Preface xv

Acknowledgements xix

Definitions of Risk and Risk Management xxi

Introduction: The Art of Prediction and the Creation of Order 1

Risk and Risk Management 1

Defining Risk 2

Managing Risk: Our Purpose, Plan and Goals 4

Recent Tragic Outcomes 6

Power Blackouts, Space Shuttle Losses, Concorde Crashes, Chernobyl, Three Mile Island and More 6

How Events and Disasters Evolve in a Phased Development: The Human Element 8

Our Values at Risk: The Probable Improvement 10

Probably or Improbably Not 11

How this Book is Organised 12

References 14

Technical Summary 15

Defining the Past Probability 15

Predicting Future Risk: Sampling from the Jar of Life 16

A Possible Future: Defining the Posterior Probability 21

The Engineers Have an Answer: Reliability 22

Drawing from the Jar of Life: The Hazard Function and Species Extinction 23

Experiencing Failure: Engineering and Human Risk and Reliability 25

Experience Space 27

Managing Safely: Creating Order out of Disorder Using Safety Management Systems 29

Describing the Indescribable: Top-Down and Bottom-Up 30

What an Observer will Observe and the Depth of our Experience 31

References 33

1 The Universal Learning Curve 35

Predicting Tragedies, Accidents and Failures: Using the Learning Hypothesis 35

The Learning Hypothesis: The Market Place of Life 37

Learning in Homo-Technological Systems (HTSs): The Way a Human Learns 39

Evidence of Risk Reduction by Learning 41

Evidence of Learning from Experience: Case Studies 42

Evidence of Learning in Economics 43

Evidence of Learning in Engineering and Architecture: The Costs of Mistakes 44

Learning in Technology: the Economics of Reducing Costs 46

Evidence of Learning Skill and Risk Reduction in the Medical Profession: Practice Makes Almost Perfect 48

Learning in HTSs: The Recent Data Still Agree 50

The Equations That Describe the Learning Curve 52

Zero Defects and Reality 54

Predicting Failures: The Human Bathtub 55

Experience Space: The Statistics of Managing Safety and of Observing Accidents 55

Predicting the Future Based on Past Experience: The Prior Ignorance 57

Future Events: the Way Forward Using Learning Probabilities 58

The Wisdom of Experience and Inevitability 59

The Last, First or Rare Event 59

Conclusions and Observations: Predicting Accidents 60

References 61

2 The Four Echoes 63

Power Blackouts, Space Shuttle Losses, Concorde Crashes, and the Chernobyl and Three Mile Island Accidents 63

The Combination of Events 64

The Problem Is the Human Element 65

The Four Echoes Share the Same Four Phases 66

The First Echo: Blackout of the Power Grid 67

Management’s Role 69

The First Echo: Findings 71

Error State Elimination 73

The Second Echo: Columbia/Challenger 75

The Results of the Inquiry: Prior Knowledge 76

The Second Echo: The Four Phases 79

Management’s Responsibility 80

Error State Elimination 82

The Third Echo: Concorde Tires and SUVs 83

Tire Failures: the Prior Knowledge 84

The Third Echo: The Four Phases 87

Management’s Responsibility 87

Error State Elimination 87

The Fourth Echo: Chernobyl 88

An Echo of Three Mile Island 88

The Consequences 92

Echoes of Three Mile Island 92

The Causes 93

Error State Elimination 94

The Fourth Echo: The Four Phases 95

Regulatory Environment and Practices 95

Case study: Regulation in Commercial Aviation 96

a) Regulations Development 96

b) Compliance Standards 97

c) Accident Investigation 97

Addressing Human Error 98

Management Responsibilities 99

Designing to Reduce Risk and the Role of Standards 99

Conclusion and Echoes: Predicting the Unpredictable 101

References 103

3 Predicting Rocket Risks and Refinery Explosions: Near Misses, Shuttle Safety and Anti-Missile Defence Systems Effectiveness 105

Learning from Near Misses and Prior Knowledge 105

Problems in Quantifying Risk: Predicting the Risk for the Next Shuttle Mission 107

Estimating a Possible Range of Likelihoods 112

Learning from Experience: Maturity Models for Future Space Mission Risk 114

Technology versus Technology 120

Missiles Risks over London: The German Doodlebug 121

Launching Missile Risk 124

The Number of Tests Required 126

Estimating the Risk of a Successful Attack and How Many Missiles We Must Fire 128

Uncertainty in the Risk of Failing to Intercept 128

What Risk Is There of a Missile Getting Through: Missing the Missile 131

Predicting the Risk of Industrial Accidents: The Texas City Refinery Explosion 132

From Lagging to Leading: Safety Analysis and Safety Culture 134

Missing Near Misses 137

What these Risk Estimates Tell Us: The Common Sense Echo 137

References 138

4 The Probability of Human Error: Learning in Technological Systems 141

What We Must Predict 141

The Probability Linked to the Rate of Errors 144

The Definition of Risk Exposure and the Level of Attainable Perfection 146

Comparison to Conventional Social Science and Engineering Failure and Outcome Rate Formulations 147

The Learning Probabilities and the PDFs 150

The Initial Failure Rate and its Variation with Experience 150

The ‘Best’ MERE Risk Values 153

Maximum and Minimum Likely Outcome Rates 155

Standard Engineering Reliability Models Compared to the MERE Result 155

Future Event Estimates: The Past Predicts the Future 157

Statistical Bayesian-Type Estimates: The Impact of Learning 158

Maximum and Minimum Likelihood 161

Comparison to Data: The Probability of Failure and Human Error 161

Comparison of the MERE Result to Human Reliability Analysis 164

Implications for Generalised Risk Prediction 168

Conclusions: The Probable Human Risk 170

References 171

5 Eliminating Mistakes: The Concept of Error States 173

A General Accident Theory: Error States and Safety Management 173

The Physics of Errors 174

The Learning Hypothesis and the General Accident Theory 176

Observing Outcomes 178

A Homage to Boltzmann: Information from the Grave 181

The Concept of Depth of Experience and the Theory of Error States 184

The Fundamental Postulates of Error State Theory 188

The Information in Error States: Establishing the Risk Distribution 189

The Exponential Distribution of Outcomes, Risk and Error States 192

The Total Number of Outcomes 193

The Observed Rate and the Minimum Number of Outcomes 195

Accumulated Experience Measures and Learning Rates 198

The Average Rate 200

Analogy and Predictions: Statistical Error Theory and Learning Model Equivalence 201

The Influence of Safety Management and Regulations: Imposing Order on Disorder 201

The Risk of Losing a Ship 203

Distribution Functions 205

The Most Probable and Minimum Error Rate 208

Learning Rates and Experience Intervals: The Universal Learning Curve 209

Reducing the Risk of a Fatal Aircraft Accident: the Influence of Skill and Experience 212

Conclusions: A New Approach 215

References 216

6 Risk Assessment: Dynamic Events and Financial Risks 219

Future Loss Rate Prediction: Ships and Tsunamis 221

Predicted Insurance Rates for Shipping Losses: Historical Losses 224

The Premium Equations 225

Financial Risk: Dynamic Loss and Premium Investments 226

Numerical Example 227

Overall Estimates of Shipping Loss Fraction and Insurance Inspections 228

The Loss Ratio: Deriving the Industrial Damage Curves 229

Making Investment Decisions: Information Drawing from the Jar of Life 231

Information Entropy and Minimum Risk 232

Progress and Learning in Manufacturing 233

Innovation in Technology for the Least Product Price and Cost: Reductions During Technological Learning 234

Cost Reduction in Manufacturing and Production: Empirical Elasticity ‘Power Laws’ and Learning Rates 235

A New General Formulation for Unit Cost Reduction in Competitive Markets: the Minimum Cost According to a Black-Scholes Formulation 237

Universal Learning Curve: Comparison to the Usual Economic Power Laws 240

The Learning Rate b-Value ‘Elasticity’ Exponent Evaluated 242

Equivalent Average Total Cost b-Value Elasticity 244

Profit Optimisation to Exceed Development Cost 246

The Data Validate the Learning Theory 247

a) Aircraft Manufacturing Costs Estimate Case 247

b) Photovoltaic Case 248

c) Air Conditioners Case 250

d) Ethanol Prices Case 251

e) Windpower Case 252

f) Gas Turbine Power Case 253

g) The Progress Curve for Manufacturing 254

Non-Dimensional UPC and Market Share 256

Conclusions: Learning to Improve and Turning Risks into Profits 259

References 260

7 Safety and Risk Management Systems: the Fifth Echoes 263

Safety Management Systems: Creating Order Out of Disorder 263

Workplace Safety: The Four Rights, Four Wrongs and Four Musts 264

Acceptable Risk: Designing for Failure and Managing for Success 265

Managing and Risk Matrices 269

Organisational Factors and Learning 272

A Practical ‘Safety Culture’ Example: The Fifth Echo 273

Safety Culture and Safety Surveys: The Learning Paradox 278

Never Happening Again: Perfect Learning 280

Half a World Apart: Copying the Same Factors 281

Using a Bucket: Errors in Mixing at the JCO Plant 283

Using a Bucket: Errors in Mixing at the Kean Canyon Explosives Plant 284

The Prediction and Management of Major Hazards: Learning from SMS Failures 286

Learning Environments and Safety Cultures: The Desiderata of Desires 289

Safety Performance Measures: Indicators and Balanced Scorecards 291

Safety and Performance Indicators: Measuring the Good 292

Human Error Rates Passing Red Lights, Runway Incursions and Near Misses 293

Risk Informed Regulation and Degrees of Goodness: How Green is Green? 294

Modelling and Predicting Event Rates and Learning Curves Using Accumulated Experience 297

Using the Past to Predict the Future: How Good is Good? 299

Reportable Events 300

Scrams and Unplanned Shutdowns 301

Common-Cause Events and Latent Errors 303

Performance Improvement: Case-by-Case 304

Lack of Risk Reduction: Medical Adverse Events and Deaths 305

New Data: Sentinel Events, Deaths and Blood Work 308

Medication Errors in Health Care 313

Organisational Learning and Safety Culture: the ‘H-Factor’ 316

Risk Indicator Data Analysis: A Case Study 319

Meeting the Need to Measure Safety Culture: the Hard and the Soft Elements 321

Creating Order from Disorder 324

References 324

8 Risk Perception: Searching for the Truth Among all the Numbers 329

Perceptions and Predicting the Future: Risk Acceptance and Risk Avoidance 329

Fear of the Unknown: The Success Journey into What We Do or Do Not Accept 333

A Possible Explanation of Risk Perception: Comparisons of Road and Rail Transport 334

How Do We Judge the Risk? 337

Linking Complexity, Order, Information Entropy and Human Actions 338

Response Times, Learning Data and the Universal Laws of Practice 341

The Number and Distribution of Outcomes: Comparison to Data 343

Risk Perception: Railways 345

Risk Perception: Coal Mining 348

Risk Perception: Nuclear Power in Japan 349

Risk Perception: Rare Events and Risk Rankings 352

Predicting the Future Number of Outcomes 354

A Worked Example: Searching out and Analysing Data for Oil Spills 354

Typical Worksheet 358

Plotting the Data 358

Fitting a Learning Curve 358

Challenging Zero Defects 359

Comparison of Oil Spills to Other Industries 362

Predicting the Future: the Probability and Number of Spills 364

Observations on this Oil Spill Case 365

Knowing What We Do Not Know: Fear and Managing the Risk of the Unknown 365

White and Black Paradoxes: Known Knowns and Unknown Unknowns 367

The Probability of the Unknowns: Learning from What We Know 368

The Existence of the Unknown: Failures in High Reliability Systems 370

The Power of Experience: Facing Down the Fear of the Unknown 371

Terrorism, Disasters and Pandemics: Real, Acceptable and Imaginary Risks 373

Estimating Personal Risk of Death: Pandemics and Infectious Diseases 374

Sabotage: Vulnerabilities, Critical Systems and the Reliability of Security Systems 377

What Is the Risk? 378

The Four Quadrants: Implications of Risk for Safety Management Systems 378

References 380

9 I Must Be Learning 383

Where We Have Come From 383

What We Have Learned 384

What We Have Shown 388

Legal, Professional and Corporate Implications for the Individual 389

Just Give Me the Facts 391

Where We are Going 392

Reference 393

Nomenclature 395

Appendices: 401

Appendix A: The ‘Human Bathtub’: Predicting the Future Risk 403

The Differential Formulation for the Number of Outcomes 405

The Future Probability 406

Insufficient Learning 408

Appendix B: The Most Risk, or Maximum Likelihood, for the Outcome (Failure or Error) Rate while Learning 411

The Most or Least Likely Outcome Rate 411

The Maximum and Minimum Risk: The Two Solutions 412

Low Rates and Rare Events 413

The Limits of Maximum and Minimum Risk: The Two Solutions 414

Common Sense: The Most Risk at the Least Experience and the Least Risk as the First Outcome Decreases with Experience 414

Typical Trends in Our Most Likely Risk 415

The Distribution with Depth of Experience 417

References 418

Appendix C: Transcripts of the Four Echoes 419

Power Blackout, Columbia Space Shuttle loss, Concorde Crash and Chernobyl Accident 419

The Combination of Events 419

The Four Echoes Share the Same Four Phases 420

Appendix. Blackout Chronology and the Dialog from Midday 14 August 2003 420

The Second Echo: Columbia/Challenger 432

Appendix: Shuttle Dialog and Transcripts 433

The Third Echo: Concorde Tires and SUVs 435

Appendix: Dialog for the Concorde Crash 436

The Fourth Echo: TMI/Chernobyl 439

Appendix: Chronology and Transcripts of the Chernobyl Reactor Unit 4 Accident 439

Conclusion and Echoes: Predicting the Unpredictable 444

Appendix D: The Four Phases: Fuel Leak Leading to Gliding a Jet in to Land without any Engine Power 447

The Bare Facts and the Sequence 447

The Four Phases 449

Flight Crew Actions 455

Initial Recognition of the Fuel Loss (04:38–05:33) 455

Crew Reaction to the Fuel Imbalance Advisory (05:33–05:45) 456

Crew Reaction to the Continued Fuel Loss (05:45–06:10) 458

Crew Reaction to the (Two) Engine Failures 460

References 463

Appendix E: The Four Phases of a Midair Collision 465

The Bare Facts 465

The Four Phases 465

References 469

Appendix F: Risk From the Number of Outcomes We Observe: How Many are There? 471

The Number of Outcomes: The Hypergeometric Distribution 472

Few Outcomes and many Non-Outcomes: The Binomial and Poisson Distributions 475

The Number of Outcomes: In the Limit 478

The Perfect Learning Limit: Learning from Non-Outcomes 479

The Relative Change in Risk When Operating Multiple Sites 481

References 482

Appendix G: Mixing in a Tank: The D.D. Williamson Vessel Explosion 483

Errors in Mixing in a Tank at the Caramel Factory: The Facts 483

The Prior Knowledge 484

Another Echo 488

References 490

Appendix H: Never Happening Again 491

The Risk of an Echo, or of a Repeat Event 491

The Matching Probability for an Echo 493

The Impact of Learning and Experience on Managing the Risk of Repeat Events 494

The Theory of Evidence: Belief and Risk Equivalence 496

References 497

Appendix I: A Heuristic Organisational Risk Stability Criterion 499

Order and Disorder in Physical and Management Systems 499

Stability Criterion 500

References 502

Appendix J: New Laws of Practice for Learning and Error Correction 505

Individual Learning and Practice 505

Comparison to Error Reduction Data 506

Comparison to Response Time Data and the Consistent Law of Practice 509

Reconciling the Laws 511

Conclusions 512

References 513

Appendix K: Predicting Rocket Launch Reliability – Case Study 515

Summary 515

Theory of Rocket Reliability 515

a) Unknown Total Number of Launches and Failures 516

b) Known Total Number of Launches and Failures 517

Results 518

Measures of Experience 519

Comparsion to World Data 520

Predicting the Probability of Failure 521

Statistical Estimates of the Failure Probability for the Very ‘Next’ Launch 523

Independent Validation of the MERE Launch Failure Curve 525

Observations 526

References 526

Index 527