CHAPTER 6
Technical Approach to the Aleutian Islands Risk Assessment

This chapter details the technical approach proposed by the committee for conducting the Aleutian Islands risk assessment. The first section describes the first five steps in the Phase A Preliminary Risk Assessment, aimed at characterizing the risk of accidental spills from maritime transportation in the Aleutian Islands region. These semiquantitative portions of the Phase A assessment, such as traffic characterization and projections and estimation of spill rates and sizes, will rely heavily on historical data and, where appropriate, experience from prior risk studies and expert opinion. The traffic and spill risk studies will help the Management Team and Advisory Panel identify geographic locations and spill scenarios for a limited number of focused environmental impact investigations, to be carried out in the Phase A consequence analysis. As noted in Chapter 5, avoiding extensive simulations and modeling and limiting the extent of the consequence assessment will control the cost of the Phase A study, allowing the majority of resources to be concentrated on the focused assessment of spill prevention and mitigation measures.

The Phase A work should yield a basic understanding of where the highest risks lie with respect to the types of hazardous substances and vessels involved, the types of accidents and the likely



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 110
CHAPTER 6 Technical Approach to the Aleutian Islands Risk Assessment This chapter details the technical approach proposed by the com- mittee for conducting the Aleutian Islands risk assessment. The first section describes the first five steps in the Phase A Preliminary Risk Assessment, aimed at characterizing the risk of accidental spills from maritime transportation in the Aleutian Islands region. These semiquantitative portions of the Phase A assessment, such as traffic characterization and projections and estimation of spill rates and sizes, will rely heavily on historical data and, where appro- priate, experience from prior risk studies and expert opinion. The traffic and spill risk studies will help the Management Team and Advisory Panel identify geographic locations and spill scenarios for a limited number of focused environmental impact investigations, to be carried out in the Phase A consequence analysis. As noted in Chapter 5, avoiding extensive simulations and modeling and limiting the extent of the consequence assessment will control the cost of the Phase A study, allowing the majority of resources to be concentrated on the focused assessment of spill prevention and mitigation measures. The Phase A work should yield a basic understanding of where the highest risks lie with respect to the types of hazardous sub- stances and vessels involved, the types of accidents and the likely 110

OCR for page 110
Technical Approach to the Aleutian Islands Risk Assessment • 111 causes of and scenarios leading to those accidents, the sizes and likely locations for spills, and the environmental impact of those spills. The intent of these studies is to provide the Management Team and Advisory Panel with sufficient information with which to identify and prioritize risk reduction measures on a qualitative basis during the continuation of Phase A. The committee wishes to emphasize that the ultimate goal of the risk assessment is to identify measures that can be taken to reduce the risk of vessel accidents and spills in the Aleutian region. As discussed below, the committee has compiled an initial list of risk reduction options than can be used by the Management Team and the Advisory Panel as a starting point for working toward this goal. Although the risk assessment is structured into discrete phases and steps, the iden- tification and prioritization of risk reduction options should be an ongoing, iterative process throughout all of these efforts, reflecting analysis results as they become available, changing circumstances, and emerging technologies and opportunities. The second section of this chapter describes the Phase A effort to evaluate the identified risks, develop a list of potential risk reduction measures, and prioritize those measures. On the basis of this quali- tative assessment, the Management Team, in collaboration with the Advisory Panel, may be able to identify certain measures as appro- priate for immediate implementation. Some measures will be dis- carded as unjustifiable, and others will be designated for more detailed analysis. The third section of the chapter describes the approaches and typical techniques to be applied in the more detailed, quan- titative analyses of Phase B that are needed to justify certain mea- sures and understand their secondary effects on the overall system. These analyses will likely involve examination of a variety of risk reduction options, numerical simulations, and elicitation of infor- mation from expert witnesses to quantify the likelihood and con- sequences of the accidents identified in Phase A with and without the risk reduction measures in place. Uncertainty and sensitivity analyses will provide a sense of the confidence warranted in the characterization of risks and the benefits to be realized. The quan- titative assessment also will supply data needed for cost–benefit analyses. The final section describes the steps needed to develop and report final recommendations for decision makers on the risk reduction

OCR for page 110
112 • Risk of Vessel Accidents and Spills in the Aleutian Islands measures to be implemented. There are many challenges to imple- mentation, including establishing sources of funding and reach- ing agreement with the various agencies and stakeholders that will influence the failure or success of a measure. The risk manage- ment process is not a one-time solution; it requires continuous monitoring and reassessment. Thus, the need for a mechanism to ensure that the risk management plan remains a living document is discussed. PHASE A PRELIMINARY RISK ASSESSMENT: SEMIQUANTITATIVE STUDIES The purpose of this portion of the Phase A Preliminary Risk Assess- ment is to identify the more significant risks related to spills from shipping and provide a basis for the identification and initial rank- ing of possible risk reduction measures. These semiquantitative studies are intended to provide a high-level understanding of relative risks, taking into consideration types of vessels and hazard- ous substances and locations where discharges are most likely to occur. Results of these studies should allow the Advisory Panel and Management Team to perform a preliminary ranking of risk control measures, taking into account such factors as effectiveness, amena- bility to implementation through regulatory or voluntary means, and costs. Some measures assessed as having a high level of effective- ness may be proposed for immediate implementation. For those measures for which the justification for implementation is not conclusive, more detailed assessments will likely be required to bet- ter quantify the likelihood of the accidents that would be addressed and the extent to which the measures would reduce their frequency and consequences. These additional analyses would be undertaken in Phase B. The Preliminary Risk Assessment should utilize relatively sim- ple tools, avoiding detailed event tree analyses and complex simu- lation models to the extent practical. These Phase A studies will rely primarily on historical data, expert opinion, and lessons learned from prior studies. Some of the initial analyses will be qualitative in nature, with increasing levels of quantification in subsequent analyses as necessary.

OCR for page 110
Technical Approach to the Aleutian Islands Risk Assessment • 113 The basic steps in this portion of the Phase A assessment are listed below and described in more detail in the ensuing text. 1. Traffic study: Perform a vessel traffic study to characterize the exist- ing fleet and traffic in the region and the quantities of oil and other hazardous cargoes being moved. Project growth in trade, changes in vessel design and size, and the impacts of known and reason- ably expected regulatory changes. Use this information to project the fleet makeup over the study period. 2. Spill baseline study: Develop the spill baseline over the 25-year study period as the product of the projected movements of oil and other hazardous cargoes and the estimated average spill rates. Frequency is developed in terms of accident return period for each type of ship and accident. Consequence is initially expressed in terms of the expected or average spill outflow, which together with the spill frequency defines the spill rate. This projection will provide an understanding of the most important hazards and serve as a baseline for later assessment of benefits. Related information, such as the maximum expected outflow (upper limit), type of substance spilled, and safety implications in terms of loss of life and serious injuries, is developed to assist in the Step 3 and Step 5 assessments. 3. Characterization of spills from the highest-risk accidents: Taking into consideration the traffic and baseline spill analysis, identify the hazardous substances, representative spill sizes, and locations of spills from the highest-risk accidents. This information will be used as input for the Phase A consequence analysis (Step 4 below). Determine which accidents (types of accidents, predominant ves- sel types, geographic locations) are of sufficient concern to merit assessment of risk reduction measures. This information will be used during the brainstorming of potential risk reduction mea- sures and as input into the accident scenario and causality analysis (Step 5 below). 4. Phase A consequence analysis: Perform a preliminary spill tra- jectory and fate analysis for the spills and locations identified in Step 3 above. The intent is to gain an understanding of the rela- tive environmental consequences of spill size, type of hazardous substance spilled, and spill location. Perform a qualitative assess- ment of the potential resource damage and socioeconomic impact of these representative spills.

OCR for page 110
114 • Risk of Vessel Accidents and Spills in the Aleutian Islands 5. Accident scenario and causality study: For the dominant accident types identified in Step 3, determine representative accident sce- narios. Develop probabilities for the principal causes and associated consequences of the significant scenarios. Where possible, historical data should be used to determine traf- fic and commodity flows, as well as the likelihood and size of spills. Care must be taken when applying these historical data. Reporting standards are rarely consistent within a given database, and no single database is comprehensive. Although data specific to the local region are generally preferred, the sparseness of accident and spill data for large vessels in the study region will necessitate use of national and international data on spill frequency and size to gen- erate statistically significant estimates. When data are unavailable or characterized by considerable uncertainty, the use of expert judg- ment, simulations, and other analytical models may be required. For instance, drift grounding simulation may be needed to gain an understanding of the likelihood of a disabled vessel drifting aground, particularly for vessels transiting to the south of the Aleutian Islands. However, the use of simulations or expert opinion to pre- dict the likelihood of major spill events should be minimized to the extent possible. The uncertainty of the estimates derived should be carefully assessed, and sensitivity analyses should be carried out as appropriate. The baseline projection developed in Step 2 should assume full implementation of the Oil Pollution Act of 1990 (OPA 90) and International Maritime Organization (IMO) regulations that have already been adopted. Examples of regulations that will affect the environmental performance of ships built during the study period include the International Convention for the Prevention of Pollution from Ships (MARPOL), Annex I, Regulation 23, Accidental Oil Outflow Performance, which specifies subdivision requirements for the cargo spaces of oil tankers, and MARPOL, Annex I, Regulation 12A, Oil Fuel Tank Protection, which specifies double-hull or equivalent protection for fuel tanks. The baseline projection should also account for future regulations that can reasonably be anticipated. For example, it is expected that IMO will implement air emission regulations that will mandate increased use of nonpersistent fuel oils. The baseline projection should assume that no additional risk reduction interventions or measures will be implemented during the study period. Thus, the baseline

OCR for page 110
Technical Approach to the Aleutian Islands Risk Assessment • 115 will represent a hypothetical future without the potentially ben- eficial effects of the risk reduction options being investigated in the Aleutian Islands risk assessment. (Future benefits from oper- ational requirements such as International Safety Management and Standards of Training, Certification, and Watchkeeping should be considered only if compliance with those requirements can be fully documented and quantified.) 1. Traffic Study 1a. Determine the makeup and traffic patterns of the fleet transiting the Aleutian Islands or operating within the study region. An analysis of traffic though the study area should be developed on the basis of the best available data. As described in Chapter 3, automatic identification system (AIS) vessel tracking data have been compiled for transits through Unimak Pass since 2006. These data provide the most accurate information on the number, types, and routing of larger vessels transiting the Great Circle Route through the Aleutians. Data for ships transiting immediately south of the Aleutian chain are less reliable. (Satellite AIS data would be useful if available as an additional data source for the assessment.) To the extent practical, other data sources, such as the U.S. Coast Guard (USCG) Puget Sound vessel tracking sys- tem (VTS) and the Canadian Coast Guard Tofino VTS should be used. These data sets provide tracking information for vessels arriving at and departing from the Seattle and Vancouver areas, respectively, and should provide an indication of the routing of vessels calling on these regions and whether tank vessels are laden or in ballast. Communication with weather routing services and shipping companies may also be required to augment these data. Determinations of concentrations of fishing vessels, locations of seafood processors, movements of barges transporting refined products to the outer Aleutian Islands, and other local vessel movements will require review of local data sources, such as the Marine Exchange and the Alaska Commercial Fisheries Entry Commission, as well as communication with pilots and industry representatives. The various data sources should be used to develop best esti- mates of traffic for vessels carrying at least 10,000 gallons of fuel or other oil product or significant quantities of hazardous cargo. These

OCR for page 110
116 • Risk of Vessel Accidents and Spills in the Aleutian Islands estimates should provide a picture of traffic patterns, categorized by vessel type, amounts and types of hazardous substances (e.g., persis- tent oil, nonpersistent petroleum products, hazardous chemicals), and seasonality. Ship data should be evaluated to determine design characteristics required for the risk analysis, such as the percentage of single-hull versus double-hull tank vessels, the extent of double- hull protection provided for fuel tanks, and the range of bunker tank capacities applicable to the various vessel types. The categories of vessel types and sizes should be sufficiently fine-grained to allow assessment of measures that may be particu- lar to a given trade or vessel type. The committee envisions that the vessel categories will include at least the following: product tankers (laden and in ballast), crude oil carriers (laden and in bal- last), tank barges (laden and in ballast), liquefied natural gas car- riers, containerships of less than 4,500 twenty-ft equivalent units (TEUs), containerships of more than 4,500 TEUs, bulk carriers of less than 60,000 tonnes deadweight tonnage, bulk carriers of more than 60,000 tonnes deadweight tonnage, roll-on/roll-off vessels and vehicle carriers, other cargo ships, government vessels, fishing vessels, tugboats, and other smaller vessels. 1b. Estimate the current movements of cargo oils, containers, bulk car- goes, bulk chemicals, and other commodities through the study region, and develop yearly estimates for the movement of cargoes through the region over the 2009–2034 study period. Commodity movements through the Aleutian Islands should be estimated on the basis of fleet and traffic data, together with data from the various national port databases documenting trade. For instance, the U.S. Army Corps of Engineers compiles statistical data on waterborne commerce covering vessels that call on U.S. ports, and Statistics Canada maintains a similar database for Canadian ports. Considerable uncertainty exists because of global climate change and peak oil concerns, and alternative growth scenarios (e.g., new oil and gas fields) should be investigated (NRC 2007). Historical growth in trade should also be reviewed. To forecast oil and dry cargo transport quantities for the period 2009–2034, data should be solicited from the various trade organizations, the U.S. Department of Energy, the Maritime Administration, ports, and other sources.

OCR for page 110
Technical Approach to the Aleutian Islands Risk Assessment • 117 1c. Project the fleet makeup over the study period, anticipating likely changes in vessel size and design. Over the 25-year study period, changes in the design of ships tran- siting the Aleutian Islands can be anticipated. For example, only a few containerships greater than 8,000 TEUs in size are currently in operation, but more than 100 ultralarge containerships are on order, ranging in capacity from 8,000 to 13,000 TEUs. The growth in ship size may reduce the number of vessels trading, but the average fuel tank capacity of ships will increase. Because container- ships represent a significant portion of the vessels on innocent passage through Unimak Pass, the growth in vessel size will have a bearing on longer-term spill risks. Regulations adopted by IMO and applicable to the international fleet also will influence the design and arrangement of ships. These regulations may apply only to newly constructed vessels, or if appli- cable to existing vessels, may have a phase-in period. The impact of these regulations on ships expected to transit the Aleutians during the study period should be considered: • By 2009, the OPA 90 and MARPOL double-hull regulations for tankers will have been largely implemented. Any further phase-in of double hulls should be considered. • For large commercial vessels, the majority of fuel oil tanks are arranged adjacent to the side or bottom shell. For new vessels con- tracted for after 2008, MARPOL Regulation 12A requires that the fuel tanks be double-hulled or that the tank arrangement be analyzed to demonstrate an equivalent level of expected mean outflow from accidents. • A previous Transportation Research Board study (TRB 2001) found that certain double-hull tankers, particularly those with single-tank-across configurations, are prone to large accidental oil outflows. IMO subsequently implemented MARPOL Regu- lation 23, which requires all newly built tankers to meet specified outflow performance requirements. 1d. Develop yearly estimates for vessel traffic and the movements of ship’s fuel oil (bunkers), cargo oil, and hazardous chemicals through the study region for the 2009–2034 period. Forecast growth in the fishing fleet. The understanding of existing vessel traffic gleaned from Step 1a above, the forecasts of growth in trade and commerce derived from

OCR for page 110
118 • Risk of Vessel Accidents and Spills in the Aleutian Islands Step 1b, and the characterization of the future fleet obtained from Step 1c should be used to project the traffic flow and fleet makeup for the study period. When projecting movements of petroleum products, consideration should be given to the anticipated increase in exploration for and production of gas and oil in the Bering Sea, the Chukchi Sea, and other Arctic regions. 2. Spill Baseline Study 2a. Estimate the spill frequency and projected spill size distribution by vessel type. The accident types most likely to lead to large spills are collisions, powered and drift groundings, structural failures, and fires and explo- sions. In the Aleutian waters, groundings and collisions, particularly those occurring during inclement weather, are frequently the cause of large oil spills. Major spills from drift groundings in the waters around Unimak Pass (e.g., the M/V Kuroshima and M/V Selendang Ayu incidents) have heightened public awareness of and concern about drift groundings. The spill baseline study should include the following accident types: collisions, drift groundings, powered groundings, allisions, struc- tural failures, founderings, and fires and explosions. Historical spill statistics for the study area should be used to determine the distri- bution of spill sizes and the frequency of accident scenarios leading to the outflow of oil and other hazardous cargoes. Data from USCG, the State of Alaska, and salvors, as well as other local records, should be reviewed. Given the scarcity of significant spill events in the region, it will be necessary to augment the local spill data with data on U.S. and international spill events. Because of the scarcity of data and the evolution of ship designs, it will be necessary to use expert opinion and limited numerical simulations to determine accident frequency. The scarcity of data on outflow from cargo tanks on double-hulled tankers as well as double- hulled bunker tanks means that probabilistic oil outflow analysis based on historical damage data or simulation will likely be needed to develop spill size distributions for collisions and groundings. These estimates should be verified against historical data for reasonableness. The overall estimate of spillage should be subdivided among major ship categories. At a minimum, the following categories should be considered: tank ships, tank barges, containerships, other large com-

OCR for page 110
Technical Approach to the Aleutian Islands Risk Assessment • 119 mercial vessels, fishing vessels, and other small craft. Separate statistics should be provided for persistent and nonpersistent oils. Multiplying the frequency of spills by the average spill size will yield an overall estimate of spillage (in terms of barrels per year). This spill study should also provide the information needed for Step 3 and Step 5 assessments. The types of accidents and the ves- sels involved should be mapped against indicators of consequence, such as the types of hazardous substances spilled, the distribution of spill size, the likely locations of spills, the seasonality (likely time of year) of spills, and safety implications in terms of loss of life and serious injuries. 2b. Develop the oil spill baseline over the 25-year study period as the product of the projected movements of oil and other hazardous materials and the estimated average spill rates. The product of the projected quantities of oil and other hazardous materials moved over the 25-year study period by each vessel type and the spill rate for that vessel type provides the oil spill baseline. 3. Characterization of Spills from the Highest-Risk Accidents Using the findings of the traffic and baseline spill studies, the Risk Analysis Team should produce a matrix that identifies for the higher-risk accidents the following information: • Type of accident (e.g., drift grounding, collision), • Type of vessel involved (e.g., containership, tank barge, fishing boat), • Type of hazardous substance spilled (e.g., heavy fuel oil, marine gas oil), • Representative spill sizes (50th and 95th percentile spill volumes), • Likely geographic locations, and • Seasonality (likely time of year). In the Phase A consequence assessment (Step 4 below), spill trajectory studies will be performed to assist in assessment of the environmental and socioeconomic impacts of spills. Each combi- nation of inputs from the above list represents a single assessment. On the basis of available resources, it is anticipated that between 10 and 15 such assessments can be carried out in the Phase A con- sequence analysis. The Management Team and Advisory Panel, in

OCR for page 110
120 • Risk of Vessel Accidents and Spills in the Aleutian Islands consultation with the Risk Analysis Team, should select a represen- tative mix of spill events for the Phase A consequence analysis and identify the dominant accident types. The latter will be subject to further causality analysis and will receive the most attention when the Advisory Panel deliberates on potential risk reduction measures and their effectiveness. 4. Phase A Consequence Analysis Although spill size serves as an indicator of consequences, it does not by itself define consequences to the extent that it can be used to compare reliably the risk posed by certain accidents and risk control measures. The type of oil or other hazardous substance, the location of the spill, and the time of year the spill occurs influ- ence the extent of damage to natural resources, cleanup costs, and socioeconomic costs, and they should be considered along with spill size when consequences are evaluated. To illustrate the importance of substance type, spills of persis- tent oils, such as the heavy fuel oil used for bunkers of large com- mercial ships, have properties different from those of the diesel oil and marine gas oil used for propulsion of smaller craft, such as fish- ing boats. The lighter refined products are more volatile, and their evaporation reduces the amount of oil remaining on the surface. Compared with spills of heavy oil, spills of diesel oil and marine gas oil generally have much lower cleanup and socioeconomic costs. Spills of diesel oil and marine gas oil also generally have less impact on seabirds and mammals, cause less shoreline contamination, and have lower cleanup costs than spills of heavier oils. On the other hand, the lighter oils dissolve and disperse more readily into the water column and can be expected to have greater impacts on fish and invertebrates in the water and on demersal fish and invertebrates in the benthic zone. Likewise, the impacts and costs of spills are highly area depen- dent. Those impacts and costs are influenced by a range of factors, such as environmental conditions (tide, current, wind, sea state), sensitivity and exposure of natural resources, and the extent of eco- nomic and societal reliance on the sea and coastal regions. To pro- vide an understanding of the relative influence of substance type, spill size, and location on spills in the study region, a scoping spill consequence analysis should be performed as part of the Phase A Preliminary Risk Assessment. At this stage, the consequence analysis

OCR for page 110
Technical Approach to the Aleutian Islands Risk Assessment • 131 these cases, the Management Team should document the fact that uncertainty in the estimates of frequencies and benefits have been accounted for. This could be achieved, for example, by demonstrating that assumptions are conservative. Other criteria could also be used for prioritization. The cost– benefit ratio is important and USCG is one of the federal agencies that is required to consider costs and benefits in risk prioritiza- tion, but it might be useful to augment this with a consideration of uncertainty because a highly promising strategy may not merit additional investigation if the Phase A analysis provides defini- tive evidence regarding its efficacy. In contrast, if there is another strategy that has a moderate expected value but high uncertainty (it could be extremely effective, but there is not enough evidence to judge), this strategy might be a higher priority for investigation in Phase B. Because one goal of Phase B is to reduce uncertainty and to identify the most promising strategies, it might be effective to prioritize strategies on the basis of risk reduction potential. At this stage, it will be possible to screen some measures out as insignificant or ineffective. Other measures will be identified as demonstrating sufficient promise, and these will be assessed in detail in the Phase B Focused Risk Assessment. PHASE B FOCUSED RISK ASSESSMENT The Phase B Focused Risk Assessment involves a more in-depth look at the potential risk reduction measures identified in Phase A, aimed at quantifying their benefits and costs and better understand- ing their secondary effects on both the overall system and the net benefits of other measures. The results of this effort should provide the Management Team and Advisory Panel with sufficient informa- tion to make recommendations with regard to the implementation of risk reduction measures and produce a report documenting the justification for these recommendations. The potential risk reduction measures identified during Phase A as warranting further study will largely dictate the scope of the Phase B assessment. Therefore, this section does not explicitly define the Phase B study, but rather describes the expectations for the effort and some of the tools and techniques that can be applied in carrying it out.

OCR for page 110
132 • Risk of Vessel Accidents and Spills in the Aleutian Islands When preparing the request for proposals for the Phase B assess- ment, the Management Team should balance the level of rigor and sophistication to be specified for the analyses with the needs of the assessment. In general, the depth of the analyses should be com- mensurate with the level of the risks being investigated. Resources must be directed at accomplishing the overall goal of the risk assessment—to reduce risk of spills to an acceptable level. Phase B can be accomplished in discrete steps as necessary on the basis of the priority of the measures to be investigated and the level of risk reduction possible. The level of rigor needed and the types of analysis techniques to be applied may depend not only on the level of risk being investi- gated but also on the complexity of the system, the role of human factors in the dominant scenarios, the uncertainty of the analysis techniques, and the extent of historical precedents and the availabil- ity of similar studies. The particular geographic area may also be a factor. Dynamic risk models incorporating comprehensive system simulations generally provide the most insight into system behavior, the influence of system changes (e.g., increases in vessel movements), and unexpected consequences of changes to the system. The dynamic nature of the system should not be underestimated. The quantities and types of cargo movements, the design of vessels, weather and environmental conditions, and regulatory requirements are some of the many factors undergoing constant changes. This complex system must be modeled with sufficient rigor and evaluated over an extended period of time to capture the implications of these changes for the risk of spills from vessels. Dynamic risk models are expensive to develop, however, and should focus on geographic areas of particular interest. Human factors analysis can also be resource-intensive and should focus on those areas where high risk has been identified and where potential risk reduction mea- sures are practical and enforceable. Phase B Analysis and Techniques The Phase B risk analysis should follow the basic steps outlined for Phase A. The specific modeling and analysis methods may differ, however, because the analysis needs to be more focused, with suffi- cient detail, precision, and data quality to allow more robust deci- sions on the selection, design, and implementation of cost-effective

OCR for page 110
Technical Approach to the Aleutian Islands Risk Assessment • 133 risk reduction measures. To the extent possible, Phase B should be a quantitative assessment. Other characteristics of the Phase B risk analysis methods are described below. Possible Need for Use of Hybrid Modeling Methods to Develop the Risk Scenarios Hybrid modeling used to develop Phase B risk scenarios may employ the following techniques: • Event sequence diagrams for the development of categories of risk scenarios at an appropriate level of abstraction (see Appendix F for further discussion of event sequence diagrams and their use in developing risk scenarios); • Fault trees, Bayesian belief networks, or similar logic modeling methods as necessary to add causal detail to the events of the event sequence diagrams; and • Physical models and simulation, particularly for environmental consequence analysis (modeling the fate and effects of spilled substances) and system simulation (capturing the complex inter- actions of vessel traffic and the changeable conditions in which the vessels operate). Need for More Detailed Causal Modeling, Particularly in Areas Where Risk Reduction Measures Are Being Considered Normally, the level of detail of risk analysis is influenced by data availability. When certain details are essential to understanding how risk reduction measures work, however, such details should be added even in the absence of data so that sensitivity analysis (see below) can be performed. Doing so usually requires a com- bination of modeling techniques (e.g., event sequence diagrams, fault trees) or system simulation supplemented by elicitation of expert opinion. Need to Consider Possible Human Errors in Critical Phases of Risk Scenario Evolution In the case of marine accidents, causality can often be traced to human factors, making evaluation of human error (e.g., in causing accidents, in implementing response plans and rescue operations) a critical part of the overall risk assessment. This analysis, at a minimum, should include identification of human failure events, with assessment of corresponding

OCR for page 110
134 • Risk of Vessel Accidents and Spills in the Aleutian Islands human error probabilities. A more detailed human reliability analysis may be necessary in some cases. See Appendix D for further discussion of human error and human reliability analysis methods. Consideration of Rare, High-Consequence Events as Well as More Frequent, Lower-Consequence Events Both types of events need to be examined. The methods used to study each type may differ. Use of Advanced Methods for Estimation of Model Parameters Such methods may encompass, for example, probabilities of events, projection of vessel traffic into the future, and metrics for the effec- tiveness of risk controls. They may also include Bayesian inference, a well-established technique for assessing rare and catastrophic events. See Appendix E for further discussion of Bayesian statistical analysis methods. Use of Formal Methods for Employing Expert Opinion Study of the risks associated with complex man-made and natural systems always involves subjects for which data or models either do not exist or are not cost-effective. In such cases, expert opinion becomes an important resource for the analyst. Formal methods for the use of expert opinion include guidelines and techniques for selection of subject matter experts, elicitation of their opinions, and use of those opinions (e.g., methods for aggregating probability esti- mates obtained from multiple experts). See Appendix C for further discussion of the elicitation and use of expert opinion. More Rigorous and Comprehensive Uncertainty and Sensitivity Analyses Conservatism is often used to offset the higher levels of uncertainty expected in the preliminary, highly qualitative stages of risk assess- ment. The more detailed assessments of the Phase B focused analy- ses should necessitate less reliance on conservative assumptions. The findings of the analysis need to be carefully scrutinized and the level of uncertainty explicitly stated, however, so that decision makers will not inadvertently be left with a false sense of precision. All important sources and types of uncertainty need to be con- sidered in generating and reporting the findings of the Phase B

OCR for page 110
Technical Approach to the Aleutian Islands Risk Assessment • 135 risk analyses. These include uncertainties stemming from the struc- ture and form of the models employed, generally referred to as model uncertainty, and uncertainties in the model parameters, gen- erally known as parameter uncertainty. In the Phase B analyses, uncertainty should be quantified when practical. Doing so imposes discipline and requires the analyst to scrutinize the robustness of each assumption made. Sensitivity analysis is a technique for assessing the influence of varying inputs on analysis results. Sensitivity analysis is appropriate for evaluating the impact of significant model assumptions and highly uncertain model parameters on the development of risk scenarios, modeling of consequences, and evaluation of the role and effec- tiveness of risk reduction measures. See Appendix E for further discussion of uncertainty and sensitivity analyses. Consequence Analysis The extent of additional consequence analysis required for the Phase B assessment will be influenced by the types of risk reduction measures to be evaluated and the possible need for monetizing of benefits, such as avoidance of natural resource damage and socioeco- nomic costs. Both spill outflow analysis and biological consequence modeling may be necessary. Spill Outflow Analysis In the Phase A assessment, the size of spills is based primarily on historical data. During Phase B, it may be desirable to obtain more detailed estimates of the distribution of spill sizes for certain acci- dent scenarios. Moreover, changes in vessel design and size and the size of bunker tanks may make it difficult to predict spill size entirely on the basis of historical data, in which case analytical techniques should be applied. A number of approaches are available for calculating spill size given a particular vessel design and type of accident. These include simula- tion of groundings and collisions and assessment of structural damage based on energy balance (TRB 2001, 250–255), as well as proba- bilistic methods based on historical damage (TRB 2001, 195–198). A broad range of factors influence outflow, such as the structural arrange- ment of the ship, the characteristics of the struck object, the speed of

OCR for page 110
136 • Risk of Vessel Accidents and Spills in the Aleutian Islands the vessel, and weather conditions. The complexity of the problem necessitates the use of many simplifying assumptions, which in turn reduces the confidence level of the results. If these approaches are applied in the Phase B analysis, their results should be benchmarked against historical spill data to the extent practical. Biological Consequence Modeling The physical fate of a number of representative spill scenarios (e.g., type of oil, size, location) is investigated in the Phase A assess- ment. For that analysis, the concentrations of the spilled substance contaminating the water, sediment, and shoreline are used as an indicator of environmental damage. Additional model runs may be needed during Phase B, because other locations or spill sizes may require evaluation. Where particularly sensitive habitats are endan- gered, estimates of fatalities to fish, invertebrates, seabirds, and mam- mals may be desired. For these analyses, a biological model is applied together with the fate model to determine exposure of habitats and organisms to lethal levels of spill components and to estimate mortality and ecological losses. Estimating natural resource dam- age requires compilation of abundance data for species of interest (e.g., fish, invertebrates, seabirds, otters). Cost–Benefit and Cost-Effectiveness Analysis The Phase A assessment includes a high-level look at costs and ben- efits, which involves assigning categories of costs and benefits largely through qualitative analysis. In Phase B, the costs and benefits should be quantified to the extent possible. The Office of Management and Budget (OMB) provides guid- ance for regulatory analysis of federal regulations (Circular A-4, September 17, 2003). Although not all of the risk reduction measures being considered would require federal intervention, the procedures for cost–benefit analysis and cost-effectiveness analysis should be applied to all the measures to facilitate decision making. In cost–benefit analysis, both costs and benefits are monetized. The results make it possible to compare all risk reduction measures according to a common metric and reduce reliance on the use of professional judgment to compare benefits qualitatively. Cost–benefit analysis should be performed to the extent practical when benefits can reasonably be monetized. Cost-effectiveness analysis is applied

OCR for page 110
Technical Approach to the Aleutian Islands Risk Assessment • 137 when it is difficult to monetize benefits, particularly when safety and health assessments come into play. For instance, cost-effectiveness can be expressed as cost per fatality avoided or cost per barrel of oil spill avoided. Results of cost-effectiveness analysis can be misleading, however, when the effectiveness measure does not properly map consequences or when disparate types of benefits are combined. For example, the consequence of an oil spill is influenced by the type of oil spilled and the size and location of the spill; this makes it difficult to compare different accidents and risk reduction mea- sures on the basis of barrel of oil spill avoided. Furthermore, a risk reduction measure may improve safety (i.e., reduce serious per- sonal injury or fatality rates) while reducing the likelihood or size of a spill. OMB recommends that costs and benefits be quantified when- ever possible. Where costs or benefits cannot be monetized, they should be expressed in physical units. If they cannot be expressed in physical units, a qualitative description of the costs and benefits should be provided. Key elements of the regulatory analysis approach recommended by OMB are as follows: • Document how a risk reduction measure will provide expected benefits. • Compare costs and benefits against a no action baseline. • Identify the expected undesirable side effects and ancillary ben- efits of the risk reduction measure, and add them to the direct benefits and costs as appropriate. The baseline spill projection developed during Phase A and sub- sequently refined during Phase B should be used as a basis for the no action baseline. An annualized stream of costs and benefits should be developed relative to this baseline and then discounted to present value for comparison purposes. The OMB guidelines provide procedures for discounting costs and benefits. In prior maritime regulatory assessments, USCG has frequently expressed cost-effectiveness in terms of barrels of spilled oil averted. For example, the methodology applied in an assessment of the use of rescue and escort tugs to avoid oil spills in the Puget Sound area (USCG 1999) assumed the following. Where a risk reduction measure was deemed effective in avoiding accidents and consequently reducing vessel damage, cargo loss, time loss, human injuries, or loss of life, these avoided losses were treated

OCR for page 110
138 • Risk of Vessel Accidents and Spills in the Aleutian Islands as benefits. These avoided costs were subtracted from the cost of compliance and enforcement to obtain the net costs. The net cost- effectiveness equaled the benefits (present value of the number of barrels of oil not spilled) divided by the net cost. In this case, a value of statistical life was applied to monetize fatalities, thereby allowing benefits to be expressed in terms of barrels of oil spill averted. The committee recommends that a similar approach for quan- tifying cost-effectiveness be applied for all the risk reduction measures evaluated in Phase B. Again, it should be remembered that cost-effectiveness ratios alone are not sufficient for deci- sion making but should be supplemented by quantified benefits, including estimates of spill size distribution, as well as mean spill size and numbers of fatalities and serious injuries avoided. The oil spill trajectory models and spill fate assessments will assist in the qualitative assessment of different spill sizes and spill types. With this approach, cleanup costs, costs of natural resource dam- age, and socioeconomic costs are treated as part of the qualitative assessment. A number of approaches and techniques are available for mone- tizing cleanup costs, costs of natural resource damage, and socioeco- nomic costs. In one approach for estimating costs of natural resource damage, for example (McCay et al. 2003), a biological model is applied as described in the section on consequence analysis. The basis for estimating these costs is the cost to restore equivalent resources. Another example is an approach described for estimating cleanup and socioeconomic costs for spills in San Francisco Bay (Etkin 2003). See Appendix I for further discussion of resource vulnerability and natural resource damage. Evaluating the costs of natural resource damage can be expen- sive, depending in part on the availability of the necessary bio- logical data. After reviewing the Phase A results, the Management Team, in consultation with the Advisory Panel and Risk Analysis Team, should decide on the extent to which the costs of natural resource damage and socioeconomic costs will be evaluated. This information may be needed to compare the relative impact of spills of different sizes or different oil and chemical types, or to justify risk reduction measures involving particularly high imple- mentation costs.

OCR for page 110
Technical Approach to the Aleutian Islands Risk Assessment • 139 DEVELOPMENT AND REPORTING OF RECOMMENDATIONS FOR DECISION MAKERS ON RISK REDUCTION MEASURES TO BE IMPLEMENTED The recommendation of risk reduction measures for implementa- tion by decision makers requires consideration of who the decision makers are in the risk management system and their capacities, both individually and collectively, to implement recommended measures. An additional consideration is the type or packaging of information needed by decision makers given their specific regulatory roles or mandates. For example, USCG rulemaking depends on consideration of benefits relative to costs, as described above. Within the bounds of their respective jurisdictions, USCG, the State of Alaska, and local municipalities all have decision-making roles. Securing federal funds involves other U.S. government branches, and IMO will have a role as well if changes to international regulations are recommended. For the more costly and far-reaching measures, successful implementation may well require a collaborative agreement among decision makers and the support of stakeholders, and some such measures, such as those that require the involvement of IMO, may take longer to implement than changes that can be made at the local level. The Management Team and the Advisory Panel should prepare a final report providing recommendations to decision makers in a way that documents the basis for those recommendations in the risk assessment. According to IMO’s Formal Safety Assessment (see Chapter 2), the purpose of this final stage of the risk assessment is to define recommendations which should be presented to the rele- vant decision makers in an auditable and traceable manner. The rec- ommendations would be based upon the comparison and ranking of all hazards and their underlying causes, the comparison and ranking of risk control options as a function of associated costs and benefits, and the identification of those risk control options which keep risks as low as reasonably practical. To meet the objectives of the risk assessment, the final report should be fully transparent, describing the risk assessment process and all relevant assumptions. It should show that the full range of relevant hazards and risks was adequately investigated, describe major uncertainties that affect the robustness of the conclusions reached, and demonstrate that the analysis was of sufficient rigor

OCR for page 110
140 • Risk of Vessel Accidents and Spills in the Aleutian Islands to represent benefits and costs accurately to the extent practical. The following should be found in the final report: • Hazards and risks should be clearly identified. For risk reduction measures that merit detailed analysis, costs and benefits should be clearly defined. • All sources of data should be documented and assumptions explained. Models and methodologies should be described in sufficient detail that a third party can understand the basic assumptions and limitations of the assessment. • Judgments applied during the assessment should be explicitly stated. The process for elicitation and analysis of expert opinion should be explained. • Uncertainty and associated sensitivity analyses should be clearly documented and explained. Results should be presented in a way that does not create a false sense of precision. • The report should be of sufficient depth to address the needs and expectations having those with expertise in risk assessment while being understandable to the layman. Ideally, all stakeholders and decision makers will reach consensus on measures to be implemented. Past experience in risk manage- ment indicates that this is not always possible, however, since some stakeholders may have strong positions on which they are unable or unwilling to compromise, or uncertainties may cloud the true worth of some risk reduction measures or the true costs associated with their implementation. When consensus cannot be reached, the report should present the differing opinions, thereby assisting decision makers in understanding the various sides of the issues. As discussed in Chapter 2, risk can be characterized as negligi- ble, tolerable, or intolerable. The goal of the risk assessment that is the subject of this report is to improve the level of safety related to spills from ships operating within the Aleutian Islands region. The implementation of risk reduction measures identified by this assessment should result in risk falling at or below the tolerable level (although it may not be possible to define such a level with precision), and decision makers must decide which measures or which combination thereof will achieve this goal. In so doing, they must balance the level of investment required to implement the measures against the projected safety level of the system that would result.

OCR for page 110
Technical Approach to the Aleutian Islands Risk Assessment • 141 Although the decision as to which measures will be adopted ultimately rests with the decision makers, the Management Team and Advisory Panel must attempt to address the concept of accept- able risk in formulating their recommendations. The acceptable level of risk will vary among different stakeholders. A collaborative effort will be necessary to understand the viewpoints of different stakeholders and meld them into an acceptance criterion. Once decisions have been made and risk reduction measures implemented, there should be a process for monitoring the effective- ness of the measures and the overall acceptability of system risk. The final report should include a proposed structure for this ongoing risk management process, which should include metrics for the continu- ous monitoring of risk. Specific institutional arrangements will likely be necessary to foster continued discussion among a broad range of stakeholders regarding the residual risk in the system and the ade- quacy of the measures instituted to control that risk. Experience with other systems having the complexity of shipping through the Aleutian Islands suggests that risk cannot be reduced to zero by risk reduction measures that can feasibly be undertaken. REFERENCES Abbreviations NRC National Research Council TRB Transportation Research Board USCG U.S. Coast Guard Etkin, D. S. 2003. Bio-Economic Modeling for Oil Spills from Tanker/Freighter Groundings on Rock Pinnacles in San Francisco Bay. U.S. Army Corps of Engineers, Sacramento, Calif. McCay, D. F., J. J. Rowe, N. Whittier, S. Sankaranarayanan, and D. S. Etkin. 2003. Estimation of Potential Impacts and Natural Resource Damages of Oil. Journal of Hazardous Materials, Vol. 107, No. 1–2, pp. 11–25. NRC. 2007. Polar Icebreakers in a Changing World: An Assessment of U.S. Needs. National Academies Press, Washington, D.C. TRB. 2001. Special Report 259: Environmental Performance of Tanker Designs in Collision and Grounding: Method for Comparison. National Research Council, Washington, D.C. USCG. 1999. Regulatory Assessment: Use of Tugs to Protect Against Oil Spills in the Puget Sound Area. Report No. 9522-002. USCG. 2008. Risk-Based Decision Making Guidelines (RBDM), Vol. 1. Washington, D.C. www.uscg.mil/hq/gm/risk/E-Guidelines/RBDMGuide.htm.