Manufacturing Production Systems & Quality Control

# Manufacturing Production Systems & Quality Control Knowledge Pack

## Lean Manufacturing Principles

### The 8 Wastes (DOWNTIME)
Lean identifies eight forms of waste that consume resources without adding customer value:

**Defects**: Products failing to meet specifications. Cost includes scrap material, rework labor, inspection time, warranty claims, and customer dissatisfaction. Addressing defects at the source (jidoka) costs 10x less than catching them at final inspection and 100x less than field failures.

**Overproduction**: Making more than the customer demands. The worst waste because it generates all other wastes — excess inventory, unnecessary transport, extra processing. Root cause is usually fear of stockouts or batch-size optimization that ignores total system cost.

**Waiting**: Idle time between process steps. Operators waiting for materials, machines waiting for setup, parts waiting in queue. VSM reveals waiting time — in most factories, value-added time is <5% of total lead time. The rest is waiting.

**Non-utilized talent**: Failing to leverage worker knowledge and skills. Frontline operators see problems daily that management never hears about. Countermeasure: structured suggestion systems (kaizen), cross-training, daily huddles.

**Transportation**: Moving materials between locations without adding value. Every move risks damage and adds lead time. Layout optimization (cellular manufacturing) minimizes transport. U-shaped cells keep all operations within arm's reach.

**Inventory**: WIP, raw materials, or finished goods exceeding immediate needs. Inventory hides problems — quality defects sit undiscovered in stockpiles, process imbalances are masked by buffers. Reducing inventory exposes problems so they can be solved. JIT delivery from suppliers, kanban pull systems internally.

**Motion**: Unnecessary movement of people. Reaching, bending, walking, searching for tools. 5S workplace organization (Sort, Set in order, Shine, Standardize, Sustain) eliminates motion waste. Shadow boards for tools, point-of-use storage, ergonomic workstation design.

**Extra processing**: Performing operations beyond customer requirements. Tighter tolerances than needed, redundant inspections, excessive documentation. DFM review during product development eliminates unnecessary complexity before production begins.

### Value Stream Mapping
VSM visualizes the entire flow from raw material to customer delivery. Draw current state first: every process step, inventory between steps, information flows, CT per station, changeover times, uptime, batch sizes, and transport methods.

Key metrics from the current-state map: total lead time (sum of processing time + waiting time), value-added ratio (processing time / lead time), total WIP inventory, number of handoffs. A typical discrete manufacturing VSM reveals value-added ratio of 1-5%.

Future-state map applies lean principles: continuous flow where possible, pull systems where flow isn't possible, level the production mix (heijunka), reduce batch sizes, implement SMED (single-minute exchange of dies) for rapid changeover.

### Takt Time and Line Balancing
TT = available production time / customer demand rate. If you have 480 minutes of production time and customer needs 240 units per day, TT = 2 minutes per unit. Every station must complete its work within TT.

Line balancing distributes work elements across stations so each station's CT is as close to TT as possible without exceeding it. Calculate balance efficiency: sum of all station CTs / (number of stations × TT). Target >85%. Imbalanced lines create bottlenecks (stations exceeding TT) and idle capacity (stations well under TT).

## Quality Control Systems

### Statistical Process Control
SPC uses control charts to monitor process stability. A process is "in control" when variation is due only to common causes (inherent random variation). Special cause variation (assignable, unusual) signals the process has changed.

**X-bar and R charts**: Monitor process mean and range. Subgroups of 3-5 consecutive parts measured at regular intervals. Control limits set at ±3 sigma from the process mean (NOT specification limits). Out-of-control signals: point beyond control limits, 7+ consecutive points on one side of center line (run), 7+ consecutive points trending up or down, 2 of 3 consecutive points beyond 2-sigma.

**Process capability**: Cp = (USL - LSL) / 6σ measures potential capability if the process were perfectly centered. Cpk = min[(USL - μ) / 3σ, (μ - LSL) / 3σ] accounts for actual centering. Cp ≥ 1.33 and Cpk ≥ 1.33 are minimum requirements for most industries. Automotive (IATF 16949) typically requires Cpk ≥ 1.67. A Cpk of 2.0 corresponds to approximately 3.4 PPM defective (Six Sigma level).

**Attribute charts**: p-chart (proportion defective), np-chart (number defective), c-chart (defects per unit), u-chart (defects per unit with variable sample size). Use when measurement is pass/fail rather than continuous variable.

### FMEA — Failure Mode and Effects Analysis
FMEA systematically identifies potential failures before they occur. Two types: Design FMEA (product) and Process FMEA (manufacturing).

For each potential failure mode, assess three factors (1-10 scale each):
- **Severity**: Impact on customer if failure occurs. 10 = safety hazard without warning, 1 = no effect
- **Occurrence**: Likelihood of the failure mode happening. 10 = near-certain, 1 = virtually impossible
- **Detection**: Likelihood of catching the failure before it reaches the customer. 10 = no detection method, 1 = certain detection

RPN = Severity × Occurrence × Detection. Maximum RPN = 1000. Prioritize actions on highest RPNs, but also address any item with Severity ≥ 9 regardless of RPN. After implementing countermeasures, re-score to verify RPN reduction.

Common pitfall: Treating FMEA as a one-time exercise. It must be a living document updated when designs change, new failure modes are discovered in production, or customer complaints reveal gaps.

### Corrective and Preventive Action (CAPA)
CAPA is the systematic process for investigating quality failures and preventing recurrence.

**8D Problem Solving**: D1 (form team), D2 (describe problem — IS/IS NOT analysis), D3 (interim containment — protect customer immediately), D4 (root cause analysis — 5 Whys, fishbone diagram, fault tree), D5 (choose permanent corrective action), D6 (implement and verify), D7 (prevent recurrence — update FMEA, control plans, work instructions), D8 (congratulate team).

Root cause analysis must go deep enough. If the root cause is "operator error," you haven't gone far enough. Why did the operator err? Missing training? Confusing work instruction? Tool that allows incorrect assembly? The countermeasure should make the correct action easy and the incorrect action impossible (poka-yoke / mistake-proofing).

## Industrial Automation

### PLC Programming
PLCs are the backbone of factory automation. They read inputs (sensors, switches, encoders), execute logic, and control outputs (motors, valves, actuators) in deterministic scan cycles (typically 1-20 ms).

IEC 61131-3 defines five programming languages: Ladder Diagram (LD — most common, relay-logic visual), Function Block Diagram (FBD), Structured Text (ST — high-level, like Pascal), Instruction List (IL — assembly-like, deprecated), Sequential Function Chart (SFC — state machine logic for sequential processes).

Best practices: modular program structure (main routine calls subroutines), consistent naming conventions (I:1/0 = input module 1, terminal 0), fault handling routines for every actuator (confirm movement within timeout, else alarm), and version control of PLC programs.

### SCADA Systems
SCADA provides centralized monitoring and control across the production facility. Architecture: field devices → PLCs → SCADA server → operator HMI (human-machine interface). Modern SCADA uses OPC UA (Unified Architecture) protocol for secure, platform-independent communication between PLCs and supervisory systems.

Key SCADA functions: real-time process visualization, alarm management (prioritized, acknowledged, with escalation), historical data logging (trends, batch records), recipe management (product-specific setpoints downloaded to PLCs).

### MES Integration
MES bridges the gap between ERP (business planning) and the shop floor (PLCs/SCADA). ISA-95 standard defines the functional hierarchy and data exchange between enterprise and manufacturing levels.

MES functions: production scheduling and dispatching, WIP tracking (part genealogy from raw material to finished product), quality data collection and SPC, labor tracking, equipment performance monitoring (OEE), and electronic batch records for regulated industries.

## Equipment Effectiveness

### OEE — Overall Equipment Effectiveness
OEE = Availability × Performance × Quality. It measures how effectively a machine produces good parts compared to its theoretical maximum.

**Availability** = Run time / Planned production time. Losses: unplanned downtime (breakdowns), planned downtime (changeovers, setups). Target: reduce changeover time with SMED methodology. Single-minute means under 10 minutes — achieve through separating internal setup (machine stopped) from external setup (done while machine runs), converting internal to external, and streamlining remaining internal steps.

**Performance** = (Ideal CT × Total pieces) / Run time. Losses: minor stops (jams, misfeeds), reduced speed (running below rated speed due to wear, vibration, or operator caution). Often the hardest loss to measure because operators don't record 30-second micro-stops.

**Quality** = Good pieces / Total pieces. Losses: scrap and rework. First-pass yield is the true measure — parts that pass inspection on the first attempt without rework.

World-class OEE is 85% (90% availability × 95% performance × 99.5% quality). Average manufacturing is 60%. The gap represents enormous untapped capacity — improving OEE from 60% to 85% is equivalent to buying 40% more equipment capacity at a fraction of the cost.

### TPM — Total Productive Maintenance
TPM shifts maintenance from reactive (fix when broken) to proactive. Eight pillars:

1. **Autonomous maintenance**: Operators perform daily cleaning, lubrication, inspection, and minor adjustments. They know the machine best and detect abnormalities earliest (unusual sounds, vibrations, temperatures, smells)
2. **Planned maintenance**: Scheduled preventive maintenance based on time or usage intervals. Predictive maintenance uses condition monitoring (vibration analysis, thermography, oil analysis) to replace parts based on actual condition rather than arbitrary schedules
3. **Focused improvement**: Cross-functional teams tackle chronic losses identified by OEE data. Use PDCA (Plan-Do-Check-Act) cycles
4. **Early equipment management**: DFM principles applied to equipment design. Maintenance-friendly access, standardized components, modular design for rapid replacement

Key metrics: MTBF (target: increasing trend), MTTR (target: decreasing trend), maintenance cost as percentage of replacement asset value (target: 2-3% for world-class), ratio of planned to unplanned maintenance (target: >90% planned).

## Production Planning

### Master Production Scheduling
The MPS translates demand forecasts and customer orders into a weekly/daily build plan. It must respect capacity constraints, material availability (BOM explosion through MRP), and inventory targets.

Rough-cut capacity planning validates the MPS against critical resource bottlenecks before detailed scheduling. If the bottleneck CNC department has 400 hours available per week and the MPS requires 450 hours, the plan is infeasible — adjust by leveling load across weeks, authorizing overtime, or subcontracting.

### Kanban Pull Systems
Kanban signals upstream processes to produce only when downstream consumes. Number of kanban cards = (daily demand × lead time × safety factor) / container size. Start with more cards than calculated (larger buffer), then systematically remove cards to expose and solve problems.

Two-bin system is the simplest kanban: when the first bin empties, it becomes the signal to replenish. Production continues from the second bin. Electronic kanban (e-kanban) replaces physical cards with digital signals from MES barcode scans, enabling real-time visibility and automated supplier signals.