TR N

EWS

286

MAY

–JUN

E 20

13

12

Barkan is Professor, Department of Civil andEnvironmental Engineering, and ExecutiveDirector, Rail Transportation and EngineeringCenter, University of Illinois at Urbana–Champaign. Saat is Research AssistantProfessor, University of Illinois at Urbana–Champaign. González is Hazardous Materialsand Tank Car Project Manager, Office ofResearch and Development, Federal RailroadAdministration, Washington, D.C. Treichel isDirector, Railway Supply Institute–Associationof American Railroads Railroad Tank Car SafetyResearch and Test Project, Leesburg, Virginia.

R ailroad tank car safety in North America has improved continuously throughcooperative testing, research, and standards development by industry andgovernment. Although much of this progress has been evolutionary, in recent

decades more revolutionary approaches have taken hold. The railroad, tank car, and petrochemical industries have worked together with

the government to develop and improve safety design standards for tank cars sincethe early 20th century (1). In 1903, the Master Car Builders’ Association formed theCommittee on Tank Cars, composed of the mechanical officers from several railroadsand a representative from Union Tank Line, then the major tank car owner. The com-mittee recommended practices that were soon established as industry standards forthe construction and repair of tank cars.

The American Railway Association and its successor, the Association of AmericanRailroads (AAR), later adopted the standards. The AAR Tank Car Committee ischarged with reviewing and revising standards to advance tank car safety.

Cooperative Research in Tank Car Safety DesignHow Science and Engineering Are Reducing the Risk of Rail Transport of Hazardous MaterialsC H R I S T O P H E R P. L . B A R K A N , M . R A P I K S A A T,

F R A N C I S C O G O N Z Á L E Z , I I I , A N D T O D D T. T R E I C H E L

Railroads andResearch Sharing Track

Testing of tank car thermal protection at theTransportation Technology Center near Pueblo,Colorado, in the early 1970s helped quantify the rate ofheat transfer into a tank under intense fire conditions.

00_TRN_286_TRN_286 7/11/13 5:04 PM Page 12

TR NEW

S 286 MAY–JUN

E 2013

13

The public’s interest was represented early on—in 1912, the Interstate Commerce Commission ref-erenced the tank car standards as the basis for federalregulations. The public sector’s oversight role—nowunder the auspices of the U.S. Department of Trans-portation (DOT)—has expanded (2), as private- andpublic-sector stakeholders work toward the com-mon goal of ever-safer transportation of hazardousmaterials.

Improving Tank Car SafetyThe substantial economies offered by the safe, reliablebulk transport of petroleum and chemical products ledto a proliferation of increasingly specialized tank cardesigns to accommodate an extraordinary variety ofhazardous and nonhazardous liquid products. As thetank car has evolved, new materials, designs, and man-ufacturing technologies have contributed to technicalsolutions for a variety of challenges.

Tank cars today are the second most commontype of railroad freight car in North America,accounting for approximately 20 percent of the railcar fleet. Each year, tank cars transport more than 1.6million shipments of hazardous materials for a rangeof products and processes essential to the nation’seconomy, public health, and quality of life.

Nearly all of these shipments arrive safely at theirdestinations. Nevertheless, a train accident involvingtank cars may release a hazardous material with apotential to harm humans, property, and the envi-ronment.

Building on a century of cooperative efforts, gov-ernment and industry continue working together toimprove tank car safety; recent design advances have followed three parallel and complementaryapproaches:

u Statistical analysis and optimization of safetydesign,

u Structural modeling, andu Physical testing.

Quantitative Analysis A series of catastrophic tank car accidents in the late1960s and early 1970s released flammable gases andtoxic materials. Industry and government did notsufficiently understand the factors affecting theseaccidents and the principal failure modes that causedthe releases. Two new cooperative research programswere initiated; one focused on train accident pre-vention and the other on tank car safety improve-ment.

The Railroad Tank Car Safety Research and TestProject started in 1970 under the auspices of theRailway Progress Institute—now the Railway Supply

Institute—and AAR. The project conducted researchand testing with U.S. DOT to identify and evaluatedesign concepts for improving the damage resistanceof tank cars in accidents. This research led to suchnow-common safety features as head shields, shelfcouplers, and thermal protection on tank cars carry-ing materials that pose the highest hazard; these fea-tures protect against the most likely failure modes.

U.S. DOT regulations and AAR standards incor-porating these safety features have reduced tank carreleases in accidents substantially. As the first majordesign elements with the sole purpose of protectingtank cars from damage in accidents, these featureswere revolutionary in their time.

Although effective in tests, the new design ele-ments required proof on cars in service. The RSI-AARSafety Project therefore launched a parallel effort torecord extensive information about tank car perfor-mance in accidents. In 43 years, the effort has col-lected data on more than 40,000 damaged tank carsand 26,000 accidents (3).

Complementing this database is the RailroadAccident–Incident Reporting System, which the Fed-eral Railroad Administration (FRA) revamped andexpanded in 1975 to improve analyses of accidentcauses and trends.

Together, these two databases—one on accidentcauses and characteristics, the other on damage to

An accident at CrescentCity, Illinois, in 1970,released and ignitedliquefied petroleum gas;industry and governmentsoon launched new,cooperative researchprograms to improverailroad and hazardousmaterials transportationsafety.

Tank car built in 1924 byAmerican Car & Foundrywas state of the art forrail transport of chlorine.

Photo: AMERICAN CAR & FOUNDRY COMPANY, EDWARD S. KAMINSKI COLLECTION

00_TRN_286_TRN_286 7/11/13 5:04 PM Page 13

TR N

EWS

286

MAY

–JUN

E 20

13

14

the vehicles involved—provide an inferential capac-ity that is unparalleled in the safety databases for anyother U.S. transportation mode or in any compara-ble rail safety database in the world. The databasesenable detailed quantitative understanding of the fre-quency and severity of tank car accident failuremodes and of the effects of different design features.

Optimizing Safety Design The expansion and refinement of the RSI-AAR data-base has allowed increasingly robust statistical analy-ses of the performance of tank car designs andvariations. For the first time, the relative benefits ofalternative tank car designs could be evaluated with“what if” analyses. The combinations of changesmost likely to maximize safety benefits could bequantitatively assessed, leading to a new approach toimproving tank car safety.

The traditional approach was to overpackage haz-ardous products—that is, to transport them in tankswith higher pressure specifications than necessary.

Now that the performance of each part of the tankcar affecting safety could be quantified, an opti-mization model could be developed, combining thestatistical estimates with data on tank car engineer-ing design and economics, to assess the costs andpotential benefits of candidate designs (4). The com-binations offering the greatest benefit for the leastcost—primarily represented as additional weight—could be identified (Figure 1, page 15).

Most tank car safety design enhancements involvethicker steel, which increases weight. Increasing acar’s weight, however, reduces its carrying capacitybecause of the maximum allowable gross rail load ortotal weight. This in turn may require more ship-ments and more railcars to move the same quantityof goods.

Informing StandardsThe optimization model revealed which combinationof design features offered the greatest safety benefitfor the least amount of incremental weight, helping

PHO

TO: T

RIN

ITYR

AIL (Left:) Modern,

nonjacketed tank carequipped with a half-height head shield.Many cars are builtwith a full-height headshield that is integralwith a steel jacketenveloping the tank,to provide insulationor thermal protection.

(Right:) Double-shelfcouplers are designedto prevent disen -gagement duringderailments, so thatadjacent cars cannotbatter and puncturethe tank car.

PH

OTO: U

NIO

NT

AN

KC

AR

Physical tests of tank carswith head shields (left)and without head shields(right) were conducted inthe 1970s. The headshield is designed toprotect the end—orhead—of the tank carfrom impacts inaccidents.

00_TRN_286_TRN_286 7/11/13 5:04 PM Page 14

TR NEW

S 286 MAY–JUN

E 2013

15

to identify the most efficient approaches to enhanc-ing safety. The AAR Tank Car Committee used thetank car safety design optimization model results todevelop several new standards, including designrequirements for tank cars with higher carryingcapacity (5).

One petition for U.S. DOT rulemaking led to newstandards for toxic-inhalation hazard (TIH) tank cars(Figure 2, below). More than 1,600 new cars havebeen built since, and the risk of transporting TIHproducts in these cars has dropped by an estimated60 to 65 percent. The AAR Tank Car Committee hasused the results from the model to develop anotherpetition for rulemaking for new, improved standardsfor flammable materials.

The optimization technique helps determinewhich combination of features will most efficientlyachieve a given level of safety performance but doesnot answer the question, “how safe is safe enough?”Performance requirements of tank car designs varywidely, depending on the hazards associated withthe material being transported. Industry and gov-ernment have grappled with this question fordecades, as understanding of different hazards hasbecome more sophisticated, shipping patterns havechanged, and societal expectations of tolerable riskhave evolved.

Clarifying Trade-OffsAssigning relative value to harmful impacts can betechnically challenging and sometimes controver-sial. Nevertheless, with improved quantitative rigor,decision making becomes more objective, traceable,and accountable, so that all parties are informedabout the necessary trade-offs.

Analysis of the FRA and RSI-AAR databases yieldsinformation about risk and helps determine anappropriate level of safety to incorporate into tankcar design. Safety design should be commensuratewith the hazard posed by the materials, with morehazardous materials warranting greater protection.

As noted, tank cars can be made safer by increas-ing the damage resistance of various components;however, this generally can have the effect of mak-ing the tank cars less efficient for transportation andmore costly to purchase. University of Illinoisresearchers developed a quantitative framework toassess the cost of losses of different hazardous mate-rials—the incremental benefit of avoided costs wascompared with the cost of more robust tank cars.Higher-hazard materials offered larger benefits foran equivalent level of tank car safety improvement.

New Safety Design Concepts From 1980 to 2012, the rate of hazardous materialsreleases caused by train accidents declined by morethan 90 percent as a result of tank car safety enhance-ments and of a dramatic reduction in accidents, asshown in Figure 3 (page 16). Several accidents in the mid-2000s, however, caused fatal releases of

FIGURE 1 Pareto optimal analysis of various combinations of tank car safetyenhancements. The blue “nondominated” points represent the most efficientfamily of options to improve safety, minimizing the increase in tank car weightand consequent loss in capacity (4).

FIGURE 2 Results ofsimulated rolloveranalyses of (a)conventional top-fittingsprotection comparedwith (b) a new design forpressure tank cars.

IMA

GE: T

RIN

ITYRA

IL

(a) (b)

00_TRN_286_TRN_286 7/11/13 5:22 PM Page 15

TR N

EWS

286

MAY

–JUN

E 20

13

16

hazardous materials and stimulated renewed interestin tank car safety design.

Although further improvements were possible bymaking tank cars thicker and heavier, statistical analy-sis indicated diminishing returns to this approach. Aconsensus emerged that a more effective approachmight be to consider new materials, structural designs,and components that would yield substantial safetybenefits without as much additional weight.

In 2006 Dow Chemical, Union Pacific Railroad,and Union Tank Car Company formed a partner-ship to develop the next-generation rail tank car(NGRTC). The coalition soon expanded to includeseveral other industry and academic partners, as wellas U.S. DOT and Transport Canada, with the goal ofimproving tank car safety more effectively and efficiently. Extensive research explored innovativeconcepts in tank car crashworthiness, including

FIGURE 3 The occurrenceof hazardous materialsreleases caused by rail -road accidents hasdeclined more than 90percent since 1980, withimprovements in tank carsafety design andsubstantial reductions inaccidents (Source: FRA).

Pho

to: T

RIN

ITYRA

IL

TIH tank car thatconforms to new, morerobust standardsrequired by AAR andFRA. These cars areapproximately 60 to 65percent less likely torelease their contents inan accident than carsconforming to theprevious standard.

(Below, left:) Thermalprotection can shield atank and its contentsfrom a buildup of heat-induced pressure in anaccident that triggers anengulfing fire.

(Below, right:) Lower-profile protection for topfittings of TIH tank cars.

PH

OTO: T

RIN

ITYRA

IL

PH

OTO: T

RIN

ITYRA

IL

00_TRN_286_TRN_286 7/11/13 5:05 PM Page 16

TR NEW

S 286 MAY–JUN

E 2013

17

computer modeling of the dynamics of train derail-ment and tank car response (as shown in Figure 4,at right), materials testing, full-scale crash testing,and tank car design optimization modeling (4, 6, 7).

Impact TestsThis work included a series of full-scale impact teststhat examined the puncture resistance of the tank carhead and shell and evaluated the performance of sev-eral designs hit by impactors with different sizes,shapes, and speeds. Accelerometer measurementswere converted into force and displacement historiesto characterize the force-crush response of the tank.The data were compared with results from finite ele-ment analysis models developed to simulate the testsand were found to be in reasonable agreement.

FRA also conducted impact tests on high-strength, low-alloy steels in welded sandwich panelsas a possible means of protecting tanks duringimpacts (8). In addition, FRA is studying the vul-nerability of tank car fittings—such as valves andother appurtenances—in accidents. Full-scalerollover tests have quantified the nature and magni-tude of the forces on the cars and appurtenances (9).The test data can be used to refine and validate mod-els now in development to predict tank car behaviorand performance in accidents (Figure 5, page 18).

Union Tank Car Company or UTLX has con-structed several “Tank Cars of Tomorrow,” incorpo-rating a tank-within-a-tank or sandwich design,along with other new safety features derived from the

Hopper Car Model

Tank Car Model

(a)

(b)

FIGURE 4 Computer simulation models of the dynamics of train derailment weredeveloped to understand the force of impacts on different parts of tank cars inaccidents: (a) 36-car train model and (b) calculated response of train 25 secondsafter derailment (7).

Full-scale impact test setup at the TransportationTechnology Center for evaluating various safetyimprovement concepts in the Next-Generation RailTank Car project (7).

00_TRN_286_TRN_286 7/11/13 5:22 PM Page 17

TR N

EWS

286

MAY

–JUN

E 20

13

18

research and development under the Next-Genera-tion Rail Tank Car Project.

Simulation TestsIn 2009 a larger coalition was formed to continue thework on the NGRTC. The Advanced Tank Car Col-laborative Research Program (ATCCRP) includesAAR, RSI, the American Chemistry Council, the Fer-tilizer Institute, and the Chlorine Institute—repre-

senting private-sector stakeholders—and U.S. DOT,the U.S. Transportation Security Administration, andTransport Canada.

Informed by the extensive safety data and by theresults of the physical testing and modeling research,the ATCCRP partners developed an extensive list ofpotential projects. The first two were (a) to identifythe most appropriate failure criteria in modeling theperformance of tank steels, as well as the materialproperties to support accurate use of those criteria,and (b) to simulate a variety of scenarios for tankhead and shell impacts, to estimate how much energyeach tank design could absorb.

Both projects aimed to improve assessments ofthe relative performance of tank car designs byimproving the accuracy of the models in the finiteelement software, providing greater fidelity in pre-dicting the failure process. For each type of tanksteel, the most appropriate failure criteria—that is,the set of assumptions about how that material’s fail-ure will unfold at the microstructural level—wereidentified for use in larger impact scenarios.

The tank impact simulation project sought torefine the design of physical tests for developing aperformance standard and to understand the forcesacting on a tank in an accident. Many interestingfindings emerged. For example, the size and shape ofthe impacting object in the tests started out as amajor topic of debate, but the simulations made clearthat larger impactors—including those with irregu-lar shapes and angles—were essentially equivalent tosmaller, sharper impactors in this context (Figure 6,at left). The element of a large impactor that makesinitial contact with the tank acts like a small, sharpimpactor, doing much of its damage quickly.

Follow-Up ProjectsSeveral follow-up projects are now under way to

u Derive up-to-date, empirical estimates from theRSI-AAR Safety Project and FRA databases for theprobability that a derailed car will lose some or all ofits contents or lading and through which components;

u Simulate the performance of tank protectionsystems fabricated from composite materials and com-pare the results with those for different types of steel;

u Develop mathematical relationships betweenthe empirical lading-loss probabilities and the esti-mated energy absorption results from tests and simu-lations;

u Develop testing protocols to determine whethera new design meets specified performance criteria; and

u Evaluate new protective design systems, includ-ing additional layers of protective material surround-ing the tank.

FIGURE 6 The effect ofdifferent sizes andshapes of impactors wasmodeled to understandthe relationship betweenfailure mode and thegeometry of objects thatmight strike a tank car inan accident: (a)calculated puncturebehaviors with a 3-by-3-inch and a 12-by-12-inchimpactor; and (b) initialimpact conditions (left)and puncture responseof the head (7).

(a)

(b)

FIGURE 5 A finite element analysis modelsimulating the response of the top fittings and thetank structure during a tank car rollover afterderailment (9).

00_TRN_286_TRN_286 7/11/13 5:05 PM Page 18

TR NEW

S 286 MAY–JUN

E 2013

19

The findings from these research projects may beused to design and build a prototype for a new gener-ation of tank cars for TIH materials, with a much-improved accident performance. The lessons learnedalso can be applied to tank cars transporting otherhazardous materials. As these new design concepts aredeveloped, tested, and perfected for implementation,the optimization techniques can help decide whichcombinations will offer the most effective design fortank car safety.

Impressive AdvancesIn 2012 the accident rate for mainline freight trainsreached an all-time low. Although technical challengesremain, the vision is that when the new tank car designconcepts now under development are implemented,further significant improvements will be possible.

Working together for more than a century, indus-try and government have conducted research anddevelopment that has generated impressive advancesin tank car safety. These advances have served thepublic interest by making the transportation of haz-ardous materials safer.

DedicationThe authors dedicate this article to the memory of thelate William J. Harris, Jr., who played a critical role inestablishing the modern era of cooperative tank carsafety research. Harris was a leader in the formation ofthe Railroad Tank Car Safety Research and Test Proj-ect of the Railway Progress Institute and the Associa-tion of American Railroads. Much of the progress intank car safety improvements in the past four decadescan be attributed to his visionary leadership.

References1. Heller, F. J. Evolution of Tank Car Design Through Engi-

neering, Proceedings of the 1970 Annual ASME PetroleumConference, Denver, Colorado, 1970.

2. Rader, J. H. A Standard of Government Responsibility: Over-seeing the Tank Car Committee. Unpublished report. FederalRailroad Administration, Washington, D.C., July 2000.

3. Treichel, T. T., J. P. Hughes, C. P. L. Barkan, R. D. Sims, E.A. Philips, and M. R. Saat. Safety Performance of Tank Carsin Accidents: Probability of Lading Loss, RSI-AAR RailroadTank Car Safety Research and Test Project, Association ofAmerican Railroads, Washington, D.C., 2006.

4. Saat, M. R., and C. P. L. Barkan. Generalized Railway TankCar Safety Design Optimization for Hazardous MaterialsTransport: Addressing the Trade-Off Between Transporta-tion Efficiency and Safety. Journal of Hazardous Materials,Vol. 189, pp. 62–68, 2011.

5. Barkan, C. P. L. Improving the Design of Higher-CapacityRailway Tank Cars for Hazardous Materials Transport:Optimizing the Trade-Off Between Weight and Safety. Jour-nal of Hazardous Materials Vol. 160, pp. 122–134, 2008.

6. Jeong, D. Y., Y. H. Tang, H. Yu, M. L. Lyons, J. E. Gordon,O. Orringer and A. B. Perlman. Engineering Studies onStructural Integrity of Railroad Tank Cars Under AccidentLoading Conditions. Final Report, DOT/FRA/ORD-09/18.Federal Railroad Administration, Washington, D.C., Octo-ber 2009.

7. Kirkpatrick, S. W. Detailed Puncture Analyses of VariousTank Car Designs. Final Report–Revision 1, AppliedResearch Associates, Inc., Mountain View, California, 2010.

8. Carolan, M. E., D. Y. Jeong, A. B. Perlman, Y. H. Tang, andF. González, III. Deformation Behavior of Welded SteelSandwich Panels Under Quasi-Static Loading. Proceedingsof the ASME/ASCE/IEEE/AREMA/TRB/INFORMS 2011 JointRail Conference, Pueblo, Colorado, March 16–18, 2011.

9. Trent, R., A. Prabhakaran, D. Brabb, and V. Sharma. Sur-vivability of Railroad Tank Car Top Fittings in Rollover Sce-nario Derailments–Phase 2. Final Report, DOT/FRA/ORD-09/20. Federal Railroad Administration, October 2009.

PHO

TO: U

NIO

NTA

NK

CA

R Several “Tank Cars ofTomorrow” wereconstructed in 2012applying findings fromthe Next-Generation RailTank Car project; thetank cars are undergoingfield tests.

00_TRN_286_TRN_286 7/11/13 5:05 PM Page 19