铁路英语论文

Cost-EffectivenessofRailwayInfrastructure

RenewalMaintenance

GeorgeAveryGrimes,Ph.D.,P.E.

1

;,Ph.D.

2

Abstract:railwaysmaintaintheirinfrastructurethroughamixofordinarymaintenanceandperiodicrenewalofinfrastruc-

entrailwaysusedifferentproportionsofordinarymaintenanceandperiodicrenewalwithlittleconsensusastothe

rmore,thecost-effectivenessofemphasizingonemethodovertheotherhasnotbeenanalyzedusingempirical

ectiveofthisresearchistoinvestigatethecost-effectivenessofrenewal-basedmaintenancestrategiesusinghigh-level

fiultsindicatethatmaintenancestrategiesthatplacemoreweightonrenewalresultinlower

unitmaintenancecosts,atleastwithinaspecifiultsimplythatifrailroadsconstrainrenewalmaintenanceto

reduceoverallcapitalexpenditures,increasingmaintenanceexpenseswillmorethanoffsettemporaryreductionsincapitalspending.

DOI:10.1061/͑ASCE͒0733-947X͑2006͒132:8͑601͒

CEDatabasesubjectheadings:Railroadengineering;Maintenancecosts;Infrastructure;Capitalcosts

.

D

o

w

n

l

o

a

d

e

d

f

r

o

m

a

s

c

e

l

i

b

r

a

r

y

.

o

r

g

b

y

L

a

n

z

h

o

u

J

i

a

o

t

o

n

g

U

n

i

v

e

r

s

i

t

y

o

n

0

3

/

2

8

/

1

3

.

C

o

p

y

r

i

g

h

t

A

S

C

E

.

F

o

r

p

e

r

s

o

n

a

l

u

s

e

o

n

l

y

;

a

l

l

r

i

g

h

t

s

r

e

s

e

r

v

e

d

.

Introduction

Sincetherailwayinfrastructureinvestmentboomofthemid-

1980s,allClassIrailroadshavemadesignificantefficiencygains

ininfrastructuremaintenancethataretheresultofimprovements

logicaladvancementsininfrastruc-

turecomponentssuchascleanerandhardersteelhavereduced

edcomponentmanagementhasalso

reducedcosts,forexample,newdevelopmentsinrailgrinding

andlubrication͑IHHA2001͒.Infrastructuremaintenancedelivery

systemsandmaintenanceequipmenttechnologyhavechanged

measurementtoolsandcross-functional

teamworkhastransformedtraditionalengineeringpractices.

Railroadsmaintaintheirinfrastructureusingacombination

rymain-

tenancegenerallyincludesthereplacementofsmallquantities

ofinfrastructurecomponentsusingrelativelysmalltrackgangs

andsmallequipment,whereasrenewalmaintenancetechniques

involvethereplacementoflargerquantitiesofcomponentswith

largergangsandbigger,moresophisticated,andmoreexpensive

rymaintenanceactivitiesarenormallycharged

tooperatingexpenseandrenewalmaintenanceprogramsto

capitalexpendituresaccordingtoSurfaceTransportationBoard

͑STB͒accountingrequirements͑1995͒.

noisatUrbana-Champaign,NewmarkCivilEngineering

Laboratory,MC250,Urbana,IL61801;presently,1572LeydenSt.,

Denver,CO80220.E-mail:avegrimes@

2

AssociateProfessorandDirectorofRailroadEngineeringProgram,

landEnvironmentalEngineering,noisat

Urbana-Champaign,NewmarkCivilEngineeringLaboratory,MC250,

Urbana,IL61801.E-mail:cbarkan@

sionopenuntilJanuary1,tediscussions

ndtheclosingdateby

onemonth,awrittenrequestmustbefiledwiththeASCEManaging

uscriptforthispaperwassubmittedforreviewandpos-

siblepublicationonFebruary5,2004;approvedonNovember8,2005.

ThispaperispartoftheJournalofTransportationEngineering,Vol.

132,No.8,August1,2006.©ASCE,ISSN0733-947X/2006/8-601–608/

$25.00.

1

Overthepast20years,allClassIrailroadshaveincreased

theiruseofrenewal-basedcomparedtoordinarymaintenance,but

thedegreetowhichtheydosovariessubstantially͑Fig.1͒.A

ClassIrailroadisonethatmetarevenuethresholdof$277.7

millionin2004intheUnitedStates͑AAR2005͒.Forthepur-

posesofthisanalysiswecalculatedrenewal-basedmaintenance

costusingaprocedure͑describedinthesectiononmethodology͒

thatseparatesrailroadcapitalexpendituresintocapacity-and

maintenance-relatedcomponents.

Usingthisdefinition,wefoundsubstantialvariationintheway

adsallocatedtheirmaintenance-relatedexpenditures

l

capitalspendingrepresentsthelargestsingleportionofthecapital

budget,withrenewalsaccountingfor67%oftotalcapitalspend-

ingin2002͑Fig.2͒.Thereisalsosubstantialvariationinrenewal

regimesamonginternationalrailroads͑Burns1983͒.

Bothrenewalcapitalexpendituresandordinarymaintenance

expensesrepresentcostsincurredformaintenanceofinfrastruc-

ture,buttheengineeringmanagementstrategyofeachdiffers

l-basedmaintenanceresultsinbetteraver-

lsasproportionoftotalmaintenancecost͑line

indicatesweightedaverageandbarsindicaterangeamongindividual

ClassIrailroads͒edaveragewas

calculatedonbasisoftotaldollarsexpendedbyallClassIrailroads.

JOURNALOFTRANSPORTATIONENGINEERING©ASCE/AUGUST2006/601

J. Transp. Eng. 2006.132:601-608.

tructureCosts

Purpose

Infrastructuremaintenance

Capacityexpansion

a

Excludingdepreciation.

Study

classification

Ordinary

Renewal

Additions

Accountingcategory

Operatingexpense

a

Capitalexpenditures

Capitalexpenditures

D

o

w

n

l

o

a

d

e

d

f

r

o

m

a

s

c

e

l

i

b

r

a

r

y

.

o

r

g

b

y

L

a

n

z

h

o

u

J

i

a

o

t

o

n

g

U

n

i

v

e

r

s

i

t

y

o

n

0

3

/

2

8

/

1

3

.

C

o

p

y

r

i

g

h

t

A

S

C

E

.

F

o

r

p

e

r

s

o

n

a

l

u

s

e

o

n

l

y

;

a

l

l

r

i

g

h

t

s

r

e

s

e

r

v

e

d

.

lcapitalexpendituresaspercentageoftotaland

infrastructurecapitalexpenditures,andpermilliongrosstonmiles

͑2001dollars͒

agetrackconditionoverthelife-cycleofthetrackbutalsogreater

variationintrackquality͑Fig.3͒.Selectiveordinarymaintenance,

ontheotherhand,isgenerallyusedtomaintaintracktoaconsis-

tentminimumstandard͑Burns1980͒.Botharerequired,butan

emphasisononeortheothercanresultinawidevariationofunit

-qualitytrackmightsupportrelativelyhigh

axleloadswithahigh-maintenanceregime;conversely,higher

investmentcanmeanhigheraxleloadsandrelativelylowmain-

tenance͑AustralianGovernment2003͒.Therearealsosubstantial

differencesintheequipmentemployedandthescheduleofwork.

Ingeneral,renewalsinvolvecapitalexpendituresmadetore-

placeand/orimproveinfrastructurecomponentsinresponseto,or

anticipationof,wearandtearcausedbyoutput͑definedhereas

grosstonmiles͒.Bycontrast,capitalexpendituresforexpansion

offacilities͑terminalsandyards,sidingormainlinetrackage,

signalordispatchingsystems,etc.͒aremadetoaccommodaterail

traffir,postfactorail-

roadfinancialstatementsdonotsegregatecapitalexpenditures

purposesofthisstudy,ordinary

maintenanceisclassifiedasmaintenancethatisexpensed,re-

newalmaintenanceasmaintenanceactivitythatiscapitalized,and

additionsascapacityexpansion͑Table1͒.

Thequestionaddressedinthispaperiswhetherarelationship

canbedemonstratedbetweentheengineeringmanagementstrat-

egyandtheoverallcosteffectivenessofthemaintenancefunction

usinghigh-levelfinancialdata.

Background

Trackmaintenancebyrenewalisnotnewbutwasoriginally

developedintheUnitedStatesintheearly1900s,andeventhen

itwasbelievedtobelessexpensive͑Burns1981͒.Renewalwas

originallyperformedbyhandorwithrelativelysimplemachines.

isonoftemporalrelationshipbetweenrenewaland

ordinarymaintenanceandtrackquality

Recentchangesintechnologyandpracticehaveledtoimprove-

mentsinoverallefficiencyforbothordinaryandrenewal-based

maintenancetechniques,buttheefficiencydifferencebetween

smallsectiongangsperformingselectivemaintenance͑character-

isticofordinarymaintenance͒andlargemechanizedgangs

͑characteristicofrenewalmaintenance͒ffer-

enceresults,inpart,fromimprovementsindeliverytechnology

includingtrackrenewalsystems,tie-handlingequipment,surface

andliningequipment,raillayingequipment,andballastdelivery

systems.

Newermaintenance-of-wayequipmentissafer,cleaner,

easiertomaintain,andeasiertooperatethanearliermodels

͑Judge1999͒.Advancesincomputerizationhaveimproved

thereliabilityofthisequipment͑BrennanandKramer1997͒.

Althoughimprovementshavebeenmadeinalltypesofmachin-

ery,thehigh-end,high-productionequipmenthasprovidedmuch

oftherecentproductivityimprovement͑Kramer1997͒.These

advancesandthelargerscaleofequipmentandgangspermit

greatereconomiesofscalecomparedtoordinarymaintenance.

Renewalprogramsalsotendtohaverelativelylongplanning

horizonssothattrackpossessionscanbecoordinatedwithtrans-

ro-

gramsmaytargetvarioustrackcomponentsforreplacement,and

mple,

atieprogrammaytargetreplacementofcrosstieswithoutrenew-

ingtheballastsectionofthetrackstructure,whileatracksurface

andliningprogrammayrenewbothcrosstiesandballast.

Maintenance“blitzes”or“jamborees”areanultimatekindof

maintenanceblitzisusedtorenewinfrastructureinamanner

intendedtominimizetrackdowntime͑Stagl2001͒.InNorth

America,themaintenanceblitzgenerallyresultsintrackclosures

between4and12days͑BurnsandFranke2005͒.Engineering

departmentscoordinatethelargerenewalprojectswithtranspor-

tationandmarketingdepartments͑Foran1997͒.Maintenance

planninghasimprovedthroughadvancementsininformation

technology͑BrennanandKramer1997͒,andrailroadshavetrans-

formedmaterial-handlingsystemsaswellason-siteproduction

͑Kramer1997͒.

Renewalactivitiesnormallyrequiresignificanttrackposses-

sionwindowsthatcanbedifficulttoobtainathightraindensities.

Spotorselectivemaintenanceactivitiesnormallyrequireshorter

trackpossessiontimesandthusarelessdifficulttoobtainevenat

uently,hightraindensitiescanlead

toareducedrelianceonrenewalwork͑Kovalev1988͒,although

thenatureoflargeClassIrailroadstodaymaypermitalternative

onally,renewalmaintenance

ofteninvolveshigh-cost,high-maintenanceequipmentthatneces-

sitateshighutilizationratesthataredifficulttojustifyforsmall

sandotherreasons,routineordinary

maintenancecontinuestobeanimportantactivityinconjunction

withrenewalregimestominimizeunitmaintenancecost͑Grassie

andBaker2000͒.

602/JOURNALOFTRANSPORTATIONENGINEERING©ASCE/AUGUST2006

J. Transp. Eng. 2006.132:601-608.

D

o

w

n

l

o

a

d

e

d

f

r

o

m

a

s

c

e

l

i

b

r

a

r

y

.

o

r

g

b

y

L

a

n

z

h

o

u

J

i

a

o

t

o

n

g

U

n

i

v

e

r

s

i

t

y

o

n

0

3

/

2

8

/

1

3

.

C

o

p

y

r

i

g

h

t

A

S

C

E

.

F

o

r

p

e

r

s

o

n

a

l

u

s

e

o

n

l

y

;

a

l

l

r

i

g

h

t

s

r

e

s

e

r

v

e

d

.

Studiesonrailwaymaintenancecostsdonotprovideinforma-

tionontherelativeefficiencyofemphasizingrenewal-based

eperiod1994to2000,

maintenancecostsinEuropedecreasedwhileexpendituresfor

renewalsincreased,andenhancedrenewalactivitygenerallyre-

sultedinlowerunitmaintenancecost͑UIC2002͒.Anotherstudy

foundthatmaintenanceandrenewalpracticesonTheNether-

lands’railwaysystemhadadirectinfluenceonitsfinancialand

operationalperformanceandthattheappropriatecombination

wascriticaltooveralloperationalperformance͑Swier2004͒.

However,neitheroftheseEuropeanstudiesprovideddatatosup-

portorquantifyitsconclusions.

Thesedevelopmentsleadtothequestion:doesrelianceon

renewal-basedmaintenancestrategyreduceunitmaintenance

cost?Presumablythetrendtowardrenewal-basedmaintenance

reflectsabeliefthatitismoreefficientoreffectiveinsome

r,quantitativeanalysesofdataevaluatingthis

ytical

methodisdevelopedtoevaluatethisissueusingacross-sectional

analysisofClassIrailroadfinancialandoperatingdatareported

totheAssociationofAmericanRailroads͑AAR1978–2002͒

underrulespromulgatedbytheSTB͑1995͒.

1.

Methodology

FinancialandoperatingdataforindividualClassIrailroadswere

modifiedtopermitstudyofthemaintenancecomponentsofthese

adfinancialstatementsdonotsegregatecapitalexpen-

dituresintorenewalsandadditions,andthereforeamethodwas

developedtoestimaterenewalcapitalexpendituressothattotal

maintenancecost,includingbothrenewal͑capitalexpense͒and

ordinarymaintenance͑operatingexpense͒,couldbecombinedto

eofconsolidationsin

theindustryduringthestudyperiod,railroadfinancialandoper-

atingdatawereconsolidatedtoreflectthe2001industrystructure.

Aseriesofstandardlinearregressionanalysesandjointhypoth-

esistestswereconductedtocompareseveralalternativemodels

regardingtheeffectonunitmaintenancecost,includingtheeffect

ofrenewalstrategy,railroadsize,percentageoflight-densitytrack

miles,walstrategyisasignifi-

cantandinfluentialvariableinthebestmodel,thehypothesiscan

beaccepted.

DataPreparation

AARfinancialdataforindividualClassIrailroadsweremodified

topermitstudyofthemaintenancecomponentsofthesedata.A

linearregressionanalysiswasperformedandstandardstatistical

ativehypothesesweretested,includ-

ingtheinfluenceofrailroadsize,averagedensity,andtheper-

centageoflight-densitytrackmiles.

InfrastructureCostIndex

Arailroadinfrastructurecostindexwasdevelopedfromcompo-

nentsoftheAARcostrecoveryindex͑AARRCR͒.Thiswas

termedthemaintenance-of-wayrailroadcostrecoveryindex

͑MOXRCR͒.TheAARRCRisbasedondataprovidedbyall

ClassIrailroads͑AAR1980–2002͒andconsistsof10compo-

nents,whicharethencombinedintofourgroups:͑1͒labor,͑2͒

fuel,͑3͒materialandsupplies,and͑4͒ationofthe

infrastructurecostindexconsideredthesecostgroupsasfollows:

Thelaborcostindex͑labor͒reflectschangesintheaverage

unitpriceofwagesandfringebenefiragewagefor

maintenance-of-wayemployeescomparedtoallrailroad

employeeshasremainedfairlyconstantovertheperiodof

study,andtheoveralllaborindexwasthereforeappropriate

foraninfrastructurecostindex.

lcostindex͑fuel͒wasnotincludedintheMOW

RCRbecausemaintenance-of-wayfuelexpenseisnot

separatelyidentifiedinfinancialreports,andasaresultthe

proportionoffuelcosttooverallcostcouldnotbecalculated.

Additionally,maintenance-of-wayequipmentisoftenfueled

directlyfromlocomotivedieselstoragetanksthatarenot

pensesrepresentarelatively

smallpercentageoftotalmaintenance-of-wayexpenditures,

andthisexclusionshouldnotaffecttheoverallresults.

erialandsuppliescostindex͑M&S͒measurescost

changesinagroupofitemsthatrepresentthepreponderance

dexcomponent

wasincludedintheMOWRCRbecauseM&Scostsarea

significantportionoftotalmaintenance-of-waycosts.

ercostindex͑other͒includesequipmentrents,

depreciation,purchasedservices,taxesotherthanincome

andpayroll,dexcomponent

wasincludedintheMOWRCRbecausethesecostsarea

substantialportionoftotalmaintenancecosts.

TheMOWRCRwasthendevelopedbymultiplyingeach

index͑labor,M&S,andother͒bytherelativeproportionofeach

componentoftotalmaintenance-of-wayexpenseforeachyear.

Thiscalculationisshownbelow

MOWRCR=͓͕R

L

͑M

L

/M

T

͖͒+͕R

M

͑M

M

/M

T

͖͒+͕R

O

͑M

O

/M

T

͖͔͒

whereR

L

ϭAARlaborindex;M

L

ϭClassIRRMOWlabor

expense;M

T

ϭClassIRRtotalMOWexpense;R

M

ϭAARM&S

costindex;M

M

ϭClassIRRMOWmaterialandsupplyexpense;

R

O

ϭAARothercostindex;andM

O

ϭClassIRRMOWother

expense.

Thisannualindexwasthencalibratedwith2001astherefer-

enceyear͑e.g.,2001indexϭ100%,1978indexϭ36.22%͒sothat

allexpensescouldbereferencedintermsofrelativelycurrent

nance-of-waynominalexpensesandinvestments

werethendividedbyeachyear’sindextoobtainconstant2001

dollars.

DefiningMaintenanceCostandRenewalStrategy

Grosstonmilesandtrackmilesarestandardunitsofmeasure-

onnageisthetotalweightofall

locomotives,railcars,andladingthatpassoveraparticularloca-

tion,andagrosstonmileis1grosstonmovingover1mileof

intenancecostwasdefinedastheunitcostofmain-

tainingtrack,thatis,ordinarymaintenanceexpensesplus

renewal-basedcapitalexpenditurespermilliongrosstonmiles

͑MGTM͒producedbyrailroads.

C

M

=͑E

O

+C

R

͒/Q

whereC

M

ϭunitmaintenancecost͑costperMGTM͒;

E

O

ϭordinarymaintenanceoperatingexpense;C

R

ϭrenewal

capitalexpenditures;andQϭmilliongrosstonmiles͑MGTM͒.

JOURNALOFTRANSPORTATIONENGINEERING©ASCE/AUGUST2006/603

J. Transp. Eng. 2006.132:601-608.

isonofRenewalStrategyandUnitMaintenanceCost

Renewalstrategy͑%͒

Road

US

UP

BNSF

CSX

NS

IC

KCS

SOO

GTW

1978–1982

19.3

23.1

20.9

16.3

20.9

19.2

21.5

11.5

15.2

1983–1987

44.5

48.2

44.7

41.5

40.2

46.0

44.7

21.5

20.7

1988–1992

41.9

47.4

34.8

40.8

44.2

58.2

48.9

35.4

24.0

1993–1997

49.6

55.5

54.3

40.6

43.0

74.5

54.2

36.5

26.4

1998–2002

52.8

62.6

62.7

40.2

38.1

69.1

53.5

41.7

50.9

1978–1982

5,803

4,885

4,966

6,349

5,167

7,330

6,659

7,228

7,747

Unitmaintenancecost͑s͒

1983–1987

4,737

4,537

3,982

4,815

5,529

3,520

4,329

4,730

5,115

1988–1992

3,499

3,140

2,908

3,376

5,011

2,118

4,543

3,985

5,159

1993–1997

2,662

2,153

2,565

2,547

3,522

2,341

4,079

4,221

4,217

1998–2002

2,217

1,969

1,875

2,589

3,270

2,053

2,639

3,255

2,698

D

o

w

n

l

o

a

d

e

d

f

r

o

m

a

s

c

e

l

i

b

r

a

r

y

.

o

r

g

b

y

L

a

n

z

h

o

u

J

i

a

o

t

o

n

g

U

n

i

v

e

r

s

i

t

y

o

n

0

3

/

2

8

/

1

3

.

C

o

p

y

r

i

g

h

t

A

S

C

E

.

F

o

r

p

e

r

s

o

n

a

l

u

s

e

o

n

l

y

;

a

l

l

r

i

g

h

t

s

r

e

s

e

r

v

e

d

.

Renewalstrategywasdefinedasthepercentageofunitmain-

tenancecostthatwasallocatedtorenewalcapitalexpenditures.

RS=C

R

/͓͑E

O

+C

R

͒100͔

whereRSϭrenewalstrategy.

EstimatingRenewal-BasedCapitalExpenditures

Becauserailroadcostaccountingsystemsdonotitemizerenewal

capitalexpenditures,weusedamodificationoftheprocedure

developedbyIvaldiandMcCullough͑2001͒toestimatethese

aredtheannualpercentageoftiesandrail

laidinreplacementtracktothetotalamountoftiesandraillaid.

Railroadfinancialreportsdistinguishbetweentiesandrail“laidin

replacementtrack”versus“laidinadditionaltrack”fromAAR

reports͑lines344–372͒͑ARR1978–2002͒.Althoughtheannual

capitalprogramhasotheraspects,thelargestportionofcapitalis

forrailandties͑bothpurchaseandinstallation͒.Anadditional

stepwastakentodifferentiallyweightrailandtiepercentages

because,onaverage,capitalprogramsnormallyallocateaslightly

higherbudgetfortiesthanforrail.

Railroadfinancialdatasegregatecapitalinvestmentforroad

communications,roadsignalsandinterlocker,androadother,

assumedthatcapitalexpendituresforsignalsandcommunications

systemswereprimarilyfornewtechnologyandmajorsystem

upgrades,suchasreplacingextantwire-andrelay-basedsystems

withfiberoptic,wireless,anddigitaltechnologiesandwereap-

propriatelyclassifiedasadditions.

Renewalcapitalexpenditureswerecalculatedasfollows:

P

T

=T

E

/͑T

E

+T

N

͒

whereP

T

ϭpercentagerenewaltieprogram;T

E

ϭnumberofties

laidinexistingtrack;andT

N

ϭnumberoftieslaidinnewtrack.

P

R

=R

E

/͑R

E

+R

N

͒

whereP

R

ϭpercentagerenewalrailprogram;R

E

ϭtonsofraillaid

inexistingtrack;andR

N

ϭtonsofraillaidinnewtrack.

P=͓͑0.6P

T

͒+͑0.4P

R

͔͒

C

R

=C

O

·P

whereC

O

ϭroadcapitalother;andPϭoverallpercentrenewal.

RailroadGroupings

Thenumberofrailroadsreportingfinancialandoperatingdata

͑inR1standardformattotheAAR͒declinedfrom36in1978to

thisreductionoccurredthroughmergersand

combinations,althoughtherewerealsoseveralbankruptciesand

deletionsbychangesinClassIrailroaddefidualrail-

roaddatafrom1978through2002werecombinedintothe2001

r2002forGrandTrunkWesternand

theIllinoisCentralarenotincludedbecausetheseweremerged

withCanadianNationalRailway.

RenewalStrategyAsSingleIndependentVariable

Thestudyperiod͑1978–2002͒wasdividedinto5-yearincre-

mponent͑renewalcapital

expenditures,ordinarymaintenanceoperatingexpense,MGTM͒

el

testedwas

Model1:C

M

=a+bRS+␧

whereC

M

ϭunitmaintenancecost͑dollarsperMGTM͒;

aϭintercept;bϭcoefficientforRS;RSϭrenewalstrategy;and

␧ϭerrorterm.

Renewalstrategyandunit͑infrastructure͒maintenancecost

werecalculatedforeachrailroadovereachtimeperiod͑Table2͒.

DataforallClassIrailroadsintheUnitedStateswereaggregated

andlabeled͑U.S.͒.

Aseriesoflinearregressionswereconductedforeachtime

periodwithrenewalstrategyastheindependentvariableandunit

maintenancecostasthedependentvariable͑Model1͒.Theresults

indicatethattherewasasignificantrelationshiponlyforthelast

timeperiod,withanR

2

of0.78,apvalueof0.003,andF/F

c

of

3.62withF

c

calculatedata95%confidencelevel͑Table3͒.

Onlythelastperiod͑1998–2002͒hasanF-testresultindicat-

ingsignificance;italsohasthelowestpvalueandthestrongest

fluenceofRenewalStrategyonUnitMaintenanceCost

Period

1978–1982

1983–1987

1988–1992

1993–1997

1998–2002

R

2

0.49

0.23

0.28

0.42

0.78

F/F

c

0.97

0.30

0.39

0.72

3.62

p

0.052

0.227

0.178

0.083

0.003

a

10,134

5,637

6,056

5,105

4,706

b

−20,674

−2,777

−5,455

−3,945

−4,130

604/JOURNALOFTRANSPORTATIONENGINEERING©ASCE/AUGUST2006

J. Transp. Eng. 2006.132:601-608.

D

o

w

n

l

o

a

d

e

d

f

r

o

m

a

s

c

e

l

i

b

r

a

r

y

.

o

r

g

b

y

L

a

n

z

h

o

u

J

i

a

o

t

o

n

g

U

n

i

v

e

r

s

i

t

y

o

n

0

3

/

2

8

/

1

3

.

C

o

p

y

r

i

g

h

t

A

S

C

E

.

F

o

r

p

e

r

s

o

n

a

l

u

s

e

o

n

l

y

;

a

l

l

r

i

g

h

t

s

r

e

s

e

r

v

e

d

.

r,theR

2

,Ftests,andpvaluessuggestatrend

towardthisrelationshipthroughthelate1980sand1990s.

Whywasthisrelationshipsignificantonlyinthelastperiod,

andwhatcouldaccountforthisapparenttrend?Althoughtrack

renewalsystemshavebeenemployedbyrailroadsformany

years,anumberofchangescouldexplainwhythisrelationship

wouldbestatisticallysignificantonlyinthemostrecentperiod:

ationshipwouldnothavebeenapparentintheperiod

priortodepreciationaccounting͑1978to1982͒becausea

largeportionofrenewalcostswereaccountedforasordinary

maintenanceoperatingexpenseduetoBettermentAccount-

ingrulesineffectduringthatperiod.

ryandinformationsystemsandplanningtechnology

havecontinuedtoimproveinrecentyears,increasingthe

relativeefficiencyofrenewal-basedmaintenanceinrelation

toordinarymaintenance.

tcostdifferencesbetweenordinaryandrenewal-

basedmaintenancemaynothavebeenstatisticallyapparent

untilreductionsinordinarymaintenancegangsweregradu-

allyrealizedtotheirpresentlevel.

singtraindensitiesmayhaveincreasedtherelative

1978

to1987,averagetraindensityincreasedbylessthan1%per

year;from1988to2001traindensityincreasedbyalmost

6%ionoflight-densitytrackthroughsaleor

abandonmentmayalsohavehadaneffectonthestatistical

relationships.

lroadswereconsolidatingtofewerandlarger

networks.

Plotsofthedatafromthelastthreeperiodsalongwiththeir

trendlinesareshowninFig.4.

AlternativeHypothesis:InfluenceofSize

Previousstudiesthatevaluatedtherelationshipbetweenoverall

tal.

͑1985͒foundslightlyincreasingreturnstoscalewhileBarbera

etal.͑1987͒andLeeandBaumol͑1987͒foundconstantreturns

uatethispossibilitywithrespecttoinfrastructure

maintenancecosts,astatisticaltestwasconductedcomparing

theoriginalmodeltooneincludinganewvariable,trackmiles

͑TM͒.Theresultsindicatethatwhilerailroadsizehadsignificant

effect͑pϭ0.05277͒,ithadfarlessinfluencethanrenewalstrategy

͑pϭ0.00164͒ultsofthejoint

hypothesistest͑pϭ0.02056;F/F

c

ϭ1.1589͒indicatethatthe

interactionbetweenthevariableswaspositive,meaningthat,in

combination,thesevariableswerebetteratpredictingunitmain-

ultssuggestthat

͑1͒a10%increaseintrackmilesfortheaveragerailroad͑equal

toanadditional2,091trackmilesin2001͒wouldresultina

reductionof$20perMGTMtotalmaintenancecost,and͑2͒an

increaseof10%inrenewalstrategywouldresultinareductionof

$398perMGTMtotalmaintenancecost,ora12to21%cost

reduction,rmore,the

resultssuggestthatthetrackmilevariablewassignificantonlyin

combinationwithrenewalstrategy͑atthe95%confidencelevel͒.

,

largerrailroadsmayhavebeenslightlymorecost-effective

intheirmaintenanceprogramsbecausetheycouldemployre-

uldhaveresultedfrom

moreproductiveuseofspecializedequipmentbyoptimizing

componentrenewalcyclesforanygivenpieceoftrack,using

equipmentonayear-roundbasis͑i.e.,workingsouthinwinter

onshipofrenewalstrategyandunitmaintenancecost

͑2001dollars͒:͑a͒1988–1992;͑b͒1993–1997;and͑c͒1998–2002

andnorthinsummer͒,and/orhavingmoreoptionstodetour

traffic,therebypermittinglongertrackpossessionwindows.A

secondexplanationforthiseffectisthataquasi-fixedoverhead

͑engineering͒costmayhavebeenassociatedwithmaintaining

infrastructureregardlessofrailroadsize.

AlternativeHypothesis:InfluenceofLightDensity

TrackMiles

Anotherhypothesisisthatlight-densitylinesareresponsible

forthevariationinunitmaintenancecostsbetweenrailroads.

ClassIrailroadshavereducedthenumberoflow-densityroutes

throughsale,abandonment,orleaseinordertoreducethe

rofstudiesfound

economiesofdensityforrailroads͑Braeutigametal.1984;Caves

etal.1987;Barberaetal.1987;LeeandBaumol1987;Dooley

etal.1991͒,butdifferedastothesignificanceofthedensity

ghthesestudiesconsideredoverallrailroadoperat-

ingandmaintenancecosts,weconsideredwhetheradensityef-

fectmightbeapplicabletoinfrastructurecostsseparateandapart

oryisthateachtrackmilehas

JOURNALOFTRANSPORTATIONENGINEERING©ASCE/AUGUST2006/605

J. Transp. Eng. 2006.132:601-608.

isonofOrdinaryMaintenanceExpenseandRenewalCapitalExpendituresperMillionGrossTonMiles͑MGTM͒

OrdinarymaintenanceexpenseperMGTM

Road1978–19821983–19871988–19921993–1997

1,346

958

1,181

1,515

2,007

606

1,859

2,686

3,111

1998–2002

1,046

738

703

1,548

2,018

631

1,239

1,897

1,322

1978–1982

882

841

823

841

851

1,206

1,163

756

912

RenewalcapitalexpendituresperMGTM

1983–1987

2,105

2,181

1,784

1,996

2,214

1,605

1,940

1,021

1,066

1988–1992

1,466

1,488

1,010

1,379

2,213

1,229

2,208

1,401

1,240

1993–1997

1,316

1,195

1,384

1,033

1,515

1,735

2,220

1,536

1,106

1998–2002

1,171

1,231

1,172

1,041

1,251

1,422

1,400

1,359

1,377

D

o

w

n

l

o

a

d

e

d

f

r

o

m

a

s

c

e

l

i

b

r

a

r

y

.

o

r

g

b

y

L

a

n

z

h

o

u

J

i

a

o

t

o

n

g

U

n

i

v

e

r

s

i

t

y

o

n

0

3

/

2

8

/

1

3

.

C

o

p

y

r

i

g

h

t

A

S

C

E

.

F

o

r

p

e

r

s

o

n

a

l

u

s

e

o

n

l

y

;

a

l

l

r

i

g

h

t

s

r

e

s

e

r

v

e

d

.

US4,9212,6332,033

UP4,0442,3561,653

BNSF4,1432,1991,898

CSX5,5082,8191,997

NS4,3163,3152,798

IC6,1251,915889

KCS5,4962,3892,334

SOO6,4723,7092,584

GTW6,8344,0493,919

Note:MGTMaregivenin2001constantdollars.

aquasi-fixedcostassociatedwithitthatincludesamaintenance-

relatedcomponent,andthoseroadsthatwereabletoshedmoreof

theselow-densitylinesmayhavehadaninherentmaintenance

costadvantage.

Toevaluatethispossibility,astatisticaltestwasconducted

comparingtheoriginalmodeltooneincludinganewvariable,the

percentageoflight-densitytrackmiles͑D

L

͒.Light-densitytrack

wasdefined,forthesepurposes,astrackwithlessthan10million

grosston-milespermileperyearandwasbasedonBureauof

TransportationStatisticsdatafrom2000͑USDOT2001͒.

Resultsfromthejointhypothesistest͑pϭ0.2926;

F/F

c

ϭ0.25444͒indicatethattheinclusionofavariableforlight-

densitytrackmilesdidnotimprovetheoriginalmodel͑Model1͒,

andthisnewmodelwasrejected.

AlternativeHypothesis:InfluenceofAverageDensity

Weconsideredathirdalternativehypothesisthataveragetraffic

densityisresponsibleforthevariationinunitmaintenancecosts

rtothehypothesispresentedinthepre-

vioussection,thishypothesisisrelatedtothetheorythateach

trackmilehasaquasi-fiuatethis

possibility,astatisticaltestwasconductedcomparingtheoriginal

modeltooneincludinganewvariable,averagedensityasmea-

suredinMGTMperClassIrailroadtrackmile͑line343,AAR

reports͒.

Resultsofthejointhypothesistest͑pϭ0.29891;

F/F

c

ϭ0.25015͒indicatethattheaveragedensityvariable͑D

A

͒

didnotimprovetheoriginalmodel,andthisnewmodelwas

rejected.

CombiningStrategy,Size,andDensityVariables

Afinaltestwasconductedcombiningrenewalstrategy,average

density,softhejointhypothesistest͑pϭ0.10961;

F/F

c

ϭ0.46501͒indicatethatthiscombinationofvariablesdid

notimprovetheoriginalmodel͑Model1͒,andthisnewmodel

wasrejected.

Discussion

Theseresultsindicatethatmostofthevariationinunitmainte-

nancecostsamongClassIrailroadscanlargelybeexplainedby

variationinthedegreetowhichtheyemphasizerenewalandde-

emphasizeordinarymaintenanceintheirengineeringstrategies.

Whyisarenewalmaintenancestrategycost-effective?Asprevi-

ouslydescribed,largemechanizedtrackgangsaremoreproduc-

tive,notonlyintermsoflaborandmaterials,butalsowithuseof

orkisbetterplannedand

executedduetoengineeringmanagementsystemsandcanbe

programmedinadvancesothattrafficpatternscanbeadjustedto

providelongtrackpossessionwindowsthatmaximizeresource

productivity.

Italsoappearsthatanemphasisonreducingordinarymainte-

rymaintenanceexpensewas

comparedtorenewalcapitalexpenditures͑perMGTM͒forthe

fourtimeperiodsbetween1982and2002foreachrailroad͑Table

4͒.Somerailroadsmadegreaterreductionsinordinarymainte-

hanaveragedensityandsystem

size,noobviouscharacteristicsappearedtoofferasatisfactory

alternativeexplanationforoverallunitmaintenancecostother

ghtherewassomeappearanceofan

east–westgeographiceffectforthelargeroads,resultsforsmaller

roadswerenotconsistentwiththis,andwearenotawareofany

apriorireasonforsuchaneffect.

Thisanalysisnecessarilymadethesuppositionthatrail

infrastructurequalityforeachroadovereachtimehorizonwas

ederalRailroadAssociation

guidelines,trackconditionscanonlyvarywithinapredetermined

etal.͑2003͒and

AndersonandBarkan͑2004͒foundthatthesafetyrecordof

theserailroadsimprovedoverthistimeperiod,whichwouldbe

increasingrelianceonrenewalmaintenancemayindicatethat

lysisalsomakes

thesuppositionthatrailroadsuserelativelysimilaraccounting

methodsandthatanydifferencesarerelativelyminoranddonot

affecttheoverallresultsoftheanalysis.

Althoughadistinctionwasmadebetweencostsforcapacity

expansionandmaintenance,capacityandunitmaintenancecost

ndensitiesincrease,track

possessionsformaintenancemaybecomelimitedindurationand

frequencybecausetrackgangsmustcompetewithtrainsfortrack

uently,capacitylimitationsincreaseunitcostbe-

causeofthemorefrequentneedforgangstogetonandofftrack.

Capacityexpansionmaythushaveasecondaryeffectofdecreas-

ingunitmaintenancecost.

rtant

considerationforanyrailroadistheeffectthatdifferentmainte-

nancestrategieshaveontransportationcostsandservicequality.

Initialtestswereinconclusiveinregardtotransportationcost,

probablybecauseofmoreinfluentialeffectsoffactorsnotrelated

606/JOURNALOFTRANSPORTATIONENGINEERING©ASCE/AUGUST2006

J. Transp. Eng. 2006.132:601-608.

D

o

w

n

l

o

a

d

e

d

f

r

o

m

a

s

c

e

l

i

b

r

a

r

y

.

o

r

g

b

y

L

a

n

z

h

o

u

J

i

a

o

t

o

n

g

U

n

i

v

e

r

s

i

t

y

o

n

0

3

/

2

8

/

1

3

.

C

o

p

y

r

i

g

h

t

A

S

C

E

.

F

o

r

p

e

r

s

o

n

a

l

u

s

e

o

n

l

y

;

a

l

l

r

i

g

h

t

s

r

e

s

e

r

v

e

d

.

tomaintenance,forexample,reductionofcrewsize,changes

intransportationlaborworkrules,improvementsinmotive

power,andfueleffiequalityfactors,suchascoor-

dinationofmaintenancewindowswithcustomercommitments,

werenottested,andtheserelationshipsaresuggestedforfuture

investigation.

Thisanalysisisonlyvalidfortherangeofdatapresented.

Extendingitbeyondthelimitsofdemonstratedvaluesmaylead

ionedpreviously,a100%

renewalstrategyisneitherattainablenordesirablebasedoncur-

analysisisintendedforusebyrailroadengineeringprofessionals

asonetool͑ofmany͒inthedeterminationoftheappropriate

balancebetweenordinaryandrenewalmaintenanceoptions.

Twofinalquestionsareproposedforfurtherresearchanddis-

,whatarethereallimitsofcostefficienciesgener-

atedbyrenewalstrategies?IfUP,BNSF,andICcanachieve

renewallevelsinthe60%range,wouldafurthershiftfromop-

eratingexpensetorenewalinvestmentresultinevenlowerunit

cost?Second,whatbarriersexistforotherrailroads,suchasCSX,

NS,andSOO,fromgainingtheapparentbenefitsofshiftingmore

ordinarymaintenancetorenewalregimes?Couldthesebarriersbe

technical͑i.e.,infrastructurecharacteristics͒,financial͑i.e.,tight

capitalbudgets͒,philosophical͑i.e.,safety,management͒,opera-

tional͑i.e.,traindensities͒,oracombination?

Conclusions

Theresultsareconsistentwiththehypothesisthatanemphasison

renewalprogramsfortrackmaintenancewascost-effectivefrom

anengineeringviewpointandprovideanexplanationforwhy

railroadshaveconsistentlyincreasedtheiruseofrenewalmainte-

onally,apparent

differencesinunitmaintenancecostscanbelargelyexplainedby

thedegreetowhichindividualfirmsapplyrenewalstrategies.

Thesefindingshaveimportantimplicationsforrailroadfinan-

1998,railroadshavebecomemoreconserva-

tivewithcapitalspendingasinvestorshavebecomeincreasingly

skepticalabouttheindustry’sfinancialcompetitiveness͑Flower

2003a,b;Gallagher2004;Hatch2004͒.Recallingthatrenewal

capitalexpenditurescomprisethelargestshareofoverallcapital

spending,ifrailroadsundulyconstrainrenewalmaintenanceinan

efforttoconservecapitalresources,theywillfindthatordinary

maintenanceexpenseswillrisedisproportionatelyinrelationto

suchtradeoffsmay

improvefreecashflowtemporarily,buttheeffectwillonlybe

shortlivedasoverallmaintenancecosteventuallyincreases.

Acknowledgments

lson,entryandthe

reviewersforhelpfuladviceandinputonthisresearch.

References

AssociationofAmericanRailroads͑AAR͒.͑1978–2002͒.Analysisof

ClassIrailroads,AAR,Washington,D.C.

AssociationofAmericanRailroads͑AAR͒.͑1980–2002͒.Railroadfacts,

AAR,Washington,D.C.

AssociationofAmericanRailroads͑AAR͒.͑2005͒.ClassIrailroad

statistics,AAR,Washington,D.C.,͗/PubCommon/

Documents/AboutTheIndustry/͑͘Jan.15,2005͒.

Anderson,R.,andBarkan,C.P.L.͑2004͒.“Railroadaccidentratesfor

useintransportationriskanalysis.”TransportationResearchRecord.

1863,TransportationResearchBoard,Washington,D.C.,88–98.

AustralianGovernmentBureauofTransportandRegionalEconomics.

͑2003͒.Railinfrastructurepricing:Principleandpractice,͗

/docs/r109/btre_͑͘Jan.6,2004͒.

Barbera,A.,Grimm,C.M.,Philips,K.A.,andSelzer,L.J.͑1987͒.

“Railroadcoststructures—Revisited.”,28͑1͒,

237–244.

Barkan,C.P.L.,Tyler,C.T.,andAnderson,R.͑2003͒.“Railroadderail-

mentfactorsaffectinghazardousmaterialstransportationrisks.”

TransportationResearchRecord.1825,TransportationResearch

Board,Washington,D.C.,64–74.

Braeutigam,R.R.,Daughety,A.F.,andTurnquist,M.A.͑1984͒.“Afirm

specificanalysisofeconomiesofdensityintherailroadindustry.”

.,33͑1͒,3–20.

Brennan,C.,andKramer,J.͑1997͒.“Hightechnology:M/Wgainsfrom

softwareadvances.”truct.,93͑12͒,24.

Burns,D.͑1980͒.“Thecasefortotalrenewal.”ads,35͑9͒,

38–42.

Burns,D.͑1981͒.“Gettingthebestoutoftrackrenewalmachines.”

.,137͑3͒,183–185,189,190.

Burns,D.͑1983͒.“RT&Sforeignscene:Trackmaintenance.”Railw.

TrackStruct.,79͑3͒,52–54,56,58.

Burns,D.,andFranke,M.͑2005͒.“Analyzingtheblitzapproachto

M/W.”truct.,101͑10͒,33–38.

Caves,D.W.,Christensen,L.R.,Tretheway,M.W.,andWindle,R.J.

͑1985͒.Networkeffectsandthemeasurementreturnstoscaleand

density:Analyticalstudiesintransporteconomics,Cambridge

UniversityPress,Cambridge,Mass.

Dooley,F.J.,Wilson,W.W.,Benson,D.E.,andTolliver,D.D.͑1991͒.

“PoststaggersproductivityforClassIrailroads.”.91-6,

MountainPlainsConsortium,Fargo,S.D.͗in-

/pubs/͑͘Aug.1,2003͒.

Flower,S.͑2003a͒.“Containervolumeisgrowing,butrailroads’inter-

modalprofitmarginsaremarginalatbest.”ProgressiveRailroading,

46͑5͒,10.

Flower,S.͑2003b͒

.“UPfocusesonmaintainingoperationaldiscipline

tolurelong-termbusiness,avoidunsustainableshorttermgrowth.”

ProgressiveRailroading,46͑6͒,10.

Foran,T.͑1997͒.“Re-engineeringtheengineeringdepartment.”Progres-

siveRailroading,40͑3͒,47,49,52–55,58,59.

Gallagher,J.͑2004͒.“Railingoverrates:Railroadpriceincreasesanger

shippersasrecordfreightlevelstesttrack,cars,terminalcapacity.”

TrafficWorld,268͑22͒,36.

Grassie,S.,andBaker,P.͑2000͒.“Routinemaintenanceextendsraillife

andofferslong-termsavings.”.,156͑2͒,88–89.

Hatch,T.͑2004͒.“Thecapitalquestion—Again.”ProgressiveRailroad-

ing,100͑5͒,10.

InternationalHeavyHaulAssociation͑IHHA͒.͑2001͒.Guidelinestobest

practicesforheavyhaulrailwayoperations:Wheelandrailinterface

issues,,hn,en,y,and

ov,eds.,IHHA,VirginiaBeach,Va.

InternationalUnionofRailways͑UIC͒.͑2002͒.InfraCost:Thecostof

railwayinfrastructure,UIC,Bern,Switzerland.

Ivaldi,M.,andMcCullough,G.͑2001͒.“Densityandintegrationeffects

trailroads.”.,19͑2͒,161–82.

Judge,T.͑1999͒.“Standardizingcomingtoheavyequipment.”Railw.

TrackStruct.,95͑

11͒17–19.

Kovalev,V.I.͑1988͒.“Managingmaintenanceontheworld’slongestrail

network.”.,144͑3͒,159–61.

Kramer,J.͑1997͒.“High-productionmaintenance:Lookingforgains.”

truct.,93͑12͒,21–22.

JOURNALOFTRANSPORTATIONENGINEERING©ASCE/AUGUST2006/607

J. Transp. Eng. 2006.132:601-608.

Lee,T.,andBaumol,C.P.͑1987͒.“ad

industryunderderegulation.”,28͑1͒,245–53.

Stagl,J.͑2001͒.“Details:ClassIs.”ProgressiveRailroading,44͑3͒,40,

42,44,46,47.

Swier,J.͑2004͒.“Exposingtherealinfrastructurecostdrivers.”Railw.

.,160͑1͒,34–36.

CommitteeonCommerce,Science,andTransportation.

͑1995͒.ICCterminationactof1995,104thCongress,1stSession,

GPO,Washington,D.C.

sportation͑USDOT͒.͑2001͒.Nationaltransportation

atlasdatabase,NationalRailwayNetwork,BureauofTransportation

Statistics,Washington,D.C.

D

o

w

n

l

o

a

d

e

d

f

r

o

m

a

s

c

e

l

i

b

r

a

r

y

.

o

r

g

b

y

L

a

n

z

h

o

u

J

i

a

o

t

o

n

g

U

n

i

v

e

r

s

i

t

y

o

n

0

3

/

2

8

/

1

3

.

C

o

p

y

r

i

g

h

t

A

S

C

E

.

F

o

r

p

e

r

s

o

n

a

l

u

s

e

o

n

l

y

;

a

l

l

r

i

g

h

t

s

r

e

s

e

r

v

e

d

.

608/JOURNALOFTRANSPORTATIONENGINEERING©ASCE/AUGUST2006

J. Transp. Eng. 2006.132:601-608.