APPENDIX D Applicability of the Prince William Sound Study to Other Geographical Areas
The applicability of the Prince William Sound (PWS) marine transportation risk assessment study to other areas raises two questions. One is the extent to which the results of the study for PWS can be applied directly to other ports. Another concerns the transferability of the study’s analytic approach to studies in other areas. Within these questions are more subtle considerations. For example, in attempting to apply either the results or the methods of the PWS Study to other marine ports, one needs to ask whether the goals and problem boundaries that influenced its design are appropriate for assessing risks in other regions.
To answer the last question, one must take into account the extent to which stakeholders in the region’s marine transportation system will support and participate in the proposed study. The PWS Study could not have been done without a very high level of commitment to providing data and access to information by the major shippers whose vessels were the focus of the study.
One can give simple answers to these questions. First, the method is more likely to be applicable in other areas than the study results. Second, the basic goals and problem boundaries of the PWS Study may not fully reflect the elements and sources of risk that constituencies in locations find significant. Finally, it may be difficult to secure the same high level of participation from shipping interests in other areas, particularly if public dialogue involving a broad cross-section of industry interests on questions of maritime safety has not been well established.
Addressing these questions in more detail requires considering the PWS Study from several points of view. These include the particular characteristics of the PWS maritime transportation system that facilitated and defined the approach that was taken, the aspects of the maritime safety problem in the region that were analyzed, data and data sources, models and modeling approaches, and finally, broader issues like study goals, problem definition, and participation by groups outside of the study team.
CHARACTERISTICS OF THE PRINCE WILLIAM SOUND MARITIME TRANSPORTATION SYSTEM THAT FACILITATED THE STUDY
Transiting vessels in PWS include tankers and their escort vessels, tugs and tows, passenger vessels, and ferries. Most tankers visiting the sound are regular visitors, and most are U.S. flag vessels. Only 42 different tankers visited the sound in 1994. In a typical month, most vessels are tankers and their escorts; relatively few involve cargo vessels (PWS Study 2.8–2.9, TD I:1.27–1.28). The system is characterized in the PWS Study as being composed primarily of known participants who have a high degree of interaction.
A number of the safety enhancements put into place since the EXXON Valdez spill have had the secondary effect of providing the kind of data on system performance that are not generally available in other areas. These include SERVS vessel logs on tanker transits, weather, and sea state and the automated dependent surveillance system capabilities of the U.S. Coast Guard’s vessel traffic service for PWS, which makes it possible to record digital images of vessel tracks for tankers.
ASPECTS OF THE PROBLEM THAT WERE ANALYZED
Like all real-world modeling studies, the PWS Study necessarily reflects numerous compromises between the desire to capture the reality of the system as fully as possible and the need to economize on the level of detail to make the analysis tractable, given the available time and resources. Decisions on the scope of the study and the degree of model resolution might not be the same decisions made in other regions. Some of the decisions made for the PWS Study are listed below:
Oil outflow (rather than the environmental effects of oil outflows) was treated as the “end state” consequence of concern.
Oil outflows at all locations within the study area were considered equally bad. Local tides and currents or seasonal or location-specific sensitivities were not taken into account.
Loss of life was not included as a specific consequence to be avoided.
Economic losses, such as the loss of tourism or extensive shoreline property damage, were not included as specific consequences of concern.
Spills from other facilities, including operational discharges in ship-to-shore or ship-to-ship transfers, were not included as outcomes to be avoided.
The consequences of various mixes of foreign and domestic traffic or other possible changes in the makeup of the fleet were not considered.
Marine incidents, such as fire and explosion, were not analyzed.
Although none of these decisions necessarily limits the value of the study to its clients in the PWS region, they would have to be reexamined before the study design could be applied elsewhere. Ports like Puget Sound, San Francisco, Los Angeles-Long Beach, and New Orleans have much more general cargo traffic than PWS and much larger human populations, which make considerations of risks to life and limb for nearby populations much more pertinent. Shorelines elsewhere are also considerably more developed, and
much of the development is very expensive; tourism may be a major component of local economies. The maritime systems in other ports may already be much more “open” than the system in PWS, and the prospect of replacing domestic supplies of crude oil that now arrive at these ports aboard U.S. flag vessels with supplies carried from abroad by foreign vessels would open them even further.
DATA AND DATA SOURCES
The semiclosed nature of the PWS transportation system facilitated cooperation among the study participants through the PWS steering committee. Much of the data used in the study was location-dependent, and the spirit of cooperation fostered by the steering committee appears to have led to unusual types of data, and uncommonly detailed data, being made available. Examples include proprietary data on tanker fleet failure rates provided by the Trans-Alaska Pipeline System shippers, which proved to be an important component of the study’s database. Unfortunately, from the standpoint of anyone wishing to replicate the study, it was agreed that these data would be returned to the original owners upon completion of the study and that no archival records would be kept.
Cooperation among study participants also included active participation by local experts. The judgments of these experts about inherent risks in the system, as well as the relative risks posed by a host of specific vessel and waterway situations, were instrumental to the study’s basic design. Expert opinions were elicited through questionnaires and became major components of the study’s database. Moreover, the collective experiential base upon which the 162 participating experts drew appears to represent a substantial portion of the total current operating experience with the system, another fortuitous consequence of the system being relatively closed.
Relatively little truly generic data appears to have been used in the PWS Study. Tanker casualty data from the International Maritime Organization and Lloyd’s Register were used, but only for events that were regarded by the study team to be truly generic and not dependent on situational factors. Thus, worldwide casualty data were used for fire, explosion, structural failures and founderings, but not for groundings, collisions, or allisions (PWS Study 4.19–4.20). The study team attempted to gather location-specific data first and to substitute worldwide casualty data only when local data were not reliable.
MODELS AND MODELING APPROACHES
Each of the three modeling approaches used in the PWS Study was chosen for providing a view of system performance that was not available from the other models. Of the three, the marine accident risk computation system (MARCS) model was the least dependent on location-specific information, and the George Washington University system simulation model was the most dependent on location-specific information. Small panels of experts at Det Norske Veritas were relied upon to supply probabilities for many of the junctures in the fault tree models. (When multiple models are used, an overarching framework to combine them should be developed.)
The choice of modeling approach depends on the questions to be asked, and it is
probably unwise to assume at the outset that risk assessment studies of other maritime ports will be applicable. Location and situation-specific “ground up” modeling appears to be preferable to generic or theory-based approaches. No matter which model is chosen, it should be calibrated or verified against real-world data. If multiple models that use common data and descriptions are calibrated against each other, the results may not be valid.
PARTICIPATION IN THE PRINCE WILLIAM SOUND STUDY AND CONSENSUS BUILDING
The most valuable aspect of the PWS Study may be the consensus-oriented process that developed around it, which was necessitated, in part at least, by the need for diverse interests to supply data and information to the study team. The study seems to have shifted the oil spill debate from conflicts among stakeholders to a scientific and technical debate about the merits of particular risk reduction measures. This kind of performance-based dialogue about oil spill prevention measures has not been evident elsewhere. From this point of view, replicating the analysis-oriented process for considering measures to reduce the threat of major oil spills is highly desirable.
The level of stakeholder involvement that was developed for the PWS Study was extremely important to the study and could be just as important to studies of other regions. Access to local experts and data on the local operating environment, including the characteristics of waterways and information about vessels and facilities and their management, are vital. The dialogues over maritime safety that have developed through the SMART forum in the Puget Sound and San Francisco Bay regions may indicate that appropriate conditions for studies at similar levels of detail can be conducted in these regions.
The conclusions and recommendations of the PWS Study are highly location-dependent and cannot be assumed to apply to other regions without careful consideration of the extent to which the situations in the region are similar to the ones analyzed in the PWS Study. The considerations for applicability range from the level of broad features of the operating environments to the handling of individual variables that appear to influence the results. Because relatively little sensitivity analysis was done on the models used in the PWS Study, the influence of individual variables on study results are not easy to determine.
Replicating the PWS modeling approach would also require collecting and analyzing similar types of location-specific data. Some of the data will be proprietary and will require the cooperation of the firms that own them, and some will depend on access to local experts. Some data types, such as information on vessel tracks derived from the automated dependent surveillance system capabilities of the PWS vessel traffic system, may be difficult or impossible to obtain for some ports.
The goal for other regions should not be to replicate the study that was done for PWS. The goal should be to identify and reduce or eliminate the significant risks in the system, recognizing that unacceptable states of risk and acceptable risk reduction measures will
always be influenced by local considerations. The purpose of a risk analysis should be to identify the factors and situations that tend to increase or decrease risk and to determine the effectiveness of particular measures for reducing risks, taking into account system-wide effects. The PWS Study did demonstrate that extensive stakeholder participation is extremely beneficial.