TRE Gol of Concrete Bridge Superstructures M. Radolll, M, M, Dillon Ltd., Cambridge, Ontario R. Green, ‘Department of Civil Engineering, changes are gineally considered inthe dostn of Bessure of dally change In both ab ne tersperatre and itanity of solar radiation, temperature differential, Stout ‘Supartuetores, These tarperatre difeenls indoca stress throughout te dapth of concrete structures, which sre rneraliy not include incurentdsign procedures. This pper describes {host ronsferprocases that occur betwaen the atmosphere and @ con. les supertrcture and sa tha omate eantonsnecesay fr th velopment of tomporatra cifras during both surnmer and winter. ‘Temaratare tivo analyst, computed by usinga one-dimensional heat low analysis, nase thatthe distribution of tarpersture throughout {he doptt of @ superstructure ls nonlinear and ita funetion of syparstrus- {ure dapth, Stee otolated nt ho nonlinear temporsture gradients [re desrlbads This stress con bs saveral time thos du ove oad, ‘Spsclly in continuous yetoms, Tha srestes predicted from the de [2ed dstlbutone are compared to thor obtained by using the hest fo nals The rooltsinefeate that tho fealzad dstbution hove liitad Grr valea, Simple empirical design exprasons are developed for both ‘hermelsesss and curvature, These aro Baer on typist climate data {or surer and winter condtiens and ean bo spied to varity of ‘Gosssetionel superstructure geometso. An exampla of the stresses Seedby omaictgchs tesa smears Limited data are available to assist the bridge designer ‘concerned with the stresses induced in a concrete bridge superstructure by heating and cooling effects (1, 2, 3 4). Carrent AASHTO specifications include probable Teinperatare ranges to be used in design and generally offer guldance only with respect to expansion and con fraction of straight structures. Mean temperature con- ditions in actual structares can differ apprectably from the ranges suggested (4, §). No guidance 1s provided to designers coneerned with serviceability problems in deep concrete superstructures where significant temperature differentials are possible: For oxample, temperature ifferentials approsehing 22°C (40°8) la frequently spect {ed temperature rise) have been observed between the top and bottom of a deep (1.4-m or 4.5-t) conerete box. gizder (0) Publication ofthis paper sponsored by Committe on Generel Srctures. University of Waterloo, Ontario ome bridge dosign specttications or recommendations do consider differentist ag well as mean temperature ef- fects 2, 3, 4). German Industrial Standard (OIN) 1072 nd British dlandasd 189 recognize the presence of a temporature grattent through the depth of a bridge super- Structure and give design values for gradionts in both Steel and conercte structures (3), However, little or no guidance 19 offered regarding the vertical distribution of Rio tomporature alfterential throughout the depth of a Concrete bridge superstructure. Procedures for clea {ating the forees and streages induced in the superstruc~ tare by such temperature differentiais are also lacking. ‘hy Is there concern about Lerporature gradients in eonoreto bridge superstructures? Much of the design Sxperience embodied in curzeat specifications ts based onthe study of structures with both cross-sectional and Gian geometries designed and constructed soveral years Ryo. “The results of thoge studies may nat be typical of Siperstructures currently dealgned and constructed. For vample, concrete box gltder superstrustures aro a re- font fnngvation, and Held observations indicate tompora- ture differentials of more than 22°C (40°F) can exist be- tween the upper and lower tlanges of a box gindor eyatem (Gh. This temperature differential gives 1iso to local Strosaea that avo nearly four times those attributed to fill live load. Studies of a two-span prestressed con rete structure have indlgated tonsile stresses of moro than 3450 kPa (600 Ibf/in’) directly above the intermedi- So support as a result of moderate temperataro differ Sntiala (7) Such stress values are of importance in the enign of prestressed concrete structures where orack- fg inthe absonce of live load is undestrable. Further Sridonce ig avatlable where damage has been atributed to thermally induced atressos (@, 3, 10, 11) HEAT FLOW THROUGH A CONCRETE SUPERSTRUCTURE ‘The distribution of teraperature throughout the dopth of a eonerete superstructure must be known if the resulting stresses, reactions, and deformations are to be calew- fated, It hag been possible to correlate weathor data vith tke surface temperatures of exposed pavements dur~ Ing heating and cooling eyeles (12). However, the in-fduence of changes fn elimatie conditions, wlth both time and location, and the factors controlling the heat galnod or fost bya superstructore do not allow the dl- oct application of the results obtained from the study of pavements to concreta superstructures. ‘An expoved eoncrote bridge deck is continually Losing and gaining heat~Crom sola radiation, radiation to or iroin the sky oF surrounding objects, and eonvection to or trom the surrounding atmosphere. Ta the daytime, tnd eopecialy duving the summer, the heat gain is greater than the heat Toss, resulling In a temperature inerease throughout the depth. During a typieal winter night, the converse is true, and the temporature In the npotatructure decreases.” Hest inp typical of a sum- mor day reoulte in positive tomporatare gradients tn tho dock in which the top surface fs warmer than tho bottom. Negative gradients, in which the top sutlaco 19 cooler {han the bottom, result from a net heat loss. “Tho heat flow processes for typical summer and win ter conditions are shown in Figure 1. Because ofthe oor thermal conduetivily of concrete, these processes Ean result in tomporature gradionts in a concrete struc- fue and changes in moan temperature. Large positive temperature gradients occur during a day with high solar radiation, clear skies, a large range Sthmbient temperature, and’a Ught wind. High meen temperatares, on the olor hand, are assoctated with a high intensity of solar radlation combined with « hgh, almost constant, amblent alr temperature. Negative temperature gradients develop during cooling periods fseociated with evening conditions. When the mesh bridge fomporature exceeds the smbtent temperature at {a particular time, more radiant energy is lst from the Saposed top surface of abridge experstructare than from the sheltered bottom surfaca. ‘Weather data sultable for ose tn a heat tlow analysis considering both radiant and convective heat exchange on tho upper and lower surfaces of bridge deck vary with toth imo and location. Conditions at a given time and position must bo considered, Obsorvations forthe tnontho of May to Augustin Toronto indieate thatthe change in daily ambiental fomperature is approx mately 10°C (16°F), and the daily intensity of solar radi- Aton gn a horizontal surface has « moan value of 28 ‘MJ/m* (550 langleys). Consideration of these values lenis to the choice ofthe climatic data shawn In Figure ans boing sepresentativo ofa day of high-Intensity Sclae radiation and above-avorage temperature change. “\milaniy, data roproventative of elesr, still winter sight in January were eelocted forthe heat loss eonition ‘Pigare 20). Heat flow through a bridge superstructure varies dur- Ing the diurnal oyete and is a aon-steady-stale process. Honee, laws governing stondy-state hest flow are not nplieable to exposed structures, and any analysis of the Tenponse of puch structires lo heat low rust consider tho variation of tomperature throughout the dogth of the structure with times ‘A one-dimensional heat low analysis similar to that sed by Emerson (6) was developed, and the resulting arti diforential equation was solved by using thefinite fitference method. the appropriate boundary conditions considered for the hent flow analysis and the material Constante ascumed for the concrete 276 given In Table 1. ‘Tho asoumplion of unidirectional hext tlw from tho exposed surfaces of tho concrete superstructura to the ierior is not stritly correct. Comparisons vith ob- served and predicted values of temperature (19) indi- ato good correlation with snidtrectional heat How anal- yses, notwithstanding small transverse temperature gradients “Tho application of the hest flow analysts for the rep~ resentative climatte conditions (Figure 2) resulted in the temperature dlatributions chown in Figure 3 for solid slabs of different depths. Temperature distribution is = funetion of member depth for both winter and summer conditions. For the summer conditions, the surface temperatures are nearly 11°C (20°F) greater than the maximum amblent'air temperature, and the temperature fistributign is nonlinear for slab depths greater than ap- proximately 30 em (12 in). The temperature at the mid- epth of 78 and 122-em (38 and 48-In) sections is not in- Huenced by the exterfor heat, and the change in the moan temperature of these sections ts small. Winter tompera~ ture gradients do not appear to be so severe as those as- soclated with summer conditions. Thus, a slab with a depth of more than approximately 30 cma (12 in) with, ‘when heated ar cooled ag a consequence of changing radi- ation and temperature conditions, be subjected to non- Bnear temperature distribution. THERMAL STRESSES ‘A nonlinear temporatuce distribution and the strains as- sociated with this gradient lead to some complexities in the computation of stresses. Structural designers do not tasually consider the effect of initial strains in the anal ysis of sections, Figure 4 shows a typical member with ‘an arbitrary crose goction and vertical temperature dis~ tribution. The temperatura varies only in the vertical irection in this enalyels, Full restraint is provided to the ends of the member by the moment M and the axial force P. ‘A member analysis made by assuming that (a) the material Is elastic and has tempsrature-independent properties, (b) plane sections remain plane after bending, fand {c) the principle of elastic superposition is valid leads to the following equation for the longitudinal stresses: Ota for a fully restrained member with a noniinear tempefature-induced strain ge Bot 0 where ‘oye = longitudinal stress at a fiber located a distance x from the conter of gravity, « = coettictent of thermal expansion, AT = change in temperature, and E = modulus of elasticity. ‘The restraining end moment M can be evaluated by and the longitudinal stress assoctated with this monient rye is given by accone([aemalfes) 0 ‘Similarly the restralaing end force P acting on the area A and associated stress o,,, are nae feast ioe “ (ff sero fors) “ 5 = PIAFigur 1. Heat sin and loss procasss for (a) summer and (b) winar conditions. rn ‘ton Sotee wre ae ego i Yh finesse? Flours 2, Types climate data for (a highintnsiy and (low toasty redtion, Figure 3. Typical vmpecaturedsebutions during {a} summer nd (wits eee (2) earns Sere CT) q] ic] « LO x & | ita to =e Figure 4, Restiained boar with aonlager temparatur ditibution. J eT om e = al iets = Figure 5, Seotona stsne and asociatedsrosses, Tebla 1, Thermal and elastic properties wid in one dimensional heat How analy. + = ober 8 ae, 23 aes Banowe ae : Ec om he10 ‘The Longitudinal thermal stress ¢ present in a fiber 1o~ cated a distance x above the center of gravity of a long thin member with no end restraint (13, 14 15) has the form © Subtraction of the strosses inducod by the restrain ing forces P and M (equations 3 and 6) leads to equation 8. This lalter equation describes the stress state cor- responding to a nonlinear temperature gradient in an un- restrained member. "The concept of 2 stress state in an unrestrained mem= ber that 18 not eubjected to external forces may be new for the bridge designer. Rolerenco to Figure § may: clarify this coneopt. The nonlinear temparature distri- bution attempts to induce e nonlinear strain distribution proportional in magnitude to the temperature distribu- Hon. Additional strains develop in the section as a con~ sequence of the Euler-Bornoulli hypothesis concerning plane sections. These strains havo associated self- equilibrating stressas similar in form to the residual stresses found in fabricated steel mombors. ‘The self equilibrating stressos dopend diréetly on the daily hoat- ing and cooling cycle and material properties; they do not depend on the support conditions of a structure and, until a satisfactory English-language term is established, ‘vill be referred to as eigen stresses. Ifthe curvature of a member induced by a tempera tare gradient is restrained by the interior supports of @ continuous beam, additional stressos develop. These additional streasos will bo referred to as continuity stresses,’ and the form of the moments and reactions as~ sociated with these stresses is shown in Figize 6, ‘The value of maximum thermal continuity stress for the two~ span beam is an thete o where continuity stress at the support, thermal eurvaiure, and aistance tothe extreme fiber from the centroldal axis, ‘and, in general, 7 CxCa$dE ® where C1 = constant developed from the span geometzies (1.5 for a two-span system, 1.0 for the interior span of a multispan system), -anstant doveloped from the sectional geometry (0.5 for a rectangular section), and = member depth. G ‘The total stresses developed In a two-span member for typical summer and winter conditions (Figure 7) con= sist of both eigen stresses and continuity stresses, As mentioiied previously, the eigen strosses are sell= cquilibrating stresses formed as a consequence of the nonlinear temperature distribution throughout the section dopth, and contiiuily stresses develop because the de~ lection at point B (Figure 7) 1s zero. Inspection of the total stress patterns (Figure Ta) indicates that large tonsile stresses develop in the web of the member 98 a consequence of the summer conditions. ‘Thus, a prestressed concrete member designed for zero tension at support B under dead load plus prestress should be provided with reinforcing steal in the web to cater for these thermally inducod tensile stra; Leonhardt has also noted this requiremont based on a simpler analysis (2), Woll-distributed steel is required ‘in the flange to ealer for the tensile stresses due to the winter conaition (Figure 7b. tis apparent (rom Figure 6 that summer heating ef~ fects change the values of reaction. Some experimental data are available describing reaction change for a three span bridge, but regrettably values of tho intensity of In- Coming solax radiation were not available (16). ‘Typical values were assumed, and the reaction changes calculated from derived curvature values appear to predict the ob~ served values closely for a day with high solar radiation (18). The observed reaction change results In a moment change of nearly 60 percent of the dead load moment for the double girder throe-span structure with spans of 13, 47, and 13 m (44, 58, and 44 tt), DESIGN CONSIDERATIONS 1a the Mterature, several tomperature distributions havo been suggested for design (7, 14, 11). The adequacy of thee distributions with respoat to the prediction of eigen stress and continully stress values in concrete super- structures does nol soon fo hare been examined provi~ ously. Figure 8 shows sovoral suggested temperatura distributions, The elgen stresses and curvature valios Induced by these distributions were computed for sum= ter eonditions. 18 apparent that the form and magal— tude of both the elgen stresses and curvatures are strongly dependent on both assumed temperature differ~ ence and temperature gradient. Linear gradients do not Induce eigen stresses. Tomperalure measurements ob- tained from prototype structures with depths of approx mately 1 m (lt) correspond most closely to thoso pr dlcted by a slath-degree parabola (6). The stresses corresponding to this distribution are very similar in Iagnitile and distribution Uo those recommended by Maher, who considera a linear gradient throughout the depth of the top slab of a box girder superstruc- fuze (7), Tho temperature elference (23°C or 40°F) tsed inthe example ehould be considered as an upper Tmt @. ‘Three representative temperature distributions and the one-dimensional heat flow analysis were used to com- pile elgen stresses and nondimenslonal curvatures (6d), {or various depths of a sold slab. The distributions ‘sed and the results obtalned are given in Figures 9 ant 10." Not ali concrete superstructures are solid, and corrections to the analysis for collular structures aro avaliable (18), “Two valies of temperature differential (10 and 20°C or 18 and 36°F) wero considered for the linear gradient {the Leonhardt gratient). For the Prlestley and Maher gradients, a temperature differential of 19°C @5°F) as ised, ‘The linear Mahor gradient vas assumed to extend to 20 om (@ in) below the surface of the solid suporstruc~ ture, ‘The sixth-order parabolic distribution suggested by Priestley (14) was used. Figure @ shows that the Meher distribution and ono~ dimensional analysis compare favorably for the predle~ tion of elgen stress Values for a wide range of depths, ‘The Priestley distribution appears to have a limited range of applleability, and eigen stresses cannot be pre~ dicted by woing a linear temperature distribution. The hondimensional curvature versus depth predictions {Figure 10) stow thal the Maher distribation will over- estimate curvature compared fo the one-dimensional ualysis for shallow members; the Priestley distriaution applies only to member depths of between 0,75 and 1.0 m {G0 to 40 in), and the Leonhardt distribution is indepen-Figure 6, Restainad cureature of a member with {ah tomparatre gradint, (0 Induced curvature, {6) appli nsding, and (nal mmomeets and reactions, fe fe «a I ae [eee ce perenne wager Flow Tamperate dubs tne on ese od = av. Figura 9, Eigen tees comparisons for ‘rove diwituions and depth.12 ent of member depth, contrary to the Maher and one~ dimensional analysis predictions. For design purposes, the calculation of the elgen and continuity stresses associated with a given temperature distribution would be tedious. Slmple empirical design expressions were developed from Radolli's analyses. (18) for use as first approximations in design. Tha re- sults for summer heating conditions are givon in Figures, 8 and 10 and allow calculation of both eigen stresses and nondimensional curvature for various depths of slab with rélative ease. For depths less than 90 em (20 in), the nondimensional curvature ts assumed fo be constant. These expressions are based on climatic conditions for Toronto and are therefore valld for locations of similar climate, Figures 9 and 10 can be used to predict values asso- ciated with winter conditions. Tt was found that eigen streaa values for winter hoat flow conditions are nearly identical to the values for summer conditions but of op~ posite sign; also curvature values are of the opposite ign and are only 60 percent of the summer values, Thus, fone set of simple ealeulations can be used to obtaln the Figure 10. ¢ for various ditibtions and depths, [rset OE asgpe test Figure 11, Obsered sad computed tamperstue grants daring le summer ad fo inte. stress values associated with thermal loading for both eurmmer and winter conditions. ‘The results prosonted apply to a variety of eross~ sectional geometries including cellular and T-beam sections (18, 19). Variations in material properties and heat transfer coefficients will influence stress values However, the changes in the stress values for a wide range of properties and coefficients were found to be within 25 percent of the values presented here (18). DISCUSSION OF RESULTS ‘The analyses and results presented for stress aro based {fm part on a one-dimensional heat transfer analysis for agolld slab, A comparison of measured and computed temperature values (Figure 11) confirms the general validity of the analysis, Tho effect of thermal loadings associated with summer ‘conditions on the design stress envelopes for typical con- ‘rete bridge superstructures has been discussed (19), Figure 12 provides an example of the stresses developed {in a tyo-span posttensioned box girder as a consequence fof winter cooling conditfons. Surface tensile stresses develop throughout the length of the structure as a cor sequence of the nonlinear temperature distribulJon. ‘Those tensile stresses influence both the span and sup~ port regions of the structure, The serviceability of such fa box girder structure could well be affected if additional bonded relnforeing stoct is not added for exack control sn zones of high tensile stress. ‘Thermal loadings will not affect the overall strength Figure 12, Stress ina box gee brid: () banding momant due to thermal fea, (8 tot foal Ss (fond dip ss ond (erin f i eo} }000 i , | = [ie rine po (4100 rouscens | reese SA jensof a continuous superstructure Inasmuch as failure of the superstructure will rasult in a release of the restraints Causing thermal etrosses. However, heating and cooling Of the structure during a normal diurnal eyele combined {with Iifetime creep and shrinkage must be considered in the serviceability analysis of concrete superstructures. CONCLUSIONS Significant flonural stresses are developed in concrete bridge superstructures as a consequence of the heating or cooling of conerete—a material of relatively poor thermal conductivity, Nonlinear temperature gradtonts, develop during a dally heating or cooling eyclo. These Gradients lead to both elgen and continuity stresses. Simple design formulas are developed that allow the pro- diction of these stresses and do not require a Knowledge of the temperature gradient. ‘Thermal loading of a superstructure ts a serviceabil- ity problem and should bo 0 considered. Proporttoning of reinforcement in a concrete bridge superstructure Should reflect the stress-indueing thermal effects. ACKNOWLEDGMENTS ‘rho work was carried out with the support of the Trans portation Development Agency of Canada, the Natlonal Research Counell of Canada, and the Department of Civil Engineering, University of Waterloo, This sup~ port is gratefully acknowledged. REFERENCES 1, Standard specifications for Highway Dridgeo, Eleventh Ed, AASHO, Washington, D.C.,, 1973, 2. Design of Highway Bridges. Canadian Stindards Association, Rexdale, Ontario, CSA standard 88~ 4974, 1974. 8. D.J. Lee. ‘The Theory and Practice of Bearings and Expansion Joints for Bridges. Cement and Conerete Assoctation, London, 1971. 4, M. Emerson, Bridge Temperation and Movements in the British Isles. U.K. Road Researeh Labora tory, Crowthorne, Berkshire, England, RRL Rept, L228, 1968. 5. M. Emerson, The Calculation of the Distribution of Temperature in Bridges. U.K, Transport and Road Research Laboratory, Crowthome, Berkshire, England, TRL Rept. LR6O1,, 1973, 6, H.Bosshart. Thermal Stress in Prestressed Con~ crete Bridges, Symposium on Design of Concrete Structures for Creep, shrinkage and Temperature Ghangos, Mads, Pinal Rept Vol: 6 197s yo 13-60, F, Leonhardt and W. Lippoth. Folgerungen aus Schiiden an Spannbetonbriicken. Beton-und stahl~ betonbau, Vol. 65, No. 10, Oct. 1970, pp. 281-244. 8, J.N, Deserio, Thermal and Shrinkage Stresses— ‘Thoy Damage Structures! ACI, Detroit, Publ. P21, 1971, pp. 43-49. 9, 7, Monier, Cases of Damage to Prestressed Con- orete, Heron, Vol. 18 No. 2 1972. 10, K. H. Weber. "Causes of Crack Formation Near In- termediate Supporta of Continuous Prestressed Con- terete Beams and Their Limitations to Allowable Crack Width, Univ, of Stultgart, dissertation, 1967. LL, RD, Larrabee, D. P. Billington, and J. F. Abel. ‘Thermal Loading of Thin-Shell Coneréte Cooling, ‘Towers, Journal of Structural Division, Proc ASCE, Vol. 100, No, ST12, Dee, 1974, pp. 2367- 2382. E.$ Barber, Calculation of Maximum Pavement 1. 13. 14 15. 16. 1, 16. 13. 13 ‘Temperature From Weather Reports, HRB, Bul Ietin 188, 1957, pp. 1-8, W.1.J, Briee and RG. Tyler, Ettects of Creep, Shrinkage and Temperature on Highway Bridges in the United Kingdom, Symposium on Dosign of Con~ crele Structures for Creep, Shrinkage and Tempera- ture Changes, Madrid, Final Rept., Vol. 6, 1970, pp. 81-83. MM. J.N. Priestley. Effects of Transvorso Tempora- ture Gradients on Bridges. New Zealand Miniotry of Works, Central Laboratories, Rept, 394, Oct. 107, R. Hoyle, Plane Strain and Plane stress. tn Ther~ mal Stress, Pitman and Sons, London, 1964 pp. 45-51. WN. Krishnamurthy, Temperature Bifects an Con~ tinuous Reinforced Concrete Bridges. Alabama Highway Research, HPR Rept. 58, July 1071, D. RH. Maher. The Bitects of Differential Tom~ perature on Continuous Prestressed Concrete Bridges. Institution of Engineers, Australia, Civil Engineering Trans., Vol. CE{2, No. 1, Paper 212, April 1970, pp. 29-82. MM, Radolli, “Thermal Stresses in Conerete Bridge Superstructures, Univ. of Waterloo, Ontario, MASc thesis, March 1975, M. Radollf and R. Green, Thermal stress in Con exete Bridge Superstruciures Under Summer Con ditions. TRS, Transportation Research Record 547, 1075, pp. 23-88,BRIDGES: Loading, Analysis, Design and evaluation. TEMPERATURE AND SHRINKAGE LOADING TEMPERATURE Introduction ‘The temperature of a bridge deck varies throughout its mass. The variation is caused by: the position of the sun the intensity of the suns rays thermal conductivity of the concrete and surfacing wind the cross sectional make-up of the structure The effects are complex and have been investigated in the UK by the Transport Research Laboratory (TRL). Changes occur on a DAILY (short term) and ANNUAL (long term) basis. Daily there is heat gain by day and heat loss by night. Annually there is a variation of the ambient (surrounding) temperature, Daily variations, Temperatures near the top are controlled by incident solar radiation, and temperatures near the bottom are controlled by shade temperature. The general distribution is indicated in Figure TI. Positive represents a rapid rise in temperature of the deck slab due to direct sunlight (solar radiation). Negative represents a falling ambient temperature due to heat loss (re-radiation) from the structure. ae Ti TARR surfacing L deck slab beam Heating (positive) Cooling (negative) Figure T1 Typical temperature distributions Research has indicated that for the purposes of analysis the distributions (or thermal gradients) can be idealised for different ‘groups’ of structure as defined in Figure 9 of BD37/88 Cl. 5.4 or Figure 22 of TMH7 C14.5. The critical parameters are the thickness of the surfacing; the thickness of the deck slab and the nature of the beam. Concrete construction falls within Group 4. ‘Temperature and shrinkage stresses MAR 1999 Page No.1BRIDGES: Loading, Analysis, Design and evaluation. Temperature differences cause curvature of the deck and result in internal primary and secondary stresses within the structure. Primary stresses occur in both simply supported and continuous bridges and are manifested as a vatiation of stress with depth. They develop due to the redistribution of restrained temperature stresses which is a self-equilibrating process. They are determined by balancing the restrained stresses with an equivalent system of a couple and a direct force acting at the neutral axis position. The section is divided into slices, and the restraint force in each slice determined, ‘The sum of the moments of each force about the neutral axis and the sum of the forces gives the couple and the direct force respectively. ‘This is best illustrated by an example later. Secondary stresses occur in continuous bridges only and resut dea change inthe global reactions and bending moments. They are determined by applying the couple and the force at each end of the continuous bridge and determining the resulting reactions and moments These are then added to the self weight and live load reactions and moments. Primary stresses are not necessarily larger than secondary stresses. Both can be significant and depend on a whole range of variables. Once calculated they are included in Combination 3. ‘The thermal expansion coefficient Br = 12x10* for normal aggregates and 9x10° for limestone agaregate, and the short term modular ratio ot is used. EXAMPLE 1 (Primary stress calculation) Determine the stresses induced by both the positive and reverse temperature differences for the concrete box girder bridge shown in Figure ST1. {A=940000mm? , =102534x10° mm* , Depth to NA=409mm, Br = 12x10%, E= 34 kN/mm’ ), 2000) a TT Th 7 aL aes wet sai AD ito Figure ST1 Box girder dimensions & temperature distribution 1. Calculate critical depths of temperature distribution From BS5400 Figure 9 this is a Group 4 section, therefore: hI=0.3h=0,3x1000=300>150 Thus hl = 180mm h2=0.3h = 0.3x1000=300>250 Thus h2 = 250mm ‘Temperature and shrinkage stresses MIR 1999 Page No.2BRIDGES: Loading, Analysis, Design and evaluation, 100 + OCF] Fr h3=0,3h = 0.3x1000=300>170 Thush3= 170mm 2. Calculate temperature distribution Basic values are in Figure 9 which are modified for depth of section and surface thickness by interpolating from Table 24. ‘TL =17,8 + (17,8-13.5)20/50= 16.1°C \ wy ol ty TA = 4.0 + (4.0-3.0)20/50 3.60°C | jesse 226°C vp PLA 2A + (2.5-2.1)20/50 pt Calculate restraint forces at critical points This is accomplished by dividing the depth into convenient elements corresponding to changes in the distribution diagram and/or changes in the section. (see Figure ST2) F=E.BrT Ay ~ FL = 34000 x 12 x 10% x (16.1- 3.6) x 2000x150 /1000x2 = 765KN F2 = 34000 x 12 x 10% x (3.6) x 2000x150 /1000 441KN 4000 x 12 x 10* x [(3.6+2.6)/2] x 2000x(220-150/1000 = —«I77KN 34000 x 12 x 10% x(2.6/2) x 2 x (250-70)x250 /1000 48kN FS = 34000 x 12 x 10% x(2,26/2) x 1000x170 /1000 T8KN TOTALF = 1509KN (tensile) Top stab 220 — Be — F4 Sead Ly —_FS Figure ST2 Element forces 4, Caloulate restraint moment about the neutral axis, M = [765(409-50) + 441(409-75) + 177(409-185) + 48(409-270) - 78(591-170x2/3)]/1000 M = 431kNm (hogging) 5, Calculate restraint stresses PE. Br Ts fey = -34000x12x10* x16,1 ‘Temperature and shrinkage stresses MIR 1999 Page No.3 u -6.56 Nimm? (¢°BRIDGES: Loading, Analysis, Design and evaluation, fia = -34000x12x10% x 3.6 1.47 Nimm? (6) fis = -34000x12x10° x 2.6 1,06 N/mm? fy = -34000x1 2x10 x 0 0.00 Nimm? + fis = -34000x12x10° x 0 0.00 Nimm?* fs = -34000x12x10* x 2.26 -0.92 Ninn?» 6 Calculate balancing stresses cl a Direct stress fo = 1509 x 10° /940000 = 1,61 N/mm. Bending strosses fa = My/I: 6 = —4ilel0- e409 = 1.71. Nimm? 102534x106 431x106 a = 431x108 = 0 fa 9534x100) 1.08 N/ma 6 fy — 431210" 9180 = 0.75 Nim? 102534x10' 431x108 2 = Bde = .06 N) f= Tossax100 0.06 Nira 6 431x107 5421 = -1,76 Nimm* 102534x106 6 fy =~ 4311059 = -2.47 Nimm* 102534x106 7. Calculate final stresses “Restraint Balancing Balancing | Final stresses stresses direct stress bending ___stresses i 6.56 16 Ln 3.24 (©) 2 =147 1.61 1.08 1.14 (1) 3 -1.06 1.61 0.75 4 0.06 3 6. 6 Table STI Summary of stresses ‘The final stress distribution is shown in Figure ST3. Similar calculations for the cooling (reverse) situation are shown in Figure ST4. ‘Temperature and shrinkage stresses MIR 1999 Page No.4BRIDGES: Loading, Analysis, Design and evaluation, 856 4.81 474 324 0.92 287 AB Restrained Stresses due Stresses due Final self stresses torelaxing toxelaxing’ —_equllibrating force moment stresses Figure ST3__Final stress distribution (Positive) 3.827 138-086 1.89 287 0.83) 2.00 Restrained Stressesdue Stressesddue Fouls stresses torelasing toreloxing —_equilbrating force moment sesso Figure ST4 Final stress distribution (Reverse) Annual variations, ‘Annual (or seasonal) changes result in a change in length of the bridge and therefore affects the design of both bearings and expansion joints, ‘The thermal movement of a bridge takes place about a point called the thermal centre or the stagnant point and defines the length and direction over which movement occurs. The deck movement A=, .T.L ian b> oe pet ‘Temperature and shrinkage stresses we wpe MIR 1999 Page No.5BRIDGES: Loading, Analysis, Design and evaluation, where T = temperature change and is based on a total possible range of movement given by the difference of the MAXIMUM and MINIMUM shade temperatures and specified in the Code for a given bridge location as Tsotherms in Figures 7 and 8 of BD 37/88 or Figures 20 os and 21 of TMH. | ‘These are further modified to take account of the bridge construction in Tables 10 and llof ayy BD 37/88 or Tables 11 and 12 of TMH to give the effective bridge temperature. ‘The bearings and expansion joins are set'in postion to account for actual movements which | ‘Mil depend upon the fime of year in which they are installed. This is shown graphically in | Figure STS in relation to the ‘setting’ of an expansion joint. : loaned iene nasi prithon eB My eee aslacen ef int at which expansion joint installed Contraction ‘Expansion Te Tein) Tmax) TOTAL RANGE _ -| Figure STS Eifect of installation on expansion joint movement ‘The two extreme situations illustrate the problem, Ifthe joint is set fully extended at T(max) thon ae the deck cools the joint will over-extend resulting in failure of the joint, Conversely if the joint is set filly closed at T(min) then asthe deck expands it will jam. deally « graph showing the local temperature variation with time of year is needed to acourately set the joint but more often than not ths is not available and an engineering judgement has to be made. I ie aleo-wise to install a joint that has a slightly greater range than calculated in order to provide an extra margin of safety. EXAMPLE 2 (Overall movement calculation) Determine the expected range of moveiient for the bridge deck in Example 1 assuming that it is simply supported over a span of 25m and is located in East London From Figure 20 of IMH7 Minimum shiade air temperature = -1.075C From Figure 21 of TMH7 Maximum shade air temperature = 49.00°C From Table 1! of IMH7 Minimum effective bridge temperature = 1.00°C ‘Temperature and shrinkage stresses MIR 1999 Page No.6BRIDGES: Loading, Analysis, Design and evaluation, From Table 12 of TMH7 Maximum effective bridge temperature = 43.00°C ow Therefore range = 43 - 1 = 42°C. Movement range = 12 x 10*x 42 x (25 x 10?) = 12.6mm. Theoretically this should be modified to account for stiffitess or friction of the bearing by an amount A = PL/EA, P = the frictional resistance (WR) of a sliding bearing or stiffness resistance of a laminated rubber bearing. fas tbo fly O84 FL Secondary Stresses The effect of temperature variations on continuous bridges is illustrated in Figure STO. (A teeter Continuous deck rR Restrained deck Restrained moments Restrained moments oT INT Tm in continuous span Relaxing flexural i bm Sane Mil Tm. + = 1 Final flexural ARETE T moments Figure ST6 Secondary temperature effects. These moments (and associated reactions) modify the stress resultants duc to dead and superimposed loads, and hence the associated stresses within the depth of the structure will vary. ‘Temperature and shrinkage stresses MIR 1999 Page No.7il Temperature and prestress loading 11.1 INTRODUCTION ‘This chapter describes the actions of temperature loading and prestress oon bridge decks, and shows how these loads can be simulated in the analytical methods. outlined in previous chapters. Although temperature loading and prestress are described below one after the other, itis important to emphasize the very different principles of these two types of loading, _ 11.2 TEMPERATURE STRAINS AND STRESSES IN SIMPLY SUPPORTED SPAN ‘i ‘The rise of temperature in an element of material causes the clement to, expand if itis unrestrained as shown in Fig. 11.1(a). Alternatively, ifthe element is prevented from expanding as shown in Fig. 11.1(b) the risein temperature causes an increase in stress which depends on Young's modulus of the material. Either the increase in strain or the increase in stress can be taken as the starting point for the calculation of the distribution’ of temperature ‘stresses jnl the structure. For a simply supported: deck with linear variation in temperature between top and bottom surfaces, the simplest way to calculate the flexure is to work from the temperature-induced free strain, which causes flexure without ‘stress, as shown in Fig. 11.1(c). However, for the more general bridge problems with complicated deck geometry and non-linear temperature distributions, it is simplest to start from the assumption that the deck experiences temperature changes which set up stresses while the deckis, rigidly. restiained throughout, as shown in Fig. 1,1(d), and then calculate the effects of removing the theoretical restraints. ~ Figure 11.2(a) shows an element of deck with non-linear temperature distribution varying from +21°C at the top surface through 0 to-+7°C at the soffit. (The temperature distribution is derived from a heat flowFo, e ly Ik hut ure ‘Temperature strains and stresses in simply supported span 223 Fig. 11. Bitects of temperature distribution and restraint conditions on movements and stresses in elements of, bridge deck. Rig. 11.2 Temperature @iamentne calculation as described in references [1]-[3], or from the appropriate Alistibution; (b) strains code of practice.) inuntestrained slice; If the coefficient of thermal expansion is a, the unrestrained thermal ‘ (©) stressesin restrained strains are (genre expansion ¢ = @AT. (ata) and () reateined rmontent fe and residual gat concrete 4 4 Sa > @ ‘onee 224 ‘Temperature and prestress loading plane sections do not remain plane Ifa= 12x 10°C, the temperatures of Fig. 11.2(a) cause the strains shown in (b) in an unrestrained thin slice of the deck. With such strains, "When expansion is prevented, with plane sections held plane, then the locked-in stresses are compression ¢ = aEAT (11.2) where E is Young's modulus. For E=30000MNm*?, the temperatures of Fig. 11.2(a) cause the locked-in stresses of (c) in a rigidly restrained slice of deck. Tt is important to note that the strains of Fig, 11.2(b) and stresses of (©) are alternative primary effects of temperature change dependent on boundary conditions, Primary stress and strain are not proportional in accordance with Hooke’s law, as stress is Targe when strain is kept small ‘and vice versa, The temperature has in effect induced a Tack of fit” or a ‘'jacked-in force’. ‘Yn contrast, the secondary stresses and strains due to redistribution of primary moments are related by Hooke’s law if the material is elastic. ‘The locked-in stress distribution of Fig. 11.2(c), shown again ind), can be thought of as composed of two parts which affect the structure in different ways: | 1. am average compression stress shown in Fig. 11.2(e) causing resultant foree on cross-section of P; 2. a nonlinear stress, distribution shown in Fig. 11.2(f) which comprisesa moment Mand residual stresses with no net resultant. For the particular stresses of Fig. 11.2(d), the average compressior stress of (c) is simply ‘ i obde : 6 a (1.3) where = breadth. Here oe 7 [2 oat (.6+0) oat 042904] = 18MNm™*. 7 Figure 11.2(f) is simply (d) ininus this average stress in (e). ‘When: a length of deck is rigidly restrained as in Fig. 11.3(a), the stresses of Fig. 11.2(d) act on every cross-section. On internal cross sections the stresses on two sides balance, so that it is only at the ent faces that thése stresses must be balanced by the externally applic restraining forces, When longitudinal expansion restraint isremoved, a |bee ia ion 3) 5 ed as Fig, 11.3 Progressive release of temperature forces: (a) fully restrained; (©) compression release; and 0) flexure reléate. ‘Temperature strains and stresses in simply siupported span Lotked in. Released": stresses’ stress" G (al ef Der in Fig, 11.3(b), the deck expands and the compression stresses of Fig. 11.2(¢) are relaxed. Only the moment and residual stresses of Fig. 11.2(f) remain, When the moment restraint is removed asin Fig. 11.3(6) the deck flexes. In effect, the restraining moments have been cancelled by equal and opposite relaxing moments which cause flexure. The magnitudes of the restraining (and opposite relaxing) moments are equal to the moment of the stress diagram of Fig. 11.2(f) (about any point). 7 ‘ : M= f obzdz. (1.4) b x 0.198 x 0.064 ~ 92: x 0,007 «0.198 = (0241.8) 0.4 x 0.45 — 1.0 + 0.14 x 0.95 2 = —0.24MNmm"* width, ‘The stresses and strains caused by the relaxing moment are related by Hooke’s law and exhibit the clastic behaviour described in Chapters 2226 Temperature and prestress loading | Fig. 11.4 Temperature moment and residual stresses: a) restrained ‘moment My and residual (same as Fig. 11.2(0)5 (b) relaxing moment Mas and (€) residual. somun! —-bannt ane 7 | =e -3 x | ta 1 ‘a | and 3, In particular, the stress distribution is linear as shown in Fig. 11.4(b) with stress related to M by on Xt, (1.6) For a rectangular beam Z = bd"/6, and hence the stress relaxation at surfaces in Fig. 11.4(b) due to M= ~0.24 is stresses of (a), we obtain the residual stress distribution of (c). This has no-net compression or moment, as can be checked by reapplying equations (11.3) and (11.4). “The residual stress diagram Fig. 11.4(c) shows the final distribution of temperatuté-induced stresses on a cross-section of a simply supported span, The maximum stresses are large, even though the deck is simply supported, because the temperature disiribution is so non-linear. Itis only when the temperature distribution is linear that its induced stresses are linear in (a) so that when relaxed by an equal and opposite linedr stress distribution of (b), no residual stresses (c) remain on the section, ‘At the end faces of the deck, the residual temperature stresses are not resisted by external forces, Consequently, these forces redistribute by local elastic distortion over a longth of deck approximately equal to the depth of the section; This redistribution is accompanied by local high Jongitudinal shear forces which tratisfer the residual compression forces near the top and bottom faces to the opposed residual tension in the middle. In this example, the residual compression force below level Xin Fig. 11.4(0) is I 2. . FX 05. = 0.3MN m™ width. | | I ‘When the relaxing stress distribution of (b) is added to the restygined | I l l ‘Thus the longitudinal shear force at level X near the end of the deck is 1 \ also 0.3 MN m™ width,h at ‘Temperature stresses in a continuous déck 227: Fig. 1L5 Release of ‘temperature moments in continuous deck: (@) restrained deck; {b) restrained moments ineach span; (@) restrained moments, in connected spans; (@) relaxing flexural ‘moments; and (e) final inoments = (6) + (€).. = A (ol Inthe above example it is assumed that the deck has thesame breadth at all levels. Ifthe deck is made up of beams with breadth b varying with depth, the variation in b is included in’ equations (11.3) and (11.4). Further worked examples of calculations of temperature stresses are- included in references [3}-[6]. ‘The preceding discussion has assumed that the bridge material is able to transmit compression and tension forces at all levels within the cross section. This snot the case for reinforced concrete which is assumed foi the purposes of design to have cracks in regions subject to tension, ‘These cracks are generally advantageous with respect to temperature effects because they can accommodate the thermal strains: without: developing stresses. References [3] and [4] explain the calculation, procédure which follows the principles of this chapter. 11.3 TEMPERATURE STRESSES IN A CONTINUOUS DECK _ The relaxation of temperature stresses in a continuous déck is only Slightly more complicated than in a simply supported span. Figure 11.5(a) shows a three-spani deck which is fully restrained with a locked in non-linear temperature distribution. an228 ‘Temperature and prestress loading ig. 11.6 Temperature stresses at internal support of continuous deck of Fig, 1.5: (@) restrained moment ‘Mand residual (Same as Fig. 11.200) () selaxing moment = 0.2 M5 (6 total steoses. | ‘The compression force due to average temperature compressior siress of Fig. 11,2(¢) is relaxed by letting all the spans expand freely ir the longitudinal direction. Figure: 11.5(b) -shows the restrainec temperature moments on each span similar to Fig. 11.3. Since the moments on the two sides of the internal supports balance, itis possible to join the spans together asin (c) without restraints at internal support: and without any relaxation taking place. In other words, only moment: at the ends of the deck are needed to restrain a continuous deck agains temperature flexure. Relaxation of these end restraining moments i achieved by superposing equal and opposite relaxing moments asshow1 in Fig, 11,5(€). In contrast to a simply supported span which hogs wit! uniform moment when subjected to eid relaxing moments, th: continuous deck distributes the relaxing moments in accordance wit! the continuous beam theory of Chapter 2 as shown in Fig, 11.5(d) Consequently, superposition of the stresses from the restrainec temperature moments of (¢), which are uniform along the deck, and th varying relaxing moments of (d) give different stress distributions a different points along the deck. 4 ‘At the ends of the end spans, the relaxing moment is equal anv opposite to the restrained temperature moment, and the stres distribution is similar to Fig. 11.4(c). However at the next support, th: ‘relaxing moment in Fig. 11.5(d) has an opposite sign to thatat the end and so doesnot counterbalance the restrained temperature moment an in fact increases the combined top stresses as shown in Fig. 11.6. If the deck has a section that varies along its length or if th temperature distribution changes along the deck, the restraine temperature morent also changes along the deck. Figure 11.7 show two alternative thethods of representing a haunched deck. If the deck i represented by a number of connected uniform segments of differen section as on the left side of (b), the temperature flexure will b restrained by different moments in each segment, When the segment are connected as in (c), the moments do not balance at intern: connections. The relaxing moments applied to the continuous dec must cancel the external restrained temperature moments, and sq ou! | |ws. is int ts jal r " Fig. 117 Release of temperature moments in haunched deck; (2) deck elevation; (©) restrained lemperatur® forcés.on each span (0) restrained temperature fotees on connected épans;and (@) relaxing flexural, forces. Grillage analysis of temperature moments .229 a | ; 5 Cc) cee cr =D pep eas ¥ ta) of-balance moments must be applied at changes of section of continuous beam as in (c). Alternatively, if a segment is tapered, the restrained temperature moments are different on its two ends, To maintain®: equilibrium, opposed vertical forces must be applied to the two ends of the segment, as on the right of (b). (This shear force is statically equivalent to a uniformly distributed moment applied along the taper.) When the segments are conriected, the restrained tempetature,”, moments now balance at the conections but the vertical restraining. forces remain. As a consequence, the cancelling relaxing forces applied to the continuous beam for distribution include vertical forces at the, ends of tapers as well as moments at the end supports. 7 Wide bridges experience stresses duc to the thermal expatision of a element being resisted transversely as well as longitudinally. The. behaviour is the same in the two directions. However if Poisson’s ratiois significant, it may be necessary to investigate the interaction of stresses, and flexure in the two directions. This is best done with a finite element * analysis in which the interaction is taken account of automatically, 11.4 GRILLAGE ANALYSIS OF TEMPERATURE MOMENTS: : ‘The analysis of distribution of temperature moments in a two- dimensional deck by grillage is the same in principle as that for !#Project OMG RN /e Page). Procter GAA, Fie: Designed By: Git Date 474 [iy Checked Bye Date: coun Reference Document / Drawing Design Description; TEMPERATURE CALE 150] 20] 1s(Fae Poa Designed By: Date: Checked By: Date: Spex Reference Dooumnent/ Drawing Design Deserpton:Project Page S Project No: File: G Designed By. Dale: Checked By Bate: GoBA Reference Document Drawing Design Deserpton: |. L oeProce Pager Project Nar Fe G Designee By: Date Checked By ate: copa Reference Document / Drawing Design Description: