Click for next page ( 45


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 44
3 DESCRIPTION OF THE PROPOSED METHODOLOGY A s noted in Chapter 1, the committee developed an overall method- ology for use by USCG in evaluating the relative environmental performance of alternative tanker designs following a collision or grounding accident. This methodology is intended for use in conjunction with other factors to determine whether USCG should approve an alternative design to a double-hull tanker. The key components of the methodology, the data needed for its implementation, and the nature of the results to be ex- pected are described in this chapter. The status of the methodology's development and the efforts required to enable its routine application are also reviewed. OVERVIEW If a tanker runs aground or collides with another vessel,1 the severity of the damage to the ship caused by the impact and the amount of oil spilled depend on the design of the tanker, its loading condition, mitigation ef- forts by its crew, the location of the impact, and the type of accident. Once the oil is in the water, the environmental, economic, and financial conse- quences of the spill will depend on the volume of oil spilled, the type of oil, the location of the spill, resources at risk, seasonality, and the weather conditions at the time of and after the accident, as well as any recovery and cleanup efforts. A rigorous evaluation of the environmental perfor- mance of a tanker should take all these variables into account. Moreover, all possible accident scenarios and their outcomes should be considered. In developing its methodology, the committee considered all the above factors, but to accomplish its mandate had to adopt some simplifica- 1Collisions with fixed objects, such as piers and bridges, are not considered within the committee's illustration of the methodology. However, a more complete future evaluation should take this type of incident into account. 44

OCR for page 44
DESCRIPTION OF THE PROPOSED METHODOLOGY 45 tions. The simplifications were necessary because of the limitations of existing technology for modeling physical phenomena and the redun- dancy of factors that did not introduce new information into the analysis. Throughout the study, the committee took care to ensure that the simpli- fications were not introducing bias into the methodology. The methodology has three main components or steps: 1. Structural damage and oil outflow calculation, 2. Consequence assessment, and 3. Design comparison. The results of the first two steps feed into the design comparison. The di- vision of each step into tasks is shown in Figure 3-1. Each task involves both theoretical and methodological challenges, which are discussed later. An overview of each step is provided in the remainder of this section; each step is then reviewed in detail in the sections that follow. Structural Damage and Oil Outflow Calculation The first step in evaluating the environmental performance of a tanker design is calculation of the structural damage and oil outflow in pos- 1. Structural Damage and Oil Outflow Calculation Collect data on collision and grounding accidents Develop distributions of accident factors Generate accident scenarios Calculate structural damage Determine oil outflow 3. Design Comparison Calculate differences in consequence metrics Analyze design differences 2. Consequence Assessment Model oil fate and transport Simulate environmental conditions Select impact measures Develop consequence metric (ratio based) FIGURE 3-1 Components of the methodology.

OCR for page 44
ENVIRONMENTAL PERFORMANCE OF TANKER DESIGNS 46 sible accident scenarios. To this end, the accident scenarios, or collision and grounding events, must be selected. The parameters (accident fac- tors) that define the collision and grounding events include the speed of the vessel, its loading condition, and the type of obstacle or colliding vessel. The accident factors used by the committee were selected to rep- resent conditions in U.S. waters, specifically in areas with a high density of tanker traffic. The factors are defined with distributions that represent the range of their possible values, as well as the frequencies at which these values occur and their correlation structure. By sampling from these distributions, a large number of collision and grounding events were generated that model possible accidents and their relative fre- quencies in the selected areas of tanker traffic. For example, the com- mittee chose a distribution of vessel speeds that are likely to occur in the areas where tankers operate. This distribution of speeds was then used to construct a distribution of accidents that are more or less likely to occur at those speeds. Once the collision and grounding events have been defined, ship damage models are used to determine the structural damage for each accident and each design. The inputs into the models include the accident factors and a description of the vessel (definition of the hull form, compartments, and hull structure).2 The output from the damage models is the damage extent and location, in other words, the size of the hole and its location on the hull of the ship. Once the size and location of the hole on the bottom or side of the ship are known, the resulting oil outflow can be determined. The calculation is carried out for each accident scenario. Consequence Assessment The second component of the evaluation process is the assessment of environmental consequences from an accidental spill. If the environ- mental impact of a spill were to increase linearly with the volume spilled, the oil outflow could be used to measure the performance of alternative designs. However, since the impact of the spill is dependent on many factors other than the volume (e.g., product type, environmental condi- tions, location), one cannot assume that the relationship between the spill volume and the consequences is linear (see Chapter 2). Therefore, 2The committee developed this methodology for evaluating new tankers. If several simplifying assumptions about aging and fatigue can be made, however, it may be possible to study retrofit options for existing ships as well.

OCR for page 44
DESCRIPTION OF THE PROPOSED METHODOLOGY 47 to perform an unbiased evaluation of alternative designs, it is necessary to assess the consequences of the oil outflow by taking into account all relevant variables that affect the outcome. The committee concluded that past efforts to use the reported cost of a spill as a surrogate for its overall consequences did not provide a ra- tional basis for comparison of designs. Therefore, the committee decided to carry out environmental impact modeling using SIMAP, a modifica- tion of a stochastic version of the National Resources Damage Assessment Model for Coastal and Marine Environments (NRDAM/CME).3 SIMAP models oil fate and transport and allows for random sampling of weather conditions on the basis of historical weather data. It provides a number of consequence measures, such as the area of the sheen, the toxicity in the water column, and the length and area of oiled shoreline. Spill cleanup costs were included in the analysis at first but were later excluded because of the uncertainty in the cost models. The economic impacts due to cargo loss and third-party damages are excluded from the methodology: the impact of cargo loss is considered negligible compared with the environ- mental impact, while third-party costs, which can be significant, are as- sumed to be linear with the environmental consequence measures. These assumptions are discussed in more detail later in the report. Because a distribution of weather events is included, the envi- ronmental impact modeling produces distributions of values for each consequence measure and for each simulated spill. In addition, the con- sequences are measured at a number of threshold levels, which represent the intensity or the quantity of the physical measure (e.g., the concentration of toxic components in the water column or the thickness of the slick on the water). The committee decided to limit the evaluation of the conse- quences to the physical measures instead of extending it to the impact on biological resources. The intent was to keep the analysis as rigorous as pos- sible without necessitating difficult decisions on what threshold levels would damage biological resources and how different types of biological resources are valued. This decision is also addressed in more detail in later sections of this chapter. These distributions of impacts are then com- pared with each other to determine a single overall consequence metric (or function) that will capture the relative damage caused by spills of dif- ferent sizes. 3NRDAM/CME and the SIMAP modification were developed by Applied Science Associates (ASA) for the De- partment of the Interior.

OCR for page 44
ENVIRONMENTAL PERFORMANCE OF TANKER DESIGNS 48 Design Comparison In the third and final component of the methodology, oil outflow values from the first step are transformed using the consequence metric from the second step. For each accident scenario, the performance of the alter- native design is compared with that of the reference double-hull tanker, and a difference is calculated. These distributions of differences capture the difference in the environmental impact of the two designs. Equivalency in performance can be determined by evaluating these distributions either qualitatively or quantitatively. STRUCTURAL DAMAGE AND OIL OUTFLOW CALCULATION The determination of oil outflow after collision or grounding is divided into the following tasks: 1. Definition of the collision or grounding event and its probability of occurrence, 2. Determination of the structural damage that results from the given collision or grounding event, 3. Determination of the outflow that results after the structure is damaged, and 4. Evaluation of active systems that modify outflow when applicable. Each of these tasks is discussed in detail in the following subsections. Definition of the Collision or Grounding Event and Its Probability of Occurrence The methodology includes the assumption that the probability of alter- native designs encountering a potential collision or grounding scenario is the same. Features of a design that reduce the risk of encountering dan- gerous situations, such as navigational aids, can be incorporated into any alternative design. However, the probability of occurrence (or frequency) of an event relative to the frequency of all other events is included in the methodology through the description of the accident factors that define the collision and grounding events. In addition, whether the ship runs aground in a given scenario depends on its draft, and the effect of pos- sibly different drafts of alternative designs is taken into account in the methodology. Accident factors, which define the condition of the tanker before the accident and the type of colliding vessel, are described in more detail in Chapter 4.

OCR for page 44
DESCRIPTION OF THE PROPOSED METHODOLOGY 49 Accident factors, which are defined by a distribution, must be se- lected so that they represent conditions in the studied area. For example, vessel speed should have a distribution that includes all possible speeds along with their likelihood. Transit and maneuvering speeds will have a high likelihood in the distribution. Speeds outside of the typical range must also be included, but their likelihood will be small. An example of the distribution for speed is shown in Figure 3-2. Once the initial distri- butions are known, the factors are sampled from the distributions to gen- erate accident scenarios. Collision and grounding incidents that are more likely will have a higher frequency of occurrence. Determination of Structural Damage During a collision between two ships or a grounding (when a ship strikes a fixed object on the bottom), contact forces occur between the two bodies. The magnitude of the contact force depends on a number of characteristics of the striking and struck bodies. The effects of these char- acteristics are to decelerate the motion of the moving ship, perhaps suf- ficiently to bring it to a stop, and to damage structural members in the region of contact. In an extreme case, the damage may result in internal flooding, leading to loss of buoyancy and stability. If sufficiently severe, the damage may cause the ship to capsize or founder, resulting in its total loss. Even if the ship remains afloat, interior tank spaces may be pene- trated, allowing spillage of any liquids they contain. 0.08 0.07 Events 0.06 0.05 Grounding 0.04 of 0.03 0.02 Proportion0.01 0.00 0 2 4 6 8 10 12 14 16 18 20 Speed of Ship (knots) FIGURE 3-2 Distribution of grounding vessel speeds.

OCR for page 44
ENVIRONMENTAL PERFORMANCE OF TANKER DESIGNS 50 Analytical Methods for Assessing Damage During Collision or Grounding The magnitude and extent of any structural damage depend on the lo- cation and magnitude of the forces of contact, as well as the strength properties of the structural members in that vicinity. The collision or grounding event occurs over a time interval ranging from a few seconds to a few minutes, during which time the forces of contact undergo vari- ations in direction and magnitude. The behavior of these contact forces depends on the initial speed and mass, as well as other properties of the ship, and on the behavior of the affected structural members. The forces will change in an irregular manner with time as the ship decelerates and as different structural members undergo deformation and rupture. An analysis of ship behavior in a collision or grounding has the twin goals of predicting the forces of contact and the behavior of the ship structure under the action of those forces. Since these effects, force and structural response, are interrelated, it is necessary to perform simultaneous analyses of the two. This is in contrast to the usual structural design problem in which the forces acting on the structure are predicted first, and the struc- tural members are then designed to withstand those forces. The basic problem to be solved in the collision or grounding may be stated in terms of energy relations as follows. Before the collision or grounding, the ship is moving forward at velocity V, and its kinetic energy (KE ) is given by the following expression: KE = 1 2 2(m + m)V where m is the actual mass of the ship and m is the added mass of an equivalent quantity of water, which accelerates with the ship. During the collision or grounding, KE is transformed or dissi- pated through several mechanisms. A part of the KE is transformed into potential energy of the ship corresponding to changes in draft, heel, and trim. KE is dissipated through friction, acting principally between the con- tacting surfaces of the two ships or the ship and the object upon which grounding occurs. KE is also transformed into strain energy of plastic de- formation, fracture, and tearing of structural members. Finally, hydro- dynamic effects, including water friction and radiating waves, dissipate some of the KE, but this is a minor term and may usually be neglected. Plastic deformation and fracture of structural members involve highly nonlinear aspects of material behavior; therefore, the forces of contact between ship and ship or ship and ground are strongly nonlinear functions of the relative motions of the two. As a result, the problem of

OCR for page 44
DESCRIPTION OF THE PROPOSED METHODOLOGY 51 finding the time history of vessel motion (and damage progression) after collision or grounding occurs must usually be solved as a step-by-step integration in time of the equations of vessel motion. An example is provided by a common model of grounding in which a rock pinnacle is modeled as a circular vertical cone with a rounded vertex. As the bottom of the ship contacts the cone tip, vertical and horizontal forces act on the ship at the point of contact. These forces cause rigid body motions of the ship in its six degrees of freedom, but principally in pitch, heave, roll, and surge. The forces also cause defor- mation, possibly including yield and fracture, of the ship's structural members in the vicinity of the contact point. The problem is then solved in a series of time steps. At each step, using the velocity at the end of the previous step, an incremental motion is predicted, structural deforma- tions are determined that are consistent with the relative motion between the ship and struck object, and the force associated with those deforma- tions is determined. The integral of this force over the increment of motion equals the work done by the contact force in this step. This work must, in turn, equal the incremental reduction of KE. From the remaining KE, the velocities at the next step are determined. The entire process ends when the remaining KE equals zero and the ship has come to rest. Obviously, the analysis of a process as complex as this requires certain simplifying assumptions to keep the magnitude of the analysis within practical bounds. These simplifications include assumptions con- cerning the external kinematics of the ship and the behavior of the ma- terial undergoing deformation and fracture. Methods that have been developed and used for this purpose include the following: Statistical analysis of previous casualty data, Detailed nonlinear finite-element methods, and Macroscopic finite-element or superelement methods. In general, the statistical methods inherently involve the behavior of structures typical of the ships forming the database and may include sim- plifying assumptions regarding the structural arrangements. They cannot be applied with any confidence to comparative studies in which the subject ship has an innovative structural arrangement. Since such comparison of alternative arrangements with a known arrangement, the double hull, is a key element of the present work, the statistical approximation methods cannot be applied in all evaluations. Finite-element methods are widely accepted in the analysis of complex structures such as those used in ships. The most advanced im-

OCR for page 44
ENVIRONMENTAL PERFORMANCE OF TANKER DESIGNS 52 plementations are capable of treating nonlinear material and structure be- havior, including yield, fracture, and buckling. A detailed nonlinear finite- element analysis requires great effort to build the structural model itself and considerable computational effort to arrive at a solution. The repet- itive analyses needed to investigate large numbers of casualty scenarios render these methods impractical for purposes of the present study. Superelement methods are essentially a variation on finite-element methods. Whereas the latter methods utilize large numbers of small simple elements to represent the structure, superelement methods use relatively few large, sophisticated elements, each incorporating the material and be- havioral properties of a relatively large portion of the structure. An example might be the entire stiffened plate panel contained between two web frames. The necessary computational effort is greatly reduced with such methods, yet if appropriately defined elements are used, the results are of sufficient accuracy for the present application. A limited number of computer codes based on superelement methods have been developed. The program DAMAGE, developed by the Joint Massachusetts Institute of Technology (MIT)Industry Project on Tanker Safety, which falls into this category, was used to analyze grounding damage during the application of the committee's method- ology. DAMAGE is available as software and has the widest range of ap- plicability of published simplified methods (Tikka and Chen 2000). Other published work includes that of Wang et al. (1997), which is applicable to raking damage only, and that of Pedersen (2000) on grounding on soft soil. The collision study was carried out using a simplified collision analysis tool, SIMCOL, which was developed by Brown and his students at Virginia Tech. Other simplified approaches for analyzing collision in- clude (a) the collision module in the program DAMAGE; (b) the ALPS/ SCOL simplified finite-element code, developed at Pusan National Uni- versity, Korea; and (c) the simplified approach developed at the Tech- nical University of Denmark. (More details on these methods can be found in Alternative Tanker Designs Collision Analysis on the accompa- nying CD.) SIMCOL was selected because of its availability and applica- bility to analyzing a large number of collision scenarios. The validation and limitations of DAMAGE and SIMCOL are discussed later in this chapter and in Chapter 4. Limitations of Modeling the Physical Phenomena Several factors limit the ability to achieve an accurate model of collision or grounding scenarios:

OCR for page 44
DESCRIPTION OF THE PROPOSED METHODOLOGY 53 Random errors: Virtually every aspect of any attempt to model the behavior of a ship and its structure in a collision or grounding event is subject to random uncertainties, including a. Randomness of the conditions in the environment (e.g., waves, current), b. Ship conditions at the time of collision (e.g., speed, draft, heading), and c. Structural conditions (e.g., corrosion, construction tolerances, material imperfections). Systematic errors: These errors are due to shortcomings of analytical methods, imperfect knowledge of the relevant phenomena, and the need to reduce the computations to a manageable level. The random errors of Types a and b can be addressed by the choice of conditions under which simulations are run. This is accomplished by conducting simulations for a very large number of cases covering all expected conditions of loading and the environment. Type c includes effects that are routinely treated when a reliability analysis is performed on a structure. These effects, such as corrosion, unrepaired dents, and cracks, are a consequence of deterioration of the structure over the life of the ship. Type c also includes construction tolerances and material imper- fections, which are due to the ship not being built precisely as designed: the thickness of steel plate as received from the mill is not exactly as spec- ified by the designer because of rolling imperfections; welded members are not in the exact locations and alignments shown on the plans; and ma- terial properties may not be precisely as assumed or specified as a result of chemistry, heat treatment, and other aspects of the steel-making process. Many of these items are covered by specifications on dimensions plus al- lowable tolerances in manufacture, and these tolerances can be incorpo- rated in the analysis procedure. The systematic errors are of more concern since they are charac- teristic of the analysis methodology. In a complex analysis such as that using DAMAGE or SIMCOL, especially when applied repeatedly to a large number of cases, simplifying assumptions must be made to keep the com- putational effort within reasonable bounds. In DAMAGE, simplified analytical expressions are developed for the behavior of the different structural elements under various conditions. For example, for a panel of bottom plating, expressions are derived for plate indentation without fracture in the initial stages of contact with an obstacle, for plate splitting as the ship continues to move over the obstacle with increased penetration, and for tearing and wrinkling of the plate

OCR for page 44
ENVIRONMENTAL PERFORMANCE OF TANKER DESIGNS 54 ahead of the obstacle. Similarly, expressions are derived for the behavior of stiffeners and webs attached to the plating. Two types of assumptions are involved in modeling the response-- those concerning the material itself and those concerning the structural deformations. Material behavior is defined by a stressstrain curve, which describes the stressstrain relationship up to the fracture strain. The kine- matic assumptions concern the geometry of deformation. An example is the behavior of a plate being deformed by contact with the vertex of the cone, which models the obstruction. The model assumes that the deflected shape consists of "flaps" with plastic hinges at their edges, whereas the real deformation pattern involves curved edges. The structural model in DAMAGE includes only the cargo block. The effect of the bow and the stern on the structural behavior in the damage region is neglected. The model is built with conventional struc- tural members, and the materials used are limited to those that can be described with the stressstrain curve and the assumed failure modes. In- novative structural designs using new materials would require extensions to the current program. DAMAGE is limited to modeling powered grounding on a single pinnacle. The obstruction, modeled as a pinnacle, is defined by its apex angle and the tip radius. Other types of obstructions or groundings (e.g., grounding on a reef or soft soil, drift grounding) currently cannot be modeled in DAMAGE. In general, the validity of the combined assumptions can be tested only by experiment. Validation of DAMAGE has been limited to a few test cases, but it has been found to predict the overall damage extent well. It predicts average forces, but it does not capture peak forces as the ship advances relative to the obstruction. (For details see Alternative Tanker Designs Grounding Analysis on the accompanying CD.) The SIMCOL collision model uses modified procedures derived from the statistical work of Minorsky (1959) to determine energy ab- sorption by the horizontal members; reaction forces on the vertical members are determined by applying simplified analytical models that take into account the mechanics of the structural behavior. Total forces are a superposition of the forces acting on vertical and horizontal members. Only conventional structural members can currently be analyzed by SIMCOL, and assumptions on material are similar to those in DAMAGE. SIMCOL can be used to analyze collisions between two vessels, but not collisions with a solid object, such as a bridge or a pier. The striking ship bow is modeled in SIMCOL as wedge-shaped and rigid. Only the

OCR for page 44
DESCRIPTION OF THE PROPOSED METHODOLOGY 59 risks that necessitate a higher degree of evaluation of the performance of these systems. CONSEQUENCE ASSESSMENT While the first component of the methodology (damage and outflow cal- culation) is based on techniques that have been developed previously, the consequence assessment component is a new approach. As discussed in Chapter 2, the committee determined that data on historical spills were not sufficient to allow for the development of a response function de- scribing the relationship of spill volume, oil type, and other factors to spill consequence. Instead, the committee chose to apply existing models of oil spill transport and fate, as well as consequence, to generate such a re- lationship. The committee's approach to this effort, which involved the following three tasks, is described in this section. 1. Selection of an existing oil fate and transport model and its appli- cation to generate estimates of the expected physical consequences of a broad range of hypothetical spills (e.g., meters of shoreline oiled, area of slick). 2. Consideration of whether response functions could be developed that describe the relationship of spill size to Value of lost product, Response cost, Environmental consequence, and Economic and social consequences. 3. Using the results of Tasks 1 and 2, establishment of equivalency ratios that describe the expected consequence of a spill in terms of the consequence of a standard-sized spill (referred to as the reference spill). These tasks are discussed further below. It is important to note that this approach to spill consequence modeling differs significantly from earlier efforts (e.g., Astrup et al. 1995; Michel and Moore 1995; Michel et al. 1996; Sirkar et al. 1997). These other efforts involved using reported costs of historical oil spills as surrogates for environmental damage, an approach rejected by the committee. Sirkar et al. also recognized the limitations of available cost data and instead con- sidered a range of hypothetical cost curves. They called for the compilation of additional data on spill costs to support the generation of experience- based cost relationships. However, the committee does not believe that an accurate and consistent set of cost data can be compiled for purposes of

OCR for page 44
ENVIRONMENTAL PERFORMANCE OF TANKER DESIGNS 60 consequence assessment, and thus pursued the alternative approach de- scribed below. Modeling of Oil Spill Transport and Fate Three primary decisions were required to model the expected physical consequences of a broad range of hypothetical spills. First, a tool to model the fate of oil in the environment had to be selected. Such tools allow for an understanding of the likely trajectory of oil in the envi- ronment under any number of hypothetical conditions. Second, the pa- rameters of this model needed to be defined, including the case study locations for which the model would be run, spill sizes, oil type, and weather. Finally, the metrics that would be used to describe the physical consequences of each modeled release event had to be identified. Several models are available that simulate the transport and fate of oil in the marine environment. Three were considered by the com- mittee: (a) NRDAM/CME, developed by the U.S. Department of the In- terior for purposes of natural resource damage assessment; (b) the Outer Continental Shelf model, developed by the Department of the Interior's Minerals Management Service for use in assessing the environmental im- plications of decisions concerning off-shore oil leases; and (c) USCG's TAP model, developed for oil spill response planning. Of these three models, NRDAM/CME is the only one that (a) ad- dresses a range of sites throughout the United States (the TAP model is available only for San Francisco Bay) and (b) allows for thresholds of concern to be varied. In addition, as noted earlier, ASA has developed a stochastic version of NRDAM/CME, SIMAP, that makes it possible to ef- fectively generate a large number of modeled scenarios with a reasonable level of effort. It is this model that the committee applied to the present analysis. The SIMAP model is described in detail by French et al. (1996, 1999) and French and McCay (2001). The model has undergone ex- tensive peer review, been applied widely for both damage assessment and spill response planning purposes, and been accepted for use in natural resource damage assessment by both the U.S. Department of the Interior and NOAA. The physical consequences of a spill event will depend largely on the location of the event, the volume and type of oil spilled, and the weather at the time of and following the spill. For example, the location of a spill will determine shoreline oiling potential; the characteristics of the product will determine the extent to which it partitions to the water

OCR for page 44
DESCRIPTION OF THE PROPOSED METHODOLOGY 61 column and sediments; and the weather following a spill will determine the direction and speed at which the slick will travel, the distance it will travel, and the extent to which the oil will be dispersed in the envi- ronment.4 The committee selected ranges for each of these factors in modeling spill consequence across various spill volumes, as discussed in Chapter 4. The physical consequences of a spill can be described by a range of metrics. Commonly used measures include extent of shoreline oiling, area of slick, oil constituent concentrations in the water column, residual oil constituent concentrations in sediments, and various measures of the "dose" of oil borne by different environmental media [e.g., hours of exceedance of a threshold polycyclic aromatic hydrocarbon (PAH) concentration in the water column]. The committee used four physical consequence metrics (the area of the slick, the length and area of oiled shoreline, and the toxicity in the water column) to generate the equiv- alency ratios, as described in Chapter 4. Response Functions of Spill Size to Consequence The ultimate implications of an oil spill are defined in terms of the envi- ronmental, economic, financial, and social consequences of the event. For example: The owner(s) and insurer(s) of the ship's cargo suffer a financial loss. The responsible party and other entities incur costs associated with the response to the spill and its cleanup. The public may suffer loss of use of an oiled resource (e.g., a beach closure). The public may suffer a loss associated with the reduced quality of an environmental resource (e.g., the loss of nursery habitat for an endan- gered fish species). Private parties may suffer losses (e.g., a commercial fishery may be closed for some period of time). An important resource for a local community (e.g., an artesianal fishery) may be lost or diminished in value. The committee gathered information on the relationship of these factors to spill size. In particular, the committee considered the extent to which 4Weather here is used to refer to wind, currents, tides, wave energy, and similar factors.

OCR for page 44
ENVIRONMENTAL PERFORMANCE OF TANKER DESIGNS 62 these spill outcomes are more or less a function of the physical conse- quence measures discussed above. One option available was to attempt to model each of these mea- sures of consequence individually. For example, it is possible to estimate the likely number of visitor days that would be diminished in value as a result of a hypothetical spill event given information on the location, extent, and duration of beach oiling; the time of year of the spill event; and beach use levels. Similarly, given the volume of oil spilled, the spill location, and the weather conditions at the time of a spill, it is feasible to generate an estimate of expected cleanup costs for a hypothetical spill event. The committee examined a range of measures of spill consequence, as discussed below. Environmental Impact Measures The committee considered the use of several environmental impact mea- sures. In particular, SIMAP provides two categories of environmental con- sequence metrics. The first involves measures of impacts to wildlife, water column organisms, and so on. The second involves measures of impact across shoreline type (e.g., rocky versus wetland). These mea- sures may be better than the physical impact measures discussed above at reflecting the likely true consequence of a spill event. For example, equivalent areas of surface water swept by a slick at two locations may affect dramatically different numbers of biota. Despite the potential advantages of considering environmental consequences,5 the committee chose not to do so, for several reasons. First, there is a great deal of uncertainty in the available measures of species abundance and other factors used to describe environmental re- sources present at each modeled site, and these measures will vary with weather, time of year, and other factors. That is, the environmental con- sequences predicted by SIMAP may significantly over- or understate the actual consequences that would occur from a given spill. Therefore, the results generated by SIMAP for these measures of damage are generally less well accepted by the professional community than are physical impact measures. Second, in selecting case study sites for use in modeling spill con- sequence, the committee's goal was not to obtain precise measures of loss 5Environmental consequences refer to the impact of an oil spill on the affected habitat, whereas physical consequences refer to physical measures of oil concentration in the water or on the shore. In the committee's methodology, physical consequences are used to measure environmental consequences, and the term envi- ronmental impact is used in the report to describe this measure.

OCR for page 44
DESCRIPTION OF THE PROPOSED METHODOLOGY 63 at these specific locations, but to model consequence under a range of pos- sible physical conditions. Including the environmental consequence mea- sures in the model would result in an increased risk of assigning too much weight to the specific conditions found at these sites. Finally, there is reason to believe that environmental consequence will, on average, be linearly related to physical consequence. That is, the physical consequence measures discussed above probably serve as good indicators of likely environmental consequence when considered across all potential spill locations and conditions. Value of Lost Product One category of economic loss associated with oil spills is the value of the lost product. Early in the analysis, the committee determined that, while larger quantities of product may have lower value per unit (and thus the value of lost product from a large spill may not be as great as that from a small spill, per unit volume), the overall effect of omitting this factor from the analysis would be quite small. Thus, this measure was not included in the analysis. Response Costs The committee expended a great deal of effort generating a relationship that would describe oil spill response and cleanup cost as a function of spill size. The committee considered historical spill cost data and ex- isting models of spill cleanup and response costs, as well as primary es- timates of spill response and cleanup costs generated for each of the hypothetical spill locations. The committee asked a recognized expert in the field, Dr. Dagmar Etkin, to prepare a set of reports summarizing available information on spill response and cleanup costs and to provide information that would make it possible to estimate response costs for each of the modeled spills. (See Shoreline Cleanup Cost Modeling and Mechanical Containment and Recovery Cost Models on the accompa- nying CD.) After reviewing these reports, the committee concluded that (a) the historical record of spill response and cleanup costs is not suffi- cient to allow for the estimation of a cost consequence function; (b) esti- mates generated using existing models of spill response and cleanup cost demonstrate a great deal of uncertainty, especially for spills at the low and high ends of the modeled spill size range; and (c) in many cases, the best predictive models are simply linear transformations of the physical con- sequence measures already being considered by the committee.

OCR for page 44
ENVIRONMENTAL PERFORMANCE OF TANKER DESIGNS 64 All efforts to estimate spill response costs for modeled spill events rely to some extent on historical cost data. Since the number of spills for which such data are available is small (especially for larger spills), either inconsistencies in the reported cost estimates or case-specific circum- stances could introduce significant bias into the modeling effort. In ad- dition, the committee determined that the degree of uncertainty inherent in any modeled cost estimate will be greatest for the smallest and largest hypothetical spill events, a factor that could have led to additional bias in the final consequence function. Since the best predictive models are linear transformations of physical consequence measures, the committee decided that these measures are sufficient proxies for response and cleanup costs. As noted below, the committee recommends that additional consideration be given to the development of a robust spill cleanup and response cost function, and that the consequence model developed here be tested for sensitivity to inclusion of alternative measures of consequence. Third-Party Damages The committee considered the development of detailed estimates of third- party damage for each of the hypothetical spill events. The focus of this discussion was on such consequences as beach closures, recreational fishing closures, interruption of other recreational activities, commercial fishing/shellfishing closures, and interruption of marine transportation. As for the biological consequence metrics discussed above, the committee determined that (a) the level of precision that could be achieved in establishing these consequence measures would be low and could vary significantly across case study sites (e.g., determining the im- pacts of a beach closure would require information on numbers of vis- itors to the beach at the time of the spill, which would generally not be known with any certainty); (b) there is good reason to believe that these measures of consequence are, in general, linearly related to physical consequence, already captured in the modeling efforts as described above; and (c) the case study sites were not chosen as specific locations, but as indicative of the kinds of locations at which oil might be spilled (that is, shifting the case study site a few miles could result in a different pattern of expected third-party losses). Thus, as for specific measures of environmental consequence, consideration of site-specific third-party losses might have resulted in assigning disproportionate weight to a par- ticular modeled location.

OCR for page 44
DESCRIPTION OF THE PROPOSED METHODOLOGY 65 Establishment of Equivalency Ratios As discussed above, the committee ultimately selected a modeling ap- proach that (a) uses SIMAP to estimate the physical consequences of a range of hypothetical oil spills at a set of case study sites and (b) assumes that a set of physical consequence measures are reasonable proxies for the financial, economic, social, and environmental consequences of oil spills. Since the committee's approach does not rely on a single metric to report consequence (such as dollar damages), it was necessary to combine the physical consequence measures used into a single mea- sure that could be employed to compare the relative consequences of spills of different sizes. The approach selected involves the use of "equiv- alency ratios," which define the expected consequence of a spill of a given size relative to the expected consequence of a standard reference spill. That is, physical consequence of modeled spill Equivalency ratio = (e.g., meters of shoreline oiled) physical consequence of reference spill (e.g., meters of shoreline oiled) If consequence is constant across spill size per gallon spilled, the equivalency ratio should equal the ratio of the size of the modeled spill to the size of the reference spill. That is, if the equivalency ratio is greater than the ratio of spill size, the consequence per gallon of oil spilled is greater for the modeled spill than for the reference spill. If the equiva- lency ratio is less than the ratio of spill size, the consequence per gallon of oil spilled is less for the modeled spill. The specific analytic steps the committee followed in generating these ratios and the ultimate consequence function are described in Chapter 4. DESIGN COMPARISON The design comparison involves combining the results of the other two steps in the methodology (outflow and consequence analyses). Since both the alternative design and the reference double-hull vessel have been run through the same set of accident scenarios, their performance for each scenario can be directly compared. However, since the conse- quence analysis showed that a direct oil outflow comparison would not

OCR for page 44
ENVIRONMENTAL PERFORMANCE OF TANKER DESIGNS 66 be adequate, the outflow values need to be transformed using an overall summary consequence metric based on the equivalency ratio. By taking this step for each accident scenario, not only can the better design be de- termined, but also the relative impact on the environment can be as- sessed. If both designs perform poorly for a particularly severe accident scenario, both may spill considerable, though different, amounts of oil into the environment. This difference in environmental impact may not be significant; massive damage has already occurred. However, the same difference could be very important if one design did not leak and the other did for a much less severe accident scenario. When repeated across thousands of realistic scenarios, this pairwise analysis done at the scenario level using the overall consequence metric allows for a comprehensive evaluation of the designs. The resulting distribution of differences describes the relative environmental performance of the alternative design and allows for a determination of equivalency. LIMITATIONS OF THE METHODOLOGY While the committee's methodology is sound, it has some limitations in its current stage of development. The four most significant limitations are discussed below, along with suggestions for further development work to address them. Vessel Structure Analysis of a vessel of nonsteel or nonconventional construction would require the use of material performance characteristics dissimilar to those used in the methodology and would necessitate modifications to the for- mulae used in the analysis. To this end, the programs used in the analysis of outflow, DAMAGE and SIMCOL, would have to be modified. Making these modifications would require specific determination of both the ma- terial properties and the performance of those materials. Site-Specific Factors The committee's approach involves modeling a large number of hypo- thetical releases from tanker accidents at given case study sites. To the extent that these case study sites are not representative of typical spill lo- cations, this approach will produce biased results. The sensitivity of the final consequence function to the chosen set of sites could, however, be tested by considering a larger number of case study sites.

OCR for page 44
DESCRIPTION OF THE PROPOSED METHODOLOGY 67 Physical Consequence Measures The committee's approach involves estimating the extent to which physical consequence predicts total consequence. As noted above, the committee assumed that various physical consequence metrics can be used as proxies for the expected environmental and third-party costs of a spill, as well as the expected response and cleanup costs. Several limitations may be intro- duced by this assumption. The most significant of these is that third-party costs and response and cleanup costs are likely to be disproportionately larger for small spills. That is, the physical consequence measures used by the committee are likely to understate third-party and response and clean- up costs for small spills. Further investigation of this limitation could lead to more confidence in the final outcomes. Modeled Grid Size Used in SIMAP SIMAP estimates the physical consequences of spills using a geographic grid size that varies by location. In the case of small spills, the amount of oil in the water may not be sufficient to register the physical consequence measures. Thus while small spills do cause environmental harm, the version of SIMAP used by the committee assigns some modeled spills a physical consequence of zero. The result is understatement of the con- sequences of some small spills. The SIMAP modeling approach does adjust for this factor to a certain extent by assuming that the physical con- sequence of a spill can never be less than the smallest measurable unit. However, it would be useful to do additional modeling with smaller grid sizes to understand the potential bias introduced by this factor and pos- sibly modify the model for future uses. SUMMARY OF THE METHODOLOGY In summary, the methodology proposed by the committee involves two initial steps--one to calculate oil outflow following a tanker accident and the other to assess the consequence or impact of an oil spill. These two steps are then combined in a third and final step--the comparison of tanker designs. The oil outflow calculation involves the computation of a series of accident scenarios causing specific damage to a ship that, in turn, re- sults in a specific quantity of oil spilled. Each design to be evaluated will be subjected to the same series of accident scenarios, but because they are different designs, different amounts of oil will be spilled.

OCR for page 44
ENVIRONMENTAL PERFORMANCE OF TANKER DESIGNS 68 The consequence assessment involves assigning a rational metric for the environmental consequence of a series of oil spill events of varying quantities, locations, and conditions, and then constructing and using a multiplier to transform the outflow differences for two designs determined in the first step into consequence differences. The committee carried out this step, and the resulting consequence curve can be applied to different design comparisons. The consequence curve needs further refinement, as discussed above, but once the refinement has been completed, this step does not need to be repeated for each design comparison. The third and final step, design comparison, involves comparing the results of the measurement of differences in consequence for the design being evaluated with the reference vessel. Under some conditions, one design may have lower consequences and thus be a better performer, but under other conditions the opposite may be true. The methodology provides the data and analyses needed to make this comparison for a large number of individual conditions and to conclude whether one design is equivalent, inferior, or superior to another. As noted above, the committee believes its methodology is sound, systematic, and well specified, and represents a significant improvement over methods proposed and used in the past. Chapter 4 describes the ap- plication of the committee's methodology in more detail and provides ex- amples of the necessary computations, descriptions of the analyses, and illustrations of the resulting graphics that would be used to compare actual designs. REFERENCES ABBREVIATION SNAME Society of Naval Architects and Marine Engineers Astrup, O., I. Monnier, and J. Sirkar. 1995. Framework for Evaluating Alternative Designs and Configurations for Tankers. Proc., International Conference on Technologies for Marine Environment Preservation (MARIENV 95), Society of Naval Architects of Japan, pp. 183190. French, D., M. Reed, K. Jayko, S. Feng, H. Rines, S. Pavignano, T. Isaji, S. Puckett, A. Keller, F. W. French III, D. Gifford, J. McCue, G. Brown, E. MacDonald, J. Quirk, S. Natzke, R. Bishop, M. Welsh, M. Phillips, and B. S. Ingram. 1996. The CERCLA Type A Natural Resource Damage Assessment Model for Coastal and Marine Environments (NRDAM/CME). Technical Documentation, Volumes IVI. Final report. Office of Environmental Policy and Compliance, U.S. Department of the Interior, April. French, D., H. Schuttenberg, and T. Isaji. 1999. Probabilities of Oil Exceeding Thresholds of Concern: Examples from an Evaluation for Florida Power and

OCR for page 44
DESCRIPTION OF THE PROPOSED METHODOLOGY 69 Light. Proc., 22nd Arctic and Marine Oil Spill Program Technical Seminar, Environment Canada, pp. 243270. French, D., and D. McCay. 2001. Development and Application of an Oil Toxicity and Exposure Model, OilToxEx. Final report. NOAA Damage Assessment Center, Silver Spring, Md., Jan. Michel, K., and C. Moore. 1995. Application of IMO's Probabilistic Oil Outflow Methodology. Presented at SNAME Cybernautics 95 Symposium, New York. Michel, K., C. Moore, and R. Tagg. 1996. A Simplified Methodology for Evaluating Alternative Tanker Configurations. Journal of Marine Technology, Vol. 1, pp. 209219. Minorsky, V. U. 1959. An Analysis of Ship Collisions with Reference to Protection of Nuclear Power Plants. Journal of Ship Research, Vol. 3, No. 1. Pedersen, P. T. 2000. Risk Assessment Procedures for Fixed Structures in Shipping Lanes. Technical University of Denmark, Copenhagen. Sirkar, J., P. Ameer, A. Brown, P. Goss, K. Michel, F. Nicastro, and W. Willis. 1997. A Framework for Assessing the Environmental Performance of Tankers in Accidental Groundings and Collisions. Presented at the SNAME Annual Meeting, Oct. Tikka, K. K., and Y. J. Chen. 2000. Prediction of Structural Resistance in Grounding: Application to Structural Design. Ship Structures for the New Millennium: Sup- porting Quality in Shipbuilding, Arlington, Va., June. Wang, G., H. Ohtsubo, and D. Liu. 1997. Simple Method of Predicting the Grounding Strength of Ships. Journal of Ship Research, Vol. 41, No. 3, Sept.