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.

103   Experimental Research Approach and Findings 4.1 Introduction This chapter reports the second phase of the experimental study in which full-scale girders having 0.7-in. strands were cast and tested. Twelve full-scale girders having 0.7-in. strands were cast and tested to (1) determine transfer length (AASHTO LRFD Bridge Design Specifications Article 5.9.4.3.2); (2) evaluate development length, particularly for partially debonded strands (AASHTO LRFD Bridge Design Specifications Articles 5.9.4.3.2 and 5.9.4.3.3); (3) understand flexural and shear behavior (AASHTO LRFD Bridge Design Specifications Section 5); (4) investigate potential interaction between 0.7-in. strands spaced on a 2-in. grid (AASHTO LRFD Bridge Design Specifications Article 5.9.4.1); (5) evaluate current detailing requirements for end-region vertical splitting reinforcement (AASHTO LRFD Bridge Design Specifications Article 5.9.4.4.1) and bottom flange confinement reinforcement (AASHTO LRFD Bridge Design Specifications Article 5.9.4.4.2); and (6) examine the applicability of current design procedures (AASHTO LRFD Bridge Design Specifications Section 5) (AASHTO, 2020). An overview of the test specimens is provided in Table 4.1. The first five girders were designed to allow testing of both ends of each beam. The following seven girders, which were shorter, were tested once. Hence, 17 sets of data were used to evaluate the performance of pretensioned girders with 0.7-in. strands. 4.2 Girder Details Four shapes were tested: (1) T-beam, (2) Iowa BTB-36, (3) AASHTO BI-36, and (4) NU-1100; the dimensions of these shapes are shown in Figure 4.1. To expedite construction, the box girders were produced without shear keys, which should not affect the test results. All specimens used normal weight concrete (NWC). The design concrete strength was 10 ksi for all specimens except G11 and G12, which were designed using 15-ksi concrete. Nonprestressed reinforcement was typically ASTM A615 Grade 60. To ensure flexural failure, ASTM A1035 Grade 100 stirrups were used in girders G7, G8, G10, and G12 to obtain greater shear capacity without significant steel congestion. The specimens were designed according to the 9th edition of AASHTO LRFD Bridge Design Specifications (AASHTO, 2020) with the following exceptions: • For end B of girders G3–G5, (1) a revised version of the Gergely-Sozen (1967) model (see Appendix F) was used to calculate the required splitting reinforcement, and (2) the bottom flange confinement reinforcement was determined using the strut-and-tie model (STM) described in Appendix G. • The bottom flange confinement reinforcement in girders G6 and G8–G10 was designed according to the needs of this project; this is described in Section 4.6.3 and Section 4.6.4. C H A P T E R   4

104 Use of 0.7-in. Diameter Strands in Precast Pretensioned Girders Girder ID Length (ft) Girder shape Target failure mode Debonding (both ends) Splitting reinforcement Confinement reinforcement Design (ksi) End A End B End A End B 10 G1 46 T-beam Bond/flexure 0% AASHTO AASHTO AASHTO AASHTO 10 G2 46 T-beam Bond/flexure 33% for 3 ft AASHTO AASHTO AASHTO AASHTO 10 G3 45 Iowa BTB-36 Bond/flexure 0% AASHTO G-Sa AASHTO STM 10 G4 55 Iowa BTB-36 Bond/flexure 14% for 3 ft AASHTO G-S AASHTO STM 10 G5 55 Iowa BTB-36 Bond/flexure 43% for 3 ft29% for 3 ft to 6ft AASHTO G-S AASHTO STM 10 G6 30 Iowa BTB-36 Shear 33% for 6.5 ft AASHTO AASHTO Revised AASHTOb Revised AASHTO 10 G7 30 Iowa BTB-36 Bond/flexure 33% for 6.5 ft AASHTO AASHTO AASHTO AASHTO 10 G8 30 Iowa BTB-36 Bond/flexure 33% for 6.5 ft AASHTO AASHTO Revised AASHTO Revised AASHTO 10 G9 30 AASHTO BI-36 Shear 36% for 3 ft 21% for 3 ft to 6.5 ft AASHTO AASHTO Revised AASHTO Revised AASHTO 10 G10 30 AASHTO BI-36 Bond/flexure 36% for 3 ft 21% for 3 ft to 6.5 ft AASHTO AASHTO Revised AASHTO Revised AASHTO 10 G11 30 NU-1100 Shear 40% for 3 ft20% for 3 ft to 6.5 ft AASHTO AASHTO Revised AASHTO Revised AASHTO 10 G12 30 NU-1100 Bond/flexure 40% for 3 ft20% for 3 ft to 6.5 ft AASHTO AASHTO Revised AASHTO Revised AASHTO 15 aGergely-Sozen model. b“Revised AASHTO” indicates that confinement reinforcement extended to 1.5 beyond the location where the longest debonded strands begin to be bonded. Splitting reinforcement for all specimens is based on AASHTO LRFD Bridge Design Specifications Article 5.9.4.4.1 (AASHTO, 2020) unless noted. NWC was used in all specimens. Table 4.1. Overview of test specimens with 0.7-in. strands. Figure 4.1. Cross sections of test girders (6-in. slab added to NU-1100 and BTB-36 girders not shown).

Experimental Research Approach and Findings 105   With the exception of girders G1–G4 (T-beams) and G9 and G10 (box girders), a 6-in.-thick cast-in-place composite slab was added to the girders before testing. The slab reinforcement was determined using the empirical design process (AASHTO LRFD Bridge Design Specifica- tions Article 9.7.2 in AASHTO, 2020). Specimen details are shown in Figure 4.2 to Figure 4.9. All strands are 0.7-in. diameter. Bent reinforcing bar details are provided in Figure 4.10. Additional specimen details along with photographs of representative reinforcing cages are provided in Appendix H. Girder G1 Girder G2 #4 stirrups @ 6″ DEBOND STRAND 3′-0″ FROM END Figure 4.2. Specimen details— G1 and G2. Figure 4.3. Specimen details—G3. Figure 4.4. Specimen details—G4.

106 Use of 0.7-in. Diameter Strands in Precast Pretensioned Girders Figure 4.5. Specimen details—G5. Figure 4.6. Specimen details—G6. Figure 4.7. Specimen details—G7 and G8.

Experimental Research Approach and Findings 107   Figure 4.8. Specimen details—G9 and G10. Figure 4.9. Specimen details—G11 and G12. Figure 4.10. Bent reinforcing labels.

108 Use of 0.7-in. Diameter Strands in Precast Pretensioned Girders Girder ID Release At girder test Age (days) f 'ci (ksi) End Girder age at test (days) Compressive strength (ksi) (ASTM C39) Splitting tensile strength (ksi) (ASTM C496)Girder Deck slab Cylinder age at test (days) Measured Cylinder age at test (days) Measured Cylinder age at test (days) Measured G1 4 4.24 A 44 39 8.37 N/A 57 0.563 B 46 45 G2 6 5.64 A 41 40 9.08 N/A 41 0.545 B 44 43 9.23 44 0.578 G3 2 9.23 A 161 156 12.8 127 10.7 Not measuredB 169 167 12.8 134 10.2 G4 1 8.43 A 215 209 11.3 161 9.90 215 0.701 B 194 188 12.2 182 9.03 194 0.830 G5 1 8.43 A 248 250 10.3 217 9.98 250 0.705 B 241 G6 1 8.01 N/A 23 24 12.3 15 6.51 Not measured G7 1 8.01 N/A 29 30 11.8 22 6.39 30 0.692 G8 1 8.01 N/A 36 37 12.0 29 7.04 37 0.713 G9 1 7.65 N/A 28 32 10.5 N/A 32 0.642 G10 1 7.65 N/A 34 34 9.87 34 0.607 G11 2 12.7 N/A 35 36 16.7 15 6.27 36 0.783 G12 2 12.7 N/A 42 42 17.9 42 0.868 Girder ID Bara Material testing fy (ksi) fu (ksi) G1, G2 #3 74.2b 105b #4 66.0 97.4 G3, G4, G5 301, 302 67.6 111 401, 402, 403 66.0 97.4 501 68.8 107 G6, G7, G8 301, 302 76.3 106 403 66.0b 97.1b 502 106 151 G9, G10 304 to 310 76.6 108 404 to 410 122 155 G11, G12 311 76.6 108 411 65.1 95.6 503 64.1 93.5 504 109 153 aSee Figure 4.10 and Appendix H for the bar labels. bThese bars were not tested; the reported values are from mill reports. Table 4.2. Concrete material properties. Table 4.3. Reinforcing bar material properties. 4.2.1 Material Properties The material properties for concrete (compressive and splitting tensile strength) and reinforcing bars (yield and ultimate strength) used for test specimens are summarized in Table 4.2 and Table 4.3, respectively. Concrete mix designs are provided in Appendix I. Strand properties were reported previously in Table 3.1. Reported concrete properties are the aver- age values of three specimens; reinforcing bar properties are the average of five specimens.

Experimental Research Approach and Findings 109   Reported nominal specimen capacity is based on measured values of material properties (mill report values are used in the cases in which measured values for reinforcing steel were not obtained). 4.3 Transfer Length Transfer length was established by measuring concrete surface strains over a distance from the end of each girder before and after prestress release. A demountable mechanical (DEMEC) strain gauge was used for this purpose. Previous researchers (e.g., Russell and Burns, 1996) epoxied DEMEC targets to the concrete surface after removing the side forms. Although this time-consuming process of attaching DEMEC targets may be feasible in a laboratory setting, it is not practical in the field. At the time of the release of prestressing, the girder is usually still warm from the heat of hydration and steam curing (when used). The forms insulate the girder to some extent and once the forms are stripped, the girder begins to cool. If the strands are not cut soon after the forms are removed, the tensioned strands restrain the cooling girders, and the girders may crack; there is insufficient time to install surface-mounted DEMEC gauges in this case. Therefore, the research team developed an apparatus to expedite the installation of DEMEC targets; this is described in Appendix J. Strands were flame detensioned in all girders except G1 and G2. In those two girders, grinding wheels were used to cut the individual strands simultaneously at both ends. Before detensioning, a high-resolution caliper, designed specifically for DEMEC targets, was used to measure the distance between DEMEC targets attached to the concrete surface, as shown in Figure 4.11. The distances were measured again following prestress release. The difference between the readings was used to determine concrete surface strains associated with prestress transfer and thereby determine the transfer length. The 1.97-in. (50-mm) target spacing is used with a 5.91-in. (150-mm) gauge length DEMEC caliper to obtain overlapping strain data on 2-in. centers. The measured “raw” strains were “smoothened” by using the algorithm shown in Eq. 4.1 (Ramirez-Garcia et al., 2018) and plotted as a function of distance from the girder end. In this equation, ε0raw, ε1raw, and ε2raw are the measured strains at the first three DEMEC targets. These readings were measured to obtain ε1smooth, which was used to determine transfer length. To obtain Figure 4.11. Operation of DEMEC strain gauge.

110 Use of 0.7-in. Diameter Strands in Precast Pretensioned Girders the smooth strains at location i (εismooth), the measured values at this location (εiraw) and the values at one location before (εi−1raw) and one location after (εi+1raw) were averaged. The smooth data at the last location n (εnsmooth) were obtained by averaging the measured strain at this location (εnraw) and the preceding value (εn−1raw). Eq. 4.1 3 for 1 3 for1 1 2 for 1 0 1 2 1 1 1 i i n i n smooth raw raw raw i smooth i raw i raw i raw n smooth n raw n raw ε = ε + ε + ε = ε = ε + ε + ε < < − ε = ε + ε = − + − A representative plot that demonstrates the approach used to determine the transfer length from the smoothed data is shown in Figure 4.12. This figure represents the data for end B of G1. The measured concrete strain gradually increases linearly before reaching a relatively constant value at the end of the transfer length. The extent of the linear best fit (passing through the origin as shown in Figure 4.12) was established by maximizing the R2 of the data along the transfer length. The remaining data points (to the right in Figure 4.12) were averaged to obtain the average maximum strain (AMS). The intersection of the resulting best-fit line and 95% AMS corresponds to the transfer length (Russell and Burns, 1996). The measured development lengths are summarized in Table 4.4. None of the measure- ments were deemed to be outliers, and all data points were used for each beam. The mean and 5th percentile transfer lengths, calculated with 95% confidence, were determined to be 36.9db and 50.3db, respectively. The mean value is appreciably lower than 60db specified in AASHTO LRFD Bridge Design Specifications (AASHTO, 2020) and, despite the scatter, the 5th percentile value is also below this value. Figure 4.12. Illustration of measured concrete strain to obtain transfer length.

Experimental Research Approach and Findings 111   Transfer lengths were also computed using the equation from NCHRP Report 603 (Ramirez and Russell, 2008) as shown in Eq. 4.2. Eq. 4.240 120l maximum of d and d f t b b ci = ′     Using the release strengths ( f ′ci) shown in Table 4.2 (repeated in Table 4.4), the value of transfer length from NCHRP Report 603 equation was calculated and is tabulated in Table 4.4. The average value of transfer length calculated from this equation is 44db, which is between the mean and 5th percentile transfer lengths obtained from the experimental data. Calculation of the distribution of stresses at release depends on the assumed transfer length. The tensile stresses at the top of the girder section calculated based on the AASHTO transfer length of 60db would be smaller (unconservative, in this case) than the values calculated using shorter measured transfer lengths. As a result, unanticipated cracking could occur even though the calculations based on 60db suggest tensile stresses are at or lower than permitted values. Because of the measured transfer lengths, use of the NCHRP Report 603 equation (Eq. 4.2) is more appropriate than the 60db value specified in AASHTO LRFD Bridge Design Specifications (AASHTO, 2020). 4.4 Loading Each girder was instrumented and tested monotonically to failure. All girders were tested over a simple span in three-point bending. Both ends of girders G1–G5 were tested individu- ally as shown in Figure 4.13a. An air jack was used to balance the weight of the overhang to remove the effect of the overhang weight on the bending moment in the test span and to prevent negative moment cracking at the central support. Both ends of girders G6–G12 were tested simultaneously in a simple span arrangement, as shown in Figure 4.13b. Girder ID End Aa End B f'ci (ksi) NCHRP Report 603 G1 46db 37db 4.24 58db G2 48db 34db 5.64 51db G3b NA NA 9.23 40db G4 36db 30db 8.43 41db G5 33db 20db 8.43 41db G6 31db 22db 8.01 42db G7 19db 20db 8.01 42db G8 31db 12db 8.01 42db G9 41db 37db 7.65 43db G10c NA NA 7.65 43db G11b 27db NA 12.7 40db G12 23db 30db 12.7 40db aEnd A corresponds to the “live end,” except for girders that were cast simultaneously in the same bed (G4–G12). In those cases, end A corresponds to the end that was at or closest to the live end, depending on the position of the girder in the bed. bDue to instrumentation error, transfer lengths from G3 and end B of G11 could not be measured. NA = Not available. cThe devices holding the weld nuts (see Appendix J) were damaged as the side form was removed so transfer lengths could not be measured. Table 4.4. Summary of measured transfer lengths.

112 Use of 0.7-in. Diameter Strands in Precast Pretensioned Girders The location of load (Le measured from the end of the girder) was varied to determine strand development length and induce different modes of failure (flexure/bond or shear). The location of the load point (Le) was also selected to ensure that the shear span-to-depth ratio (a/dv) exceeded 2 (see Table 4.5). In all cases, the shear span, a, is 6 in. (one-half the bearing pad length) shorter than Le. Development length was calculated using the following two methods: AASHTO (2020) , = κ − 2 3 κ =1.6 for full-length bonded strands, κ = 2.0 for partially debonded strands Eq. 4.3 NCHRP Report 603 (Ramirez and Russell, 2008) , = 120 + 225 ≥ 100 Eq. 4.4 In Table 4.5, the location of load (Le) is also reported in terms of distance x at which all strands in the section (full-length bonded and partially debonded) are assumed to be developed for design purposes. Thus, a value of Le/x > 1 indicates that the strands should be fully developed for flexure based on the location of the load. For Le/x < 1, the strands are not fully developed based on the development lengths calculated using Eq. 4.3 and Eq. 4.4. Furthermore, the ratio Le/x represents the proportion of flexural capacity that should be attainable based on the method of determining development length. Considering the NCHRP Report 603 development equation (Eq. 4.4), the load was applied at a location where all strands were not fully developed in girders G2, G4, and G5 in which the ratio Le/x ranged between 0.54 and 0.98. Based on the AASHTO calculated development length (Eq. 4.3), no girders tested had all strands fully developed at the load point, with the ratio Le/x ranging between 0.29 and 0.92. 4.4.1 Support Conditions Girders G1 and G2 were supported on 3-in.-thick neoprene pads with two layers of internal steel plates. For the other girders, 13⁄8-in.-thick neoprene pads with a single steel plate layer (a) Girders G1–G5. (b) Girders G6–G12. Figure 4.13. Loading and test setups.

Experimental Research Approach and Findings 113   were used. Except for girders G3, G4, and G5, for which the bearing pad extended over only 53% of the bottom flange width, at least 75% of the bottom flange width of the other single- web girders was supported, as shown in Figure 4.14. Narrower supports were used in G3–G5 to study the potential influence of using methods other than AASHTO to detail the ends (i.e., Gergely-Sozen for splitting reinforcement and an STM for bottom flange confinement reinforcement). Two 7-in.-wide bearing pads were placed under each 5-in.-wide web of girders G9 and G10. All the supports were 12 in. long and located at the ends of the girders, resulting in the test spans (L′) shown in Table 4.5. Girder ID Test ID (ft) ′ (ft) (ft) / / a NCHRP Report 603 AASHTO G1 G1A 46 24 13'-3" 5.77 1.67 0.92G1B 24 8'-0" 3.85 1.01 0.56 G2 G2A 46 24 7'-0" 3.33 0.68 0.37G2B 24 5'-6" 2.56 0.54 0.29 G3 G3A 45 26 6'-6" 2.00 1.09 0.52G3B 26 6'-6" 2.00 1.09 0.52 G4 G4A 55 32 9'-0" 2.86 0.98 0.46G4B 32 9'-0" 2.86 0.98 0.46 G5 G5A 55 32 9'-0" 2.87 0.72 0.40G5B 32 9'-0" 2.87 0.72 0.40 G6 G6 30 29 14'-6" 4.73 1.14 0.63 G7 G7 30 29 13'-1" 4.25 1.03 0.57 G8 G8 30 29 13'-1" 4.25 1.03 0.57 G9 G9 30 29 14'-4⅜" 7.66 1.09 0.72 G10 G10 30 29 13'-4" 7.09 1.01 0.57 G11 G11 30 29 14'-4" 3.98 1.16 0.63 G12 G12 30 29 12'-4¼" 3.41 1.00 0.54 a is the distance computed by each of the two methods given by Eq. 0.45 and Eq. 0.46, at which all strands (full-length bonded and partially debonded) are expected to be developed. Table 4.5. Span, calculated development lengths, and location of load point. Figure 4.14. Placement of bearing pads.

114 Use of 0.7-in. Diameter Strands in Precast Pretensioned Girders 4.5 Instrumentation In each beam, strain gauges were bonded to multiple strands as well as confining and shear reinforcement. The locations of all strain gauges are provided in Appendix E. Additionally, external sensors were installed to measure (1) applied loads, (2) vertical deflections at three locations along the span and at the bearings, (3) strand slip (Appendix K), (4) average shear strain, and (5) concrete compressive strain at approximately 1 in. below the top of the girder. The external instrumentations are shown in Figure 4.15 and Figure 4.16. 4.6 Test Results and Discussions The test results of the girders are presented and discussed in five groups: (1) G1 and G2; (2) G3, G4, G5, and G7; (3) G6 and G8; (4) G9 and G10; and (5) G11 and G12. These groups were selected based on the differences in (1) girder shapes, (2) methodologies used for calcu- lating the required amount of splitting and detailing of bottom flange confinement reinforce- ment, and (3) concrete strengths. Girder G7 is grouped with G3–G5 because the splitting and confinement reinforcement were determined according to AASHTO in these girders although a larger portion of the bottom flange was supported in G7 (see Figure 4.14). As part of evaluating performance, the expected capacity of each girder was computed by the following two approaches. The measured material properties were used in both approaches. 1. Method 1: This analysis was conducted using a bespoke spreadsheet developed by the research team that had also been used to design the test specimens. This spreadsheet accounts for the development of strand stress based on a selected development length equation. For analysis Method 1, AASHTO development length (Eq. 4.3) was used. All reduction factors were taken as unity because the test girders were fabricated under controlled conditions, the loading was well defined and known a priori, and the purpose of the calculation was to determine a predicted capacity rather than a design capacity. The capacity obtained from this method is referred to as the AASHTO capacity. 2. Method 2: Response 2000 (Bentz, 2000) was used for the second analysis in which all strands were assumed to be fully developed. This program is based on the modified com- pression field theory and accounts for shear-moment interaction. The capacity obtained from this method is referred to as All-strands capacity. Figure 4.15. Schematic layout of external instrumentation.

Experimental Research Approach and Findings 115   (a) Measurement of support vertical deformation. Pressure transducer (c) Measurement of applied force. (d) Measurement of vertical deflection. (e) Measurement of concrete surface strain. (f) Measurement of average shear strain. (b) Measurement of strand slip. Figure 4.16. Photographs of typical external instrumentation.

116 Use of 0.7-in. Diameter Strands in Precast Pretensioned Girders The applied shear or load point deflections corresponding to slip of full-length bonded or partially debonded strands are presented in the following discussions. These values are average slip measurements from the instrumented strands, which were in the same row or two rows.* The following discussion presents applied shear (i.e., reaction at bearing nearest applied load) versus deflection (at the location of applied load) plots for each test; capacity comparisons are summarized for all specimens in Table 4.6 (see page 134). 4.6.1 Group 1: Girders G1 and G2 As evident from Figure 4.17, failure of the four tests (with Le/x ranging between 0.29 and 0.92) was flexure-dominated in each case. In the case of G2B, which had one partially debonded strand (33%), one full-length bonded strand fractured followed by fracture of the second full-length bonded strand (see Figure 4.17d), thus indicating that these strands were fully developed. *The instrumented strands are shown in Appendix K. (a) G1A. (b) G1B. (c) G2A. (d) G2B. fractured strands Figure 4.17. Failure mode of girders G1 and G2.

Experimental Research Approach and Findings 117   The applied shear versus deflection at the point of load application is compared for all tests in Group 1 in Figure 4.18. The capacities based on Method 1 (labeled AASHTO capacity) and Method 2 (labeled All-strands capacity) are also plotted in this figure. The occasional drops seen in this figure occurred while loading was paused to mark and document cracks. In this figure, shears corresponding to 0.01-in. and 0.10-in. strand slip, if observed, are indicated. The value of 0.10-in. slip was selected for consistency with ASTM A1081 (2015b) discussed in Section 3.4, and 0.01-in. slip was considered to be the smallest value that could be reliably measured (even though the sensors were calibrated to 0.001 in.). These threshold values have also been used by others (Morcous et al., 2012). The following observations are based on the applied shear-deflection curves: 1. The capacity determined based on the AASHTO development length equation (AASHTO capacity) could be developed and reached in all specimens. This observation is noteworthy considering that for test G2B, the value of Le/x (Le is the location of applied load measured from the end of girder, and x is the distance at which all strands are fully developed) was only 0.29. Accounting for the two full-length bonded strands and one partially debonded strand, AASHTO development equation indicates that 67% of the total strand area was available to resist the applied load for the selected location of the load. However, the beam resisted 1.58 times the capacity based on analysis Method 1, suggesting the conservative nature of the AASHTO development equation. (a) G1A (b) G1B (c) G2A (d) G2B Figure 4.18. Applied load versus load point deflection: G1 and G2.

118 Use of 0.7-in. Diameter Strands in Precast Pretensioned Girders 2. Except for test G2B, both girders resisted and exceeded the capacity calculated assuming all strands were fully effective (All-strands capacity). The first occurrence of 0.01-in. slip of the debonded strand in G2A was beyond the “yield point,” defined as the point at which the shear-deflection curve became nonlinear. After experiencing 0.10-in. slip, the shear-deflection curve plateaued, but the girder could still resist additional deflection. The debonded strand continued to provide resistance despite its slipping. A similar observation regarding the continued contribution of slipping strands to flexural capacity has been made by others (e.g., Naito et al., 2015). Loading was stopped after the beam deflection reached 3.5 in. (L′/83). 3. Once again, in test G2B the debonded strand slipped 0.01 in. after the “yield point” and attained a slip of 0.10 in. The debonded strand continued to provide resistance despite slipping as indicated by the additional load that was resisted. The north full-length bonded strand fractured shortly after slipping 0.01 in., resulting in a sudden drop of the applied shear. The maximum measured girder resistance was 94% of the capacity predicted based on the full development of all strands. The girder exhibited residual capacity before the south full- length bonded strand fractured at which point loading was stopped. The sensor measuring slip of the debonded strand was removed after reaching a 0.40-in. slip at a beam deflection of 1.35 in. (L′/213). Therefore, the slip of the 3-ft debonded strand was not measured after the second full-length bonded strand fractured. 4. For test G2B, the debonded strand slipped 0.01 in. after reaching and exceeding the capacity determined with the AASHTO development length equation. This observation reaffirms the assessment that the AASHTO development length equation is conservative. 4.6.2 Group 2: Girders G3, G4, G5, and G7 4.6.2.1 Applied Shear versus Deflection The applied shear versus deflection at the load point is shown in Figure 4.19 for this group of tests. The expected capacities calculated by Method 1 and Method 2 are also provided in this figure. Splitting reinforcement and amount of confinement reinforcement met AASHTO LRFD Bridge Design Specifications (AASHTO, 2020) for tests G3A, G4A, and G5A whereas the girder ends for tests G3B, G4B, and G5B had been designed according to the Gergely-Sozen (1967) model [Appendix for an STM (Appendix G)]. In all cases, girder ends B have more splitting and confinement reinforcement than required by AASHTO LRFD Bridge Design Specifications (AASHTO, 2020). The following observations are made from the shear-deflection curves: 1. None of the girders could develop the capacity corresponding to the full development of full-length bonded and partially debonded strands (All-strands capacity) although the maximum shear for tests G3B and G4B was close to this capacity. 2. For girder G3, which did not have any partially debonded strands, the capacity computed based on Method 1 (AASHTO capacity) was exceeded. The load resisted in test G3B is larger than that in test G3A, as evident by the maximum applied shear being 70% and 99% of All-strands capacity for test G3A and test G3B, respectively. The full-length bonded strands in test G3A slipped 0.01 in. at a shear force close to AASHTO capacity. In test G3B, a 0.01-in. slip occurred at a noticeably larger applied shear. At both end A and end B, the girder lost its load-carrying capacity at or shortly after the 0.10-in. slip. 3. Similar to girder G3, shear resisted by girder G4 exceeded the AASHTO capacity. Similarly, test G4B resisted a larger load, for which the maximum applied shear reached 96% of All-strands capacity in comparison to 88% for test G4A. For test G4A, the strands debonded 3 ft from the end slipped 0.01 in. at almost the AASHTO capacity. In contrast, 0.01-in. slip for the same strands occurred at a larger applied shear in test G4B. After the 3-ft debonded strands slipped 0.10 in., the capacity began to drop for test G4A; however, the girder exhibited

(a) G3A (b) G3B (c) G4A (d) G4B (e) G5A (f) G5B (g) G7 Figure 4.19. Applied load versus load point deflection: G3, G4, G5, and G7.

120 Use of 0.7-in. Diameter Strands in Precast Pretensioned Girders a ductile behavior in test G4B, i.e., the capacity did not drop as the deflection continued to increase. The strength for test G4A degraded at a more severe rate once the full-length bonded strands slipped 0.10 in. In contrast, the strength for test G4B did not drop appre- ciably after 0.10-in. slip of the full-length bonded strands. Loading was stopped after the deflection at the load point reached 3 in. (L′/128) to avoid excessive damage that could affect the subsequent test of G4A. 4. The load resisted by both ends of girder G5 marginally exceeded the AASHTO capacity. The maximum shear was 1.11 and 1.05 times AASHTO capacity for test G5A and G5B, respectively. For test G5A, the full-length bonded strands and those debonded 3 ft and 6 ft from the end slipped 0.01 in. at a shear that was very close to AASHTO capacity. The partially debonded strands (3-ft and 6-ft debonded) slipped 0.10 in. almost simultaneously, and the resistance gradually began to drop thereafter. The full-length bonded strands did not slip during test G5B; however, the partially debonded strands (3-ft and 6-ft debonded) slipped 0.01 in. at a shear less than AASHTO capacity. The 3-ft and 6-ft debonded strand slipped 0.10 in. at slightly below AASHTO capacity. The girder exhibited significant ductility with little strength degradation during both tests although the strength was lower in comparison to girders G3 and G4. 5. Girder G7, which was tested once, had a behavior similar to that of girder G5. The maximum shear was 94% of the capacity corresponding to all strands being fully developed at the load point, which was greater than that achieved by girder G5 (77% and 73% for test G5A and G5B, respectively). Only the strands debonded 6.5 ft from the end slipped more than 0.01 in. These strands slipped 0.01 in. and 0.10 in. at nearly the same load. The strength degraded gradually after 0.10 in. slip and “stabilized” at almost the AASHTO capacity. The deflection at the load point continued to 2.9 in. (L′/120) at an applied shear of 198 kips when the displacement transducer was lost. 4.6.2.2 Failure Modes Examining failure modes provides a tool to better understand the responses captured in the applied shear-deflection curves (Figure 4.19). Girder 3 is discussed separately from girders G4, G5, and G7 in the following. 4.6.2.2.1 Girder G3. Significant diagonal cracks and damage extended into the support region for test G3A, for which the end had been detailed according to AASHTO LRFD Bridge Design Specifications (AASHTO, 2020) (see Figure 4.20a). For the other end having more end region reinforcing detailed using the Gergely-Sozen formulation (Appendix A) and STM approach (Appendix B), damage during test G3B was concentrated within the shear span with a few diagonal cracks extending toward the girder end. The reinforcement at the end tested in G3A was insufficient to control the growth of cracks near the support, which eventually led to loss of anchorage of strands and a brittle failure. On the other hand, the additional vertical and confining reinforcement at the end of G3B maintained the load-carrying integrity near the support, resulting in a combination of shear-tension and shear-compression failure at a larger load than test G3A. It should be noted that the bearing pads in girder G3 supported only about one-half the width of the bottom flange (see Figure 4.14). Importantly, the bearing pads did not support the portion of the bottom flange containing strands, as shown in Figure 4.21. Calculations using the strut-and-tie approach (Harries et al., 2019) indicate that the failure observed in test G3A would most likely have been prevented had the bearings been wider, supporting nearly the full width of the bottom flange as recommended by Shahrooz et al. (2017). 4.6.2.2.2 Girders G4, G5, and G7. Photographs of damage patterns are provided in Figure 4.22, Figure 4.23, and Figure 4.26 for girders G4, G5, and G7, respectively. The soffit of

Experimental Research Approach and Findings 121   (a) G3A (detailed based on AASHTO). (b) G3B (detailed based on Gergely-Sozen and STM). Figure 4.20. Failure of girder G3. Figure 4.21. Support width for girder G3.

122 Use of 0.7-in. Diameter Strands in Precast Pretensioned Girders these girders experienced extensive cracking and damage. Each girder is discussed separately in the following. 1. Girder G4. In test G4A, extensive damage occurred on the sides of the bottom bulb. Upon unloading, the cover concrete spalled between approximately 5 ft and 8 ft from the end of the girder (Figure 4.22a) and the partially debonded (debonded for 3 ft) and full- length bonded strands were found to have “buckled.” This behavior is an indication that at the end of the test all strands had experienced considerable slip and could not return to their original position upon unloading the girder. On the other hand, the girder expe- rienced only diagonal shear cracks and primarily longitudinal cracks on the soffit in test G4B (Figure 4.22b). The cracks on the soffit could not be marked during loading, and the photos shown were taken after the beam had been unloaded. Loading was stopped after the deflection at the load point reached 3 in. (L′/128) in test G4B to avoid excessive damage that could affect G4A, which was tested subsequently. Therefore, it is possible the soffit would have experienced similar extensive damage as occurred in test G4A had the loading been continued. 2. Girder G5. The crack and failure patterns after both tests (G5A and G5B) were similar (Figure 4.23). The bottom bulb near the load point was damaged significantly. Upon unload- ing, the cover spalled off over approximately 3 ft. The debonded strands, which had slipped appreciably, buckled upon the elastic rebound of the girder as seen from Figure 4.24. The large level of slip is attributed to loss of cover confinement (as evident from spalling) and not to “poor” bond quality; adequate bond is suggested by the imprints of strands in the concrete seen after spalling of the concrete cover (Figure 4.25). 3. Girder G7. Failure of this girder is attributed to the formation of a strut extending from the loading point to the point at which the confinement reinforcement had been terminated (see Figure 4.26a). This strut, which had begun to form before reaching the peak load, pushed the strands downward, resulting in spalling of the soffit (Figure 4.26b). The downward strut force caused a noticeable permanent deformation and bending of debonded strands, which occurred on both sides, while the full-length bonded strands maintained their original position (Figure 4.26c). The excessive damage and spalling in the soffits of girders G4, G5, and G7 are attributed to the relative lack of bottom flange confinement reinforcement. This is discussed in Section 4.6.4. 4.6.3 Group 3: Girders G6 and G8 According to AASHTO LRFD Bridge Design Specifications Article 5.9.4.4.2 (AASHTO, 2020), “For the distance of 1.5d from the end of the beams other than box beams, reinforcement shall be placed to confine the prestressing steel in the bottom flange. The reinforcement shall not be less than No. 3 deformed bars, with spacing not exceeding 6.0 in. and shaped to enclose the strands.” Girders G4, G5, and G7 met this requirement. After examining the behavior and failure modes of these girders, however, the research team hypothesized that the larger 0.7-in. strand requires the bottom flange confinement reinforcement to be extended further into the beam for partially debonded strands. To evaluate this hypothesis, the bottom flange confinement reinforcement was extended to 1.5d beyond the location where the longest debonded strands begin to be bonded in girders G6 and G8 instead of 1.5d from the end of the beam (in G4, G5, and G7). As evident from Figure 4.27, both girders reached and developed the capacity corresponding to all strands being developed (labeled as All-strands capacity). Moreover, the capacity of G6 and G8 was 1.40 and 1.21 times their AASHTO capacity, respectively. For both girders, strands slipped above the “yield point,” defined as the point at which the shear-deflection curve became nonlinear.

Experimental Research Approach and Findings 123   Side (a) G4A. Side Bottom flange Soffit (Grid line numbers are distances in ft from the end near the load point.) (b) G4B. Figure 4.22. Crack and failure patterns for girder G4.

124 Use of 0.7-in. Diameter Strands in Precast Pretensioned Girders Side (a) G5A. Side Soffit (Grid line numbers are distances in ft from the end near the load point. Note that 2 out of 14 strands were debonded up to grid line 6.) Soffit (Grid line numbers are distances in ft from the end near the load point. Note that 2 out of 14 strands were debonded up to grid line 6.) (b) G5B. Figure 4.23. Crack and failure patterns for girder G5.

Experimental Research Approach and Findings 125   (a) G5A. (b) G5B. Figure 4.24. Buckling of debonded strands in girder G5 after unloading (note that full-length bonded strands remain straight in each image). Figure 4.25. Imprints of strands in concrete seen after spalling of cover in girder G5. The full-length bonded and partially debonded strands in girder G6 slipped 0.01 in. This girder failed soon after the full-length bonded strand slipped 0.01 in.; however, this slip did not initiate failure. As seen from Figure 4.28, girder G6 failed in shear, as intended. This observation is further seen from Figure 4.29 in which the loading curve corresponding to the maximum applied shear is compared against the shear-moment interaction diagram (generated using Response 2000). The loading curve is the relationship between the applied moment (M) and shear (V), i.e., M = (a − dv)V in which a is the shear span and dv is effective shear depth computed based on AASHTO LRFD Bridge Design Specifications (AASHTO, 2020). The value of shear depth is subtracted from the shear span to reflect the disturbed region under the applied load. This method is applied in Response 2000. The loading curve reaches the shear-moment interaction diagram before reaching the maximum moment capacity. The maximum shear resisted by G6 was 9% larger than the calculated capacity. This difference is well within the expected variability of shear strength of prestressed beams.

126 Use of 0.7-in. Diameter Strands in Precast Pretensioned Girders (a) Overall view. (b) Bottom flange from below. Support is to the left, and loading point is to the right where the top flange is crushed/spalled. (c) Permanent deformation of debonded strands. Figure 4.26. Crack and failure patterns for girder G7.

Experimental Research Approach and Findings 127   (a) G6 (b) G8 Fractured stirrup Figure 4.27 Applied load versus load point deflection: G6 and G8. Figure 4.28. Failure of girder G6. Figure 4.29. Evaluation of failure mode of girder G6.

128 Use of 0.7-in. Diameter Strands in Precast Pretensioned Girders The failure mode of Girder G8 was a “typical” flexural failure, i.e., exhibiting large deflec- tion before compression failure of the top flange as shown in Figure 4.30. Girder G8 reached 1.02 times All-strands capacity followed by a gradual loss of capacity. The strands debonded for 6.5 ft from the end slipped 0.01 in. at 97% of AASHTO capacity. These strands slipped 0.10 in. essentially at the All-strands capacity (1.2 times AASHTO capacity). The beam exhibited excel- lent ductility, and 0.10-in. slip did not result in failure, which was observed in the girders with a shorter extension of bottom flange confinement reinforcement (Figure 4.19). 4.6.4 Effect of Extent of Bottom Flange Confinement Reinforcement Girders G7 and G8 were identical except for how far the bottom flange confinement reinforce- ment extended along the beam. The benefits of extending confinement reinforcement 1.5d beyond the location where the longest debonded strands begin to be bonded (G8) versus 1.5d from the end of girder (G7) are evident by comparing shear-deflection curves (Figure 4.31) and modes of failure (G7: Figure 4.26, G8: Figure 4.30). The difference between these two girders is further discussed below. The relationships between the applied shear and deflection at the load point for girders G7 and G8 are compared in Figure 4.31. This figure demonstrates these two girders performed differently. This difference is attributed to the extension of bottom flange confinement in girder G8. Figure 4.30. Failure mode of girder G8. Figure 4.31. Comparison of the overall behavior of girders G7 and G8.

Experimental Research Approach and Findings 129   The force in the confinement reinforcement was determined from the strain data (from which stress was calculated as discussed in Appendix F). Figure 4.32 shows the confinement reinforcement force corresponding to the peak applied load at various locations along the girder length. The data suggest the confinement reinforcement was engaged for girder G8 at the critical section near the support but was not for girder G7. This trend is consistent with the observed cracking also shown in Figure 4.32: diagonal cracks extended to the top of the bottom bulb of girder G8, but the bulb did not crack. Well-distributed cracks and lack of observable permanent deformation in girder G8 are in contrast to girder G7 in which the bottom bulb experienced significant cracking (Figure 4.32). The force spiked near the termi- nation point of the confinement reinforcement in girder G7, reaching 4.5% Vmax where Vmax is the maximum shear resisted, but the confinement reinforcement force at the same location was essentially zero in girder G8. A smaller spike, equal to 2.9% Vmax, occurred for girder G8 near where the confinement reinforcement was terminated further into the span. Girder 8: 23 #3 confinement bars @ 6" Girder 7: 10 #3 confinement bars@ 6" Girder 7 Girder 8 Extent of confinement reinforcement (Girder 7) Extent of confinement reinforcement (Girder 8) End of debonded strands 2.9% of Vmax 4.5% of Vmax C on fin em en t r ei nf or ce m en t f or ce (k ) 0 1 2 3 4 5 6 7 8 9 10 Distance from end (in.) 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 Figure 4.32. Variation of bottom flange confinement reinforcement force along the girder length and damage pattern.

130 Use of 0.7-in. Diameter Strands in Precast Pretensioned Girders The spike for girder G7 occurred within the region with 33% debonded strands whereas in girder G8 the confinement reinforcement force spiked once all the strands were bonded. The introduction of stresses from the debonded strands is essentially the same as that of full-length bonded strands at the girder ends. Prestress transfer results in Hoyer effect expansion at the location at which strand bonding begins, whether at the girder end or into the span for partially debonded strands. At the girder ends, confinement is required. The same Hoyer effect occurs, although to a lesser degree (due to concrete confinement at both sides of the section of initial bonding), at the end of the debonded strand. In the past, the use of smaller strands and limits on debonded strands kept these bursting forces small. Using larger strands and more debonding will increase these forces and, thus, the need for confinement. Although the cracking may not occur at strand release (as happens at girder ends), the stress is still there, and the effects may not actually become significant until the girder is loaded close to its ultimate load-carrying capacity. 4.6.5 Group 4: Girders G9 and G10 In both box beam specimens, the bottom flange confinement reinforcement was extended to 1.5d beyond the location where the longest debonded strands begin to be bonded. As seen from applied shear-deflection curves (Figure  4.33), in both cases, the full-length bonded strands and those debonded for 3  ft from the end of the girder did not slip. The strands debonded for 6.5-ft at each end slipped 0.01 in. after the specimen reached the “yield point,” defined as the point at which the shear-deflection curve became nonlinear. For girder G9, both 0.01-in. and 0.10-in. slip occurred after the AASHTO capacity was reached (Figure 4.33a). After 0.10 in. slip, the girder stiffness became negligible; the girder failed after undergoing an additional deflection of 1.5 in. The 6.5-ft debonded strand slipped only 0.01 in. for girder G10. This small slip occurred at 95% of AASHTO capacity (Figure 4.33b). The analyses conducted using Response 2000 indicate a combination of shear and flexural failure for girder G9 as the loading curve reaches the shear-moment interaction envelope and maximum moment almost simultaneously, as shown in Figure 4.34a. The total web thickness for the box girder is 10 in., which is relatively large in comparison to the total member depth (27 in.). Although this girder had been designed to fail in shear, it can be argued that the failure mode is dominated by flexure if the dip in the shear-moment interaction (point A) is neglected. However, the loading curve is also very close to the interaction diagram at point B. It is, therefore, not easy to clearly distinguish between flexure and shear failure modes. Post-failure photographs (Figure 4.35a) also suggest a combination of the failure of the top flange and web. (a) G9 (b) G10 Figure 4.33. Applied shear versus load point deflection: G9 and G10.

Experimental Research Approach and Findings 131   Nevertheless, the actual girder capacity was 99% of the All-strands capacity and 1.38 times AASHTO capacity. Considering the challenges of modeling the shear behavior of pretensioned girders, a 1% difference between the measured and calculated shear capacities is not considered significant. Failure of girder G10 was flexure-dominated, which was according to its design. This failure mode is evident from (1) observing that the loading curve reaches the shear-moment inter- action curve first in pure flexure (with zero shear) (see Figure 4.34b) and (2) by formation of an obvious plastic hinge at the load point (Figure 4.35b). The capacity of girder G10 was 1.02 and 1.15 times the calculated All-strands capacity and AASHTO capacity, respectively. 4.6.6 Group 5: Girders G11 and G12 These girders were designed to evaluate the performance of pretensioned girders with high- strength concrete (at least 15 ksi) and 0.7-in. strands. Similar to group 4 girders and girders G6 and G8, the bottom flange confinement reinforcement was extended to 1.5d past the location where the longest debonded strands begin to be bonded. (a) Girder G9. (b) Girder G10. (a) Girder G9. (b) Girder G10. Figure 4.34. Shear-moment interaction diagram and loading curve for girders G9 and G10. Figure 4.35. Post-failure photographs of girders G9 and G10.

132 Use of 0.7-in. Diameter Strands in Precast Pretensioned Girders The applied shear-load deflection curves (Figure 4.36a) indicate the applied shear in girder G11 reached 1.02 times All-strands capacity and 1.20 times AASHTO capacity before failing in shear-tension (Figure 4.37a) shortly after the full-length bonded strands slipped 0.10 in. The mode of failure is consistent with that predicted by modeling the girder with Response 2000: the loading curve reaches the shear-moment interaction curve before reaching the maximum moment capacity (see Figure 4.38a). The strands debonded for 3-ft and 6.5-ft slipped 0.01 in. at nearly 97% of AASHTO capacity (the 3-ft debonded strands slipped slightly sooner), but the full-length bonded strands slipped the same amount at 1.09 times the AASHTO capacity. The 3-ft debonded strands slipped sooner than those debonded 6.5 ft from the end, particularly at 0.10-in. slip. This apparent anomaly cannot be explained with the available data. Consistent with design, girder G12 failed in flexure, as evident from the large number of flexural cracks (Figure 4.37b) and by recognizing the loading line reaches the shear-moment interaction first in pure flexure (Figure 4.38b). Despite the flexural mode of failure, a portion of the web close to the load point spalled after the formation of diagonal cracks. The maximum shear resisted by girder G12 was 1.20 times AASHTO capacity and 98% of All-strands capacity, as seen from the applied shear-deflection curve shown in Figure 4.36b. In girder G12, the strands debonded for 3-ft and 6.5-ft slipped 0.01 in. at almost the same shears, both of which occurred below the AASHTO capacity (59% and 62% of the capacity for the strands debonded for 3 ft and 6.5 ft from the end, respectively). The full-length bonded strand experienced 0.01-in. slip once the applied shear reached 97% of AASHTO capacity. The debonded strands slipped 0.1 in. at or slightly above AASHTO capacity, at 1.01 times AASHTO capacity for 3-ft debonded strands, and at AASHTO capacity for strands debonded 6.5 ft from the end. The full-length bonded strands slipped 0.10 in. on the descending branch of the shear- deflection curve when the deflection reached 4.37 in. (L′/80). 4.7 Assessment of Development Length and Capacity Calculation Based on AASHTO As discussed previously, the measured material properties were used to calculate capacity based on the following two methods. 1. AASHTO LRFD Bridge Design Specifications (AASHTO, 2020) in terms of transfer and devel- opment length (Article 5.9.4.3), longitudinal reinforcement requirement (Article 5.7.3.5), and shear resistance (Article 5.7.3.3). AASHTO de facto accounts for shear-moment (a) G11 (b) G12 Figure 4.36. Applied load versus load point deflection: G11 and G12.

Experimental Research Approach and Findings 133   (a) Girder G11. (b) Girder G12. Towards load point Figure 4.37. Post-failure photographs of girders G11 and G12.

134 Use of 0.7-in. Diameter Strands in Precast Pretensioned Girders interaction by using longitudinal strain to determine β and θ for calculating concrete shear strength and θ for the shear resistance of transverse reinforcement. 2. Response 2000 (Bentz, 2000) in which all the strands were assumed to be fully developed. These capacities, which were discussed in Section 4.6, are tabulated in Table 4.6 along with the loading location (Le; see Figure 4.13) normalized by x, the distance at which all strands are fully developed. Distance x was established using the development equation in AASHTO LRFD Bridge Design Specifications (AASHTO, 2020) and the equation provided in NCHRP Report 603 (Ramirez and Russell, 2008) given in Eq. 4.3 and Eq. 4.4. The results for G3A–G5B and G7 are excluded because the failure of these specimens has been attributed to inadequate extension of bottom flange confinement reinforcement, dis- cussed in Section 4.6.4. These results are discussed in Section 4.8. In all the tests, the load was applied at a location where the strands would not be fully developed according to the AASHTO development length equation (Le/x < 1). Except for tests G2A and G2B, the load was applied at or beyond the location where the strands would be fully developed based on the NCHRP Report 603 development equation (Le/x > 1). It should be noted that no additional strength beyond what all strands can provide is achieved if Le/x is greater than 1. (a) G11. (b) G12. Figure 4.38. Shear-moment interaction diagram and loading curve for girders G11 and G12. Test ID Measured/Calculated / AASHTO All strands AASHTO NCHRP Report 603 G1A 1.16 1.04 0.92 1.67 G1B 1.59 1.18 0.56 1.01 G2A 1.61 1.11 0.37 0.68 G2B 1.58 0.94 0.29 0.54 G6 1.40 1.09 0.63 1.14 G8 1.21 1.02 0.62 1.03 G9 1.38 0.99 0.72 1.09 G10 1.15 1.02 0.57 1.01 G11 1.32 1.02 0.63 1.16 G12 1.20 0.98 0.54 1.00 Average 1.36 1.04 COV 22.3% 7.4% Table 4.6. Summary of measured versus calculated capacity and location of the loading point.

Experimental Research Approach and Findings 135   Assuming all the strands are fully developed, the ratio of measured to calculated capacity is greater than 1 for all the tests except for G2B, G9, and G12; however, the average value of this ratio is 1.04 with a coefficient of variation of 7.4%. On other hand, the calculated capacities based on the AASHTO development length equation underestimated the specimen capacity resulting in the measured to calculated capacity ratio being above 1 for all the tests including G2B, G9, and G12. For test G2B, the load was applied at 54% of the NCHRP Report 603 development length; therefore, it is expected that not all the strands could be fully developed although the mea- sured capacity achieved was still 96% of that calculated, assuming all the strands are developed. The T-beam used in test G2B had three strands, with one of them being partially debonded. The specimen during test G2B experienced a single dominant crack at the load point (Figure 4.17d), which eventually led to fracture of the two full-length bonded strands. For this lightly reinforced girder, the full-length bonded strands were clearly developed. Shear behavior and capacity were the focus of test G9. The measured capacity was 99% of the capacity calculated based on the development of all strands. Considering the challenges of modeling the shear behavior of prestressed girders, it can be argued that the specimen ostensibly reached the capacity corresponding to all strands being developed. The concrete strength for test G12 was 17.9 ksi. A comprehensive material characterization of such a high-strength concrete has been shown to be difficult. Although Response 2000 is an excellent program for analysis of prestressed members, it has not been fully calibrated to model very high-strength concrete such as that in girder G12. Considering the challenges of model- ing very high-strength concrete (fib Bulletin 42, 2008) and the fact that the measured capacity was just 2% lower than the predicted capacity of all strands, it is the research team’s contention that, again, ostensibly the strands were fully developed, with the load being applied at the NCHRP Report 603 development length. Given the tabulated ratios of measured to calculated capacity, the development length calculated from NCHRP Report 603 is more representative of the development length observed for 0.7-in. strands. While it is applicable to 0.7-in. strands, the AASHTO development equation provides a conservative prediction. 4.8 Impact of Extension of Bottom Flange Confinement Reinforcement on Capacity The measured responses for girders G3A–G5B and G7 (Section 4.6.2.1) and their modes of failure (Section 4.6.2.2) indicate an important influence of the length over which bottom flange confinement is provided. This influence is further observed by comparing the load at the onset of 0.01-in. and 0.1-in. slip. The values of the maximum measured shear normalized with the capacity calculated based on all strands being developed, i.e., Method 2 discussed in Section 4.6, are shown in Figure 4.39 (the bars in this figure are graphical representations of the “All strands” column in Table 4.6). Figure 4.39 also shows the normalized shears at the onset of 0.01-in. and 0.1-in. slip for fully bonded strands, strands debonded for 3 ft from the end of the girder, and strands debonded for 6 ft or 6.5 ft from the end of the girder. As expected, the fully bonded strands slipped at higher shears than the partially debonded strands. Generally, the fully bonded strands slipped 0.1 in. at, or nearly at, the maximum measured shear. By extending the bottom flange confinement 1.5d beyond the location where the longest debonded strands begin to be bonded, strand slippage was delayed. Table 4.7 shows the shear at which the fully bonded strands slip 0.01 in. and 0.1 in. was increased, on average, 22% and 20%, respectively, by extending the bottom flange confinement beyond the current length of 1.5d from the end

136 Use of 0.7-in. Diameter Strands in Precast Pretensioned Girders (a) At 0.01 in. slip. (b) At 0.1 in. slip. Note: Gray bars (G3A–G5B, G7) are those girders having confinement reinforcement extending only from the girder end while the white bars (G6, G8–G12) are those girders having confinement extending beyond the location where the longest debonded strands begin to be bonded. Figure 4.39. Normalized shear capacity and shear at the onset of 0.01-in. and 0.1-in. slip. ⁄ (0.01-in. slip) Bonded 3-ft debonded 6/6.5-ft debonded Currenta 0.77 0.74 0.74 Extendedb 0.95 0.84 0.80 % increase 22% 14% 8% ⁄ (0.1-in. slip) Bonded 3-ft debonded 6/6.5-ft debonded Current 0.84 0.81 0.79 Extended 1.01 0.94 0.98 % increase 20% 15% 24% aBottom flange confinement extended to 1.5d from the end of the girder. bBottom flange confinement extended to 1.5d from the location where the longest debonded strands begin to be bonded. Table 4.7. Comparison of normalized shear at the onset of 0.01-in. and 0.1-in. slip.

Experimental Research Approach and Findings 137   of the girder. Slippage of strands debonded for 3 ft and 6 or 6.5 ft from the end of the girder was also delayed by extending the bottom flange confinement beyond AASHTO LRFD Bridge Design Specifications Article 5.9.4.4.2 (AASHTO, 2020). On average, the debonded strands slipped at 8% to 24% larger shears when the length over which the bottom flange confinement reinforcement is extended (Table 4.7). 4.9 Example Highlighting Additional Extension of Bottom Flange Confinement The impact of extending bottom flange confinement to 1.5d beyond the termination of the longest debonded length was illustrated by examining one of the design case study girders (discussed in Section 2.2). The case was 110-ft-long NU-1100 girders spaced at 12 ft with 15-ksi concrete and 0.7-in. strands. As seen from Figure 4.40, 19 additional sets of confinement reinforce- ment data are required to meet the proposed specification, nothing more. 4.10 State of Practice on Bottom Flange Confinement To gauge the practicality of extending bottom flange confinement beyond the current require- ment, the research team conducted a brief survey of state departments of transportation regarding confinement reinforcement per AASHTO LRFD Bridge Design Specifications Article 5.9.4.4.2 (AASHTO, 2020). Thirty responses were received. The research team supplemented data for the states not responding, adding an additional nine sets of data. The survey questions and verbatim responses are provided in Appendix M. Thirty (77%) states, for which data are available, apply AASHTO LRFD Bridge Design Speci- fications Article 5.9.4.4.2 without revision (Figure 4.41). New York and Pennsylvania report standard spacing details that require No. 4 bars rather than No. 3 bars. Four states (California, Colorado, Missouri, and Nebraska) report confinement details (sometimes relaxed from that required in first 1.5d) extended over the entire beam length; Pennsylvania reports extension over one-third of the beam length. There are contradictory data regarding two states: Delaware and West Virginia. In their survey responses, Delaware and West Virginia indicated no difference from AASHTO. However, from standard details, it appears that both Delaware and West Virginia use the overall beam height rather than depth. Additionally, Delaware requires confinement to extend for 1.5 times the height of the beam when 0.5-in. strands are used and 2 times the height of the beam when 0.6-in. strands are used. 4.11 Summary of Full-Scale Girder Test Results and Observations The focus of the second phase of the experimental program was on the evaluation of full-scale prestressed girders with 0.7-in. strands. Seventeen sets of data were generated through the testing of 12 girders. The following observations are made through synthesis of the measured responses and observations of the performance of the test girders: 1. The mean and 5th percentile transfer lengths with 95% confidence of 0.7-in. strands were determined to be 36.9db and 50.3db, respectively. The mean value is appreciably less than 60db specified in AASHTO LRFD Bridge Design Specifications Article 5.9.4.3.2 (AASHTO, 2020). The average value of transfer length from NCHRP Report 603 equation was found to be 44db. Overestimation of transfer length underestimates concrete tensile stresses at release in the area affected by the transfer length, which could potentially result in cracking even though

138 Use of 0.7-in. Diameter Strands in Precast Pretensioned Girders Figure 4.40. Comparison of current and proposed extent of bottom flange confinement reinforcement.

Experimental Research Approach and Findings 139   the calculations based on a larger transfer length indicate the stresses are within the limits specified in AASHTO LRFD Bridge Design Specifications (AASHTO, 2020). Considering transfer length is a part of development length, a two-tier approach may be used: a reduced transfer length to check tensile stresses at prestress release and the longer development length to determine design load-carrying capacity. 2. The development length as calculated by AASHTO LRFD Bridge Design Specifications Articles 5.9.4.3.2 and 5.9.4.3.3 (AASHTO, 2020) is conservative. NCHRP Report 603 provides a more accurate estimate of the development length. It should be noted that the NCHRP Report 603 and AASHTO equations with κ = 1 are both representations of the mean value. By introducing κ = 1.6 or 2.0, as specified by AASHTO LRFD Bridge Design Specifications Articles 5.9.4.3.2 and 5.9.4.3.3 (AASHTO, 2020), the calculated development length is shifted toward a 95 percentile or better value. However, the use of κ is understood to capture some observed outlier data. Assuming the intent is to design using some “characteristic value,” the appropriate value of development length should be determined from statistical analysis of the body of work. This goal may be achieved from either the AASHTO or NCHRP Report 603 equations although the NCHRP Report 603 equation is deemed to be more rigorous as it includes the effect of concrete strength. 3. The 2-in. minimum center-to-center strand spacing, specified in AASHTO LRFD Bridge Design Specifications Article 5.9.4.1 (AASHTO, 2020) for up to 0.6-in. strands, was found to be adequate for 0.7-in. strands both in terms of release stresses and development of strands. No local cracking or other deleterious effects associated with strand spacing were observed. 4. Vertical splitting reinforcement (AASHTO LRFD Bridge Design Specifications Article 5.9.4.4.1 in AASHTO, 2020) is sufficient for 0.7-in. strands to resist bursting stresses at release. 5. The amount of confinement reinforcement (minimum of No. 3 at a maximum spacing of 6 in.) as required by AASHTO LRFD Bridge Design Specifications Article 5.9.4.4.2 (AASHTO, 2020) was found to be sufficient to confine 0.7-in. strands. The resulting reinforcement may, however, not be sufficient for cases with support-bearing widths covering less than 75% of the bottom flange width, particularly if a wheel load is close to the support. Therefore, confinement reinforcement required to resist potential lateral splitting of the bottom flange (and associated loss of strand anchorage) at the ultimate limit state should be checked using the strut-and-tie approach discussed in Section 2.4.1. The required amount of confining reinforcement is the Figure 4.41. Summary of survey responses and supplemental data.

140 Use of 0.7-in. Diameter Strands in Precast Pretensioned Girders greater of that determined using the strut-and-tie approach and that required by AASHTO LRFD Bridge Design Specifications Article 5.9.4.4.2 (AASHTO, 2020). 6. The extension of bottom flange confinement reinforcement (AASHTO LRFD Bridge Design Specifications Article 5.9.4.4.2 in AASHTO, 2020) was found to be inadequate for cases with partially debonded 0.7-in. strands. Extension of bottom flange confinement reinforcement to 1.5d beyond the end of the girder (AASHTO LRFD Bridge Design Specifications Article 5.9.4.4.2 in AASHTO, 2020) was adequate for cases with no debonded strands. Extension of bottom flange confinement reinforcement to 1.5d beyond the location where the longest debonded strands begin to be bonded was adequate for cases with partially debonded 0.7-in. strands. 7. The test specimens did not have many debonded strands. Within the scope of the presented experimental data, AASHTO LRFD Bridge Design Specifications Article 5.9.4.3.3 (AASHTO, 2020) is adequate for the detailing of partially debonded 0.7-in. strands. 8. Based on the reported tests, current procedures for calculating flexural and shear capacity [AASHTO LRFD Bridge Design Specifications Section 5 (AASHTO, 2020)] for girders with 0.7-in. strands were found to be adequate.