A SYSTEMS APPROACH TOWARDS RISK INTERVENTION PRIORITIZATION IN MARITIME ENVIRONMENT Dr. T.A. Mazzuchi, Dr. J.R. van Dorp, Dr. J.R. Harrald Dr. J Merrick (VCU) Dr. M. Grabowski (RPI) Engineering Management & Systems Engineering Department 1 THE RISK OF RIVER BOAT GAMBLING A Risk Assessment for the Port of New Orleans Port Authority “Are you odds of winning better than your odds for dying” Joint Work: The George Washington University Rensselaer Polytechnic Institute 2 The Prince William Sound Risk Assessment A Risk Assessment for ADEC, APSC/SERVS, PWS Regional Citizens Advisory Council, US Coast Guard, PWS Shipping Companies) Joint Work: Det Norske Veritas The George Washington University Rensselaer Polytechnic Institute The Washington State Ferry Risk Assessment Washington State Department of Transportation The George Washington University Rensselaer Polytechnic Institute Virginia Commonwealth University The San Francisco Bay Traffic Density Analysis San Francisco Bay Ferry Associations The George Washington University Virginia Commonwealth University Citations of Work Presented “A Bayesian paired comparison approach for relative accident probability assessment with covariate information”, European Journal of Operational Research, Vol. 169, Issue 1, 2006, pp. 157-177. “A traffic density analysis of proposed ferry service expansion in San Francisco Bay using a maritime simulation model”, Reliability Engineering and System Safety, Vol. 81, Issue 2, 2003, pp. 119-132. "The Prince William Sound risk assessment", Interfaces, Vol. 32, No. 6, 2002, pp. 25-40. “A risk management procedure for the Washington State Ferries", Risk Analysis, Vol. 21, 2001, pp. 127-142. “Risk modeling in distributed, large scale systems", IEEE Transactions in Systems Man, and Cybernetics Part A: Systems and Humans, Vol. 30, 2000, pp.651-660. “A systems approach to managing oil transportation risk in Prince 6 William Sound", Systems Engineering, Vol. 3, 2000, pp. 128-142. Examples Risk Intervention Questions • Port of New Orleans Risk Assessment: “Is it safer for a gambling boat to be underway or at the dock?” • Prince William Sound Risk Assessment: “Should we tighten weather based closure restrictions for outbound tankers?” • Washington State Ferry Risk Assessment: “Is it (cost\risk) efficient to invest in addition survival craft capacity on Washington State Ferries?” • San Francisco Ferry Analysis: “Can the current system maintain an acceptable risk level under future modifications and projected traffic increase” 7 Maritime Accidents • • • • • • • COLLISION POWERED GROUNDING DRIFT GROUNDING ALLISION FOUNDERING STRUCTURAL FAILURE FIRE\EXPLOSION Stakeholders • Shipping (Oil/Passenger) Companies • US Coast Guard • US Departments of Transportation, Commerce, Interior and Environmental Protection Agency • Local Port Authorities • Fishing Industry • Pleasure Craft Industry • Environmentalist Groups • Local Community A Risk Assessment Approach for Dynamic Transportation Systems Organizational Risk Factors High Performing Low Performing Risk Averse Risk Prone Organizations Organizations Situational Risk Factors High Risk System States Low Risk System States • Organizational Risk Factors - influence the likelihood of the occurrence of triggering events. • Situational Risk Factors - influence the likelihood of occurrence of accidents given the occurrence of a triggering event. The Dynamic Risk Profile of the System • The situational and organizational factors lead to the dynamic profile of system risk. RISK 10-2 10-3 10-4 10-5 10-6 10-7 TIME FIGURE 3: Dynamic Risk Profile • The peak risk spikes in the system may to 100 to 1000 times riskier than the average system risk level. • Identifying how and when these risk spikes occur is a fundamental objective of the dynamic risk assessment methodology. The Maritime Accident Event Chain Stage 1 Basic/Root Causes E.g. Inadequate Skills, Knowledge, Equipment, Maintenance, Management Stage 2 Immediate Causes E.g. Human Error, Equipment Failure, Stage 3 Incident Stage 4 Accident E.g. E.g. Loss of Power, Collisions, Loss of Steering , Groundings, Dangerous Navigation Fire/Explosion ORGANIZATIONAL FACTORS Vessel type Flag/classification society Vessel age Management type/changes Pilot/officers on bridge Vessel incident/accident history Individual/team training Safety management system Stage 5 Consequence Stage 6 Delayed Consequence E.g. Oil Outflow, Persons in Peril E.g. Environmental Damage, Loss of Life SITUATIONAL FACTORS Type of waterway Wind Speed Traffic situation Wind Direction Traffic density Current Visibility Time of day Risk Reduction Interventions Stage 1 Basic/Root Causes E.g. Inadequate Skills, Knowledge, Equipment, Maintenance, Management Stage 2 Immediate Causes E.g. Human Error, Equipment Failure, Risk Reduction/ Prevention 1. Decrease Frequency of Root/Basic Causes Stage 3 Incident E.g. E.g. Loss of Power, Collisions, Loss of Steering, Groundings, Dangerous Navigation Fire/Explosion Risk Reduction/ Prevention 2. Decrease Frequency Immediate Causes Stage 4 Accident 3. Decrease Exposure to Hazardous Situations Stage 5 Consequence Stage 6 Delayed Consequence E.g. Oil Outflow, Persons in Peril E.g. Environmental Damage, Loss of Life Risk Reduction/ Prevention Risk Reduction/ Prevention 4. Intervene to Prevent Accident if Incident Occurs 5. Reduce Consequence (Oil Outflow) if Accident Occurs Risk Reduction/ Prevention 6. Reduce Impact if Oil Outflow Occurs E.g. E.g. E.g. E.g. E.g. E.g. Closure Conditions, Inspection Program, Emergency Repair or Double Hull, Pollution ISM, One-way Zone, Double Engine, Assist Tug, Double Bottom Response Vessel, Training, Traffic Sep. Scheme, Double Steering, Emergency Response Oil Boom, Better Traffic Management, Redundant Nav Coordination, Pollution Maintenance Nav. Aids for Poor Aids, VTS Watch Response Visibility Work Hour Limits, Coordination Drug/Alcohol Tests Data and The Maritime Accident Event Chain Stage 1 Basic/Root Causes Stage 2 Immediate Causes Stage 3 Incident SPARSE DATA Stage 4 Accident Stage 5 Immediate Consequence DATA BASES Stage 6 Delayed Consequenc e Modeling the Causal Chain: Collision Risk Vessel Attributes Waterway Attributes Opportunity for Incident Incident Pr(OFI) Pr(Incident|OFI) Simulation + Counting Model Collision Pr(Collision|Incident,OFI) Data on technological failures Expert Judgement on Human Error Data + effect of waterway attributes from expert judgment Information Flow Prince William Sound Simulation Quest. I & II Vessel Ops. Questionnaires Quest. III & IV Failure/Error Questionnaires Determination of Relative Incident Probabilities Quest. V & VI Calibration Questionnaires Vessel Reliability & Appropriate Incident Data Simulation Weather Data Traffic Data System Description Calibration of Vessel and Situational Relative Incident Probabilities Characterization of PWS Accident Profiles Modeling Traffic Movements Published Data VTS Way Point Data 17 Modeling Traffic Movements Rules of the Road Nuisance Traffic: Fishing Openers and Regattas 18 Modeling Rules of the Road and Weather (PWS Risk Assessment) 19 Continuous vs. Discrete System Risk PWS OFI = 5 minutes WSF OFI = 2.5 minutes SFF OFI = 1 minute Time 20 Interacting Vessels 21 OFI Counting Opportunity For Incident 10-mile radius probability density 2-mile radius 2 miles D 10 miles Distance 22 Not Every Interaction is the Same 1 Ferry, 1 Container Vessel Crossing Tracks Scenario 1 Ferry 2 Ferries, Parallel Tracks Scenario 2 Container Vessel 23 OFI Counting Model FRONT CROSSING BACK PASSING PASSING (OVERTAKING) (MEETING) - CROSSING BACK Vessel Ferry - FRONT 24 Modeling Conditional Failure Probabilities Using Expert Judgment Given a propulsion Failure, Asses the likelihood of Collision (PWS Risk Assessment) Traffic Type: Tug with Tow Traffic Prox.: Vessels 2 to 10 Miles Tanker Size & Direction: Inbound more than 150 DWT Wind Direction: Perpendicular/on Shore Wind Speed: More than 45 Visibility: Greater than 1/2 mile No Bergy Bits “System States” Bergy Bits within a mile 25 Expert Judgment Questionnaire (Example: WSF Risk Assessment Vessel Reliability Failure Will Lead to Collision?) Question: 2 Situation 1 Issaquah SEA-BRE(A) High Speed Ferry Crossing astern < 0.5 miles No Vessel No Vessel No Vessel > 0.5 Miles Along Ferry 0 9: 7: 5: 3: 1: 34 Attribute Ferry Class Ferry Route 1st Interacting Vessel Traffic Scenario 1st Vessel Traffic Proximity 1st Vessel 2nd Interacting Vessel Traffic Scenario 2nd Vessel Traffic Proximity 2nd Vessel Visibility Wind Direction Wind Speed Likelihood of Collision Avoidance 9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 VERY MUCH MORE LIKELY to allow collision avoidance. MUCH MORE LIKELY to allow collision avoidance. MODERATELY LIKELY to allow collision avoidance. SOMEWHAT MORE LIKELY to allow collision avoidance. EQUALY LIKELY to allow collision avoidance. Situation 2 Meeting - 26 Example Result Responses Question 51 Average Answer out of 6 experts : 5.23 Regression Fit: 5.29 2.5 Expert 10 Expert 9 2 # Responses Expert 8 Expert 7 1.5 Expert 6 1 Expert 5 Expert 4 0.5 Expert 3 0 -9 -8 -7 -6 -5 -4 -3 -2 1 2 Response Values 3 4 5 6 7 8 9 Expert 2 Expert 1 27 Accident Probability Model - Regression Traffic Scenario 1 X 1 Paired Comparison 1 2 Traffic Scenario 2 X Pr( Accident | Propulsion Failure, X ) P0 e 1 T X 1 T X 1 3 Pr( Accident | Prop . Failure, X ) P0 e T X 1 X 2 e 2 T X 2 Pr( Accident | Prop. Failure, X ) P0 e 4 Pr( Accident | Prop . Failure, X1 ) LN 2 Pr( Accident | Prop. Failure, X ) 1 2 T X 1 X 2 28 Example Regression Analysis Fit R2 of Regressions in the order of 75% to 80% Note: This is fit for representing expert data not fit to actual values 29 The responses can be enumerated through the use of the exponential risk equation Location Central Sound Traffic Proximity Vessels 2 to 10 Miles Traffic Type Tug with Tow Tanker Size & Direction Inbound More Than 150DWT Escort Vessels Two or more Wind Speed More Than 45 Wind Direction Perpendicular/On Shore Visibility Greater Than 1/2 Mile Ice Conditions Bergy Bits Within a Mile LIKELIHOOD OF COLLISION 98765432123456789 No Bergy Bits in a Mile Location Traffic Proximity Traffic Type Tanker Size & Direction Escort Vessels Wind Speed Wind Direction Visibility Ice Conditions Enumerating the exponential yields Relative Pr(Collision) = 313.2 Relative Pr(Collision) = 136.0 The risk model says ice in this condition is 2.3 times more dangerous What remains is to determine the scaling factor from data 30 Simulation Analysis Tool (PWS Risk Assessment) 31 Simulation Analysis Tool (WSF Risk Assessment) 32 Simulation Analysis Tool (SFF Analysis) 33 Prince William Sound Mitigation Analysis Case A Baseline Case Case B, B1 MINIMUM SAFEGUARDS Case 1 CAT I ROOT CAUSES Case 1.1 CAT IA & ID H/O ERROR Case 1.4 1A 2,3,4 BRIDGE TEAM Case 3 CAT III EXPOSURE Case 2 CAT II IMMEDIATE CAUSES Case 1.2 CAT IB & 1C VRF Case 1.5 1A1 DIM. ABILITY Case C, C1 MAXIMUM SAFEGUARDS Case B2 BASE CASE W/O ESCORTS Case 1.3 CAT ID INFORMATION Case 3.1 III 1.2 CLOSURE Case 3.2, 3.2A III 1.4 ICE Case 4, 4A CAR IV INTERVENTION Case 3.3 III3 FISHING VS Case 5A, 5B CAT V CONSEQUENCES Case 4.1 IV B2 TUG Case 4.2 IV A4 FAILURE CAPTURE Risk Reduction Cases Analyzed INTERVENTIONS Stage I Interventions: Reduce basic or root causes. Stage II Interventions: Decrease frequency of immediate (triggering) events Stage III Interventions: Decrease exposure to hazardous situations. Stage IV Interventions: Intervene to prevent accidents if incidents occur. Stage V Interventions: Reduce oil outflow if accidents occur. RISK REDUCTION CASES EVALUATED Case Case Case Case Case Case Case 1. Improve human and organizational performance (SS) 1.1 Reduce human and organizational error (M/FT, SS) 1.1.1 Improve training and navigation information (SS) 1.1.2 Improve bridge team management (SS) 1.1.3 Reduce incidence of diminished ability (SS) 1.2 Reduce vessel reliability failures (M/FT, SS) 2 Improve internal vigilance. (SS) Case 3 Reduce exposure to hazardous weather, ice, traffic conditions (SS) Case 3.1 Impose stricter closure conditions (M/FT, SS) Case 3.1.1 Impose stricter closure conditions at Hinchinbrook Entrance (SS) Case 3.2 Revise ice navigation procedures (FT,SS) Case 3.3 Revise fishing vessel/tanker interaction rules (SS) Case 4 Revise escort program using pre-positioned tugs (M/FT,SS) Case 4A Revise escort program, pre-position salvage tug at Hinchinbrook Entrance (M/FT, SS) Case 4.1 Tug in indirect mode in Narrows: (FT) Case 4.2 Capture/correct 50% of all steering and propulsion failures (M/FT) Case 5A Replace entire fleet with double hulled tankers.(M/FT) Case 5B Hydrostatically load all single hulled vessels in existing fleet (M/FT). 35 (M/FT) Displaying Results 36 Interactions by Route and Interacting Vessel Washington State Ferry Analysis NON - WSF WSF 37 Average Collision Probability per Interaction by Route and Interacting Vessel (WSF) NON - WSF WSF 38 Statistical Expected Number of Collisions per Year by Route and Interacting Vessel (WSF) NON - WSF WSF 39 Traffic Density Maps San Francisco Bay Analysis 40 Traffic Density Comparisons San Francisco Bay Analysis SFB Alternative 3 SFB Alternative 1 41 Expected Frequency Accident Scenarios Cumulative % of Total Accidents (PWS) Risk Mitigation Effectiveness - PWS Tug in Ind irect M o d e in Narro ws Only Hyd ro st at ic Lo ad ing Revised Esco rt & Fishing \ Tanker Rules Revised Fishing \ Tanker Rules Imp ro ved Human\ Org anizat io n Perf o rmance Red uce Pro p . & St eer. Failure b y 50 % Esco rt wit h Salvag e Tug Revised Esco rt Red uced Human Erro r Red uced V RF (A ll t o b est ) Imp ro ved M anag ement & Crew A d d it io nal Perso n o n t he B rid g e St rict er Clo sure at HE Imp ro ved Training & Navig at io n Inf o rmat io n Red uced Diminished A b ilit y Only Do ub le Hull B ase Revised Ice Pro d ecures Red uced Exp o sure St rict er Clo sure -100% -80% -60% -40% -20% 0% 20% 40% 60% 80% 100% Ave r age % Change in # Accide nts fr om Bas e Cas e DNV : Average % Change in # Accidents GWU : Average % Change in # Accidents 43 Risk Mitigation Effectiveness - WSF 44 Lessons Learned • Use of local experts is very important for acceptance • Experts can impart useful knowledge for risk analysis • Other data sources are always available • In many instances risk is a dynamic function of the system • Risk needs to be addressed system wide – avoid local focus • Risk Management questions must be established before risk modeling is conducted • Each system will have a certain uniqueness and new modeling challenges 45