Strengthening of a Long Span Prestressed Segmental Box

Bruno Massicotte The Grand-Mere Bridge in the province of Ph.D., P.Eng. Quebec, Canada, is a 285 m (935 ft) long, Associate Professor cast-in-place, segmental bridge that Department of Civil Engineering Ecole Polytechnique de Montreal experienced several problems which resulted Montreal , Quebec, Canada in distress characterized by an increasing de­ Formerly, Bridge Design Engineer at the Quebec Ministry of Transportation flection combined with localized cracking. These defects were due mainly to insufficient prestressing causing high tensile stresses in the deck and possible corrosion of the pre­ stressing steel. To remedy this situation, the Quebec Ministry of Transportation strength­ ened the bridge by adding external prestress­ Andre Picard ing equivalent to 30 percent of the remaining Ph.D., P.Eng. internal prestressing. The paper describes the Professor Department of Civil Engineering causes of the distress and focuses on the as­ Universite Laval sumptions adopted in the analyses to deter­ Quebec City, Quebec, Canada mine the current state of the bridge. The tech­ Yvon Gaumond, P.Eng. nique and design criteria used in strengthening Bridge Design Chief Engineer the Grand-Mere Bridge are described. Also, Bridge Department the construction aspects and the various prob­ Quebec Ministry of Transportation lems met during the external prestressing op­ Quebec City, Quebec, Canada eration are discussed. The new technology and experience gained in strengthening this structure can be applied to both pretensioned and post-tensioned .

he Grand-Mere Bridge, a 285 m (935 ft) long, cast-in­ Claude Ouellet, P.Eng. place, post-tensioned segmental structure built in 1977 (Fig. 1), consists of a single-cell box girder. It is Senior Bridge Engineer T Bridge Department located 200 km (125 miles) northeast of Montreal on High­ Quebec Ministry of Transportation way 55, where it crosses the St. Maurice River near the Quebec City, Quebec, Canada town of Grand-Mere.

52 PCI JOURNAL An article on the design and con­ struction of the Grand-Mere Bridge appeared in the January-February 1979 PCI JOURNAV and the project won a Special Jury Award in the 1978 PCI Awards Program. This long and slender bridge is an elegant structure which blends in well with the sur­ rounding landscape. Unfortunately, this bridge experi­ enced various problems and distress during and after its construction, result­ ing in localized cracking and increasing deflection in the 181.4 m (595 ft) cen­ tral span. These problems were due mainly to insufficient prestressing as a result of construction problems, opti­ mistic design assumptions, and a lim­ ited state of knowledge at that time, es­ pecially regarding thermal stresses. Fig. 1. The Grand-Mere Bridge. Numerous studies showed that the short-term bridge safety was adequate. However, because of the potential risk SCOPE OF THE PAPER tant monitoring program carried out of cracking in the deck, the studies on the bridge during the strengthening also indicated that the long-term in­ This paper is the first of two com­ process. This program studied the var­ tegrity could be affected if corrective panion papers describing the correc­ ious aspects of the bridge's behavior measures were not taken immediately. tive measures applied to the Grand­ before, during and after the strength­ Based on these technical evaluations Mere Bridge. The scope of this first ening operation. and considering the structure's impor­ paper is to present the history of the tance, the owner, the Quebec Ministry bridge, describe the various problems of Transportation (QMT), decided to met during construction, and discuss BRIDGE DESCRIPTION strengthen the bridge. Additional lon­ the subsequent distresses, their causes The Grand-Mere Bridge, a 285 m gitudinal prestressing corresponding to and their remedies. It presents the as­ (935 ft) long structure, consists of 30 percent of the remaining amount sumptions used in the analyses to es­ three continuous spans of 39.6, 181.4 corrected the lack of sufficient pre­ tablish the bridge state and later to de­ and 39.6 m (130, 595 and 130ft), stressing. The strengthening design, in sign the strengthening program. The completed with a wedge-shaped, solid which the authors of this paper played use of individually lubricated strands cantilever of 12.2 m (40ft) at each end an active role, was done by the QMT and details of the cable anchorage sys­ acting as counterweights (Fig. 2). The Bridge Department. tem to the existing structure are de­ length of the central span was a re­ Construction began in May 1991 scribed. The construction aspects and markable achievement in 1977, and is and the additional prestressing was fi­ problems faced during the strengthen­ still among the longest for this kind of nally applied to the bridge in Novem­ ing process are also discussed. The bridge in North America. T~e bridge ber of the same year. The technology paper deals with practical considera­ is an elegant, slender structure span­ used in the strengthening of the bridge tions and is addressed to bridge engi­ ning the St. Maurice River. can be applied to all prestressed con­ neers who may face similar problems. In the central span, the depth of the crete structures, whether pretensioned The companion paper/ on the other box girder varies parabolically from or post-tensioned. hand, presents the details of an impor- 9.75 m (32 ft) at the piers to 2.90 m

Exterior pier [m] 1 m = 3.28 ft

Fig. 2. Longitudinal elevation of the Grand-Mere Bridge.

May-June 1994 53 cantilever construction proceeded. The box girder was prestressed 4 ----12 800------+1:1 142060 using 32 mrn (1.25 in.) diameter pre­ r stressing bars with 1030 MPa (150 ··~ ...,,l, .., .., .., ., .., .,..,. , .. , .., '\~/", ...... ~ ksi) ultimate stress. The bridge was prestressed longitudinally during the cantilever construction with 284 straight bars located in the upper deck. The continuity prestressing comprises 80 slightly curved bars in the bottom 0 LO flange and 48 draped bars in the webs {61 0 end spans ...... :~~ :i:~ 0) ::::::: 356 central span ::::::: (Fig. 4). Shear reinforcement consists of [mm) straight prestressing bars, either verti­ ~~~~~ ~ ~~.i cal or inclined, and of conventional 152 --~ 1000 mm =1m mild reinforcing steel. The deck is pre­ 1 m = 39.37 ln. stressed transversely with straight ; = 3.28 ft bars. Moreover, passive reinforcement ~ fFJ!!:{~/'t/Ji of each segment was extended through adjacent segment joints. The specified concrete strength was 34 MPa (5000 psi), except in the middle portion of 11) Cross section Ill the interior piers the central span, where 38 MPa (5500 psi) concrete was specified.

INITIAL PROBLEMS AND SUBSEQUENT DISTRESS The bridge had experienced several [mm] problems originating from three sources: construction problems, design assumptions, and the limited state of knowledge at that time. The bridge de­ b) Cross section at midspan sign included two distinguishing fea­ tures: the slenderness of the 181.4 m Fig. 3. Cross sections of the Grand-Mere Bridge at interior piers and midspan. (595 ft) central span and the exclusive use of straight and curved prestressing bars for the longitudinal and trans­ (9.5 ft) at midspan (Fig. 3), giving piers as a consequence of the unusual verse prestressing. span-to-depth ratios of 18.6 and 62.6, center-to-end span ratio of 4.58. respectively. The depth of the box girder in the end spans varies slightly, Construction Problems Construction and Materials from 9.75 m (32ft) at the interior piers Construction problems originated to 8.53 m (28 ft) at the exterior piers. The bridge construction lasted two from three sources: coupling of nu­ The total width of the deck is 12.8 m years, from 1976 to 1977, interrupted merous bars, concreting, and grouting (42 ft), including a 6.7 m (22 ft) wide during the winter season to avoid cold the prestressing bars. top flange of the single-cell box girder weather concreting. The end spans First, to join 10 to 15 m (33 to 50ft) and two 3.05 m (10 ft) cantilevers. were erected the first year and were long bars over a length of up to 100 m This torsionally stiff box cast on scaffolding. The central span (330 ft), the couplers used were some­ has only 1.83 m (6 ft) thick di­ was built in 1977, using the progres­ times unevenly screwed on two adja­ aphragms over the main piers. sive cantilever method with traveling cent bars, causing some to fail. Dam­ The designer planned to place 930 m3 forms for the cast-in-place segments. aged ducts, due to insufficient care (1216 cu yd) of gravel ballast in the After construction of the two segmen­ during concreting, restricted the free chambers of both end spans. The bal­ tal cantilevers of the central span, a sliding of bars and couplers during lasted chambers and the wedge-shaped, keystone segment was cast and the stressing. To overcome these two solid cantilevers would counterbalance continuity prestressing in the bottom problems, holes had to be made in the the weight of the main span during the flange and webs was added. Gravel top flange to replace or free some cou­ cantilever construction and avoid any ball ast in the end spans was applied plers and allow adequate prestressing. uplift of the box girder at the exterior progressively, in three stages, as the These necessary corrective measures

54 PCI JOURNAL 300r------~ Cantilever construction 250 prestressing bars

200 0 150 z0 100 284 top flange bars

50

~ 100r------, jg Continuity prestressing bars 0 50 48 draped web bars ~ ~ ~~(~~---~ ~'-rr--...... ----- .-::::::::::::=__ -~ 0 ~ ) Longitudinal ct Transverse p bars 1 bars (web lti..bars) : / -r j: (Long: bars) ! t \: p I Not to scale (bottom flange bars)

Cross section at interior pier

Bottom flange Web bars bars" I(! r Cross section at midspan

Fig. 4. Ori ginal prestressing layout of bridge. reduced the concrete quality in por­ top deck, eliminating the need for up to 20 percent of the top bars re­ tions of the top deck. gravel ballast at the west end. How­ mained virtually unprotected against Second, concrete casting problems ever, this solution required additional corrosion problems in the longer term. in the webs of the ballast chamber on prestressing and thickening of the bot­ Although the construction technique the west side forced the designer to tom flange (Fig. 5) in the end span. that was used was adequate in princi­ modify his original design due to Finally, duct injection of grout after ple, the various construction problems weakened walls. The retained option prestressing was completely achieved reduced the durability potential of the consisted of widening the wedge­ with certainty only on 80 percent of structure, and the required quality shaped cantilever span acting as coun­ the bars, the remainder being partially level for this type of construction was terweight with concrete, flush to the filled or not injected at all. Therefore, not reached.

May-June 1994 55 Widening of the cantilever (see detail A)

Thickening of the bottom flange West side East side

Detail A Additional concrete

Fig. 5. Alternative scheme for west side of bridge.

200

- :r - 6 I I / ~ - - ~ 100 " Approximate deflection after closure 3 , ..--- (as per the as-built drawings) "":"' -E c::::: E 0 0 ;::.. \ - c::::: c::::: cu Q. !. (I) ,en -3 , ·e -100 ·e 1ii 1U -6 c::::: c::::: 0 0 -200 ts -~ Strengthening C1) -9 =C1) cai 0 0

-300 -12

-15 ~ ~~~--~--~~--~~--~--L-_L __L_~--~--L-~~L__L __L_~__J 1975 1980 1985 1990 1995

Years

Fig. 6. Midspan deflections of bridge over time.

Design Assumptions original design. At that time, the val­ measured on short bars and used in the ues recommended by the Canadian design6 does not reflect the actual be­ Some of the problems in this bridge Bridge Design Code3 were 0.0026 per havior of long bars, which contain up were due to the adoption of overly op­ m (0.0008 per ft) and 0.30 per radian, to eight couplers in some cases. timistic design assumptions. Wobble respectively, very close to the pre­ A more appropriate value of 193000 and curvature coefficients of 0.0007 scribed values by CEB-FIP' and MPa (28,000 ksi)l-' would be recom­ per m (0.0002 per ft) and 0.30 per ra­ AASHT0. 5 Moreover, the elastic mended in such a case. If the theoreti­ dian, respectively, were used in the modulus of 213000 MPa (30,892 ksi) cal curvature coefficient is assumed to

56 PCI JOURNAL be adequate, the calculated wobble co­ efficients, according to measurements 5 during construction,6 were approxi­ Maximum allowable 500 stress In tension 'll mately 0.0053 perm (0.0016 per ft) in ------the top and bottom flanges and 0.0145 I? 0 0 a.. per m (0.0042 per ft) for the web pre­ ~ ! stressing bars. These values are, re­ -500 U) U) spectively, 7 to 20 times larger than U) -5 .... _ U) the design value. ~ - ~ -U) -1000 1ii a; a; c: c: State of Knowledge '5 -10 -1500 '5 ::J ::J .t: Minimum stress ~ The knowledge related to long span, C) envelopes C) c: c: 0 -2000 0 segmental bridges ....I ------\ ....I was limited or not yet published at that -15 Maximum allowable ;t '------stress in compression time. The length- to-midspan depth -2500 ratio of 62.6 for the main span is sig­ nificantly higher than a more recent -~~------~ maximum value of 50 recommended by Podolny and Muller.7 ~,o------~------~ The most slender structure in Eu­ ~ rope, as reported by Mathivat,8 has a slenderness ratio of 47. Although the Fig. 7. Stress variati on in top flange. slenderness at midspan can theoreti­ cally reach values up to 60, Mathivat recommends a much smaller maxi­ Maximum allowable Excessive tensile mum value to reduce the creep and 500 thermal gradient effects. s~~i~t~s~n-~-- __ ~- _ The Post-Tensioning Institute (PTI) 0 suggests a span-to-depth ratio at Before 9 midspan of about 42. It is obvious strengthening -500 that the 2.90 m (9.5 ft) bridge depth at midspan was too small and a value of a; -1000 at least 3.70 m (12 ft) would have c: been more appropriate. '5 .~ -10 -1500 An important design consideration c:0 of concrete box concerns .9 stresses induced by thermal gradients -2000 -15 between the top and bottom slabs. Lit­ tle was known about thermal stresses Minimum stress -2500 in 1976 and they were not accounted envelopes for in the bridge design. For the -~~------~ Grand-Mere Bridge, a linear thermal gradient of 10°C (l8°F) creates a posi­ ~,------~------tive bending moment at midspan of the same order of magnitude as a full Fig. 8. Stress variation in bottom flange. two-lane live load on the central span. Such loading is included in modern box girder bridge design. ciently abnormal to proceed with ex­ were fine, their width being about 0.1 tensive studies. to 0.2 mrn (0.004 to 0.008 in). Despite this unusual deflection, a Wider transverse cracks were found Observed Distress careful inspection of the bridge in 1988 in the top slab of the east end span, Shortly after its completion, the cen­ did not show any evidence of signifi­ with some traces of chloride efflores­ tral span of the bridge underwent un­ cant distress and cracking was ob­ cence. In 1985, coring of concrete in expected deflection. Measurements served in only two areas. At the third various locations on the top and bot­ were then taken regularly, and, by points of the central span, tiny fish­ tom slabs indicated average compres­ 1986, the average midspan deflection, bone cracks were found on the internal sive strengths of 51 and 43 MPa (7400 fluctuating with seasonal temperature face of the bottom flange, at the loca­ and 6200 psi), respectively. The cored changes, had reached 300 mm (12 in.) tion where the unexpected deflection concrete showed little sign of deterio­ (Fig. 6). This was considered suffi- begins (Fig. 6). However, these cracks ration and the waterproofing mem-

May-June 1994 57 brane underneath the asphalt appeared by a = E1 I E; , varies between 20 and Quebec's Past Experience in to be in good condition, except along 30 percent. These values correspond Strengthening Long Span parapets. to creep factors of 4.0 and 2.3, respec­ Prestressed Concrete Bridges tively - not surprising for concrete In 1977, the Grand-Mere Bridge prestressed at an early age. Such high was the third long span prestressed Structural Safety values would be predicted by the 1970 concrete bridge built in Quebec, hav­ and Behavior CEB-FIP Model Code,' as reported in ing the longest span. The first one, the Cracking observed in some areas Ref. 11. Thus, the final state of stress, St. Adele Bridge, was, in 1964, the was not considered to impair the due to dead load and prestressing first North American cast-in-place bridge safety. The location of the first forces applied before closing the struc­ . 9 This bridge suf­ set of cracks coincides with the dead­ ture, can be obtained as the average fered from insufficient prestressing end anchorage of the bottom flange between the cantilever state and the causing excessive deflection and continuity prestressing bars (Fig. 4) continuous state with the following cracking. It was strengthened in 1988 7 and was caused by the prestressing equation: by external prestressing.13 Since then, force being transferred to the concrete. the bridge deflection rate and cracking The second set of cracks was at­ (JD = aCJcantilever + ( l - a) (Jcontinuous has stabilized. (l) tributed to differential shrinkage be­ The second one, the Lievre River tween the webs and the top flange of This equation takes into account the Bridge, built in 1967, was the first the box girder. 10 North American precast segmental effect of creep and allows subsequent Nonetheless, the excessive deflec­ 14 easier computation of stresses. In the bridge. This bridge was strengthened tion of the central span was a major various studies on the Grand-Mere in 1987 because insufficient prestress­ concern for the QMT authorities. Nu­ Bridge, different professional opin­ ing led to the opening of joints be­ merous analyses of this structure by ions were given about the amount of tween segments. In 1968, the joints the QMT and various independent creep that had already taken place in were not glued with epoxy as recom­ consulting firms concluded that struc­ 15 1986. Depending on the assumptions mended today. However, that bridge tural safety was not compromised in used in the time-dependent analysis to did not suffer from creep deflection as the short-term period. The most recent evaluate the current state of stress, the much as the Grand-Mere and the study6 showed that the prestress losses total creep in 1986 ranged between 65 St.Adele bridges, as expected for pre­ due to shrinkage, elastic shortening and 85 percent of the long-term value. cast segmental construction. The ade­ and relaxation were 18 percent larger Nevertheless, all the analyses were quate behavior of the bridge after than assumed in the original design. calibrated on the observed deflection strengthening indicates the success of The average final prestressing steel at that time and, except for the re­ the operation. stress was estimated to be 48 percent maining creep, they indicated, with a The last two studies on the Grand­ of the ultimate prestressing bar 6 6 sufficient degree of confidence, the Mere Bridge ·' indicated the need for strength (0.48fpu), whereas the design current state of stress before strength­ early corrective action on the bridge, value was 0.59fw ening. After these studies were com­ i.e., additional prestressing, before This significant reduction of pre­ pleted, the deflection increased by more distress develops. Based on the stressing force does not impair bridge more than 90 mm (3.5 in.) in four experience gained in the strengthening safety. However, it enhances creep ef­ years ( 1986 to 1990), showing that of the first two bridges, considering fects and it can affect bridge behavior creep was still occurring. the advice from experts indicating cor­ and safety in the longer term. The high rective actions, and due to its impor­ tensile stresses in the upper flange tance, the QMT decided to strengthen over the interior piers and in the bot­ Conclusions on the the Grand-Mere Bridge. tom flange at midspan (Figs. 7 and 8) Current State are sufficiently in excess of the allow­ able values to cause concrete cracking From the various studies conducted STRENGTHENING and probably lead to serious corrosion on the Grand-Mere Bridge, it can be PROGRAM problems in the future. Moreover, the concluded that the main source of increasing unusual deflection of the problems was the insufficient pre­ Design Assumptions central span, caused by reduced pre­ stressing due to the following factors Code provisions or guidelines for stressing, was not showing any sign of (in order of importance): the strengthening of existing bridges stabilization. 1. Overly optimistic design assump­ are not covered in any bridge code. Various groups conducted studies tions about friction coefficients and an The strengthening was based on re­ on the effect of creep using time-step underestimation of prestress losses. quirements of various recent bridge analyses and nonlinear models. Based 2. Construction problems which design codes such as the Canadian on calibration with measured deflec­ worsened the design unconservatism. Code 17 and the Ontario Code.' 8 The ex­ tion, they concluded that the ratio of 3. Lack of code specifications about pertise of the French Department of the final creep elastic modulus (E1) to thermal gradients and a maximum Transportation (SETRA) in the the initial elastic modulus (E;), given span-to-depth ratio. strengthening of segmental prestressed

58 PCI JOURNAL Table 1. Load cases and load factors 0.6 f pu 0.6 f pu considered in the strengthening design. 0.5 f pu Load case 1 Nature of loads

number D L Mmin L Mmax T p I I 1.0 1.0 ~lor------~------~ 2 1. 05 0.95 ~ 3 0.95 1. 05 4 1.0 1.0 1.0 Fig. 9. Stress variation in the original prestressing steel of the top flange. 5 1.0 1.0 1.0

6 1.0 1.0 -0.4 1.0 1. 0 0.8 1.0 concrete bridges was also considered the bridge. It varies between 0.20 and 7 1.0 in establishing the general concept. 0.40. Based on Refs. 6 and 16, these­ 8 1.0 -0.5 1.0 In the strengthening of existing lected DLA was 0.20 and 0.40 for the 9 1.0 1.0 1.0 structures, the assumptions determin­ negative and positive bending mo­ D = dead load effec ts ing the current state of stresses should ments, respectively. The lane load LMmi" = li ve load effec ts associated with the negati ve account for some degree of uncer­ was the governing loading for this bending moment LMma.< = li ve load effects associated with the positive tainty. In stress computations, a range bridge. It consists of a CS-600 loading bending moment of stress values was used rather than model, with each axle load reduced to T = temperature gradi ent effects specific values. Variable load factors 60 percent (with the corresponding P = prestressing force effects and load combinations were consid­ DLA), superimposed on a uniform ered. Such an approach could also be load of 12 kN/m (822 lb per ft) with a adopted for the design of new bridges DLA ofO.lO. Several loading cases were used in the considering construction problems and Thermal gradient - The most re­ analyses to determine the worst load­ some degree of uncertainty often met cent French regulation recommended ing conditions, as listed in Table 1. in long span, prestressed concrete a short-term linear thermal gradient of Allowable stresses - The allow­ bridges. The following assumptions l2°C (21.6°F) between the top and able tensile stresses are 0.25 and about the Grand Mere Bridge were bottom flanges and a reduced long­ R made. term thermal gradient of 6°C (10.8°F). 0.50 { t.,' , expressed in MPa (3.0 {!; State of stresses before strength­ The short-term gradient is applied and 6.0 -.J J;,' in psi), for severe and ening - Based on observed evidence, with dead load only, whereas the long­ normal exposure, respectively - the a linear effective stress varying from term gradient is combined with live first applying to the top slab and the 0.5fpu to 0.6fpu in longitudinal pre­ load. In the analysis for the strength­ second, to the bottom slab. These lim­ stressing steel contained in the top ening operation, either 100 or 50 per­ its are given in OHDBC-83,18 which al­ slab was assumed (Fig. 9). For the ad­ cent of the positive gradient [l2°C or lows some tensile stress in precom­ ditional 20 bars on the west side, an 6°C (21.6°F or 10.8°F)] was consid­ pressed joints if waterproofing is effective stress of 0.6fpu was assumed. ered for positive bending moment cal­ present, as in this case. The stresses The corresponding forces over the in­ culations. For negative bending mo­ were calculated considering the aver­ terior piers are equal to 117600 and ments, when a negative thermal age of the transformed and non-trans­ 127500 kN (26,440 and 28 ,670 kips) gradient occurs, minus half of these formed section properties (including or for the east and west sides, respec­ values were considered, giving gradi­ not including reinforcing steel at joints, tively. On the other hand, the effective ents equal to -6°C and -3°C ( -10.8° respectively). stress level in the continuity prestress­ and -5.4°F) for short- and long-term ing steel was taken as 0.5fP"' whereas occurrences, respectively. Strengthening Prestress for the vertical and inclined web bars, Load cases - According to the a value of 0.6fpu was considered. The CSA-S6,'7 a unique load combination Strengthening work constraints - stress redistribution due to creep, for (D + L + 0.8T, where D =dead load, L In strengthening this bridge, several loads acting on the structure in the =live load, and T =temperature) shall constraints had to be considered. cantilever stage, was considered using be considered as the governing condi­ First, access to the inside of the box Eq. (1) with values of a ranging from tion for stress computations at service girder was restricted to an existing 0.2 to 0.4. load level. Although such an approach 600 x 600 mm (2 x 2 ft) opening in Live load - For live load, the cur­ may be acceptable for the design of the bottom flange of the east side end rent CSA loading CS-600, 17 a four­ new bridges, its application becomes span. The possibility of cutting an ac­ axle 600 kN (67.4 ton) loading model, questionable for existing bridges cess of approximately 1 x 2 m (3 x 6 was used. The dynamic load al­ where the uncertainty about prestress­ ft) in the bottom flange of the west lowance (DLA), or impact factor, ap­ ing forces is larger. It was felt that a side end span was studied and al­ plied to the CS-600 loading is a func­ broader range of values should be as­ lowed. In the top flange, only local­ tion of the first natural frequency of sumed in the strengthening design. ized small holes, with diameters not

May-June 1994 59 mm (0.6 in.) diameter strands], each strand having a nominal section of 140 mm2 (0.217 sq. in.); 15S15 cables were used for Cables 1 and 2. Strands were stressed at 70 and 82 percent of their nominal ultimate strength, equal to 1860 MPa (270 ksi), for 12S15 and [m] 15S 15 cables, respectively. Notes: - Anchorage of Cables 6 and 8 at Blocks 1 1m = 3.28ft The 12S 15 cables transfer a pre­ - Anchorage of Cables 5 and 7 at Blocks 2 stressing force of 2190 kN (492 kips) - Anchorage of Cables 3 and 4 at Blocks 3 each, whereas the prestressing force is -Anchorage of Cables 1 and 2 at Blocks 4 (see also Fig. 11) 3205 kN (720 kips) for 15Sl5 cables. At the section over each interior pier, Fig. 10 . Strengthening cables and anchorage bl ocks. the total force added by the 16 strengthening cables is equal to 39100 kN (8790 kips), which corresponds to 31 and 33 percent of the prestressing force of the original design, at the west and east interior piers, respectively. Final prestressing- Creep, relax­ ation, friction and elastic losses were estimated at an average of 11 percent of the added prestressing. Creep losses for load application to an old concrete were assumed equal to 20 percent of [mm] the creep losses of a 28-day concrete, based on extrapolation of experimen­ tal data reported in Ref. 11, leading to 1000mm = 1m creep losses of 0.3 percent of the 1 m = 39.37 ln. =3.28ft added prestress. The elastic shortening of the existing internal prestressing bars due to the external prestressing Fig. 11. Location of strengthen ing cables inside cross section. produced losses of about 7 percent. Relaxation and friction losses were es­ exceeding 180 mm (7 in.), were northern industrial region of Quebec. timated at 2 percent each. drilled for pouring concrete. Prestressing cable layout - The Individually lubricated sheathed For practical purposes and aesthet­ strengthening operation was performed strands were used. All cables, made of ics, the prestressing cables were local­ using 32 cables, 16 from each end (Fig. either 12 or 15 strands, were inserted ized inside the box girder. Moreover, 10). These straight cables were placed in individual PVC ducts which were the bridge had to remain open, with just underneath the top slab near the later grouted prior to tensioning. minimum interference to traffic; the webs, eight on each side (Fig. 11). Ca­ Final str esses - Final stresses, bridge is on a main access road to a bles 3 to 8 are 12S 15 cables [twelve 15 computed with the above assumptions,

Cables 1 & 2 anchorage lengths 8.0 and 7.0

Anchorage chamber (proposed by the designer)

1 m - 3.2811

Fig. 13. Dead-end anchorages in solid trapezoidal Fig. 12. Schematic section of dead-end anchorage chamber. cantilevers.

60 PC I JOURNAL [mm]

Fig. 14. Anchorage device at dead end of each strand. were equal to 0.96 0"au and 1.08 0"au for ...... ······· ·· ······ . ······ .. the top and bottom slab, respectively, within the allowable limits in tension ~ ~: 1000 mm =·1 m and compression for most sections 1 m =39 .37 in. Section A-A along the bridge. In compression, the =3.28 ft allowable stress was exceeded by about 10 percent over a very limited area. Al­ Fig. 15. Schematic view of anchorage blocks and diaphragms. lowable stresses were computed based on original concrete design strength at 28 days. Final stress variations are to obtain a sound concrete mass. Un­ bars stressed at 412 MPa (60 ksi), 40 shown in Figs. 7 and 8 for the top and fortunately, this alternative delayed percent of their ultimate strength, were bottom slab, along with the correspond­ the strengthening operation by more used to fasten the blocks to the webs. ing values before strengthening. than two months but minimized traffic The webs at the block location were interference. chipped 25 mm (1 in.) deep before concreting the blocks. CABLE ANCHORAGES To compute the transverse prestress­ Anchorage Blocks ing force, the friction coefficient at the Dead-End Anchorages Inside the box girder, the 32 added interface was assumed equal to 1.0. The cable dead-end anchorages cables were anchored to the webs by Some details of the blocks and di­ were located at both ends of the means of 14 blocks distributed along aphragms are shown in Fig. 16. The bridge. In the QMT design, anchorage the bridge (Fig. 10). At each anchorage outside view of the Dywidag bars, be­ chambers (Fig. 12) were planned at section, the concrete blocks were fore they were protected by a 250 mm both ends. To minimize traffic distur­ linked by two diaphragms (Fig. 15) to (10 in.) concrete cover, is shown in bance, the contractor suggested an­ eliminate any bending moment in the Fig. 17. A significant amount of rein­ choring the cables in drilled holes, box girder webs. Tensile stresses pro­ forcement was used in the blocks. Re­ subsequently grouted, in the 12 m duced in the front diaphragm by the inforcing bars were placed in the ( 40 ft) long solid trapezoidal can­ couple due to the longitudinal pre­ forms in three main directions, giving tilevers (Fig. 13). This accepted alter­ stressing were eliminated by prestress­ a total reinforcement volume varying native required the removal of the ing the diaphragm with Dywidag bars from 3.0 percent in Block No. 1 (Fig. sheath and grease on all strands and running from one web to the other. The 18) to 2.4 percent in Block No.4. the use of a special anchorage device back diaphragm, subjected to a com­ In computing load transfer from the squeezed on each strand (Fig. 14) to pression force, was only reinforced. blocks to the webs, various assump­ increase bond. At cable tensioning, a total force of tions were made. The design was During drilling, it was found that 5315 kN (1195 kips), corresponding to · based on the shear-friction theory. the concrete of the solid cantilevers 0.85fP"' was transferred to the webs by Also, as a design check, the shear was in questionable condition. This each anchorage block, except for the force produced in the transverse Dy­ required additional grout injection central ones. Dywidag prestressing widag bars by the longitudinal pre-

May-June 1994 61 create additional tensile stresses in the web equal to 70 percent of the allow­ able limit. This figure was obtained by assuming a vertical dispersion angle of 45 degrees of the prestressing force in the anchorage block. Calculations also indicated that the stresses in the com­ pression struts and nodal zones were smaller than the maximum limits.

PRESTRESSING APPLICATION Prestressing Sequence Bridge strengthening work began in June 1991 and lasted until November 1991. Dead-end anchorage of the ca­ bles and slippage problems delayed the work. The prestressing was ap­ Fig. 16. Anchorage Blocks No. 3 at 27.4 m (90 ft) from interior support. plied strand by strand with a monos­ trandjack. To reduce as much as possible any lateral bending moment due to uneven prestressing in the bridge, pairs of Ca­ bles 8 to 3 were tensioned in the fol­ lowing sequence from anchorage Blocks No. 1 to No. 3 (see Fig. 10). Two cables were tensioned simultane­ ously, one on each side of the box girder. At each pair of anchorage blocks, one pair of cables was first tensioned at the west side of the bridge, followed by two pairs on the east side and, finally, the remaining pair on the west side. The prestressing progressed at the rate of four pairs of cables per day. The sequence is given in Table 2 for Cables 6 and 8. The tensioning sequence for Cables 1 and 2 at Blocks No. 4, at the center Fig. 17. Outside view of transverse Dywidag bars at anchorage block. of the bridge (Fig. 10), was different. There was no diaphragm between Blocks No. 4 for clearance reasons. stressing was computed assuming an Also, there were no Dywidag bars to T = C"' p e- w I 2- ew elastic force distribution between the L-t/2-(x/2)tan() (4) fasten the blocks to the webs. How­ bars, as one would do in shear-eccen­ ever, the anchorage blocks were bear­ tric bolted connections in steel struc­ In this truss model, it is assumed that ing longitudinally at their bottom tures. Finally, in the verification pro­ the overall bending moment induced against existing anchorage blocks in cess, the force induced in the webs by the eccentricity of the applied force the bottom flange used to anchor the and the corresponding bending mo­ with respect to the web centroidal axis continuity prestressing (Fig. 20). ments were obtained using the simpli­ is equilibrated solely by the additional The blocks were also bearing against fied truss model shown in Fig. 19. force, T, developed in the prestressing the top deck. Cables 1 and 2 were ten­ From this figure, by geometry and bars of the front wall and the corre­ sioned from both sides of each block equilibrium: sponding compression force in the simultaneously such that no horizontal back wall. Theoretically, the initial force was transmitted to the web. The w+bP -t x= ---"---- prestressing force in the front wall, Pb, tensioning was done alternately from 1 +tan() (2) should remain constant. one web to the other, for each pair of These verifications indicated that cables. To avoid too much uneven pre­ Mweb =P ew = P (w -x)/2 (3) the bending moment in the web would stressing, the cables were tensioned to

62 PCI JOURNAL 50 percent of their final tensioning force and then brought to their final value of 0.82 fpw The sequence for Ca­ bles I and 2 is presented in Table 3.

Duct Grouting and Prestress Losses Grouting cable ducts when individu­ ally lubricated sheathed strands are used is controversial. Although grout­ ing is not strictly required with exter­ nal prestressing cables located inside the box girder, it may be beneficial. Past experience'9 indicates that strand sheaths can be damaged under contact stresses at cable deviators. Grouting before tensioning reduces contact stresses and avoids sheath damage. Fig. 18. Reinforcing steel in anchorage block. However, for cables longer than 30 m (100ft), it is recommended to initially tension each strand at approximately 0.1 fpu before grouting so as to align the strands inside the cable duct. For the Grand-Mere Bridge, the contractor decided to grout the cable 1pb c ducts before tensioning. For perfectly straight cables, this would not have caused a problem. However, the loose Front Back wall wall 1 strands were not perfectly aligned in ------... ~ the ducts and some were probably zone twisted. It followed that friction losses Block were larger than expected for strands in individually lubricated sheaths. p As an example, the wobble coeffi­ cient for Cables I and 2, anchored at Nodal zone ~ Blocks No. 4, computed from the mea­ sured force and elongation, was equal p --~~ to 0.00124 perm (0.00038 per ft), close to 0.001 per m (0.0003 per ft) , a value suggested in Ref. 19. However, the curvature coefficient was evaluated Compression strut at 0.20 per radian, more than four times the value of the French regula­ tions.19 For all cables in this project, it would have been advisable to tension the strands at O.lfpu prior to grouting.

Construction Site p 1::--- The width of the bridge deck per­ _ w __ _ mitted two lanes of traffic to be kept • open for most of the construction pe­ riod, except during the curing of the anchorage blocks (Fig. 21). During the first four hours after concrete pouring, Detail of the nodal zone only one lane was open on the bridge, and escorting vehicles limited the bridge speed crossing to 15 krnlh (1 0 Fig. 19. Truss model at anchorage block. miles per hr).

May-June 1994 63 Table 2. Prestressing sequence at anchorage Blocks No. 1. CONCLUSIONS Side of I Side of the Prestressing force The experience and findings gained the bridge cross section Cable (percent of final value) f- from the Grand-Mere Bridge project West North and south 6 100 - from the construction to the East North and south 6 100 strengthening operation - are sum­ East North and south 8 100 marized here: West North and south 8 100 1. The span-to-depth ratio of box girder bridges at midspan should not exceed 50. 2. Thermal gradients should be con­ sidered in the design and analysis of Cables 1 & 2 tensioned symmetrically long span concrete bridges. from both sides of the block 3. Curved prestressing bars are not recommended. 4. Precast segmental construction, with epoxied joints, is preferable to cast-in-place construction because of a better control on materials and work­ manship, of reduced creep effects, and of higher concrete strength. Fig. 20. Anchorage Block No. 4 at midspan. 5. Cables more than 30 m (100 ft) long made of individually lubricated sheathed strands inserted in ducts Table 3. Prestressiing sequence at anchorage Blocks No.4. should be initially tensioned to 10 per­ Side of Side of the Prestressing force cent of their final tensioning force be­ the bridge cross section Cable (percent of final value) fore grouting. 6. The strand-by-strand symmetrical East and west North 1 50 application of the prestressing force East and west South I 50 worked satisfactorily for both construc­ East and west South 2 50 tion purposes and bridge behavior. East and west North 2 50 7. Dead anchorage in the wedge­ East and west North 1 100 shaped cantilever span was not as suc­ East and west South 1 100 cessful as expected. Questionable East and west South 2 100 concrete quality and slippage of ca­ East and west North 2 100 bles delayed the work significantly. However, traffic disruption was avoided. 8. Cast-in-place anchorage blocks did not show any sign of distress. The anchorage block-and-diaphragm as­ sembly and the transverse prestressing bars appeared to work efficiently. 9. The technology used for strength­ ening the Grand-Mere Bridge can be applied to both pretensioned and post­ tensioned concrete bridges, either for the strengthening of existing bridges or for the construction of new structures. 10. The objectives of the strengthen­ ing operation - that is, to create a more favorable state of stress and sta­ bilize the deflection - were achieved based on the first two-year deflection survey shown in Fig. 6. 11. The strengthening project, which cost $1.3 million, is expected to Fig. 21. Bridge strengthening operations, working from deck. extend the bridge's useful life.

64 PCI JOURNAL ACKNOWLEDGMENT 6. Dessau, "Expertise sur le comporte­ Developments in Short and Medium ment du pont en beton precontraint de Span Bridge Engineering '90, Bakht, The authors would like to thank the Grand-Mere construit par encorbelle­ B., Dorton, R. A. , and Jaeger, L. G. , authorities of the Quebec Ministry of ment" (Expertise on the Behavior of Editors, Third International Confer­ Transportation for their support dur­ the Grand-Mere Prestressed Concrete ence on Short and Medium Span ing this project. The financial support Bridge), Report 3571, Les Consul­ Bridges, V. 2, Toronto, Ontario, of the Natural Science and Engineer­ tants Dessau Inc., Montreal, Quebec, Canada, August, 1990, pp.259-269. ing Research Council of Canada and Canada, 1988 (in French). 14. Precast Segmental Box Girder the Quebec Fund for the Promotion of 7. Podolny, W., and Muller, J. M., Con­ Bridge Manual, Post-Tensioning In­ Research (Le Fonds FCAR) is also struction and Design of Prestressed stitute, Phoenix, AZ, and Precast/ Concrete Segmental Bridges, John Prestressed Concrete Institute, acknowledged. Wiley & Sons, New York, NY, 1982. Chicago, IL, 1978. 8. Mathivat, J., "Construction par encor­ 15. MacGregor, R. J. G. , Kreger, M. E., REFERENCES bellement des ponts en beton precon­ and Breen, J. E., "Strength and Duc­ traint" (Cantilever Construction of tility of a Three-Span Externally 1. "Quebec's Grand-Mere Bridge- Prestressed Concrete Bridges), Ey­ Post-Tensioned Segmental Box 935 ft Long Post-Tensioned Segmen­ rolles, Paris, 1979 (in French). Girder Bridge Model," ACI Special tal Structure," PCI JOURNAL, V. 24, 9. Post-Tensioned Box Girder Bridge Publication SP-120, American Con­ No. 1, 1979, pp. 94-99. Manual, Post-Tensioning Institute, crete Institute, Detroit, MI, 1990, 2. Massicotte, B., and Picard, A. , "Mon­ Phoenix, AZ, 1978. pp. 315-338. itoring of a Prestressed Segmental 10. Ouellet, C., and Gaumond, Y. , Le 16. Massicotte, B., "Verification du ren­ Box Girder Bridge During Strength­ renforcement du pont de Grand-Mere forcement du pont de Grand-Mere" ening," PCI JOURNAL, V. 39, No. 3, (Strengthening of the Grand-Mere (Verification of the Grand-Mere May-June 1994, pp. 66-80. Bridge), Proceedings of the Canadian Bridge Strengthening), Notes de cal­ 3. CSA-S6-74, Design of Highway Society for Civil Engineering Annual cui du Ministere des Transports du Bridges, Canadian Standards Associ­ Conference, Quebec City, Quebec, Quebec, Canada, 1990 (in French). ation, Willowdale, Ontario, Canada, Canada, V. I, pp. 363-372, May 1992 17. CAN/CSA-S6-88, Design of High­ 1974. (in French). way Bridges, Canadian Standards As­ 4. CEB-FIP, Model Code for Concrete 11. Neville, A.M., Dilger, W. H., and sociation, Willowdale, Ontario, Structures: CEB-FIP International Brooks, J. J., Creep of Plain and Canada, 1988. Recommendations, Second Edition, Structural Concrete, Construction 18. OHDBC-83, Ontario Highway Cornite Euro-International du Beton, Press, New York, NY, 1983. Bridge Design Code, Ontario Min­ Paris, 1970. 12. PTI Post-Tensioning Manual, Fifth istry of Transportation and Commu­ 5. AASHT0-73, Standard Specifica­ Edition, Post-Tensioning Institute, nications, Toronto, Ontario, Canada, tions for Highway Bridges, Eleventh Phoenix, AZ, 1990. 1983. Edition, American Association of 13. Ouellet, C., and Gaumond, Y., 19. Precontrainte Exterieure (External State Highway and Transportation "Strengthening of Two Prestressed Prestressing), SETRA (French DOT), Officials, Washington, D.C., 1973. Segmental Box-Girder Bridges," in Paris, France, 1990 (in French).

APPENDIX - NOTATION

C = additional axial force in fpu = ultimate tensile stress of elastic modulus (E;) back wall due to longitu­ prestressing bars or (Jail code allowable stress dinal prestressing force strands a cantilever longitudinal stresses DlA = dynamic load allowance p = longitudinal prestressing computed in cantilever factor (impact factor) force structure e, ew eccentricities of applied pb initial prestressing force (]continuous = longitudinal stresses prestressing force and in front wall computed in continuous web reaction, respec­ T additional axial force in structure tively front wall due to longitu- (JD = stresses due to dead load E; = initial elastic modulus dina! prestressing force and prestressing acting E1 = final elastic modulus ac­ a = ratio of final creep elas- on structure in cantilever counting for creep effects tic modulus (E1) to initial construction

May-June 1994 65