Hazardous Materials Transportation Risk Assessment: State of the Practice (2013)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

88 The following section summarizes the major findings from the literature review. It is organized into four primary headings: • New Modeling Techniques and Approaches • Data-Driven Risk Assessment • The Use of Risk Analysis and Route Choice • Economic Risk Analysis The documents and articles summarized were found to be the most comprehensive and detailed in each of these areas, but other resources reviewed are listed with brief notations at the end of each section. A.1. New Modeling Techniques and Approaches The amount of research and data collected concerning haz- ardous materials transportation continues to grow every year. As more data exists, newer and more accurate models can be developed to measure the risk associated with transport- ing hazardous materials. The following articles demonstrate how modeling techniques and approaches can be developed by collecting and analyzing new data. A.1.1. Cross-Analysis of Hazmat Road Accidents Using Multiple Databases Martin Trépanier, Marie-Hélène Leroux, and Nathalie de Marcellis-Warin, Accident Analysis and Prevention, Vol. 41, Issue 5, 2009; pp. 1192-1198. Route-based risk researchers often have difficulty in finding strong data pertaining to hazmat accidents; while datasets do exist, not all of them collect the same data or even all the data that is needed to analyze accidents. As a result, the authors set out a methodology and a tool that integrate multiple data sources to analyze hazmat road accidents in Quebec, Canada. The article looks at three databases for Quebec, Canada: Dan- gerous Goods Accident Information System (DGAIS) from Transport Canada, Road Accident Database from the Societe de l’assurance – automobile de Quebe – SAAQ, and Community Database on Work Accidents from the Commission de la santé et de la securite du travail du Quebec. DGAIS has information on spills, injuries, death counts, and other parameters related to hazmat, but the information reported is only for instances in which Canadian laws require it. Data of the Road Accident Database is collected by Quebec police officers for each acci- dent involving human or large financial consequences. The Community Database on Work Accidents contains accident information where workers were injured or killed on duty. Once the data was gathered, the databases were integrated using the Transportation Object-Oriented Modeling (TOOM) approach which allows analysis of data in relation to other trans- portation sources. The approach is structured around four meta-classes of objects: dynamic objects, kinetic objects, static objects, and system objects. The authors create a master table for each dataset that contains the raw, unmodified data, includ- ing dates, time, materials, etc., after which, they use the Hazmat Event Cross-Observer Tool (HECOT), which identifies the same event through different databases and the possible cause of the event, to observe data about single events across several datasets. HECOT focuses on identifying events based on spatial and temporal data contained within the three databases. Unfortunately, the Database on Work Accidents does not identify the location, so no match could have been called exact, and the authors focused on the first two databases. Even then, it was difficult to make true matches without further investi- gation using information from Quebec Ministry of Transport. Finally, the temporal and spatial criteria were widened to allow for imprecision of events’ location, time, and general data. As a result, 41 true matches were found between the two databases. Those matches accounted for 28.1% of the accidents reported in the DGAIS and 2.9% of the reported accidents in the Road A p p e n d i x A Literature Review Results

89 but it is not feasible to prepare for every possible accident on every ship. The risk of each possible incident must be deter- mined so that a shipper can determine which risks it would like to prepare against. For example, relative probabilities of hull failure can be determined based on ocean conditions, including wave frequency. Portions of the ocean with a wave frequency associated with a high relative probability of hull failure can be avoided, if desired. A.1.4. Comprehensive Risk Assessment for Rail Transportation of Dangerous Goods: A Validated Platform for Decision Support Adrian V. Gheorghea, Reliability Engineering & System Safety, 2005; pp. 247-265. Most risk assessment models are reactive and only estimate risk after an event has already occurred. Research is needed to help determine causes of incidents that force the release of hazardous materials. By determining why incidents occur, this research will help limit the number of accidents and releases involving hazardous materials. Calculating the consequences of a release of hazardous materials is discussed. Factors such as wind speed, air pressure, and average temperature contrib- ute to the release of airborne particles. Assumptions are also made for incidents involving fires and shock waves. Expected consequences of an incident can influence route choice to avoid hot spots, locations with a large population center or high risk of an incident. This study analyzes past incidents and categorizes each incident into one of approximately 30 groups. These groups include items such as rail failure, a foreign object on the track, maintenance on the track, and excessive speed. Identifying common causes of the release of hazardous materials can help inform future planning and risk assessments. Data regarding route choice and accident probabilities can be adjusted to be more accurate based on past incidents. A.1.5. The Weighted Risk Analysis Shahid Suddle, Safety Science, 47, 2009; pp. 668-679. Suddle discusses how buildings in The Netherlands are being erected near and above hazardous materials transpor- tation routes. Suddle expands on the quantitative risk analy- sis framework that the Dutch regulations require to assess the safety of such projects to allow other aspects of risk in the decision-making process. He proposes a ‘weighted risk analysis’ (WRA) methodology that would allow the effect of safety measures to be optimized with regards to environment, quality, political, and economical aspects, thereby expanding the decision criteria from economic and human risks. Accident Database. The authors conclude that there is under- reporting of accidents in both databases. A.1.2. A Behavioural Model for the Level Crossing Collision Risk Assessment M. Ghazel, Safety and Security Engineering, III, 2009; pp. 637-645. Many accidents occur every year at grade-level rail inter- sections. At these crossings, trains cross over roads on the same level as vehicular traffic. There is inherent risk associated with these crossings, despite safety measures which include physical barriers and flashing lights and bells that alert drivers to approaching trains. Vehicles often stop in the train crossing zone when a train is not nearby. When a train then approaches, the vehicles are unable to leave the crossing zone and cause a collision. To address this problem, behavioral models were developed to examine the risk caused by the multiple participants, includ- ing the train, the vehicles, and the railway equipment, such as the lights and bells. Because it is believed that people’s behavior causes most of these collisions, like when a person drives their vehicle on to the tracks, these behavioral models will allow the risk associated with each participant to be calculated. Attempt- ing to quantify people’s behavior to measure risk will allow some subjective criteria to be judged more quantitatively. Modeling the risk of a collision at a level crossing involves multiple smaller models. One elementary model was created to demonstrate the relationship between the train and the signals, while a separate model was created to show the rela- tionship between vehicular traffic and the signals. A complex model was then developed combining both basic models and to examine a possible accident between a train and vehicular traffic. The model is particularly complex because the trains and vehicular traffic never have any direct interaction. By quantifying individual behaviors, it becomes much easier to evaluate the risk of an accident at a level crossing. A.1.3. Risk Assessment in Maritime Transportation C. Guedes Soares, A.P. Teixeira, Reliability Engineering and System Safety, 2001; pp. 299-309. Maritime transportation carries many risks not typically associated with transportation by rail or truck. Transport- ing cargo across oceans requires more manpower and larger equipment than any other mode. If a ship has an accident in the middle of the ocean, there is a greater potential for fatali- ties to the crew than with other modes of transportation. This study addresses major causes of ship loss, including fire, explosion, and foundering. Probabilistic approaches can be utilized individually to address each major cause of ship loss,

90 The population-at-risk factor is then calculated by multiply- ing the hazmat accident probability and the population-at- risk, as determined by evacuation distances, for each route. Alternatively, risk can be assessed through the use of use of the potential hazard rating (PHR), which measures the potential hazard posed by hazmat based on the volume of materials and the evacuation distance by class. Including a PHR in a risk analysis makes it easier to inject a more sensi- tive measurement of incident severity into any risk equation. A.1.7. Review of the Department of Homeland Security’s Approach to Risk Assessment Committee to Review the Department of Homeland Secu- rity’s Approach to Risk Analysis; National Research Council, http://www.nap.edu/catalog/12972.html, 2010. The committee was tasked by the U.S. Congress to assess risk assessment approaches across the Department of Homeland Security and to offer suggestions on ways to improve upon the department’s approaches. The approach for the report was to review six illustrative risk models in use in the department. While the committee does not explicitly discuss hazardous materials transportation, they do make over-arching state- ments about the approaches that DHS entities use to assess risk. The committee found that DHS’ risk approach (as a func- tion of threat, vulnerability, and consequence): . . . appears appropriate for decomposing risk and organiz- ing information, and it has built models, data streams, and pro- cesses for executing risk analyses for some of its various missions. However, with the exception of risk analysis for natural disaster preparedness, the committee did not find any DHS risk analysis capabilities and methods that are yet adequate for supporting DHS decision making, because their validity and reliability are untested. As a result, the committee recommended that the depart- ment’s risk assessments for terrorism need to incorporate a peer review process that includes technical experts that are not DHS employees. The document continues along this call for great transparency by recommending that: To maximize the transparency of DHS risk analyses for decision- makers, DHS should aim to document its risk analyses as clearly as possible and distribute them with as few constraints as possible. Further, DHS should work toward greater sharing of vulnerability and consequence assessments across infrastructure sectors so that related risk analyses are built on common assessments. The paper goes on to further express concern about DHS’ all-hazards approach because the authors do not feel that terrorism risk and natural hazard risk can be combined and analyzed based on the same metrics. The authors recommend Suddle takes the Risk = Frequency × Consequence risk model and introduces weighting with the final overall risk being equal to the sum of the risks of each aspect: 11 Risk R where the monetary value per considered loss cost unit final j ij ij ∑∑ ( ) =   = == ~ ~ This risk value is then put into a Cost equation to be minimized: 1 , , 0 1 0 C C y Risk r where C y investment in the safety measure y j number of the year r real rate of interest total final j j ∑( ) ( ) ( )= + + = = = = The author states that the monetary value per considered loss, ~, can be found through research; however, varying the values given for each considered loss in the weight can have a strong impact on the final weighted risk value and, thus, over the total decision-making process. A.1.6. Risk Assessment of Transporting Hazardous Material: Route Analysis and Hazard Management K. David Pijawka, Steve Foote, and Andy Soesilo, Transporta- tion Research Record 1020, 1985. In this article, the authors delve into the consequence side of hazmat transportation risk by discussing the vulnerability of communities to hazardous material accidents. The authors describe that the lower the vulnerability of a community, whether through preparedness or risk mitigation, the lower the possible consequences. To emphasize their point and further discuss overall indus- try risk analysis, the authors use hazmat transportation on major highway routes in Arizona. The first step in the risk analysis is to identify the hazardous materials being shipped, the amount of hazmat cargo, and the routes used to ship the cargo. Next, exposure-miles, the total number of miles tra- versed annually by vehicles carrying hazmat on a route-by- route basis, is calculated by: (1) applying the load-per-vehicle factors to the weight of hazmat transported by hazard class, which gives the number of trips by class; (2) summing the number of trips by class for an entire route; and (3) multiply- ing the number of trips by real travel miles along individual routes. Then, the authors calculate the probability of a hazmat accident by multiplying the prevailing accident rate by the number of total number of miles of exposure on each route.

91 Analyzing Mitigation of Container Security Risks Using Six Sigma DMAIC Approach in Supply Chain Design Sameer Kumar, Heidi Jensen, Heather Menge. Transportation Journal, Lock Haven: Spring 2008. Vol. 47, Iss. 2; pp. 54-67. • Discussion of container supply chain safety; how to reduce risk. • While there is not much mention of measuring risk, it does address reducing risk and improving safety. Real-Time Crash Risk Reduction on Freeways Using Coor- dinated and Uncoordinated Ramp Metering Approaches Mohamed Abdel-Aty, Vikash Gayah. Journal of Transporta- tion Engineering, New York: May 2010. Vol. 136, Iss. 5; p. 410. • Reducing risk of crashes on freeway. • Measuring risk and reducing crashes through various algorithms. • Solid analysis of risk determination for highway crashes. A Tool for Risk Analysis and Protection Design of Railway Infrastructures Angela Di Febbraro, Federico Papa, Nicola Sacco. Transporta- tion Research Board Annual Meeting, Washington, DC: 2010. Paper #10-0540. • Presents an “easy-to-use” risk analysis tool, based on a modular architecture, to evaluate risk and provide mitiga- tion indicators. • Uses real world experiences made by authors on the Italian high-speed rail lines to apply the tool. • The tool can be tailored to other real-world situations. Environmental Risk Analysis of Railroad Transportation of Hazardous Materials M Rapik Saat, Christopher P.L. Barkan, Charles Werth, David Schaeffer, Hongkyu Yoon. Transportation Research Board 89th Annual Meeting, Washington, DC: 2010. Paper #10-2174. • Research sponsored by Association of American Railroads to estimate environmental risk when transporting LNAPL chemicals via freight rail. • Uses the Hazardous Materials Transportation Environmental Consequence Model (HMTECM) with a geographic infor- mation system (GIS) for probabilistic estimates of exposure to different spill scenarios. • Risk analysis incorporated the estimated clean-up cost based on HMTECM, route-specific probability distribu- tions of soil type and depth to groundwater, annual traffic, railcar accident rate, and tank car safety features. a move from decision-making approaches based on pure quantitative analysis to ones with scientific guidelines with different approaches for different risks. The committee conducted an in-depth review of the Ter- rorism Risk Assessment and Management (TRAM) toolkit, which is a software-based approach to assessing risk primar- ily within the transportation sector. The tool is used by the industry and has six steps: Criticality Assessment, Threat Assessment, Vulnerability Assessment, Response and Recov- ery Capabilities Assessment, Impact Assessment, and Risk Assessment. The tool uses user-input data, often arrived at through elicitation of subject-matter experts, to determine the values in the first five steps. Currently, TRAM’s main use is in regards to terrorism; however, progress is being made to expand this to man-made and natural disasters. In addition to expanding TRAM’s risk scope beyond ter- rorism, the Committee states that the tool is overly complex despite the subjective and speculative nature of the inputs. Despite their concerns about TRAM’s complexity, the authors did claim that the outputs aid in conceptualizing the risk space, ranking different risks, and showing how risk mitiga- tion activities affect those rankings. The only TRAM-specific recommendation made in the report was to have the tool vet- ted through a peer review process to evaluate its reliability and identify any improvements that could be made. A.1.8. Additional Resources The Freight Transport Portfolio: A New Way to Analyze Inter- modal Freight Transport as Compared to Single-Mode Road Transport Bart W Wiegmans. Transportation Journal, Lock Haven: Spring 2010. Vol. 49, Iss. 2; pp. 44-53. • Calculations of expected performance and reliability. • Discussion of risk of poor performance from rail trans portation. • Little mention of safety or risk assessment from safety viewpoint; mostly financial risk and reliability considered. Risk Assessment for the Security of Inbound Containers at U.S. Ports: A Failure, Mode, Effects, and Criticality Analysis Approach Sameer Kumar, Janis Verruso. Transportation Journal, Lock Haven: Fall 2008. Vol. 47, Iss. 4; pp. 26-42. • Discusses developing a risk assessment for cargo at ports; can be extended to hazmat. • Includes risk of catastrophic event caused by sabotage, a security lapse, equipment malfunction or human error. • Suggestions for reducing risk as well as to more accurately measure risk.

92 dled in a safe manner, as they pose a tremendous amount of risk to the surrounding areas. In this paper, researchers have developed a decisional support system to identify risks asso- ciated with transportation hazardous materials. This model integrates a database covering each mode of transportation in the study region with GIS capabilities to illus- trate potential risks. The database includes what materials are being transported, the length that they are being transported, the conditional release probabilities of all materials, and popu- lation data for surrounding areas. By developing an algorithm that includes how dangerous a material is and how close that material travels near population centers, this model allows the risk to be determined with regards to the number of people potentially affected. A.2.2. An Expeditious Risk Assessment of the Highway Transportation of Flammable Liquids in Bulk Theodore Glickman, Transportation Science. 1991; pp. 115-123. There are many regulations concerning the transportation of hazardous materials through New York City. Almost all hazardous materials incidents occur because of an accident or because of container failure. Accidents tend to be more dan- gerous because of the force involved. Container failure nor- mally only involves a spill of material, which tends to be less lethal than the explosion or ignition of hazardous materials. In this study, differing regulations are analyzed to determine the impact of container and route choice. Conditional release and accident probabilities must be cal- culated so that expected outcomes can be determined. This study focused on the use of two different containers, two routes (one considered typical and one considered the “most hazard- ous”), and two risk scenarios (the average case and the worst case). These scenarios were then analyzed to determine the risk of a release, a release that leads to a fire, and an explosion. An estimated number of fatalities was also determined based on the expected outcome of the above scenarios. By using data to determine risk, a risk model can estimate the potential number of fatalities and other consequences of various transportation scenarios. A.2.3. A New Approach to Hazardous Materials Transportation Risk Analysis: Decision Modeling to Identify Critical Variables Renee M. Clark and Mary E. Beserfield-Sacre, Risk Analysis, Vol. 29, No. 3. 2009. The authors design a methodology to assess hazmat trans- portation risks at the loading and unloading stage through the use of a probability and statistics-based approach. The • Annual per car-mile and per ton-mile risk was calculated, too, which allowed financial comparisons to be drawn between shipping different chemicals. • Risk reduction estimates were also investigated based on the use of damage-resistant tankers. Risk Analysis and Reliability Based Design in Tunnel Fire Safety M Guarascio, M Lombardi, G Rossi, G Sciarra. Third Inter- national Conference on Safety and Security Engineering, Rome, Italy: Jun. 2009; pp. 575-584. • This article covers transportation risk assessment. • Authors develop a quantitative risk analysis procedure that focuses on smoke control system effectiveness and the spall- ing effect and structural reliability of the liners. Transportation Risk Analysis Tool for Hazardous Substances (TRANS) – A User-Friendly, Semi-Quantitative Multi-Mode HazMat Transport Route Safety Risk Estimation Methodol- ogy for Flanders G L L Reniers, Katleen De Jongh, Bob Gorrens, Dirk Lauwers, Maarten Van Leest, Frank Witlox. Transportation Research Part D: Transport and Environment, Dec. 2010. Vol. 15, Iss. 8; p-489-496. • Describes a methodology for assessing the relative risk levels in moving hazardous materials by various modes. • TRANS divides routes into small segments using multi- criteria analysis and incident likelihood scores. A.2. Data-Driven Risk Assessment Many researchers believe that the most accurate risk assess- ments are based on data and are not subjective. As such, researchers often attempt to use as much data as possible in their risk assessments. The following papers demonstrate algo- rithms and calculations using large amounts of data. Larger datasets remove some uncertainty from risk assessments and may more accurately calculate risk. A.2.1. A Decisional Support System to Quantify Risk Due to the Transportation of Dangerous Substances A. Romano and G. Romano, Safety and Security Engineering III, 2009; pp. 565-573. More hazardous materials are being manufactured than ever before and the transportation of hazardous materials carries a large amount of risk. These materials must be han-

93 A.3. The Use of Risk Analysis and Route Choice Individual route choice contributes significantly to the amount of risk associated with transporting a particular ship- ment. Determining which routes have a greater risk allows a shipper to determine if they are willing to accept higher risk in exchange for lower operating costs. The seemingly short- est and cheapest route might not be best if it travels through a large population center or poses other risks. The following papers illustrate the usefulness of determining routes by bal- ancing risks with costs. A.3.1. A Framework for Risk Assessment and Decision-Making Strategies in Dangerous Goods Transportation B. Fabiano*, F. Cuffô, E. Palazzi, R. Pastorin, Journal of Haz- ardous Materials. 2002; pp. 1-15. While transporting hazardous material by rail involves large quantities of hazardous materials, transport by road is often more dangerous because roads tend to travel through higher populated areas. Data must be collected to describe the population on potential transport routes. The authors created a model to analyze the impact of route choice in various populations. For example, a route can be determined to have a small, yet vulnerable population. This could influence transportation planners and policy makers to avoid this smaller population and steer hazardous materials towards a larger population with a greater chance of survival. Many of these decisions are entirely subjective and politi- cal, but the model offers an objective look at the potential impacts of various route choices. A.3.2. Road Transportation of Dangerous Goods: Quantitative Risk Assessment and Route Comparison Philippe Cassini, Journal of Hazardous Materials, 61. 1998; pp. 133-138. The author sets out to compare choices in trucking routes between urban areas and alternatives that include tunnels between 2 and 9 km in length. The key data points are popula- tion density along the route, all vehicular traffic along the pos- sible routes, the dangerous goods itself (how it is contained), global annual traffic, weather along the route, the lay-out of the open air routes, and, if applicable, disposition taken for the design and equipment of the tunnels. The program looks at 10 scenarios to evaluate the risk along the route. The author uses a Fortran-written program to evaluate the risks for open-air travel by developing F/N curves that show yearly frequency against number of fatalities. For tunnels, the author uses a spreadsheet tool to determine the F/N curves. methodology first simplifies, in part through the use of latent class analysis (LCA), the variables from data collected through the DOT’S HMIRS database, which contains data on hazmat releases. Next, the authors measure the relationships between variables using a log-linear analysis. The authors then develop a Bayesian network decision model. After aggregating container- failure variables with natural subgroups (area type, land use, geographic division, season, shift, material type, container type, release quantity, dollar loss), the authors use the LCA to simplify the five sets of binary variables (contributing action, causing object, failure mode, failure item, and failure area). Through the log-linear analysis, the authors found relationships between: material type and container type, season and material type, and shift and location. Finally, 40,191 accident records were divided into five sets so five-fold cross-validation could be done, result- ing in five Bayesian networks. The networks were accurate 70% of the time in regards to “dollar loss” and 87% of the time for “release quantity.” The authors used GeNIe software to determine that “caus- ing object” was the leading explanatory variable based on five networks with regard to medium, small, and zero dollar loss. “Failure item-area” and “contributing action” were, respec- tively, the second and third leading explanatory variable. With regards to release quantity, contributing action, failure mode, and failure item-area were first, second, and third, respectively, leading explanations. The authors conclude with two tables that explore the best targets, according to their model, for reducing risks by miti- gating the explanatory variables. As a result, their decision model can support decision making at the Office of Hazard- ous Materials Safety. A.2.4. Additional Resources Risk-Based Volume Warrants for Free Right-Turn Lanes on Two-Lane Roadways Jidong Yang. Journal of Transportation Engineering. New York: Apr 2008. Vol. 134, Iss. 4; p. 155. • Risk assessment of vehicle crashes as cars slow to make a right-hand turn. • Determining probability of likelihood of crash based on modeled scenarios; can be extended to hazmat. Crashing, Smashing, Evaluating Kathi Kube. Trains. Milwaukee: Dec 2010. Vol. 70, Iss. 12; pp. 16-17. • FRA R&D testing specialized rail cars, designed to trans- port hazmat. • Describes FRA testing research. • Physical testing of pressure cars.

94 regulator sets a toll where it minimizes population exposure and travel costs, with the use of tolls allowing for the differen- tiation between carriers. The toll (either positive or negative value) can be set in a way that the regulator and the carriers’ optimal routes are the same. The authors looked at hazmat trucking along the high- way system in Western Ontario, Canada, and show how toll-setting policies can achieve higher reductions in the associated transport risks while only slightly increasing the carriers’ costs. A.3.5. Utilization of Accident Databases and Fuzzy Sets to Estimate Frequency of HazMat Transport Accidents Yaunhua Qiao, Nir Keren, M. Sam Mannan, Journal of Haz- ardous Materials. 2009. The authors present . . . a methodology to estimate the accident frequency for differ- ent types of roads by incorporating the effects of a larger number of parameters, including the nature of truck configurations, oper- ating condition, environmental factors, and road condition. The authors consider the HMIS, which contains informa- tion pertaining to hazmat spills and accidents that occur on interstates; state Department of Public Safety (DPS) accident databases, which can contain information about the route and environmental conditions surrounding the accident; and the Commodity Flow Survey, which has information about miles traveled. The methodology has two sets of parameters: route-dependent (lane number, weather, population den- sity) and route-independent (truck configuration, container capacity, driver experience). Their procedure to estimate accident frequency is: (1) Number of accidents is derived from the DPS databases as a function of route-dependent parameters. (2) The corre- sponding vehicle-mile data are obtained from state DOT’s or transportation institutes and from the 2002 CFS . . . (3) The basic accident frequency is modified by considering the effects of route-independent parameters. Fuzzy logic is employed to incorporate expert knowledge. The authors used a highway in Texas to determine the fol- lowing relationships: increases in population density and in number of lanes lead to an increase in the frequency of acci- dents, weather conditions (clear or other) affect frequency, increase in both the complexity of vehicle configuration and container capacity resulted in higher frequency. In the end, the methodology provides information and data that improves upon the frequency term in hazmat truck- ing risk assessments. The author does point out that the research is limited by the low number of scenarios corresponding to a very small num- ber of hazardous materials. Also, there is difficulty in maintain- ing and forecasting data for future traffic patterns; however, the author points out that it is possible to judge the acceptability of the risks due to the transportation of dangerous goods on the route by comparing the ‘F/N’ curve with acceptance criteria. A.3.3. Preliminary Study on the Transport of Hazardous Materials Through Tunnels Roberto Bubbico, Sergio Di Cave, Barbara Mazzarotta, Barbara Silvetti, Accident Analysis & Prevention. 2009; pp. 1200-1205. The authors conducted a risk analysis to determine the dif- ferences in risk for the shipping of hazardous materials via rail and road through tunnels versus open air. They used the American Institute of Chemical Engineers (AIChE)’s 2000 “Guidelines for Chemical Process Quantitative Risk Analy- sis” and fault tree analysis to determine the risk scores for two potential release scenarios—15 mm hole, 15 minute duration, and whole tanker release via 220 mm hole—of gasoline, LPG, liquefied chlorine and liquefied nitrogen. The risk analysis used TrHazGis software for risk assessment and management. For rail, tunnels have a lower societal risk than whole route in the open, perhaps due to the lower number of people at risk inside the tunnel (in highway tunnels, the presence of additional vehicular traffic contributes to the risk). For truck- ing, tunnels increase the societal risk due to the changes in the fault tree of the tunnel scenarios. For instance, LPG may create a jet fire in the open, but in the tunnel there is the pos- sibility of more hazardous outcomes such as BLEVE, vapor cloud explosion (VCE), or fireball. Finally, in the case of an inert gas (nitrogen), the risk associated with rail is extremely low, and for road, the risk is somewhat greater. A.3.4. Toll Policies for Mitigating Hazardous Materials Transport Risk Patrice Marcotte, Anne Mercier, Gilles Savard, Vedat Verter, Transportation Science, Vol. 43, No 2, May 2009; pp. 228-243. The paper focuses on governments’ attempts to regulate the trucking routes that hazardous materials are shipped through by closing road segments based on minimizing population exposure and transport costs. The authors pro- pose not closing road segments based on regulator decision but instead through the use of tolls for transporting hazmat through certain road segments. By using tolls, the govern- ment can set fees that would impact the shippers’ cost-benefit analysis and create incentive for the shipper to find alterna- tive, less-populated routes. With a toll-setting policy, the

95 Quantitative risk analysis focuses on basic concepts and methodology, calculation techniques, data requirements and limitations, results and presentation formats, and common pitfalls. Where data and availability are concerned, the authors suggest that simple scenarios are usually better for transpor- tation risk analysis. Data used should be the most applicable, statistically sound data available. For example, specific carrier accident rates are preferred to generic truck accident rates. Generic rates should not be used for specialized conditions, such as for tank truck shipments. The authors caution about using rates derived from different sources for the numerator and denominator. There is also a discussion about endpoint criteria for consequence analysis, including toxic chemical exposure, vapor cloud explosion, and flammability hazards. Risk presentation includes the use of risk contours, risk tran- sects, and F-N curves as well as a discussion of uncertainty. Transportation security is covered separately from safety but the similarities are clearly highlighted. There is a discussion of the synergies and tradeoffs between the two. Security prioritization and vulnerability assessment are included. The book includes a treatment of risk reduction strategies that addresses balancing safety and security, factors influ- encing risk reduction options, pre-shipment risk reduction options, and selecting strategies. The final chapter discusses sustainability of a risk management program that addresses incorporating advances in analysis techniques and best practices. A.3.7. City of Boston Hazmat Route Evaluation Battelle, City of Boston Department of Transportation. April 2011. This report is one of the few route risk analyses publicly available and examines alternatives necessitated to the exist- ing (and grandfathered) routing restrictions after completing of the “Big Dig” project that depressed some of the major roadways through Boston; some of the previously designated routing alternatives no longer existed. This study imple- ments the routing guidelines for non-radioactive hazardous materials specified in 49 CFR 397.71(b) and in accordance with the DOT document Highway Routing of Hazard- ous Materials, Guidelines for Applying Criteria to Designate Routes for Transporting Hazardous Materials (Publication No. FHWA-HI-97-003). The report discusses the iterative approach for identifying alternate routes for consideration and eliminating those with insufficient roadway width, clearance restrictions, or bridge conditions. Truck accident rates were determined separately by the University of Massachusetts from truck flows provided by the authors and accident data from the City of Boston. The A.3.6. Guidelines for Chemical Transportation Safety, Security, and Risk Management Center for Chemical Process Safety of the American Institute of Chemical Engineers and John Wiley & Sons, Inc., Hoboken, New Jersey. 2008. This resource is designed to augment Guidelines for Chemi- cal Transportation Risk Analysis published in 1995. It adds a broader perspective and includes “more qualitative and prac- tical techniques for screening, identifying, and managing higher-level risk issues that balance both safety and security.” Recognizing the nature of many operations, the book exam- ines risk from an international perspective. The book puts risk analysis into the overall context of safety and security management systems and discusses the key risk assessment concepts of (1) identification and prioritization, (2) risk analy- sis, (3) risk evaluation, and (4) risk reduction. The authors present a concise differentiation between risk analysis (the pro- cess of evaluating likelihood and consequence and estimating risk) and risk assessment (the process of taking risk analysis results and using them to help make decisions). The authors discuss the scope of a potential assessment, to consider all movements and materials or with some restric- tions to material, mode, route, or some smaller subset of all shipments. Four types of initial prioritization options are listed, with each based on (1) hazard, (2) consequence, (3) likelihood, or (4) risk, respectively. Qualitative and semi-quantitative risk assessments are pre- sented as good practices to focus the more intensive detailed quantitative risk assessments on those operations where they can be most beneficial. A key factor discussed is the impor- tance of company-specific evaluation criteria to guide the decision making process. These criteria need to be consistent with the organizations risk tolerance and established before any analyses are conducted. The use of benchmarking comparisons in qualitative analy- ses is presented as a good way to begin and help identify other areas where additional detail or analysis is needed. Qualita- tive reviews should also consider the nature of any anticipated changes in operations, including new materials, modes, pack- aging, carrier, quantities, and many others. Semi-quantitative analyses have a benefit of not requiring risk management experts to conduct; many others are pre- sented and include ease of application and update, ability to address a range of consequences and likelihoods, and efficient use of resources. The book includes a wide range of topic areas and questions to assist in conducting these types of analyses. Ultimately, the results of individual elements are brought together and one approach is to use a risk index. A second approach presented is a risk matrix. Both are very helpful in identifying the key areas for further focus and analysis.

96 B. Fabiano, E. Palazzi. International Journal of Heavy Vehicle Systems. Genève, Switzerland: 11 Oct 2010. Vol. 17, Num- bers 3-4; pp. 216-236. • Use an Italian case study to assess the risk associated with transporting hazardous materials by heavy vehicles. • Analytical model for solving the ventilation design for both plane and sloping tunnels is shown. Strategic Thinking and Risk Attitudes in Route Choice: Stated Preference Approach Michael Razo, Song Gao. Transportation Research Record: Journal of the Transportation Research Board, No. 2156, Wash- ington, DC: 2010; pp. 28-35. • Conducts a stated preference study to investigate route choice behavior in networks with risky travel times and real-time information. • Two maps are used to determine if routes were adapted as new information about the designated route becomes available: – Map one measures the basic risk attitude of the subject by offering the choice between a stochastic route and a deterministic route. – Map two allows for strategic planning and measures the effect of this opportunity on subject’s choice behavior by containing real-time information and an available detour. • Data gathered is used to estimate several choice models based on travel times and standard deviations as explana- tory variables. Risk Assessment for Transportation of Hazardous Materi- als Through Tunnels E G Nathanail, S Zaharis, N Vagiokas. Transportation Research Record: Journal of the Transportation Research Board, No. 2162, Washington, DC: 2010; pp. 98-106. • Presents a methodology to estimate the risk associated with the transportation of hazmat through tunnels and identifies remedial measures to minimize unacceptable risk levels. • A societal risk assessment is used to determine if alternative routes should be considered and then an overall risk assess- ment is conducted on the tunnel route and alternative routes. • If the tunneled route has the lowest risk, risk reduction measures are implemented. Routing Hazardous Materials around the District of Columbia Area Shih-Miao Chin, Ho-Ling Hwang, Bruce E Peterson, Lee D Han, Charles Chin. Journal of Transportation Safety & Secu- rity, 09 Dec. 2009. Vol. 1, Iss. 4; pp. 296-313. types and quantities of non-radioactive hazardous materials were estimated from PHMSA incident data; inspections con- ducted in Boston; permits and applications for transporting hazmat in Boston; hazmat shippers and carriers registered within 75 miles from Boston; and Census commodity flow survey data. Generally, Class 3 materials were used as a proxy for all materials in the analysis. Bulk gasoline pool fires were used to determine the potential impact area. Residential population data from the Census were deter- mined from transportation analysis zone data at the Census tract level and day and night population estimates were derived from other Census data. The authors determined that 30 per- cent of the nighttime population would be home during the day. Employment population were obtained from city officials and adjusted by broad estimates of the percentage of work- ers that worked during the day (83%) versus as night (17%). School, hotel, hospital, nursing home, and visitor populations were determined from various GIS analyses. Ultimately, population density was determined along each potential route. Travel times were determined by city staff that actually drove each of the routes. The analysis first considered through hazmat routing. A comparative analysis using the ratio of the current route to that of each proposed alternative was used. Risk and distance were used and compared to the guidance to determine if the alternate route should be prescribed. Risk was calculated by multiplying the average population density within one-half mile along the route, the accident rate, and distance. Uncertainty analysis was used to determine the thresh- old at which the risk ratio was significant; only alternatives that exceeded this threshold were considered for adoption. Additional sensitivity analyses were also performed, includ- ing varying the percentage of residents that stay home during the day and varying the percentage of workers at night. Some qualitative assessments were also considered and discussed. Finally, the burden to commerce from increased travel times was estimated for different operations. The authors conclude that there is ample justification for monitoring, controlling, or prohibiting through shipments from down- town Boston during the day. They select a specific route as a leading candidate for designation as a through hazmat route. Route risk is population based and the authors state that other factors, such as emergency response capability, location of sensitive environmental features, climate, and burden to commerce cannot “be used to effectively discriminate among the alternative through routes.” A.3.8. Additional Resources HazMat Transportation by Heavy Vehicles and Road Tun- nels: A Simplified Modeling Procedure to Risk Assessment and Mitigation Applied to an Italian Case Study

97 The risk is based on the probability that train meets with an accident and the consequence of the accident, which is a function of probability (given that an accident has happened: probability of derailment: 0.2347, probability of derailed car is hazmat: 0.4087, probability of a derailed hazmat car has a hazmat release: 0.3952, and population exposure due to hazmat). Four hazardous materials are used: ammonia, chlorine, hydrochloric acid, and petroleum. After assessing the prob- abilities, Verma optimized the data for the Norfolk Southern network and found that hazmat cargo is being shipped along neither the least expensive nor the least risk routes – showing that neither objective (cost or risk) is dominating. The article lists three key contributions that it makes to the industry: 1) it is the first tactical planning model for railroad transportation of hazmat where transport risk calculation incorporates the sequence of events leading to hazmat release from multiple sources; 2) it is the only work that throws light on yard and line activities, given the incorporation of risk in the decision making framework; 3) it is the only work that suggests a number of non-dominated solutions to generate risk-cost trade-off frontier that could be used to both plan and manage railroad shipments. A.4.2. Valuation of Road Safety Effects in Cost-Benefit Analysis Wim Wijnen, Paul Wesemann, Arianne de Blaeij, Evaluation and Program Planning, 32. 2009; pp. 326-331. The VoSL plays an important role in cost-benefit analysis for safety measures and in consequence variable in risk assessments. VoSL is measured via the concept of “willingness to pay,” which is the maximum amount people are willing to pay for a given decrease in fatality rate, i.e. if two in 10 people die from an event, the cost paid to lower the rate to one in 10 would represent the “willingness to pay” for one statistical life; therefore, VoSL is the value of a decrease in the fatality rate, not a specific life. The human capital approach is an alternative method to cal- culate VoSL. Here, the VoSL is based on the economic loss of the deceased’s productive capacity. Additionally, VoSL can be calculated through “quality adjusted life years” (QALYs) which expresses the benefit of saving a life in the number of healthy life years gained. This value “is calculated by a reduction of the num- ber of life years lost due to early death and the number of years someone lives with a disability, weighted for their severity;” how- ever, pure QALYs do not address whether or not an investment is socially profitable. In order to do so, a willingness to pay for a QALY lost or gained needs to be established, and by doing so, the costs in a cost-benefit analysis can include fatalities and injuries. The “willingness to pay” can be measured through revealed preference (RP)—value risk reductions based on actual behavior—and state preference (SP) methods—uses surveys in which people are asked how much they are willing to pay • Lays out a methodology for evaluating hazmat shipment routing options on via freight rail networks various situa- tions similar to D.C.’s shipment ban. • Methodology is applied to three alternative routes with population and other vulnerable people within a 0.8 km radius buffer zone values being used to evaluate the poten- tial risk from ultra-hazardous materials. • It is concluded that rerouting results in moderate increases in ton-km and time in transit, but that the overall popula- tion at risk will be lowered; however, the population-at-risk burden is shifted from one location to other locations. A.4. Economic Risk Analysis Shippers and transportation companies strive to create as much profit as possible, while maintaining a safe workplace and con- forming to applicable regulations. Economic risk analysis helps a shipper to maximize profitability while determining the amount of risk they are willing to accept. The following papers explain common techniques when considering economic risk analysis. A.4.1. A Cost and Expected Consequence Approach to Planning and Managing Railroad Transportation of Hazardous Materials Manish Verma, Transportation Research, Part D 14. 2009; pp. 300-308. The author looks at the risk and costs associated with the transportation of hazardous materials via rail using a bi- objective optimization model. Verma assesses transporting risk, the primary concern of regulators, by using hazmat-specific expected risk and uses values based on other researcher’s efforts to calculate cost to operators. Verma makes three assumptions: “first, demand is expressed in terms of the number of railcars to be shipped per week; sec- ond, operation level details such as congestion is beyond our scope and is ignored; third, the hazmat being shipped possess the same chemical property.” The optimization minimizes the risk associated with hazmat transit and costs to operators such that demand for hazmat and non-hazmat railcars is met and that operating constraints of the train service, classification yard, and transfer yards are not exceeded. Costs to operators is based on Ahuja et al.19 research value of $0.50 to move a railcar one mile and $50 per intermediate handling. The hourly fixed cost of operating a train is $500, and the average speed of a train is around 22 mph based on Railroad Performance Measures data from 2008. 19 Ahuja, R.K., Jha, K.C., Liu, J., 2007. Solving real-life railroad blocking problems. Interfaces 37, 404-419.

98 • Uses a random utility model to examine stated preferences for valuation of public risk of fatalities from terror and natural disasters. • Two series of pair-wise risk-risk tradeoff choices are made using traffic-related deaths. • Nationally representative sample used stated preventing terrorism deaths is almost twice as highly valued as pre- venting natural disaster-related deaths. • Might be extended to hazmat. Economic Evaluation of Routing Strategies for Hazardous Road Shipments F. Saccomanno and A. Y.-W. Chan, Transportation Research Record 1020. 1985. pp. 12-18 • This early article discusses how freight companies must constantly balance their costs with the risks their ship- ments cause. The cheapest option is often the most danger- ous and these shippers must find a way to remain profitable without significantly increasing their risk. • Potential routes were analyzed on the basis of cost- effectiveness. • Factors that contribute to a route’s risk include visibility, congestion, and pavement conditions. • The authors found that always choosing a route based on minimum accident likelihood will not be cost-effective. for safety measures. Of the two, RP is seen as being more sci- entifically rigorous since there could be a difference in real and theoretical behaviors seen in SP methods. A Canadian study examined 28 VoSL road safety studies and found an average VoSL more than $3.5m (US $, year 2000). Another VoSL road safety study looked at some European countries and the United States. It found an average VoSL of $4.4m (price level 1997). A third study was looked at where the research used regression modeling to calculate the VoSL in 49 countries based on data from 13 and found that the VoSL in Europe was $2.7m (price level in 1995) and $2.2m for North America. Within the EU, member states’ transport departments use values between 1.4m Eurodollars and 2.6m Eurodollars, with member governments pegging the VoSL at 1.5m Eurodollars. Taking The Netherlands, where the CBA accounts for medi- cal costs, production loss, human costs (injury and fatality), property damage, settlement costs and cost of traffic delays, as a case study, the authors conclude that VoSL measurements should include human losses for severe injuries. This is based on the observation that in the Dutch case, the total human costs of serious injuries is higher than the total human cost of fatalities. A.4.3. Additional Resources Valuing Risks of Death from Terrorism and Natural Disasters W Kip Viscusi, Journal of Risk and Uncertainty, Jun. 2006. Vol. 38, Iss. 3; pp. 191-213.