Control of Concrete Cracking in Bridges (2017)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

5 Cracks in reinforced concrete members can be classified into two main categories (Leonhardt 1977): • Cracks caused by externally applied loads, and • Cracks that occur independent of the loading conditions. Cracks caused by external loads are generally flexural and shear cracks and occur after the con- crete has hardened. Cracks independent of the loading condition include plastic shrinkage cracks, settlement cracks, drying shrinkage cracks, thermal cracks, and map or pattern cracks. Cracks may also be described by their orientation, such as longitudinal, transverse, diagonal, or random (Patnaik and Baah 2015). In the survey for this synthesis, agencies were asked what lessons they have learned about con- trolling concrete cracking. Several agencies mentioned the necessity of timely and proper water or moist curing. Several agencies also mentioned that drying shrinkage was a major source of cracking, and that modifications to the concrete constituent materials to reduce shrinkage are beneficial. Modi- fications that were mentioned included the use of a shrinkage-reducing admixture (SRA), includ- ing fibers in the concrete mix, limiting cement or paste content, and internal curing. However, one agency stated that “nothing will eliminate cracking in concrete.” Plastic shrinkage cracks Plastic shrinkage occurs near the surface of freshly placed concrete when moisture evaporates from the surface faster than it is replaced by bleed water [American Concrete Institute (ACI) Commit- tee 224 2007]. Plastic shrinkage cracking is more likely to occur under conditions that produce high evaporation rates, such as high air and concrete temperatures, low humidity, and high wind velocity over the concrete surface. In addition, concrete mixes with lower amounts of bleed water, such as those containing supplementary cementitious materials (SCMs), have a greater tendency to exhibit plastic shrinkage cracks (ACI Committee 224 2008). In bridges, plastic shrinkage cracks are most likely to occur in the decks because of the relatively large surface area compared with the thickness. Plastic shrinkage cracks can be unsightly but do not normally affect the structural performance of the concrete member. In most cases, the cracks do not penetrate the full depth of the member but may act as initiators for full-depth cracks (TRB 2006). Plastic shrinkage can be minimized through the use of fibers (Kosmatka and Wilson 2016). The most effective solution is to prevent such cracks from occurring by providing a saturated atmosphere over all exposed surfaces during the curing process. Thus, it is important to cover the top surface of the concrete with a moisture-proof cover as soon as concrete placement and finishing are complete. Plastic settlement cracks Plastic settlement cracks occur when concrete continues to consolidate under its own weight after initial placement, vibration, and finishing. The cracks are most likely to occur when the vertical settlement is restrained by horizontal reinforcing bars. Settlement cracking increases with larger bar sizes, higher concrete slump, and smaller concrete cover (Dakhil et al. 1975; Weyers et al. 1982; Babaie and Fouladgar 1997). chapter two tyPes and causes of concrete cracking in Bridges

6 The likelihood of settlement cracking can be reduced by proper vibration of the concrete, use of the lowest possible slump, and increasing the concrete cover in conventional concrete or using viscosity-modifying admixtures in self-consolidating concrete (SCC). Plastic settlement cracks can be reduced through the use of fibers (Kosmatka and Wilson 2016). As with plastic shrinkage cracks, it is best to prevent plastic settlement cracks by using proper construction procedures. autogenous shrinkage Autogenous shrinkage is a reduction in volume caused by the chemical process of hydration of cement. It is most prominent in concretes with water/cementitious materials (w/cm) ratios of less than about 0.40. Consequently, high-strength concretes may have large amounts of autogenous shrinkage. Autog- enous shrinkage can contribute to plastic shrinkage cracking (TRB 2006). Concretes susceptible to large amounts of autogenous shrinkage should be cured with external water for at least 7 days to minimize crack development (Kosmatka and Wilson 2016). drying shrinkage cracks Drying shrinkage is caused by the loss of moisture from the cement paste in the concrete. When the drying shrinkage is restrained by other components, tensile stresses develop in the shrinking concrete and can lead to cracking. Such is the case for concrete decks cast on steel or concrete beams. The mag- nitude of the tensile stress is influenced by many factors, including the amount and rate of shrinkage, the degree of restraint, the modulus of elasticity, and the amount of creep (ACI Committee 224 2007). The amount of shrinkage is influenced by the concrete constituent materials and the member size. Thin- ner members have more shrinkage and shrink at a faster rate than do thicker members. thermal cracks Temperature differences between different components in a concrete structure are caused by dif- ferent heats of hydration, different cooling rates, and ambient temperature changes. For example, a bridge deck will heat more quickly from sunshine than will the girders supporting the deck. When these temperature differences occur, tensile stresses result, which can lead to thermal cracks. Ther- mal cracks can also occur when the two components have the same temperature change but have different coefficients of thermal expansion. This rest of this chapter addresses the causes and types of cracking in hardened concrete for each of the major components of a bridge. cracking in Bridge decks Cracks in concrete bridge decks are generally characterized by their orientation with respect to the longitudinal axis of the bridge (ACI Committee 345 2011). The major types are transverse, longitu- dinal, diagonal, and random. Transverse cracks are illustrated in Figure 1. In addition, there is map cracking, which is also called pattern cracking or crazing. In hardened concrete, cracks form where the tensile stress in the concrete exceeds the tensile strength of the concrete. Tensile stresses are caused by applied loads such as vehicles or restraint to the length changes caused by shrinkage or temperature changes. The tensile strength of the concrete depends on the concrete constituent materials, curing environment, and concrete age. Some cracking in nonprestressed concrete bridge decks is inevitable because the cross section is expected to crack before the reinforcement becomes effective. Such cracks include the following: • Transverse cracks over intermediate supports caused by negative moments from dead and live loads; • Diagonal cracks caused by torsional forces in the acute corners of skew bridges, as illustrated in Figure 2;

7 • Cracking in curved bridges caused by torsional forces; • Longitudinal cracks at the ends of spans, particularly where the bridge deck is integral with the abutment; and • Cracks at construction joints. Because some cracks are inevitable, their width and spacing need to be controlled through the use of reinforcement. This is discussed in chapter five. The incidence of cracking increases with span length [Larson et al. 1968; Axon et al. 1969; Portland Cement Association (PCA) 1970], angle of skew (Larson et al. 1968), and use of continuous structures (Axon et al. 1969; PCA 1970). full-depth, cast-in-Place concrete decks Cracking in concrete bridge decks is not a new phenomenon. In 1961, the PCA began a study of concrete bridge deck durability (PCA 1970). The study included a survey of 1,000 bridges selected FIGURE 1 Transverse cracking in a bridge deck (Courtesy: Michigan Department of Transportation). FIGURE 2 Diagonal cracking in an acute corner of a skewed bridge (Courtesy: Henry G. Russell, Inc.).

8 at random in eight states and a detailed survey of 70 bridges in four states. The study concluded that transverse cracking was the dominant type of cracking. These cracks typically were located above transverse reinforcing bars. Several other studies have identified that longitudinal and transverse cracks tend to fall directly above reinforcing bars in the top layer of reinforcement because the pres- ence of the reinforcement acts as a stress raiser, but this is not always the case (Cheng and Johnston 1985; Perfetti et al. 1985; Kochanski et al. 1990). The formation of these cracks can be initiated by the presence of plastic shrinkage cracks. A 1996 survey indicated that more than 100,000 bridge decks in the United States have experienced early-age transverse cracking (Brown et al. 2007). In the survey for this synthesis, agencies were asked which types of concrete cracking their bridges had experienced in the past 5 years in cast-in-place (CIP) concrete decks with removable formwork or stay-in-place steel forms on both steel and concrete beams. The results are summarized in Figure 3. On the positive side, nearly one quarter of the agencies reported that cracks occurred infrequently. On the negative side, more than one-half of the agencies reported that cracks occurred frequently. Fewer agencies reported frequent cracking when stay-in-place metal formwork was used compared with removable formwork. This could be the result of fewer states using stay-in-place formwork. The agencies were asked to identify the strategies they are using to minimize cracking in CIP con- crete bridge decks. Their responses are provided in Table 1 along with the results of surveys in 2012 (Russell 2013) and 2003 (Russell 2004). Based on this table, the strategies used by at least 85% of the agencies are to specify minimum concrete compressive strength, minimum concrete temperature at placement, maximum concrete tem- perature at placement, maximum w/cm ratio, maximum slump, and a minimum wet curing period. These strategies are the same as those that were rated highly in a 2012 survey (Russell 2013), as shown in Table 1. One difference between the 2012 and 2016 surveys is that more agencies now specify a maximum concrete temperature during curing. Agencies were asked to identify the strategies that were most or least effective in minimizing cracking in full-depth, CIP concrete decks. Although 23 strategies were listed as most effective, the strategy cited most often was to apply wet curing early and provide a minimum wet curing period for the deck, followed by the application of a curing compound. The strategy cited second most often was the use of fogging to reduce evaporation rates during concrete placement. The third most often FIGURE 3 Frequency of cracking in full-depth, CIP concrete bridge decks.

9 cited strategy was a reduction in the cement and paste content. The full list of strategies is provided in the answer to Question 7 in Appendix B. Fifteen strategies were identified as least effective in minimizing cracking in full-depth CIP concrete bridge decks. The strategy listed most often as least effective related to not requiring or enforcing the use of wet curing per the specifications. The strategies cited next most often were the use of fogging and the use of fibers. The full list of strategies is provided in the answer to Question 8 in Appendix B. In a 2012 survey for NCHRP Synthesis 441: High Performance Concrete Specifications and Prac- tices for Bridges (Russell 2013), state highway agencies identified that drying shrinkage cracking was a dominant issue in using high-performance concrete (HPC) in CIP bridge decks. It appeared that the use of HPC had not eliminated the concerns about deck cracking, although the use of HPC resulted in better performance overall. Individual agencies also reported that use of the following contributed to increased deck cracking: • High early strength concrete; • High-strength concrete; • Silica fume; • Larger cement content to produce lower permeability; • Fly ash; and • Evaporation retardant. One agency reported in 2012 that the use of an SRA had helped reduce cracking but not to a satis- factory degree. The responses did not show any consensus. The one practice that was not successful in three states was the use of silica fume. As a means of limiting the free shrinkage of concrete, the criteria listed in Table 2 have been used. Based on the numbers in Table 2, a reasonable criterion would be less than 300 to 350 millionths after 28 days of drying. Most of these criteria are based on tests using AASHTO T 160: Length Strategy to Minimize Bridge Deck Cracking 2016 Survey 2012Survey 2003 Survey No. % % % None 2 10 – – Specify minimum cementitious materials content 23 59 76 – Specify maximum cementitious materials content 20 51 54 33 Specify minimum concrete compressive strength 34 90 94 – Specify maximum concrete compressive strength 4 10 13 4 Specify a ratio between 7- and 28-day compressive strengths 3 8 16 – Specify minimum concrete temperature at placement 34 85 83 – Specify maximum concrete temperature at placement 37 90 94 80 Specify maximum concrete temperature during curing 23 61 30 – Specify maximum water-cementitious materials ratio 39 95 94 – Specify maximum slump 36 88 86 98 Specify maximum water content 20 53 42 – Specify the use of a shrinkage-reducing admixture 9 23 – – Specify the use of a shrinkage-compensating concrete 6 16 – – Specify the use of fibers 10 25 – – Require use of the ACI surface evaporation nomogram 25 63 55 – Require wind breaks during concrete placement 15 39 38 22 Require evaporation retardants 13 33 28 29 Specify internal curing 4 11 – – Require fogging during placement when evaporation rates are high 27 68 77 67 Specify a minimum wet curing period 40 95 100 – Other 11 48 54 – – = Strategy was not listed in the survey. TABLE 1 STRATEgIES TO MInIMIzE CRACKIng In CIP BRIDgE DECKS

10 Change of Hardened Hydraulic Cement Mortar and Concrete, but the initial curing period may be different. A low shrinkage alone does not guarantee that cracking will not occur. However, it does reduce the likelihood of cracking. In Kansas, 59 bridge decks were investigated to identify factors that contribute to cracking (Schmitt and Darwin 1995, 1999; Miller and Darwin 2000; Lindquist et al. 2005). The investigations concluded that concrete shrinkage or restraint of concrete shrinkage was a major contributor to bridge deck cracking. Additional details of the Kansas activities are provided in chapter seven. In 1994, new York State developed an HPC designated as Class HP (Streeter 1999). One of the goals of Class HP was to reduce cracking. The newly developed concrete was achieved by using fly ash and silica fume to reduce the cement content, lowering the w/cm ratio, and using normal range water-reducing admixtures. Streeter (1999) reported that cracking in the HPC bridge decks resulted for a variety of reasons. If there was not sufficient retardation during placement, cracks developed, primarily on multispan continuous structures. Shrinkage cracks occurred when curing was delayed or fresh concrete was placed on existing concrete that was not in a saturated, surface dry condition. This latter problem was prevented by placing soaker hoses or sprinklers on the existing concrete for 12 or more hours below concrete placement. To quantify the effects of the use of Class HP concrete, 84 bridge decks, built from 1995 to 1998, were inspected (Owens and Alampalli 1999). Deck ages ranged from 1 to 4 years. The results of the study showed that 49% of the inspected decks exhibited no cracking. Transverse cracking was found on 48% of the decks and longitudinal cracking on 44%. Forty percent of the bridge decks exhibited both transverse and longitudinal cracking. It was observed that most cracks occurred within 2 weeks of the deck placement. Visual inspections revealed that the HPC decks cracked with less frequency and exhibited narrower and shorter cracks than did their non–high-performance counterparts. Krauss and Rogalla (1996) monitored the temperatures and strains in the replacement deck of the Portland–Columbia Bridge between Pennsylvania and new Jersey. During the first 12 hours, the temperature of the deck climbed from 80°F to as high as 131°F from the heat of hydration. After 48 hours, the temperature rise had dissipated. The authors concluded that the differential temperature alone was not sufficient to cause deck cracking. However, significant cracking occurred after 18 to 41 days of air drying. A survey of 72 bridges for transverse deck cracking in the Minneapolis-St. Paul metropolitan area was reported by French et al. (1999). The survey included 34 simply supported, prestressed concrete girder bridges; 34 continuous, steel girder bridges; and four continuous, rolled steel wide-flange girder bridges. Overall, the decks of bridges with simply supported prestressed concrete girders were observed to be in better condition than decks on continuous steel girder bridges. This was attributed to reduced end restraint and the beneficial creep and shrinkage characteristics of the prestressed con- crete girders. The few prestressed concrete girder bridge decks that consistently performed poorly Shrinkage (millionths) Length of Drying Application Reference 300a 28 days Caltrans specifications Maggenti et al. (2013) 450 180 days Caltrans specifications Maggenti et al. (2013) 350 56 days Angeles Crest Bridge, Calif. Higareda (2010) 360 21 days SEAOC Class M Higareda (2010) 600 32 weeks Route 36 Highlands Bridge, N.J. Kolota (2011) 320 28 days WSDOT Gaines and Sheikhizadeh (2013) 700 4 months Bridge decks Purvis et al. (1995) 500 180 days Benicia-Martinez Bridge, Calif. Murugesh and Cormier (2007) 450 180 days San Francisco Oakland Bay Bridge, Calif. Brown (2007) 350 28 days Virginia DOT Nair et al. (2016b) 450 28 days Oregon DOT research Ideker et al. (2013) aLater changed to 320 millionths (0.032%). TABLE 2 SHRInKAgE REQUIREMEnTS

11 were either bridges with reconstructed or re-overlaid decks, or bridges that had decks placed during extreme temperature conditions. Cracking as a result of deck reconstruction was attributed to shrink- age of the deck being restrained by the aged prestressed concrete girders. For the steel girder bridges, end restraint and shrinkage were the most significant factors contrib- uting to deck cracking. The steel girder bridges had more cracking on interior spans than end spans, more cracking in curved bridges compared with straight bridges, more cracking with no. 6 bars than no. 5 bars as transverse reinforcement, and more cracking with increased restraint owing to steel configuration, girder depth, and closer girder spacing. A november 2002 multistate survey for the Michigan DOT showed that 30 (97%) of the 31 respond- ing states had detected early-age cracking in reinforced concrete bridge decks, and 25 (81%) of the states reported that the cracking was observed in the first few months (Aktan et al. 2003). Almost all states reported that transverse cracking was most prevalent. The Michigan research (Aktan et al. 2003) established the mechanism by which the cracks formed as thermal stresses that develop during the cement hydration process and, in most cases, during the first 12 to 24 hours after concrete placement. A temperature difference of approximately 20°F was established as sufficient to initiate cracking. An increase in drying shrinkage arising from delays in both concrete placement and application of wet curing also affected deck cracking. Based on a survey of 36 state transportation agencies, Fu et al. (2007) reported that most diagonal corner cracks in skewed bridges were observed in the first 3 months after bridge construction. The cracks generally occurred in the acute corners in a circumferential direction around the corner of the bridge, as illustrated in Figure 2. After inspection of 40 bridge decks in Michigan, instrumentation of two bridge decks, and finite element analysis, Fu et al. (2007) concluded that the main cause of corner cracking in skewed bridges was the temperature rise in the deck caused by the heat of hydration. Measured tempera- tures in two bridge decks had maximum values of approximately 30°F above the initial temperature of the concrete. Fu et al. recommended that additional reinforcement be used in the corner areas. The presence of cracks in newly constructed concrete bridge decks prompted the Colorado DOT to initiate a study and subsequently to introduce a Class H concrete for exposed bridge decks (Xi et al. 2003). The new concrete required a Type II cement, fly ash, and silica fume for a total cementitious materials content of between 518 lb/ft3 and 640 lb/ft3, air content between 5% and 8%, and a w/cm ratio between 0.38 and 0.42. In addition, the concrete had to not exhibit a crack at or before 14 days when tested in accordance with AASHTO PP 34 (now AASHTO T 334 2012): Standard Method of Test for Estimating the Cracking Tendency of Concrete. A subsequent article (Pott and Elkaissi 2009) reported that the Class H mixes had achieved the objective of less cracking, but at a cost. The capability of testing per AASHTO PP 34 was available at only two facilities in the state. This presented a challenge for the first two projects. The addition of new capabilities at other testing facilities reduced this challenge. Pott and Elkaissi (2009) also reported that completely crack-free bridge decks, curbs, and side- walks are difficult to obtain because of the restraint of drying shrinkage. Even with the elimination of negative moment cracking at the piers through alternative structural designs, shrinkage cracking still created challenges. HPC can mitigate this cracking but is only one component in making the deck system last 75 to 100 years. Secondary protection systems, such as corrosion protection of the reinforcement and bridge deck waterproofing systems, also are important components. Mokarem et al. (2009) reported on the inspection of 19 bridges in 14 states in which the concrete deck was constructed with HPC. The bridges had been in service for 5 to 10 years and were located in different climatic regions. A detailed crack survey of each bridge deck was made to document the number, length, and width of the transverse, diagonal, and longitudinal cracks. Using the crack survey data from each bridge, the lengths of the transverse, diagonal, and longi- tudinal cracks on each deck were calculated. The average crack lengths for all the bridge decks were

12 0.073 ft/ft2 transversely; 0.008 ft/ft2 diagonally; and 0.042 ft/ft2 longitudinally, indicating that most cracks were in the transverse direction. The results also indicated that, in some cases, the use of HPC reduced bridge deck cracking, whereas in other cases the crack lengths were greater. When the structural system of the bridge included skewed supports, diagonal cracks developed near the supports. When the structural system included continuity over the supports, transverse cracks developed in the negative moment regions. The 14 bridges that used precast, prestressed concrete beams with a full-depth CIP concrete deck exhibited a wide range of total crack densities. The bridges in georgia, Louisiana, new Hampshire (Route 104), and Tennessee (Hickman) had relatively low total crack densities. In contrast, the bridges in Alabama, nebraska, new Mexico, north Carolina, South Dakota, Tennessee (Porter), Virginia, and Washington had at least twice as much cracking. On average, the latter group of bridges had about eight times as much cracking as the former group. For most bridges, the highest crack density occurred for cracks running in the transverse direction. The exceptions were two bridges in Virginia, which are discussed later. The cracking densities in each span of each bridge were compared with span lengths, beam spacings, deck thickness, girder types, clear deck spans, and beam span-to-depth ratios in an attempt to identify any overall correlations. none were identified. However, some comparisons between crack densities on spans of individual bridges were relevant. The georgia bridge comprises four simply supported spans. The bridge exhibited different patterns of deck cracking in different spans. The eastbound span 3 and westbound span 2 showed little crack- ing compared with westbound span 3 and eastbound span 2. Some diagonal cracking perpendicular to the skewed diaphragms at the end of the spans was present. The Charenton Canal Bridge in Louisiana has five continuous spans with an average length of 73 ft. Each span consists of five Type III AASHTO girders evenly spaced at 10-ft centers. The CIP concrete deck is 8 in. thick. A visual inspection of the bridge 228 days after casting did not reveal any cracks (Bruce et al. 2001). A second visual inspection of the bridge deck, performed about 4 years after the bridge opened to traffic (Mokarem et al. 2009), revealed some cracks. Most of the cracks were located in the negative moment regions over the intermediate piers. A total of 46 transverse cracks were recorded on the bridge with a combined total crack length of 187.4 ft over a bridge deck area of 16,060 ft2. However, all were hairline cracks with a width less than 0.016 in. no diagonal or longitudinal cracks were observed. The authors reported that the structural system of the Charenton Canal Bridge is flexible compared with conventional bridges because of the wider beam spacing and longer span length. This relatively flexible structural system might have contributed to the development and widening of some cracks. The Route 104 bridge over the newfound River in Bristol, new Hampshire, is a single simple-span bridge that was completed in 1996 (Waszczuk and Juliano 1999). The superstructure consists of a 9-in.- thick CIP concrete deck supported on five precast, prestressed concrete girders spaced at 12.5 ft on center. Until 2000, only some hairline longitudinal flexural cracks over the girder lines were observed; no transverse or shrinkage cracks were found. A visual inspection of the bridge deck was performed in 2004 (Mokarem et al. 2009). Only two longitudinal cracks were recorded on the bridge with a com- bined total crack length of 10 ft over a bridge deck area of 3,218 ft2. The maximum crack width was 0.02 in. no transverse or diagonal cracks were observed. The concrete utilized 660 lb/yd3 of cementitious material, including 8% silica fume. The w/cm ratio was 0.39. The concrete compressive strength at 28 days ranged from 8.16 to 9.61 ksi. The modulus of elasticity ranged from 4,200 to 4,300 ksi. The authors commented that the modulus of elasticity was lower than expected. In Tennessee, two similar bridges were inspected. Both bridges were constructed in 2000 and consist of a two-span continuous structure with span lengths ranging from 139 to 159 ft. The superstructures

13 consist of 8¼-in.-thick concrete decks with stay-in-place forms on four precast concrete bulb-tee gird- ers at 8 ft 4 in. centers. The bridges have skew angles of 27 and 17.5 degrees. The Porter Road Bridge deck was cast in January 2000 when the ambient temperature at time of placement was 35°F to 40°F. Heaters were used on the Porter Road Bridge. The Hickman Road Bridge was cast in May 2000 when the ambient temperature at time of placement was 70°F. Both bridges had transverse and diagonal cracks and no longitudinal cracks. However, the total length of cracks on the Porter Road Bridge was 860 ft compared with 110 ft on the Hickman Road Bridge. These crack lengths corresponded to crack densities of 0.085 and 0.013 ft/ft2. Most of the cracks were transverse and on a line along the middle of the deck. Some diagonal cracks were present at the skewed abutments. The authors speculate that the larger number of cracks on the Porter Road Bridge was the result of differential temperatures between the heated deck and the cooler beams. Each of the two bridges in north Carolina consists of two pairs of continuous spans. Most of the cracking was in the transverse direction and occurred in the half of each span adjacent to the continuity connection over the pier. Two bridges in South Dakota were three-span continuous structures with a similar amount of cracking. Most of the cracking was in the transverse direction with some diagonal cracking at the skewed abutments. Two bridges in Virginia had simple spans and were the only two bridges with a full-depth CIP concrete deck on precast, prestressed concrete beams that had more longitudinal cracking than trans- verse cracking. The reason for this was unclear in that the structural system for these bridges is similar to that of the other 12 bridges with full-depth, CIP concrete decks. Patnaik and Baah (2015) studied the negative moment cracking behavior of 13 reinforced concrete continuous span slab bridges in Ohio and reported cracks as wide as 0.14 in. under dead load alone. The measured crack widths were reported to be more than 15 times the maximum limit of 0.007 in. recommended in ACI 224R-01 for bridge decks exposed to deicing salts (ACI Committee 224 2008). Based on an investigation into the early-age cracking of concrete bridge decks in California, Araiza et al. (2011) recommended the following changes to their specifications: • Replace the minimum cement content of 675 lb/yd3 with a maximum cement content of 600 lb/yd3. • Specify a maximum paste content of 27% by volume. • Specify a minimum compressive strength of 3.5 ksi at 28 days unless otherwise required for structural design. • Consider a maximum compressive strength of 4.5 ksi at 7 or 14 days. • Reduce the maximum shrinkage from 0.045% to 0.035% at 28 days. • Specify an air content of 6% to 8% irrespective of exposure content. • Avoid the use of silica fume. • Wet cure the deck for 14 days. • Apply a white curing membrane after the wet curing period. • Hold a pre-job conference with the contractor to discuss curing and cracking. Stringer and Burgueno (2012) inspected 16 bridge decks in Michigan. Transverse cracking was present in bridge decks supported on concrete beams or steel beams. Longitudinal cracking was present only in the decks on concrete beams. Cracking was more prevalent in the negative moment regions. The authors identified that the cause of deck cracking in jointless bridges was restrained concrete shrinkage caused by the end restraint conditions. They concluded that restrained shrinkage cracking in concrete decks of jointless bridges is unavoidable in steel and concrete beam bridges. The lowest amount of cracking was predicted for bridges with nonintegral abutments or low-shrinkage concrete mixes. Based on a limited review of bridges in Iowa and finite element modeling, Phares et al. (2015) concluded that longitudinal and diagonal cracking in the deck on an integral abutment bridge in Iowa

14 was caused by the restraint of the abutment and the temperature differences between the abutment and the deck. Although not likely to induce cracking, shrinkage of the deck concrete may have exac- erbated cracks that developed from thermal effects. Longitudinal and diagonal cracks were prevalent in integral abutment bridges but not as prevalent in bridges with stub abutments. Bridge width and skew had minimal effect on the bridge deck strain resulting from restrained thermal expansion. Pier type, girder type, girder spacing, and number of spans also appeared to have no influence on the level of restrained thermal expansion strain in the deck near the abutment. Based on research for the Indiana DOT, Frosch et al. (2002) recommended that the total amount of reinforcing steel, As, to control crack widths in concrete bridge decks should satisfy the following equation: ≥ ′6 (1)A ff As c y g where Ag = gross area of concrete section (in.2); As = area of reinforcement in the cross-section (in.2); f ′c = specified compressive strength of concrete (psi); and fy = specified yield strength of reinforcement (psi). The purpose for this quantity of reinforcement is to prevent yielding of the reinforcement that can result in uncontrolled crack growth. For a 4.0-ksi concrete compressive strength with a 60-ksi yield strength reinforcement, this requirement results in a reinforcement percentage of 0.63 or 0.61 in.2/ft in an 8-in.-thick deck. This compares with 0.27 in.2/ft and 0.18 in.2/ft for bottom and top layers, respec- tively, required by the empirical design procedure of the AASHTO LRFD Specifications (AASHTO 2016). Frosch et al. (2002) also recommended a maximum bar spacing of 6 in. to control crack widths. In contrast, the Texas DOT recently reduced the amount of reinforcement used in the top mat of reinforcement in concrete bridge decks 8 in. and 8.5 in. thick (Holt 2014). Previously, no. 5 bars at a 6-in. spacing were used in the transverse direction and no. 4 bars at a 9-in. spacing in the longitu- dinal direction. The new requirement is no. 4 bars at 9-in. spacing in both directions, supplemented with short no. 5 bars at 9-in. spacing in the overhang portions to ensure adequate strength for the overhang and forces on the traffic railings. The selection of a 9-in. spacing for the top mat was based on inspections and observations of in-service decks where adequate crack control was being obtained by using no. 4 bars at 9-in. spacing in the longitudinal direction. This is equivalent to 0.27 in.2/ft or 50% more than 0.18 in.2/ft required by the AASHTO specifications for top bars in the empirical design method. Riding et al. (2009) measured concrete and ambient temperatures during the casting of a bridge deck in Austin, Texas, in August 2006. The measured temperature at middepth of the deck ranged from about 85°F at time of concrete placement to a maximum of about 123°F. The measured tempera- tures were then used in the laboratory with restrained specimens to determine the induced concrete stresses in a variety of concrete mixes. The testing revealed that the early-age thermal stresses were reduced by as much as 50% when a coarse aggregate with a lower coefficient of thermal expansion was used. A simulation of casting the concrete late at night versus in the morning also reduced the tensile stresses. In 2011, the Washington State DOT introduced a performance-based specification to reduce cracking in their bridge decks (gaines and Sheikhizadeh 2013). Instead of placing requirements on minimum cementitious material content, the new mix has the following requirements: • 28-day concrete compressive strength not less than 4.0 ksi; • 28-day drying shrinkage of 320 millionths per AASHTO T 160; • 56-day rapid chloride permeability of 2,000 coulombs per AASHTO T 277; and • Scaling resistance with a visual rating less than or equal to 2 after 50 cycles per ASTM C672.

15 Although implementation of the new specifications reduced deck shrinkage cracking, minor cracking was still evident. This was attributed to the differential temperature between the deck peak hydration temperature and the ambient temperature. gaines and Sheikhizadeh (2013) indicated that strain caused by this temperature differential can be as much as 300 millionths. A differential tem- perature limited to 24°F maximum has resulted in a significant reduction in transverse deck cracking. The Washington state performance-based approach is discussed in chapter seven. In Pennsylvania, data from a field investigation of 203 bridge decks were used to identify factors that contribute to early-age cracking and to assess the effect of cracks on the long-term durability of bridge decks (Manafpour et al. 2016). The following main conclusions were reported: • Higher concrete compressive strengths resulted in higher crack densities. A maximum allow- able concrete strength of 4.0 or 5.0 ksi at 7 or 28 days, respectively, was suggested. • Lower total cementitious materials content and higher quantities of SCMs resulted in less cracking. A maximum cementitious materials content of 620 lb/yd3 and the use of SCMs to reduce heat of hydration were suggested. • Decks constructed with half-width procedures to allow one half of the bridge to remain open cracked four times more often than did decks constructed with the full-width cast at one time and the use of traffic detours. Lwin and Russell (2006) recommended the following practices to reduce cracking in concrete bridge decks: • Decrease the volume of water and cementitious paste consistent with achieving other properties. • Use the largest practical maximum size aggregate. • Use aggregates, when locally available, that result in a lower concrete shrinkage. • Use the smallest transverse bar size and minimum spacing that are practical. • Avoid high concrete compressive strengths. • Design the concrete mix to produce a low modulus of elasticity and high creep. • Implement surface evaporation requirements and use windbreaks and fogging equipment, when necessary, to minimize surface evaporation from fresh concrete. • Apply wet curing immediately after finishing and cure continuously for at least 7 days. • Apply a curing compound after the wet curing to slow the shrinkage and enhance the concrete properties. Partial-depth, Precast concrete Panels with a cast-in-Place topping This bridge deck system consists of precast concrete panels that span between the top flanges of adjacent steel or concrete beams. The panels generally are pretensioned in the direction of their span and initially supported on bearing strips along the beam flanges. Below casting the topping, it is important that the precast panel top surface be thoroughly cleaned and saturated with water without ponding. The CIP deck placed over the beams and panels forms the complete deck and makes a composite system of the CIP concrete deck, precast panels, and the bridge beams. The system is often used to accelerate bridge deck construction. In the survey for this synthesis, agencies were asked about the frequency of cracking in CIP con- crete decks on precast concrete deck panels. The results are summarized in Figure 4, which indicates that less than half of the agencies use partial-depth precast concrete panels and that cracking occurs slightly more frequently than infrequently. In Texas, this system has been used extensively where the design has been standardized. Lon- gitudinal cracking over the girders has been reported as the most significant problem associated with the use of partial-depth concrete panels because it can result in a reduction in deck stiffness over the girders, which could compromise the deck’s load-transfer mechanism (Merrill 2002). The cracks have been caused by insufficient support of the panels on the beams and shrinkage of the CIP concrete being restrained by the precast concrete panels. Shrinkage of the CIP deck concrete and panel restraint have also led to transverse cracking. The widths of longitudinal and transverse cracks

16 are controlled by the reinforcement in the CIP concrete. In comparative laboratory tests, Tsui et al. (1986) showed that a deck with partial-depth panels was stronger, stiffer, and more crack-resistant than a deck with full-depth, CIP concrete. As part of their inspection of 19 HPC bridges, Mokarem et al. (2009) reported the inspection of three bridges that used precast, prestressed concrete panels supporting a composite CIP concrete deck. The three bridges that included panels are the Route 3A Bridge in new Hampshire and the Louetta Road and San Angelo bridges in Texas. The new Hampshire Route 3A Bridge uses four longitudinal nE 1000 simple-span girders spaced at 11.5-ft centers and supporting 3.5-in.-thick precast concrete panels and a 5.5-in.-thick CIP com- posite concrete deck. Cracking in the main span of the bridge consisted of five cracks with a total length of 18.5 ft. This was a low amount of cracking and indicates that the use of precast concrete deck panels is not always a contributing factor in bridge deck cracking. The Texas Louetta Road Bridge consists of separate northbound and southbound structures. Both structures use precast, prestressed concrete U-beams supporting 3.5-in.-thick precast concrete deck panels and a 3.75-in.-thick composite CIP concrete deck. Beam spacing varies from 11.5 to 16.6 ft. The panels span between the two top flanges of individual beams as well as between the flanges of adjacent beams. The specified concrete strengths for the CIP decks on the northbound and southbound structures were 4.0 ksi at 28 days and 8.0 ksi at 28 days, respectively. Measured compressive strengths were about 5.7 ksi at 28 days for the northbound structure and about 9.1 ksi at 28 days for the southbound structure. Overall, both structures exhibited a similar and relatively high total cracking density, with the north- bound having less transverse cracking and more longitudinal cracking than the southbound bridge. Most of the longitudinal and transverse cracking appears to occur above the edges of the precast deck panels and occurs throughout the length of each span. Factors that contribute to this cracking could be shortening of the precast panels as a result of creep and differences in the coefficient of thermal expan- sion between the CIP concrete and the precast panel concrete. It appears that the different concrete strengths used in the two bridges did not play a significant role in the amount of cracking in the decks. The Texas San Angelo Bridge consists of separate eight-span eastbound and nine-span westbound structures. Both structures consist of precast, prestressed concrete I-beams supporting 4.0-in.-thick precast concrete deck panels and a 3.5-in.-thick composite CIP concrete deck. AASHTO Type IV beams are used for most spans with Texas Type B beams for two short spans. Beam spacing varies from 5.4 to 11.0 ft. For the eastbound structure, HPC was used for the beams, panels, and CIP con- crete deck, except for the CIP deck of spans 6 through 8. For the westbound structure, HPC was used only for the CIP deck of spans 1 through 5. FIGURE 4 Frequency of cracking with partial-depth, precast concrete panels.

17 Overall, both structures exhibited the largest total crack density of the 19 bridges included in the investigation (Mokarem et al. 2009). However, the total crack density in the eastbound structure was about 60% of that in the westbound structure. In both structures, about 65% of the cracking occurred in the transverse direction. Similar to the Louetta Road Bridge, the presence of the precast concrete panels influenced the location of the cracks. Of the 17 spans included in both structures, eastbound spans 1 through 4 exhibited the least total crack density. All four spans are rectangular in plan. Spans 1 through 3 have a constant beam length and spacing. Span 4 has a slightly variable beam length to accommodate a change in the roadway width and skew angle of the bents. Spans 2 through 4 are the longest three spans in the bridge. By contrast, east- bound spans 5 through 7 are shorter and have a larger change in beam spacing and length; in addition, there is a large skew at the end of span 7. Most of the cracking, which is transverse, occurred above the beams with the longer span lengths. In addition, span 7 is relatively short but has a large skew at one end and exhibited the second highest total crack density in all of the bridge spans inspected. Westbound span 9, which has a square abutment on one end and skew bent at the other, had the highest total crack density. These observations indicate that bridge geometry influences the amount of concrete cracking, particularly when the geometry results in torsional stresses. In Missouri, several bridge decks with partial-depth panels were observed to have spalling on the underside at the edges of the panels (Spraggs et al. 2012). A detailed investigation indicated that reflective cracking in the deck above the transverse joints had allowed water and chlorides to penetrate to the interface between the CIP concrete and the precast concrete panels. The water then penetrated into the precast panel causing corrosion of the prestressing strand located nearest to the panel joints. According to Kwon et al. (2014), approximately 200,000 ft2 of deck panels composed of 3,000 pieces measuring 8 ft by 8 ft by 4 in. are rejected annually in Texas because of cracking during fab- rication, handling, or transporting. The most common cracks form above and below the prestressing strands after strand release. The researchers determined that the likelihood of this type of cracking could be reduced by lowering the initial strand stress from 189 to 169 ksi. The weakness of the system is the reflective cracking that occurs and allows water and chlorides to penetrate into and below the deck. Research is needed to identify the primary factors causing the reflective cracking and ways to reduce or eliminate it. full-depth, Precast concrete Panels In the full-depth, precast concrete deck system, panels span transversely across several bridge beams, as shown in Figure 5. The length of a panel along the roadway is usually 8 to 12 ft. The width of a FIGURE 5 Erection of full-depth, full-width, precast concrete panels [Courtesy: ENTRAN PLC (now Stantec) and ASPIRE– The Concrete Bridge Magazine].

18 panel is usually the full width of the bridge, unless that makes the panel too long or heavy to ship or staged construction is used. The panels generally are pretensioned in the transverse direction and may be posttensioned in the longitudinal direction. The panels usually are made composite with the beams using studs or reinforcement as shear connectors. The shear connectors fit into pockets in the panels. grouting the pockets accomplishes the composite action [Precast/Prestressed Concrete Insti- tute (PCI) 2011]. The use of precast panels facilitates faster construction for deck replacements and new construction. The advantage of a full-depth panel system is the prestressing in two horizontal directions, which reduces the likelihood of cracking. For simple span bridges, the PCI State-of-the-Art report (PCI 2011) suggests a precompression across the transverse joint between panels of 0.250 ksi after all losses. For continuous spans, the amount is 0.300 to 0.850 ksi, depending on the magnitude of the negative moment. On nebraska’s Skyline Bridge (Fallaha et al. 2004), the precompression across the transverse joints from the longitudinal posttensioning was about 0.800 ksi. This allowed for no tension in the deck even in the negative moment regions over the piers. Swenty et al. (2014) recommended that the transverse joints have sufficient prestressing force across them to limit the tensile stress to a maximum of 3√f ′c psi under superimposed dead loads and live loads. Carter et al. (2007) reported that Wisconsin’s first full-depth precast panel bridge had no cracks of any kind 1 year after completion. Other installations are described in PCI (2011). In addition to the panels being designed to carry live load and superimposed dead loads, it is important that the stresses during lifting and shipping be checked to ensure that the panels do not crack under such loading conditions. In the survey for this synthesis, slightly less than half the agencies reported on the frequency of cracking in full-depth, precast concrete deck panels. The results are summarized in Figure 6, which indicates that cracks occur above the connections and at other locations. Apparently, the use of this system has not always eliminated deck cracking. A 2003 survey (Badie and Tadros 2008) identified that nine states had used full-depth panels in the previous 10 years. In the survey for this synthesis, 16 states provided information about cracking in full-depth panel systems. FIGURE 6 Frequency of cracking in full-depth, precast concrete deck panels. 30 25 20 15 10 5 0 No. of Agencies Ne ve r Inf req ue ntl y Fre qu en tly Alw ay s Un kn ow n No t A pp lic ab le

19 cracking in adjacent Box Beam Bridges and slaB Beam Bridges Adjacent box beam and slab beam bridges consist of precast, prestressed concrete beams that are placed next to each other (Russell 2009). Adjacent units generally are connected by a longitudinal grouted keyway. For box beams, transverse ties are incorporated to hold the beams together. The ties may be grouted or ungrouted and vary from a limited number of nontensioned threaded rods to sev- eral high-strength tendons posttensioned in multiple stages. A transverse diaphragm is provided at the location of each transverse tie (Russell 2009). Adjacent slab beam bridges are similar to adjacent box beam bridges, except the adjacent units are usually connected with reinforcement protruding from the beams into the joint region between the adjacent units. A noncomposite topping or a composite concrete slab may be added as the riding surface. The com- posite topping contains transverse and longitudinal reinforcement to control cracking and provides lateral transfer of shear between adjacent beams or slabs. According to a 2008 U.S. survey, approxi- mately two-thirds of the states use adjacent box beam construction (Russell 2009). As illustrated in Figure 7, longitudinal cracking occurs at the location of the joint between the adjacent box beams. The transverse ties and composite topping, if used, are provided to control this type of cracking. In the survey for this synthesis, agencies provided information about the frequency of cracking with this system. The results are shown in Figure 8. It appears that the frequency of cracking is about the same for a system with and without a CIP concrete deck. FIGURE 7 Longitudinal cracking in an adjacent box beam bridge (Courtesy: New York State DOT). FIGURE 8 Frequency of cracking in adjacent box beam bridges.

20 Ahlborn et al. (2005) reported on an inspection of 15 box beam bridges in Michigan. Longitudinal deck cracking reflecting from the shear keys was identified as the leading cause of distress. Similar longitudinal cracking has been observed in slab beam bridges in Florida (Alfonso et al. 2006). In a 2008 survey of state agencies, 76% of the respondents who used box beam bridges indicated that the most common type of observed distress was longitudinal cracking along the grout and box beam interface (Russell 2009). For bridges with a CIP topping, reflective cracks often were visible in the riding surface. The impact of the cracking was to allow water and salt leakage through the longitudinal joint. This was reported as the second most common type of observed distress. The presence of longitudinal cracking in adjacent box beam bridges has been reported by others, as described here. Attanayake and Aktan (2008) reported that cracks along the shear key and box beam interfaces were observed before and after transverse posttensioning. Fifteen days after the deck was cast, through-thickness cracks that stemmed from the top surface of the deck above the abutments were observed. Mokarem et al. (2009) reported on the inspection of two adjacent precast, prestressed concrete box beam bridges: one in Ohio and one in Colorado. The Ohio bridge consisted of a single span with twelve 42-in.-deep box beams and a 3-in.-thick asphalt riding surface. With the exception of three short diagonal cracks, the entire crack pattern consisted of longitudinal cracks. The Colorado bridge consisted of two spans with twenty-four 29.5-in.-deep box beams and a 6.9-in.-thick CIP concrete deck with a 3- to 4-in.-thick asphalt overlay. no visible leakage on the underside of the box beams was observed during the inspection. The lack of visible cracking above the edges of the box beams may have been the result of using a 6.9-in.-thick CIP concrete deck that acts as a transverse tie. Stringer and Burgueno (2012) reported that longitudinal cracking was common in bridges in Michigan with both adjacent box beams and spread box beams. The cracks typically ran the entire length of the bridge and were spaced at the same spacing as the beams. The longitudinal cracks in the adjacent box beam bridges likely were caused by differential settlement between the beams and the grout filler or loss of the transverse posttensioning force between the beams. For the spread box beam bridges, longitudinal cracking likely occurred because of a concentration of longitudinal shear forces at the edges of the beams. This was predicted because the spacing of the cracks matched the beam spacing. These types of cracks were not caused by restrained concrete shrinkage. Transverse cracking was not evident in adjacent box beams but was present in spread box beam bridges. A Precast/Prestressed Concrete Institute report (PCI 2012) confirms that the predominant distress observed in adjacent box beam bridges is reflective cracking of the deck along the shear keys between beams and the associated degradation below the cracks. The cracking allows water and deicing salts to penetrate through the deck and may cause freeze-thaw damage or corrosion of the transverse tie. Researchers differ on the causes of the longitudinal cracks. Miller et al. (1999) reported that stresses caused by temperature changes crack the keys along the interface. grace et al. (2012) concluded that a positive temperature gradient (top surface hotter than the bottom surface) and not live load was the major contributing factor in the initiation of longitudinal cracks. On the other hand, other researchers have developed design procedures to prevent cracking on the basis of force transfer (El-Remaily et al. 1996; Hanna et al. 2007; Badman and Liang 2007; Hansen et al. 2012). Field inspection data for bridges in Michigan revealed that a high level of posttensioning force did not prevent reflective crack- ing (Attanayake and Aktan 2008). In a study of new York State bridges, Lall et al. (1997) concluded that the frequency of longitudinal cracking in adjacent box beam bridges was unrelated to maximum span length, total bridge length, and bridge skew. One practice that has the potential to eliminate the longitudinal cracking at the keyways is the use of ultra–high-performance concrete (UHPC) as the grout between the adjacent box beams

21 (graybeal 2014). UHPC has the ability to provide a much stronger bond strength with the precast elements than do conventional grouts. The use of UHPC in connections together with a reinforce- ment lap splice can provide a connection that is capable of transferring shear, moment, and axial forces across the connection. The UHPC detail eliminates the need for transverse posttensioning or a structural concrete overlay. The Sollars Road Bridge over Lees Creek in Fayette County, Ohio, was the first box beam bridge in the United States to use the UHPC detail. The bridge consists of seven 21-in.-deep by 48-in.-wide adjacent box beams. no transverse posttensioning was provided, so the intermediate diaphragms were eliminated. no cracking in the joint was observed before application of the waterproofing membrane (Steinberg et al. 2015). nCHRP Project 12-95 has the objective of developing guidelines for the design and construction of connection details for adjacent precast concrete box beam bridges to eliminate cracking and leak- age in the longitudinal joints between adjacent boxes. cracking in Pretensioned concrete Beams end Zone splitting cracks End zone cracking occurs in pretensioned concrete girders during or after release of the pretensioned strands. The strands may be released by flame cutting, gradual release using hydraulic jacks, or a combination. The draped strands usually are released first and then the hold-down anchorage devices at the harp points are removed. The straight strands are then released (PCI 2014). Various types of end zone cracking have been observed, as illustrated in Figure 9. Cracking at the ends of pretensioned girders is not a new phenomenon. In 1962, Marshall and Mattock reported that horizontal web cracks had been observed at the ends of pretensioned girders. These cracks occurred FIGURE 9 End zone cracks in a prestressed concrete beam (Courtesy: University of Wisconsin and Concrete Bridge Views, published jointly by FHWA and the National Concrete Bridge Council).

22 more frequently in girders with draped strands. Based on experimental research, the following equa- tion was proposed to control the size of these cracks: = ×0.021 (2)A Tf h lt s t where At = total cross-sectional area of stirrups required (in.2); T = effective prestressing force (kip); fs = maximum allowable stress in stirrups (ksi); h = overall depth of the girder (in.); and lt = strand transfer length (in.). If h/lt is taken as 2, the maximum value for which the equation was developed, the equation reduces to designing for about 4% of the prestressing force, which is the amount of splitting reinforcement required by Article 5.9.4.4.1 (formerly 5.10.10.1) of the AASHTO LRFD Bridge Design Specifications (2017). The authors recommended that this reinforcement be located as close as possible to the ends of the girder. Based on the results of the survey for this synthesis, as shown in Figure 10, end zone splitting cracks occur infrequently. Mirza and Tawfik (1978) reported that narrow vertical cracks were observed at the ends of beams during the prestressing transfer after approximately one-half of the strands had been detensioned. The cracks were located within a few inches of the beam end and extended from the bottom of the lower flange into a portion of the web. They determined that the restraining effect of unreleased strands can cause the cracking. When some of the strands were cut, the beams shortened and cam- bered, but the uncut strands restrained the shortening. Increasing the free length of the strand in the bed or debonding some strands at the ends was proposed as a means for eliminating the cracking. Kannel et al. (1997) reported vertical cracks similar to those reported by Mirza and Tawfik. The cracks occurred after the draped strands were cut and before any straight strands were cut. They also reported that diagonal cracks at an angle of 45 degrees formed in the tapered portion of the bottom flange during cutting of the straight strands. Short horizontal cracks also formed at the web-to-bottom flange intersection during cutting of the final strands. The straight strands were flame cut from the outside face of the flange toward the center of the beam. The cause of the cracks was attributed to the restraining effect of unreleased strands as the girders shorten from the partially transferred prestress and from shear stresses generated by the cutting order of the strands (Kannel et al. 1997). FIGURE 10 Frequency of end splitting cracks.

23 Based on analytical and experimental research, Kannel et al. (1997) recommended the following sequence for cutting strands: • Cut some of the bottom straight strands before all of the draped strands. A general rule of thumb is to precut one pair of straight strands for every three pairs of draped strands. • Cut the bottom straight strands in alternating columns from the interior of the cross section toward the outside face. The outer column of strands should not be the last to be released. • Where debonding is used to control cracking, cut a few straight strands first. According to Cook and Reponen (2008), six types of cracks were evident during site visits to three Florida manufacturers of precast, prestressed concrete beams. • Vertical end cracks: These cracks appear on the vertical face of the bottom flange within a few inches of the end of the beam. • Radial cracks: These cracks form a radial pattern that extends over the full depth of the web at the end of the beam. The authors attribute this cracking to the change in support location when the beams are lifted. Before the beams are lifted, the support is below the bottom flange. When lifted through lifting hooks, the lifting location moves to the top of the beam. • Angular cracks: These cracks originate in the sloped part of the bottom flange a few inches away from the end of the beam and propagate upward at an angle toward the web. These are similar to the cracks discussed by Kannel et al. (1997). • Strand cracks: These cracks originate at the end of the prestressing strand and propagate to the outer surface of the beam. The authors attribute the causes of these cracks to a combination of the Hoyer effect, strand rust occurring after casting, and the strand cutting pattern. • Horizontal top flange cracks: These cracks begin at the end face of the upper flange and move inward. Field personnel indicated that these cracks are caused by formwork pressing against the beam as it cambers during transfer. • Horizontal web cracks: These cracks begin at the end of the beam near the interface between the web and the bottom flange and extend a short distance into the beam. The authors’ primary recommendation to reduce vertical end cracks was to install steel bearing plates at the beam ends to reduce the friction force that develops between the end of the beam and the cast- ing bed. Florida now uses galvanized bearing plates in all I-beams. Field surveys of the ends of pretensioned concrete bridge beams in Virginia indicated that many of the precast bulb-tee beams developed cracks within the anchorage zone region (Crispino et al. 2009). The lengths and widths of these cracks ranged from acceptable to poor and in need of repair. Field observations also indicated deeper cross sections, heavily prestressed sections, and beams with lightweight concrete tended to be most susceptible to crack formation. Based on their analyses, the authors developed design tables that provide the area of stirrups required within h/4 and between h/4 and 3h/4 from the end of the beam, where h is the depth of the beam. Ronald (2015) has stated that bearing zone tearing cracks will not occur in every beam, only those in which the friction force exceeds the tensile strength of the concrete. The problem can be eliminated by using bearing plates, but there are likely thresholds that can be quantified to avoid the problem in less heavily stressed girders. Detensioning of all prestressing strands at the same time may also prove to be as effective as steel bearing plates in eliminating these cracks. In NCHRP Report 654, Tadros et al. (2010) identified the following possible causes for end zone cracking: • Method of detensioning: flame cut or hydraulic. • Release of the top straight or draped strands before the bottom straight strands. • Order of release of bottom strands with flame cutting. • Length of free strand in the prestressing bed. • Friction between the beam end and the bottom form of the prestressing bed. • Heat concentration from flame cutting.

24 • Lifting the beam from the bed. • Hoyer effect. • Use of larger diameter strands. • Inadequate design of end zone reinforcement. • Concrete type: lightweight or normal weight. • Strand distribution: draped or straight. Based on the results of structural testing of eight full-scale girders and field inspection of five bridges, the following proposed crack width limits were developed for acceptance, repair, or rejection of beams with web splitting cracks at the ends of beams (Tadros et al. 2010): • Cracks narrower than 0.012 in. may be left unrepaired. • Cracks ranging in width from 0.012 to 0.025 in. should be repaired by filling the cracks with approved specialty cementitious materials and coating the end 4 ft of the beam web side faces with an approved sealant. • Cracks ranging in width from 0.025 to 0.050 in. should be filled by epoxy injection and the end 4 ft of the beam web coated with an approved sealant. • For beam webs exhibiting cracks wider that 0.050 in., the beams should be rejected unless shown by detailed analysis that the structural capacity and long-term durability are sufficient. The report also recommended that vertical splitting reinforcement at the ends of prestressed con- crete beams be provided to resist at least 2% of the prestressing force at transfer and located within the distance h/8 from the end of the beam, where h is the overall depth of the precast member. In addition, the total amount of vertical reinforcement located within the distance h/2 from the end of the beam shall be provided to resist at least 4% of the prestressing force at transfer. The existing requirement of the AASHTO LRFD Specifications is that reinforcement to resist 4% of the prestress- ing force at transfer be distributed over a distance of h/4. Ross at al. (2013) conducted experimental and analytical research programs to evaluate and quan- tify the effects of different end region detailing practices on end zone cracking. The programs included the testing of 14 Florida I-beam specimens and finite element analyses. The authors concluded that the Florida DOT’s confinement reinforcement requirement is adequate for beams to at least an FIB-63. The current detail generally requires no. 3 bars at 3.5-in. spacing over a distance of approximately 0.3d from the end of the beam and then no. 3 bars at 6-in. spacing to a distance of approximately 1.5d from the end of the beam, where d is the distance from the compression face to the centroid of the tension reinforcement. Florida DOT (2016) has design standards detailing the reinforcement for each beam type and size. The current detail also includes an embedded steel bearing plate, which functions as a confining element. For the use of partially debonded strands, Ross et al. (2013) recommended that the fully bonded strands be placed as close as possible to the centerline of the web to prevent bottom flange split- ting cracks. They also indicated that the AASHTO limitation of no more than 40% of the debonded strands, or four strands, whichever is greater, be terminated at any section should control flange splitting cracks within the transfer length. To prevent or control web splitting cracks, Ross et al. (2013) recommended that the Florida DOT requirement of vertical end region reinforcement be used. This requirement is generally two no. 5 bars at 3.5-in. spacing over a distance of approximately 0.3d from the end of the beam and then two no. 5 bars at 6-in. spacing to approximately 1.5d from the end of the beam, where d is the distance from the compression face to the centroid of the tension reinforcement. Florida DOT has design standards detailing the reinforcement for each beam type and size. Okumus and Oliva (2013) identified three types of cracking in the ends of prestressed concrete girders, as shown in Figure 9. These were inclined longitudinal cracks in the webs, horizontal longitu- dinal cracks in the webs, and Y- and T-shaped cracks at the web-bottom flange intersection as seen at the ends of the girders. They evaluated methods to control these cracks using nonlinear finite element analysis methods. Results were compared with the observed crack patterns. The most effective crack

25 control methods were debonding the strands at the ends rather than using draped strands, locating the lifting loops at a distance equal to the girder depth from each end, and detensioning the strands beginning with the innermost ones. Increasing the vertical reinforcement area in the end zone alone was not recommended because it did not eliminate cracking, although it did help control crack widths. The two sets of bars clos- est to the girder end were determined to be the most effective bars in controlling the size of cracks in the webs. Additional bars further into the section were not useful in controlling web crack widths. For strand debonding, the authors recommended that the innermost strands of the bottom flange should be fully bonded and the rest of the bonded strands evenly distributed across the bottom flange. Debonding the strands within 12 in. of the end of the girders was highly recommended to control horizontal web and Y cracks. However, this approach results in the end of the beam being nonpre- stressed, a practice that may not be acceptable by state bridge engineers. It also makes it more dif- ficult to satisfy the longitudinal reinforcement requirements of Article 5.7.3.5 (formerly 5.8.3.5) of the AASHTO LRFD Bridge Design Specifications. shear cracks Compared with the number of publications about cracking in bridge decks and cracking at the ends of prestressed concrete beams, there are few publications about shear cracks in prestressed concrete beams in actual bridges. This probably is because the principal tensile stress in the web of a prestressed concrete beam at service load does not exceed the tensile strength of the concrete. To ensure that shear cracks do not occur with the use of higher strength concretes, the eighth edition of the AASHTO LRFD Bridge Design Specifications (AASHTO 2017) requires that the principal tensile stress in the webs of pretensioned girders with a compressive strength of concrete for use in design greater than f ′c = 10.0 ksi shall not exceed 0.110l√f ′c when the superstructure element is subjected to the loadings of Service III limit state, where l = concrete density modification factor. cracking Before transfer Vertical cracks extending the full depth of the webs of precast, prestressed concrete beams have been observed to develop before transfer of the prestressing force. These cracks generally are in the midspan region of the beam and are more prevalent in deep, long-span beams that have higher strength concrete and a large quantity of prestressing strands (Baran et al. 2004). They are more likely to occur if the beams are left in the precasting bed for a longer time period between the end of heat curing and transfer of the prestressing force. The cracks close up and may not be visible after the prestressing force is transferred. The cause of these cracks has been attributed to shortening as the beams and exposed strands between the beams cool. This shortening is restrained by the strands that are anchored at both ends of the bed. As a result, tensile stresses develop in the beam before transfer of the prestressing force. The likelihood of these cracks occurring can be reduced by providing sufficient free lengths of strand between adjacent beams and between the end beams and the abutments in the casting bed. A prolonged period of time between form stripping and transferring the prestressing force should be avoided (zia and Caner 1993; Baran et al. 2004). Based on the survey for this synthesis, as shown in Figure 11, vertical cracks occur infrequently before transfer. cracking in nonPrestressed concrete Beams In the survey for this synthesis, 11 agencies reported that cracking occurred infrequently or never in nonprestressed concrete beams. Eight agencies reported frequent cracking, and 25 agencies reported they did not know or the question was not applicable.

26 Flexural cracking in nonprestressed concrete beams is inevitable because cross sections are designed to be cracked. Crack control is provided by using a minimum amount of reinforcement and a maximum bar spacing. According to Higgins et al. (2004), approximately 500 CIP, reinforced concrete deck-girder bridges in the Oregon DOT inventory are identified as exhibiting diagonal-tension cracking. Most of the cracked bridges were built between 1947 and 1962. The cracks were attributed to an overestima- tion of the concrete contribution to shear that was permissible in the design specifications at the time. Subsequently, revisions were made to the specifications to alleviate the cause of the cracks. Shear cracking in some nonprestressed substructures is discussed in the next section. cracking in suBstructures Cracking in substructures appears to be uncommon but can occur, as illustrated in Figure 12. In the survey for this synthesis, 28 agencies identified that they use the same crack control criteria for substructures that they use for superstructures. Eleven agencies use different criteria. These include FIGURE 11 Frequency of vertical cracking before transfer. FIGURE 12 Cracking in a bent cap (Courtesy: Florida DOT).

27 different concrete mixes, shorter curing period, or a different exposure factor in Equation 5.6.7-1 (formerly 5.7.3.4-1) of the AASHTO LRFD Specifications (AASHTO 2017). Pier or Bent caps As part of the survey for this synthesis, 31 agencies responded that cracking in pier caps occurred infrequently or never. Ten agencies identified that their bridges had experienced frequent cracking in pier caps, and one agency identified that cracking always occurred. The cracking was attributed to shrinkage, shear, flexure, or temperature. One agency mentioned that the design stress for flexural reinforcement in all substructures under Service III loading is limited to a maximum of 24 ksi. Fu et al. (1992) reported that cracks occurred in eight pier caps of the governor Thomas Johnson Memorial Bridge in Maryland. The bridge was opened to traffic in December 1977, and the cracks were observed during the first inspection in October 1979. It was subsequently determined that the tension reinforcement was inadequate for the design condition. The design had used working stress design assuming simple flexural behavior, whereas the behavior was predominantly one of shear in a deep beam. Top and side face cracking at outside column locations in reinforced bent caps has been reported for bridges in Texas (Bracci et al. 2000). This cracking, as illustrated in Figure 13, occurs under ser- vice load conditions and, in some cases, is initiated under dead load alone. Based on experimental and analytical research, the researchers recommended that the stress in the longitudinal reinforce- ment under service load be limited to 36 ksi for moderate exposure and 30 ksi for severe exposure conditions. In addition, horizontal side face reinforcement should be provided within the web tension region per the AASHTO Specifications. To control cracking in structures or regions thereof designed by the strut-and-tie method, the AASHTO LRFD Bridge Design Specifications (AASHTO 2017) require the use of orthogonal grids of bonded reinforcement. The spacing of the bars in these grids shall not exceed the smaller of d/4 and 12.0 in., where d is the effective depth of the member, and the reinforcement shall be distributed evenly near the side faces of the strut. This reinforcement is intended to control crack widths and ensure a minimum ductility for the member. columns As part of the survey for this synthesis, 35 agencies responded that cracking in columns and abut- ments occurred infrequently or never. Six agencies reported that their bridges had experienced FIGURE 13 Diagonal cracking in a nonprestressed, concrete bent cap (Courtesy: Texas DOT).

28 frequent cracking in columns or abutments, and one agency reported that cracking always occurred. The cracking was identified as being caused by flexure, shear, shrinkage, formwork settlement, or foundation settlement. Columns are generally in compression. Thus, it is unusual for them to expe- rience cracks under service loads. Pile caps Most pile caps tend to be relatively large components in bridge construction. As such, they are suscep- tible to cracking caused by the heat of hydration of the cementitious materials. The likelihood of this cracking can be reduced by control of the temperature differential between the surface concrete and the interior concrete temperature. Cooling of the surface concrete can be prevented through the use of surface insulation. The temperature rise of the internal concrete can be controlled by using cementi- tious materials that produce a low heat of hydration, such as fly ash, and lowering the temperature of the concrete at the time of placement through the use of ice or liquid nitrogen, the use of internal cooling pipes, or a combination of these methods. Specifications for larger projects generally require that the contractor develop a thermal control plan showing how thermal cracking will be controlled. A chart indicating when concrete should be considered mass concrete has been developed by gajda and Feld (2015). As part of the survey for this synthesis, 32 agencies reported that cracking in concrete pile caps occurred infrequently or never. Several agencies reported that their bridges experienced minor crack- ing in concrete pile caps. In general, the cracking was associated with shrinkage and temperature and occurred infrequently. effective Practices for control of concrete cracking Two general approaches for crack control are possible. The first approach is to prevent the cracks. This is a goal that may not always be achievable. The second approach is to ensure that adequate reinforcement is present to control crack widths if cracking occurs. full-depth, cast-in-Place concrete Bridge decks Many factors are known to affect deck cracking, including bridge design; concrete mixture propor- tions; concrete constitutive materials; environmental conditions; and placing, finishing, and curing practices. Studies have shown that the primary causes of bridge deck cracking are shrinkage (plastic, autogenous, and restrained drying) and temperature differences between the deck concrete and the supporting beams. Practices that can reduce shrinkage and associated shrinkage cracking in CIP concrete bridge decks are as follows: • Using the lowest quantity of water and cement paste in the concrete consistent with achieving other required properties. • Using the largest practical maximum size coarse aggregate to reduce the water demand. • Avoiding concrete compressive strengths greater than 6.0 ksi. • Specifying a minimum shrinkage of 300 to 350 millionths after 28 days of drying when tested in accordance with AASHTO T 160. Practices that can reduce cracking caused by temperature differences include the following: • Minimizing the temperature difference between the CIP concrete deck and the supporting steel or concrete beams.

29 • Specifying and ensuring minimum and maximum concrete temperatures at the time of placement as 55°F and 75°F, respectively. • Minimizing cement content. • Using a Type II cement. • Using aggregates with low modulus of elasticity, low coefficient of thermal expansion, and high thermal conductivity. Construction practices that can reduce the likelihood of deck cracking include the following: • Applying wet curing procedures immediately after concrete finishing and maintaining the surface wet for at least 7 days. • Applying a curing compound after the wet curing period to slow the shrinkage and enhance the concrete properties. • Using windbreaks and fogging equipment, when necessary, to minimize surface evaporation from fresh concrete. Other practices some agencies have used that offer potential solutions are the use of SRAs, internal curing, and SCMs. After cracking, crack widths can be controlled with appropriate amounts of reinforcement. This is most effectively achieved with smaller bars and a specified maximum spacing of reinforc- ing bars. Partial-depth, Precast concrete Panels with a cast-in-Place topping The main concern with the use of partial-depth, precast concrete panels with a CIP topping is the reflective cracking that occurs in the topping above the edges of the panels. This type of cracking may be reduced by saturating the surface of the panel before casting the topping, special joint detailing, and delaying erection of the panels until most of the creep and shrinkage have occurred. Crack widths can be controlled by the reinforcement in the topping. Research is needed to identify the primary factors causing the reflective cracking and ways to reduce or eliminate it. full-depth, Precast concrete Panels Although two-directional prestressing in full-depth, precast concrete panels offers the potential to pro- duce a crack-free concrete deck, the results from the survey indicate that this is not always accomplished. Research is needed on this topic to provide a solution for crack control. adjacent Precast Box Beam Bridges and slab Beam Bridges Lateral ties are used in adjacent precast box beam bridges to tie together adjacent beams. As such, the ties function to control the crack widths. The degree of restraint varies from requiring a minimum compressive stress across the longitudinal joint to providing a passive tie. nCHRP Project 12-95 has the objective of developing guidelines for the design and construction of connection details for adja- cent precast concrete box beam bridges to eliminate cracking and leakage in the longitudinal joints between adjacent boxes. The use of UHPC is one potential solution. Pretensioned concrete Beams End zone cracking in pretensioned concrete beams occurs infrequently and can be prevented by modifying the detensioning sequence. Cracks that occur can be controlled through the use of split- ting and confinement reinforcement in the end zone region. Vertical cracking that occurs in the webs before transfer can be reduced by providing longer lengths of free strand in the prestressing bed.

30 non-Prestressed concrete Beams Cracking in nonprestressed concrete beams is almost inevitable but is controlled by providing mini- mum amounts of reinforcement to control the widths of cracks caused by flexure or shear. substructures In general, cracking in substructures occurs infrequently. Limiting the maximum stress in the rein- forcement under service loads provides a means of crack width control. Design of deep components using the strut-and-tie method, rather than the sectional design method, along with the required reinforcement should provide improved crack control. For large members, a thermal control plan should be developed.