Manufacturing Engineering Series Part 7: Lean Manufacturing & Operational Excellence

Lean Foundations & TPS

The Toyota Production System (TPS), developed by Taiichi Ohno and Shigeo Shingo at Toyota Motor Corporation between 1948-1975, revolutionized manufacturing worldwide. TPS rests on two pillars: Just-in-Time (make only what's needed, when it's needed, in the quantity needed) and Jidoka (automation with a human touch — machines detect defects and stop automatically). Western manufacturers adopted and renamed these principles as "Lean Manufacturing" after the 1990 publication of The Machine That Changed the World.

5S Workplace Organization

5S is the foundation of lean — a systematic method for organizing the workplace to eliminate waste, improve safety, and create visual standards. It's not housekeeping; it's the discipline that makes all other lean tools possible.

8 Wastes (Muda, Muri, Mura)

Lean identifies three types of waste: Muda (non-value-adding activities), Muri (overburden — pushing people or machines beyond capacity), and Mura (unevenness — variation in workload). The 8 wastes (originally 7, with "Unused Talent" added later) form the DOWNTIME acronym:

Flow, Pull & JIT Systems

Value Stream Mapping (VSM) is the single most powerful lean tool for visualizing the entire flow of material and information from supplier to customer. A current-state map reveals the ratio of value-adding time to total lead time — typically a shocking 1-5%. This means 95-99% of lead time is waste (waiting, batching, queuing, transportation).

Just-in-Time & Kanban Pull

Just-in-Time (JIT) means producing exactly what the customer wants, in the exact quantity, at the exact time needed — with zero inventory buffers. JIT uses a pull system: production is triggered by actual downstream consumption, not forecasts.

Kanban (literally "signboard" in Japanese) is the pull mechanism. When a downstream process consumes a container of parts, the empty container (with Kanban card attached) is sent upstream as an authorization to produce exactly one container of replenishment.

import numpy as np

# Kanban Quantity Calculator
# Formula: K = (D × L × (1 + S)) / C
# K = number of Kanban cards, D = demand rate, L = lead time
# S = safety factor, C = container quantity

demand_per_day = 480        # parts per day (one per minute across 8 hours)
lead_time_days = 0.25       # replenishment lead time (0.25 day = 2 hours)
safety_factor = 0.10        # 10% safety buffer
container_qty = 20          # parts per Kanban container (bin)

K = (demand_per_day * lead_time_days * (1 + safety_factor)) / container_qty
K_rounded = int(np.ceil(K))

# Calculate WIP with Kanban system
wip_kanban = K_rounded * container_qty
# Compare to traditional batch system (typically 2-5 days inventory)
wip_batch = demand_per_day * 3  # 3 days WIP typical in batch

print("Kanban System Design Calculator")
print("=" * 55)
print(f"Daily demand:           {demand_per_day} parts/day")
print(f"Replenishment lead time: {lead_time_days} days ({lead_time_days*8:.0f} hours)")
print(f"Safety factor:          {safety_factor*100:.0f}%")
print(f"Container quantity:     {container_qty} parts/bin")
print(f"\nK = ({demand_per_day} × {lead_time_days} × (1 + {safety_factor})) / {container_qty}")
print(f"K = {K:.2f} → round up to {K_rounded} Kanban cards")
print(f"\nWIP Inventory Comparison:")
print(f"  Kanban system:  {wip_kanban} parts ({wip_kanban/demand_per_day:.1f} days)")
print(f"  Batch system:   {wip_batch} parts ({wip_batch/demand_per_day:.1f} days)")
print(f"  Reduction:      {(1-wip_kanban/wip_batch)*100:.0f}%")
print(f"\nKanban Rules (Taiichi Ohno's 6 Rules):")
rules = [
    "1. Downstream pulls from upstream (never push)",
    "2. Upstream produces only what Kanban requests",
    "3. Do not send defective products downstream",
    "4. Number of Kanban cards limits WIP (reduce over time)",
    "5. Kanban is used to fine-tune production (smooth flow)",
    "6. Stabilize and rationalize the process continuously"
]
for rule in rules:
    print(f"  {rule}")

SMED & Quick Changeover

SMED (Single-Minute Exchange of Die), developed by Shigeo Shingo, reduces machine changeover time to under 10 minutes ("single-minute" refers to single digit). The method converts internal setup (tasks requiring machine stoppage) to external setup (tasks performed while machine runs), then streamlines both.

Six Sigma & Process Improvement

Six Sigma is a data-driven methodology for eliminating defects by reducing process variation. Developed at Motorola (1986) and popularized by GE under Jack Welch, Six Sigma targets 3.4 defects per million opportunities (DPMO) — a 99.99966% yield. The core framework is DMAIC:

Kaizen & Continuous Improvement

Kaizen (改善, "change for better") is the philosophy that every process can be improved, and everyone should participate in improvement. Unlike Six Sigma's project-based approach, Kaizen emphasizes small, daily improvements by the people who do the work.

OEE & Performance Metrics

Overall Equipment Effectiveness (OEE) is the gold standard metric for equipment productivity, combining three factors: Availability × Performance × Quality. World-class OEE is ≥85%, but most factories operate at 40-60%.

import numpy as np

# OEE (Overall Equipment Effectiveness) Calculator
# OEE = Availability × Performance × Quality

# Shift data for a CNC machining center
shift_minutes = 480          # 8-hour shift (480 minutes)
planned_breaks = 30          # minutes of planned breaks
planned_maintenance = 0      # scheduled maintenance (counted separately)
planned_time = shift_minutes - planned_breaks - planned_maintenance

# Downtime events (minutes)
downtime_events = {
    "Machine breakdown": 25,
    "Tooling change (unplanned)": 15,
    "Material shortage": 10,
    "Quality investigation": 8,
}
unplanned_downtime = sum(downtime_events.values())
operating_time = planned_time - unplanned_downtime

# Performance
ideal_cycle_time = 2.5       # minutes per part (theoretical best)
actual_parts_produced = 135  # parts actually produced in operating_time
ideal_parts = operating_time / ideal_cycle_time

# Quality
total_parts = actual_parts_produced
defective_parts = 4
good_parts = total_parts - defective_parts

# OEE Calculation
availability = operating_time / planned_time
performance = (actual_parts_produced * ideal_cycle_time) / operating_time
quality = good_parts / total_parts
oee = availability * performance * quality

print("OEE Analysis — CNC Machining Center")
print("=" * 55)
print(f"\n--- AVAILABILITY ({availability*100:.1f}%) ---")
print(f"  Planned production time:  {planned_time} min")
print(f"  Unplanned downtime:       {unplanned_downtime} min")
for event, mins in downtime_events.items():
    print(f"    → {event}: {mins} min")
print(f"  Operating time:           {operating_time} min")

print(f"\n--- PERFORMANCE ({performance*100:.1f}%) ---")
print(f"  Ideal cycle time:         {ideal_cycle_time} min/part")
print(f"  Ideal parts in {operating_time} min:   {ideal_parts:.0f} parts")
print(f"  Actual parts produced:    {actual_parts_produced} parts")
print(f"  Speed loss:               {ideal_parts - actual_parts_produced:.0f} parts")

print(f"\n--- QUALITY ({quality*100:.1f}%) ---")
print(f"  Total parts:    {total_parts}")
print(f"  Good parts:     {good_parts}")
print(f"  Defective:      {defective_parts} ({defective_parts/total_parts*100:.1f}%)")

print(f"\n{'='*55}")
print(f"  OEE = {availability*100:.1f}% × {performance*100:.1f}% × {quality*100:.1f}%")
print(f"  OEE = {oee*100:.1f}%")
print(f"  Rating: {'World Class (≥85%)' if oee >= 0.85 else 'Good (65-85%)' if oee >= 0.65 else 'Needs Improvement (<65%)'}")

# Six Big Losses breakdown
print(f"\n--- SIX BIG LOSSES ---")
print(f"  1. Breakdowns:           {downtime_events.get('Machine breakdown', 0)} min")
print(f"  2. Setup/Adjustments:    {downtime_events.get('Tooling change (unplanned)', 0)} min")
print(f"  3. Small Stops:          {(ideal_parts - actual_parts_produced) * ideal_cycle_time / 2:.0f} min (est)")
print(f"  4. Reduced Speed:        {(ideal_parts - actual_parts_produced) * ideal_cycle_time / 2:.0f} min (est)")
print(f"  5. Process Defects:      {defective_parts} parts")
print(f"  6. Startup Rejects:      0 parts")

Advanced & Agile Manufacturing

Classical lean (TPS) was designed for high-volume, repetitive manufacturing — Toyota makes millions of identical vehicles. But what about job shops making 1-50 custom parts? Or aerospace manufacturers with 200+ part numbers and 18-month lead times? Advanced lean adapts lean principles for these challenging environments.

Lean Supply Chain Integration

Lean doesn't stop at the factory walls. Lean supply chain extends JIT and pull principles to suppliers, logistics, and distribution. Toyota's supply chain is the benchmark — suppliers deliver parts directly to the assembly line in sequence, multiple times per day, with zero incoming inspection.

Agile Manufacturing & Resilience

Agile manufacturing extends lean by adding responsiveness to change. Where lean optimizes for efficiency in stable environments, agile manufacturing excels in volatile, uncertain environments — sudden demand spikes, product mix changes, supply disruptions, and short product lifecycles.