• Crisis expediting: Reduced 81%

Load Balancing Within HMLV Constraints:

Yamazumi Charts for Process Balancing:

Created Yamazumi (load balancing) charts for platform assembly:

• Decomposed assembly into 23 discrete work elements
• Measured time for each element (standard work combination sheet)
• Redistributed work to achieve 85-90% balance efficiency
• Established standard work-in-process (SWIP) between stations

Balance Efficiency Calculation:

BE = (Sum of station times) / (Number of stations × Maximum station time)

• Achieved BE = 0.87 (target: >0.85 for HMLV)

Workforce Flexibility Development:

• Created skill matrix mapping all operators × all operations
• Implemented training matrix with competency levels (1-5 scale)
• Minimum requirement: Each operator proficient in 3+ operations
• Achieved average proficiency: 3.8 operations per operator

Setup Reduction Using SMED (Single-Minute Exchange of Die):

Applied SMED methodology to critical operations:

Setup Time Analysis:

• Baseline setup time: 3.2 hours (CNC machining fixture changes)
• Target: <10 minutes (single-minute)

SMED Steps Applied:

1. Separate Internal and External Setup:
   • Internal (machine must stop): 2.1 hours
   • External (can do while running): 1.1 hours
2. Convert Internal to External:
   • Pre-staged fixtures, tools, programs
   • Standardized fixture interface (3-2-1 principle)
   • Quick-change tooling (minimag systems)
3. Streamline Remaining Internal Setup:
   • Eliminated adjustment through precise positioning systems
   • One-touch clamps replacing bolted fixtures
   • Color-coded setup procedures

Results:

• Setup time reduced to 18 minutes (94% reduction)
• Setup frequency now economically viable for batch size = 1
• Setup time no longer a constraint consideration

Phase 4: Supply Chain Synchronization (Months 10-16)

Strategic Sourcing and Supplier Rationalization:

Pareto Analysis of Procurement Spend:

Applied ABC analysis:

• A-items (80% value, 15% items): 142 critical components
• B-items (15% value, 25% items): 287 moderate-value components
• C-items (5% value, 60% items): 421 low-value, high-volume items

Supplier Segmentation Matrix (Kraljic Portfolio):

Category	Strategy	Suppliers (Before → After)	Contract Type
Strategic (A-items, sole source)	Partnership	12 → 8	Long-term, VMI
Leverage (A-items, multiple sources)	Competitive bidding	28 → 15	Annual contracts, call-off
Bottleneck (B-items, limited sources)	Secure supply	34 → 22	Framework agreements
Non-critical (C-items, abundant)	Efficiency	68 → 33	E-procurement, 2-bin Kanban

Total vendor consolidation: 142 → 78 (45% reduction)

Inventory Management System Redesign:

Platform Components (70% of machine):

Implemented Make-to-Stock (MTS) with Reorder Point (ROP) system:

ROP = (Average daily demand × Lead time) + Safety stock Safety stock = Z × σ_demand × √Lead time

• Maintained 3-machine equivalent safety stock
• ABC classification for storage location optimization
• Cycle counting (daily for A, weekly for B, monthly for C)

Module Components (20% of machine):

Implemented Assemble-to-Order (ATO):

• Held common module base components in stock
• Variant-specific components: Make-to-Order
• Postponement strategy (delayed differentiation)

Custom Components (10% of machine):

Pure Make-to-Order (MTO):

• Procurement release only post-design freeze (Gate 2)
• Supplier development for lead time reduction
• Electronic data interchange (EDI) for PO transmission

Vendor-Managed Inventory (VMI) for Strategic Suppliers:

• Consignment stock for high-value, long-lead items
• Supplier maintains inventory at our location
• Payment on consumption (improves working capital)
• Real-time visibility through ERP integration

Bullwhip Effect Mitigation:

1. Information Sharing: Shared production schedule visibility with Tier-1 suppliers (4-week rolling horizon)
2. Order Batching Reduction: Eliminated EOQ-based ordering for platform components, implemented daily replenishment
3. Price Stabilization: Negotiated fixed-price annual contracts removing promotion-based ordering
4. Rationing Elimination: Transparent capacity allocation during constraints

Supplier Quality Management:

• Implemented Supplier Quality Index (SQI) scoring
• SQI = f(Defect PPM, On-time delivery %, Lead time variance, Responsiveness)
• Quarterly business reviews with strategic suppliers
• Joint FMEA sessions for critical components
• First Article Inspection (FAI) protocol for new parts

Material Requirements Planning (MRP) Optimization:

Challenge: MRP nervousness due to design changes

Solution:

• Established planning time fence (30 days – firm zone)
• Demand time fence (15 days – frozen zone)
• MRP parameters tuning:
  • Planning horizon: 90 days
  • Replanning frequency: Weekly (vs. daily previously)
  • Lot sizing: Lot-for-lot for custom, Fixed Order Quantity for standard
  • Safety stock: Strategic for A-items only

Inventory Performance Metrics:

Metric	Before	After	Improvement
Inventory Turnover Ratio	2.1	4.8	129%
Days Inventory Outstanding	174 days	76 days	56% reduction
Obsolete Inventory	₹11.2 Cr	₹0.4 Cr	96% reduction
Stockout Frequency	8-12/machine	0.8/machine	92% reduction
Inventory Accuracy	76%	98.5%	30% improvement

---

IMPLEMENTATION RESULTS: QUANTITATIVE IMPACT ANALYSIS

Throughput Time Reduction (Primary Objective)

Manufacturing Lead Time (MLT) Transformation:

Phase	Before (days)	After (days)	Reduction	Improvement %
Design Engineering	128	38	90	70.3%
Procurement	285	52	233	81.8%
Manufacturing	247	48	199	80.6%
Rework/Rectification	62	8	54	87.1%
Total MLT	722	62	660	91.4%

Little’s Law Validation:

WIP = Throughput Rate × Lead Time

• Before: 18 machines WIP = (6 machines/year) × (722/365) = 11.9 ≈ 12 machines (validated)
• After: 3.2 machines WIP = (32 machines/year) × (62/365) = 5.4 ≈ 5 machines (observed: 3-4, excellent)

Manufacturing Critical Path Ratio (MCPR):

MCPR = Touch time / Total throughput time

• Before: 82/722 = 0.114
• After: 48/62 = 0.774 (578% improvement in flow efficiency)

Queue Time Reduction:

Analyzed using Kingman’s VUT equation for queue time:

Q = (u/(1-u)) × (Ca² + Cs²)/2 × te

Where:

• u = utilization
• Ca = coefficient of variation of arrivals
• Cs = coefficient of variation of service time
• te = mean service time

Constraint Resource Queue Time:

• Reduced utilization from 142% to 84% (eliminated overload)
• Reduced process time CV from 2.5 to 0.8 (standardization effect)
• Queue time reduced from 67 days to 4 days (94% reduction)

Production System Performance Metrics

Overall Equipment Effectiveness (OEE) – Cell Level:

OEE = Availability × Performance × Quality

Platform Manufacturing Cell:

Component	Before	After
Availability	68%	89%
Performance	62%	84%
Quality (FTQ)	73%	96%
OEE	30.7%	71.8%

Improvement: 134% increase in OEE (world-class benchmark: >85%)

Throughput Accounting Results:

Metric	Before (₹ Cr)	After (₹ Cr)	Change
Throughput (T)	18	53	+194%
Operating Expense (OE)	22	24	+9%
Inventory (I)	28	14.5	-48%
Net Profit (NP = T – OE)	-4	29	Profitable
ROI (NP/I)	Negative	200%	–
Productivity (T/OE)	0.82	2.21	+170%

Drum-Buffer-Rope (DBR) Performance:

Post-implementation of constraint-based scheduling:

• Constraint utilization: Optimized at 84% (sustainable)
• Buffer penetration frequency: <5% (excellent protection)
• Due date performance: 89% (vs. 42% previously)
• Constraint efficiency: 94% (minimized starving/blocking)

Setup Time Performance (SMED Impact):

Machine Type	Setup Time Before	Setup Time After	Reduction
CNC Machining	192 min	18 min	91%
Press Brake	145 min	12 min	92%
Welding Fixtures	87 min	8 min	91%
Assembly Tooling	56 min	6 min	89%

Economic Batch Quantity (EBQ) Impact:

Using Harris-Wilson formula: EBQ = √(2DS/H) × √(P/(P-D)) × √((H+πC)/(H))

Where setup cost (S) reduced by 90%, economic batch size approaches 1 unit, enabling true one-piece flow for custom machines.

Quality Performance Transformation

First Time Yield (FTY) Metrics:

FTY = (Units produced – Defects) / Units produced

Process Stage	FTY Before	FTY After	Improvement
Fabrication	82%	97%	18.3%
Machining	88%	98%	11.4%
Sub-assembly	76%	95%	25%
Final Assembly	68%	94%	38.2%

Rolled Throughput Yield (RTY):

RTY = FTY₁ × FTY₂ × … × FTYn

• Before: 0.82 × 0.88 × 0.76 × 0.68 = 0.372 (37.2%)
• After: 0.97 × 0.98 × 0.95 × 0.94 = 0.848 (84.8%)

Impact: 128% improvement in overall process yield

Defects Per Million Opportunities (DPMO):

DPMO = (Defects / Opportunities) × 1,000,000

Metric	Before	After	Sigma Level Change
DPMO	87,500	12,300	2.8σ → 3.9σ
Cost of Poor Quality	13.3% sales	4.8% sales	64% reduction

Design FMEA Risk Priority Number (RPN) Reduction:

RPN = Severity × Occurrence × Detection

• Average RPN (critical failures): 420 → 98 (77% reduction)
• High-risk items (RPN >400): 23 → 2 (91% reduction)

Financial Performance Indicators

Revenue and Profitability:

Metric	Year 0	Year 1	Year 2	CAGR
Revenue (₹ Cr)	60	78	95	25.7%
Gross Margin %	-6.7%	8.2%	18.5%	–
EBITDA (₹ Cr)	-4	6.4	11.4	–
EBITDA %	-6.7%	8.2%	12.0%	–
Net Profit (₹ Cr)	-4	4.1	6.8	–
Net Margin %	-6.7%	5.3%	7.2%	–

Working Capital Optimization:

Cash Conversion Cycle (CCC) Transformation:

Component	Before	After	Days Saved
Days Inventory Outstanding (DIO)	485	76	409
Days Sales Outstanding (DSO)	125	98	27
Days Payable Outstanding (DPO)	68	72	-4
Cash Conversion Cycle	542	102	440

Working Capital Released: ₹14.2 Crores

• (440/365) × ₹60 Cr annual revenue × 1.28 growth factor

Return on Net Assets (RONA):

RONA = Net Operating Profit After Tax / Net Assets

• Before: Negative
• After: 28.4% (excellent for capital goods sector)

Economic Value Added (EVA):

EVA = NOPAT – (Capital × WACC)

• Before: Negative ₹8.2 Crores (value destruction)
• After: Positive ₹4.7 Crores (value creation)

Cost Structure Transformation:

Cost Category	Before (% Revenue)	After (% Revenue)	Change
Direct Material	45%	38%	-7% points
Direct Labor	18%	14%	-4% points
Manufacturing Overhead	24%	18%	-6% points
Engineering/Design	8%	6%	-2% points
SG&A	12%	11%	-1% point
Total Cost	107%	87%	-20% points

Operational Capacity Metrics

Constraint-Based Capacity Analysis:

Using Theory of Constraints (TOC) methodology:

Theoretical Throughput Capacity:

T = (Available time × Efficiency) / (Processing time per unit)

Platform Manufacturing Cell:

• Before: 6 machines/year (actual), 24 machines/year (theoretical at 100% utilization)
• Capacity utilization: 25% (severe underutilization indicating constraints elsewhere)

After Constraint Elimination:

• Actual: 32 machines/year
• Theoretical: 42 machines/year
• Capacity utilization: 76% (healthy, sustainable)

Scalability Index:

SI = (Incremental revenue) / (Incremental fixed cost)

• Achieved SI = 4.2 (₹35 Cr additional revenue / ₹8.3 Cr additional fixed cost)
• Indicates excellent operating leverage

Labor Productivity:

Revenue per Employee = Total Revenue / Total Employees

Metric	Before	After	Improvement
Total Employees	68	52	23.5% reduction
Revenue/Employee	₹88 lakhs	₹183 lakhs	108% improvement
Value-added/Employee	₹26 lakhs	₹102 lakhs	292% improvement

---

SUSTAINABILITY MECHANISMS AND CONTROL SYSTEMS

Statistical Process Control (SPC) Implementation

Control Charts for Key Processes:

Implemented X-bar and R charts for critical dimensions:

• Monitored 47 critical-to-quality (CTQ) characteristics
• Process capability studies: Target Cpk ≥ 1.33
• Achieved average Cpk: 1.68 (indicating robust processes)

Control Limits Calculation:

UCL = x̄ + A₂R̄
LCL = x̄ – A₂R̄

Where A₂ is constant based on sample size, R̄ is average range

SPC Results:

• Special cause variation incidents: Reduced 84%
• Process stability index: 0.94 (target: >0.90)
• Out-of-control signals: <2% of observations

Performance Management Dashboard (Hoshin Kanri Deployment)

Strategic Objective Decomposition:

Level 1: Corporate Breakthrough Objective

• Reduce MLT to 62 days while achieving 12% EBITDA

Level 2: Departmental Key Performance Indicators (KPIs)

• Design: ECO <3 per machine, Design cycle time <38 days
• Manufacturing: OEE >70%, On-time delivery >85%
• Supply Chain: Inventory turns >4.5, Stockouts <1 per machine
• Quality: FTY >95%, Customer complaints <4 per machine

Level 3: Team/Individual Metrics

• Daily management metrics cascaded to shop floor
• Visual management boards with real-time status

X-Matrix (Hoshin Planning Matrix):

Linked strategic objectives → annual objectives → improvement priorities → metrics → owner accountability

PDCA (Plan-Do-Check-Act) Review Cadence:

• Daily: Tier 1 huddles (15 min) – production metrics review
• Weekly: Tier 2 meetings (60 min) – VSM metrics, problem escalation
• Monthly: Tier 3 reviews (half-day) – financial performance, strategic initiative status
• Quarterly: Hoshin review (full-day) – strategic objective progress, catchball for next quarter

Continuous Improvement (Kaizen) Culture Establishment

Structured Problem-Solving Methodology:

A3 Thinking Deployment:

Trained 85% of workforce in A3 problem-solving:

• Background/Current condition
• Goal/Target condition
• Root cause analysis (5-Why, Ishikawa)
• Countermeasures
• Implementation plan
• Follow-up/Confirmation

Kaizen Metrics:

• Kaizen suggestions: 247 submissions (Year 1)
• Implementation rate: 68%
• Annual savings from kaizen: ₹2.4 Crores
• Average suggestion value: ₹97,000

Gemba Walks:

Established gemba walk protocol:

• Senior management: Weekly gemba walks (2-3 hours)
• Focus on “go, see, ask why, show respect”
• Documented observations, immediate countermeasures for safety/quality
• Long-term improvements captured in A3 format

Knowledge Management and Standard Work

Standard Operating Procedure (SOP) Documentation:

Created comprehensive SOP library:

• Assembly SOPs: 127 documents (platform + modules)
• Machining process sheets: 89 documents
• Quality inspection procedures: 56 documents
• Maintenance procedures: 34 documents

Standard Work Documentation:

For each operation:

• Standard Work Combination Sheet: Sequence, time, walking distance
• Standard Work Chart: Layout, material placement, quality checks
• Job Instruction Sheet: Key points, reasons, safety

Lessons Learned Database:

Implemented structured knowledge capture:

• Engineering change history with rationale
• Design failure modes and solutions
• Manufacturing problem-solving database
• Searchable by component, process, failure mode

Impact:

• Training time for new operators: Reduced from 6-9 months to 2-3 months
• Knowledge retention despite 18% attrition: Maintained through documentation
• Cross-training efficiency: 72% reduction in time-to-competency

Design Standardization Governance

Product Data Management (PDM) System:

Implemented PLM/PDM for configuration control:

• Single source of truth for all CAD models
• Automated BOM generation
• ECO workflow with approvals
• Revision control and audit trail

Design Reuse Metrics:

Tracked design efficiency:

• Standard component library: 1,847 items
• Design reuse percentage: 68% (target: >60%)
• New design requirements: Only 32% of total components
• Design time per new component: Reduced 54%

Modular Design Standards:

Established design rules:

• Interface standards (mechanical, electrical, pneumatic)
• Dimensional coordination (module mounting patterns standardized)
• Connector standardization (M12 for sensors, M23 for actuators)
• Software modularity (function block library for PLC programming)

---

ADVANCED INSIGHTS: THEORETICAL FRAMEWORKS APPLIED

Theory of Constraints (TOC) – Beyond Implementation

Five Focusing Steps Application:

1. IDENTIFY the constraint: Dynamic bottleneck identification using real-time capacity analysis
2. EXPLOIT the constraint: DBR scheduling, setup reduction, quality improvement at constraint
3. SUBORDINATE everything else: Non-constraint resources scheduled per constraint rhythm
4. ELEVATE the constraint: Capacity expansion where exploitation insufficient (added 2nd CNC machine). Best scale manufacturing business
5. Return to Step 1: Continuous constraint reassessment (constraint shifted from CNC to assembly)

Constraint Migration Management:

Established control mechanism to prevent “constraint whiplash”:

• Predictive constraint modeling using simulation
• Planned constraint elevation before full exploitation exhausted
• Buffer management to detect emerging constraints early

Throughput Accounting Integration:

Decision-making shifted from cost accounting to throughput accounting:

• Make/buy decisions: Based on throughput per constraint minute, not unit cost
• Product mix: Optimized for maximum throughput (₹/constraint hour)
• Investment decisions: ROI calculated using throughput impact, not cost savings

Mass Customization Theory – Pine’s Framework

Four Approaches to Mass Customization:

Our implementation combined multiple approaches:

1. Collaborative Customization: QFD-based dialogue to identify customer needs, translate to standard platform + custom modules
2. Adaptive Customization: Standard product with user-adjustable features (box size servo adjustment, interchangeable folding cassettes)
3. Cosmetic Customization: Standard functionality with presentation customization (HMI screens, labeling, color schemes)
4. Transparent Customization: Engineered solutions without customer awareness of customization effort (automatic parameter calculation in PLC based on box dimensions)

Solution Space Development:

Defined feasible solution space:

• Box dimensions: 50mm × 50mm to 400mm × 600mm (parametric range)
• Folding configurations: 2/3/4-fold (discrete options)
• Throughput speeds: 30-120 boxes/min (performance spectrum)
• Automation level: Manual, semi-auto, full-auto (tiered offerings)

Postponement Strategy:

Delayed differentiation point:

• Generic platform manufactured to forecast (push system)
• Customization modules triggered by order (pull system)
• Final configuration at assembly (postponement point)
• Lead time visibility: Platform (stock), Modules (3 weeks), Assembly (1 week) = 62 days total

Queueing Theory Application – Kingman’s Equation

Analytical Queue Modeling:

Validated queue time improvements using queueing theory:

Kingman’s VUT Equation:

E[Wq] = (u/(1-u)) × ((Ca² + Cs²)/2) × E[S]

Where:

• u = utilization factor
• Ca = coefficient of variation of arrival process
• Cs = coefficient of variation of service process
• E[S] = mean service time

Before State (CNC Machining Constraint):

• u = 1.42 (overloaded – unstable queue)
• Ca = 2.1 (highly variable arrivals due to unpredictable design releases)
• Cs = 2.5 (high service time variation across custom parts)
• E[S] = 2.8 hours
• Predicted queue time: Unbounded (unstable system)

After State:

• u = 0.84 (sustainable utilization)
• Ca = 0.6 (stabilized by design freeze, batched releases)
• Cs = 0.8 (reduced through standardization)
• E[S] = 1.4 hours (reduced through setup reduction)
• Predicted queue time: E[Wq] = (0.84/0.16) × (0.85) × 1.4 = 6.2 hours
• Observed queue time: 4-6 hours (validated model)

Key Insight: Variability reduction (Ca, Cs) had greater impact than utilization reduction on queue time

Product Platform Strategy – Meyer & Lehnerd Framework

Platform Leverage Metrics:

Degree of Commonality (DC):

DC = (Number of common components) / (Total unique components across product family)

• Achieved DC = 0.72 (excellent for ETO environment)

Platform Efficiency (PE):

PE = (Development cost per new variant) / (Development cost for standalone product)

• Achieved PE = 0.22 (78% cost reduction for new variants)

Platform Effectiveness (PF):

PF = (Number of derivatives launched) / (Platform development cost)

• Achieved PF = 12.4 derivatives per ₹1 Crore platform investment

Scalability Index (SI):

SI = (Range of performance variation) / (Range of platform variation)

• Box size range: 800% variation (50mm to 400mm)
• Platform variation: 0% (single platform)
• SI = 800% / 0% = Infinite scalability (theoretical)

Design Structure Matrix (DSM) – Eppinger Methodology

Task Dependency Analysis:

Created 47×47 DSM for design task sequencing:

• Rows/Columns: Design tasks
• Matrix cells: Information dependencies (binary or weighted)

Partitioning Algorithm Applied:

• Reordered tasks to minimize feedback marks (iterations below diagonal)
• Identified strongly coupled tasks (clustered blocks on diagonal)
• Sequenced independent tasks in parallel

Results:

• Design iteration loops: Reduced from 14 to 3
• Critical path length: Reduced 38%
• Concurrent engineering opportunities: Identified 23 parallel task sets
• Design cycle time: Validated reduction from 128 to 38 days

---

LESSONS LEARNED: THEORETICAL AND PRACTICAL SYNTHESIS

HMLV-Specific Insights

1. Impossibility of Traditional Flow in Pure HMLV

Conventional lean “one-piece flow” assumes:

• Repetitive products
• Predictable demand
• Stable routing
• Uniform processing times

HMLV Reality:

• Non-repetitive (every product unique)
• Stochastic demand
• Variable routing (jumbled flow)
• Heterogeneous processing

Solution Paradigm Shift:

• Flow at platform level (standardized)
• Flexibility at module level (customized)
• Hybrid push-pull production control

2. Constraint Migration is Inherent in HMLV

Unlike stable production environments with static bottlenecks, HMLV systems exhibit dynamic constraints due to:

• Product mix variation
• Design complexity variation
• Resource skill mismatches
• Material availability dependencies

Management Approach:

• Real-time constraint identification
• Rapid constraint exploitation
• Predictive constraint modeling
• Buffer management as early warning system

3. Variability Reduction Trumps Capacity Addition

From Kingman’s equation and empirical validation:

• Reducing utilization from 142% to 84% had linear queue time impact
• Reducing variability (Ca, Cs) from >2.0 to <0.8 had exponential queue time impact

Implication: Variability reduction through standardization yields greater ROI than capacity investment

4. Group Technology Enables Mass Customization

Critical enabler was classification of product universe into:

• Common components (standard)
• Variant components (modular)
• Unique components (custom)

Statistical Clustering Validation:

• Silhouette coefficient: 0.68 (good cluster separation)
• Within-cluster similarity: >85%
• Between-cluster differentiation: Sufficient for market segmentation

Design-Manufacturing Interface

5. Quality Function Deployment (QFD) Prevented Requirement Creep

Formal VOC (Voice of Customer) → CTQ (Critical to Quality) translation:

• Reduced interpretation errors
• Created measurable design targets
• Enabled objective design review
• Reduced customer-designer mismatch (primary ECO driver)

6. Design Freeze Non-Negotiability

Single most impactful intervention:

• ECO reduction: 83.8%
• BOM stability enabled JIT procurement
• Process planning certainty enabled cellular layout

Enforcement Mechanisms:

• Contractual agreements
• Cost transparency for post-freeze changes
• Schedule impact communication
• Escalation protocols

7. Modular Architecture Requires Interface Standardization

Platform modularity succeeded due to:

• Mechanical interface standards (ISO tolerance stack-up, bolt patterns)
• Electrical interface standards (connector types, voltage levels, signaling protocols)
• Software interface standards (function block encapsulation, API definitions)
• Data interface standards (HMI parameter structures)

Without interface standards: Modules become product-specific, defeating mass customization objective

Production Control in HMLV

8. Heijunka Adaptation: Capacity-Based vs. Mix-Based

Traditional Heijunka (mix leveling) infeasible when:

• Product mix entropy >4.0 (ours: 4.8)
• Batch size = 1 (ETO reality)

Successful Adaptation:

• Capacity leveling using equivalent units (SME)
• Time-phased capacity planning (not takt-based sequencing)
• Visual load balancing (Heijunka board showing capacity, not specific products)

9. Drum-Buffer-Rope (DBR) Superiority Over Kanban in HMLV

Kanban assumptions violated in HMLV:

• Repetitive withdrawal
• Standard containers
• Stable demand

DBR advantages:

• Works with custom orders
• Optimizes constraint utilization
• Accommodates variable processing times
• Provides schedule visibility

Hybrid Approach:

• DBR for custom assemblies (main production flow)
• Kanban for standard components (C-items replenishment)

10. Buffer Management as Control System

Time buffers (not inventory buffers) used for:

• Constraint protection
• Lead time reliability
• Constraint starvation prevention

Buffer zones:

• Green (>⅔ buffer remaining): Normal operation
• Yellow (⅓-⅔ consumed): Expedite if necessary
• Red (<⅓ remaining): Crisis management, root cause analysis

Buffer penetration tracking:

• Identified chronic vs. sporadic causes
• Targeted improvement efforts
• Validated constraint effectiveness

Supply Chain Configuration

11. Inventory Positioning Strategy (Decoupling Points)

Applied CODP (Customer Order Decoupling Point) analysis:

Before (Pure MTO):

• CODP at raw material
• All components procured post-order
• Lead time = Design + Procurement + Manufacturing

After (Hybrid MTS-ATO-MTO):

• Platform components: CODP at finished goods (MTS)
• Module components: CODP at component inventory (ATO)
• Custom components: CODP at raw material (MTO)
• Lead time = Manufacturing only (for 70% of machine)

12. Supplier Relationship Segmentation (Kraljic)

Different supplier management by category:

• Strategic: Partnering, VMI, joint development
• Leverage: Competitive bidding, multiple sources
• Bottleneck: Security of supply, long-term contracts
• Non-critical: Efficiency, e-procurement, consolidation

Error to Avoid: Treating all suppliers uniformly (transactional vs. relational approach)

Organizational Change Management

13. Cross-Functional Ownership Essential

VSM revealed:

• 35% delays from Design (ECOs)
• 28% delays from Supply Chain (material)
• 22% delays from Manufacturing (quality)
• 15% delays from HR (workforce)

Implication: Siloed improvement insufficient

Solution:

• Cross-functional improvement teams
• Shared metrics (MLT, throughput)
• Joint problem-solving (A3 reviews)
• Aligned incentives (company profitability, not departmental efficiency)

14. Sales Incentive Structure Misalignment

Original problem:

• Sales incentivized on revenue (price × volume)
• No accountability for profitability, delivery, or complexity

Result:

• Over-promising on delivery
• Accepting unprofitable customization
• No pushback on design changes

Corrected Incentive Structure:

• 60% weight: Gross margin (not revenue)
• 20% weight: On-time delivery (customer satisfaction)
• 20% weight: Design freeze compliance (operational efficiency)

Impact:

• Sales behavior aligned with operations capability
• Customer expectation management improved
• Profitability focus institutionalized

15. Skill Matrix and Polyvalence Critical for HMLV

HMLV workforce requirements:

• Broad skill sets (multiple operations)
• Problem-solving capability (non-routine work)
• Adaptability (varying product mix)

Polyvalence Index Target: >0.75 (each operator proficient in 75%+ of cell operations)

Training Investment:

• 180 hours/operator/year (vs. 40 hours previously)
• Cross-training incentive pay
• Competency-based progression

Result:

• Flexible deployment across products
• Reduced dependency on specific individuals
• Improved schedule adherence despite absenteeism

---

SCALABILITY VALIDATION AND FUTURE TRAJECTORY

Scalability Proof Points

Volume Scaling (Achieved):

• Throughput: 6 → 32 machines/year (433% increase)
• Headcount: 68 → 52 (-23.5%, demonstrating productivity scaling)
• Revenue per employee: ₹88 lakhs → ₹183 lakhs (108% improvement)

Variety Scaling (Potential):

• Platform can accommodate:
  • 8 box dimension ranges (vs. 3 currently marketed)
  • 5 folding configurations (vs. 3 current: 2/3/4-fold)
  • 3 automation levels (vs. 1 standard)
• Theoretical product configurations: 120 (8×5×3)
• Current configurations marketed: 12
• Scalability headroom: 10× variety expansion without platform redesign

Revenue Scaling Model:

Regression analysis (R² = 0.94): Revenue = f(Throughput capacity, Product variety, Market penetration)

Projected capacity (Year 5):

• Throughput: 50 machines/year (constraint elevation plan)
• Revenue: ₹162 Crores
• EBITDA: 14.5%
• Working capital requirement: ₹18 Crores (vs. ₹28 Crores at ₹60 Cr revenue – efficiency gain)

Digital Transformation Roadmap

Phase 1: ERP Integration (Months 18-24)

• Fully integrated MRP-II system
• Real-time shop floor data capture (MES integration)
• Automated constraint identification algorithms
• Predictive analytics for bottleneck forecasting

Phase 2: PLM Maturity (Months 24-36)

• Configurator-based quoting (CPQ – Configure-Price-Quote)
• Automated BOM generation from customer specifications
• Digital twin for virtual commissioning
• Simulation-based design validation (FEA, CFD, MBD)

Phase 3: Smart Manufacturing (Months 36-48)

• IoT sensors on manufacturing equipment
• Predictive maintenance (reducing unplanned downtime)
• Automated quality inspection (machine vision)
• Augmented reality (AR) for assembly guidance

Market Expansion Opportunities

Geographic Expansion:

• Current: 85% domestic (India), 15% exports
• Target (Year 5): 50% domestic, 50% exports
• Enabled by: Faster delivery (competitive advantage), standardized designs (easier certification), modular shipping (reduced logistics cost)

Vertical Integration:

• Aftermarket service revenue (AMC contracts)
• Spare parts standardization (high-margin business)
• Performance upgrades (modular retrofits to existing machines)
• Projected service revenue: 20% of total revenue (Year 5)

Horizontal Expansion:

• Adjacent applications: Blister packaging, labeling, serialization machines
• Leveraging: Common platform, design capabilities, manufacturing infrastructure
• Product portfolio expansion: 3 product lines → 8 product lines (Year 5)

---

CONCLUSION: HMLV TRANSFORMATION BLUEPRINT

Core Thesis Validated

Hypothesis: High-Mix Low-Volume (HMLV) Engineer-to-Order (ETO) environments can achieve lean manufacturing benefits through mass customization paradigm, despite inapplicability of traditional lean tools designed for repetitive manufacturing.

Evidence:

• 91.4% throughput time reduction (722 → 62 days)
• 433% volume increase (6 → 32 machines/year) with 23.5% headcount reduction
• Transition from losses to 12% EBITDA margin
• 440-day working capital cycle improvement
• ₹14.2 Crores cash released

Mechanism: Platform-based architecture creating 70% standardization enabling economies of scale, while retaining 30% customization preserving market differentiation and premium pricing.

Critical Success Factors – HMLV Context

1. Group Technology Classification

• Systematic product family identification
• Component commonality analysis (incidence matrices, cluster analysis)
• Platform architecture derivation (70-20-10 rule: platform-module-custom)

2. Voice of Customer Formalization (QFD)

• Structured requirement capture preventing interpretation errors
• Engineering specification translation reducing design iterations
• Design freeze protocol with change control governance

3. Theory of Constraints (TOC) vs. Line Balancing

• Dynamic bottleneck identification and exploitation
• DBR scheduling superior to takt-based flow in HMLV
• Buffer management as control system for lead time reliability

4. Hybrid Production Control

• MTS for platform (standard components)
• ATO for modules (semi-custom sub-assemblies)
• MTO for unique components
• Postponement strategy minimizing inventory while maximizing responsiveness

5. Variability Reduction as Leverage Point

• Design variability: Standardization, modularity
• Process variability: SMED, SOP, Poka-yoke
• Demand variability: Design freeze, batched releases
• Supply variability: Supplier development, VMI

Impact validated through Kingman’s equation: Variability reduction (Ca, Cs) had exponential impact on queue time vs. linear impact of utilization reduction. Best scale manufacturing business

Applicability to Other HMLV Environments

Sectors with Similar Characteristics:

• Custom machinery manufacturing (injection molding, CNC machines)
• Defense equipment (low-volume, high-variety)
• Shipbuilding (one-of-a-kind, long lead time)
• Aerospace (complex assemblies, custom configurations)
• Industrial automation (unique system integration)

Transferable Methodology:

1. VSM adapted for product families (not single product flow)
2. Group Technology for platform identification
3. QFD for requirement formalization
4. Design freeze governance
5. Modular architecture with interface standardization
6. Constraint-based scheduling (DBR)
7. Hybrid inventory positioning (MTS-ATO-MTO)
8. Polyvalent workforce development

Contextual Adaptations Required:

• Product complexity level (aerospace >machinery >apparel)
• Regulatory environment (defense/pharma vs. commercial)
• Customer power (procurement-driven vs. technology-driven)
• Supply chain maturity (developed vs. emerging markets)

Contribution to Industrial Engineering Body of Knowledge

1. HMLV Lean Framework

• Documented methodology for lean in jumbled flow environments
• Quantified effectiveness: 91.4% lead time reduction, 12% EBITDA achievement

2. Mass Customization Economics

• Validated 70-20-10 platform-module-custom architecture
• Demonstrated scalability: 10× variety potential, 5× volume achieved

3. Design-Manufacturing Integration

• QFD as ECO prevention (83.8% reduction)
• Design freeze protocol as throughput enabler

4. Variability as Primary Constraint

• Queueing theory validation: Variability reduction >> Capacity addition
• Quantified impact: CV reduction from >2.0 to <0.8 enabling 91.4% queue time reduction

5. Throughput Accounting in ETO

• Validated TOC financial metrics in HMLV
• Decision frameworks: Make/buy, product mix, investment prioritization

---

APPENDIX: TECHNICAL TOOLS AND CALCULATIONS

Value Stream Mapping Symbols (HMLV Adapted)

• Customer/Supplier Icon: Modified to show multiple customers (variety indicator)
• Process Box: Added “complexity rating” (1-5 scale) for process variability
• Data Box: Included CV (coefficient of variation) alongside mean cycle time
• Inventory Triangle: Distinguished raw material, WIP, finished goods by shading
• Push Arrow: Dashed (MRP-driven scheduling)
• Pull Arrow: Solid (Kanban replenishment)
• Constraint Icon: Red octagon (bottleneck indicator with utilization %)

Group Technology Classification Code

Hierarchical Classification Structure (8-digit code):

Position	Attribute	Example Values
1	Product family	2=2-fold, 3=3-fold, 4=4-fold
2-3	Box length category	01=50-100mm, 02=101-200mm, etc.
4-5	Box width category	01=50-100mm, 02=101-200mm, etc.
6	Automation level	1=Manual, 2=Semi-auto, 3=Full-auto
7	Throughput class	1=30-60 bpm, 2=61-90 bpm, 3=91-120 bpm
8	Special features	0=Standard, 1=Vision system, 2=Serialization, etc.

Example: 3-04-03-2-2-1 = 3-fold machine, 301-400mm length, 201-300mm width, semi-auto, 61-90 bpm, vision system

Usage:

• Enables rapid similarity search
• Facilitates platform component identification
• Supports automated BOM generation in configurator

Little’s Law Application – WIP Calculation

Formula: WIP = Throughput Rate (λ) × Lead Time (L)

Validation Example:

Before State:

• Observed WIP: 18-22 machines on shop floor
• Annual throughput: 6 machines/year = 6/365 = 0.0164 machines/day
• Lead time: 722 days

Calculation: WIP = 0.0164 × 722 = 11.8 machines

Discrepancy: 18-22 observed vs. 11.8 calculated

Root Cause: Design iterations and rework occurring outside main flow (not captured in linear throughput)

Corrected Model: WIP_total = WIP_design + WIP_manufacturing WIP_design = 3.2 machines (average design queue) WIP_manufacturing = 11.8 machines WIP_total = 15 machines (closer to observed 18-22, remaining variance explained by rework loops)

After State:

• Observed WIP: 3-4 machines
• Annual throughput: 32 machines/year = 0.0877 machines/day
• Lead time: 62 days

Calculation: WIP = 0.0877 × 62 = 5.4 machines

Match: 3-4 observed validates model (excellent flow)

Takt Time vs. Cycle Time – HMLV Distinction

Takt Time = Available time / Customer demand rate

Inapplicability in HMLV:

• Customer demand rate unknown (stochastic order arrival)
• Product mix variability prevents single takt time
• Batch size = 1 eliminates repeating cycles

Alternative Metric: Cycle Time

Cycle Time = Processing time per unit

Platform Manufacturing Cell:

• Station 1 (Frame fabrication): 4.8 hours
• Station 2 (Machining): 3.6 hours
• Station 3 (Sub-assembly): 4.2 hours (longest)
• Station 4 (Platform integration): 3.9 hours

Bottleneck station: Station 3 at 4.2 hours Cell cycle time: 4.2 hours (determines throughput)

Throughput capacity = Available time / Cycle time = (8 hours/day × 6 days/week × 50 weeks/year) / 4.2 hours = 2,400 hours/year / 4.2 hours = 571 platforms/year

Actual requirement: 32 machines/year Utilization: 32/571 = 5.6% (extremely low, indicates platform cell not constraint)

True constraint: Custom module manufacturing (longer cycle time, variable)

Overall Equipment Effectiveness (OEE) – Detailed Calculation

OEE = Availability × Performance × Quality

Platform Manufacturing Cell – Before:

Availability: = (Operating time) / (Planned production time) = (Planned time – Downtime) / (Planned production time)

Downtime breakdown:

• Breakdown: 12%
• Setup/changeover: 8%
• Minor stoppages: 6%
• Waiting for materials: 6%

Availability = (100% – 32%) = 68%

Performance: = (Actual cycle time) / (Ideal cycle time) = (Ideal cycle time × Total count) / (Operating time)

Losses:

• Slow cycles: 18%
• Small stops: 12%
• Reduced speed: 8%

Performance = (100% – 38%) = 62%

Quality: = Good count / Total count

First Time Yield = 73% Quality = 73%

OEE = 0.68 × 0.62 × 0.73 = 0.307 = 30.7%

Platform Manufacturing Cell – After:

Availability: Downtime breakdown:

• Breakdown: 4% (TPM implementation)
• Setup/changeover: 1% (SMED)
• Minor stoppages: 3% (5S, standardization)
• Waiting for materials: 3% (Kanban pull)

Availability = (100% – 11%) = 89%

Performance: Losses:

• Slow cycles: 6%
• Small stops: 5%
• Reduced speed: 5%

Performance = (100% – 16%) = 84%

Quality: First Time Yield = 96% (Poka-yoke, SPC) Quality = 96%

OEE = 0.89 × 0.84 × 0.96 = 0.718 = 71.8%

Improvement: 71.8% / 30.7% = 2.34× (134% increase)