10r

Thermal Effects in Concrete Bridge Superstructures

ROY A. IMBSEN and DAVID E. VANDERSHAF

ABSTRACT

The findings of a current research projectentitled Thermal Effects in concrete BridqeSuperstructures are highlighted. The re-search project is sponsored by NCHRP. A

brief discussion of the mechanis¡ns of heattransfer as they relate to bridge structuresis presented, including the ther¡nal coeffi-cients of various types of different con-crete conposed of different aggregates.AIso inclualecl are brief discussions on neaneffective bridge temperatures' temperaturedífferentiats, and the response analysis fornonlinear temperature gra¿lients. Two casestudies of longitudinal thernal responseselected from those conducted in the re-search project are íncLuded ín this paper.the thermal gradients used in Nev¡ zealand,England, and Ontario, Canada, plus thosereconmended' by the Post-Tensioning Insti-tute, are included in the response analy-sis. In addition, specific recon¡nendationsfor inproving the U.S. design provisions forthermal effects are included.

Traditionalty, bridges have been designed to resistonly Èhe overall longituclinal movernent arísing fromtemperature strain. However ' with the recentchanges in bridge types, it has become apparent thattenperature differentials also exist in bridge su-perstructures. These tetnperature dífferentialscause stresses that should be inclucled ín the designprocedures. Although the current AASHTo specifica-tions include probabLe temperature ranges of meantemperature conditions that affect expansion andcontraction of concrete bridge superstructures'there is no recom¡nendation for tenperature differen-tials thât rnay occur in individual superstructuresections.

The purpose of this paper is to highlight some ofthe findings in a current research project' ThernalEffects in Concretê Bridge supêrstructures (f) 'which reviews and evaluates the various proceduresthat have been proposed for use in considering thêr-maL effects on bridge superstructures. The ultímategoal of the research is to upgrade the currentAÀsHTo code for thermal effects.

There are only a few publishe¿l accounts of con-crete bridges danaged by diffêrential temperatureeffects. In l98I zichner (!) described the funda-nentals for determining temperature effects ín con-crete bridges and indicated that cracks such asthose shown in Figure I were observed duringthorough inspections of several different bridges.The cracks, located in the botton sl"âbs and girdersterns of these box-gircler bridges, resulted in partfron temperature differences that exísÈed erithin theindividual brirlge superstructures.

Recently in Colorado, cracking was experienced ínthe webs and bottom deck soffíts of four cantilever,segmental, prestressed bridges. T\do of the bridgesare approximately 747 ft long, the third is about516 ft long, and the fourth ís about 449 ft long.The three longer brÍdges have four spans, whereasthe shorter one has three. The three-span bridget

shown schematically in Figure 2, exhibited thègreatest anount of cracking (3). The crack patternson the single-cel1 bottom deck soffit and webs arealso shown in rigure 2. The largest crack wídthreporte¿l ís approximately 0.13 in.

HEAT TRANSFER BY RADIATION

Heat transfer by radiation is gênerally consideredto be the most inportant contribution of heat energyexchânge on concrete bridge superstructures. Duringthe daylight hours when the structure is exposed tothe sun, especially during the war¡û summer months, a

net gain of heat energy occurs throughout the depthof the structure' prirnarily as a result of solarradiation impinging on the surfâces of the struc-ture. Conversely, prinarily as ã result of reradia-tion to the surrounding environrnent of the heat en-ergy stored in the structure' a net loss of heatenergy occurs during the nÍght. During the sunnerthe ternperature in the top surface of the bridgedeck is warmer than the soffit' which results in apositive gradient. A negaÈive gradient develops ontypical winter níghts when the top surface is coolerthan the soffit.

THERMÀL PROPERTIES OF CONCRETE

The thermal coefficient of expansion for concrete isgreatly depen¿lent on its aggregate type and nix pro-portions (4-l). The cenent paste of normal concreteusually has a higher thernal coefficient of expan-sion than the aggregate in the nixi howeverr becausethe aggregate occupies about 75 percent of the vol-ume, it is the thertnal expansion characteristics ofthe aggregate that díctate the anticipated volunet-ric change during a given tempèrature change.

l4ost co¿les specify an average thermal coefficientof 0.000011 to 0.000012 per degree Celsius (about0.000006 per degree Fahrenheit) for reinforce¿l con-crete. The actual coefficients frotn laboratorytests on concrete sa¡nplès (Table 1) vãry by äs muchas 22 percent äbove and 64 percent below the highervalue, depending on the aggregate type.

!{EAN TEI4PERATURES

Mean or effective bridgê tempêratures are associ,atedwith the long-term (seasonal) movements of a

bridge. Bridge codes typically provide detailedguidelines for the calculation of overall l-ongitudi-nal movements by specifying a range of temperaturesthat depend on the geographical location of thebridge and the structure type. The specified rangeof effêctive temperatures rèpresents the averagerange to be considered in design. At tines, a givenrange of effective teÍìperatures tnay have to bè ad-justed to compensate for unusual conditÍons' such asfrost pockets or shel-tered, low-lyíng areas.

EÍìerson (!) defines the effective temperature ofa bridge as that tenperature that governs the longi-tudinal movement of the bridge deck. The effectivetemperature may be derived by performing a calcula-tion that includes both the product of the areas be-tween isotherns and their rnêan temperatures divídedby the total area of cross section of the deck.Emerson (9) an¿l Black et aI. (!q) have correlatedthe extreme values of effecÈive bridge temperatures

ro2

FIGURE I Gacks in a multispan box-girder bridge.

with shåde temperatures. Ernerson correlated shadetemperatures with the tenperatures obtained fro¡nstructures instrumented with thermocouples, srhereasBlack et a1. correlated shade ternperatures withbridge rnovements to obtain the extrene values of ef-fective bridge temperatures.

TEMPERATURE DIFFERENTIALS

Because of the unsatisfactory thermal conductivityof concrete, the diurnal temperature effects on aconcrete bridge superstructure usually produce ten-perature gradíènts. Large, positive tenperaturegrådients occur during days with high solar radia-tion, clear skies, a large range of atnbient tem-peråtures, and a Iight wind. Negative temperaturegradients develop during periods associated withnight and winÈer conditions. The ternperature gradi-ents that for¡n in a given structure are governed byheat flow through the bocìy and are a funcÈion of thedensity, specific heat, and thermal conductivity ofthe concrete.

Transportation Research Record 950

One-dimensional heat flow in the vertical direc-tion is generally consiclered to be sufficientlyaccurate to conduct most analyses of briclge super-structures. Researchers have conducted bothanalytical and experinental studies to deterrnínetemperature differentials Èhat occur in bridge su-perstructures. The research efforts Èhat led to thedevelopment of the codes in New Zealand, England,and Ontario, Canada, are briefly discussed.

New Zealand

Priestley (+r12) analyzed the effècts of severalassumed thermal gradíents and conpared the resultswith measured data available at the time. One ofthe assumed thermaL graclients consisted of. a lineartenperature distribution through the top deck slab,as proposed by l4aher (ll), and which was supportedby measured tenperãtures fron three bridges locatedin the British Isles (L4,15). Other assumed thernalgradients included the temperature distribution pro-posed by the !4inistry of works of New zealand, anddistríbutions in which temperatures vary vrith depth

i

l.l

I

I

-(,:.r I

(.lro, k.u itlth (nnt)

4 55'- 3"

Irnbsen and Vandershaf

L P|ER 3

FIGURE 2 Miller Creek Bridge in Colorado.

TABLE I Thermal Coefficient of Concrete (-0.000001 per

degree Celsius)

103

i PIEP 2

early from a maxi¡nun at the top surface of the deckstab to a mininum at a ¿lepth of 1200 tm. The non-1ínear variation is represented by a fifth-degreeparabola. The second Part of the revised distribu-tion appli"es only to a ileck slab over an encloseclcell of a box girder, in which case tenperatures areassuneal to decrease 1inear1y. The third and finaLpart of the revised distribution assutnes a linearvariation of tenperatures over the bottorn 200 ¡nm (8in.) of the cross section.

Priesttey also found that the naximum temperatureat the top of the concrete deck slab woulil decreaselinearly wíth the thickness of bituninous overlaybecause of the insulating properties of thís ¡na-ter ial.

creât Britain

In 1973 E¡nerson (!rlZ) describecl a method for calcu-lating the one-dinensional heat flow within a con-crete-slab briclge by using an iterative' finite-ilif-ference solution schene. The method relates thebridge temperature to solar radiation, ambient airtetnperature, and wind speed. the rnoclel of thestructure in this case is conposeil'of several lay-ers, and a starting tíme is assuned, at which point

i anrooE

I

AggregateType

PCA(q) Emerson

Browne

ç) Ontario

QuartziteQuartzSandstoneG¡avelGraniteDoleriteBasaltMa¡bleLimestone

11.91.t.710.8

9.5

12.7

tt.713.29.69.6

7.3

rz.8

tr.7l 3.l9.59.5

1.4

8.6

6.8

1t .7- t4.69 .0- t3.29 .2- t3.39 .0- 13.78. 1- 10.3

7 .9- \0.44.4- 7.44.3- r 0.3

as second-degree, fourth-degree, and sixth-¿legreeparabolas. The sixth-¿legree parabola was found tobe in satisfactory âgreement with neasured ¿latã' andits use was recommencled for superstructure dePthsbetween 1200 anal 1500 rm (47 and 59 in.).

In 1976 Priestley (15'16) proposed a revised te¡n-perature distribution that consisted of three indi-vidual parts. as shown in Figure 3. In the firstpart' ternperatures are assumed to decrease nonlin-

_¡J_O_8rH ELEVATTON:

l'

104

DE

E

oEão0

o

f/--L\5'\t2oo/

5-.O5d"C

32-O-2d"C

Etockrop thickness ( mm )

Soffit

FIGURE 3 Temperature difference from the New Zealand designbrief.

the equations governing the boundary conditions areapplied.

The assunption of boundary conclitions requires anestirnation of the tines at v¡hich the nonlinear dif-ferential distribution is at a minimum. ft is fur-ther assuned that the tenperature throughout thestructure is a constant at this tine. Enerson (g)estinated that the beginning tirne for concretebrídges rdas 0800 hours for the heating phase and1600 hours for the coolíng phase. By using Èheseinput paraneters, a nonlinear differential tenpera-ture distribution was computed at l5-nin intervalsuntil a naxi¡num gradient was reached at approxi-rnately 1500 hours. Tenperatures predícted by thenodel correlated well with measured proÈotype sutunerand winter temperature distributions. The currentBritish Standard BS 5400 (18) has been updated toinclude the tenperature distributions deternined ínthese research efforts, ag shown in Figure 4.

Ontario

T:.ansportation Research Record 950

used by Emerson (l). Àlthough acknowledging thatthe assumption of one-dinensional heat flow is nottechnically correct, they cited conparisons that in-dicated that satisfactory correlations exist betweenobserved and predicted tenperature gradients ob-tained fro¡n a one-dimensional heaÈ-flow analysis.They were able to use Èhis approach to develop sin-plifíed formulas for use in desígn.

Conparisons betr/reen the British standar¿ls (lg),Irtaher (13), the New Zealand Uinistry of Works (20) ,and Priestleyrs sixth-degree parabola to an l-gírderindicate that the resutting stresses are strongtydependent on the temperature differences and tenper-ature gradients. Comparisons betlreen the gradientsproposed by Leonhardt et al. (21), priestley, trlaher,and the one-dimensional heat flow were presented forvarying superstructure depths. The results were de-composeil into continuity and setf-equilibratingstresses. Radolli and creen proposèd that simpleflesign for¡nulas be used for clesigns that do not re-quire an understanding of the tenperature grailient.

RESPONSE ANALYSIS

Having selected a gjven tenperature gradient orloading, the bridge designer is next faced wíth per-forming the response analysis. There are severalways to accomplish this; the two most useful methodsfollow.

General- llethod

The general nethod procedure, shown in Figurê 5a,involves separating the arbitrary gradient intothree components: axial, bending, and residual. ftis first necessary to solve these three problens,and then to superimpose the resultíng three stressdistributions, as shown ín Figure 5b.

Equivålent Prestress ltethod

The equivalent prestress ¡nethod procedure, which ísmuch easier to apply than the preceding oDêr isbased on an analogy betereen thermal strains and pre-stress strains. In this metho¿l the stresses from aconpletely restrained structure are superimposedwith the stresses that result from the removal ofthe restraints, as shown in Figure 5c. Rernoval ofthe restraint is accomplished by dividing the re-strained stress field into a seríes of equivalentnegative prestress forces, as shown in Figure 5d.

ly

tv

T

d

In 1975 Radolli and Green Ggl developeddimensional heât-flow analysis sinilar to

Crnc,c¡c rl¡b o. conc,e¡e deckort coôcrct€ b?¿mr or box g¡rdcrr

- S0mm rurlacing

a one-the one

lffi,.-ff h¡=0.6h '0.2r¡'hr = 0.4(h .hrl ' 0.3n'h:.0.4(h-h¡l >0. lmt! = h-(h,.hrr lq) ,0.?rn

tì' - 0.4hhr. O fh

> C.ln'

'0.3mh1.0.4(h-hr l>0.1mlì.= h-{hr.hrh¡l'0-2m

fr Tr T:qCcnr,9,øcT.

o-?0..o.60.8

>09

9.5l].5r 5.or6.Or6.o

o.5o.r5r.or. r5?.o

0.?o.5l.or.5| .t¿

o5

..05.05.0

TT T¡ T:oCcñ t,g. ¡d.

T.

0.70..0.6o8

> o.9

1.56.58.5

I O.OI O.5

0.5 I O.?51 r.5o.rsl o-s I l.sr.o I r.s I s-or.s lz.s le.sr.o l:.s l.ro

FIGURE 4 Temperature difference from the British Code of practice, BS 5400.

t_0sImbsen and Vandershaf

:EËE

Axial

Môst of the case studies presented in the follo'wingsections were analize¿l by this method.

ALTERNATIVE APPROACHES

The bridge design coiles of õlifferent countries pro-vi¿le the designer with nany ilifferent approaches toteip ...ount lor thernal effects' The codes basí-cafiy aiffer in the refinemenÈ used to deter¡nine therãi"åt"r"gi.a1 conditions at proposed bridge sites'ft. typ"" of ther¡nal loadings consiilered' and thenethods used to accommodate these ther¡nat loadings'

The cocles in most countries are si¡nil-ar ín thatthey contain provisions for some sort of thermalgraåient that producês a corresponding stress ¿lis-

tribution; this corresponding stress distríbution isthen grouped with other stresses (e'g', dead load'live load, prestress) so as to rnodify the ¿lesign'

This usually results in an increase in the prestressforce, sometirnes by a relatively large amount'

However, researchers in Switzerland and GerÍìany

are currently ¿leveloping a new approach based on theàoncept thaC partíaI prestressing is sufficient toacconmoclate therrnal gradient stresses and strains'This design procedure leaves the prestress force un-ãhanged, ãnd relies insteaël on nilil steel reinforce-mènt to resist the thernal stresses' Thist at leastin part, explains why bhe large nurnber of existinqprestressed concrete brídges that were designed and

tuilt without consicleration given to thermal gradi-ent effects have continue¿l to Perforn satisfactorilyfor so rnany Years of service.

\=\l-r-\+

t\t\l_

Bendi ng

VAHEI

s iduaArbitrary Gradlent

(sa)

{<Í

=\FH+\+fE - \'lCÞ õ

ç.:t7' :VFl+\t\LI I tototu

restraine-d fcs'traint

(a) separate themal gradient' geoeral nethod

(b) Superinposed stre6sea' general Eethod

(c) Súperioposed 6tresses' equivalent Preatreas nethod

(d) Equivalent prestre66' equlvaleot prestress Eethod

(5d)

FIGURE 5 Response analYses.

(sb)

.v'

êL __.1

\-.]\L\6

(5c)

ç---+7rIt

The mini¡nu¡n design requirenents for briclges inthe United States are governeat by the AASHTO design

specifications. rn certain cases additional' more

aetaifea design criteria rnay be used' With respectto temperature, the tlesign recommendations of the

Post-Te;sioning Institute (PTI) are often used inthe design of prestressed-concrete bridges' severalindividual states have al-so developecl their own dè-

=iqn proceilures for considering thernal effects'which in most cases arè similar to those reco¡nmended

by PTI. A iletailed description of the procedures'cãnpitecr from a survey of the states' is containedin the NCHRP report (1).

CASE STUDIES

Longitudinal and transverse temperature effects were

applied to a selected group of U.S' bridges as partof the NCRHP research prÕjêct. Four thermat gradi-ents were used to study the longituclinal effects¡whereas two ther¡nal gradients were used to study thetransverse effects.

Four thermal gradients (Figure 6) werè selectedfor the case studies on longitudinal temperature ef-fects. These gradients were selected because Èhey

are representative of those currently being used'they include those specified in the New zeal-and'

British' and Ontario co¿les. In ad¿lition' the gradi-ent reconrnended by PTI was included because it issomevrhat representative of the gradients currentlyusecl in the uníted States. A surnmary of the briclgesincluded in the NCHRP project on case studies for

I06

57.6'F

Transportation Research Record 950

r8'F7. F

FIGURE 6 Thermal gradients used for case studies.

TABLE 2 Summary of tsridges lncluded in Case Studies for Longitudinal Temperature Bffects

?7

3

BNITISH?.7' F

NEW Z EALANO ONTÂRIO USA. PTI

CaseNo. Name and Location Superstructure Substructure Length

Depth/SpanDepth Ratio

No. of No. ofSpans Hinges

No. ofFrames Comment

IL

2L

3L

4L

5L

6L

Colorado River Bridge, Precast, prestressedCalifornia and Arizona l-girder

West Silver Eagle Road Cast-in-place pre-overhead, California stressed box girder

Turkey Run Creek Precast, prestressedBridge, Indiana segmental box girder

Kishwaukee Rive¡ Precast, prestressedBr.idge, Illinois segmental box girder

Columbia River Bridge, Cast-in-place pre-Washington stressed segmental

box gûderWesf Silve¡ Eagle Road Cast-in-place pre-ove¡head (falsework), stressed box ghderCalifo¡lia

East Connector, Cast-in-place rein-Califo¡nia forced box girder

Miller Creek F-l l-AK, Precast, prestressedColorado segments and box

girder

Pie¡ wall

Doublecolumn

Singlecolumn

Singlecolumn

Singlecolumn

Doublecolumn

Singlecolumn

Singlecolumn

5ó5 ft, 6 in.

750 ft, 0 in.

317ft,0in.

1,096 ft, 0 in.

1,870 ft, 0 in.

750 ft, 0 in.

I,104 ft,0 in.

445 ft,3 in.

6 ft, 3 in. 0.056

5 ft, 0 in. 0.037

9 ft, 0 in. 0.057

I I ft,8 in. 0.047

9 ft,0 in.a 0.05324 ft,O in9

5 ft, 0 in. 0.037

6 ft, 0 in. 0.055

8 ft,0 in. 0.041

5060205050

RepresentativeI-girder

Representativebox girder

Segmentalcantilever

Segmentalcantilever

Segmental andhaunchedcantilever

Falsewo¡kloadirg

Multiframe

Case history

1L

8L

11 2

30

uMini^um bMaximum

longitudinal effects is given Ín TâbIe 2. The re-sults of two of Èhese case studies (case numbers 5Land 8t) are included in this paper. The âpplicableportions of the codes are those that pertain to thepositive gradients that occur during the day whenthere is high solar radiation.

PloÈs of top and botton fiber stress versus thedistance longitudinally along the bridge are pre-sented in this paper for both of the case studies.In addition, section stresses are included at se-lected points atong the bridge to show the stressvariations at different depths.

The plots show thãt naximurn fiber stresses usu-ally occur at the pier or column supports adjacentto the abutments. The plots also show that changesin superstructurê cross sections caused by flares inthe bottom slab and girder stens or in the haunchedsuperstructure cause significant changes in fiber3 tresses.

The bridges analyzed were assumed to have a coef-ficient of thernal expansion of 0.000006 per degreeFahrenheit, uncracked section properties, and no re-ductions in thernal gradients for surfacing, eleva-tion, and so forth.

Case 5L: Columbia River Bridge

The Colu¡nbia River Bridge was included in the casestudies because it is a najor structure and becauseit uses a customized and optimized cross section.Another reason for selecting this bridge is to eval-uate the effect of the pronounced variation in thedepth of its structure caused by the haunches at theinterior supports. This bridge was designect accord-ing to the ¡{ashington State criteria on thermal ef-fects. For longitudÍnal thermal effects of box

girders, the criteria specífy a temperature increaseof. 20oF in the top slab. The stresses caused bythis increase in tenperature are combined with deaãload for a service loading. In additíon, anotherservice load condition, which results from one-hâlfthe temperature gradient (i.e., I0oF), is combinedwith the dead load and full live 1oad.

This structure, sho$rn schenatically in Figure 7,is a five-span, 1,870-ft-1ong bridge.

Figures 8 and 9 are plots of top and botton fiberstresses, respectively, along the bridge center_line. The Ontario code was not considered in thiscase because ít ivâs not apparent how this codeshould be applied to nonprismatic sections.

Figures 10 and 1I show plots of the variatÍon instresses, \¿ith section depth for the three gradientsconsidered for 9- and 24-ft-deep sections, respec_tively.

Case 8L: MiIIer Creek Bridge

The Miller Creek Bridge and several other bridges inthe same area developed severe cracking problernsshortly after conpletion of construction. Becausethermal gradient effects are suspected as being oneof the causes of this distress, this bridge rr,as in_cluded in the casè studies.

Thís structure, shown schemâtically in Figure 12,is a three-span, 455-ft-long bridge.

Figurès 13 and 14 are plots of top anil bottomfiber stresses, respectively, along the bri¿lge cen_terline.

The I{iller Creek Bridge is of special interestbecause the structure developed retatively severecracking in the bottom flange and girder sterns atapproxinately the one-guarter ånd threè-guarter

i

I

l

3

@@@

ELEVATION

]5'FROM 49U1, SPAI{S I A 5t sPA¡¡s 2'4

HAUNCH

I

2l4361I9roilt2r3t4r5

9.OO9.¡99.60Io.t8to.90tt.7712.7 0rt,0 7I 5.lor6.44r ?.89t9.442r.ro??.0324.OO

6.OO6.r36. 4l6.797.r87,5 0?.ôl6.22c.659.t 39.69r0.29ro.9 3I r.66il.70

r2l.¡2r2r.90l21.2 3r25.1Ir29.l4r 16.7 5

t a 1,60r51. ? 5I 66.35t77.24I 87.65r98.4 5209.0 6

2 r9.502 23.1 0

r 309.84ß75.6ô1524.73t73t.092r12.8427?5.s3J6r9.G l4 706.6460 60.877 7 20.869705.5 4

r2 rO9.O 5t4971.47r8l3r.4ôr8540.0o

SUPERSTRUCfURE

SUgSTRUCfURE

l'c.5ooo Þ!il'c. 400O 9ri

SUPERSTRUCTURE A COLUMN SECTION

FIGURE 7 Case 5L: Columbia River Briilge superstructures and substructure details used for analysis'

Po{

I

7'MtN3'.6' MAX

9'. o' r{lN24'-O'MAX

CASE 5L COLUMBIA RIVERTOP FIBER STRESS

o u.s.^+ ¡E9 :[^t^Í0x G. aRt IA¡f

BR DGE

^o-qØEg'

u)ou) c2

LlJ 8'É@

tsU)

oeio.ñ

ooci

oo<so

o^q-oØ?Èr

Ø3frg&,U)

it/

* *-i-*

I

j

* "-r'* *

i

:

,tt**,

i\i\

oooo'.

oo

FIGURE B Case 5L: longitudinal variations in top fiber stresses along a girder centerìine caused by positivetemperafure gradients.

CASE 5L COLUIIBiA RIVER BRlDGE8OT TON F i BER SIRESS

't

I

.|ie

oqêo9.

I:tt

:

:

i:

:: \/v

'o'. oo as . c0 s0. 00 2s . 00 ¡ 0o. oo r 25. oo r so. oo r 7s. oo zoo. ooDISTANCE ALONG BRIDCF cENIERLINE (FT) .IO'

FIGURE 9 Caee 5L: longitudinal variation in bottom fiber stresses along a girder centerline caused by positivetemperature gradients.

109Imbsen and Vandershaf

+

-¿00 0

usA

FIGURE l0 Case 5L, 9'ft deep cross section:

temperature gradients.

+

Tens ionCompress ion

\-

L

-¿OO 0 1400

USA

FIGURE l1 Case 5L,24'1t deep cross section:

temperatule gradients.

r400 -800 -{00 0 1100 +800

New Zealand

variation in girder cross-section stresses with depth caused by positive

TenslonCompress ion

-.¿Oo o r¿00 ( psl )

England

-800 -4OO 0 1400 1800 -¿00

llerv Ze al and

variation in girder cross-section stresses Ìrith depth caused by positive

0 r¿00(Psi)

England

poínts of span 2 (see Figure 2) a short time afterãornpletion of construction. Thernal gradient ef-fecls are thought to have been a significant con-tributing factor to this distress' The crackwiilthsr in fact, were observed to be opening and

closing on a daily basis, generally correlating rea-sonably wetl with daily tenperature fluctuations'

This structure was constructed by the segnental¡balanced-cantílever nethoat. Prestress tendons aretypically placed in the bottom stab wíthín the cen-ter portion of the span to resist positive bendingmonents that can result from creep after the canti-levered portions of the suPerstructure are Èied tÕ-gether. In this bridge these tendons terninated in[.ne vicinity of the cracking. several othêr sirnilarbrÍdges "ori"trr.t"d

nearby at about the same tinehave developed similar cracking patterns'

A plot that shows dead-, prestress-, and U'S'thernal-load stresses is shown in Figure 15' Thisplot shoyrs that tensile stresses occur in the botton

fiber at the same location grhere the cracks devel-oped in the Miller Creek Brialge.

Baseal on the iletails of construction and thecharacteristics of the observed cracking, it ispostulated that the reasons for the cracking appear

to be some conbination of the following:

I. Greater inetastic reclistribution of stress(i.e.r increase of positive monent from creep) thananticipated,

2. Stress concentratíons in the prestress an-choragê zone, and

3. Thernal gradient stresses.

Àlthough the prinary cause of the distress cannotbe precisely determined, it appears that thê inelas-tic redistributíon of stress that results in an ín-crease of positive moment (reason I) is probably themost inportant single factor. Local tensionstresses caused by prestress anchorages an¿l thernal

r10

TYpteÂr sFcTtoN

FIGURE 12 Miller füeek Bridge superstructure and substructurs {sfail¡ used for analysis.

MI LLER CREEK BR I DGETOP FIBER STRESS

o us^ ci^DlEil. x€9 ¿Ê^r^iD Gt^Dl€rf+ 8nlilsx rt^o1¡¡r++---+++

i\++ _ *.+-i\ \

60.00 i20.00 r80.00 240.oo J00.30 360.ooD I STANCE ALONC BRI DGF CENTERL I NE (FT)

FIGURE 13 Case 8L: longitudinal variation in top fiber stresses along a.girder centerline eaused by positivetemperature gradients,

/r

/

^o-qØ3g-

<no'ctq[!9üNts(n

00

Õ.o'oo

o(ÍrqôãuJ-cr: r

(n

11I L L FR CREEK BR 1 DGEBOT TOM F I BER STRESS

0 !sÀ cn^orÉ¡r n€v ¡É^!^N0 çñÐlÉrf+ litrttr ct¡or€nt

^o-qØe

.oo t20.oo t80.00 240.00 3q0.0o J60'00DiSTANCE ALONG BRIDGE CENTERLINE (FT)

FIGURE 14 Case BL: longitudinal variation in bottom ffüer stresees along a girder centerline eaused by

positive temperaürre. gradients.

MI LLER CREEK BR I DGEBOTTOM FIBER STRESS

o DEAO LO^D . rRÉSf€SSè Dl . PS . !6^ CR^g¡t¡!

Å

/.\

'60.00 t20.oo t80.c0 240.00 300.co J60.cc 420.co 480.00DISTANCE ALONG BRIDGE CENTERLINÉ (FT)

FIGURE 15 Case BL: Iongitudinal variation in hottom fiber stresees along a girder eenterline caused by

eombined dead loado positive temperature gradierrt, antl p.restressing.

00

oodo'o

aa.o

U)a.Dar¡30loa

o<;o

r

I

I -:Á-

00

ótit

0

rt2

gradient êtresses, although significant, are prob-ably of secondary inportance.

The cracking probably couldl have been avoitted byextending the bottorn slab prestress tendons in thezones of high bottom-fiber cornprêssion, thus anchor-ing then much closer to bents 2 ancl 3.

In surnmary, it appears that although the thernalgradient stresses do contribuÈe to the crackingproblem, they are not the basic cause of it.

CONCLUSIONS AND RECOM¡,iENDATIONS

Several câse studies were conducted to investígatethe effects of various assumed thermal gradients onthe longitudinal and transverse fiber stresses in-duced in various. types of concrete bridge super-structures. The following conclusions about thenature of ther¡¡al effects on concrete bridges aredrawn fro¡n an asEessnent of the results of thesecase studies.

1. Although fiber stresses induced in differentbrídges by any single thernal grailient may vary inmagnitude, the stress patterns are generally similar.

2. Cross-sectional changes such as bottom slabflares and haunches have a significant effect on thefiber stresses induced by therrnal gradients.

3. The calculated fiber stresses were sensitiveto the type of tenperature gradient assume¿I. À1-though some gradients produced similar, extrenìefiber stresses, the tlifferences in fiber stressesbetneen the extrene fibers within the girder sectionwere sígnificant. fn nany cases this difference ¡vassufficient to affect the requirements for longitudi-na1 reinforcenent. Thís could be a contributingfactor in some cases where concrete cracking hasbeen observed.

Based on the results fron these case studies anda review of the ilesign approaches used in othercountries, it is apparent that U.S. bridge designrequirenents should be expanded to include the ef-fects of a thermal gradient. The current AASHTOprocedures for considering Longitudinal movenent arebasically adequate, a1-though it would be tlesirableto include a mor€ accurate rnethod for deterrniningseasonal variaÈions in tenperature. In Iight ofthese observations, the following recom¡nendationsare nìade for i.mproved thermal design procedures inthe united states.

1. Maps that reflect climatic variation, similarto maps used in the British design stândard and theOntario bridge code, should be included in theÀÀSHTO specifications. Data contained ín these mapscan be used to develop the nuch-needed methocls fordeternining structure tetnperatures.

2. Àn AASHTO guide specification should. be de-veloped to províde for the effects of thernal gradi-ents. Because there is little or no evidence ofdistress caused by a tenperature gra¿lient in conven-tionally reinforcedl-concrete bridges, this guidespecificâtion should be directed toward the Larger,prestressêd-concrete bridges, which past experiencehas denonstrated are nore likely to devel_op problensfron temperature differentials. This guide specifi-catíon ehoul¿l include the following:

a. l¡lethods for deter¡nining both longítudinaland transverse stressesi

b. A nonlinear temperature gradient with ashape similar to thât used in the Britishdesign standard; actual te¡nperature dif-ferentials will vary;

c. An option to accommodate the qradientstresses by increasing the prestressforce or by deterrníning the anount of

Transportation Research Record 950

auxilíary ¡ni1d reinforcement by the par-tlâ1 prestressing concept;

d. A nap that indicates the maxirnum probablesolar radiation in different geographicalareas to help deterrnine t,he maximum ten-perature differential for a given bridgesite; and

e. Transverse analysis consideration of theeffects of a temperature differentialcâused by solar radiation on the top deckslab.

3. rn developing a method for designing bridgesfor thermäl effects, there are several points that¡nust be kept in mind. These pointsr which relate tothe current state of the art and design practices,are as follows:

a. The format of the design specifícationsshould be general enough to include ad-vances in the state of the art as theyare developed and allow for extensíon ofthe procêdures to other bridge types inthe future,

b. The proce¿lures should provicle for maxi¡nunsimplícity without sacrifícing signifi-cant accuracy,

c. The need for future tralning to inplementthe procedures nust be considered, and

d. In light of the línited number of casesin which tenperature-índuced ¿listress hasbeen observe¿I, there is a potential prob-lem with designer acceptance of elaboratetherrnal design procedures.

REFERENCES

l. R.A. Imbsen, D.E. Vandershaf, and C.F.Stewart. Thernal Effects in Concrete BridgeSuperstructures. NCHRP Interim Report 12-22.TRB, National Research Council, Washington,D.C.., Iqay 1983.

2. fng. T. Zichner. Thermal Effects on ConcreteBridges. Presented at Comite Europeen du BetonEnlargeal Meeting, Corimission 2, pavia, ItaIy,oct.1981.

3. Iuemorandum: fn-Depth Inspection of the Creeksand Spa11s Developing in the Walls and BottomSlabs of the F-ll-AK, F-II-AL Structures overIrliller Creek on f-70. Colorado Department ofHighways, Denver, July 29,1982.

4. s.G. Fattal, T.A. Reinhold, ancl B. Ellingwood.Analysis of Thernal Stresses in InternatlySèa1ed Concrete Bridge Decks. Report FHWA/RD-80,/085. HRs-20, Office ôf Resealch and Devel-opment, FHWA, U.S. Departnent of Transporta-tion, April 1981.

5. R.D. Bro\dne. Thernal Movernent of Concrete.Concrete, Nov. 1972.

6. Building Movements an¿l Joints. portland CenentÀssociation, Skokie, I11., 1982.

7. Concrete Manual, 7th ed. Bureau of Reclana-tion, U.S. Department of the Interior, Denver,1966.

8. U. Ernerson. The Calculation of Èhe Distribu-tion of Tenperature in Briclges. TRRL Report LR561. Transport and Road Research Laboratory,Departnent of the Environment, Crowthorne,Berkshire, Englanil, 1973.

9. 14. Emerson. Bridge Tenperatures Estimated fronthe Shade Ternperature. TRRL Report 696.Transport and Road Research Laboratory, Depart-¡nent of the Environnent, Crowthorne, Berkshire,England,1976.

10. W. Black, D.S. Mossr and 14. Emerson. Bridge' Tenperatures Derived frorn Measurements of Move-

nent.. ÎRRL ReporÈ 748. Transport and Road Re-search Laboratoryr Departnent of the Environ-nent, Crovrthorner Berkshirer England, 1976.

11. !t.J.N. Prlestley. Effects of Transverse Ten-perature Grailients on Bridges. Report 394.Ministry of wÕrksr welllngton' New Zealand,Sept.1972.

12. M.J.N. Priestley. Tenperature Gradients inBridges--some Design Consí¿lerations. New zea-land Engineering, voI. 27. Patt 7, July 1972,pp. 228-233.

13. D.R.H. I{aher. The EffecÈs of Differential Tem-perature on continuous concrete Bri¿lges. CivilEngi.neerÍng Transaetíons, Institute of Enqi-neers of Australia' voI. CEI2, Part Ir April].970, pp. 29-32.

14. Vl.I.J. Price. Discussion of Papers EntitledMedway Briilge: Design, by O.A. Kerensky and G.tittle; and Medway aridge: Constructionr bytil.F. Hansen and J.A. Dunster. Proc., Institu-tion of Civil Engineers, vol. 3I, June 1965,pp.162-156.

15. !4.J.N. Priestley. tinear Heat-Flow and ThermalStress Analysis of concrete Bridge Decks. Res.Report 76./3. Department of Engineering, Uni-versity of Canterburyr Christchurch' New Zea-lancl' Feb.1976.

113

I6. ùl.J.N. Priestley. Design Therrnal Gradients forConcrete Bridges. New zealan¿l Engineering,Vol. 31' Part 9' Sept. 1976' pp. 213-219.

L7. M. Emerson. Ternperature Differences inBridges: Basis of Design Reguirements. TRRLReport 765. Transport and Road Research Labo-ratory, Departnent of the Environnent, crow-thorne, Berkshire, Englandr 1977.

18. steel, concrete, and composite Bridges--PartI: General state¡nent. British stan¿lard BS5400. British Stan¿lards Institution, Crow-thorner Berkshire, Englandr 1978.

19. ì{. Radolli and R. Green. Thernal Stresses inConcrete BrÍdge Superstructures Under SurunerConditions. fn Transportation Research Record547. TRBr National Re6earoh CounciI, llashing-ton, D.c., 1975, pP. 23-36.

20. I'l.J.N. Priestley and I.G. Buckle. A¡nbientThermal Response of concrete nridges--BridgeSeminar, 1978. Summary volune 2. structureco.¡mittee' Road Research Unitr NationåL RoadsBoard, Wellington, New Zealandr 1979.

2L'. F. Leonhardt, G. Kolbe' and J. Peter. Ternpera-ture Differences Endanger Prestresse¿l ConcreteBriilges. Brefon-und Stahlbetonaur No. 7, July1965, pp. 157-163.

Publicøtion of thís paper sponsored by Comnittee on Concrete Bridges.

Before the refine¡nent of the steel-making process,wroughÈ iron was widely used as the princÍple struc-tural material in briclge construction. lt was usedin many railroad structures during the perioil fromthe late 1800s to the early 1900s. Although thenaterial properties of wrought iron have been wellknown since the beginníng of its use, the weldedfatigue behavior has hever been adequately quanti-fied. Thus a study on the fatigue behavior ofwelded wrought-iron splice plate repairs on Norfolkand Western Raihray Bridge No. 651- in Hannibal,Iqissouri, is prèsenteil. Welded repairs are knor.¿n toresult in l-ow fatígue strength details with steelcomponents. Because the structural membêrs werewrought iron rsith steel reinforcementr it wasdesired to evaluate the seriousness of the resultingwelded details and to assess the degree of cumula-tive danage that rnay have occurred.

BRIDGE DESCRIPTION

The Norfolk and lilestern Railway Bridge was origi-nally built for the wabash Railroad in 1888 by theDetroit lron and Bridge works. The bridge spans the!4ississippi River nith seven truss spans and onecontinuous swing span for a total length of I'580 ft(Figure 1). It carries a single track and has atruss spacing of 18 ft. The bridge members are con-

Fatigue Behavior of Welded Wrought-Iron Bridge Hangers

PETER B. KEATING, JOHN W. FISHER, BEN T. YEN, ANd WILLIAM J. FRANK

ABSTRÀCT

The behavior of fatigue crack growth andfatigue strength of welded lap splicewrought-iron hangers in a railroad bridgewas studied. The origínal wrought-ironhangers were lap spliced with steel platesfor the purpose of tightening the members.Fíeld inspection revealed cracks in thewelded lap splices. Examination of simu-Iated test joints and cracked hanger splicesin the laboratory indícated that fatiguecracks would develop from weld deposits atthe cut of the wrought-iron bar, propagateinto the steel splice plate, and cause fail-ure. Fatigue cracks could also propagateinto the wrought-iron bars but would be ar-resteil by the slag (íron silicate) string-ers. Breaking of the wrought-iron bar onlyoccurred when the applied stress was quitehigh in conparison with the yield point.l'leasured live-load stresses in the actualbridge member were relatively Iow. Eval-ua-tion of traffic and load records índicatesthat the effective live-load stresses wouldbe well below the fatígue strengths of thesespli.ced joínts. No inminent problen of fa-tigue failure is expected in the hangers.