Cast-in-Place Concrete Connections for Precast Deck Systems (2011)

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.

162 Chapter 6 PCSSS: Subassemblage Investigation of Crack Control Reinforcement 6.0 Introduction Seven specimens were designed to investigate the effect of spacing, size, and placement of transverse reinforcement on the development and propagation of reflective cracking in the precast composite slab span system. This chapter describes the design and cracking behavior of these seven subassemblage specimens. Each specimen in this section is generally referenced by the specimen number and representative design parameter, i.e., Specimen 1, the first control specimen, is commonly referenced as: SSMBLG1-Control1. 6.1. Selection and Design of Laboratory Subassemblage Specimens Each subassemblage specimen was selected to investigate a single attribute (e.g., spacing of transverse reinforcement, bar size, absence of reinforcement cage, specimen depth) that was expected to influence the crack control performance of the system. The specimens were designed as simply-supported panel elements with a 10 ft. span with the precast joint located at midspan, as illustrated in Figure 6.1.1. The widths of the specimens (i.e., in the direction parallel to the precast joint) ranged from 62.75 to 67.25 in. and was dependent on the transverse reinforcement spacing. Each specimen was designed such that the center of the transverse reinforcement closest to one face was located a distance of 3.5 in. from that face, defined as the origin face, except for in SSMBLG6-Frosch where the center of the hoop reinforcement was located 3.25 in. from the face. The specimen width was selected such that the opposite face, designated as the end face, was located a distance of half of the transverse reinforcement spacing from the center of the nearest transverse hooked pair, or in the case of SSMBLG6-Frosch it was the distance from the end face to the center of the nearest cage hoop. This configuration enabled the investigation of crack width both at a constant distance from the interior reinforcement, as well as an estimate at the maximum crack width, which was expected to occur at midpoint between adjacent bars. The specimen supports were parallel to the longitudinal joint between the precast elements, rather than transverse to the joint, in order to flexurally crack the specimens along the joint. It was found during the testing of the first specimen, that the stiff flanges of the precast section rotated and caused delamination between the precast flange and CIP concrete, resulting in propagation of a crack at the precast-CIP concrete interface. The test setup was modified to clamp the precast flanges to the CIP concrete a distance of approximately 1.25 in. from the longitudinal joint in both directions. This test setup with the clamping system was believed to more realistically emulate the field conditions because in a bridge system, the pier supports are normal to the longitudinal joint and would constrain the relative rotation between the precast flange tips and the CIP at the ends of the span. The selection of the reinforcement details for each specimen was completed using a range of sources, primarily Frosch (2006), AASHTO (2010), and ACI (318-08). The design parameters of each subassemblage specimen are summarized in Table 6.1.1. The reinforcement ratio shown in Table 6.1.1 is the reinforcement ratio defined for crack control. This value accounts for all reinforcement traversing the longitudinal joint located near the bottom of the trough area near the precast flanges, which accounts for contribution from both the transverse hooked bars and the reinforcing cage. The reinforcement ratios for load transfer and crack control are discussed in detail in Section 5.1.

163 The subassemblage specimens were designed to provide insight into the relative performance of variations in the transverse reinforcement details for crack control and load transfer. For this reason, the specimen designs were limited to the transverse reinforcement selection. The primary depth of the precast members used for the subassemblage specimens was selected to be 12 in. to maintain consistency with the bridge specimens. The depth of the deck was minimized in an effort to reduce the transverse cracking moment of the specimens so that yielding of the transverse reinforcement immediately at cracking could be avoided. A minimum deck thickness of 2 in. was used to provide concrete cover over the deck reinforcement placed to resist potential shrinkage cracking at the surface. The reinforcement to control shrinkage cracking consisted of a total of five No. 3 bars, oriented perpendicular to the precast joint, in the deck region of each of specimen, in addition to eight No. 3 bars oriented parallel to the precast joint. The deck reinforcement was placed a clear distance of 1/4 in. from the precast members in all specimens except for SSMBLG3-HighBars, where the deck reinforcement was placed a clear distance of 3/4 in., due to a localized increase in the section depth away from the joint, which is discussed in more detail in Section 6.1.3. A photograph illustrating the deck reinforcement is shown in Figure 6.1.2. Each subassemblage specimen was designed to include four sets of transverse hooked bars to ensure at least two sets of interior transverse hooked bars were present, while the cage reinforcement design and placement was varied among the specimens. The specific design parameters of each subassemblage specimen are described in the following sections. The design and section calculations for each of the subassemblage specimens are given in Appendix H, which utilize the measured material properties, as documented in Section 6.4. Two spatial relationships were used throughout the construction and testing of the subassemblage specimens. The North direction was defined to originate from the origin face and point in the direction of the precast joint. In addition, a three-dimensional grid, with the origin located at the bottom of the section, at the precast joint, on the origin face, was used. The x-direction corresponded to the north direction and pointed along the direction of the precast joint and was therefore always positive. The y- and z- directions followed the right hand rule, thus y pointed in the west direction, with zero at the precast joint, and z pointed upwards, with zero at the bottom of the section.

164 Table 6.1.1: Subassemblage specimen design details Specimen Identification Width Depth Transverse Bars (Load Trans.) Cage (Crack Control) Max Spacing2 R/F Ratio3 [in] [in] Size Spacing Depth1 Presence Size Spacing ρcr SSMBLG1- Control1 62.75 14 #4 18 in. OC 4 1/2 in. Cage #3 18 in. OC 9 0.0031 SSMBLG2- NoCage 67.25 14 #4 18 in. OC 4 1/2 in. No Cage 0 0 18 0.0025 SSMBLG3- HighBars 62.75 14 #4 18 in. OC 7 in. Cage #3 18 in. OC 9 0.0031 SSMBLG4- Deep 62.75 18 #4 18 in. OC 4 1/2 in. Cage #3 18 in. OC 9 0.0022 SSMBLG5- No.6Bars 62.75 14 #6 18 in. OC 4 1/2 in. Cage #3 18 in. OC 9 0.0061 SSMBLG6- Frosch 64 14 #4 18 in. OC 4 1/2 in. Cage #3 4.5 in. OC 4.5 0.0052 SSMBLG7- Control2 62.75 14 #4 18 in. OC 4 1/2 in. Cage #3 18 in. OC 9 0.0031 1The depth of the transverse reinforcement was taken from the bottom of the precast section to the center of the reinforcement 2The maximum spacing was the maximum nominal distance between reinforcement traversing the longitudinal joint, regardless of type (i.e., transverse hooked bars or cage) 3The reinforcement ratio shown is that corresponding to crack control, see above and Section 5.1

165 Figure 6.1.1: Elevation and plan views of subassemblage specimen. The x-axis was aligned along the North direction and corresponded with the longitudinal joint. Positive x pointed North, positive y pointed West, positive z, was vertically upward

166 Figure 6.1.2: Photograph of deck reinforcement utilized for the subassemblage specimens 6.1.1. Subassemblage 1 – Control Specimen 1 The first subassemblage was deemed the control specimen because it had detailing similar to that of the transverse hooked reinforcement in the Concept 2 laboratory bridge specimen. The control specimen had a reinforcement ratio for transverse load transfer of 0.0010, while the cage and additional transverse hooked bar provided a reinforcement ratio of 0.0031 for crack control. This crack control reinforcement ratio was considered a practical lower bound value for the system when constructed with a cage, as constructability issues may arise if reinforcement smaller than No. 4 transverse hooked bars and No. 3 cage hoops are selected. The maximum spacing between transverse reinforcement was reduced to 9 in. by offsetting the cage reinforcement from the transverse hooked bars. The reinforcement layout for SSMBLG1-Control1 is shown in Figure 6.1.3. The cage reinforcement is shown in green. The portion of the transverse hooked bars protruding into the CIP concrete is shown in blue. For simplicity, the part of the hooked bars embedded in the precast elements is not shown. Reinforcement perpendicular to precast joint Reinforcement parallel to precast joint

167 Figure 6.1.3: Layout for SSMBLG1-Control1, SSMBLG3-HighBars, SSMBLG4-Deep, SSMBLG5-No.6Bars, and SSMBLG7-Control2. Transverse hooked bars are shown in blue; cage reinforcement is shown in green 6.1.2. Subassemblage 2 – No Cage Reinforcement The reinforcement cage was intended to provide additional reinforcement for reflective crack control above the longitudinal joint. In order to be able to evaluate the effectiveness of the presence of the cage with the test series, one of the specimens, SSMBLG2-NoCage, was designed without cage reinforcement. The overall width of the specimen was increased to 67.25 in. to allow for the end transverse hooked bar to be half of the maximum spacing (9 in.) away from the end face. The reinforcement ratio for transverse load transfer was equal to 0.0010, while the reinforcement ratio for crack control was equal to 0.0025, which corresponded to the lower bound reinforcement ratio for crack control investigated during the study. The absence of the reinforcement cage increased the maximum spacing to 18 in. The specimen layout is shown in Figure 6.1.4.

168 Figure 6.1.4: Layout for SSMBLG2-NoCage. Transverse hooked bars are shown in blue 6.1.3. Subassemblage 3 – Increased Distance Between Transverse Hooks and Precast Flange The third subassemblage specimen was designed with the centroid of the transverse hooked bars located higher in the section than for the other specimens. In SSMBLG3-HighBars, the reinforcement was located 7 in. from the bottom of the section, which placed the hooked bars 55 percent higher than those in the other specimens. The proximity of the transverse hooked bars to the precast flange was expected to alter the performance of the system in two primary ways. The transverse moment of inertia of the cracked section decreased as the distance between the compression fiber and the reinforcement was decreased. The reduced moment of inertia of SSMBLG3-HighBars was expected to increase the curvature, and subsequently the stress demands on the transverse reinforcement of the section relative to the control specimen. Secondly, the lower in the section that the transverse reinforcement intercepts the reflective crack, the more effective that reinforcement was expected to be in controlling the crack. As a consequence, it was expected that SSMBLG3-HighBars would be less effective in controlling the crack than the control specimen. The specimen layout for SSMBLG3-HighBars was identical to that shown in Figure 6.1.3 for the control specimen. The only difference was the vertical position of the transverse hooked bars and cage. Because SSMBLG3-HighBars was constructed and tested first, and the capability of the 2 in. deep deck to develop adequate bond for the shrinkage reinforcement used in the deck was unknown, this specimen was constructed with an increased section depth away from the precast joint. The depth of the section near the joint was maintained at 14 in., and increased to 15-½ in. starting at 15 in. from the joint in both directions.

169 An elevation view of SSMBLG3-HighBars with the increased depth of the deck is shown in Figure 6.1.5. The reinforcement ratios for crack control and load transfer were equal to 0.0031 and 0.0010, respectively. Figure 6.1.5: Elevation view of SSMBLG3-HighBars and increased deck depth to provide additional cover for the shrinkage reinforcement in the deck As previously noted, SSMBLG3-HighBars was tested first, and delamination of the precast-CIP interface was observed early in the test, before cracking was observed near the precast joint. Though subsequent use of a clamping assembly was anticipated to better emulate the expected restraint of a real structure, the presence of delamination at those locations greatly influenced the behavior near the precast joint, and subsequently, data collected during testing on this section is not considered to be reliable, and therefore is not included in the analysis of the data from the remaining specimens. Because little information was expected to be learned from the specimen after delamination of the joint, SSMBLG3-HighBars was utilized as a general test specimen to investigation the behavior of the overall subassemblage concept in the load frame. Therefore, the specimen was loaded in a wide range of sequences and load levels, so much so that the specimen failed by fracturing of the transverse hooks near the CIP - precast web interface at an applied load of approximately 111 percent of the predicted cracking load, or 32 kips. The specimen after failure is shown in Figure 6.1.6.

170 Figure 6.1.6: Failure of SSMBLG3-HighBars due to fracture of the transverse hooked reinforcement near the CIP - precast web interface 6.1.4. Subassemblage 4 – Increased Depth of Precast Section As the span length of the PCSSS increases, the depth of the precast section also needs to increase in order to meet stress requirements at transfer and service. The 12 in. deep precast section was expected to perform well for moderate spans. The maximum practical span of the system was estimated to be approximately 65 ft., requiring a precast section depth of approximately 22 in. SSMBLG4-Deep was designed with an identical reinforcement plan as that of the control specimen, such that the effects of the deeper section could be investigated with an unchanged reinforcement design, although the depth of the cage stirrups was increased to the same relative depth as in the 12 in. deep precast sections (i.e., the top of the cage stirrup was aligned with the top of the precast web). The load transfer and crack control reinforcement ratios of the section were reduced to 0.0007 and 0.0012, respectively, due to the increase in the area of concrete considered in the calculation. The specimen layout of SSMBLG4-Deep was identical to that shown in Figure 6.1.3 for the control specimen.

171 6.1.5. Subassemblage 5 – Increased Transverse Hook Size The size of the reinforcement used for transverse load transfer must be sufficient to provide adequate flexural capacity after cracking. Article 5.7.3.3.2 of the 2010 AASHTO LRFD Design Specification requires that the nominal flexural moment capacity of the section be at least 120 percent of the cracking moment. No. 6 transverse hooked bars represented the minimum bar size that could satisfy this requirement while maintaining a hook spacing of 18 in. The specimen layout of SSMBLG5-No.6Bars was also identical to that shown in Figure 6.1.3. The cage remained unchanged from the control specimen; the only difference was that No. 6 hooked bars were used rather than No. 4 hooked bars. The reinforcement ratio for crack control and load transfer was 0.0033 and 0.0061, respectively, both of which represented the upper bounds investigated during the study. 6.1.6. Subassemblage 6 – Frosch Design Recommendations Frosch et al. (2006) provided relevant design recommendations for crack control reinforcement in bridge decks which is summarized in Section 2.2. Frosch et al. provided guidelines for both the spacing and reinforcement ratio required for crack control via an experimental and numerical parametric study. The spacing limits were developed to provide sufficient reinforcement such that the crack widths would remain less than 0.021 in. The crack width selected represented a 1/3 increase in the maximum crack width of 0.016 in. suggested by ACI 224 (2001) for aesthetics. The authors stated that the increase in the selected maximum crack width was done due to the wide scatter generally observed in crack widths. The reinforcement spacing design proposed by Frosch et al. when grade 60 mild steel reinforcement was used is given in Eqn. (6.1.1). (6.1.1) where cc is the depth of concrete cover from the extreme tensile fiber of the concrete to the center of reinforcement in inches. The depth of cover, cc , considered in Eqn. (6.1.1) was calculated to the bottom of the CIP concrete above the flanges, thus disregarding the depth of the precast flange. This was deemed a reasonable assumption because of the discontinuity between the precast flanges along the longitudinal joint where the flanges in the adjacent precast panels abutted. The precast flanges themselves were reinforced as part of the design of the precast inverted-T sections. Frosch provided a recommended reinforcement ratio for crack control reinforcement, as given in Eqn. (6.1.2). The reinforcement ratio was developed to ensure that sufficient reinforcement is present upon the introduction of cracking such that all tensile loads can be transferred through the reinforcing bars. (6.1.2) where fc’ is the specified concrete strength and fy is the specified reinforcement yield strength. With the depth of cover assumed to be 1.5 in., the maximum spacing was determined to be the 9 in. bound of Eqn. (6.1.1), and with grade 60 reinforcement and a 28-day concrete compressive strength of 4000 psi, the reinforcement ratio for crack control reinforcement was determined to be 0.0063 from Eqn. (6.1.2).

172 The reinforcement design for SSMBLG6-Frosch was selected such that the transverse reinforcement for load transfer (i.e., hooked bars protruding from the precast webs) was identical to that of the control specimen. Therefore, the cage reinforcement was modified to meet the above limits; however the cage bar size was maintained at No. 3 hoops to provide consistency among the specimens. The hoop spacing for the SSMBLG6-Frosch specimen was provided at 4.5 in. because an even multiple of 18 in. (i.e., the hook spacing) was desired, primarily to avoid interference between the cage and transverse hooked reinforcement, and a cage design requiring closer spacing, or spacing that caused intermittent interference between the cage and hooked bars, would be expensive and difficult to implement in the field. The 4.5 in. spacing was well within the maximum 9 in. spacing recommended by Frosch et al. (2008), but because of the design constraints associated with spacing the cage reinforcement to facilitate constructability, the reinforcement ratio for crack control provided in the SSMBLG6-Frosch specimen (i.e., 0.0052) was smaller than that strictly required by Frosch et al. (2008) (i.e., 0.0063 for Grade 60 steel and a concrete compressive strength of 4,000 psi). The specimen reinforcement ratio was considered sufficient to explore the increased benefit of the design recommendations provided by Frosch et al. (2008). The reinforcement ratio for load transfer was 0.0010. The specimen layout of SSMBLG6-Frosch is shown in Figure 6.1.7. It should be noted that the specimen SSMBLG5-No.6Bars had a reinforcement ratio (i.e., 0.0061) that was closer to that of the Frosch requirements. Figure 6.1.7: Specimen layout for SSMBLG6-Frosch. Transverse hooked bars are shown in blue; cage reinforcement is shown in green

173 6.1.7. Subassemblage 7 – Control Specimen 2 The seventh subassemblage specimen was originally designed to provide means to investigate a debonded flange surface. Observations of the performance of SSMBLG3-HighBars and SSMBLG5-No.6Bars, which were constructed and tested prior to the construction of the remaining five specimens, suggested that inadequate bond of the precast flange and CIP concrete could hinder the performance of the subassemblage specimens and limit their value by promoting the formation of a crack at the CIP-precast web interface. The intent of the subassemblage specimens was to investigate the crack control characteristics of the transverse reinforcement and cage above the longitudinal joint above the interface of the adjacent precast flanges. No instrumentation was located across the CIP-precast web interface, and only the transverse hooked bars could provide crack control at that location rather than the combined effect with the cage. For these reasons, the seventh subassemblage specimen was developed as a redundant control specimen. The reinforcement ratios for crack control and load transfer were identical to SSMBLG1-Control1, with values of 0.0031 and 0.0010, respectively. The specimen geometry of SSMBLG7-Control 2 was identical to SSMBLG1-Control1, however the flange surfaces of this specimen had been patched to provide a smooth surface in an effort to minimize the bond at the interface before the decision to abandon the debonded specimen was made. This represented the only difference in the design and construction between SSMBLG7-Control2 and SSMBLG1-Control1. The specimen layout for SSMBLG7-Control2 is shown in Figure 6.1.3. The flange surface before and after the surface irregularities were patched is shown in Figure 6.1.8, which illustrates the representative level of smoothness present in this specimen. (a) Prior to patching (b) After patching Figure 6.1.8: Precast flange surface condition in SSMBLG7-Control2 before and after patching of the flange to provide a smooth surface condition A manufacturing error was observed after the fabrication of the precast elements for SSMBLG7-Control2 was completed. The second set of transverse hooked bars, measured from the origin face, was constructed with the bars protruding from the adjacent precast members in reverse order, as shown in Figure 6.1.9. Because of this, two adjacent bars protruding from one precast member were spaced slightly larger than 18

174 in., and the two respective bars from the opposite member were spaced at slightly less than 18 in., however the center to center distance between adjacent sets of transverse hooked bars remained constant, at 18 in. Figure 6.1.9: Photograph of SSMBLG7-Control2 to illustrate manufacturing error in placement of transverse hooked bars 6.2. Instrumentation of Subassemblage Specimens Each of the seven subassemblage specimens were instrumented similarly, with the only exception being the presence of additional instrumentation in SSMBLG3-HighBars to provide adequate comparison among specimens. Each specimen had a concrete embedment VW strain gage oriented perpendicular to the joint near the middle of the specimen directly over the precast joint at the same vertical depth as the transverse hooked bars (i.e., a nominal distance of 4.5 in. from the bottom of the precast section). Because vibrating wire gages do not drift over time, these were selected to monitor the effects of shrinkage as well as provide an absolute value of strain at the joint to quantify any potential shrinkage or cracking during handling of the specimens in the laboratory (see Section 6.6). An analysis of the shrinkage strains measured via the concrete embedment VW strain gages is included in Section 6.6.1. Each specimen contained a total of 17 concrete embedment resistive strain gages oriented perpendicular to the joint, near the joint. The 17 gages were split between two cross sections. Nine of the gages were located near the middle of the specimen, with the remaining eight gages located near the origin face of the Hook from right PC panel is north of hook from left panel Hook from right PC panel is south of hook from left panel

175 specimen. The instrumentation layout for the subassemblage specimens is shown in Figure 6.2.1. Each vertical layer of gages consisted of three concrete embedment resistive gages, with one centered over the precast joint and the remaining two centered 4.5 in. to either side of the precast joint. Figure 6.2.1: Instrumentation layout for subassemblage specimens. Overlap of gages not shown for clarity The nine and six gage layouts were installed to track the height of the crack from the precast – CIP concrete interface above the flange. All instruments in the nine and six gage layouts were 120 mm concrete embedment resistive strain gages. The bottom three strain gages were located at the same depth as the transverse hooked bars. The second layer was located midway through the depth of the reinforcement cage, while the third layer, where present, was located at the top of the reinforcement cage. The additional instrumentation near the origin face included two 60 mm concrete embedment resistive strain gages oriented perpendicular to the joint, one to either side of the six gage layout; 60 mm gages were chosen for this location because they were available from a past project. A 1 mm strain gage was attached to the transverse hooked bar nearest the origin face with a second gage attached to the hooked bar located 18 in. away; both were located directly over the precast joint. The two resistive steel gages and three concrete embedment resistive gages across the joint allowed for the transverse strain to be measured at five Origin Face

176 locations between adjacent transverse hooked bars. The instrumentation at the origin face of SSMBLG7- Control2, which was representative of all specimens, is shown in Figure 6.2.2. Figure 6.2.2: Plan view of instrumentation near origin face of subassemblage specimens Two additional strain gages were added to SSMBLG3-HighBars to facilitate comparison among the specimens. The two additional strain gages were located 1.5 in. up from the precast flange centered over the joint, one at each instrumented cross section, to correspond with the lower level of gages in the other specimens. The instrumentation naming scheme and nominal and as-placed locations for the instrumentation in each specimen are tabulated in Appendix I. In addition to the embedded instrumentation, external LVDT’s were utilized to monitor the joint opening. Four LVDTs were placed on both the origin and end faces of the specimen, and were located such that the center of the gage length of each LVDT was vertically aligned with the precast joint. The vertical locations of the LVDT instrumentation was measured from the bottom of the 45 degree, 1 in. chamfer near the horizontal precast-CIP interface, as shown in Figure 6.2.3. Three LVDTs were utilized to measure displacement locally near the precast joint, each with a nominal range of ±0.050 in., except for SSMBLG5- 1 mm. steel gage 1 mm. steel gage 60 mm. concrete embedded gage 60 mm. concrete embedded gage 6 - 120 mm. concrete embedded gages (in 2 layers) 18 in.

177 No.6Bars, which was instrumented with LVDTs with a nominal range of ±0.1 in. (referred to as LVDT050 or LVDT100 hereafter) and a relatively constant gage length of 9.5 in, which was measured from the center of the anchor block that secured the LVDT to the center of the corresponding anchor block to which the core was attached, and varied by no more than 1/4 in. The three LVDT050s were located a distance of -0.75 in. (Low LVDT), 2 in. (Mid LVDT), and 5 in. (High LVDT) from the bottom of the 1 in. precast chamfer. A fourth LVDT spanned between the adjacent vertical precast webs, which was selected to provide a measurement of the total opening associated with the vertical web interfaces, as well as entire CIP region. The fourth LVDT had a nominal range of ±0.5 in. (referred to as LVDT500 hereafter), a gage length of 30 ±1/4 in., and was located a nominal distance of 3 in. from the bottom of the precast chamfer. Also included on both the origin and end faces was a vertical grid to assist with the documentation of cracking expected on those faces. The grid was also measured from the bottom of the precast chamfer, and consisted of solid horizontal lines in increments of 1 in., along with dots in increments of 1/2 in. The dots were located along a vertical projection of the precast joint. The LVDT layout and vertical grid on the origin face of SSMBLG6-Frosch, which was representative of both faces on all of the specimens, is shown in Figure 6.2.4. Figure 6.2.3: Location of LVDT instrumentation utilized for subassemblage tests. Vertical measurements for placement of instrumentation originated from bottom of 1 in. precast chamfer

178 Figure 6.2.4: LVDT layout on origin face of SSMBLG6-Frosch. Vertical line above precast joint is shown by series of dots; measurements were taken from bottom of the precast chamfer 6.3. Clamping System Each subassemblage specimen was considered to represent a portion of a PCSSS bridge. The specimens were essentially 5 ft. “long” bridge sections, two panels “wide”; however the panels were supported parallel to the longitudinal joint between the precast panels to facilitate investigation of the effectiveness of the transverse reinforcement in controlling flexural cracks that could be induced along the longitudinal joint with the test setup. As described in Section 6.1, it was found during the testing of the first specimen, that the stiff flanges of the precast section rotated and caused delamination between the precast flange and CIP concrete, resulting in propagation of a crack at the precast-CIP concrete interface. The test setup was subsequently modified by developing a system to clamp the precast flanges to the CIP concrete on either side of the longitudinal joint. Although the test setup induced compressive forces through the depth of the section at the faces, it was believed to better emulate the field conditions because in a bridge system, the pier supports would be normal to the longitudinal joint, preventing the relative rotation of the precast flanges with respect to the CIP in the trough above the precast flanges at the ends of the sections. The vertical rods that connected the top and bottom steel members used to clamp the section were located a clear distance of between 2 and 3 in. from the face of the specimen. Consequently, curvature was induced LVDT050s LVDT500 Bottom of 1 in. precast chamfer

179 in the longitudinal clamping members, which tended to concentrate the compressive force at the ends of the members. This served to better simulate the effects of restraint in the bridge system (i.e., clamping the subassemblage specimens near the ends simulated the effect of the bridge supports transverse to the longitudinal joint, and relieved the compressive stress across the subassemblage). An illustration of the clamping system, viewed in elevation perpendicular to the precast joint, is shown in Figure 6.3.1, while a cross section (i.e., section AA in Figure 6.3.1) parallel to the precast joint with the curvature of the clamping system shown exaggerated, is shown in Figure 6.3.2. Figure 6.3.1: Clamping system developed to simulate restraint near joint region on subassemblage specimens. Section AA is shown in Figure 6.3.2 A A

180 Figure 6.3.2: Section view of clamping assembly and subassemblage specimen, parallel to joint, illustrating exaggerated curvature of L-section (top) and wide flange section (bottom) due to eccentricity of tensioned threaded rods The clamping system was developed with structural steel members, as shown in Figure 6.3.1 placed as near the precast joint as possible, while providing sufficient clearance for the loading strip between the clamps. A total of four 1 in. diameter threaded rods connected the upper and lower steel members near each face. The rods were tightened using an 18 in. long standard spud wrench. The concrete surfaces at the top and bottom of the specimens were adequately smooth and level such that the longitudinal steel clamping members were placed directly on the specimens as constructed without the addition of a grout bed. As mentioned previously, the clamping assembly was not initially utilized during the testing of the first specimen, SSMBLG3-HighBars. The necessity for the clamping system was quickly realized, as the separation of the horizontal precast flange – CIP concrete interface and the vertical web interface was observed early in the test, shown in Figure 6.3.3. Cracking and/or separation at these locations greatly reduced the applicability and value of the tests because the crack was not able to develop above the longitudinal joint where the crack control reinforcement and instrumentation was located. In the field study of Center City PCSSS bridge (No. 13004) documented in Smith et al. (2008), instrumentation had indicated the initiation of reflective cracking above the longitudinal joint between the precast flanges. While instrumentation was not present in that bridge at the vertical CIP-precast web interface, it would be unlikely that a crack would be present both at the joint and web face, because the presence of one crack would tend to relieve tensile stresses at the other locations. For these reasons, it was necessary to provide the clamping system to ensure that the boundary conditions imposed on the subassemblages were adequately representative of the conditions observed in the implementation of the PCSSS in the field. The clamped assembly fully installed with the HSS sections and neoprene bearing pad in place is shown in Figure 6.3.4.

181 Figure 6.3.3: Separation of East precast section from CIP concrete during testing of SSMBLG3-HighBars before the implementation of the vertical clamping assembly Separation at the precast – CIP interface

182 Figure 6.3.4: Clamping system used to provide rotational restraint of the precast members from the CIP concrete during the subassemblage tests and loading apparatus consisting of 1 in. HSS and neoprene bearing pad 6.4. Construction of Subassemblage Specimens and Material Properties The seven subassemblage specimens were constructed in two groups. The CIP concrete for SSMBLG3- HighBars and SSMBLG5-No.6Bars was placed on July 27th, 2009. The remaining five specimens were cast on October 1st, 2009. The CIP concrete used in all specimens was Mn/DOT state mix 3Y33HE, which had a nominal 28-day strength of 4,000 psi and was provided by a single supplier. The two specimens constructed during the first pour were built with concrete from two separate ready-mix trucks to allow for sufficient time between pours to finish each specimen. The vertical precast webs near the joint were not prewetted prior to placement of the CIP concrete. The five specimens constructed during the second pour used concrete from a total of three ready-mix trucks. SSMBLG2-NoCage and SSMBLG6-Frosch used concrete from the same truck, while SSMBLG1- Control1 and SSMBLG7-Control2 used concrete from the second truck, and SSMBLG4-Deep used concrete Tensioned rods Loading beam 1” HSS tube x 2 Neoprene bearing pad Steel L-sections Wide flange sections LVDT instrumentation

183 from the third truck. All the specimens were prewetted with water and allowed to reach a surface dry condition prior to the placement of the CIP concrete. All seven specimens were moist cured for 8 days immediately after placement of the CIP concrete according to ACI 308.1-98 Standard Specification for Curing Concrete (ACI 308.1-98). The curing process consisted of the placement of prewetted burlap against the concrete surface, with a 4 mil plastic sheet above. The burlap was wetted in 12 hour increments to ensure the surface was continuously moist during the 8 day period. The material properties of each specimen were measured at two key times. The concrete compressive strength was measured via 4x8 in. concrete cylinders 28 days after the concrete was placed and on the day testing was initiated for each specimen. Furthermore, the tensile strength of the concrete was measured both via 6x6x24 in. beams and 6x12 in. concrete cylinders on the first day of testing for each specimen. In addition, the modulus of elasticity of the concrete was measured on the first day of testing using 4x8 in. concrete cylinders. The measured material properties for the specimens when the CIP concrete was at an age of 28 days are included in Table 6.4.1. Note that the order of the specimens listed in the table is the order in which the specimens were cast; consequently, they are not listed in numerical order. For the sets of specimens cast with the same batch of concrete (i.e., specimens 1 and 7, and specimens 2 and 6), a single value is listed in the table for the 28-day strength. The material properties measured on the first day of testing for each specimen are given in Table 6.4.2. The order in which the specimens were tested was: SSMBLG3-HighBars, SSMBLG5-No.6Bars, SSMBLG1-Control1, SSMBLG6-Frosch, SSMBLG7-Control2, SSMBLG2-NoCage, and lastly SSMBLG4-Deep. Table 6.4.1: Measured subassemblage CIP concrete material properties at an age of 28 days Specimen # Description fc’ [psi] 3 High Bars 4714 5 No. 6 Bars 5201 1 Control 1 5670 7 Control 2 2 No Cage 6238 6 Frosch 4 Deep Section 6326

184 Table 6.4.2: Measured subassemblage CIP concrete material properties on first day of specimen testing Specimen # Description CIP age on 1st day of testing fc [psi] ft -Beam [psi] ft – Split Cylinder [psi] Ec [ksi] 1 Control 1 95 days 6552 840 594 4740 2 No Cage 122 days 6577 746 521 5633 3 High Bars 37 days 4726 678 497 3933 4 Deep Section 127 days 7151 757 546 4703 5 No. 6 Bars 53 days 5204 609 501 4112 6 Frosch 103 days 6898 806 529 5270 7 Control 2 113 days 7005 732 548 4194 The transverse hooked reinforcement utilized in the subassemblage specimens was tested using the exterior transverse hooks in the Concept 2 laboratory bridge. The No. 4 reinforcing bars embedded in the precast panels for the Concept 2 bridge and subassemblage tests was procured by County Materials at the same time, was of the same bar size, and was epoxy coated; therefore it was expected that the embedded reinforcement in the Concept 2 bridge adequately represented the embedded #4 bars in the subassemblage specimens. Three tensile tests were completed to measure the yield strength of the reinforcement, with average yield strength of approximately 70 ksi. Note that the reinforcing bars were specified as standard Grade 60 steel. Tensile tests were not conducted on the No. 6 bars. 6.5. Laboratory Testing Program An extensive laboratory testing program was developed to investigate the performance of each subassemblage specimen. The primary concern regarding the design of the testing program was to create a testing plan that would not be heavily influenced by the age of the specimen or the differences in the material strengths and properties of each specimen at the time of testing. The primary objective of the testing plan was to provide a means to investigate the effectiveness of the transverse reinforcement above the longitudinal joint between the precast flanges in controlling the crack growth generated through both static and cyclic load tests. Therefore, the test plan was constructed to monitor crack growth, especially at loads above those which initiated flexural cracking near the joint. In an effort to minimize the variations of the concrete strengths and other material properties on the results of the tests, the modulus of rupture of the CIP concrete in each specimen, measured on the first day of testing, according to ASTM C78-09 Standard Test Method for Flexural Strength of Concrete (Using Simple Beam with Third-Point Loading) (ASTM C78-09), was used to determine the predicted transverse cracking moment of the subassemblage specimen. The modulus of rupture of each specimen is repeated in Table 6.5.1 along with the respective predicted cracking moment, Mcrack , and predicted cracking load, Pcrack , for each specimen.

185 Table 6.5.1: Subassemblage specimen measured modulus of rupture and predicted cracking moment and load Specimen # Description ft [psi] MCR-pred [in-kip] PCR-pred [kip] 1 Control 1 840 1117 33.0 2 No Cage 746 1050 30.8 3 High Bars 678 872 28.9 4 Deep Section 757 1855 56.9 5 No. 6 Bars 609 851 22.7 6 Frosch 806 1121 32.2 7 Control 2 732 972 28.2 The testing plan for the subassemblage specimens was based on the load at which cracking was observed to occur. Because effects such as restrained shrinkage in the test specimens may have reduced the cracking load from the predicted value, loading was conducted in small increments, followed by cyclic loading, in order to capture the initiation of visual cracking. After the specimen was taken to each increased load level and held at the peak, the specimen was examined at each face (i.e., origin and end face) for evidence of visual cracking. After cracking was observed, the specimens were subjected to cyclic loading at a lower load level to investigate the stability of the crack (i.e., potential durability of the test specimens) under repeated loads. The specimens were subsequently subjected to increased levels of static load to grow the crack, and the process of cycling the specimen at a lower load level continued. The specific testing plan, which was consistent for each specimen (with the exception of SSMBLG3- HighBars), consisted of an initial load to 15 percent of the predicted cracking load, PCR-pred, followed by an increase of 5 percent of the predicted cracking load during each subsequent load step. At each load step, the load was applied and removed at a quasi-static rate three times; followed by 1,000 cycles to the displacement that occurred concurrent with load applied at the current load step at a dynamic rate of 0.7 Hz via a displacement-controlled test. The displacement limits were selected such that the range of load that was applied at each load step was initially between 2 kips at the lower bound up to the load applied during the quasi-static loading at each step. A quasi-load controlled testing scheme was used in the initial cracking phases of the tests to better control the crack propagation. Upon observation of a visual crack at either the origin or end faces, the load was recorded as the observed cracking load, PCR-meas, after which the specimen was subjected to 5,000 cycles of displacement-controlled loading to the displacement that occurred concurrent to the load level previously reached. As previously mentioned, displacement limits were selected such that the applied load was cycled between 2 kips and the target load at that particular load step. The specimens were subsequently subjected to an additional 2,000 displacement-controlled load cycles with the displacement limits selected such that they initially occurred concurrent with the “base” level load. The displacement limits were periodically adjusted during the completion of cyclic loading at each load step such that the applied load range was kept relatively constant; thereby producing a load- controlled type testing environment with the benefit of a controlled lower bound displacement to prohibit sudden collapse of the specimens. The base load was taken as approximately 55 percent of PCR-meas (i.e., “BASE” ≈ 0.55*PCR-meas), rounded to the nearest 5 percent of PCR-pred, because the applied load values were

186 tabulated and plotted as percentages of the predicted cracking load. The use of 55 percent of the observed cracking load was selected to provide an adequately small base level (i.e., small enough such that initial cracking was reasonably expected not to have occurred prior to exceeding the base level load) which would allow for the investigation of degradation of the specimen due to cyclic loading at the base load while not inducing large levels of strain. The load was incremented above the observed cracking load, PCR-meas, in load step increments of 5 percent of PCR-pred consisting of both quasi-static and 1,000 cycles of quasi-load-controlled fatigue load at each load step, until the crack length observed on either the origin or end face reached a length of 8 in., measured from the origin of the vertical grid on each face (i.e., the bottom of the chamfer at the precast joint, see Figure 6.2.3 for additional details). A total of 5,000 quasi-load-controlled cycles were then completed at a displacement consistent with the load required to grow the crack to a length of 8 in., and the base-level load was repeated. Load was then incremented until the maximum load, Pmax, was achieved, which was selected as the lesser of: (1) a proportion of the predicted cracking load, taken as approximately 195 percent of PCR-meas (i.e., Pmax ≈ 1.95*PCR-meas), rounded to the nearest 5 percent of the predicted cracking load, or (2) when delamination was observed at the vertical precast – CIP web interface. Because the subassemblage tests were designed to induce a maximum moment at the longitudinal precast joint, data measured near the precast joint after the observation of cracking or delamination at the vertical precast-CIP web interface was considered to be irrelevant. Therefore, Pmax was selected based on when cracking was observed at the vertical precast-CIP web interface during testing on SSMBLG5-No.6Bars, which was at an applied load of approximately 195 percent of PCR-meas. The maximum applied load, Pmax, was then followed by 2,000 cycles of quasi-load-controlled loading at the base level. The loading procedure for an example specimen with a predicted cracking load of 40 kips is given in Table 6.5.2. The only specimen to deviate from this plan was the first one tested, SSMBLG3-HighBars, where the initial load step was 50 percent of the predicted cracking load, and the lack of the clamping assembly allowed separation at the precast flange-CIP concrete interface. During each load step, visual observations were recorded. The crack width and general observations were documented after the application of the third quasi-static load cycle with the load applied. In addition, after the completion of all cyclic loading at a given load step, the crack widths and other observations were again recorded with the magnitude of load equal to that which was applied during the cyclic tests at that load step.

187 Table 6.5.2: Subassemblage loading plan for example specimen with predicted cracking load of 40 kips Predicted Cracking Load PCR-pred = 40 k Applied load when cracking was observed PCR-meas= 24 k ( = 60% of PCR-pred) Base ≈ 0.55*PCR-meas = 0.55(24 k) = 13.2 k = 33% of PCR-pred rounded to 35% of PCR-pred = 14 k Reflective crack width reached length of 8 in. during load step to 105% of PCR-meas Maximum Load ≈ 1.95*PCR-meas = 47 k = 117% of PCR-pred rounded to 120% of PCR-pred = 48 k Load Step Applied Load [% of PCR-pred] Applied Load [kip] # Cycles Completed at applied load 1 15 6 1000 2 20 8 1000 3 25 10 1000 4 30 12 1000 5 35 14 1000 6 40 16 1000 7 45 18 1000 8 50 20 1000 9 55 22 1000 101 60 (100% of PCR-meas) 24 5000 11 “base” load 35 (≈55% of PCR-meas) 14 2000 12 65 26 1000 13 70 28 1000 14 75 30 1000 15 80 32 1000 16 85 34 1000 17 90 36 1000 18 95 38 1000 19 100 40 1000 20 105 42 5000 21 “base” load 35 (≈55% of PCR-meas) 14 2000 22 110 44 1000 23 115 46 1000 24 120 (≈195% of PCR-meas) 48 1000 25 “base” load 35 (≈55% of PCR-meas) 14 2000 1cracking visually observed at this load level

188 6.5.1. Data Acquisition The primary instrumentation utilized during the subassemblage tests were resistive-type concrete embedment strain gages and LVDTs, and also included a single transversely oriented concrete embedment VW gage. All data from the resistive and LVDT instrumentation were collected via a single National Instruments data acquisition (DAQ) system. The DAQ collected data at all times during the testing process at a rate of 15 Hz. The DAQ was equipped with a low pass filter which produced readings of satisfactory quality, with an approximate total gage noise of ±6 µε for the 120 mm concrete embedment resistive strain gages at a steady state condition. Data from the concrete embedment VW gages was collected via a CR10X data logger from Campbell Scientific at 2 minute intervals. The approximate noise of the concrete embedment VW gages instrumentation was approximately ±0.5 µε. 6.6. Results of Laboratory Testing A thorough analysis of the data collected both visually and via the DAQ was completed at the conclusion of the laboratory testing of each specimen. Seven metrics were developed in an effort to quantitatively and qualitatively characterize the performance of the subassemblage specimens relative to each other. Three of the seven metrics were defined based on visual observations and crack measurements during each test, which consisted of the documentation of general observations related to cracking observed during testing, and the measurement of the width and length of the crack observed on the origin and end faces of each specimen. Three metrics were determined through analysis of the data collected with the DAQ during the tests, and included an analysis of the strain due to shrinkage and placement of the subassemblage specimens in the load frame, width of opening of the joint measured via the LVDT instrumentation, and an analysis to investigate the vertical and horizontal generation and propagation of reflective cracking internally. The seventh metric consisted of the investigation of the predicted transverse reinforcement stress demands in the specimens. In all cases, the results were tabulated in terms of the predicted cracking load, PCR-pred. The section is organized as follows: • Investigation of transverse strain due to shrinkage and handling of the specimens (Section 6.6.1) • General observations regarding cracking observed during load tests (Section 6.6.2) • Analysis of visually observed crack widths on the origin and end faces of each specimen (Section 6.6.3) • Analysis of opening of joint region measured at the origin and end faces via LVDT instrumentation (Section 6.6.4) • Analysis of the rate of increase in the vertical length of cracking measured with a crack gage on the origin and end faces of each specimen (Section 6.6.5) • Investigation of the vertical and horizontal generation and propagation of reflective cracking near the longitudinal joint between the precast flanges measured via concrete embedment resistive strain gages (Section 6.6.6) • Predicted transverse reinforcement stress demands through range of loading (Section 6.6.7)

189 6.6.1. Transverse Strains Near Joint Region due to Shrinkage and Handling of Subassemblage Specimens The transverse strain near the joint region was monitored during the curing process, as well as during handling of the specimens, to provide an estimate of the state of strain near the joint at the initiation of load testing. A single transversely oriented concrete embedment VW strain gage was installed near the middle of each section (measured along the precast joint), as shown in Figure 6.2.1. The strain in each gage was recorded in two hour increments from 1 hour after placement of the CIP concrete until the conclusion of all load testing. The initial state of strain, defined to be the strain at the initiation of load testing, was of interest because the introduction of cracking due to applied load was a critical element of the testing plan, and therefore a record of existing cracking present prior to the start of testing was of interest. The transverse mechanical strains measured via the concrete embedment VW strain gages as a function of the number of days after placement of the CIP concrete, up until the first day of laboratory testing for each specimen, is shown in Figure 6.6.1. The horizontal axis represents the number of days between the placement of the CIP concrete and the first day of load testing for each specimen, and ranged from 37 days (SSMBLG3-HighBars) to 127 days (SSMBLG4-Deep). The transverse strains given in the figure originate at zero strain, and increase quickly over a short period of time, giving the illusion that the initial strain was nonzero. In addition, the DAQ was initiated at the same absolute time for all of the specimens, therefore there was a variation of several hours between when the CIP concrete was placed in the first and last specimen of the day, resulting in a small relative difference in the age of the specimens at what is referred to as zero time (when the DAQ was initiated). The vertical offsets among the readings may be attributed to the different stages of hydration occurring in the different specimens when the DAQ was initiated. Each specimen was transferred from a construction staging area (where the CIP concrete was placed) to a loading frame, which is shown in the figure by the positive (i.e., tensile) increase in strains near the rightmost point of most of the data series. The specimens were cured and stored in the University of Minnesota Structures Laboratory. The maximum temperature variation measured via the thermistors in the concrete embedment vibrating wire gages was +/- 7 degrees Fahrenheit. No sudden change in temperature was observed during the time between casting and testing of the specimens, which suggested that the mechanical strains reported in Figure 6.6.1 were primarily due to shrinkage.

190 Figure 6.6.1: Measured transverse mechanical strains in the subassemblage specimens based on the number of days after the placement of the CIP concrete The sign of the transverse strain data remained “compressive” indicating shortening or shrinkage throughout the curing and handling periods for each specimen. No evidence of cracking was visually observed on either the origin or end faces of the specimens after placement of each subassemblage into the load frame. 6.6.2. General Observations of Cracking Behavior during Load Testing Cracking was induced in the CIP region near the precast joint in each of the seven subassemblage specimens as a result of applied mechanical loading in accordance with the testing program. In addition, cracking was also observed at the vertical precast web-CIP interface, and occurred after the initiation of cracking near the joint in each of the specimens except SSMBLG3-HighBars, due to the absence of the clamping assembly on that specimen. Of the remaining six specimens, a single primary vertical crack was observed on both the origin and end faces throughout the range of applied loads on all but SSMBLG6- Frosch, where two distinct vertical cracks were observed on both faces. For this specimen, a second vertical crack was observed at an applied load of approximately 49.0 k (152 percent of PCR-pred) and 41.9 k (130 percent of PCR-pred) on the origin and end faces, respectively. A photograph of the dual vertical cracks observed on the origin face of SSMBLG6-Frosch, which is representative of both faces, is shown in Figure Positive (tensile) change in strain due to transfer to load frame

191 6.6.2. In the discussion regarding visually measured crack widths and lengths in the following sections, the crack width was taken to be equal to the largest observed crack on each face when two cracks were present (i.e., when the load was large enough such that a second crack had formed), and the recorded crack length was taken to be the longest crack among the two, though little variation was observed in either the crack widths or crack lengths after the second 5 percent increase in applied load above the load step in which the secondary cracks were initiated. Figure 6.6.2: Photograph of development of two primary vertical cracks near the precast flange on origin face of SSMBLG6-Frosch. Applied load was 49.0 k (152 percent of PCR-pred ) Cracking was also observed near the vertical precast web-CIP interface at large levels of applied load. Cracking at these locations generally initiated at the bottom of the precast section, near the embedded end of the precast flange, as illustrated in Figure 6.6.3. The observed cracking near the bottom of the precast flanges was likely primarily due to the added restraint provided by the clamping assembly, whereas if the clamps were not present delamination of the horizontal precast flange-CIP interface was expected, as observed during testing of SSMBLG3-HighBars before the installation of the clamping mechanism. The introduction of cracking was generally observed at the vertical web interfaces on both sides of the precast joint, and tended to initiate at roughly the same applied load. Precast joint 2nd crack to be developed 1st crack to be developed

192 In some cases, at loads near the maximum applied to each specimen, diagonal cracking was also observed in the CIP region, initiating at the outside edge of the bottom wide flange section of the clamping assembly, and extending to the vertical precast web-CIP interface, as noted in Figure 6.6.3. Figure 6.6.3: Photograph of cracking near the vertical precast web-CIP interface, including cracking through the precast flange and diagonal cracking due to the clamping assembly in SSMBLG6-Frosch 6.6.3. Width of Cracking Near Joint Region Measured with Crack Gage The crack widths observed on both the origin and end faces of each subassemblage were documented throughout the majority of the tests. The procedure for visually documenting the crack growth in the subassemblage specimens was finalized during testing of the second specimen, or SSMBLG5-No.6Bars. For this reason, crack widths obtained for the SSMBLG3-HighBars test and crack widths from the origin face of SSMBLG5-No.6Bars are not compared with those of the other specimens. For the remaining specimens, the crack width was measured at each 5 percent increment relative to the predicted cracking load on both faces. A standard crack gage with a range of 0.002 to 0.032 in., inclusive, was used to measure the crack width at the lowest possible location in the CIP concrete in each specimen. The crack width was measured directly at the interface between the silicone caulk and CIP concrete, or as close as possible, and was recorded in association with the loading that corresponded to the respective load step applied to the section. The width of observed cracking on the faces of the specimens generally tended to be reduced with the removal of the applied load. Diagonal cracking from edge of clamping assembly Vertical precast web-CIP interface Cracking through precast flange Prestressing strand

193 The detail near the joint and location of crack measurement is shown in Figure 6.6.4. A representative image of the crack width measurement is shown in Figure 6.6.5. Figure 6.6.4: Location of measurement of width and length of crack observed on origin and end faces of subassemblage specimens

194 Figure 6.6.5: Measurement of crack width during subassemblage testing Because the performance of the transverse hooked bars and cage reinforcement was of primary interest in the subassemblage specimens, and the presence of cracking near the vertical precast web face was undesirable and could have an influence on the crack width measurement above the longitudinal joint between the precast flanges, the observed crack widths were tabulated up to and including the load step in which the first crack or separation of the vertical precast web was observed. Figure 6.6.6 shows the width of the crack observed near the joint on the origin face before each set of cycles in the selected specimens. Each specimen is labeled in the legend using the specimen number, descriptive name, and transverse reinforcement spacing (considering both the transverse hooked bars and cage as transverse reinforcement). The horizontal axis illustrates the measured crack width in inches, while the percent of predicted cracking load is shown on the vertical axis. It is to be noted, that the percent of the predicted cracking load at which the crack width is indicated as “0 in.” corresponds to the visually observed cracking load. In other words, for 6-Frosch-4.5”, visual cracking was observed at 65% of the predicted cracking load. One of the reasons for the observed cracking load being less than the predicted cracking load was attributed to potential restrained shrinkage effects in the subassemblages. The crack was observed to originate at roughly the same percentage of predicted cracking load, with the exception of SSMBLG6-Frosch, which required an additional increase of approximately 15 percent of the predicted cracking load before cracking was first observed. Subsequent analysis of the internal strain data (see Section 6.6.6) suggested that small crack widths of SSMBLG6-Frosch hindered early visual detection of the crack, which caused the observed results to be inaccurately recorded at higher loads. Because of the larger perceived measured cracking load, Bottom of precast chamfer Precast joint Observed crack

195 SSMBLG6-Frosch was consistently loaded to higher levels of load in comparison with some of the other test specimens. This was also the case for the SSMBLG5-No.6Bars, for which results are not present in Figure 6.6.6 because crack width measurements were not recorded on the origin face of that specimen. Figure 6.6.6: Maximum1 crack widths measured on the origin face2 of selected specimens3 1Maximum crack width was measured at the horizontal precast-CIP flange interface 2Clear distance from origin face to first transverse bar was approximately 3.1 in. for SSMBLG5-No.6Bars and SSMBLG6-Frosch, and 3.25 in. for the other specimens 3SSMBLG5-No.6Bars is not shown because crack widths were not measured on the origin face of that specimen. SSMBLG3-HighBars is also not shown because the clamping apparatus was not installed prior to testing and therefore those results are not comparable to the remaining specimens before each set of cycles The performance of each specimen can be deduced from Figure 6.6.6 by selecting a value for the percent of the predicted cracking load and comparing the measured crack widths at that load. As expected, the observed crack widths tended to be bounded by the specimens with the closest (Frosch) and sparsest (No Cage) maximum reinforcement spacing. The variation in the observed crack widths between the two control specimens, SSMBLG1-Control1 and SSMBLG7-Control2 might be attributed to the smooth surface condition achieved in the SSMBLG7-Control2 specimen. The smooth flange surface was expected to better distribute the tensile stresses across the width of the CIP concrete in the trough between the precast webs (transverse to the longitudinal joint) reducing the effect of the discontinuity created by the interface between the adjacent precast flanges. A smooth or debonded surface however may not support an 0% 20% 40% 60% 80% 100% 120% 140% 0 0.005 0.01 0.015 0.02 Pe rc en t o f P re di ct ed C ra ck in g Lo ad Maximum Crack Width [in] 1-Control1-9" 2-NoCage-18" 4-Deep-9" 6-Frosch-4.5" 7-Control2-9"

196 improved overall PCSSS because if a reflective crack would initiate at the vertical precast web-CIP concrete interface, the cage reinforcement would not contribute to the crack control resistance. The only reinforcement crossing such a crack would be the transverse hooked bars protruding from the precast webs. Figure 6.6.6 clearly shows that SSMBLG6-Frosch outperformed the other specimens at high loads. The crack widths on the end face of the specimens were documented at increased load levels prior to the subsequent cycling, and the values are plotted in Figure 6.6.7 with respect to the percentage of the cracking load applied. The crack widths measured on the end face followed a similar trend to those documented on the origin face. The transverse reinforcement terminated a distance of half of the maximum spacing from the end face of each specimen, essentially providing an estimate of the crack width at the point farthest from adjacent transverse reinforcement in the specimen; whereas on the origin face, the clear distance to the transverse reinforcement was approximately 3.1 in. for SSMBLG5-No.6Bars and SSMBLG6-Frosch and 3.25 in. for the remaining specimens. As observed near the origin face, the Frosch and Control2 specimens performed well, however the inclusion of the results from SSMBLG5-No.6Bars also showed very good performance of this specimen. As stated earlier in this section, the relatively higher initial visual cracking loads recorded for SSMBLG5-No.6Bars and SSMBLG6-Frosch were found to be much larger than those identified with the embedded instrumentation which was attributed to the relatively smaller crack widths generated in these specimens (see Section 6.6.6). The SSMBLG5-No.6Bars and SSMBLG6-Frosch were subjected to relatively higher levels of subsequent loads than the other specimens because the applied load history was associated with the visual crack initiation.

197 Figure 6.6.7: Maximum1 crack widths measured on the end face2 1Maximum crack width was measured at the horizontal precast flange -CIP interface; and in the case of SSMBLG6-Frosch, where two vertical cracks were observed at high loads, the largest crack width was recorded 2Reinforcement spacing from end face of specimen was half of the maximum reinforcement spacing of selected specimens before each set of cycles According to Figure 6.6.7, SSMBLG5-No.6Bars, SSMBLG6-Frosch, and SSMBLG7-Control2 appeared to outperform the other specimens, especially at higher levels of load in the case of SSMBLG6-Frosch and SSMBLG7-Control2. Also, in the case of SSMBLG6-Frosch, recall two vertical cracks were observed on each face of the specimen, which was generally preferred because the development of many, smaller cracks is often favored to the development of few, larger cracks, especially when the ingress of water and corrosive materials are of concern. Also observed from the figure, SSMBLG2-NoCage showed large increases in the measured crack width with each 5 percent increase in the applied load, especially at higher levels of load, highlighting the faster rate of degradation of this specimen compared to the others that were well reinforced for crack control. Furthermore, it is evident that an increased area of reinforcement can also provide satisfactory crack control performance, as shown by SSMBLG5-No.6Bars. The section constructed with a deeper precast beam, SSMBLG4-Deep, was observed to perform relatively well, especially considering that the reinforcement ratio for crack control in this section was smaller than 0% 20% 40% 60% 80% 100% 120% 140% 160% 180% 0 0.005 0.01 0.015 0.02 Pe rc en t o f P re di ct ed C ra ck in g Lo ad Maximum Crack Width [in] 1-Control1-9" 2-NoCage-18" 4-Deep-9" 5-No.6Bars-9" 6-Frosch-4.5" 7-Control2-9"

198 the other specimens due to the increase in member depth while the section contained similar reinforcement details as were used in the control specimens. The overall performance of the six subassemblage specimens based on the crack widths measured using a crack gage was good. Note that at an applied load of 100 percent of PCR-pred, the worst performing specimen was SSMBLG2-NoCage, with a measured crack width of approximately 0.014 in. An estimate of acceptable crack widths was determined based on recommendations from ACI Committee 224, which recommended a maximum crack width of 0.007 in. for structures exposed to corrosive environments, and 0.016 in. as a maximum limit for aesthetics (ACI 224, 2001). Also recall that Frosch et al. selected a target maximum crack with of 0.021 in. during their study (Frosch et al., 2006). The crack widths were also measured after the completion of all cyclic loading at each load step. The potential durability of each specimen subjected to cyclic loading might be inferred by examining the change in measured crack width on the end face before and after the completion of cyclic loading at a given load step, as illustrated in Table 6.6.1. Negligible to small increases in measured crack widths after cyclic loading at each load step indicated good performance. Many specimens showed no increase or only small increases in crack width (i.e., 0.001 in.) due to the application of cyclic loading. The four largest increases in the measured crack widths are highlighted in yellow.

199 Table 6.6.1: Increase in the measured crack width on the end face as a result of cyclic loading at each load step 1No cracking was visually observed at associated load step 2Measurements were not available at given load step due to error in data collection 3Delamination was observed at vertical precast web – CIP interface The relative performance between each face of a given specimen was also of interest. Each face had reinforcement located a specified distance away. The reinforcement near the origin face was a clear distance of 3.1 in. away in SSMBLG5-No.6Bars and SSMBLG6-Frosch and 3.25 in. in the remaining specimens, while the center of the reinforcement or paired hooks near the end face was half of the maximum spacing away, ranging from ideally 2.25 to 9 in. Figure 6.6.8 shows the difference in the measured crack widths between the origin face and the end face; a positive value in Figure 6.6.8 indicates that the crack width measured on the origin face was larger than that measured on the end face. For all Increase in measured crack width (in.) Percent of P cr-pred 1-Control1 2-NoCage 4-Deep 5-No.6Bars 6-Frosch 7-Control2 45 NCO1 0 NCO 50 0 0 0 55 0 0.001 0.001 60 0.001 0.002 0.001 0 65 0.001 0.002 0 0 0 70 0 0 0.003 0 0 75 0.004 0.002 0.003 0 0.001 80 0.001 0 0.002 0.001 0 85 NA2 0 0 0.001 0.001 90 0 0.001 0 0 0 95 NA 0 0 0 0 100 0.002 0.001 0 0 0.002 105 0 0.001 0.001 110 0 0 0.002 115 0 0 120 0 0 125 0.001 130 0.003 135 0 140 0.001 145 0.001 150 0 155 0 160 0 Delam NCO Delam NCO Delam NCO Subassemblage Specimen Delam3 Delam

200 specimens shown other than SSMBLG6-Frosch, the proximity of the reinforcement to the origin face (3.1 in.) was smaller than the proximity of the reinforcement to the end face, and therefore the crack widths on the origin face would be expected to be smaller than those on the end face, resulting in negative values in Figure 6.6.8. On the contrary, SSMBLG6-Frosch data showed a negative trend, suggesting that the crack width at the end face, where the reinforcement was nearest the face of the specimen was larger than on the origin face; however, there was very little difference in the distance from the reinforcement to the origin and end faces in the SSMBLG6-Frosch specimen. Both control specimens showed a trend towards positive differences in the crack widths, again suggesting that the face with the reinforcement located closer exhibited larger crack widths. Furthermore, contrary to expected results, SSMBLG2-NoCage, which had the largest difference between the proximity of the reinforcement on each face at 5.75 in. (9 in. on end face – 3.25 in. on origin face), showed the least variation in the crack widths measured between the faces. Because the crack width could only be measured at the external faces, and only one measurement was taken on each face, potential variations in the measured crack widths with respect to the distance from the reinforcement were expected, which may correspond to some of the unexpected trends documented in Figure 6.6.8. Figure 6.6.81 1Measured crack widths were not available on the origin face of SSMBLG5-No.6Bars : Difference in crack width between the origin and end face (origin minus end) 0% 20% 40% 60% 80% 100% 120% 140% -0.008 -0.006 -0.004 -0.002 0 0.002 0.004 0.006 0.008 Pe rc en t o f P re di ci te d Cr ac ki ng L oa d Crack Width [in] 1-Control1-9" 2-NoCage-18" 4-Deep-9" 6-Frosch-4.5" 7-Control2-9"

201 6.6.4. Width of Cracking Near Joint Region Measured with LVDTs The origin and end face of each specimen was instrumented with LVDTs oriented across the joint to measure the total displacement, or opening, within their gage lengths. The placement and LVDT size used was included in Section 6.2. The Mid LVDT displacement measurements (from LVDT050, or LVDT100 for SSMBLG5-No.6Bars) located a distance of 2 in. from the bottom of the precast chamfer on the origin and end faces are shown in Figures 6.6.9 and 6.6.10, respectively. The figures illustrate the measured LVDT displacement as a function of the applied load; therefore improved performance can be identified by data closer to the upper left corner of the plot, and steeper slope. As before, the specimens exhibiting smaller crack widths were SSMBLG6- Frosch, SSMBLG7-Control2, and SSMBLG5-No.6Bars, especially at larger levels of load, which was especially evident by the steeper slopes of those three data series, compared to the slopes of the SSMBLG1-Control1, SSMBLG2-NoCage, and SSMBLG4-Deep data series. The displacements measured with the Mid LVDTs correlated relatively well with the visually recorded crack widths measured with a crack gage at the intersection of the chamfers at the joint, documented in Section 6.6.3. It was generally expected that the displacements measured with the LVDTs would be slightly larger than the crack widths measured with a crack gage because the LVDT measurements included the concrete strains across the gage length, as well as the widths of each of the vertical cracks, where multiple cracks were present (i.e., SSMBLG6-Frosch), though this was not generally observed. Note however that the Mid LVDTs were located a vertical distance of 2 in. above the horizontal precast – CIP interface (which is where the visual observations were recorded). Because the specimens were subjected to flexural stresses, it was expected that the strain would vary linearly with the depth, which suggested that the Mid LVDTs could measure a smaller crack width than what was visually observed, depending on the amount of straining and additional cracking occurring between the gage blocks of the LVDT instrumentation. The maximum crack width measured with a crack gage and maximum displacement measured with the LVDT instrumentation are given for both the origin and end faces in Table 6.6.2. The measured LVDT displacements correlated relatively well to the crack widths measured visually using a crack gage.

202 Figure 6.6.9: LVDT displacement measured via the Mid LVDT at the origin face 0% 20% 40% 60% 80% 100% 120% 140% 160% 180% 0 0.005 0.01 0.015 0.02 Pe rc en t o f C ra ck in g Lo ad [% ] Measured Displacement of Mid LVDTs 1-Control1-9" 2-NoCage-18" 4-Deep-9" 5-No.6Bars-9" 6-Frosch-4.5" 7-Control2-9"

203 Figure 6.6.10: LVDT displacement measured via the Mid LVDT at the end face 0% 20% 40% 60% 80% 100% 120% 140% 160% 180% 0 0.005 0.01 0.015 0.02 Pe rc en t o f C ra ck in g Lo ad [% ] Measured Displacement of Mid LVDTs 1-Control1-9" 2-NoCage-18" 4-Deep-9" 5-No.6Bars-9" 6-Frosch-4.5" 7-Control2-9"

204 Table 6.6.2: Maximum crack widths via crack gage (from Section 6.6.3) and LVDT displacements measured on the origin and end faces Specimen Description 1 Percent of predicted cracking load applied2 Origin Face End Face Measured with crack gage Measured with LVDTs Measured with crack gage Measured with LVDTs 1-Control1-9" 100% 0.016 in. 0.009 in. 0.011 in. 0.010 in. 2-NoCage-18" 100% 0.014 in. 0.011 in. 0.014 in. 0.015 in. 4-Deep-9" 80% 0.008 in. 0.011 in. 0.008 in. 0.012 in. 5-No.6Bars-9" 160% NA3 0.012 in. 0.014 in. 0.017 in. 6-Frosch-4.5" 110% 0.006 in. 0.005 in. 0.01 in. 0.007 in. 7-Control2-9" 120% 0.014 in. 0.007 in. 0.008 in. 0.007 in. 1Note that the locations of measurements were inconsistent between the crack gage and LVDT measurements (see Section 6.2) 2The tabulated applied percentage of the predicted cracking load was the largest applied load prior to cracking at the vertical precast web-CIP interface 3Crack widths measured with a crack gage were not available on the origin face of SSMBLG5-No.6Bars The difference in the measured LVDT displacements between the origin and end faces at each load step was investigated in a similar method to that used in Section 6.6.3. Recall that all reinforcement near the origin face was a clear distance of 3.1 in. away from the face in SSMBLG5-No.6Bars and SSMBLG6-Frosch and 3.25 in. from the face in the remaining specimens, while the center of the hooked pair or center of the cage hoop reinforcement was half of the maximum spacing away from the end face, nominally ranging from 2.25 to 9 in. Figure 6.6.11 shows the difference in the LVDT displacements between the origin face and the end face; a positive value in Figure 6.6.11 indicates that the LVDT displacement measured on the origin face was larger than that measured on the end face. For all specimens other than SSMBLG6-Frosch, the proximity of the reinforcement to the origin face was less than that to the end face, therefore, it was expected that the results for these specimens would have a negative sign because crack widths were expected to increase as the distance from the nearest reinforcement increased, which generally corresponded with the observed results. Also note that SSMBLG6-Frosch was expected to have the least variation in the measurements from opposite faces because the difference in the proximity to the reinforcement to the exterior faces was the smallest for that specimen. Furthermore, the observed relative LVDT displacements between the origin and end faces matched the expected results better than the same analyses of the crack widths measured with a crack gage, which highlighted the variability in those measurements.

205 Figure 6.6.11: Difference in LVDT displacements between the origin and end face (origin minus end) 6.6.5. Rate of Increase in the Length of Cracking Near Joint Region via Visual Observation The vertical length of the reflective crack near the joint region was also recorded during the subassemblage tests. The origin of the measured crack length was taken at the intersection of the chamfers at the joint in the precast flanges, as shown in Figure 6.6.4. As before, crack lengths were not documented for SSMBLG3- HighBars (which underwent delamination between the precast flange and the CIP concrete without the clamping system rather than developing a crack above the adjacent precast flange tips); crack lengths were also not documented on the origin face of SSMBLG5-No.6Bars using the same methodology as for the other specimens; therefore, those locations are not included in the specimen comparison. The length of the crack was measured using either the grid that was transferred onto each specimen, a tape measure, or both. Measurements were generally taken with a precision to within ¼ in. A representative image of the measurement of the length of the crack is shown in Figure 6.6.12. The measured crack length, normalized by the total section depth of each specimen (to include analysis of the deep section in the comparison), versus the applied load in terms of a percent of the predicted cracking load before each set of cycles measured on the origin and end face is shown in Figure 6.6.13 and Figure 6.6.14, respectively. Also included is the range of the predicted cracked neutral axis locations for the specimens shown in the figures, with the specimens representing the upper and lower bounds annotated on the plots. The predicted cracked neutral axis locations are also tabulated in Table 6.6.3. The location of the cracked neutral axis was calculated using measured concrete compressive strengths and the measured modulus of rupture for each specimen. 0% 20% 40% 60% 80% 100% 120% 140% 160% 180% -0.008 -0.006 -0.004 -0.002 -2E-17 0.002 0.004 0.006 0.008 Pe rc en t o f C ra ck in g Lo ad [% ] Measured Displacement of Mid LVDTs 1-Control1-9" 2-NoCage-18" 4-Deep-9" 5-No.6Bars-9" 6-Frosch-4.5" 7-Control2-9"

206 Figure 6.6.12: Measurement of the length of crack during subassemblage testing. Red dots illustrate path of crack Crack length is approx. 5.5 in.

207 Figure 6.6.13: Normalized crack length on the origin face of selected specimens before each set of cycles Figure 6.6.14: Normalized crack length on the end face of selected specimens before each set of cycles 0% 20% 40% 60% 80% 100% 120% 140% 160% 0 0.2 0.4 0.6 0.8 Pe rc en t o f P re di ct ed C ra ck in g Lo ad Normalized Crack Length 1-Control1-9" 2-NoCage-18" 4-Deep-9" 6-Frosch-4.5" 7-Control2-9" 0% 20% 40% 60% 80% 100% 120% 140% 160% 0 0.2 0.4 0.6 0.8 Pe rc en t o f P re di ct ed C ra ck in g Lo ad Normalized Crack Length [in] 1-Control1-9" 2-NoCage-18" 4-Deep-9" 5-No.6Bars-9" 6-Frosch-4.5" 7-Control2-9" Cracked NA location for SSMBLG6-Frosch Cracked NA location for SSMBLG4-Deep Cracked NA location for SSMBLG5-No.6Bars Cracked NA location for SSMBLG4-Deep

208 Table 6.6.3: Predicted locations of the subassemblage cracked section neutral axes Specimen Cracked Section Neutral Axis Depth measured from the compression fiber Normalized Depth of Cracked Section Neutral Axis (measured from bottom of section) 1-Control1 1.74 in. 0.71 2-NoCage 1.43 in. 0.73 4-Deep 2.09 in. 0.76 5-No.6Bars 2.48 in. 0.66 6-Frosch 2.01 in. 0.69 7-Control2 1.84 in. 0.71 For each of the specimens, normalized crack length increased to a maximum of between 50 and 65 percent of the total section depth (i.e., precast depth plus depth of the deck). The primary behavior of interest regarding the length of the crack was its rate of increase relative to increases in the applied load, or the slope of each data series, particularly over a select range of applied loads. The final, or maximum, crack length was of little significance because the crack length was expected to be driven up to the cracked neutral axis (NA) in each case, though the results for each specimen showed that the maximum visually observed length of the crack did not reach the location of the cracked neutral axis As illustrated in Figures 6.6.13 and 6.6.14, the well reinforced sections (i.e., SSMBLG5-No.6Bars and SSMBLG6-Frosch controlled the length and rate of increase in the length of the crack, especially at high loads. Note however, that the predicted neutral axis locations suggest that the maximum length of the crack in the well reinforced sections would be lower than in the remaining specimens. The variation in the crack length measured on the origin face and end face before each set of cycles provided further insight regarding the effect of reinforcement spacing. The difference in the normalized crack length on the origin and end faces (origin minus end face) is illustrated in Figure 6.6.15; a positive value implies that the crack length on the origin face was greater than that on the end face at a given load step. As before, all specimens other than SSMBLG6-Frosch had reinforcement located closer to the origin face (3.1 in. clear from face) than the end face (2.25 in. from center of reinforcement to end face), suggesting that the difference in the normalized crack length should be negative for these specimens. However, for the majority of the specimens, the difference in the crack length between the origin and end faces was almost always positive, suggesting that the proximity of transverse reinforcement to a face for the situations investigated did not directly affect the length of the crack. This further supports the notion that the crack length is better controlled by the amount of reinforcement for crack control as opposed to the spacing of the reinforcement for the spacings included in this study.

209 Figure 6.6.15: Difference in normalized crack length between the origin and end face (origin minus end) of selected specimens 6.6.6. Investigation of the Vertical and Horizontal Generation and Propagation of Reflective Cracking near the Precast Joint Measured via Concrete Embedment Resistive Strain Gages The concrete embedment resistive strain gages provided a quantitative means to determine the internal crack initiation and propagation within the subassemblage specimens. The presence of cracking at a particular gage was identified based on the slope of the load versus strain relationship for each gage, which is illustrated in Figures 6.6.16 through 6.6.20. Using this method, the location of the crack could be determined using an “on-off” type of metric. Furthermore, when the load was small (i.e., when the percent of cracking load was on the order of 40 or 45 percent) large increases in the measured strain were also considered to be representative of cracking. As an example, consider the slope of the load versus strain data for SSMBLG1-Control1, as shown in Figures 6.6.16 through 6.6.20. Each figure illustrates the slope of the load versus strain data for each of the three transversely oriented concrete embedment resistive gages across the width of the CIP region near the precast joint at each vertical level (i.e., 1.0, 1.5, and 2.0). The figures are also annotated with the load at which load cracking was assumed to have occurred, which included some level of subjectivity. 0% 20% 40% 60% 80% 100% 120% 140% 160% -0.1 -0.05 0 0.05 0.1 0.15 0.2 Pe rc en t o f P re di ct ed C ra ck in g Lo ad Change in Normalized Crack Length [Δ crack length / total section depth] 1-Control1-9" 2-NoCage-18" 4-Deep-9" 6-Frosch-4.5" 7-Control2-9"

210 Figure 6.6.16: Slope of linear fit line for load versus 1.0 level strain data at middle cross section in SSMBLG1- Control1 Cracking was detected in the middle gage (directly over the precast joint) at an applied load of 50 percent of PCR-pred

211 Figure 6.6.17: Slope of linear fit line for load versus 1.5 level strain data at middle cross section in SSMBLG1- Control1 Cracking was detected in the middle gage (directly over the precast joint) at an applied load of 50 percent of PCR-pred

212 Figure 6.6.18: Slope of linear fit line for load versus 2.0 level strain data at middle cross section in SSMBLG1- Control1 Cracking was detected in the middle gage (directly over the precast joint) at an applied load of 50 percent of PCR-pred

213 Figure 6.6.19: Slope of linear fit line for load versus 1.0 level strain data at origin cross section in SSMBLG1- Control1 Cracking was detected in the middle gage (directly over the precast joint) at an applied load of 45 percent of PCR-pred

214 Figure 6.6.20: Slope of linear fit line for load versus 1.5 level strain data at origin cross section in SSMBLG1- Control1 The load at which cracking was first observed was taken as the first data point where a clear difference in the slope of the data occurred. For example, in Figure 6.6.16, the crack was assumed to be present at the 1.0 location at the midspan cross section at a load of 50% of the cracking load, with cracking observed only in the gage centered over the joint. Furthermore, when cracking was observed in two adjacent gages at similar levels of applied load, it was more likely that a single crack was developed in the region of overlap between the gages as opposed to two independent cracks. Using this process, the load required to drive a crack to the various vertical levels of each specimen was determined. Recall (from Section 6.2) that the concrete embedment resistive gages were placed at three vertical locations designated 1.0, 1.5 and 2.0. The 1.0 level gages coincided with the depth of the transverse hooked bars which were nominally located 4-½ in. from the bottom of the specimen (for all but SSMBLG3- HighBars). The 1.5 level gages were located 8 in. from the bottom of the section (10 in. for SSMBLG4-Deep), and the 2.0 level gages coincided with the top of the cage (i.e., top of the precast web) at 12 in. from the bottom of the section (16 in. for SSMBLG4-Deep). The middle cross section contained the 1.0, 1.5, and 2.0 level gages, while the origin cross section contained only the 1.0 and 1.5 level gages. Cracking was detected in the middle gage (directly over the precast joint) and the gage centered 4.5 in. east of the joint at an applied load of 50 percent of PCR-pred

215 Figure 6.6.21 and Figure 6.6.22 illustrate the load at which cracking was observed for each specimen at the gages through the depth at the midspan and origin cross sections, respectively. A specimen exhibiting superior performance would have a steeper slope at each gage level, that is, larger levels of load would be required to drive a crack to a given depth. Better performance was observed in SSMBLG6-Frosch and SSMBLG5-No.6Bars, which corresponded with expected results, as the transverse reinforcement details of these two specimens included an increased area of reinforcement (achieved either through tighter spacing or larger bars, respectively). Also pertinent to these specimens was the fact that the cracked section neutral axis was slightly lower in the section than in the other members, as outlined in Table 6.6.3, though the variation in cracked neutral axis depth normalized to the total section depth was relatively small. An analysis of the remaining four specimens suggested that little to no increase in load was required to drive a crack from the lowest level of gage to the highest, indicating a less than ideal reinforcement design. As with the other analyses, SSMBLG3-HighBars were not included in this analysis because this specimen was initially tested with the absence of the clamping system. Figure 6.6.21: Load at which cracking was first observed in gages at middle cross section as determined from concrete embedment resistive strain gages 30% 40% 50% 60% 70% 80% 90% 1.0 1.5 2.0 Pe rc en t o f P re di ct ed C ra ck in g Lo ad Gage Depth 1-Control1-9" 2-NoCage-18" 4-Deep-9" 5-No.6Bars-9" 6-Frosch-4.5" 7-Control2-9"

216 Figure 6.6.22: Load at which cracking was first observed in gages at origin cross section as determined from strain gages The improved performance of SSMBLG5-No.6Bars and SSMBLG6-Frosch based on the analysis of the strain gage data supported the results from the previous sections. Furthermore, SSMBLG5-No.6Bars outperformed the Frosch specimen as larger levels of load were applied as the crack was driven up towards the top level of instrumentation. The specimen with larger reinforcing bars, which had a slightly larger overall reinforcement ratio for crack control, at 117 percent of the crack control reinforcement ratio of the Frosch specimen, further supported the notion that increased reinforcement area was superior to reduced reinforcement spacing when working to limit the depth of reflective cracking in the sections studied (although the reinforcement ratio provided in SSMBLG5-No.6Bars was closer to that specified by Frosch et al., 2006, because of the desire to maintain the same transverse bar sizes and hook spacing among the specimens). This may have resulted because the reinforcement spacings used in the study were all no larger than 9 in. with the exception of SSMBLG2-NoCage. With adequate reinforcement spacing, the dominant factor observed in controlling the cracking was the amount of reinforcement. In addition to the investigation of the vertical propagation of cracking, the horizontal propagation was also considered. A longitudinally oriented spreader beam (i.e., parallel to the precast joint) was utilized to apply load during the subassemblage tests to provide a relatively uniform transverse stress gradient (i.e., stresses perpendicular to the precast joint) along the length of the precast joint. The spreader loading was designed 30% 40% 50% 60% 70% 80% 90% 1.0 1.5 2.0 Pe rc en t o f P re di ct ed C ra ck in g Lo ad Gage Depth 1-Control1-9" 2-NoCage-18" 4-Deep-9" 5-No.6Bars-9" 6-Frosch-4.5" 7-Control2-9"

217 to provide a stress field that was constant along the length of the precast joint over which the reinforcement details of each particular specimen could be investigated. Furthermore, uniform loading applied along the structure above the longitudinal joint between the precast elements was necessary to provide an unbiased investigation of the crack mapping completed on the origin and end faces of each specimen, where the influence of the proximity of the embedded reinforcement on surface cracking was monitored. In addition, the investigation of horizontal crack propagation (both along the joint and transverse to the joint) provided a useful metric to identify details that promoted the development of more cracks, which were expected to be smaller in size. Several transversely oriented concrete embedment and steel resistive strain gages were installed in each subassemblage specimen at various locations along the precast joint, as outlined in Section 6.2. The largest number of gages in a single layer were vertically located at approximately 1.5 in. from the horizontal precast – CIP interface, which corresponded to the vertical location of the center of the transverse hooked bars (in all cases except SSMBLG3-HighBars). A total of ten transversely oriented gages were located at this depth, and were distributed between the middle of the section near the precast joint, and towards the origin face. The behavior of cracking along the length of the precast joint was investigated using these ten gages, which provided insight into the uniformity of loading as well as a means of comparing the results from the visual observations and DAQ analysis. A plan view of the ten strain gages discussed in this section is shown in Figure 6.6.23. The 120 mm concrete embedment resistive strain gages over the joint, and to the east or west of the joint are illustrated using a square, circle and x symbol, respectively. The 60 mm concrete embedment resistive strain gages over the joint are shown with a star. The steel resistive strain gages are illustrated using a diamond symbol, while cracks observed visually on the end and origin faces of each specimen are denoted by a triangle. The results of the analysis for each subassemblage specimen are presented in two complementary plots. In both plots, only gages in which cracking was detected are shown, and subsequently the number of data points in each plot varied. The first plot illustrates the load at which cracking was detected by the internal instrumentation or visually on each face versus the location along the precast joint where each of the ten strain gages of interest were located. The horizontal axis corresponds with the “x” axis dimension utilized for the duration of the subassemblage tests, and therefore a distance of zero corresponded to the origin face and the end face was located at either 67.25 in. (SSMBLG2-NoCage), 64 in. (SSMBLG6-Frosch), or 62.75 in. (remaining specimens). The first plot provides a means of identifying how uniform cracking was along the length of the precast joint in terms of the level of applied load required to induce cracking, and also provided insight into the uniformity of applied loading using a spreader beam. The second plot presented for each specimen provides a means of visualizing the geometric location of the crack in terms of approximate lateral distance from the longitudinal joint between the precast flanges. The horizontal axis matches the previously described plot, and corresponds with the length of the precast joint. The vertical axis represents the width of the precast trough region, which is 24 in. Both plots also indicate the transverse hook and cage reinforcement in blue and green vertical lines, respectively. In this case, the gages near the middle and origin face of the specimen provided insight into where the crack was detected internally. The reported location of the crack represents the center of the gage in which cracking was detected, and therefore the location of the crack was given as the nominal “y” dimension for the gages, which corresponded to +/- 4.5 in. and 0 in. for the three gages located at each of the two cross sections. Also because of the overlap between adjacent gages, cracking detected in two adjacent gages at similar levels of load suggested that the crack was located in the region of overlap.

218 The results of the above analysis for each of the subassemblage specimens are shown in Figures 6.6.24 – 6.6.29. Figure 6.6.23: Strain gage identification utilized for investigation of uniformity of cracking along the length of the precast joint. Pairs of hooks spaced at 18 in. and the cage reinforcement are not shown in the drawing for clarity

219 (a) Load at which cracking was detected at various locations along the length of the precast joint (b) Lateral location of instrumentation in which cracking was detected at various locations along the length of the precast joint Figure 6.6.24: Load and location at which cracking was detected for SSMBLG1-Control1 The results from SSMBLG1-Control1 showed relatively good uniformity in the location and load at which cracking was observed. The load at which cracking was detected near the origin cross section and the middle cross section varied by only one load step or 5 percent of the predicted cracking load. Cracking was detected internally near the origin cross section at 45 percent of the predicted cracking load, while not visually observed on the origin or end face until the following load step, at 50 percent of the predicted cracking load. Also, as shown in Figure 6.6.24(b), the instrumentation in which the crack was detected was that which was centered over the longitudinal joint between the precast flanges. 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 Pe rc en t o f C ra ck in g Lo ad [% ] Distance from Origin Face, Measured Along Precast Joint Hooks Cage -12 -9 -6 -3 0 3 6 9 12 0 10 20 30 40 50 60 70D is ta nc e fr om P re ca st Jo in t (in ) Distance from Origin Face, Measured Along Precast Joint Hooks Cage

220 (a) Load at which cracking was detected at various locations along the length of the precast joint (b) Lateral location of instrumentation in which cracking was detected at various locations along the length of the precast joint Figure 6.6.25: Load and location at which cracking was detected for SSMBLG2-NoCage Relatively good uniformity in the location and load at which cracking was detected was also observed for SSMBLG2-NoCage. As with the first specimen, cracking was observed via the embedment instrumentation one load step prior to the visually observed cracking load. Also, cracking was detected solely in the instrumentation which was centered directly over the longitudinal joint between the precast flanges. 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 80 Pe rc en t o f C ra ck in g Lo ad [% ] Distance from Origin Face, Measured Along Precast Joint Hooks -12 -9 -6 -3 0 3 6 9 12 0 10 20 30 40 50 60 70 80D is ta nc e fr om P re ca st Jo in t (in ) Distance from Origin Face, Measured Along Precast Joint Hooks

221 (a) Load at which cracking was detected at various locations along the length of the precast joint (b) Lateral location of instrumentation in which cracking was detected at various locations along the length of the precast joint Figure 6.6.26: Load and location at which cracking was detected for SSMBLG4-Deep As observed in the previous two specimens, the visually observed cracking load was larger than the load at which cracking was detected via the embedded instrumentation for SSMBLG4-Deep. Cracking was relatively uniform along the length of the longitudinal joint between the precast flanges, and varied by 10 percent of the predicted cracking load between the internal and visually observed results. Cracking at the set of three concrete embedment resistive strain gages near the origin face was detected in the gage centered 4.5 in. west of the precast joint, though the remaining gages near the origin face (which were all centered over the precast joint) also detected the crack. 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 Pe rc en t o f C ra ck in g Lo ad [% ] Distance from Origin Face, Measured Along Precast Joint Hooks cage -12 -9 -6 -3 0 3 6 9 12 0 10 20 30 40 50 60 70D Is ta nc e fr om P re ca st Jo in t (in .) Distance from Origin Face, Measured Along Precast Joint Hooks Cage

222 (a) Load at which cracking was detected at various locations along the length of the precast joint (b) Lateral location of instrumentation in which cracking was detected at various locations along the length of the precast joint Figure 6.6.27: Load and location at which cracking was detected for SSMBLG5-No.6Bars The results for SSMBLG5-No.6Bars are shown in Figure 6.6.27. For this specimen, visual observations were not recorded on the origin face of the specimen, and are therefore not present in the figure. Cracking was first detected in SSMBLG5-No.6Bars with the internal instrumentation at an applied load of approximately 45 percent of the predicted cracking load, which was observed at both the origin and middle cross sections; however the load at cracking detected by the internal gages did not correlate with the load at which cracking was visually observed on the end face, which was approximately 85 percent of the predicted cracking load. The discrepancy between the DAQ and visually observed data is likely due to an error in the load at which cracking was visually observed, possibly as a result of the crack not being immediately identified. This might be attributed to crack widths being smaller in this specimen than those of other 30 50 70 90 110 130 150 0 10 20 30 40 50 60 70 Pe rc en t o f C ra ck in g Lo ad [% ] Distance from Origin Face, Measured Along Precast Joint Hooks Cage -12 -9 -6 -3 0 3 6 9 12 0 10 20 30 40 50 60 70D is ta nc e fr om P re ca st Jo in t (in .) Distance from Origin Face, Measured Along Precast Joint Hooks Cage

223 specimens where the internal instrumentation and external visual measurements better correlated in terms of crack initiation. Also observed in SSMBLG5-No.6Bars was that the crack traversed two gages at each of the two multi- instrumented cross sections. Near the origin face, the crack traversed the gage centered over the joint first, and then at the next load step was detected in the gage 4.5 in. west of the joint. This may have indicated that the crack was slightly offset from the longitudinal joint between the precast flanges. At the middle cross section the crack was detected in the gage centered over the joint at roughly the same load as other locations in the section, and then cracking was detected in the gage 4.5 in. east of the joint at a significantly higher load (i.e., a approximately 130 percent of the predicted cracking load). The indication of cracking on the opposite side of the joint compared to the other gage that had previously indicated cracking, at a much later time in the test after the load had been increased significantly, suggests that multiple cracks developed in the section. The presence of multiple cracks was expected to be a beneficial characteristic of PCSSS bridges, because more cracks near the joint suggested that the crack widths may be smaller, and cracks located in the region of the cage reinforcement was preferred to cracking occurring near the vertical precast web, where the cage provided no benefit.

224 (a) Load at which cracking was detected at various locations along the length of the precast joint (b) Lateral location of instrumentation in which cracking was detected at various locations along the length of the precast joint Figure 6.6.28: Load and location at which cracking was detected for SSMBLG6-Frosch Cracking in SSMBLG6-Frosch was detected in the most number of gages, a total of eight, among the specimens. There was a relatively large variation in the load at which cracking was detected, though the grouping of data points near 40 and 65 percent of the predicted cracking load suggested that multiple cracks were present in the section. Also, as discussed in Section 6.6.2, two cracks were observed near the longitudinal joint between the precast flanges on both the origin and end faces of this specimen, with the secondary cracks observed visually at approximately 140 percent of the predicted cracking load. These 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70P er ce nt o f C ra ck in g Lo ad [% ] Distance from Origin Face, Measured Along Precast Joint Hooks Cage -12 -9 -6 -3 0 3 6 9 12 0 10 20 30 40 50 60 70D is tn ac e fr om P re ca st Jo in t (in .) Distance from Origin Face, Measured Along Precast Joint Hooks Cage These two gages likely indicate the development of multiple cracks in the section

225 observations suggest that the relatively large reinforcement ratio provided in SSMBLG6-Frosch (which was accomplished through a tight spacing of No. 3 cage stirrups) encouraged the development of more, smaller cracks, as for the case of SSMBLG5-No.6Bars. The presence of multiple cracks was further supported by the results shown in Figure 6.6.28, specifically at the middle cross section, where cracking was detected in the gages centered directly over the joint, as well as the gages centered 4.5 in. in either direction of the joint at different load levels. (a) Load at which cracking was detected at various locations along the length of the precast joint (b) Lateral location of instrumentation in which cracking was detected at various locations along the length of the precast joint Figure 6.6.29: Load and location at which cracking was detected for SSMBLG7-Control2 The second control specimen exhibited behavior similar to SSMBLG1-Control1, SSMBLG2-NoCage, and SSMBLG4-Deep, in that a single crack was detected along the length of the longitudinal joint between the 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 Pe rc en t o f C ra ck in g Lo ad [% ] Distance from Origin Face, Measured Along Precast Joint Hooks Cage -12 -9 -6 -3 0 3 6 9 12 0 10 20 30 40 50 60 70 D is ta nc e fr om P re ca st Jo in t (in .) Distance from Origin Face, Measured Along Precast Joint Hooks Cage

226 precast flanges, as shown in Figure 6.6.29(b), at roughly the same applied load despite the presence of the smooth surface condition. In general, the behavior of the specimens as categorized in this section can be divided into two categories, based on the relative area of reinforcement provided for crack control. SSMBLG1-Control1, SSMBLG2- NOCage, SSMBLG4-Deep, and SSMBLG7-Control2, which had similar amounts of reinforcement for crack control, had cracking that was observed at nearly the same applied load at the various locations along the longitudinal joint between the precast flanges. Furthermore, cracking was generally detected in the gages that were centered over the precast joint. Visual cracking loads observed for these specimens also correlated well with the internal strain measurements. This behavior provided confirmation that the loading setup shown in Figure 6.3.4 adequately induced a region of maximum moment uniformly along the precast joint. The two specimens with relatively larger reinforcement areas for crack control, SSMBLG5-No.6Bars and SSMBLG6-Frosch appeared to develop multiple cracks in the precast joint region based on the results from the embedded instrumentation. In the case of SSMBLG6-Frosch, this was confirmed with the visual observation of two unique cracks observed on both the origin and end faces of the specimen, as discussed in Section 6.6.2. The introduction of more cracks, each of which was expected to have smaller crack widths, was expected to provide an improved system because the smaller crack widths tend to be more resistant to the ingress of chlorides. The visual cracking loads recorded for these specimens were significantly larger than the cracking loads identified with the embedded instrumentation, which indicates that the crack widths generated in these more heavily reinforced specimens were much smaller than those generated in the more lightly reinforced specimens (SSMBLG1-Control1, SSMBLG2-NOCage, SSMBLG4-Deep, and SSMBLG7-Control2). 6.6.7. Calculation of Expected Tensile Reinforcement Stress in Subassemblage Specimens An analytical investigation of the stress demands on the tensile reinforcement in each subassemblage specimen was conducted to investigate the expected reinforcement stress ranges among the specimens during loading. The sectional analysis tool BIAX (Wallace, 1989) was utilized to construct moment-curvature diagrams for each specimen. Two models were constructed for each specimen, one with the concrete tensile strength included in the analysis – which was utilized to identify the cracking moment and stress in the reinforcement before cracking, while the second model neglected the effects of the concrete tensile strength and was utilized to model the behavior after cracking. The stress-strain model for the reinforcement included the effects of strain hardening, which was based on the equation presented by Saenz (1964). The ultimate strength of the reinforcement was assumed to be 100 ksi, with a fracture strength of 90 ksi. The strain at the onset of strain hardening was assumed to be 1 percent. The strain at the ultimate stress was assumed to be 8 percent, with an assumed fracture strain of 10 percent. The initial modulus of elasticity for the strain hardening region was assumed to be 1500 ksi. The measured 28-day concrete compressive strengths, concrete tensile strengths, and reinforcement yield strengths were utilized in the program inputs. The reinforcement yield strength was measured to be approximately 70 ksi for the No. 4 bars, and was assumed to be similar for the No. 6 bars. The reinforcement stress was calculated for increasing moment until a concrete compressive strain of 0.003 (inclusive) was achieved, which was assumed to represent the maximum available concrete compressive strain before crushing occurred.

227 The predicted tensile reinforcement stress in each specimen is shown in Figure 6.6.30. The vertical axis represents the applied loading, and is given as the ratio of the applied load to the predicted cracking load. Figure 6.6.30: Predicted tensile reinforcement stress demands as a function of applied loading in subassemblage specimens `1The black ‘x’ symbols indicate the maximum applied load for each specimen during testing. The ‘x’ symbol was not shown for each specimen for clarity; the maximum applied load for each specimen is given in Table 6.6.4. 1 The plot shown in Figure 6.6.30 represents the expected tensile reinforcement stress range in each subassemblage specimen for increasing load up to when concrete crushing in compression was expected according to the analysis. The plot is truncated at a ratio of applied load to predicted cracking load of 2, which cuts off the load at which crushing would have been observed for SSMBLG5-No.6Bars. All of the tests were terminated before concrete crushing was observed. Test terminations are denoted by the black “x” symbols for three of the specimens. For clarity, these symbols were left off the plot for the specimens that were expected to undergo concrete crushing near the cracking load. 0 0.5 1 1.5 2 0 20 40 60 80 100 A pp lie d Lo ad / P re di ct ed C ra ck in g Lo ad Reinforcement Stress [ksi] SSMBLG1-Control1 SSMBLG2-NoCage SSMBLG4-Deep SSMBLG5-No6Bars SSMBLG6-Frosch SSMBLG7-Control2 Maximum applied loading was 175% of Predicted cracking load Yield stress

228 Table 6.6.4 summarizes the maximum tensile reinforcement stress predicted for each of the specimens associated with the maximum loads applied during each of the tests. Table 6.6.4: Maximum applied loading and associated predicted tensile reinforcement stresses in subassemblage specimens Specimen Tensile Reinforcement Area, including cage [in2] Ratio of Maximum Applied Load/Predicted Cracking Load Predicted Tensile Reinforcement Stress [ksi] SSMBLG1-Control1 1.93 1.0 67.0 SSMBLG2-NoCage 1.60 1.0 75.7 SSMBLG4-Deep 1.93 1.05 76.0 SSMBLG5-No6Bars 3.85 1.75 39.3 SSMBLG6-Frosch 3.03 1.35 54.1 SSMBLG7-Control2 1.93 1.25 70.0 As illustrated in Table 6.6.4, the tensile reinforcement was expected to yield in subassemblage specimens 2 (NoCage), 4 (Deep), and 7 (Control2). As shown in Figure 6.6.30, the reinforcement stresses immediately after cracking in the two specimens with the significantly larger tensile reinforcement areas, SSMBLG5- No6Bars and SSMBLG6-Frosch, were expected to be considerably smaller than yield. This is consistent with the conclusions of the previous sections that the more heavily reinforced specimens exhibited smaller crack widths. 6.7. Destructive Testing of Subassemblage Specimens At the conclusion of the testing program, three or more cores were removed from each specimen to investigate the physical crack length. Three total cores were taken from each specimen, unless more were deemed necessary to locate a crack. The three cores were removed from the middle of the specimen on a line perpendicular to the joint, with one core centered over the joint, and the remaining two centered over each vertical precast web, as illustrated in Figure 6.7.1. The core locations were selected to facilitate measurement of the reflective crack at the joint and the depth of cracking or separation at the vertical precast web-CIP concrete interface.

229 Figure 6.7.1: Coring locations in subassemblage specimens Each of the core specimens was examined both with the naked eye and the aid of an Olympus SZX12 stereo-microscope to identify the extent of cracking on the core surface. The level of magnification used to examine the cores ranged between 2.1X to 27X, which was the full capacity of the microscope. Any and all cracking that was identified in each core was tabulated, regardless of the size or anticipated origin. The observed crack widths were documented in classification categories, defined in Table 6.7.1. The vertical depth of cracking identified in the cores was referenced from the line created by the horizontal precast flange-CIP concrete interface, as shown in Figure 6.7.2. The characteristics of each core specimen and the measured crack widths and locations for each subassemblage specimen are tabulated in detail in Appendix G. West precast web Precast joint East precast web

230 Table 6.7.1: Crack width classification categories for analysis of core specimens Crack Classification Crack Width (W) 0.002 W < 0.002 in. A 0.002 in. ≤ W < 0.008 in. B 0.008 in. ≤ W < 0.023 in. C 0.023 in. ≤ W < 0.200 in. D W ≥ 0.200 in. Figure 6.7.2: Location of reference line for measurement of vertical location of cracking in core specimens The maximum height of the crack, measured vertically upwards from the reference line, and the maximum width of the crack are summarized in Table 6.7.2. Also shown in the table are the maximum crack length and width that were measured on the origin and end faces of the specimens during the tests. The crack length and width correspond to the length and width of the cracks under the applied load in the load step immediately preceding the first observed cracks at the vertical web interface. The maximum values recorded here are the largest of the measured crack widths and lengths documented in Appendix G. Because the cores were removed after the completion of all loading, the observed cracked condition of the core specimens corresponded to the maximum level of loading applied to each subassemblage specimen. Because the cores were subsequently examined under no load, the potential cracks present in the cored samples may have been reduced in size in comparison with the maximum lengths and widths measured at the origin and end faces under load. This was corroborated by the fact that the observed crack widths on the faces of the specimens were reduced after the removal of load.

231 Table 6.7.2: Summary of maximum height and width of crack measured in core specimens Specimen Location of Core Maximum height of crack1 Maximum width of crack2 Crack length measured on face3 Crack width measured on face4 (class designation) Origin Face End Face Origin Face End Face 1-Control1- 9” Joint 7.5 in. C 8.3 in 7.8 in. 0.016 in. (B) 0.011 in. (B) East Web 7 in. B West Web NO5 NO 2-NoCage- 18” Joint 7.25 in. B 8.3 in. 8.1 in. 0.014 in. (B) 0.014 in. (B) East Web NO NO West Web NO NO 3-HighBars- 9” Joint 9.5 in. B Values not relevant because clamping assembly was not utilized during the test of this specimen East Web NO NO 4-Deep-9” Joint 11.5 in. B 10.3 in 11.3 in. 0.008 in. (B) 0.014in. (B) East Web 10.75 in. B West Web 10.75 in. A 5-No.6Bars- 9” Joint 7.5 in. B NA 8.3 in NA 0.014 in. (B) East Web 7 in. <0.002 West Web NO NO 6-Frosch- 4.5” Joint 8 in. B 7.3 in. 7.3 in 0.006 in. (A) 0.010 in. (B) East Web 2.25 in. A West Web 0.75 in. <0.002 7-Control2- 9” Joint NO NO 7.3 in. 7.3 in. 0.014 in. (B) 0.008 in. (B) Joint NO NO East Web NO NO West Web NO NO 1The height of crack was measured from the reference line, defined in Figure 6.7.2; only “upward” values are recorded here 2The width of crack was documented by crack classification, as defined in Table 6.7.1 3The crack length measured on the face of the specimens was taken as the maximum crack length observed prior to cracking at the vertical web interface, see Section 6.6.5. Values are adjusted to match reference line defined in Figure 6.7.2 4The crack width measured on the face of the specimen was taken as the maximum crack width observed prior to cracking at the vertical web interface, see Figures 6.6.6-6.6.7 5”NO” represents “No reflective cracks observed” Most of the specimens with a total section depth of 14 in. had a maximum observed crack height of approximately 7- ½ in., which corresponded to a crack length of 8.2 in. once the chamfer length was taken into account as discussed in Section 6.6.5, which correlated well with the visually observed data on the two faces of the specimens. Cracking was visually identified in all of the subassemblage specimens with the exception of SSMBLG7-Control2. The width of the crack near the joint region was generally a Class B crack, which has a relatively wide range of crack widths (i.e.: 0.008 in. ≤ W < 0.023 in.) and corresponded well to the crack widths observed on the origin and end faces of the subassemblages during testing. The only core

232 to develop a Class C crack, crack width larger than 0.023 in., was SSMBLG1-Control1, in which no specific perturbation in the history of the specimen was known to be the cause. Cracking or separation of the CIP concrete at the vertical precast web interface was observed on nearly every specimen. The presence of cracking at these locations was not known to exist in field applications of the PCSSS, and was attributed to the subassemblage test setup which induced flexural stresses in the joint region. The results discussed in Section 6.6 do not include measurements or recorded data after the presence of cracking at the vertical web interface was detected, therefore the introduction of cracking at the vertical web was not expected to influence the results in Section 6.6. An additional core specimen was removed near the joint of SSMBLG7-Control2 after no cracking was identified in any of the three original cores. Investigation of the fourth core did not reveal any cracking in the subassemblage specimen, however cracking was visible on the end faces of the specimen during laboratory testing.