Use of 0.7-in. Diameter Strands in Precast Pretensioned Girders (2022)

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.

1   1.1 Introduction In the United States, seven-wire prestressing strands conforming to ASTM A416/AASHTO M203 (ASTM, 2015a) are used to pretension concrete bridge members. Typically, Grade 270 (ksi) low-relaxation strand is used for bonded pretensioning (referred to in this context as “prestressing strand”). For a long time, the 0.5 strand was the standard in the bridge industry. Supported by research conducted in the 1990s, 0.6-in. strand is now accepted as a means of increasing pretensioning force and, thus, extending spans, increasing girder spacing, and decreasing structural depth. The use of seven-wire, 0.7-in. Grade 270 low-relaxation strands, which are primarily used as cable or strand roof anchors in the mining/tunneling industries, may provide similar benefits as the 0.6-in. strand did previously. The 0.7-in. strands also conform to ASTM A416 (ASTM, 2015a). Various properties of 0.5-in., 0.6-in., and 0.7-in. strands are compared in Table 1.1. In Japan, seven-wire, 0.7-in. epoxy-coated Grade 295 strands (with a nominal area of 0.294 in.2) have been developed (Kido, 2015), and 19-wire Grade 270, 0.7-in. strands (with a nominal area of 0.323 in.2) have been produced (Sumitomo Electric Industries, 2017). Both strand types conform to Japanese Industrial Standard (JIS) G 3536 (2014) for steel wires and strands. Current practice in Canada [Canadian Standards Association (CSA), 2015], United Kingdom (British Standards Institute, 2012), Europe [European Committee for Standardization (EC2, CEN), 2005], New Zealand (New Zealand Transport Agency, 2014), Japan [Japan Road Association (JRA), 2012], Taiwan (MOTC, 2009), South Korea [Korean Agency for Technology and Stan- dards (KATS), 2015; Korea Road and Transportation Association (KRTA), 2015], and China (China Highway Planning and Design Institute, 2004) does not permit the use of 0.7-in. strands in pretensioned bridge members. AASHTO (2020) includes a 0.7-in. strand by reference to AASHTO M203 (ASTM A416) but is otherwise silent on the use of a 0.7-in. strand. Inter- nationally, ASTM A416 appears to be the only available specification for the seven-wire Grade 270 0.7-in. strand. European standards BS 5896 (British Standards Institute, 2012) and EN 10138 (European Committee for Standardization, 2000) identify seven-wire Grade 250 0.7-in. strand products, but, anecdotally, they are not used in bridge construction. Other international speci- fications consulted do not include a seven-wire 0.7-in. strand of any strength. Nonetheless, the larger Grade 270 0.7-in. strands have been evaluated as pretensioning reinforcement in several experimental projects (Shahrooz et al., 2017; Schuler, 2009; Nebraska Department of Roads, 2015). Two projects in Japan that used strands larger than 0.6-in. are known: 19-wire Grade 270 0.86-in. strands were used as transverse strands in a post-tensioned bridge; and seven-wire, epoxy-coated Grade 323 strands were incorporated in a post-tensioned bridge (Kido, 2015). C H A P T E R 1 Background

2 Use of 0.7-in. Diameter Strands in Precast Pretensioned Girders 1.2 Motivations for Using 0.7-In. Strands Record-breaking spans have recently been achieved with 0.6-in. strands, including 205-ft- long girders in the Alaskan Way viaduct in Seattle, Washington (Concrete Products, 2011); 209-ft-long girders in the US 17-92 interchange at SR 463 in Casselberry, Florida (Lallathin, 2015); 210-ft-long girders in the Deerfoot Trail extension near Calgary, Alberta, Canada (Canadian Precast Prestressed Concrete Institute, 2002); and 223-ft-long modified wide-flange girders for a high-occupancy vehicle extension in Tacoma, Washington (West, 2019). A logical question, therefore, becomes “Does the precast/pretensioned industry need 0.7-in. strands?” From Table 1.1 it is apparent that the area of a 0.7-in. strand is 92% larger than that of a 0.5-in. strand and 36% larger than that of a 0.6-in. strand. The larger area of 0.7-in. strands, in conjunction with higher-strength concrete, has the potential to offer the following benefits. 1. The required number of strands in a girder could be reduced for the same girder span. Fewer strands would alleviate congestion in heavily reinforced pretensioned elements. The potential for greater spacing between fewer (particularly larger-diameter) strands is also expected to improve bond characteristics. Depending on the cost of the strand and the labor to install, fewer strands might be economically advantageous. However, simply reducing the number of strands in a girder by replacing 0.6-in. strands with fewer 0.7-in. strands that provide the same reinforcement area has little, if any, structural advantage, such as increasing span length or using fewer girders (i.e., increasing girder spacing). The potential benefit is related to cases in which all (or most) strand locations in a section are used and 0.6-in. strands cannot provide enough pretensioning force. The only way to achieve greater pretension is by increasing the individual strand force, which may be accomplished by replacing a 0.6-in. strand with a 0.7-in. strand on a one-to-one basis on the same strand spacing grid. 2. The total number of girders in a bridge could be reduced by using individual girders having greater pretension force. Fewer girders should shorten the construction time and cost and reduce overall energy consumption, all of which are advantageous from economical and sustainability points of view. 3. Longer spans could be achieved using girders having greater pretension force. Longer spans may reduce the number of piers required for a new bridge or permit the elimination of the central pier in typical two-span bridges. In bridge replacement projects, particularly in con- gested urban areas, eliminating the central piers on large thoroughfares or interstate crossings (1) may permit more efficient expansion of the roadways beneath the bridge; (2) eliminate the hazards associated with piers located close to the roadways; and (3) minimize the impact of the span on environmentally sensitive habitats. Nonetheless, there are practical upper limits on girder length because of size and weight limitations associated with shipping and handling (Castrodale and White, 2004). 4. Shallower girders having greater pretension force could be used for the same span. This benefit becomes particularly important in replacement projects that have to maintain or Diameter (in.) Nominal area (in.2) Nominal weight (lb/ft) Min. breaking strength (kips) Min. load at 1% extension (kips) Min. elongation in 24 in. 0.5 0.153 0.520 41.3 37.2 3.5%0.6 0.217 0.740 58.6 52.7 0.7 0.294 1.000 79.4 71.5 Table 1.1. Physical properties of Grade 270, low-relaxation strand (ASTM, 2015a).

Background 3   increase clearances beneath the bridge. Once again, clearance is a major design issue in urban areas of many states or for bridges over waterways where the bridge limits the hydraulic opening. 1.3 Objectives of Research Program The primary objective of the reported study was to identify issues related to the design and fabrication of pretensioned girders using 0.7-in. strands and to develop design methodologies and guidelines for bridge girders using such strands. In particular, the research addressed (1) bond and material characterization; (2) transfer and development length; (3) strand spacing; (4) longitudinal reinforcement requirements and details at ends of girders; (5) anchorage of 0.7-in. strands into diaphragms; (6) flexure and shear capacity and ductility of girders having 0.7-in. strands; and (7) stability of long-span girders during all phases of construction. 1.4 Review of State of the Art and Practice 1.4.1 Past Experimental Studies Akhnoukh (2008 and 2013) and Akhnoukh and Carr (2012)—As part of a study developing a self-consolidating high-strength concrete and evaluating the use of welded wire reinforce- ment (WWR) as confinement and shear reinforcement, transfer and development lengths of 0.7-in. strands were investigated through several prism tests (with nonprestressed and pre- stressed strands) and testing of two NU-900 girders. A formulation based on shear friction theory was developed to estimate the required amount of confinement around 0.7-in. strands such that they would comply with the transfer length and development length equations in AASHTO LRFD Bridge Design Specifications (AASHTO, 2008). The strands in one NU-900 girder were placed horizontally at 2 in. on center and 2.5 in. on center vertically while those in the other girder were placed on a 2-in. horizontal and vertical grid. A number of the strands were bent up 90 degrees, and all the strands were embedded in end diaphragms. Although confine- ment could affect the transfer length, the level of confinement provided was found to have only a marginal influence. The strands could be developed at AASHTO-prescribed (AASHTO, 2008) development length (Eq. 5.9.4.3.2-1) or less, and strands did not slip. The lack of slip is attributed primarily to the benefits of embedment into the end diaphragms, which may not always be present in bridges using I girders. The WWR shear and confinement reinforcement provided was found to be adequate. Briere et al. (2013)—Using an approach developed by Oh et al. (2006), Briere et al. analytically investigated the effects of various parameters on the potential for radial cracking associated with prestressing strand release and stress transfer to concrete. It was found that varying the apparent dilation ratio of the strand, υp, has a significant effect on the stresses associated with strand anchorage. As υp increases, the interface stress increases proportionally; as a result, the expected extent of the cracked region surrounding the strand becomes larger, potentially leading to the interaction of cracks from adjacent strands or the cracks being expressed at the concrete surface, both leading to potential loss of anchorage and, as the stresses become greater, longitudinal splitting failures. Cabage (2014)—The transfer and development lengths of 0.7-in. strands were examined by testing mono-strand reinforced prisms and rectangular beams, and two AASHTO Type I girders (one with 0.6-in. and another with 0.7-in. strands). The mono-strand specimens had no transverse reinforcement. The main test variables in the mono-strand tests were (1) the length of the prisms and (2) the level of pretensioning ranging from 0 to 60 kips (0 to 0.75fpu).

4 Use of 0.7-in. Diameter Strands in Precast Pretensioned Girders By comparing the results from the pretensioned and nonpretensioned strands, Cabage attempted to identify the contribution of radial expansion to bond capacity. The concrete in the prisms had a compressive strength of 9 ksi. The pretensioned and nonpretensioned strands behaved differently. As expected, the strands embedded in the longer prisms had a larger pullout capacity than those with a shorter embedment length. The contribution of adhesion could be appreciable; however, the adhesive component of the bond is expected to be lost during strand detensioning. An iterative model, which neglects adhesive bond, was developed to estimate the transfer length. The strand perimeter, along which bond is affected, is greater for larger strands and the absolute radial expansion of large strands is also greater than that for small strands. Therefore, Cabage concluded that the transfer length is expected to be shorter for larger strands. Both ends of three 12 (d) × 8 in. (b) rectangular beams having a single 0.7-in. strand were tested to investi- gate the development length of large-diameter strands. The 28-day concrete compressive strength ranged from 6.3 to 12.7 ksi. The measured development lengths were appreciably smaller than the values computed based on AASHTO LRFD Bridge Design Specifications (AASHTO, 2008). A better estimate of the measured development lengths was possible using an iterative model similar to that used for estimating the transfer length. The AASHTO Type I girder with 0.7-in. strands had a concrete strength of 10.6 ksi at release and 14.1 ksi at test. Straight strands were used at one end, and the other end had hot-bent strands cast into an end block. The measured development length was appreciably larger than that observed in the mono-strand prism tests or the value predicted by the model developed by the author. The beam behavior, however, was accurately predicted by currently available software such as Response 2000 (Bentz, 2000). Dang (2015)—In several pretensioned concrete beams, the transfer and development lengths of 0.7-in. strands were evaluated. The test specimens were 12 (d) × 6.5 in. (b) rectangular beams with one or two strands (having a 2-in. center-to-center spacing) and two No. 5 and two No. 6 reinforcing bars, respectively, at the top. The concrete compressive strength at release ranged between 5.8 and 9.9 ksi with 28-day strength between 8.8 and 13.3 ksi. Equations for computing transfer and development length were proposed. The concrete strength was found to have little or no effect on the measured development lengths. Hanna et al. (2010a) and Morcous et al. (2011)—Prior to the construction of Pacific Street Bridge, the first bridge in the United States to use 0.7-in. strands, two NU-900 girders were tested. Girder A, having a 28-day concrete strength of 8 ksi, had twenty-four 0.7-in. strands placed on a 2.2 in. (horizontal) × 2.25 in. (vertical) grid tensioned to 0.75fpu. Girder B, having a concrete strength of 15 ksi, had thirty 0.7-in. strands placed on a 2-in. grid tensioned to only 0.66fpu (due to limitations of the pretensioning bed, this number of 0.7-in. strands could not be fully prestressed). To control cracking in the top flange of Girder B, four 0.5-in. strands tensioned to 0.073fpu and four No. 5 Grade 60 nonprestressed reinforcing bars were used. The end region of Girder A exhibited some minor cracking at release. The observed transfer lengths of both girders were closer to the shorter value of (fse/3000)db (which the authors took as equal to 50db) prescribed by ACI Committee 318 (2008) than the value of 60db prescribed in the AASHTO LRFD Bridge Design Specifications (AASHTO, 2010). The transfer length of Girder B, having higher concrete strength, was shorter than that of Girder A, although the lower initial pretension stress may also contribute to this observation. Before load testing, the second layer strands were bent up 90 degrees, and a diaphragm was cast around the straight and bent strands. The bent strands and end diaphragms were intended to help the girders develop their flexural capacity without strand slip. Based on strand slip data at the maximum measured loads, it was concluded that the diaphragm-embedded strands were developed over a shorter development length than that determined by the AASHTO LRFD Bridge Design Specifications (AASHTO, 2010); however, the instrumentation was insufficient to quantify the development length.

Background 5   In Girder A, vertical No. 3 hairpins spaced at 3 in. were provided for a distance of 1.27 times the girder depth from the girder end and were adequate to prevent splitting of the bottom flange from the web. Girder A failed at 95% of its predicted shear capacity and 104% of its calculated flexural capacity. Shear failure was due to inadequate anchorage of shear reinforcement in the bottom flange (the AASHTO 0.25 fc’ bv dv shear strength limit was not satisfied). Revised details used in Girder B prevented shear failure. Maguire et al. (2013) and Tadros and Morcous (2011)—Similar to studies by Morcous et al. (2011), simply supported high-strength concrete Bridge Double Tee (BDT) girders, having f ′c = 17,500 psi, were tested. Due to limitation on prestressing capacity, the BDT girders had an initial pretension of only 0.60fpu. The strands in the BDT were harped at 0.4 L. Unsurprisingly, considering the lower-than-typical values of fpe, the experimentally determined transfer lengths were lower than those predicted by ACI Committee 318 (2008) or AASHTO LRFD Bridge Design Specifications (AASHTO, 2010). For the midspan flexure test, no cracking associated with pretension release was reported, and the girder achieved its predicted ultimate flexural capacity calculated using nominal material properties and no shear failure was reported. How- ever, bond and horizontal shear failures were reported for a girder tested to examine its shear capacity and behavior. Unlike the previously reported NU-900 girders (Morcous et al., 2011), the BDT girder had no additional end diaphragm in which to anchor strands. No strand slip was reported; however, it is noted that the BDT was a considerably more slender beam, having a span-to-depth ratio of 29.3 as compared to 13.2 for the NU-900. Jiang (2013)—Analytical and experimental studies were conducted to examine the bond performance of 0.7-in. strands. Factors influencing transfer length were studied through finite element analyses validated using pullout tests of nonpretensioned and pretensioned strands. The finite element analyses indicated that the transfer length and slip increased as the coefficient of friction (between strand and concrete) decreased. Based on the analytical studies, it was concluded that 2-in. center-to-center spacing between 0.7-in. strands would be adequate to allow bond stresses to develop adequately. By using prisms having 1-day and 2-day concrete strengths of 8.5 and 9.1 ksi, respectively, it was concluded that bond behavior is improved with greater embedment length and with greater pretension force [this is attributed to the increased normal forces developed as a result of the Hoyer effect (Hoyer, 1939; Hoyer and Friedrich, 1939)]. Morcous et al. (2010a and b); Patzlaff (2010); Patzlaff et al. (2009, 2010, and 2012); and Tadros and Morcous (2011)—Through testing of mono-strand prisms and T and NU girders, the effects of confinement around 0.7-in. strands on the transfer and development lengths and vertical shear capacity were investigated. The concrete strength ranged between 9 and 13.5 ksi for the T girders, and the nominal concrete strength of the NU 1100 was 10 ksi. The main test variables were the spacing of the confining reinforcement and, in the girders, the extent of con- fining reinforcement into the girder. The horizontal and vertical center-to-center spacing of the strands in the girders was 2 in. Similar to observations made for 0.5-in. and 0.6-in. strands (e.g., Russell and Burns, 1996), the transfer length was not noticeably affected by the amount and distribution of confinement reinforcement, and the measured transfer lengths were appre- ciably smaller than that computed based on AASHTO LRFD Bridge Design Specifications. It was concluded that 0.7-in. strands can be developed within the AASHTO-prescribed development length (Eq. 5.9.4.3.2-1 in AASHTO, 2020), provided the concrete has a compressive strength of at least 10 ksi and confinement reinforcement required in AASHTO LRFD Bridge Design Specifications Article 5.9.4.4.2 (AASHTO, 2020) is provided. A shorter development length may be achieved by increasing the ratio of confinement reinforcement. Based on flexural testing of the girders, it was concluded that the equations in AASHTO LRFD Bridge Design Specifications (AASHTO, 2020) are adequate to determine the development length of 0.7-in. strands spaced at 2 in. (horizontally and vertically) and tensioned to 0.75fpu. However, the actual development

6 Use of 0.7-in. Diameter Strands in Precast Pretensioned Girders length could not be determined due to insufficient instrumentation. The level of confinement reinforcement was found to not significantly affect the vertical shear capacity (although the shear capacity exceeded the design requirements by at least 16% in any event). On the other hand, adequate confinement increased the overall ductility of the girder and reduced the slippage of prestressing strands. Morcous et al. (2012) and Tadros and Morcous (2011)—Mechanical properties and bond characteristics of 0.7-in. strands were evaluated by performing material testing on a large number of strand samples and by conducting the North America Strand Producers (NASP) bond test (now ASTM A1081), respectively. The 1-day concrete strength in the NASP tests ranged between 4.8 and 10 ksi. The Power Method proposed by Mattock (1979) was found to accurately model the strand stress-strain relationship. The NASP test method was found to be applicable for 0.7-in. strands in mortar and concrete. The bond strength was found to change as a function of the concrete strength. Shahrooz et al. (2017)—As part of NCHRP Research Report 849, one NU-1100 girder with 0.7-in. Grade 270 strands was fabricated and tested. The 55-ft-long girder had 22 strands spaced 2 in. on center vertically and horizontally. Ten strands (45%) were debonded at one end (End A) and six (27%) at the other end (End B). The concrete strength (at the time of testing) was 14.0 ksi for End A and 13.2 ksi for End B. It was determined that the measured girder capacities exceeded those calculated based on AASHTO LRFD Bridge Design Specifications (AASHTO, 2010). No adverse effects of the use of 0.7-in. strands were observed in this study. Vadivelu (2009) and Song et al. (2014)—The transfer length and end cracking of AASHTO Type I girders using twelve 0.6-in. Grade 330 and 0.7-in. Grade 270 strands were compared. In both girders, the strands were spaced at 2 in. on center vertically and horizontally. The concrete strength at release was 10 ksi, and the 28-day concrete strengths were 14.2 and 12.3 ksi for the girders with 0.7-in. and 0.6-in. strands, respectively. The confinement and transverse reinforcement at one end of each girder were designed based on the 2008 edition of AASHTO LRFD Bridge Design Specifications (AASHTO, 2008); the reinforcement at the other end was designed by a strut-and-tie model (STM) developed by the authors. The measured transfer length, which was not affected by the amount of confinement reinforcement, for both sizes of strands was shorter than the AASHTO-prescribed values. In the case of 0.7-in. strands, the measured transfer length was approximately one-half of the value predicted by AASHTO, whereas the measured value was 75% of the prescribed 60db value for the 0.6-in. strands. However, since the higher-strength 0.6-in. strands had fpi = 0.75 and fpu = 247.5 ksi, the basic AASHTO transfer length should be increased to 82.5db. In this case, both strands performed similarly, having a transfer length of approximately one-half that prescribed by AASHTO. The end design based on AASHTO provisions exhibited slightly more cracking than the other end, although the girder using 0.7-in. strands had some spalling in the end design based on STM. An equa- tion for computing splitting force was developed, and the study concluded that the use of 0.7-in. strands having 2-in. center-to-center spacing did not result in cracking around individual strands as long as high-strength concrete (10 ksi) is used. Because a high-strength 0.6-in. strand was used, the total prestress force in the two girders differed by only 10% (714 kips for the Grade 270 0.7-in. and 645 kips for the Grade 330 0.6-in. strands). Unfortunately, there is no record of these comparable girders having been tested. Akhnoukh (2013); Morcous (2013); and Morcous et al. (2014)—Implementation of 0.7-in. strands at 2 in. × 2 in. spacing in the Oxford South Bridge in Oxford, Nebraska, was reported. The two-lane bridge has five spans, four of which are 110 ft and the fifth is 140 ft long. Four NU-1350 girders were used in each span. The girders in the 140-ft-long span had thirty-four 0.7-in. straight strands and six 0.6-in. harped strands. The 0.7-in. strands tensioned at 0.75fpu could not be harped due to the unavailability of suitable hold-down devices. Some of the 0.7-in.

Background 7   strands were, therefore, debonded to control initial cracking stresses; the maximum debonding ratio was 0.29. A total of twenty-four 0.7-in. straight strands were used in the 110-ft-long spans. Some of the strands in these girders were also debonded with a maximum debonding ratio of 0.25. The end confinement provided (4 × 4 D7 × D7 WWR) met AASHTO LRFD Bridge Design Specifications Article 5.9.4.4.2 (AASHTO, 2020) requirements. Additional confinement reinforcement (4 × 4 D4 × D4 WWR) was provided for the remainder of the span. The strands extending beyond the girder ends were bent 90 degrees and embedded in cast-in-place end diaphragms. The reported 28-day concrete strengths were 8 and 9 ksi for the 110-ft and 140-ft- long girders, respectively. The transfer length for two of the 110-ft-long girders was measured and found to be closer to the ACI 318R-14 (ACI Committee 318, 2014) value of 50db (based on long-term strand stress of 0.55fpu) than the 60db default value prescribed by AASHTO. No cracks around the strands could be found based on visual inspection of the girder ends, suggesting adequate bursting and confinement reinforcement as determined by AASHTO LRFD Bridge Design Specifications (AASHTO, 2020). The top of some of the 110-ft girders experienced crack- ing. The measured cambers at release were reasonably close to the values computed based on the Precast/Prestressed Concrete Institute (PCI, 2010) method; however, the cambers measured after 60 days were larger than the estimated value by as much as 1.5 in. 1.4.2 Past Analytical Studies Design case studies and finite element analyses have been conducted to evaluate the impacts of using 0.7-in. strands or to develop preliminary design charts for girders with 0.7-in. strands. These studies are summarized below. Akhnoukh (2008); Ross (2012); Ross et al. (2013); and Shahrooz et al. (2017)—Confinement reinforcement, specified in AASHTO LRFD Bridge Design Specifications Article 5.9.4.4.2 (AASHTO, 2020), has been investigated using two different approaches. Akhnoukh (2008) proposed a model based on shear friction for designing confinement reinforcement. Ross (2012) developed an STM to establish the likelihood of bottom flange splitting cracks under service loads and for ultimate strength design. Shahrooz et al. (2017) generated a more nuanced STM to determine the confinement reinforcement requirement for single-web girders (I, BT, and NU) reinforced with 0.5-in., 0.6-in., or 0.7-in. strands. This latter model, subsequently incorporating the work of Ross (2012) and presented in Harries et al. (2019), accounts for the support width and debond- ing pattern and can be used to rationally calculate the required amount of confinement reinforce- ment needed to prevent lateral-splitting failure at the girder ultimate load. Brown et al. (2012); Storm et al. (2013); and Tadros et al. (2011)—These studies evaluated the influence of production practices on camber with PCI forming a FAST team to evaluate these tolerances (Brown et al., 2012). The FAST team found significant scatter when compar- ing the calculated and measured cambers from more than 1,800 girders, pointing to challenges of estimating camber. Tadros et al. (2011) proposed a method that uses conventional elastic camber equations to estimate camber at release but additionally accounts for the girder self- weight after the girder is placed on storage dunnage. Camber at erection was estimated using AASHTO LRFD Bridge Design Specifications Article 5.9.3.4 (AASHTO, 2020) for refined estimates of time-dependent losses. Nevertheless, the camber was found to vary by as much as 50%. Castrodale and White (2002); Davis et al. (2005); Crispino et al. (2009); and Tuan et al. (2004)— STMs have also been developed for the design of the anchorage zone (e.g., Castrodale and White, 2002; Davis et al., 2005; Crispino et al., 2009). These models are similar to the Gergely-Sozen model (Gergely and Sozen, 1967), with the main difference being that the location of the ties in the STM has been calibrated based on the observed locations of cracks. Based on experimental data from girders, in which the strands were released gradually, Tuan et al. (2004) recommended

8 Use of 0.7-in. Diameter Strands in Precast Pretensioned Girders that AASHTO Bridge Design Specifications Article 5.9.4.4.1 (AASHTO, 2020) be amended to require 50% of the total splitting reinforcement be placed within h/8 (h is the girder total depth) from the girder end and the balance be evenly distributed over the distance h/8 to h/2 from the girder end. Ferhadi (2011)—As part of examining various means for extending the span length or increasing girder spacing, the application of 0.7-in. strands was investigated in conjunction with lightweight and normal-weight high-performance concrete and construction type. The author examined the following construction types: (1) simple spans, (2) simple spans for girder and deck weight, and continuous for superimposed dead loads and live loads, and (3) simple spans for girder weight and continuous for deck weight, superimposed dead loads, and live loads. The study focused on PCI/AASHTO bulb tee (BT-72), Washington State Super Girder wide flange (WF58G), and PCI/AASHTO Box Girder (BIII-48). The benefits of using 0.7-in. strands over 0.6-in. strands were found to depend on the type and strength of concrete, the girder shape, the type of construction, and span lengths. For example, in the case of simple spans, the ideal concrete strength is 15 ksi when 0.7-in. strands are used in BT-72 and BIII-48 girders. The use of 0.7-in. strands in simple span bridges using WF58G did not offer a significant benefit over 0.6-in. strands. On the other hand, WF58G girders using 0.7-in. strands would be more advantageous than 0.6-in. strands for spans greater than 150 ft in bridges designed to be con- tinuous for superimposed dead and live loads. Hanna et al. (2010b)—Design charts were developed for NU I-girders using 0.6-in. and 0.7-in. strands and concrete compressive strengths from 8 to 15 ksi. The preliminary design charts are applicable to simple span, two-span continuous, and three-span continuous bridges. Russell et al. (1997)—This project focused on evaluating pretensioned girders and optimized sections developed by several states in terms of their application with high-strength concrete. In addition to considering 0.5-in. and 0.6-in. strands (the latter were not as common at the time), the application of 0.7-in. and 0.9-in. strands in BT-72 girders using 10 ksi concrete was also examined. The report concluded that 0.7-in. strands spaced at 2 in. would result in the most cost-effective structures with the longest spans, while 0.9-in. strands spaced at 4 in. were found to be inefficient. The authors highlighted the importance of transfer and development length aspects of design. Shahrooz et al. (2017)—A comprehensive nonlinear finite element modeling of an NU-1100 girder with twenty-two 0.7-in. strands was developed. The computed load–deflection responses, slippage of strands, and damage patterns were found to be close to their experimental counterparts. Several additional models (not corresponding to experiments) were also developed, including cases with 0.7-in. strands. Weldon et al. (2012)—The main focus of this research was to investigate the advantages and limitations of implementing ultra high-performance concrete (with compressive strengths ranging between 15 and 22.5 ksi) into pretensioned bridge design in New Mexico. As part of this study, several design cases with 0.6-in. and 0.7-in. strands were conducted. The parametric studies examined various flexure and shear design issues in simple span bridges and two- and three-span bridges that were continuous for superimposed dead and live loads. The use of 0.7-in. strands was found to be most beneficial when high-strength concrete is used. For identical design parameters (e.g., number and length of spans, number of lanes), smaller sections provide adequate flexural strength when 0.7-in. strands are used. However, the shear capacity of the smaller sections could require sections with wider webs or additional transverse reinforcement. Vadivelu (2009) and Vadivelu and Ma (2008 and 2009)—Finite element analyses of AASHTO Type I girders reinforced with 0.7-in. and 0.6-in. strands spaced at 2 in. indicated a greater possibility of cracking at the transition between the bottom flange and web when 0.7-in. strands

Background 9   were used. Nevertheless, the authors concluded that confinement steel could reduce this possibility. On the basis of several parametric design studies, the authors concluded that a BT-72 girder using 0.6-in. strands and a BT-54 girder with 0.7-in. strands could span the same lengths. Considering the wider flange width of NU girders can accommodate a larger number of strands than BT girders, an NU-1800 (with a depth of 71 in.) can span further than a BT-72; however, the practicality of the additional span length is questionable because of difficulties associated with transporting girders longer than 160 ft. 1.4.3 Demonstration Projects Using 0.7-In. Strands The use of 0.7-in. strands in two demonstration projects has been reported. The Pacific Street Bridge over I-680 in Omaha, Nebraska (Schuler, 2009), has two spans of 110 ft-5 in. that were made continuous at the central pier. NU-900 girders (with a depth of 35.4 in.) spaced at 10 ft-8 in. were used. The girders were pretensioned with thirty 0.7-in. strands. The Oxford South Bridge in Oxford, Nebraska (Morcous, 2013), has five spans (four 110-ft spans with the central span of 140 ft). The NU-1350 girders (with a depth of 53.1 in.) are spaced at 9 ft. The 140-ft-long girders used thirty-four 0.7-in. strands with six 0.6-in. harped strands, and the girders in the 110-ft spans had twenty-four 0.7-in. strands. Considering that the NU girder bulb has 60 strand locations, neither project strictly needed 0.7-in. strands; an adequate design having 0.6-in. strands is feasible in both cases. Nonetheless, the use of 0.7-in. strands reduced reinforcing congestion in each case. 1.4.4 Synthesis of Previous Studies 1.4.4.1 Synthesis of Past Experimental Studies and Demonstration Projects 1. The majority of previous test specimens were mono-strand reinforced prisms or rectangular beams (12 × 6.5 in. or 12 × 8 in.) with at most two strands. The remaining specimens con- sisted of girders that are not frequently used, particularly for long-span girders (i.e., AASHTO Type I or T girders), or do not appear to have taken full advantage of the 0.7-in. strands. 2. Except for NU girders, the number of strands (and therefore the degree of pretension) in pre- vious test specimens does not represent the quantity needed in long-span girders. The full effect of interaction between strands on transfer and development length has, therefore, not been thoroughly evaluated. Nevertheless, a limited number of previous tests have demonstrated that 0.7-in. strands may be placed on a 2-in. center-to-center grid: (1) a 55-ft-long NU-1100 girder with twenty-two 0.7-in. strands (Shahrooz et al., 2017), (2) a 40-ft-long NU-900 girder with thirty-four 0.7-in. strands (Morcous et al., 2011), and (3) a 40-ft-long NU-1100 girder with thirty-four 0.7-in. strands (Morcous et al., 2010b). It is interesting to note that 0.7-in. strands at 2 in. center-to-center spacing have been incorporated only in NU girders. 3. An attempt was made to determine a development length of 0.7-in. strands in 17 out of a total of 102 reported test specimens. These limited tests concluded that development length is equal to or less than that computed based on AASHTO LRFD Bridge Design Specifications (AASHTO, 2020); however, the actual development length was not quantified. Furthermore, many of the extant tests provided additional anchorage (90-degree bends) and embedment into an end diaphragm, rendering conclusions regarding development length not appli- cable to conventional construction without external strand embedment. Of the long-span, appropriate shapes tested, only the NU-1100 tested by Shahrooz et al. (2017) and those reported by Morcous et al. (2010b) did not use diaphragms to improve strand anchorage at the ends of the beams.

10 Use of 0.7-in. Diameter Strands in Precast Pretensioned Girders 4. The design concrete strength was at most 10.3 ksi except for six specimens reported; one of these six was used to evaluate transfer length only. A 55-ft NU-1100 girder had a concrete strength of 14 ksi at the time of testing (Shahrooz et al., 2017). 5. Bent-up strands may be used to develop continuity for live load. Hanna et al. (2010a) and Morcous et al. (2011) embedded bent strands in diaphragms to improve anchorage at the ends of the beam. However, anchorage of bent 0.7-in. strands in diaphragms has not been evaluated systematically. The current equation in the AASHTO LRFD Bridge Design Specifica- tions (AASHTO, 2020) for the capacity of a bent strand in a diaphragm (Article 5.12.3.3.9c) was developed for 0.5-in. strands and has not been verified for larger strands. 6. Only NU girders have been used in the limited demonstration projects in Nebraska. The girders for these projects were fabricated by one producer. Potential fabrication issues of other shapes with 0.7-in. strands have not been studied, and the experience of producers with 0.7-in. strands is not available. 1.4.5 Synthesis of Past Analytical Studies Past studies have not thoroughly investigated several issues: (1) the cases for which 0.7-in. prestressing strands are more advantageous than smaller strands of equal strength or (2) the influence of girder shape, girder spacing, and concrete strength on the efficiency of 0.7-in. strands. Detailed three-dimensional nonlinear finite element modeling of girders using 0.7-in. strands has not been conducted with the exception of one study (Shahrooz et al., 2017). Moreover, interactions between 0.7-in. strands on strand-concrete bond (and, hence, transfer and devel- opment length) have not been examined. End-region cracking and confinement have been examined primarily for 0.5-in. and 0.6-in. strands. Additionally, little information is available on the camber of girders reinforced with 0.7-in. strands. 1.5 U.S. and International Codes Current practice in the United States and other countries does not address the use of 0.7-in. strands. Codes from the United States and several other countries were examined in terms of two aspects that could affect the performance of 0.7-in. strands: (1) distance between strands and (2) transfer and development length. These two issues are discussed in the following sections. 1.5.1 Minimum Distance between Strands The distance between strands affects the degree of distress in the concrete surrounding individual strands and, therefore, the transfer of pretensioning force to the surrounding concrete. To account for this phenomenon, among other reasons, design codes specify a minimum distance between strands (typically in terms of a center-to-center spacing requirement). The minimum distance is also intended to mitigate the interaction of stress concentrations around the strands at the ends of pretensioned elements at release. Minimum spacing requirements are also based on the maximum aggregate size to permit adequate consolidation of concrete. The require- ments from several U.S. and international sources are summarized in Table 1.2. In this table and throughout the report, the references to sections and equations in the AASHTO LRFD Bridge Design Specifications (AASHTO, 2020) are from the edition before the reorganization of Section 5. None of these codes address the spacing requirements for 0.7-in. strands, which have a 17% larger diameter than 0.6-in. strands. In most cases, strand spacing is proportional to strand diameter. Except for the Korean Bridge Standard, 2.0-in. spacing is adequate for 0.6-in. strands although the trends illustrated suggest that a greater spacing may be appropriate for 0.7-in. strands.

Background 11   1.5.2 Strand Spacing in U.S. Practice and the Transition from 0.5-In. to 0.6-In. Strands “Modern” strand spacing requirements appear to have been introduced in ACI 318-63R (ACI Committee 318, 1963). Article 2617(b) states that the “minimum clear spacing between pretensioning steel at each member end shall be . . . three times the diameter of strands. . . .” This is a spacing of 4db. Much of the basis for the 1963 requirements seems to come from Hanson and Kaar (1959) who report 250 Grade strand sizes up to 0.5 in. Curiously, there is no mention of strand spacing in the multistrand specimens tested. One assumes that the 0.5-in. strand was tested with 4db = 2 in. spacing (this is consistent with specimen dimensions); however, it is unclear whether smaller 0.25- and 0.375-in. strands were tested with a spacing of 4db or 2 in. The research team suspects the latter (resulting in the spacing of 8db and 5.3db, respec- tively). Tabatabai and Dickson (1993) and Buckner (1995) provide further insight, including reporting telephone conversations with Professor Mattock. In 1988, FHWA placed a moratorium on the use of 0.6-in. strands and prescribed a minimum spacing of 4db. The primary issues were that available data included only 0.5-in. strands and, critically, only 250 Grade stress-relieved strands. The 1988 memorandum spurred significant research efforts summarized by Lane (1998) and was revised in 1996 to permit the use of 0.6-in. strands and allow spacing of 2 in. (3.33db) for 0.6-in. strands and 1.75 in. (3.5db) for 0.5-in. strands. The reduced spacing permitted is based on satisfactory test results of 16 AASHTO Type II girders having either 5 or 10 ksi concrete (Lane, 1998). Another significant test program Code Requirements * AASHTO LRFD Bridge Design Specifications (2020) 0.5-in. strand > 1.75 in. [Table 5.4.4.1-1] 0.6-in. strand > 2.0 in. [Table 5.4.4.1-1] >1.33 maximum aggregate size > 1 in. [Article 5.9.4.1] Strand bundles are permitted in a vertical plane; limited to four strands/bundle in other than a vertical plane ACI 318R-19 (2019) ′ ≤ 4 ksi > 4 for all bar sizes ′ > 4 ksi <0.5-in. strand >4 0.5-in. strand >1.75 in. 0.6-in. strand >2.0 in. >1.33 maximum aggregate size > 1 in. Strand bundles are permitted in a vertical plane in the central region of the span Canada (CSA S6, 2015) > 50 mm (2 in.) − > 1.33 maximum aggregate size > 1 in. European Standard (EC2, 2005) − > > maximum aggregate size + 5 mm (0.2 in.) > 20 mm (0.75 in.) Up to eight strands may be bundled in a vertical plane; limited to four strands otherwise Japan (JRA, 2012) “Clear spacing between tendons or sheaths shall be determined in view of such factors as the type of tendon and the diameter of tendons or sheaths so that the space around tendons or sheaths can be completely filled with concrete, and concrete can be compacted in a reliable way and adequate bonding can be achieved.” China (JTG, 2004) s > 1.5db > 25 mm (1 in.) Korea (KSA, 2015) − > 1.33 maximum aggregate size 0.5-in. strand − > 44 mm (1.75 in.) [resulting in > 2.25 in.] 0.6-in. strand − > 50 mm (2.0 in.) [resulting in > 2.60 in.] Taiwan (MOTC, 2009) > 3 − > 1.33 maximum aggregate size = strand center-to-center spacing; = strand diameter; = clear spacing between strands. Table 1.2. Strand spacing requirements from U.S. and international sources.

12 Use of 0.7-in. Diameter Strands in Precast Pretensioned Girders (Unay et al., 1991), which considered 0.5-in. and 0.6-in. strands only considered spacing of 2 in. (4db and 3.33db, respectively) and 2.25 in. (4.5db and 3.75db, respectively). No difference was seen in performance, and it was concluded that 2-in. spacing was adequate. It is noted that in Unay et al. (1991), the edge distance for all strands was 2.5 in. Shahawy (1999) similarly concludes that “a spacing of 2 in. is acceptable, irrespective of strand size, up to 0.6 in. dia.” The specimens reported in this work, having either 0.5-in. or 0.6-in. strands, were provided with 2-in. spacing for both strand sizes. In summary, most studies have used 2-in. spacing for both 0.5-in. and 0.6-in. strands. Only one study can be found that (1) uses smaller, 1.75-in., spacing for 0.5-in. strands and (2) uses larger, 2.25-in., spacing for 0.6-in. strands. The 2-in. value seems to be a legacy of the 4db requirement prescribed in 1963. 1.5.3 Transfer and Development Length Design codes represent the bond strength and transfer mechanism by specifying transfer length (the length over which effective stress in pretensioned members is transferred to the concrete) and development length (the length required to develop the stress in the strand corresponding to the nominal flexural capacity of the member), as shown schematically in Figure 1.1. The current AASHTO-prescribed equation (Eq. 5.9.4.3.2-1 in AASHTO LRFD Bridge Design Specifications) (AASHTO, 2020) for strand development length is: = κ −    2 3 l f f dd ps pe b Eq.1.1 Where fpe is the effective pretensioning stress, fps is the strand stress to be developed, and db is the strand diameter. The parameter κ was introduced in 2001 to correct observed worst-case characteristics of some strands shipped before 1997; the value of κ = 1.6 for pretensioned members having a depth greater than 24 in., and κ = 1.0 for members shallower than 24 in. For strands whose transfer length is not in a girder support region (such as partially debonded strands), κ = 2.0. Figure 1.1. Idealizations of transfer and development length.

Background 13   Eq. 1.1 can be rewritten as ( )= κ + κ − 3 l f d f f dd pe b ps pe b where the first and second terms represent the transfer length and the additional length required to develop the flexural bond strength, respectively. The sum of these terms is the development length of the strand. In the AASHTO LRFD Bridge Design Specifications (AASHTO, 2020), the transfer of strand stress is idealized as shown in Figure 1.1. For simplicity, the LRFD Specifications allow the transfer length (lt) to be taken as 60db, which corresponds to fpe = 180 ksi. The transfer and development lengths from a number of U.S. and international sources [AASHTO, 2020; ACI Committee 318, 2019; Canadian Standards Association (CSA), 2015; European Committee for Standardization (EC2), 2005; Japan Road Association (JRA), 2012; China Highway Planning and Design Institute (JTG), 2004; Korean Agency for Technology and Standards/Korean Standards Association (KSA), 2015; Ramirez and Russell (NCHRP Report 603), 2008; MOTC, 2009] were computed for a 0.6-in., seven-wire, Grade 270 low- relaxation strand. The development length equations from AASHTO, ACI, JRA, MOTC, and KSA are independent of the concrete strength, while the others include this parameter. The comparison shown in Figure 1.2 was made for f ′c = 5, 10, and 15 ksi, with 70% of these values taken as concrete strength at pretension transfer to compute transfer lengths. The Chinese code (JTG, 2004) caps the transfer and development length calculations at values corresponding to f ′c = 6.3 ksi and f ′c = 7.4 ksi, respectively; as a result, the transfer and development lengths do not change for f ′c = 10 and 15 ksi. For the lower-strength concrete, f ′c = 5 ksi, with the exception of ACI 318 (ACI Committee 318, 2019) and the Taiwanese code (MOTC, 2009), all sources considered require a greater transfer length than that required by AASHTO. For those standards that include the effect of concrete strength in the calculation of transfer length, the transfer length decreases rapidly, below that required by AASHTO. A similar trend is seen in develop- ment length: at lower concrete strengths, other standards (except for Japan Road Association, 2012) require greater development lengths. As concrete strength increases, development length decreases in cases where concrete strength is included in the calculation of development length. The Japanese code (Japan Road Association, 2012) specifies a maximum value of bond strength; hence, the development lengths remain the same for all concrete strengths considered. Considering that 0.7-in. strands are anticipated to be more efficiently used in girders with higher- strength concretes, the dependency of the transfer and development lengths on concrete strength is an important issue that needs to be evaluated. 0 15 30 45 60 75 90 105 120 135 150 165 180 MOTC NCHRP 603 KSA JTG JRA EC2 CSA ACI 318 AASHTO Transfer and development length (in.) Transfer length Development length fc' = 15 ksi fc' = 10 ksi 25db 50db 75db 100db 125db 150db 175db 200db 225db 250db 275db db = 0.6 in. fci' = 0.70fc' fpi = 0.75fpu fpe = 0.54fpu Figure 1.2. Comparison of transfer and development length from U.S. and international sources.

14 Use of 0.7-in. Diameter Strands in Precast Pretensioned Girders 1.5.3.1 NCHRP Report 603 Ramirez and Russell (2008) proposed a revised development length calculation that accounts for the concrete strength at prestress transfer, f ′ci: 120 225 100= ′ + ′     ≥l f f d dd ci c b b Eq.1.2 where the first term is the transfer length: 120 40= ′ ≥l d f dt b ci b Eq.1.3 1.5.3.2 Observed Transfer Lengths and the Effects of Lightweight Concrete The previous discussion is based on the behavior of conventional normal weight concrete (NWC). Lightweight [aggregate] concrete (LWC), however, is receiving greater attention for its ability to reduce the weight of long bridge girders. Although it is occasionally specified, the use of LWC in prestressed bridge girder elements remains relatively uncommon except for very long spans. From a design perspective, the gains made in reduced weight are offset by reductions in concrete strength (Ball, 2019). There is comparatively little study of strand performance when used to prestress LWC. Greene and Graybeal (2019) assembled a data set of 350 NWC and 250 LWC girders to study the effect on transfer length. An analysis of the database was used to develop potential revisions to provisions related to LWC and NWC within Section 5 of the AASHTO LRFD Bridge Design Specifications (AASHTO, 2020). Greene and Graybeal confirmed the AASHTO definition of LWC as “concrete containing lightweight aggregate conforming to AASHTO M195 and having an equilibrium density not exceeding 0.135 kcf, as determined by ASTM C567.” Thus, 135 pcf is the threshold between LWC and NWC. They also confirmed the applicability of AASHTO Eq. 5.4.2.4-1 for modulus of concrete, Ec, as appropriate for unit weights from 90 to 155 pcf: Greene and Graybeal (2019) propose revisions to strand transfer length. They propose revising the NWC transfer length of 60db to that proposed by Ramirez and Russell (2008) for NWC (in. Eq. 1.3) and the following for LWC: = ≥ Eq.1.4220,000 40l d E dt b c b Figure 1.3 shows transfer lengths calculated for unit weights, wc, varying from 0.1 to 0.155 kcf and concrete strengths, f ′c, varying from 5 to 12 ksi. Eq. 1.3 is used for wc ≥ 135 pcf and Eq. 1.4 is applied for wc < 135 pcf. Current AASHTO practice is that lt = 60db. From Figure 1.3 it is observed that the calculated values lt < 60db for all NWC shown (wc > 0.135 kcf) and lt > 60db for LWC having lower density and lower strength. Realistically, for LWC, lt > 40db (at wc = 0.135, fci ≥ 16 ksi is required to achieve lt ≤ 60db). LWC has longer transfer length than NWC, which may exceed 60db in some reasonable cases. The data reported by Greene and Graybeal are mostly for 0.5-in. and 0.6-in. strands. There are some limited data for 0.62-in. and 0.75-in. strands in NWC. No data appear to suggest anything other than the linear correlation between transfer and development lengths and strand diameter implied by the AASHTO equations. 1.5.3.3 Effects of Confining Reinforcement on Strand Development and Transfer Lengths AASHTO LRFD Bridge Design Specifications Article 5.9.4.4.2 (AASHTO, 2020) requires confinement reinforcement around prestressing steel in the bottom flange of I-shaped girders.

Background 15   This confinement must extend at least 1.5d into the span and consist of no less than No. 3 bars spaced at 6 in. There are no known studies that specifically address the provision of prestressing flange confinement beyond the end region of the girder. Most early studies extended confine- ment only 1.0 h, while more recent studies have respected a 1.5-h requirement. A recent review (French, 2019) also concludes that there is no research indicating that confinement should be extended beyond 1.5 h. Russell and Burns (1996) report an extensive study of small-scale tests intended to assess transfer length of 0.5-in. and 0.6-in. strands. In these small-scale tests, no discernible effect on transfer length was attributed to the presence of confining reinforcement although it is noted that the small prismatic specimens were entirely uncracked; indeed, the authors appear to conclude that results from such small-scale tests are not indicative of full-scale beams. Additionally, the authors are careful to note that although confining reinforcement does not appear to affect transfer length, it serves to prevent splitting and should not be eliminated. Simi- larly, Maguire (2009; reported in Patzlaff et al., 2012) reports transfer length tests of single 0.7-in. strands in small prisms having different amounts of confinement distributed along the entire length of the prism. There was no discernible correlation between transfer length and confine- ment. On the other hand, Tadros and Morcous (2011), using prestressed and nonprestressed single 0.7-in. strand pullout tests, identified the beneficial role of confining reinforcement on strand development. Early studies of prestressed bridge girders identified the beneficial effects of providing con- finement reinforcement around the prestressing strands in the flanges on the shear capacity of girders (Csagoly, 1991; Shahawy et al., 1993). Based on a more refined understanding of behavior, the improvement is likely (at least partially) attributable to the improved anchorage of the horizontal tension tie resulting from confinement near the girder end (Shahrooz et al., 2017). Similarly, Shahawy (2001) reports an extensive study in which some girders are provided 5 6 7 8 9 10 11 12 0 10 20 30 40 50 60 70 80 90 100 110 0.1 0.105 0.11 0.115 0.12 0.125 0.13 0.135 0.14 0.145 0.15 0.155 108 98 89 82 75 69 64 59 54 54 54 54 101 92 84 77 70 65 60 56 49 49 49 49 96 87 80 73 67 62 57 53 45 45 45 45 92 84 76 70 64 59 55 51 42 42 42 42 89 81 73 67 62 57 53 49 40 40 40 40 86 78 71 65 60 55 51 47 40 40 40 40 83 75 69 63 58 53 49 46 40 40 40 40 81 73 67 61 56 52 48 44 40 40 40 40 f'ci (ksi) Density (kcf) Tr an sf er le ng th (1 /d b) Figure 1.3. Effect of concrete on the prestressing strand transfer length (data based on Greene and Graybeal, 2019).

16 Use of 0.7-in. Diameter Strands in Precast Pretensioned Girders with No. 3 bars at 6 in. over a distance of 1.0 h and others are not provided with additional flange confinement. The girders having confinement are observed to have greater capacities, and the improvement becomes more pronounced as shear becomes more dominant (i.e., at shorter shear span-to-depth ratios). Ross (2012) also demonstrated the benefit of confinement (as opposed to none) but only extended the confinement a distance of 1.65 h from the girder end. Ross reported a shear capacity increase of 15% and a 200% increase in deformation capacity resulting from providing confinement meeting the requirements of Article 5.9.4.4.2 (NB, this was Article 5.10.10.2 at the time). In small girders tests, Patzlaff (2010) and Tadros and Morcous (2011) assessed the effects of confinement although no strand debonding is considered. Of eight otherwise identical specimens, two had confinement (No. 3 bars at 4-in. spacing) extending h from the girder end although the total confinement capacity was the same as for beams in which confinement extended 1.5 h (No. 3 bars at 6 in.). The results report no premature failures and a marginally improved shear capacity when confinement is extended to 1.5 h compared to 1.0 h. Experi- mentally determined transfer lengths were improved when confinement was provided with reduced spacing although the transfer length in every case was less than h = 24 in. Tadros and Morcous also report tests of NU-1100 girders having different confinement (although only meeting AASHTO LRFD Bridge Design Specifications Article 5.9.4.4.2 require- ments) and 25% debonding at one end of their girders. Capacities are all reported to exceed those predicted, and no strand slip data are reported for debonded strands. Tadros and Morcous conclude that the requirements of Article 5.9.4.4.2 “should be used as a minimum reinforcement at the girder ends to provide anchorage of prestressing steel and reduce the probability of strand slippage at extreme loading conditions. Additional confinement reinforcement placed throughout the entire length of bridge can be used to improve ductility and reduce damage due to over-height vehicular collision.” Russell et al. (1994) reported an experimental program of modest-scale prestressed girders having a range of debonding. They identified that debonded strands have the potential to introduce spalling, splitting, and bursting stresses beyond the locations at which debonding ends. The girder geometry tested did not permit the investigation of confinement reinforce- ment; however, the authors concluded that provided debonding termination points were staggered, as is now required by AASHTO LRFD Bridge Design Specifications Article 5.9.4.4.3 (AASHTO, 2020), bond failures are mitigated. Okumus and Oliva (2014) go further, concluding, based on a numeric study: “At locations where debonding is terminated, cracking is unlikely to occur as long as these locations were spaced at least a transfer length apart from each other.” French (2019) paraphrases this conclusion as follows: “The use of staggered debonding ensures that limited stresses will be introduced in a given section along the girder such that cracking further along the length of the beam from spalling, splitting or bursting stresses is not anticipated.” However, with the use of 0.7-in. strands, the prestress force being introduced at each debonding termination is greater. Russell and Burns (1996) considered only 0.5-in. strands and Okumus and Olivia (2014) modeled 0.6-in. strands. Furthermore, the recommendation for staggering terminations, while sound, has never been contrasted with providing confinement. 1.5.4 Synthesis of Current Design Codes and Design AASHTO and codes in Canada, China, Europe, Japan, New Zealand, South Korea, Taiwan, and the United Kingdom do not provide guidance in terms of spacing and transfer/development length of 0.7-in. strands. As a result, current design provisions must be scrutinized to under- stand if new provisions have to be developed for 0.7-in. strands.

Background 17   1.6 Overview of Bond Characteristics Like any bonded internal reinforcement, stress is transferred between concrete and prestress- ing strand by bond. Bond forces are developed by chemical adhesion, friction, and mechanical interlock between deformed bars and the surrounding concrete (Figure 1.4a) (ACI 408R-03, 2012). Adhesion is small and rapidly overcome and is, therefore, neglected. In prestressing strands, there is a frictional component associated with the Poisson and Hoyer effects associated with prestressing although this occurs over a relatively short length near the end of the embed- ment. The Hoyer effect (see Section 1.7) is understood to improve transfer length behavior but is unlikely to significantly affect ultimate development length. The Hoyer effect is of more interest in terms of the radial forces generated at girder ends and the resulting likelihood of cracking (see Section 2.7). Mechanical interlock is the dominant component of bond strength. Mechanical interlock for prestressing strands is similar to that for deformed bars but also engages a helical twisting action, which generates an enhanced friction component coincident with mechanical interlock. The shallow-angle helical deformations of prestressing strands will result in a smaller radial component of bond stress than conventional deformed bars that have near-transverse-oriented deformations (Figure 1.4a). The kinematics of the seven-wire strand subject to tension is (e) Imprint of 0.6-in. strand observed in this study. (c) Probable normal or shear stress distribution around strand resulting from helical twisting action (Salmons and McCrate, 1973). (d) Imprint (bore) of a strand in concrete (Salmons and McCrate, 1973) (b) Components of bond force acting on the seven-wire strand (Salmons and McCrate, 1973). (a) Components of bond force acting on a deformed reinforcing bar (ACI 408R-03, 2012). Figure 1.4. Bond forces engaged in embedded deformed bars and strands.

18 Use of 0.7-in. Diameter Strands in Precast Pretensioned Girders described by Machida and Durelli (1973) and Utting and Jones (1987). The resulting mechanism of mechanical interlock with concrete is described by Salmons and McCrate (1973) and is shown schematically in Figure 1.4b and Figure 1.4c. The mechanical bond can be envisioned from the imprint of the strand into the surrounding concrete as shown in Figure 1.4d and Figure 1.4e. Out of practical necessity, bond characterization of strand—and comparison between strands or other conditions—is conventionally conducted in the nonprestressed state. The unstressed strand bond is, however, directly relevant in some instances. Prestressing strand extending from the ends of precast girders can be embedded in diaphragms. This may be done to develop a continuous (for live load) structure. Embedding the extensions of a bonded prestressing strand in a diaphragm may allow these strands also to develop greater stresses (fps) when resisting the longitudinal reinforcement requirement arising from shear near the ends of girders expressed by AASHTO LRFD Bridge Design Specifications (AASHTO, 2020) Equations 5.7.3.5-1 and 5.7.3.5-2. 1.7 Hoyer Effect Briere et al. (2013) provide a complete discussion of the “Hoyer effect.” Much of the following discussion is adopted from this reference. The mechanics affecting the transfer of prestressing force are partially a manifestation of the so-called Hoyer effect (Hoyer, 1939; Hoyer and Friedrich, 1939). When a straight wire or “tendon” is placed in tension, the diameter decreases due to the Poisson effect. As the stress is relieved, the tendon attempts to return to its original diameter. When not encased in concrete, the tendon returns to its original diameter; and similarly, where the prestressing force has been developed (in the member beyond the transfer length), the tendon retains its stressed diameter (Figure 1.5). In reality, the prestressing force does not develop linearly over the transfer length (a) Hoyer effect along transfer length (Briere et al., 2013). (b) Representations of stresses and cracking adjacent to the embedded strand (Briere et al., 2013). (c) Average stresses at plane cut through concrete and strand (after Ross, 2012). Figure 1.5. Schematic representation of Hoyer effect.

Background 19   but varies as seen in Figure 1.5a and, therefore, the diameter of the tendon varies similarly. As the tendon stress is reduced (toward the free end of the member) and the tendon attempts to return to its original diameter, radial forces develop along the concrete/tendon interface (Figure 1.5b). The resulting lateral expansion and development of radial forces improve the bond strength (in the sense of transfer of prestress force to the concrete) and assist in transferring the force from the tendon to the concrete. This improvement in bond strength and transfer of tendon prestress force is referred to as the Hoyer effect. For the case of a multiwire tendon (i.e., prestressing strand), an additional “tightening” of the wire bundle making up the strand also accompanies the prestress force. Analogous to the Poisson effect, the strand “unfurls” as stress is relieved. This effect (also conventionally lumped into the term Hoyer effect) also improves bond by generating some radial forces and “locking” in mechanical bond associated with the helically deformed surface of the strand. The radial expansion of the tendon or multiwire strand enhances the bond and, therefore, prestress force transfer to the concrete by (1) exerting a force normal to the strand-concrete interface, enhancing friction, and (2) developing a wedge-like geometry (Figure 1.5a) enhancing the effectiveness of this normal force. The force is developed by the restraining effect of the surrounding concrete. While the radial force is compressive, directed into the concrete, per- pendicular circumferential tensile stresses necessarily develop. These stresses can lead to radial cracking emanating from the strand, which degrades the transfer mechanism. In members having many closely spaced strands, such as bridge girders, the interaction between adjacent strands could result in severely cracked regions and spalling, particularly near the ends of girders (for example, Barnes et al., 1999 and Hamilton, 2009). The practice of partially debonding strands at girder ends partially mitigates the cumulative effect of anchoring multiple strands at the same location. Ross (2012) identifies an additional issue associated with the radial stress induced by the Hoyer effect, referred to as the “combined” condition (described in Figure 2.13c). Strands located along a vertical section produce a “zipper” of sorts. The Hoyer effect places the concrete between strands in tension as it resists the Hoyer-induced compression forces around the perimeter of the strand. 1.7.1 Concrete Stresses Due to Strand Anchorage Oh et al. (2006) performed the theoretical analysis of a single prestressing strand embedded in the center of an isotropic (i.e., uncracked) concrete cylinder having radius c, which would conventionally be taken as the concrete cover measured to the center of the strand (Figure 1.5b). In their simplification, Oh et al. (2006) considered the strand as a solid cylinder having an unstressed radius, r0, and a stressed radius, rs (Figure 1.5). Assuming strain compatibility at the strand-concrete interface, the radial pressure at this interface, p, is obtained as (Oh et al., 2006): [ ] ( ) ( ) ( ) ( ) ( ) = − − − − + − + + 1 1 1 0 0 2 2 2 2 p r v f E r v f E v r E v r c r c r E p pz p s c cz c p p c s s s c Eq.1.5 where r0 = radius of unstressed prestressing strand rs = radius of prestressing strand due to initial prestress force, fpi: ( )= −10r r v f Es p pi p Eq.1.6 vp = dilation (Poisson’s) ratio of prestressing strand Ep = Young’s modulus of prestressing strand

20 Use of 0.7-in. Diameter Strands in Precast Pretensioned Girders fpz = axial stress in prestressing strand a distance z from the free end of the concrete embedment vc = Poisson’s ratio of concrete Ec = Young’s modulus of concrete fcz = axial stress in concrete a distance z from the free end: = 02 2f f r ccz pz Eq.1.7 c = radius of concrete cylinder; c = clear cover + r0 From the interface pressure, the radial (σr) and circumferential stresses (σθ) as a function of the radial distance, r, from the strand axis are determined as (Oh et al., 2006): ( ) ( )σ = − − −1 1 1 12 2 2 2p c r c Rr Eq.1.8 ( ) ( )σ = − + −θ 1 1 1 12 2 2 2p c r c R Eq.1.9 where R is the radial distance to the inner boundary of the isotropic (uncracked) region. For an isotropic concrete cylinder, R = r0, this solution is shown schematically in Figure 1.5b. The stresses calculated at the strand-concrete interface and those near the free end of the embedment (z = 0) are high and will result in radial cracking of the concrete, as shown schemati- cally in Figure 1.5b. The stress decreases quickly with the radial distance from the strand, r. If the concrete cover is sufficient, this behavior results in a cross section having a cracked zone adjacent to the strand and an uncracked annular region surrounding the cracked zone where the stresses are below the tensile capacity of the concrete, that is, anisotropic and isotropic material regions, respectively (Figure 1.5). Oh et al. (2006) present a numeric solution to the resulting “multiphase” problem representing the transition from cracked to partially cracked to fully cracked concrete that occurs along the transfer length of an embedded strand (Figure 1.5). Applying this considerably more complex analysis results in lower calculated stresses; nonetheless, the trends established that consider only the simplified isotropic case are sufficient for the present discussion. The isotropic case is also sufficient to establish the likelihood and extent of cracking associated with the anchorage of prestressing strands (Briere et al., 2013; Ross, 2012). A parametric study of Eq. 1.5 (Briere et al., 2013) revealed that realistic variation of the con- crete parameters υc and c had little effect on the calculated results. The minimal effect of concrete cover, c, was also reported by Ross (2012). Variation of Ec results in an expected linear variation in stress at the strand-concrete interface. Variation of the dilation ratio for the strand, νp, however, has a significant effect on the stresses associated with strand anchorage and, more importantly, the expected extent of the cracked region. 1.7.2 Dilation of Prestressing Strand As described by Hoyer, for a single plain wire, prestress force transfer is affected by the normal stress generated by the radial expansion of the wire and the wedge action resulting from the transfer of prestressing force itself as described in Figure 1.6. In this case, νp is the Poisson’s ratio for the steel wire. The dilation of the seven-wire prestressing strand, however, is not wholly described by Poisson’s ratio of the steel. Rather, in a seven-wire strand placed in tension, the helical outer wires experience some torsion and “tighten,” initially “filling” any annular space in the strand and then bearing on each other at their points of contact. Machida and Durelli (1973) describe

Background 21   the geometry of an unstressed prestressing strand as one in which the helical wires are “slightly elliptical,” resulting in a small clearance between the center wire and the helical wires if the diameters of all wires are the same. It is not clear from Machida and Durelli whether this geometry and the associated annular space apply to both stress-relieved and low-relaxation strand. Briere et al. (2013) contend that the annular space will be negligible in modern low- relaxation strand since this strand is tempered with the strand under tension and the central straight wire has a marginally larger diameter (ASTM, 2015a). The strand tightening and wire interaction result in strand dilation in addition to that result- ing from the Poisson effect present in each wire. The Poisson dilation of the individual wires allows the helical wires to “tighten” further, and the bearing of the outer helical wires on the center straight wire further compresses this center wire. These effects are shown schematically in Figure 1.6. As described above, the stressed radius of the strand, rs, is critical to the stress generated in the concrete. Thus, the total dilation of the strand, accounting for the effects described, is required. This dilation includes Poisson’s effect but is more pronounced due to the “tightening” and bearing effects associated with the seven-wire geometry. These effects can, nonetheless, be described using Eq. 1.5 with νp interpreted as the apparent dilation ratio of the strand, necessarily greater than the Poisson’s ratio for steel. In an attempt to quantify νp, Briere et al. (2013) report an experimental procedure for deter- mining the dilation of prestressing strand; they reported results of tests of 0.5-in. and 0.6-in. low-relaxation strand. Using the same methodology, a limited number of tests on 0.7-in. low- relaxation strands were conducted as part of the development of this project; further tests on 0.7-in. strand are described in Section 3.2. The values of νp = 0.40, 0.34, and 0.32 were found for 0.5-in., 0.6-in., and 0.7-in. strands, respectively. The inverse relationship between seven-wire strand diameter and apparent dilation is attributed to the tightening effect being proportionally greater for smaller helical wires having the same lay angle (Briere et al., 2013), confirming the theoretical geometric analysis of Machida and Durelli (1973). 1.8 Long-Span Girder Stability By using a 0.7-in. strand, span lengths of girder shapes may, theoretically, be increased up to 20% (see Section 2.2). As girders become longer, stability considerations during lifting and handling can begin to control aspects of design. The stability of prestressed concrete girders is (a) Annular space (if any) between wires is first closed. (b) Unstressed, low-relaxation seven-wire strand. (c) Poisson effect results in contraction of wires and tightening of helical wires. (d) Contact stresses, particularly those compressing the center straight strand. (e) Stressed seven-wire strand. Figure 1.6. Contraction behavior of seven-wire prestressing strand (Briere et al., 2013).

22 Use of 0.7-in. Diameter Strands in Precast Pretensioned Girders considered in terms of the potential for rollover and the susceptibility to excessive deformations causing concrete stress limits to be exceeded. Rollover is the rigid body rotation of the girder off its seat. Bracing at girder supports restrains this. Rollover is checked when the girder is seated on cribbing, during transportation, and during erection. Rollover will typically control long girder design (Zureick et al., 2009). Mitigating rollover by providing bracing is relatively straightforward at all handling, transportation, and erection stages; this is commonly done and represents good practice. Excessive lateral displacement—similar in shape/pattern to lateral-torsional buckling (LTB)— causes biaxial bending of the girder, leading to cracking and stresses exceeding those permitted in design. Although the mechanism is the same as LTB, the allowable stress limits for prestressed concrete against which section stresses are checked are lower than those at which a buckling instability would occur, that is, no practical prestressed concrete section will exhibit LTB without first cracking and exceeding allowable stress limits. Thus, satisfying stress checks de facto mitigates LTB. Concrete tension and compression stress limits are those prescribed in AASHTO LRFD Bridge Design Specifications Article 5.9.2.3.1a (AASHTO, 2020). Excessive lateral deflections must be checked for the girder in its seated or hanging conditions. Effects of lateral forces due to wind and centrifugal forces (during road transportation) are included in these checks. Zureick et al. (2009) determined girder stability limits for simply supported AASHTO Type I through Type VI girders having a minimum concrete compressive strength of 6 ksi. The LTB calculations did not consider AASHTO stress limits and assumed that the girder ends are restrained against rollover. Taking Type VI girders as an example, Zureick et al. report that Type VI girders exceeding 140 ft require bracing against rollover and those exceeding 193 ft are susceptible to LTB. In this study (Section 2.2), the maximum achievable length of a Type VI girder with 0.6-in. strands having 15 ksi concrete and girder spacing of 6 ft is 184 ft; with 0.7-in. strands, the maximum achievable length is 202 ft. In either case, the cracking and failure stress limits are likely exceeded before theoretical LTB occurs. The lower rollover limit—well within the practical span range for a Type VI girder—reinforces the need to provide appropriate bracing at girder supports. 1.8.1 PCI Method of Stability Analysis The prestressed girder stability analysis approach prescribed by PCI (2015) was followed in this study. The fundamental steps of a stability analysis are as follows: 1. Determine girder geometry (Figure 1.7a) and concrete material properties at each erection stage; typically, these will be early age properties. 2. Consider all other factors affecting stability analysis, including camber, prestress force, lateral wind pressure, centrifugal force during transportation, etc. 3. Determine the factor of safety (FS) for stability for conditions causing cracking, failure, and rollover: =FS M M resisting acting Eq.1.10 The PCI procedure is programmed in an Excel spreadsheet (PCI, 2019). The version of the spreadsheet used in this study was revised by the research team to address several programming errors found in the original version.

Background 23   1.8.2 Hanging Girders The hanging girder condition corresponds to that in which a girder is supported (by a crane) from above by cables attached to the girder web or top flange. Support is provided near each girder end (Figure 1.7b). For long-span girders, lift cables are always vertical (as opposed to inclined as when using a basket or bridle arrangement). The critical lift condition will be when the girder is supported at only two locations (as opposed to when multiple spreaders are used to effect a multipoint lift). Therefore, only the critical two-lift point conditions were considered in this study. Two factors of safety are calculated when evaluating the stability of a hanging girder. The factor of safety against cracking, which must exceed 1.0, is calculated as: = θ θ + − +0 FS y z z e e cr r max max wind wind i Eq.1.11 Note: Figure from Recommended Practice for Lateral Stability of Precast, Prestressed Concrete Bridge Girders, http://doi.org/10.15554/CB-02-16, used by permission. (a) Geometry and free-body diagram of rotated girder. (b) Hanging girder. (c) Geometry and free-body diagram of transportation vehicle rollover. (d) Geometry and free-body diagram of a rotated girder on a transportation vehicle. Figure 1.7. Equilibrium of prestressed girders (PCI, 2015).

24 Use of 0.7-in. Diameter Strands in Precast Pretensioned Girders The factor of safety against failure, which must exceed 1.5, is calculated as: ( )( ) ′ = ′ θ ′θ + + ′θ − +1 2.50 FS y z z e e r max max wind max wind i Eq.1.12 where, regarding Figure 1.7a, yr = distance of center of gravity of girder from roll axis; θ = rotation angle of the girder from vertical; z0 = theoretical lateral deflection of the center of gravity of the beam with full deadweight applied laterally; zwind = lateral deflection due to wind force; ewind = eccentricity of girder weight due to wind force; and ei = lateral eccentricity due to sweep and eccentricity due to lifting device from the centerline. The rotation at which a girder cracks is given as: ,θ = M M max y crack z Eq.1.13 where My,crack = lateral moment applied to the girder that causes tensile cracking in the most critical flange, and Mz = gravity moment of the girder. For a cracked section, the lateral stiffness of the girder is reduced, causing more deflection under the same load. Mast (1993) defines the maximum tilt angle for the cracked section, θ′max, as: 2.5 0′θ = e zmax i Eq.1.14 1.8.3 Seated Girders The seated girder condition corresponds to that in which a girder is supported from below during transportation, storage, or on site (Figure 1.7c and Figure 1.7d). Three factors of safety are calculated for evaluating the stability of a seated girder. The factor of safety against cracking, which must exceed 1.0, is calculated as: 0( )( ) ( ) = θ − α + θ + + − + θFS K W z y z e z M cr max r max wind i CE ot Eq.1.15 The factor of safety against failure, which must exceed 1.5, is calculated as: 1 2.50( )( ) ( ) ( ) ′ = ′θ − α ′θ + − + ′θ + + ′θ + θFS K W z z z e y M max max wind CE max i r max ot Eq.1.16 The factor of safety against rollover, which must exceed 1.5, is calculated as: 1 2.50( )( ) ( ) ( ) = ′′ θ − α ′′θ + − + ′′θ + + ′′θ + θFS K W z z z e y M roll max max wind CE max i r max ot Eq.1.17 where, regarding Figure 1.7c and Figure 1.7d, Kθ = rotational constant of the transportation rig or support condition; α = maximum roadway superelevation or cross slope of the support condition; W = weight of girder; z0 __ = lateral deflection of centroid of the cross section; and zCE = deflection of girder due to centrifugal force (when present; see below).

Background 25   The rotation angles θmax and θ′max are those given by Eq. 1.13 and Eq. 1.14, respectively. The critical rotation angle is calculated as: ( ) ( )( )′′θ = − α − − + α + α θ W z h WS CE h z K max max r r max Eq.1.18 where WS = wind force. 1.8.4 Additional Lateral Forces during Transportation PCI (2015) describes the superposition of wind (WS) and centrifugal force (CE) on a girder during the transportation stage. However, in a conventional arrangement of a vehicle traveling around a curve, the CE component is counteracting the effect of the superelevation (in Figure 1.7d, CE acts to push the girder to the right). Notes related to PCI (2019) indicate that two cases should be checked: 1. A transportation vehicle stopped on a superelevated curve. In this case, CE = 0 since the design speed is also zero. 2. A transportation vehicle traveling around a curve with adverse superelevation such as when making a left turn. In this case, for the very long girders considered in this study, one would anticipate speed to be very low and controllable; hence, once again CE ≈ 0. Therefore, in this study focused on very long girders, only wind (WS) loads are applied in the transportation stage analysis. 1.8.5 Interpreting and Revising Stability Analyses In this study, several issues arose while interpreting the results of stability analyses and considering measures to mitigate unstable girder conditions. As demonstrated in the following sections, many interacting parameters affect stability. Some are easily revised in the field; these parameters can be more easily (and rationally) varied to meet the factor of safety requirements. In preliminary analyses, negative values of the factor of safety were obtained in some instances. Such results appear to indicate cases in which the girder exceeds stress limits when subjected to only the non-stability-related loads. Essentially, such girders are not meeting their design requirements before being evaluated for stability. This situation arises most often if the assumed girder support locations are such that the overhang is too long, resulting in excessive cantilever stresses. In such a case, adjusting the support location parameter, a (distance from the end of the girder), is required. As designed, an in situ girder is intended to be supported from below at its ends (assumed to be a = 6 in. in this study, representing a 12-in.-long bearing with no overhang) and is designed accordingly. However, when supported on dunnage or by a crane, support locations may be varied to improve the stability of the girder—reducing internal stresses and mitigating rollover. For example, a girder that fails to meet requirements for rollover can be braced when placed on its support. Using a shorter span, however, may permit the girder to be supported on dunnage without the need for bracing. In this study, with very long girders, the support locations are found to be crucial to ensuring adequate stability. Following an initial assumption, the value of a can be revised until adequate factors of safety are achieved. It should be noted that the value of a during transportation may be limited by the interaction of vehicle and roadway geometry; specifically, the arc “swept” out by the overhanging end of

26 Use of 0.7-in. Diameter Strands in Precast Pretensioned Girders the girder. Even for very long girders, the value of a during the transportation stage is likely limited. In this study, for transportation, a is limited to 20 ft. Transverse support stiffness is also a critical parameter, especially during the transportation phase: the so-called hauling rig stiffness, Kq, trans parameter (PCI, 2015 and 2019). This parameter can be schematically represented by the springs shown beneath the girder support in Figure 1.7c and Figure 1.7d and is assumed to include the combined effects of girder support and vehicle parameters affecting rotational stiffness such as wheelbase width and vehicle suspension char- acteristics. There is little guidance available for calculating the transverse support stiffness provided by a transport vehicle. For more common length girders, using conventional flat-bed trucks, Kq,trans ≈ 40,000 kip − in./rad appears to be recommended (PCI, 2019). For the very long girders considered in this study, a special trailer would be required. The special trailer will inevitably have a greater rotational stiffness since both the girder bearing width and vehicle wheel- base will be wider. Upon consultation with practitioners, a value of Kq,trans ≈ 82,000 kip − in./rad was selected as an initial value. In some analyses of very long (and heavy) girders, Kq,trans had to be increased further to ensure stability. Additional calculations required for stability analysis are described in Section 2.8 using specific girder examples.