Proposed AASHTO Seismic Specifications for ABC Column Connections (2020)

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

5 2.1 Introduction Accelerated bridge construction (ABC), an approach that utilizes new technologies, novel detailing, and enhanced planning, is expected to address key objectives of Federal Highway Administration (FHWA) initiatives such as Every Day Counts and Highways for Life. ABC has received substantial attention among bridge engineers and public officials for the following reasons: • Substantial reduction of onsite activities, • Reduction of the total project delivery time, • Utilization of higher-quality structural elements as compared with CIP elements, • Concurrent execution of different tasks such as site preparation while structural elements are being fabricated, • Reduction of traffic interruption and risk to travelers and highway construction workers, and • Potential reduction of the total project cost. Five components of ABC are recognized by FHWA (Culmo 2009, 2011): 1. Foundation and wall elements, 2. Rapid embankment construction, 3. Prefabricated bridge elements and systems, 4. Structural placement methods, and 5. Fast-tracked construction. Prefabricated bridge elements and their connections are particularly critical in moderate- and high-seismic zones because of their pivotal role in ensuring satisfactory seismic performance of bridges. The connection of prefabricated elements—especially bridge columns—to adjoining mem- bers is of primary concern under seismic loading because the connection should resist column plastic forces while maintaining structural integrity under inelastic deformations in large columns. The application of precast bridge columns in high-seismic zones has been limited, mainly because of the aforementioned concern as well as a lack of seismic performance test data pertaining to column connections. Nevertheless, recent studies have led to the devel- opment and proof testing of several alternative earthquake-resistant connections for precast bridge columns. In the present study, these connections are generally categorized as • Mechanical bar splice connections, • Grouted duct connections, • Pocket and socket connections, • Pin connections (e.g., pipe-pin and rebar hinges), • Rocking columns, and • Connections with advanced materials. C H A P T E R 2 State-of-the-Art Literature Review

6 Proposed AASHTO Seismic Specifications for ABC Column Connections A state-of-the-art literature review of the first three of these types of connections is presented in this chapter. Two of the excluded connections, connections with advanced materials and rocking connections, are emphasized in NCHRP Research Report 864 (Saiidi et al. 2017). The panel for NCHRP Project 12-105 did not consider the pin connection of primary interest for further investigation. However, this chapter includes a review of pipe-pin connections to demonstrate the contrast between moment-resisting and reduced-moment precast column connections. 2.2 Mechanical Bar Splices 2.2.1 Introduction Mechanical bar splices, which are commonly referred to as “bar couplers,” were originally developed to shorten the splice length of bars in reinforced concrete (RC) structures, to reduce congestion, and, potentially, to reduce cost. Within the ABC context, couplers are used to splice reinforcing bars from two separate structural components. One or both component(s) can be precast. Several types of mechanical bar splices are available. New coupler types are also emerging. Among various types, five that are commercially available and are capable of transferring both ultimate tensile and compressive forces were selected in this project because they would be the most potentially appropriate for ABC application (Figure 2-1): 1. Shear screw couplers: This type of mechanical bar splice consists of a coupling sleeve with lock-shear screws and shear rails. The shear screws are designed to shear off at a specified torque. Reinforcing bars are inserted into the sleeve, then the screws are tightened until the screw heads shear off. Tension is transferred between the bars through bearing of the bars against the shear rails inside the couplers and shear in the screws (Figure 2-1a). 2. Headed bar couplers: A head is formed on the anchoring end of each bar; then the threaded coupling pieces are torqued to complete the connection. A steel spacer may be placed between the heads to directly transfer the bar compressive forces (Figure 2-1b). 3. Grouted sleeve couplers: The bars are inserted into a steel sleeve, and then the sleeve is filled with a nonshrink, high-strength grout. Tension is transferred through bond between the bar, grout, and the sleeve (Figure 2-1c). A version of grouted couplers is threaded at one end to shorten the coupler length (threaded grouted sleeve couplers). 4. Threaded couplers: The bars are threaded at the ends and connected to a coupling sleeve with matching internal threads to complete the connection (Figure 2-1d). The threaded portion of the bar can be either straight (or parallel) or tapered, as shown in the figure. 5. Bar-grip (swaged) couplers: Equal lengths of reinforcing bars are inserted into a coupling sleeve. A hydraulic machine presses the sleeve to grip the bars. Tension is transferred between the bars through the swaged sleeve (Figure 2-1e). Due to the variation of the coupler types and their anchoring mechanisms, precast column connections using couplers can be detailed with several configurations. Figure 2-2 shows sample details for column–footing connections. Similar configurations can be used for column–cap beam connections. Couplers can also be used to connect column segments in segmental piers. In this section, the current U.S. code limitations on the application of bar couplers in struc- tural components are presented first. Subsequently, the findings of a comprehensive literature review on the performance of couplers as well as the performance of mechanically spliced columns are presented.

State-of-the-Art Literature Review 7 (a) Shear screw coupler (ancon.co.uk) (b) Headed bar coupler (hrc-usa.com) (c) Grouted sleeve coupler (splicesleeve.com) (d) Threaded coupler (erico.com) (e) Swaged coupler (barsplice.com) Figure 2-1. Mechanical reinforcing bar splices. (a) Couplers embedded in adjoining member (b) Couplers in plastic hinges (c) Shifted couplers (d) Couplers at two levels Precast Column Footing Coupler Precast Column Footing Coupler Precast Column Footing Coupler Pedestal Footing Precast Column Bar Couplers Closure Pour Figure 2-2. Practical mechanically spliced column connection detailing. 2.2.2 Couplers in U.S. Codes Characteristics and installation methods vary between coupler types. The mechanism of force transfer through the coupler also varies by coupler type. Even though bar couplers have been extensively used in RC construction, they are either prohibited or allowed with limitations in plastic hinges of ductile members of bridges and buildings. Table 2-1 presents restrictions in the current AASHTO, California Department of Transportation (Caltrans), and American Concrete Institute (ACI) codes. Caltrans provides a list of proprietary couplers for service and ultimate applications. Furthermore, ACI 439.37 (ACI Committee 439 2007) presents a list of

8 Proposed AASHTO Seismic Specifications for ABC Column Connections mechanical bar splices available in the U.S. market, identifies their limitations, and provides information regarding bar end preparation and coupler configurations, but does not provide specifications or acceptance criteria. 2.2.3 Shear Screw Couplers 2.2.3.1 Performance of Shear Screw Couplers Strength performance of shear screw couplers (SSCs) under monotonic and cyclic loading was investigated by Lloyd (2001). The couplers used in this study were Bar Lock L-Series. Two sizes of ASTM A615 Grade 60 bars—No. 6 (Ø19 mm) and No. 8 (Ø25 mm)—were tested. Monotonic and cyclic tensile tests were performed on 160 coupler specimens. The monotonic test results showed that SSCs can guarantee development of 90% of the ultimate tensile strength of reinforcing bars. In 24 of 80 connection tests, the bars pulled out. None of the bars or couplers failed in 80 cyclic tests in which 100 cycles of loading between 5% and 90% of the specified yield strength were completed. Some of these specimens were subsequently tested under an additional 100 cycles of loading. No signs of deterioration and no failure were observed. Furthermore, eight of the speci- mens that experienced 100 cycles of loading were monotonically pulled to failure. The strength of these specimens was nearly the same as that of the specimens tested under monotonic loading. Slip tests showed that there is no tendency for the rebar to move within the coupler prior to developing the full splice strength. It was concluded that Bar Lock L-Series couplers are accept- able alternative mechanical splices for nuclear safety–related applications. This specific coupler is categorized by Caltrans as a service splice, but there are other SSC couplers that Caltrans categorizes as ultimate splices. Four nickel–titanium shape memory alloy (SMA) steel bar connections were tested by Hillis and Saiidi (2009) under a slow strain rate with SSCs (Figure 2-3). This SSC type had three Code Splice Type Stress Limit Strain Limit Max Slip Location Restriction ACI 318 (2014) Type 1 1.25fy None None Shall not be used in the plastic hinge of ductile members of special moment frames neither in longitudinal nor in transverse bars (Article 18.2.7). Type 2 1.0fu None None Shall not be used within one-half of the beam depth in special moment frames but are allowed in any other members at any location (Articles 18.2.7 and 25.5.7). Caltrans SDC (2013) Service None >2% None Splicing is not allowed in the no-splice zone of ductile members, which is the plastic hinge region. Ultimate splices are permitted outside the no-splice zone in ductile members. Service splices are allowed only in capacity-protected members (Chapter 8). Ultimate None >9% for No. 10 (32 mm) and smallera >6% for No. 11 (36 mm) and largera None AASHTO (2013, 2014) Full mechanical connectionb 1.25fy None No. 3-14: 0.01 in. No. 18: 0.03 in. Shall not be used in plastic hinge of columns in SDC C and D (AASHTO 2014, Article 8.8.3). Note: fy = yield strength of the anchoring bar; fu = ultimate strength of the anchoring bar; SDC = seismic design criteria. aFor ASTM A706 reinforcing steel bars. There is also a maximum strain demand limit [e.g., 2% for ultimate splices and 0.2% (the bar yield strain) for service splices] (California Department of Transportation 2014). bAASHTO (2013), Article 5.11.5.2.2. Table 2-1. Current U.S. code restrictions on mechanical bar couplers.

State-of-the-Art Literature Review 9 screws to link the coupler to each bar. No. 4 (Ø13 mm) SMA bars were connected to the same size Grade 60 reinforcing steel bars. The SSC splice developed the full tensile capacity and ductility of the spliced SMA bars. No failure of the couplers and no slippage at the coupler region were observed. All the specimens failed due to rupture of the SMA bars at the thread fixed in the tensile test machine grip, away from the coupler region. Rowell et al. (2009) tested nine SSC specimens connecting No. 10 (Ø32 mm) ASTM A615 Grade 60 steel bars under three strain rates: slow (3,500 µe/s), intermediate (65,000 µe/s), and high (3,200,000 µe/s). Three specimens were tested under each strain rate. The couplers had seven screws to link each bar to the coupler. At the slow strain rate, bars fractured inside the coupler at the screw closest to the edge of the coupler in all three specimens, and the average ultimate load for the splice was 27% lower than that of the reference bars (Figure 2-4a). The ductility was also significantly lower than that of the control bars. At the intermediate strain rate, bars fractured inside the coupler in all three specimens, and the average ultimate load for the splice was 16% lower than that of the reference bars (Figure 2-4b). The average maximum (a) Specimens (b) Test setup (c) Failure of shape memory alloy at threads Source: Hillis and Saiidi (2009). Figure 2-3. Specimen tests for shape memory alloy steel bar shear screw couplers.

10 Proposed AASHTO Seismic Specifications for ABC Column Connections elongation was reduced to 3% in comparison to 10% for the control bars. At high strain rate, bars either ruptured inside the coupler or pulled out from the coupler (Figure 2-4c). The average ultimate load for the SSC splice was 24% lower than that of the reference bars, and the average maximum strain was significantly lower (3%) than that of the control bars (14%). In summary, the dominant mode of failure was the bar rupture under the first or the second screw for these shear seven-screw couplers. This premature failure was due to stress concentra- tion under the shear screws resulting in lower ultimate stresses and significantly lower ductility as compared with the control bars. Huaco and Jirsa (2012) (and later Huaco 2013) tested two types of SSCs—three-screw (ERICO S-series) and four-screw (ERICO B-series)—for each bar under half- and full-cycle loadings. Ten specimens in which No. 8 (Ø25 mm) ASTM A706 Grade 60 steel bars were con- nected to the couplers were tested, four under half-cycle loading and six under full-cycle loading. The bars fractured on the edge of the SSC with three screws (Figure 2-5a), while with four-screw couplers bars fractured outside of the couplers (Figure 2-5b). It was observed in the full cyclic tests that the ultimate strain in the longer SSC can be three times that of the shorter SSC. Alam et al. (2010) tested nine specimens in which No. 4 to No. 6 (Ø13 mm to 19 mm) deformed and plain steel bars were connected with three-screw Dayton SSCs. The test results (a) Low strain rate tests (b) Intermediate strain rate tests (c) High strain rate tests Source: Rowell et al. (2009). Figure 2-4. Shear screw coupler specimen tests. (a) Three-screw coupler (b) Four-screw coupler Source: Huaco (2013). Figure 2-5. Shear screw coupler specimen tests.

State-of-the-Art Literature Review 11 showed that the full ultimate strength of both deformed and smooth bars can be achieved. The bars fractured outside the splice in all specimens. Slippage of bars inside the couplers was mini- mal before yielding, but significant slippage was reported after yielding. In an attempt to utilize this type of coupler for SMA bars, a modified version of the coupler was developed in which the number of screws for the SMA bar was increased from three to nine and the screw ends were flattened to avoid stress concentration (Figure 2-6). The SMA bars fractured at a strain of 6.4% inside the coupler as a result of stress concentration. Table 2-2 presents a summary of experimental studies performed on SSCs. It can be concluded that the tensile strength of this type of coupler may be sufficient for seismic applications, depend- ing on the product and manufacturer. However, these couplers generally limit the strain capacity (a) Nine-screw coupler (b) Failure of shape memory alloy inside coupler Source: Alam et al. (2010). Figure 2-6. Shear screw coupler specimen for SMA bars. Study Coupler Bar Size Bar Type Mode of Failure Remarks Lloyd (2001) Three-screw Bar Lock L- Series No. 6 (Ø19 mm) and No. 8 (Ø25 mm) ASTM A615 Grade 60 Bar pullout, bar fracture 90% of the ultimate strength of anchoring bars was achieved. Hillis and Saiidi (2009) Three-screw Zap Screwlok Type 2 No. 4 SMA bars to steel bars NiTi SMA and Grade 60 steel bars SMA bar fractured inside grip No coupler failure and no SMA bar fracture inside the coupler was observed. Rowell et al. (2009) Seven-screw Zap Screwlok Type 2 No. 10 (Ø32 mm) ASTM A615 Grade 60 Mainly bar fracture inside couplers Lower strength and significantly lower strain capacities were observed due to premature failure of bars. Huaco and Jirsa (2012) Three-screw (ERICO S- series) and four- screw (ERICO B-Series) No. 8 (Ø25 mm) ASTM A706 Grade 60 Bar fractured inside or away from coupler Bars fractured on the edge of three- screw couplers and fractured outside of four-screw couplers. Three times higher strain capacity for longer couplers was observed. Alam et al. (2010) Three-screw Bar-Lock S Nos. 4–6 (Ø13–19 mm) Grades 40 and 60 Bar fracture Low slippage was observed before yielding; sufficient strength was reported. Note: NiTi = nickel titanium. Table 2-2. Summary of studies on shear screw couplers.

12 Proposed AASHTO Seismic Specifications for ABC Column Connections of the bars because of the stress concentration under the screws. This limitation is critical for seismic applications, especially in the plastic hinge region, where large deformations are expected. 2.2.3.2 Performance of Columns with SSCs Cruz-Noguez and Saiidi (2012) tested a quarter-scale, four-span bridge in which SSCs were used in one of the three two-column bents to connect SMA bars to steel bars anchored in the foot- ing and steel bars above the plastic hinge at the column base (Figure 2-7). The bridge was tested on shake tables under seven simulated earthquake runs. The measured force-displacement envelopes for the bent are shown in Figure 2-8. It can be seen that the columns performed in a ductile manner. The measured maximum drift ratio of the bent exceeded 5% with no bar fracture or cou- pler failure. The testing of the bridge model was terminated because of failure in a different bent. Huaco and Jirsa (2012) and Huaco (2013) incorporated short and long SSCs in the repair of two severely damaged concrete columns, which were tested in double curvature under cyclic loading. The repair of the first column consisted of the utilization of short SSCs at the column base after concrete and the original reinforcement have been removed (Figure 2-9) as well as incorporation of glass fiber–reinforced polymer (GFRP) at the top of the column. Since (a) Shape memory alloy bars connected to couplers (b) Shear screw couplers Source: Hillis and Saiidi (2009). Figure 2-7. Two-column bent with SSCs. Source: Cruz-Noguez and Saiidi (2012). Note: Long = longitudinal; Tran = transverse. (a) Longitudinal direction (b) Transverse direction Figure 2-8. Force-displacement envelope for two-column bent with shear screw couplers.

State-of-the-Art Literature Review 13 (a) Original column after testing (b) Short shear screw couplers at base (c) Repaired column Source: Huaco (2013). Figure 2-9. First repaired column with short shear screw couplers and glass fiber–reinforced polymer. (a) Original column after testing (b) Short (left) and long (right) shear screw couplers at base (c) Repaired column Source: Huaco (2013). Figure 2-10. Second repaired column with short and long shear screw couplers. the damage in the second column was severe, the column concrete and reinforcement were completely replaced with new materials, but the original column was cut into two halves, and each repaired segment was tested as a cantilever (Figure 2-10). Short SSCs were incorporated in the base of one of the columns to connect the repaired segment to the existing footing, and long SSCs were used in the second piece at the base. The difference between the short and long couplers was the length of the coupler to accommodate a higher number of screws. Three screws were used in short couplers for each bar and four screws were used in the long coupler for each bar. Figure 2-11 shows the measured original and repaired column force-drift ratio hysteresis curves. It can be seen than the columns with short SSCs exhibited lower drift capacity as compared with the reference test model but with similar lateral strength, and the

14 Proposed AASHTO Seismic Specifications for ABC Column Connections column with longer SSC couplers showed higher drift capacity and higher strength as com- pared with the reference test model. The limited drift capacity of the columns with short SSCs was attributed to the insufficient strain capacity of the short SSCs. Six bars fractured under low displacement due to stress concentration under the end screw of the short SSCs close to the column base. Flat-end screws reduced stress concentration, thus shifting the bar fracture point to outside the long SSCs and resulting in large displacement capacity for the column. Yang et al. (2014) used SSCs (similar to the long couplers in the previous study) in the plastic hinge of a severely damaged bridge column to replace buckled bars with new reinforcement (Figure 2-12). This coupler is the only SSC that is rated as the “ultimate” splice by Caltrans. Carbon fiber–reinforced polymer (CFRP) was utilized on the entire column height to increase confinement and shear capacity of the column. The repaired column (R-Calt-1) was tested under reversed cyclic loading to failure. Figure 2-13 shows the measured force-displacement Source: Huaco (2013). (a) First column (b) Columns with short versus long shear screw couplers Figure 2-11. Force-drift ratio hysteresis of repaired columns with shear screw couplers. (a) Plastic hinge of original column after testing (b) Long shear screw couplers at base (c) Repaired column Source: Yang et al. (2014). Figure 2-12. Repaired column with shear screw couplers.

State-of-the-Art Literature Review 15 relationships of the repaired and original columns. The repaired column exhibited a displace- ment ductility capacity of 4.9, which was 4% higher than that of the original column. The lateral strength of the repaired column was on average 20% higher than that of the original column. In summary, it can be concluded that columns with the long SSCs are expected to exhibit better seismic performance relative to columns with short SSCs, since the latter may prema- turely fail from stress concentration under the screws. Table 2-3 presents a summary of the findings regarding the seismic performance of columns with SSCs. (a) Force-displacement hysteresis (b) Force-displacement envelope Source: Yang et al. (2014). Figure 2-13. Response of repaired column with shear screw couplers. Reference Column Geometry Coupler Length Remarks Cruz-Noguez and Saiidi (2012) No. of columns: one two-column bent Scale factor: 25% Section: circular Diameter: 12 in. (305 mm) Longitudinal bars: nine No. 4 (Ø13 mm) SMA Transverse bars: 0.11 in. (Ø2.9 mm) spirals at 1.25 in. (32 mm) 14db: used at two levels The bent with SSCs showed a drift ratio capacity of more than 5% with no bar or coupler failure [bottom couplers embedded 12db into the footing on center; top couplers were 22 in. (1.83Dc) away from the bottom couplers on center]. Huaco (2013) No. of columns: three Scale factor: 100% Variable: repair method Section: square Side diameter: 16 in. (406 mm) Longitudinal bars: 8 No. 8 (Ø25 mm) Transverse bars: No. 3 (Ø10 mm) ties at 6 in. (152 mm) Short (6.8db): used either at two levels or one level; Long (10db): used in one level Two repaired columns with short SSCs showed lower drift capacity as compared with the original column but with comparable lateral strength. The repaired column with long couplers (couplers used only at the column base) exhibited similar performance to the original column. Yang et al. (2014) No. of columns: one Scale factor: 50% Section: interlocking Diameter: 24 in. (610 mm) by 36 in. (914 mm) Longitudinal bars: 20 No. 8 (Ø25 mm) Transverse bars: No. 4 (Ø13 mm) spirals at 2.75 in. (70 mm) 10db: used at two levels Repaired column showed 4% higher displacement ductility and 20% higher base shear capacity as compared with the original test model [bottom end of the bottom couplers was embedded 5db into the footing; top couplers were 36 in. (1.0Dc) away from the bottom couplers on center]. Note: db = longitudinal bar diameter; Dc = either the column diameter or the dimension of the column’s largest side. Table 2-3. Summary of seismic performance of column test models with shear screw couplers.

16 Proposed AASHTO Seismic Specifications for ABC Column Connections 2.2.4 Headed Bar Couplers 2.2.4.1 Performance of Headed Bar Couplers To splice bars with headed bar couplers (HCs), each bar is heated and the bar end is flattened to form a head. Experimental investigation on HCs has been presented in a few studies. Sritharan et al. (1999) reported testing of several HC specimens before utilizing these couplers in a cap beam test model. A consistent mode of failure was observed in the tests, but no additional information was reported about the coupler tests. Rowell et al. (2009) tested nine specimens with HCs. The loading rates and the bar size were the same as those presented for SSCs in Section 2.2.3. Full ultimate strength and significant strain capacity were observed in slow strain rate tests. The average maximum strain was 11%, as compared with 10% for the control bars. The bars fractured outside the heat-affected zone in these tests (Figure 2-14a). At the intermediate strain rate, the average ultimate load for the HC splice was comparable to the reference bars. The average maximum strain was reduced to 6% as compared with 10% measured for the control bars. Bars fractured either inside or outside of the heat-affected zones (Figure 2-14b). At high strain rates, the average ultimate load for the HC splice was 10% lower than that of the reference bars. The average maximum elongation was reduced to 7% in comparison to the 14% achieved in the control bar tests. Similar to the intermediate strain rate tests, the bar rupture was either in the heat-affected zones or outside the heat-affected zone (Figure 2-14c). Haber et al. (2013) performed monotonic, cyclic, and dynamic tests on No. 8 (Ø25 mm) ASTM A706 Grade 60 steel bars in which the bars were connected by HCs. Four specimens were tested under static and dynamic loading protocols, respectively, and two specimens were tested (a) Slow strain rate tests (failure shown on right side of coupler) (b) Intermediate strain rate tests (failure shown on right side of coupler) (c) High strain rate tests (failure shown on right side of coupler) Source: Rowell et al. (2009). Figure 2-14. Headed bar coupler specimen tests.

State-of-the-Art Literature Review 17 under cyclic loading protocols. The mode of failure was bar fracture outside the couplers in all specimens (Figure 2-15). Figure 2-15d shows the measured stress–strain relationship of speci- mens under static and dynamic testing. It can be seen that HCs allowed steel bars to sufficiently deform, and the ultimate strain capacities of bars were achieved in both tests. Furthermore, large strains were measured in the coupler region, which can be beneficial when these couplers are used in plastic hinges. Similar behavior was observed in cyclic load tests. A summary of the available test data on the performance of HCs is presented in Table 2-4. It can be concluded that the ultimate stress and strain capacities of bars can be achieved utilizing this coupler type. 2.2.4.2 Performance of Columns with Headed Bar Couplers Lehman et al. (2001) repaired a severely damaged column with HCs and then tested the repaired column under cyclic loading. The concrete in the column plastic hinge area as well as the upper part of the footing was removed and HCs were used to connect new reinforcement to the column and the existing footing bars (Figure 2-16a). The measured force-displacement (a) Monotonic test (b) Cyclic test (c) Dynamic test (d) Stress–strain relationship Source: Haber et al. (2013). Figure 2-15. Headed bar coupler specimen test. Study Bar Size Bar Type Mode of Failure Remarks Sritharan et al. (1999) NA NA NA Consistent mode of failure. Rowell et al. (2009) No. 10 (Ø32 mm) ASTM A615 Grade 60 Bar fractured inside or outside of couplers. Approximately full ultimate strength was achieved in all tests, but the strain capacity of the system was adversely affected by the increasing strain rate. Haber et al. (2013) No. 8 (Ø25 mm) ASTM A706 Grade 60 Bar fractured outside of couplers. Full ultimate stress and strain capacities were achieved under static, cyclic, and dynamic tests. Strains in the coupler region were more than 8%. Note: NA = not available. Table 2-4. Summary of studies on headed bar couplers.

18 Proposed AASHTO Seismic Specifications for ABC Column Connections hysteretic relationships for the original and the repaired columns are shown in Figure 2-16b. It can be seen that the lateral strength in the repaired column was higher and the displacement capacity was improved as compared with the original column. The higher displacement capacity of the repaired column was because of the 2 in. (51 mm) of extra clear cover, which increased the resistance of the column longitudinal bar against buckling and fracture. Haber et al. (2014) tested two half-scale precast bridge columns in which HCs were incorpo- rated in the plastic hinge area to connect the columns to the footings (Figure 2-17). A precast pedestal was utilized in one of the specimens to shift the couplers away from footing surface and to investigate the effects of lower moment demand on the couplers. The columns were tested under slow cyclic loading to failure, and the response was compared with that of a reference cast- in-place (CIP) column. The test results showed that both columns were viable precast columns for high-seismic areas, since similar mode of failure, plastic hinge damage, lateral strength, and displacement ductility capacities were observed as compared with the CIP column. Figure 2-17e shows the measured force-drift envelopes of the columns. It can be seen that all columns exhib- ited excellent displacement ductility capacity, which was 40% more than the AASHTO allowable displacement ductility demand for single column bents (ductility of 5, which is equivalent to a 6.75% drift ratio for CIP columns). Therefore, it can be concluded that HCs allow columns to deform freely and to develop large displacement ductility. Tazarv and Saiidi (2015c) incorporated HCs in a half-scale precast bridge column test model to link reinforcing nickel titanium (NiTi) SMA bars to reinforcing steel bars (Figure 2-18a). Other advanced materials such as ultrahigh-performance concrete (UHPC) and engineered cementitious composite (ECC) were utilized in the column–footing connection and the plastic hinge area, respec- tively. The column was tested in a manner similar to that of the columns tested by Haber et al. (2013), and the response was compared with that of CIP columns. The column with advanced materials [headed bar coupler with SMA (HCS)] showed minimal damage, insignificant residual displacement, higher lateral strength, and slightly better displacement capacity as compared with CIP columns (Figure 2-18b). The HCs accommodated the column’s large inelastic deformations without failure. Nakashoji and Saiidi (2014) tested two CIP SMA-reinforced ECC bridge columns under cyclic loading. A reference RC column was also tested. HCs were utilized to connect reinforcing SMA bars to reinforcing steel bars. The test variable was the length of the SMA bars incorpo- rated in the plastic hinge region (Figure 2-19a), in which one column (called SR99-LSE) was (a) Detail (b) Force-displacement relationship Source: Lehman et al. (2001). Figure 2-16. Repaired column with headed bar couplers.

State-of-the-Art Literature Review 19 (c) HCNP plastic hinge region (d) HCPP pedestal (e) Force-drift envelope Source: Haber et al. (2013). Note: HCPP = headed bar coupler with precast pedestal; HCNP= headed bar coupler with no pedestal. (a) HCNP drawing (b) HCPP drawing 0 50 100 150 200 250 300 350 0 10 20 30 40 50 60 70 80 0 1 2 3 4 5 6 7 8 9 10 11 L at er al F or ce ( kN ) L at er al F or ce ( ki ps ) Drift (%) CIP Column HCNP Column HCPP Column l = 7.36AASHTO Allowable Displacement Ductility Demand l = 5.0 Figure 2-17. Precast columns with headed bar couplers.

20 Proposed AASHTO Seismic Specifications for ABC Column Connections (a) Shape memory alloy steel bar cage (b) Force-drift envelope Source: Tazarv and Saiidi (2015c). 0 50 100 150 200 250 300 350 0 10 20 30 40 50 60 70 80 0 1 2 3 4 5 6 7 8 9 10 11 L at er al F or ce ( kN ) L at er al F or ce ( ki ps ) Drift (%) CIP Column HCS Column μ = 7.36 AASHTO Allowable Displacement Ductility Demand Figure 2-18. Precast columns with HCs. (a) Shape memory alloy steel bar cage (b) Force-drift envelope Source: Nakashoji and Saiidi (2014). Figure 2-19. Cast-in-place columns with headed bar couplers.

State-of-the-Art Literature Review 21 built with 1.0 Dc long SMA bars (Dc is the column side dimension) and another column (called SR99-SSE) was constructed with SMA bars 0.75 Dc long. The drift capacity in both columns was substantially higher than that of the reference column (Figure 2-19b), confirming that HCs used in the plastic hinge region allowed the columns to develop full plastic moment with high drift capacity. Table 2-5 presents a summary of the findings on the seismic performance of columns with HCs. 2.2.5 Grouted Couplers 2.2.5.1 Performance of Grouted Couplers Noureddine (1996) performed four tests on No. 18 (Ø57 mm) ASTM A615 and A706 Grade 60 steel bars in which grouted couplers (GCs) were used to splice the bars. Two speci- mens per each bar type were tested (Figure 2-20). The specimens were tested under monotonic Table 2-5. Summary of seismic performance of column test models with headed bar couplers. Reference Column Geometry Coupler Length Remarks Lehman et al. (2001) No. of columns: one Scale factor: 33% Section: circular Diameter: 24 in. (610 mm) in original column, increased to 28 in. (711 mm) in repaired column Longitudinal bars: 11 No. 5 (Ø16 mm) Transverse bars: No. 2 (Ø6 mm) spirals at 1.25 in. (32 mm) in original column, increased to No. 3 (Ø10 mm) spirals at 2.25 in. (57 mm) in repaired column 3.4db: used at two levels The repaired column showed higher lateral strength and improved displacement capacity over the original column. Higher strength was because of higher ultimate strength for longitudinal bars, and higher displacement capacity was because cover concrete was increased by 2 in. (51 mm), which delayed buckling of the longitudinal bars. Haber et al. (2014) No. of columns: two Scale factor: 50% Variable: pedestal Section: circular Diameter: 24 in. (610 mm) Longitudinal bars: 11 No. 8 (Ø25 mm) Transverse bars: No. 3 (Ø10 mm) spirals at 2 in. (51 mm) 3.13db: used at two levels Both columns showed similar seismic performance to the CIP column performance [bottom couplers were installed 5db above either the footing surface or pedestal; top couplers were installed 12 in. (0.5Dc) above the bottom couplers]. Nakashoji and Saiidi (2014) No. of columns: two Scale factor: 30% Variable: SMA bar length Section: square with circular arrangement for bars, Side diameter: 20 in. (508 mm) Longitudinal bars: 16 No. 4 (Ø13 mm) SMA Transverse bars: No. 3 (Ø10 mm) spirals at 1.8 in. (48 mm) 4.26db: used at two levels (the coupler length is smaller if the same size bars are connected) Columns showed better seismic performance than the CIP column. Couplers were used to connect SMA bars to steel bars (bottom couplers were installed in the footings; top couplers were installed 0.75Dc and 1.0Dc above the bottom couplers). Tazarv and Saiidi (2015c) No. of columns: one Scale factor: 50% Section: circular Diameter: 24 in. (610 mm) Longitudinal bars: 10 No. 10 (Ø32 mm) SMA Transverse bars: No. 3 (Ø10 mm) spirals at 2 in. (51 mm) 3.3db: used at two levels (the coupler length is smaller if the same size bars are connected) The column showed improved seismic performance over CIP. Couplers were used to connect SMA bars to steel bars [bottom couplers were installed immediately above the footing surface; top couplers were installed 20 in. (0.85Dc) above the bottom couplers]. Note: db = longitudinal bar diameter; Dc = either the column diameter or the dimension of the column’s largest side.

22 Proposed AASHTO Seismic Specifications for ABC Column Connections tensile loading to fracture. The average ultimate load for GCs was comparable to that of the control bars. The average ultimate strains were approximately 7% and 12% for A615 and A706 bars, respectively. Three specimens failed due to bar rupture away from the coupler region and one sleeve failed at 3% strain. The Michigan Department of Transportation (DOT) tested No. 6 (Ø19 mm) and No. 11 (Ø36 mm) epoxy-coated bars that were connected by two types of GCs: threaded grouted couplers (TGCs) and splice sleeve grouted couplers (SSGCs) (Jansson 2008). For TGCs, the average slip was 0.004 in. (0.1 mm) for the No. 6 (Ø19 mm) bars and 0.005 in. (0.13 mm) for the No. 11 (Ø36 mm) bars. Fatigue testing demonstrated that the splices were able to withstand at least 1,000,000 cycles with a stress range of 18 ksi (124 Mpa), as specified by the AASHTO LRFD Bridge Design Specifications, or AASHTO LRFD (AASHTO 2013). In ultimate load testing, all specimens except for one (Specimen 11AI) exceeded the AASHTO LRFD and Michigan DOT requirement of 125% of the yield strength, and Specimen 11AI failed at a lower load. For SSGCs, the average slip was 0.007 in. (0.18 mm) for the No. 6 (Ø19 mm) bars and 0.009 in. (0.23 mm) for the No. 11 (Ø36 mm) bars. Fatigue testing demonstrated that these splices were also able to withstand at least 1,000,000 cycles with a stress range of 18 ksi (124 Mpa). The average ulti- mate strength was 166% and 175% of the yield strength for the No. 6 (Ø19 mm) and No. 11 (Ø36 mm) bars, respectively. For TGCs, the threaded section of the coupler was found to be the common failure location in all splice tests (Figure 2-21a), either as a result of fracture of the bar at the reduced threaded section or by shear failure of the threads themselves (Figure 2-21b). For SSGCs, different failure modes were observed, but there was no discernable effect of the type of failure mode on the ultimate load of the splices (Figure 2-21, c and d). Since both TGCs and SSGCs performed well under testing for slip, fatigue, ultimate strength, and creep, they were recommended for use by the Michigan DOT. It should be noted that there were no data on the strain capacity, which is critical for the seismic applications. Rowell et al. (2009) tested nine No. 10 (Ø32 mm) ASTM A615 Grade 60 steel bars with grouted sleeve couplers. The specimens were subjected to the following three strain rates: slow (3,000–4,000 µe/s), intermediate (62,000–65,000 µe/s), and high (3.2 × 106 to 3.8 × 106 µe/s) tests. For each strain rate, three specimens were tested. The average yield and ultimate load for GCs were comparable to those measured in the reference bars at a slow strain rate (Figure 2-22a). The average maximum strain was 6%, as compared with 10% for the control bars. At intermediate strain rates, the average yield and the ultimate loads for GCs were comparable to those of the control bars (Figure 2-22b). The average maximum elongation was reduced to 9% as compared with 11% measured in the control bars. At high Source: Noureddine (1996). Figure 2-20. Grouted coupler specimen tests.

State-of-the-Art Literature Review 23 strain rates, the average yield and the ultimate loads for GCs matched those of the control bars (Figure 2-22c). However, the average maximum strain was lower—8% as compared with 14% for the control bars. Three different failure modes were observed in these tests: rupture of the bar away from the grouted sleeve, pull-out of the bar from the grouted sleeve, and failure of the sleeve. Haber et al. (2013) performed tensile tests on No. 8 (Ø25 mm) ASTM A615 Grade 60 steel bars connected with GCs (Figure 2-23a). The specimens with GC connections were subjected to two strain rates: slow (1,000–8,000 µe/s) and high (70,000 µe/s). Three specimens for each strain rate were tested. Furthermore, cyclic tests were performed to investigate the effect of strain reversals. The average ultimate load for GC splice was comparable to the reference bar ultimate strength. The strain in the sleeve region was very low, with maximum strain of 0.7%. Figure 2-23b shows the measured stress–strain relationship of specimens under static and Figure 2-21. Threaded grouted and grouted sleeve coupler tests. (a) Bar fracture inside threaded grouted coupler (b) Thread failure in threaded grouted coupler (c) Pullout failure in splice sleeve grouted coupler (d) Bar fracture in splice sleeve grouted coupler Source: Jansson (2008).

24 Proposed AASHTO Seismic Specifications for ABC Column Connections (a) Slow strain rate tests (b) Intermediate strain rate tests (c) High strain rate tests Source: Rowell et al. (2009). Figure 2-22. Threaded grouted and grouted sleeve coupler tests. (b) Stress–strain relationship Source: Haber et al. (2013). (a) Monotonic Test Figure 2-23. Grouted coupler specimen test.

State-of-the-Art Literature Review 25 dynamic testing. The loading rate had negligible effects on the coupler performance, and the ultimate capacities of the bars were developed. All the GC specimens failed due to bar fracture away from the coupler region. Ameli et al. (2015) tested six TGCs connecting No. 8 (Ø25 mm) Grade 60 reinforcing steel bars. Bars pulled out from the grouted end of the couplers in all samples (Figure 2-24). The splice withstood 1.2 times the bar yield strength prior to the bar pullout. This failure mode was attributed to the relatively low strength of the grout that was used in the test splice. The grout compressive strength was 9.4 ksi (64.8 Mpa), but the grout strength in the Michigan DOT tests (Jansson 2008) with the same coupler type was 14.7 ksi (101.3 Mpa). The manufacturer 28-day specified grout compressive strength is 8,500 psi (58.6 Mpa). A summary of available test data on the performance of grouted sleeve couplers is presented in Table 2-6, which is used in the following sections for the evaluation of this coupler type. 2.2.5.2 Performance of Columns with Grouted Couplers Haber et al. (2014) tested two half-scale precast bridge columns utilizing GCs (Figure 2-25, a and b). Similar to their previous test models, one column was built with precast pedestal and another column was constructed without pedestal. Both columns were tested under cyclic loading to failure. The lateral strength of the columns was the same up to the first bar fracture as compared with the reference column (CIP), but the displacement capacity and the displace- ment ductility capacity of both columns with spliced bars were only 60% of those of the CIP (Figure 2-25e). The relatively low displacement capacity was because of strain concentration under grouted sleeve couplers or precast pedestal in which bulky GCs behaved similarly to large size reinforcement, shifting the yielding below the coupler region; and also GDs incorporated in the pedestal made the pedestal very stiff, thereby shifting damage below the pedestal. Tazarv and (a) Test setup and failure Source: Ameli et al. (2015). (b) Bar pullout Figure 2-24. Threaded grouted sleeve coupler specimen tests.

26 Proposed AASHTO Seismic Specifications for ABC Column Connections Study Bar Size Bar Type Mode of Failure Remarks Noureddine (1996) No. 18 (Ø57 mm) ASTM A615 Grade 60, ASTM A706 Grade 60 Bar fractured in three tests, coupler failed in one test Ultimate loads were comparable to the reference bar load; bar fractured in three tests and large strains (e.g., 7% and 12%) were measured. Jansson (2008) No. 6 (Ø19 mm), No. 11 (Ø36 mm) Grade 60 Bar fracture, GC fracture, shear failure of threads Minor slippage occurred, but still met AASHTO LRFD requirements. Ultimate loads and fatigue testing exceeded the requirements. No data on the strain capacity were reported. Rowell et al. (2009) No. 10 (Ø32 mm) ASTM A615 Grade 60 Bar pullout, bar fracture, GC fracture For slow, intermediate, and high strain tests, all specimens resulted in maximum strain less than the control bars. Haber et al. (2013) No. 8 (Ø25 mm) ASTM A615 Grade 60 Bar fracture Bar fractured away from the coupler region in static, cyclic, and dynamic tests; ultimate stress and strain capacities of the bars were developed. Ameli et al. (2015) No. 8 (Ø25 mm) Grade 60 Bar pullout Bar pulled out in all six samples. The grout strength was 9.4 ksi, 36% weaker than the grout used in Jansson (2008). Table 2-6. Summary of studies on grouted sleeve couplers. Saiidi (2014) proposed a new detail for GC columns to enhance displacement ductility capacity by debonding the longitudinal bars within the pedestal and using a CIP pedestal (Figure 2-25, c and d). A column was constructed using the new detail and was tested under the same loading protocol as that used by Haber et al. (2014); the force-drift envelope for this column is shown in Figure 2-25e. The displacement capacity was increased by 47%, and the displacement ductility capacity was increased to 7, which met the design target. Pantelides et al. (2014) tested six half-scale precast bridge columns utilizing GCs in which normal GCs (such as that shown in Figure 2-23) were used in the column–footing connections of three specimens, and modified GCs (such as that shown in Figure 2-24) were utilized in the column–cap beam connections of the remaining three test models (Figure 2-26a). Couplers were grouted at both ends of the normal GCs, but in the modified GCs, one end of couplers was threaded to reduce the coupler length. The columns were tested under slow cyclic loading. The measured force-drift response of the columns with standard couplers at the column base is shown in Figure 2-26b. These columns exhibited 25% to 40% lower displacement ductility capacity as compared with the reference CIP column. The displacement ductility capacity of the CIP column was 8.9. The most ductile column was the one with GCs installed immediately above the footing surface and 8db-debonded bars below the coupler level. Figure 2-26c shows the force-drift relationship of column test models with modified couplers used in column–cap beam connections. It can be seen that the displacement ductility capacity of these columns was substantially lower (41% to 69%) than that of the reference column, which confirms that the modified GCs were not suitable for high-seismic regions. A new detailing for building columns was proposed by Belleri and Riva (2012) (and later by Popa et al. 2015). In this detail, corrugated steel ducts are placed in the plastic hinge of columns, footing starter bars are anchored in the ducts, and then ducts are grouted after the column installation. This connection can be considered as either a GC or a GD connection. Even though similar seismic behavior to reference specimens was reported in these studies, the application of

State-of-the-Art Literature Review 27 (a) GCNP or GCPP drawing (b) Steel cage with GCs (c) GCDP drawing (d) Debonded bars in pedestal (e) Force-drift envelope for GC columns Source: Haber et al. (2014) (a and b); Tazarv and Saiidi (2014) (c–e). Note: GCPP = grouted coupler with precast pedestal; GCNP = grouted coupler without pedestal; GCDP = grouted coupler with new detail. 0 50 100 150 200 250 300 350 0 10 20 30 40 50 60 70 80 0 1 2 3 4 5 6 7 8 9 10 11 Ba se S he ar ( kN ) Ba se S he ar ( ki ps ) Drift (%) GCNP Column GCPP Column GCDP Column CIP Column μ = 4.52 μ = 4.53 μ = 7.07 μ = 7.36 AASHTO Allowable Displacement Ductility Demand Figure 2-25. Precast columns with grouted couplers.

28 Proposed AASHTO Seismic Specifications for ABC Column Connections this detailing would be difficult for bridge columns, owing to size of the ducts and large number of reinforcing bars that are needed in bridge columns. A summary of the findings from experimental studies of columns with GCs is presented in Table 2-7. 2.2.6 Threaded Couplers 2.2.6.1 Performance of Threaded Couplers Noureddine (1996) performed tensile tests on No. 18 (Ø57 mm) ASTM A615 and A706 Grade 60 reinforcing steel bars. Four tapered threaded coupler (TC) specimens were tested under monotonic tensile loading to failure. The average ultimate load for TCs was 15% smaller than that observed for the control bars. The average ultimate strain was 2% for both A615 and A706 bar specimens, which was extremely low. All the TC specimens failed as a result of stripping of the threads (Figure 2-27). Rowell et al. (2009) tested 18 No. 10 (Ø32 mm) ASTM A615 Grade 60 mechanical bar splices in which nine tapered TCs and nine straight TCs were incorporated to connect the bars (Figure 2-28). Both type of coupler connections were subjected to three strain rates: slow (3,000–4,000 µe/s), intermediate (62,000–65,000 µe/s), and high (3.2 × 106 to 3.8 × 106 µe/s). (a) Column–footing connections (top) and column–cap beam connections (bottom) (b) Force-drift envelope for footing grouted coupler columns (c) Force-drift envelope for cap beam grouted coupler columns Source: Pantelides et al. (2014). Figure 2-26. Precast columns with grouted couplers.

Source: Noureddine (1996). Figure 2-27. Tapered threaded coupler specimen tests. (a) Tapered threaded coupler (b) Straight threaded coupler Source: Rowell et al. (2009). Figure 2-28. Threaded coupler specimen tests under intermediate strain rate. Reference Column Geometry Coupler Length Remarks Haber et al. (2014) No. of columns: two Scale factor: 50% Variable: pedestal Section: circular Diameter: 24 in. (610 mm) Longitudinal bars: 11 No. 8 (Ø25 mm) Transverse bars: No. 3 (Ø10 mm) spirals at 2 in. (51 mm) 14.6db: used at one level Both columns showed 40% lower displacement capacity as compared with the CIP column performance (couplers were installed immediately above either footing surface or pedestal). Tazarv and Saiidi (2014) No. of columns: One Scale factor: 50% Section: circular Diameter: 24 in. (610 mm) Longitudinal bars: 11 No. 8 (Ø25 mm) Transverse bars: No. 3 (Ø10 mm) spirals at 2 in. (51 mm) 14.6db: used at one level This column showed the same seismic performance as the CIP column (couplers were installed immediately above a pedestal). Pantelides et al. (2014) No. of columns: three Scale factor: 50% Variable: coupler location Section: octagonal Diameter: 21 in. (533 mm) Longitudinal bars: 6 No. 8 (Ø25 mm) Transverse bars: No. 4 (Ø13 mm) spirals at 2.5 in. (63 mm) 14.6db: standard couplers were used at one level in column–footing connections These columns showed 25% to 40% lower displacement ductility capacity as compared with a reference column. The best performance was observed for a column with couplers immediately above the footing surface and debonded bars below the couplers. Pantelides et al. (2014) No. of Columns: three Scale factor: 50% Variable: coupler location Section: octagonal Diameter: 21 in. (533 mm) Longitudinal bars: 6 No. 8 (Ø25 mm) Transverse bars: No. 4 (Ø13 mm) spirals at 2.5 in. (63 mm) 8.6db: modified couplers were used in column–cap beam connections These columns showed 41% to 69% lower displacement ductility capacity as compared with a reference column. The best performance was observed for a column with couplers embedded in the cap beam. Note: db = longitudinal bar diameter; Dc = either the column diameter or the dimension of the column’s largest side. Table 2-7. Summary of seismic performance of column test models with grouted sleeve couplers.

30 Proposed AASHTO Seismic Specifications for ABC Column Connections The average yield and the ultimate loads for both coupler types were comparable to those of the reference bars at the slow strain rate. For tapered TCs, the average maximum strain was 11% as compared with 10% measured for the control bars. For straight TCs, the average maxi- mum strain was 7%, which was lower than that of the control bar. For tapered TCs at the inter- mediate strain rate, the average yield load was comparable to that of the control bars, but the average ultimate load for the tapered TCs was 8% lower than that for the control bars. The average maximum elongation was 50% lower than that of the control bars. For straight TCs, both yield and ultimate loads were close to those for the control bar. The ultimate elongation was 10% in comparison to 11% for control bar. At the high strain rate, the average yield load for tapered TCs was the same as the control bar yield load, but the ultimate load was 24% lower than that for the control bar. The average maximum strain was 2% in comparison to 14% for control bars. For straight TCs, both the yield and ultimate loads were comparable to those measured in the control bars. The maximum strain measured was 11%, as compared with 12% measured in the control specimen. In summary, tapered TCs showed inferior performance with premature failure of the bars at the threads. However, the ultimate stress and strain capacities of the bars were developed when the straight TCs were utilized. A summary of the available test data on the performance of TCs is presented in Table 2-8. 2.2.6.2 Performance of Columns with Threaded Couplers Saiidi and Wang (2006) utilized TCs in a quarter-scale bridge column to connect reinforcing SMA bars to steel bars in the plastic hinge region (Figure 2-29a). The column was tested on a shake table under 11 runs (Run 11 was 300% of the design level earthquake), during which the drift ratio was 4.8%. The test was stopped after Run 11 to prevent SMA bar failure. The column was then repaired by replacing the conventional concrete in the plastic hinge region with ECC. The repaired column was tested under 15 runs (Run 15 was 400% of the design level earthquake), during which a drift capacity of 5.7% was reported. Figure 2-29b shows the force-displacement envelopes of the two columns. The test was stopped because the selected motion was not able to impose more deformation to the column. Minor damage of ECC and no SMA bar fracture was observed in these tests. The TCs performed well by allowing the column to deform freely and by transferring the stresses to the adjoining bars. Saiidi et al. (2009) tested two CIP columns in which TCs were incorporated in the plastic hinge region of these columns to link SMA bars to steel bars (Figure 2-30a). Conventional concrete was utilized in one of the SMA columns (RNC) and ECC was used in the plastic hinge of another SMA column (RNE). A reference column with conventional concrete and reinforcing steel bars was also tested (RSC). These columns were tested under cyclic loading to failure. Figure 2-30b shows the measured force-drift envelope for the columns. The columns with TCs showed equal or improved drift capacity compared to the reference column confirming the suitability of TCs Table 2-8. Summary of studies on shear screw couplers. Study Bar Size Bar Type Mode of Failure Remarks Noureddine (1996) No. 18 (Ø57 mm), taper threaded ASTM A615 Grade 60, ASTM A706 Grade 60 Shear failure of threads Ultimate loads were 15% lower than for the control bars; premature failure (2% strain capacity) was reported as a result of failure of the threads. Rowell et al. (2009) No. 10 (Ø32 mm) ASTM A615 Grade 60 Tapered bars: bar rupture, thread failure Straight bars: bar rupture Bars connected with tapered TCs showed significantly lower strain capacity resulting from failure of the threads, but bars connected with straight TCs exhibited large stress and strain capacities similar to those of the reference bars.

State-of-the-Art Literature Review 31 (a) Shape memory alloy steel cage (b) Force-displacement envelope Source: Saiidi and Wang (2006). Figure 2-29. Repaired column with threaded bar couplers. Figure 2-30. Cast-in-place columns with threaded bar couplers. (a) Shape memory alloy steel cage (b) Force-drift envelope Source: O’Brien et al. (2006). Reference Column SMA-ECC Col. SMA-Concrete Col. in the plastic hinge of columns located in high-seismic regions. No bar fracture was reported up to 10% drift ratio cycles. The test was continued for RNE to 14% drift ratio cycles in which one of the SMA bars fractured at the thread inside one of the bottom couplers. A quarter-scale bridge column was tested by Varela and Saiidi (2014) in which TCs were utilized to link reinforcing copper-based SMA bars to reinforcing steel bars (Figure 2-31a). The column was tested on a shake table up to 350% of the design level earthquake. The column withstood a drift ratio of 11.8%, in which two SMA bars fractured in a ductile manner. The ECC

32 Proposed AASHTO Seismic Specifications for ABC Column Connections in the column plastic hinge was removed after testing to locate the bar fracture. It was found that both SMA bars fractured away from the TCs as shown in Figure 2-31b. The authors also used TCs in the plastic hinge of six deconstructible bridge columns and a bridge model with three two-column bents. The TCs maintained the integrity of the connections in these tests and allowed the columns to deform freely (Varela and Saiidi 2016). Table 2-9 summarizes the findings from the experimental studies of columns with TCs. It is worth mentioning that all of the abovementioned test models utilized straight TCs Figure 2-31. Cast-in-place column with threaded bar couplers. (a) Cage with threaded couplers (b) Location of shape memory alloy bar fracture Source: Varela and Saiidi (2014). Reference Column Geometry Coupler Length Remarks Saiidi and Wang (2006) No. of columns: one Scale factor: 25% Section: circular Diameter: 12 in. (305 mm) Longitudinal bars: 15 No. 4 (Ø13 mm) SMA Transverse bars: 0.19 in. (Ø4.9 mm) spirals at 1.6 in. (41 mm) 4.0db: used at two levels Both original and repaired columns exhibited large displacement capacities (greater than 5.8% drift ratio) with no bar fracture [bottom couplers were installed 4 in. (102 mm) below the footing surface; top couplers were installed 1.17Dc above the bottom couplers]. Saiidi et al. (2009) No. of columns: two Scale factor: 20% Section: circular Diameter: 10 in. (254 mm) Longitudinal bars: 8 No. 4 (Ø13 mm) SMA Transverse bars: 0.17 in. (Ø4.4 mm) spirals at 1.5 in. (38 mm) 4.0db: used at two levels Both columns exhibited drift capacity (10% or 14% drift ratio) equal to or better than that of the reference column with no bar fracture up to 14% drift ratio [bottom couplers were installed 4 in. (102 mm) below the footing surface; top couplers were installed 1.4Dc above the bottom couplers]. Varela and Saiidi (2014) No. of columns: one Scale factor: 25% Section: circular Diameter: 14 in. (356 mm) Longitudinal bars: 12 Ø0.45 in. (Ø11 mm) SMA Transverse bars: 0.25 in. (Ø6.3 mm) spirals at 1.5 in. (38 mm) 3.4db: used at two levels The column exhibited 11.8% drift ratio [bottom couplers were installed 2 in. (51 mm) below the footing surface on center; top couplers were installed 0.85Dc above the bottom couplers on center]. Note: db = longitudinal bar diameter; Dc = either the column diameter or the dimension of the column’s largest side. Table 2-9. Summary of seismic performance of column test models with threaded couplers.

State-of-the-Art Literature Review 33 and that column performance with tapered TCs is yet to be studied. Furthermore, the SMA bars were machined in a dog-bone shape in these studies to shift the yielding outside the coupler region. 2.2.7 Swaged Couplers 2.2.7.1 Performance of Swaged Couplers Noureddine (1996) tested four No. 18 (Ø57 mm) ASTM A615 and A706 Grade 60 reinforcing steel bars that were connected by swaged couplers. The specimens were tested under monotonic tensile loading to failure. The average ultimate load for the splice was comparable to that of the control bars. The average ultimate strain was 8% for A615 bar specimens and 9% for A706 bar specimens. The failure mode was either bar fracture away from the splice or bar pullout from the sleeve (Figure 2-32). The study pointed out that this splice is effective in that it provides a satisfactory energy dissipation mechanism through friction without significant degradation of load capacity. However, the performance of swaged couplers under cyclic loading is unknown. Yang et al. (2014) tested three tensile specimens consisting of No. 8 (Ø25 mm) ASTM A615 Grade 60 steel bars connected by swaged couplers. The average yield and the ultimate strengths for these specimens were 65 ksi (448 Mpa) and 98 ksi (675 Mpa), respectively, which were com- parable to those measured for the reference bars. The average maximum strain exceeded 8%. All specimens failed as a result of bar rupture away from the coupler region (Figure 2-33). A summary of the available test data on the performance of swaged couplers is presented in Table 2-10. 2.2.7.2 Performance of Columns with Swaged Couplers Yang et al. (2014, 2015) used swaged couplers in the plastic hinge of a severely damaged bridge column to replace buckled bars with new reinforcement (Figure 2-34, a and b). FRP was utilized on the entire column height to increase confinement as well as shear capacity of the column. The repaired column was tested under reversed cyclic loading to failure. Figure 2-34c shows the mea- sured repaired and original force-displacement relationships. The repaired column exhibited a Source: Noureddine (1996). Figure 2-32. Swaged sleeve coupler specimen tests. Source: Yang et al. (2014). Figure 2-33. Swaged sleeve coupler specimen tests.

34 Proposed AASHTO Seismic Specifications for ABC Column Connections displacement ductility capacity of 2.4, which was 59% lower than that of the original column. The low displacement ductility of the repaired column was because of early termination of the test owing to rupture of the FRP jacket under torsional loading. It was reported that the column ductility might have been higher, as there was no bar fracture. The lateral strength of the repaired column was 30% higher than that of the original column. Table 2-11 presents the summary of the findings on columns with swaged couplers. Given the lack of extensive test data at the time of the writing of this report, it would be difficult to conclusively evaluate the seismic performance of these couplers when used in columns. 2.2.8 Mechanically Spliced Column Field Application Of the various coupler types, grouted sleeve couplers are the most common type in the bridge industry. GCs were used in three states—Utah, Florida, and Colorado—to connect precast columns to footings (Figure 2-35). The Utah and Florida DOTs developed design manuals for Study Bar Size Bar Type Mode of Failure Remarks Noureddine (1996) No. 18 (Ø57 mm) ASTM A615, Grade 60 ASTM A706, Grade 60 Bar pullout, bar fracture Ultimate loads comparable to control bars. Large strains (more than 8%) were observed in the tests. Yang et al. (2014) No. 8 (Ø25 mm) ASTM A615, Grade 60 Bar fracture Average maximum strain greater than 8%. Table 2-10. Summary of studies on swaged couplers. Source: Yang et al. (2014, 2015). Note: Calt-2 = original column; R-Calt-2 = repaired column. (a) Original column after testing (b) Swaged couplers at base (c) Force-displacement hysteresis Figure 2-34. Repaired column with swaged couplers.

State-of-the-Art Literature Review 35 Reference Column Geometry Coupler Length Remarks Yang et al. (2014) No. of columns: one Scale factor: 50% Section: interlocking Diameter: 24 in. (610 mm) by 36 in. (914 mm), Longitudinal bars: 20 No. 8 (Ø25 mm) Transverse bars: No. 4 (Ø13 mm) spirals at 2.75 in. (70 mm) 7db: used at two levels Repaired column showed 59% lower displacement ductility and 30% higher base shear capacity as compared with the original test model [bottom couplers were installed immediately above the footing; top couplers were 36 in. (1.0Dc) away from the bottom couplers on center]. Note: db = longitudinal bar diameter; Dc = either the column diameter or the dimension of the column’s largest side. Table 2-11. Summary of seismic performance of column test models with swaged couplers. Source: (a) Culmo (2009); (b) Kapur et al. (2013). (a) Grouted coupler, Florida (b) Grouted coupler, Utah (c) Grouted coupler, Colorado (d) SR-99 off-ramp bridge, Washington State Figure 2-35. Mechanical bar coupler field application.

36 Proposed AASHTO Seismic Specifications for ABC Column Connections precast columns incorporating GCs (Utah DOT 2010; Florida DOT 2019). An SMA bridge was constructed in Seattle, Washington, in which HCs were used to connect SMA bars to steel bars in the plastic hinge regions (Figure 2-35d). 2.3 Grouted Duct Connections 2.3.1 Introduction Grouted duct (GD) connections are constructed by inserting precast or CIP column lon- gitudinal reinforcing bars into ducts installed in the adjoining member; then the ducts are filled with grout (Figure 2-36). In this connection type, a separate duct is provided for each column longitudinal bar. Grouted ducts were initially developed to connect precast cap beams to columns to simplify construction for overwater bridges or to tall columns. However, recent studies explored their applications for pile-to-pile cap and footing or pile cap to column connections. 2.3.2 Previous Studies Eight pullout tests were carried out at the University of Texas, Austin, to determine the bond strength of grout-filled duct connections (Matsumoto et al. 2001). Epoxy-coated No.11 (Ø36 mm) bars were embedded in corrugated steel ducts 4 in. (102 mm) in diameter that were filled with standard grout. Two of those bars were headed. The effect on the bond behavior of the embed- ment length (8.5db, 12.8db, and 17db, where db is the bar diameter), grout brand, and bar anchorage (straight or headed bars) was investigated. Figure 2-37 shows photographs of the connection. The specimens with an embedment length of 12 in. (8.5db) exhibited pullout failure charac- terized by the development of splitting cracks in the surrounding concrete and pullout of the bar–grout mass from the duct. In other specimens, the maximum tensile force was restricted by either the test setup limitation or the bar fracture. The following simple empirical design equation was proposed: ( ) 41.7 (2-1)l d f f d b y c = ′ where ld = embedment length, db = bar diameter (in.), fy = specified yield stress (psi) of the bar, and f c′ = specified compressive strength (psi) of concrete. It can be shown that the average bond strength of this series of GD connection tests is 2.5 times the conventional strength of a concrete bond, mainly due to the confinement Extended Column Reinforcing Steel Precast Column Cap Beam Grouted Duct Column Section A-A Column Section B-B A A B B Figure 2-36. Grouted duct connections.

State-of-the-Art Literature Review 37 provided by the duct. Subsequently, Matsumoto et al. (2001) tested a full-scale precast cap beam connected to a column incorporating a grout-filled duct connection. The column was reinforced longitudinally with 12 epoxy-coated No. 9 (Ø29 mm) bars and transversely with No. 3 (Ø10 mm) spirals spaced at 4 in. (102-mm), which resulted in longitudinal and trans- verse steel ratios of 1.7% and 0.46%, respectively. However, only four bars were extended into the cap ducts. The column diameter and the clear height were 30 in. (762 mm) and 24 in. (610 mm), respectively. The embedment length of bars in the ducts was 15 in. (381 mm). Two vertical rams and one horizontal ram were used to obtain load deflection of connection at service and failure levels under different moment demands. The test results showed that the grout-filled duct connection exhibited a similar load deflection relationship to the CIP analyti- cal model with the expected strength, ductility, and bar anchorage. The connection exhibited only minor damage (Figure 2-38). Another 32 pullout tests were carried out by Brenes et al. (2006) at the University of Texas, Austin. The test specimens were similar to the specimens in the aforementioned pullout study. The effects of bar embedment length, duct material, number of ducts, bar coating, and bar eccentricity in the ducts were investigated in this study. Three types of ducts 4 in. (102 mm) in diameter were used: corrugated galvanized strip metal duct, corrugated high-density poly- ethylene duct, and corrugated polypropylene duct. Three embedment lengths were used: 8db, 12db, and 16db, where db is the bar diameter. Grade 60 uncoated and epoxy-coated deformed No. 11 (Ø36 mm) bars conforming to ASTM A615 were used. Normal strength grout with a compressive strength of 5,800 psi (40 Mpa) in 28 days was used to fill the ducts. All of the 32 specimens failed due to bar pullout. The test results showed that the initial stiffness of bond-slip curves and the ultimate bond strength of galvanized steel (GS) duct were greater than those of the plastic ducts (polyethylene and polypropylene) with the same embed- ment length. Bond-slip relationships for different ducts with 8db embedment length are shown in Figure 2-39. The embedment length had a minor effect on the initial stiffness of the con- nection. Substantial reduction of the bond strength was observed in multiple duct tests, while the initial stiffness exhibited minor variations. For example, a 25% reduction of bond strength was observed in the double GS duct system as compared with the single GS duct connection. The test results also confirmed that the duct spacing in multiple duct connections had a minor effect on the bond-slip relationship. The clear duct distances were 1dd and 2dd, where dd is the (a) Side view (b) Top view Source: Matsumoto et al. (2001). Figure 2-37. Grout-filled duct connection in cap beam before casting.

38 Proposed AASHTO Seismic Specifications for ABC Column Connections Source: Matsumoto et al. (2001). Figure 2-38. Damage to grouted duct column–cap beam connection at design-level load. Source: Brenes et al. (2006). Note: GS = galvanized steel; PE = polyethylene; PP = polypropylene. Figure 2-39. Effect of duct material in grouted duct connection bond behavior. duct diameter. Bar eccentricity had a minor effect on the initial stiffness but reduced the bond strength by 17%. The effect of the other test variables on the bond strength of the grout-filled duct system was negligible. The test result of the GS duct connections showed that in nearly all the tests, the grout failed before duct seams opened 3 in. (76 mm) or more below the surface. An example of this type of failure is shown in Figure 2-40. The lack of sufficient tensile strength of grout can be the cause of the failure. In the plastic ducts, other types of failure, such as slippage of grout relative to the duct, bar pullout with partial grout pullout, and bar pullout with complete grout pullout, were also observed.

State-of-the-Art Literature Review 39 The authors proposed a design equation for the embedment length (ld) of the bar in the grout- filled duct connection, as follows: max 8 ,12 in., 180 , 45 (2-2) ,cr l d f d f f d f d b y b c s b c = β γ ′ β γ ′     where db = bar diameter (in.), fy = specified yield stress (psi) of the bar, fs,cr = calculated tensile stress (psi) in the bar corresponding to the critical load combination, f c′ = specified compressive strength (psi) of concrete, β = modification factor for duct material taken as 1.0 for GS duct and 1.3 for plastic duct, and γ = modification factor to account for group effects, calculated on the basis of the number of ducts subjected to simultaneous tension under the design load combination. Raynor et al. (2002) investigated the bond behavior of reinforcing bars grouted in ducts sub- jected to cyclic loading. To determine the local bond-slip relationship, 13 specimens were con- structed with No. 6, 8, and 10 bars (Ø19, 25, and 32 mm, respectively), which were embedded 2 in. (51 mm) deep into steel post-tensioning ducts, each with a 3-in. (78-mm) diameter and a wall thickness of 0.019 in. (0.48 mm). Depending on the bar size, three, four, or five bar ribs were embedded in the grout. The bars were deboned intentionally to a length of 8 in. (200 mm) from the test specimen surface to eliminate the likelihood of conical pullout failure. The test bars were grouted into a precast RC block. The test results showed that the bond strength provided by GD connections was higher than that of conventional concrete. It was also found that slippage of the bar from the grout was due to compressive failure of concrete against the bar ribs, which (a) Side view (b) Top view Source: Brenes et al. (2006). Figure 2-40. Damage of grouted galvanized steel duct connections.

40 Proposed AASHTO Seismic Specifications for ABC Column Connections is in contrast to the radial bond cracks observed in pullout test of bars in concrete with no ducts. The lack of longitudinal and radial cracks showed that the duct provided sufficient confinement around the bar to prevent splitting of the grout. Ou (2007) tested 28 GD pullout specimens under monotonic and cyclic loading in three phases. No. 5 (Ø16 mm) and No. 8 (Ø25 mm) bars were anchored in corrugated steel ducts 3.2 in. (81 mm) in diameter. Phase I, which included six samples, was carried out under mono- tonic loading to investigate the bond strength of bars with three types of duct filler. The bar embedment length for all samples in Phase I was 24db, which was calculated on the basis of the design equation proposed by Matsumoto et al. (2001). The maximum strain in the bars reached the strain hardening region of the stress–strain relationship in all tests, but the tests were stopped before bar fracture because of stroke limitation. The Type 2 grout with a compressive strength of 7,250 psi (1.05 MPa) was selected for further studies because of its better workability. Sub- sequently, 14 samples were tested in Phase II under monotonic loading to investigate different embedment lengths (4db, 16db, and 24db.). Bars pulled out from the samples with 4db. Strain hardening was observed for samples with higher bar embedment lengths, but the tests were stopped because of stoke limitation. Eight samples were tested in Phase III under cyclic loading. Bars with 4db embedment length pulled out from the GDs, but the bond strength was higher than that measured in the monotonic tests. This behavior is not consistent with other bond studies, which have shown that cyclic loading adversely affects the bond strength. The bars fractured in samples with higher embedment lengths. The study concluded that an embedment length of 24db is sufficient to fully anchor a bar in GD. Ou (2007) tested five hybrid rocking segmental columns under cyclic and pseudodynamic loading. The column base segment reinforcement was connected to the footing by using GD connections with an embedment length of 24db. The GD connections showed satisfactory performance in all the tests, since bars fractured in the plastic hinge of the columns at 5% and 6% drift ratios. Steuck et al. (2009) extended Raynor’s findings to larger bars. The experimental program consisted of 17 monotonic pullout tests on bars ranging from No. 8 (Ø25 mm) to No. 18 (Ø57 mm) with embedment lengths of 2db to 14db into GDs. For the tests on No. 10, 14, and 18 bars, corrugated steel pipes 8 in. (203 mm) in diameter were used. No. 8 bars were grouted into 4-in. (107-mm)-diameter post-tensioning ducts. Both fiber-reinforced and conventional grouts were utilized as duct filler. Fourteen of the test specimens failed as a result of bar pullout from the grout accompanied by near-surface conical failure of the grout (Figure 2-41a). In the remaining specimens, a cylinder consisting of the bar surrounded by grout pulled out from the duct. Two tests resulted in concrete splitting and one bar fracture (Figure 2-41b). The test results showed that pullout resistance in tests with fiber-reinforced grout was typi- cally lower than that of comparable tests with conventional grout. On the basis of the test results of the specimens with short embedment lengths, a constitutive bond-slip relationship was developed. Steuck et al. (2009) reported that embedment lengths of 6db and 14db were sufficient to develop the yield and the ultimate capacities of No. 18 bars. Furthermore, they presented an empirical design equation: 130. 2 (2-3)l f d f d d d y b g d b( )= ′ + − where ld = embedment length, fy = bar yield stress (psi),

State-of-the-Art Literature Review 41 db = bar diameter (in.), f g′ = grout compressive strength (psi), and dd = duct/pipe diameter (in.). The second term in the equation presents the length of the cone in the conical pullout failure mode. If partial cone failure is prevented by thick pipe or sleeve, this term should be taken as zero. Pang et al. (2008) performed cyclic tests of three 0.4-scale precast cantilever columns con- nected to precast bent cap segments with GD connections. The test results were compared with those of a typical CIP reference column (DB5-RE) that had approximately the same geometry and reinforcement details. The precast columns consisted of six No. 8 (Ø25 mm) longitudinal bars, which resulted in a longitudinal steel ratio of 1.51%. Two of the three precast models employed longitudinal bars that were debonded over a length of 8db into the cap beam with tight- fit (LB8-D1) and loose-fit (LB8-D2) polyvinyl chloride sleeves. The longitudinal bars in the other model (LB8-FB) were fully grouted into the ducts. The bars were anchored in post- tensioned metal ducts 4 in. (102 mm) in diameter to a length of 17.5db in the cap beam and further anchored to a length of 18.5db into a concrete diaphragm (36db in total). For all the pre- cast models, 12 No. 3 (Ø76 mm) longitudinal bars that stopped at the interface were provided to meet the code spacing requirement. A grout pad 0.5 in. (13 mm) thick was also cast at the beam– column interface to simulate the field erection of the precast pieces. A plastic hinge was formed outside the bent cap in all test models (Figure 2-42). The precast models exhibited concentrated deformations at the column–cap interface as compared with the more distributed deforma- tions observed within the reference CIP model, owing to intentional debonding of longitudinal bars within the cap beam and the presence of supplementary bars intentionally stopped at the interface. The force-displacement response of the test models was approximately the same for all columns (Figure 2-42), which indicates that the drift ratio capacity, ductility capacity, lateral load capacity, and energy dissipation capacity of the precast columns were comparable to those of the CIP column. Debonding of the bars over a length of 8db reduced the strain concentration (a) Conical failure of grout (b) Bar fracture Source: Steuck et al. (2009). Figure 2-41. Damage of grouted steel pipe connections.

42 Proposed AASHTO Seismic Specifications for ABC Column Connections at low drift levels but did not delay the bar fracture as intended. It also had a minor effect on the overall hysteretic performance of the column. The longitudinal bars buckled and then fractured during nearly the same cycles and at the same drift ratios in all specimens. Despite the intentional debonding of the bars in Model LB8-D2, the buckling occurred in the bonded region in the column, not in the polyvinyl chloride sleeve. Pang et al. (2008) expanded their investigation by conducting three pullout tests on bars replicating the anchorage in precast connections. The specimens were labeled AD8-FB, AD8-D1, and AD8-D2, corresponding to the LB8-FB, LB8-D1, and LB8-D2 precast con- nections, respectively. The measured strain data demonstrated that the debonding methods in Specimens LB8-D1 and LB8-D2 were effective, with no significant difference. The initial (a) Plastic hinge damage of cast-in-place column (DB5-RE) (b) Hysteretic response of cast-in-place column (DB5-RE) (c) Plastic hinge damage of tight-fit debonded bar column (LB8-D1) (d) Hysteretic response of tight-fit debonded bar column (LB8-D1) Source: Pang et al. (2008). Figure 2-42. Columns with cap beam grouted duct connections.

State-of-the-Art Literature Review 43 stiffness of Specimen AD8-D2 was lower than that of the other two due to the flexibility added to the connection because of the debonding of the bars. The yield plateau and the strain hard- ening region of Specimen AD8-FB were shorter than those measured in the other two speci- mens because the bar was fully bonded in the ducts (Figure 2-43a). Contrary to Specimens LB8-D1 and LB8-D2, considerable displacement of the concrete and grout surface, metal duct, and bar were measured in Specimen AD8-FB. This difference was attributed to the fact that debonding of the bar within 8db reduced the demand on the surface concrete and grout. Figure 2-43b shows the strain profiles of the bars during 20- and 60-ksi (137.9- and 413.7-Mpa) bar stress. The average calculated bond strength for Specimen AD8-FB was lower than that of the other two specimens. Restrepo et al. (2011) tested a 0.42-scale GD column–cap beam connection under quasi- static cyclic loading and compared the results with those of a CIP reference model. Sixteen No. 5 longitudinal bars were embedded to a length of 35.2db into 1.75-in. (44.5-mm)-diameter corrugated steel ducts. The longitudinal bars were fully grouted into the ducts. To provide construction tolerance, a grout pad 1.5 in. (38 mm) thick was cast between the bent cap soffit and column. The damage to the joint region at the end of testing is depicted in Figure 2-44a. Relatively minor cracking occurred in the joint region, but significant flexural and shear cracks and spalling were observed in the columns. Although splitting cracks developed between the ducts, there was no sign of splitting failure of the grout within the ducts, pullout failure, or significant bar or duct slip. The column bars’ anchorage within the ducts was satisfactory. The force-displacement envelopes for the GD connection and CIP column are compared in Figure 2-44b. The GD connection exhibited a drift ratio capacity of 5.5%, corresponding to a displacement ductility of 8. The force-displacement response indicated stable hysteretic behavior without appreciable degradation of strength. The strength, stiffness, and ductility for the GD connection were approximately the same as those for the CIP column up to a 3.7% drift ratio. As compared with the CIP column, the GD connection achieved a higher lateral load capacity and slightly lower drift ratio capacity (5.5% versus 5.9%). A new detail for building columns was proposed by Belleri and Riva (2012), as discussed in the section on GCs. In this detail, corrugated steel ducts are placed in the plastic hinge region of Source: Pang et al. (2008). (a) Force-displacement response (b) Strain profiles Figure 2-43. Response of pullout tests.

44 Proposed AASHTO Seismic Specifications for ABC Column Connections the column, footing starter bars are anchored in the ducts, and the ducts are grouted after the column installation. Even though seismic behavior similar to that of the reference specimens was reported in the study, the application of this detail would be impractical for bridge columns, owing to the size of the ducts and the large number of reinforcing bars that are typically used in bridge columns. Tazarv and Saiidi (2014) conducted 14 pullout tests to determine the bond strength of UHPC-filled duct connections. The test variables were the embedment length, bar size, duct diameter, number of ducts, and bar bundling. Straight No. 8 and 11 bars (Ø25 and Ø36 mm, respectively) were used with embedment lengths ranging from 3db to 12db. Ducts with a nomi- nal diameter of 3, 4, and 5 in. (76, 102, and 127 mm, respectively) were used. Bundled bars consisting of two No. 8 (Ø25 mm) bars were used in three of the specimens. The effect of using double ducts spaced at a relatively small clear distance of 3 in. (76 mm), each with a single bar, was investigated in two specimens. Duct pullout was observed in two tests and bar pullout only in a specimen with 3db embedment length; bar fracture occurred in the remaining 11 specimens (Figure 2-45). The effect of bar bundling, bar size, and multiple ducts on the bond performance was found to be negligible, but the duct size had significant effects on the bond strength. The test results illustrated that the bond strength of UHPC was eight times that of the conventional concrete and that the required embedment length of reinforcing bars to fracture the bar in the UHPC-filled duct connections was at least 50% shorter than that of any other GD connections as well as conventional anchoring mechanisms. “Duct bond strength,” the ratio of the pull force capacity to the surface area of the duct, was introduced and investigated in addition to the “bar bond strength” to include all modes of failure in the connection design. The following design equation was proposed: max 27 , 120 (2-4) 2 UHPC l d f d f d f f d b s d c b s= ′ ′     where ld = development length (in.) for unhooked deformed bar in UHPC-filled duct connections, db = nominal diameter (in.) of bar, dd = inner diameter (in.) of duct, Source: Restrepo et al. (2011). (a) Plastic hinge damage (b) Force-displacement envelope Figure 2-44. Columns with cam beam grouted duct connections.

State-of-the-Art Literature Review 45 fs = bar stress (psi) (1.5fy or fu, whichever is greater, using expected properties, where fy = yield strength of anchoring bar and fu = ultimate strength of anchoring bar), f c′ = compressive strength (psi) of concrete surrounding the duct, and f ′UHPC = compressive strength (psi) of UHPC in the duct. Tazarv and Saiidi (2015d) also tested two half-scale precast column models incorporat- ing UHPC-filled duct connections at the column base under cycling loading. The test models were labeled PNC and HCS. These columns were initially hollow but were filled with self- consolidating concrete after the column installation. The test results were compared with those of a reference CIP column. In Model PNC, 11 No. 8 (Ø25 mm) longitudinal bars protruded out of a precast column base and were embedded 28db into corrugated steel ducts 3 in (76 mm) in diameter. The longitudinal bars were debonded 4db each above and below the column–footing interface to prevent premature failure of bars from strain concentration (Figure 2-46). Model PNC showed similar apparent damage as that of the CIP column with no damage to the UHPC-filled duct connection, which indicated satisfactory performance of this type of con- nections. The failure mode was buckling and fracture of the longitudinal bars with no evidence of bar and duct pullout. The force-drift ratio hysteresis response of Model PNC is compared with that of the CIP column in Figure 2-46c. Model PNC behaved approximately the same as the conventional model in terms of force-displacement hysteretic behavior, lateral load capac- ity, and degradation of strength and stiffness. However, energy dissipation was higher than in the CIP column owing to debonding of the bars. Model PNC reached a displacement ductility of 6.30, which was lower than the CIP column’s ductility capacity of 7.36, but the difference was due to the lower concrete compressive strength in Model PNC. In Model HCS, 10 No. 11 (Ø36 mm) longitudinal headed reinforcing steel bars protruded out of the column base and were embedded to a length of 20db into corrugated steel ducts 4 in. (76 mm) in diameter that were filled with UHPC. To enhance the seismic performance of the column, a combination of SMA bars and ECC was used in the plastic hinge region of the column. The SMA bars were linked to the footing steel bars and precast column steel bars with HCs. The longitudinal bars were debonded 4 in. (101.6 mm) below the column–footing interface to (a) Duct pullout Source: Tazarv and Saiidi (2014). (b) Cut-in-half view of bar pullout specimen Figure 2-45. Results of pullout tests of duct connections filled with ultrahigh-performance concrete.

46 Proposed AASHTO Seismic Specifications for ABC Column Connections (a) Connection detail (b) Column base damage after 12% drift (c) Model PNC force-drift hysteresis (d) Model HCS force-drift hysteresis Source: Tazarv and Saiidi (2015d). Extended Column Reinforcing Bars Longitudinal Reinforcement Precast Column D eb on de d Le ng th Footing D d L 4 d 4 d d b b emd Transverse Reinforcement Corrugated Galvanized Duct w/ UHPC Ba se S he ar (k N ) -356 -256 -156 -56 44 144 244 344 -80 -60 -40 -20 0 20 40 60 80 -12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12 Ba se S he ar (k ip s) Drift (%) -378 -278 -178 -78 22 122 222 322 -85 -65 -45 -25 -5 15 35 55 75 -12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12 Ba se S he ar (k N ) Ba se S he ar (k ip s) Drift (%) PNC Column CIP Column HCS Column CIP Column Figure 2-46. Columns with duct connections filled with ultrahigh-performance concrete. minimize the effects of strain concentration. Figure 2-46b demonstrates the damage pattern to the column plastic hinge region at a 10% drift ratio. The damage to Model HCS was the least as compared with that in the other two models and was limited to spalling of the cover ECC. The force-drift ratio hysteresis of HCS is compared with that of CIP in Figure 2-46d. HCS showed a lateral load capacity similar to that of the CIP column in each loading cycle. Debonding of the longitudinal bars was found to be a successful technique for spreading bar yielding and avoiding premature failure of bars from strain concentration either in the UHPC or under the coupler regions. No damage in the UHPC-filled duct connections was observed, even under 12% drift ratio cycles. The study concluded that UHPC-filled duct connections can

State-of-the-Art Literature Review 47 be considered as an alternative to CIP connections in high-seismic regions for connecting pre- cast columns to shallow cap beams and footings. However, when the footing or cap beam depth is sufficient, high-strength grouts may be used in lieu of UHPC. Mashal et al. (2014) tested one half-scale, two-column bent under cyclic loading. Grouted ducts were used in the column-to-cap beam connections (Figure 2-47). Pocket connections were used to link the columns to the footings. Concrete shear keys were incorporated at the column–cap beam interface to enhance the shear load transfer. The test results showed that the full plastic moment was developed at the column top and was maintained at up to a 3.4% drift ratio (equivalent to a displacement ductility capacity of 4.2), at which the test was stopped. Galvis et al. (2015) tested 18 pullout specimens in which two or three No. 8 (Ø25 mm) bars were bundled and then anchored in grout filled GS ducts 4 in. (100 mm) in diameter. The Source: Mashal et al. (2014). (a) Test setup (b) Column–cap beam connection (c) Bent force-displacement hysteresis Figure 2-47. Two-column bent with cap beam grouted duct connections.

48 Proposed AASHTO Seismic Specifications for ABC Column Connections development length for the bars ranged from 6 in. (150 mm) to 36 in. (900 mm). These embed- ment lengths are respectively equal to from 3.5db to 20.4db when an equivalent diameter for the three bundled bars (db = 1.73 in. or 44 mm) is used. The bars fractured in specimens with an embedment length of 16.8db or greater. However, the shorter embedment lengths were not sufficient because the bars pulled out from the ducts or ducts pulled out from the surrounding concrete (Figure 2-48). The following design equation was proposed: 60 (2-5)l f f dd b y g L b= ψ + ψ    where ld = development length; yb = parameter to account for effect of the number of bars on force; 1.0 for an individual or two bundled bars and 1.2 for three bundled bars; fy = yield strength (psi) of the bar; fg = grout compressive strength (psi); db = single bar diameter (in.); and yL = parameter to account for effect of the number of bars on anchorage length; 5.0 for an individual bar, 7.0 for two bundled bars, and 10.0 for three bundled bars. Thonstad et al. (2016) tested a quarter-scale, two-span, two-column-bent bridge on shake tables (Figure 2-49). The precast cap beams were connected to the hybrid rocking columns by GDs and pocket connections. These columns were hybrid, since both reinforcing steel bars and pre-tensioning tendons were incorporated in all column connections. Two sizes of reinforcing steel bars were used as the column longitudinal reinforcement: six No. 3 (Ø10 mm) and six No. 4 (Ø13 mm) bars. Four unbonded post-tensioning strands 3/8 in. (9.5 mm) in diameter were utilized in the columns. Steel jackets were incorporated at the ends of columns to reduce the damage during rocking. The No. 3 bars were extended 50db into the cap beam ducts [with a diameter of 1.25 in. (32 mm)], and then the ducts were filled with a grout. A portion of the embedded bars was debonded, which resulted in an approximately 34db effective embedment length into the cap beams. The No. 4 longitudinal bars were terminated at the column-to- cap beam interface. The bridge was tested under several motions. It was reported that many Source: Galvis et al. (2015). (a) Bar pullout (b) Duct pullout Figure 2-48. Grout-filled duct connection pullout tests.

State-of-the-Art Literature Review 49 column longitudinal bars fractured at high-amplitude earthquake motions, which indicated that the GD connections were able to anchor the bars fully. The peak drift ratio capacity of the bridge exceeded 10%, and the residual drifts were insignificant because of prestressing of the columns. 2.3.3 Summary of Grouted Duct Connection Tests Summaries of the findings from experimental studies of GD connection pullout tests and the seismic performance of columns with GD connections are presented in Tables 2-12 and 2-13, respectively. 2.3.4 Grouted Duct Column Field Application GD connections have been implemented in several bridges in Texas, South Carolina, Washington, and California. A few examples are shown in Figure 2-50. 2.4 Pocket and Socket Connections 2.4.1 Introduction In socket connections for columns, an opening is left in the adjoining member and a par- tially cast column with protruding longitudinal bars (Figure 2-51a) is inserted into the open- ing. Similarly, in what is known as “pocket connections,” an opening is left in the adjoining member of the column, but a fully cast column (Figure 2-51, b–d) is inserted into the opening. Grout is placed in the gap in the pocket connections, but concrete is cast in the joint in socket connections. Even though previous studies have made a distinction between the abovementioned con- nections (pocket and socket), these connections are essentially the same and can generally be categorized as “pocket/socket connections.” The term “pocket” is used in this section to refer to both types of connections. The column and its adjoining member can be either precast or Source: Thonstad et al. (2016). (a) Cap beam grouted duct (b) Test setup (c) Cap beam damage at design level earthquake Figure 2-49. Two-column bent, two-span bridge with cap beam grouted duct connections.

50 Proposed AASHTO Seismic Specifications for ABC Column Connections Reference Test Variables Design Equation Remarks Matsumoto et al. (2001) No. of tests: 8 Variables: embedment length, grout brand, and headed or straight bars 41.7 The grout strength was not included in the equation. The bar size and the duct diameter were the same in all specimens. Brenes et al. (2006) No. of tests: 32 Variables: embedment length, duct material, number of ducts, bar coating, and bar eccentricity β The grout strength was not included in the equation. The bar size and the duct diameter were the same in all specimens. Raynor et al. (2002) No. of tests: 13 Variables: embedment length and bar size An analytical equation was proposed The proposed equation was developed on the basis of a parametric study. Ou (2007) No. of tests: 28 Variables: embedment length, bar size, grout type 24db was recommended Monotonic and cyclic tests showed that 24db is sufficient to anchor bars. The cyclic bond strength was greater than that for monotonic loading, which is not consistent with findings from other bond studies. Steuck et al. (2009) No. of tests: 17 Variables: embedment length, bar size, duct type (posttensioned duct or steel pipe), and fiber-reinforced grout 130. 2 The strength of concrete surrounding the ducts was not included in the equation. No difference was recognized between the steel pipe and duct. Pang et al. (2008) No. of tests: 3 Variables: debonding effect ehtnotcefferonimadahgnidnobeDAN local bond behavior. Tazarv and Saiidi (2014) No. of tests: 14 Variables: embedment length, bar size, duct diameter, bar bundling, and multiple ducts This equation is only valid for UHPC as the duct filler. This was the only study that included the duct diameter as well as the strength of the concrete and grout. Galvis et al. (2015) No. of tests: 18 Variables: embedment length, and bar bundling The strength of the concrete surrounding the ducts was not included in the equation. The duct diameter was the same in all specimens Note: NA = not available. Table 2-12. Summary of grouted duct connection pullout tests. CIP. The embedment length of the column or the column reinforcement in the opening is a key parameter that controls the development of the full plastic moment of the column. A state- of-the-art review of pocket connections is presented in this section. 2.4.2 Previous Studies Pullout tests on single-line and double-line grouted pocket connections (Figure 2-52) were performed by Matsumoto et al. (2001). The single-line pocket connections had only one pocket, while the double-line specimens had two pockets. The following variables were investigated: • Bar anchorage (straight or headed); • Bar size (No. 6 [Ø19 mm], No. 8 [Ø25 mm], and No. 11 [Ø36 mm]); • Embedment length (5 to 18 times the bar diameter); • Number of bars per pocket (single and double bars); and • Grout type.

State-of-the-Art Literature Review 51 Reference Column Geometry Duct and Filler Remarks Matsumoto et al. (2001)a No. of columns: one Scale factor: 100% Section: circular Diameter: 30 in. (762 mm) Longitudinal bars: 12 No. 9 (Ø29 mm) Transverse bars: No. 3 (Ø10 mm) spirals at 4 in. (102 mm) Type: Galvanized PT Diameter: 4 in. (102 mm) Column–cap beam Filler: Grout Embedment length: 15 in. (13.3db) The grout-filled duct connection exhibited a similar load–deflection relationship to the CIP analytical model, with the expected strength, ductility, and bar anchorage. Ou (2007) No. of columns: five Scale factor: NA Section: hollow square Side: 34.4 in. (860 mm) Longitudinal bars: 12 No. 5 (Ø16 mm) or 12 No. 8 (Ø25 mm) Transverse bars: No. 3 (Ø10 mm) ties at 1.5 in. (38 mm) dewohs snoitcennoc DG satisfactory performance, as the hybrid rocking segmental column longitudinal bars fractured in the base segment under a 5% drift ratio or more. Pang et al. (2008) No. of columns: three Scale factor: 40% Section: circular Diameter: 20 in. (508 mm) Longitudinal bars: six No. 8 (Ø25 mm) Transverse bars: 0.24-diameter wire at 2.8 in. (70 mm) Type: Galvanized PT Diameter: 3.2 in. (80 mm) Column–footing Filler: Grout Embedment Length: 24db The force-displacement response of all three precast columns was the same as that of the reference CIP column. Restrepo et al. (2011) No. of columns: one Scale factor: 42% Section: circular Diameter: 20 in. (508 mm) Longitudinal bars: 16 No. 5 (Ø16 mm) Transverse bars: No. 3 (Ø10 mm) hoops at 1.5 in. (38 mm) Type: Galvanized PT Diameter: 1.75 in. (44 mm) Column–cap beam Filler: Grout Embedment Length: 22 in. (35.2db) The drift capacity of the precast column was 80% of that of the reference CIP column. Mashal et al. (2014) No. of columns: one two-column bent Scale factor: 50% Section: circular Diameter: 20 in. (500 mm) Longitudinal bars: eight No. 5 (Ø16 mm) Transverse bars: No. 3 (Ø10 mm) hoops at 3.0 in. (75 mm) Type: Galvanized PT Diameter: 2 in. (50 mm) Column–cap beam Filler: Grout Embedment length: 20 in. (32db) The connection performed well up to a displacement ductility capacity of 4.2, at which the test was stopped. Tazarv and Saiidi (2015d) No. of columns: two Scale factor: 50% Section: circular Diameter: 24 in. (610 mm) Longitudinal bars: 11 No. 8 (Ø25 mm) and 10 No. 11 (Ø36 mm) Transverse bars: No. 3 (Ø10 mm) spirals at 2 in. (51 mm) Type: Galvanized PT Diameter: 3 in. (76 mm) and 4 in. (102 mm) Column–footing Filler: UHPC Embedment length: 24 in. (24db and 17db) Full plastic moment was developed in both columns and was maintained up to fracture of the longitudinal reinforcement. No damage of connections was observed, even under 12% drift ratio cycles. Thonstad et al. (2016) No. of columns: three two-column bents in a bridge Scale factor: 25% Section: Octagonal Diameter: 14 in. (305 mm) Longitudinal bars: six No. 3 (Ø10 mm) and six No. 4 (Ø13 mm) Transverse bars: wire spirals and steel jackets Type: Galvanized PT Diameter: 1.25 in. (32 mm) Column–cap beam Filler: Grout Embedment length: 13 in. (34db) Many column longitudinal bars fractured, which indicated that the bars were fully anchored in the GDs. The bridge showed a drift ratio capacity of more than 10% with minimal residual displacements. Note: db = longitudinal bar diameter; PT = posttensioned; NA = not available. aThis was not a column. It was an RC stub with four bars and was not subjected to cyclic loads that represent earthquakes. Table 2-13. Summary of seismic performance of column test models with grouted ducts.

52 Proposed AASHTO Seismic Specifications for ABC Column Connections (a) Lake Belton Bridge, Texas (b) Lake Ray Hubbard Bridge, Texas (c) Carolina Bays Parkway, South Carolina (d) SR-520/SR-202 interchange, Washington State (e) US-12 over I-5 bridge, Washington State (f) Boeing North Bridge, Washington State Source: (a and b) Brenes et al. (2006); (c) Culmo (2009); (d and e) Khaleghi et al. (2012); and (f ) Banks et al. (2015). Figure 2-50. Field application of grouted duct connections.

State-of-the-Art Literature Review 53 Bar pullout and concrete breakout failure were observed in pocket specimens for straight and headed bars, respectively. The following design embedment length (Lemd) for straight bars in grouted pocket connections was proposed: 45 (2-6)emdL d f f b y c = ′ where db = bar diameter (in.), fy = specified yield strength (psi) of the bar, and f c′ = specified compressive strength (psi) of the bent cap concrete. (a) Partially cast column (b) Fully cast column (c) Precast column in footing (d) Precast column in pile shaft Partially cast Column Steel Pipe Extended Column Reinforcing BarPrecast Cap Beam Extended Column Fully cast Column Steel Pipe Precast Cap Beam Precast or Cast-in-Place Footing Extended Column Steel Pipe Precast Column Precast Column Pile Shaft Steel Pipe Figure 2-51. Pocket connections. (a) Rebar cage for single-line pocket (b) Rebar cage for double-line pocket Source: Matsumoto et al. (2001). Figure 2-52. Cap beam pocket pullout tests.

54 Proposed AASHTO Seismic Specifications for ABC Column Connections A safety factor of 1.7 was included in this equation to account for the bar overstrength capacity and the concrete strength reduction factor. A column connected to a precast cap beam by a double-line pocket system was tested by Matsumoto et al. (2001) in the next phase of their study (Figure 2-53a). The cap beam dimen- sions were 33 × 30 × 144 in. (0.84 × 0.76 × 3.66 m). The column was reinforced longitudinally with 12 No. 9 (Ø29 mm) bars and transversely with No. 3 (Ø10 mm) spirals spaced at 4 in. (102 mm), which resulted in longitudinal and transverse steel ratios of 1.7% and 0.46%, respec- tively. Only four of the column longitudinal bars were extended into the cap beam pocket. The column diameter and the clear height were 30 in. (762 mm) and 24 in. (610 mm), respectively. The embedment length of the column longitudinal bar into the cap beam was 15 in. [381 mm, or one-half of the column diameter (0.5Dc)]. Two vertical rams and one horizontal ram were used to obtain the load–deflection of connection at service and failure levels under different moment demands. Strain gauges were installed only on the column longitudinal bars, and strain data for bars in the cap beam were not recorded. Minor damage to concrete in the column and the cap beam was reported at the column yielding (Figure 2-53b). Since there was no reference test model, moment-curvature and load–deflection analyses were performed for an analytical model of an assumed CIP model and the results were compared with those of the measured precast test model. Because no reference CIP model was tested, the measured load–deflection and moment-curvature relationships of the column with the pocket connection were compared with the calculated response of the CIP model. The results were in close agreement. Zhu et al. (2006) tested three one-sixth scale concrete-filled fiber-reinforced polymer tubes (CFFT) under quasi-static cyclic loading. The three specimens included a CIP CFFT column with steel starter bars (CIP-CFFT), a precast CFFT column with steel starter bars extending out of the footing and protruding into column GDs (GD-CFFT), and a precast CFFT column with post-tensioned connection (PT-CFFT). All test columns were connected to their footing with pocket connections. The test results were compared with those of a reference CIP column. A photograph of the footing pocket is shown in Figure 2-54a. The embedment length of the CFFT into the footing pocket for all precast columns was 1.0Dc (Dc is the column diameter). The pockets were formed with Sonotubes. The study showed satisfactory performance for all three joints with minimal apparent damage. The PT-CFFT failed because of local fracture of the tube under combined shear and compression. The force-displacement envelopes of the measured hysteresis curves are compared in Figure 2-54b. All CFFT models performed better than the reference model in terms of the initial stiffness, strength, ductility, and energy dissipation capacity. The improved performance was attributed to the confinement provided by the tubes in CFFT models. The CIP-CFFT and GD-CFFT columns achieved a drift ratio of (a) Cap beam pocket detail: plan view (b) Damage at column–cap beam interface Source: Matsumoto et al. (2001). Figure 2-53. Column–cap beam pocket test.

State-of-the-Art Literature Review 55 13%, as compared with 6.7% in CIP. The post-tensioning did not improve the column peak drift capacity, but led to smaller residual displacements. Restrepo et al. (2011) tested two 0.42-scale columns incorporating cap beam pocket connec- tions under slow cyclic loading. These columns were detailed to exhibit high and low ductility and were labeled CPFD and CPLD, respectively. A reference CIP model was also tested. The pocket was constructed in both models by using a corrugated steel pipe 18 in. (457 mm) in diameter. In CPFD, stirrups were provided in the joint region outside the steel corrugated pipe. Additional steel hoops were also provided around the top and bottom of the pipe. In contrast, stirrups or additional hoops were not provided in the joint region of CPLD. Figure 2-55, a and b show the joint regions during fabrication. In both columns, 16 No. 5 (Ø16 mm) longitudinal bars were extended out of the precast columns and were inserted into the steel pipes. The embedment length of the column longitudinal bar in the pocket was 35.2db (or 1.2Dc). The pockets were filled with conventional concrete. A grout pad 1.5 in. (38 mm) thick was cast between the bent cap soffit and the column to provide a construc- tion tolerance. The damage to the joint regions at a displacement ductility of 8 is shown in Figure 2-55c. In CPFD, the plastic hinge was formed in the column out of the pocket con- nection, and the damage to the bent cap was limited to minor cracking. However, in CPLD, the response was characterized by a combination of plastic hinging of the column adjacent to the bent cap and joint shear cracking and distortion. Compared with the CPFD, the joint damage was greater, with larger crack widths and minor joint concrete spalling. Bar slippage in CPFD was slightly higher than that of the CIP column. The bar strain distribution and joint crack pattern also differed between the two models. CPLD showed the highest bar slip of the three models; however, the bar anchorage was still sufficient. The envelopes of the force- displacement hysteresis curves for the three models are compared in Figure 2-55d. The force- displacement response in the CPFD and CPLD models indicated stable hysteretic behavior without appreciable degradation of strength. The strength, stiffness, and ductility for CPFD were approximately the same as those of the CIP column up to a drift ratio of 3.2%. Overall, CPFD achieved a slightly higher lateral load capacity but a lower drift ratio capacity as compared with the CIP column (4.3% versus 5.9%). The lower drift ratio capacity was attributed to a fabrication error. CPLD achieved a drift ratio of 5.1%, corresponding to a displacement ductility (a) Footing pocket (b) Force-displacement envelope Source: Zhu et al. (2006). Figure 2-54. Concrete-filled fiber-reinforced polymer tube column–footing pocket connection tests.

56 Proposed AASHTO Seismic Specifications for ABC Column Connections (c) Damage to Column CPFD plastic hinge (d) Force-displacement envelopes Source: Restrepo et al. (2011). (a) Column CPFD pocket (b) Pocket inside view Figure 2-55. Column–cap beam pocket connection tests.

State-of-the-Art Literature Review 57 of 8, which exceeded the ductility of the CIP column. The test results illustrated close correlation between the CIP column and CPLD in terms of stiffness, strength, and ductility. However, the cap beam longitudinal reinforcement yielded in both precast specimens, which is not acceptable, since cap beams are supposed to be capacity-protected members. Motaref et al. (2011) tested a 0.3-scale precast two-column bent model on a shake table at the University of Nevada, Reno. The bent detail is shown in Figure 2-56a. The bent (desig- nated PEFB) incorporated two precast columns, a precast footing, and a precast cap beam. The columns were connected to the cap beam and footing by pipe-pin connections and pocket connections, respectively. The pockets were formed by using octagonal-shaped openings in the footing (Figure 2-56b). The embedment length of the columns in the pockets was 1.5Dc. One of the columns used ECC in the plastic hinge region and conventional RC elsewhere (RC-ECC). Headed longitudinal bars were used in the RC-ECC column to reduce the develop- ment length. The other column used a GFRP tube filled with concrete. The bent was tested to failure. The plastic hinge was formed in the RC-ECC immediately above the pocket connection (Figure 2-56c). No damage was observed in the footing near the GFRP-wrapped column, and the column failed from rupture of the GFRP tube above the footing (Figure 2-56c). The test results showed that the embedment length was sufficient to develop the column full plastic moment with no damage in the connections. The RC-ECC and GFRP columns achieved a drift ratio capacity of 11%, corresponding to displacement ductility of 7.8 and 5.8, respectively. Two full-scale pile-to-pile cap pocket connections were tested by Ziehl et al. (2011) under slow cyclic loading in a single cantilever configuration. The piles were of 18 in. (457 mm) square concrete sections that were prestressed with strands in a circular configuration. Two different pile caps were used: one to represent an interior connection, and the other to represent an exte- rior connection of a multipile bent. The reinforcement in the exterior connection was the same as that of the interior specimen, with the exception of typical South Carolina DOT end reinforce- ment as well as reduced stirrup spacing at the cantilever end. Each pile cap incorporated a corru- gated steel pipe 36 in. (914 mm) in diameter. Piles were embedded 1.44 times the pile width into the pocket (1.44Dc). The gap between the pockets and piles was filled with a flowable concrete mix. During testing, the interior joint model was subjected to a constant compressive axial load, while the exterior joint model was subjected to variable axial tension and compression from an inclined actuator connected to the interior end of the bent cap. The tests indicated good perfor- mance of both interior and exterior joint models. The force-displacement hysteresis response of the interior pile test model is shown in Figure 2-57a. Both connections sustained several cycles of large, inelastic deformations with only minor cracking within the capacity-protected pile cap segments (Figure 2-57b). The displacement ductility for the models with interior and exterior connection details was 13.5 and 20.0, respectively. It was concluded that when precast piles are embedded at least 1.44 times the pile width into the precast cap, plastic hinges of the pile occur below the cap under the design earthquake, while the pile cap would remain essentially elastic. However, the width of pile caps with pocket connections has to be larger than that of CIP pile caps to accommodate the pockets. Tran (2012) tested two 0.28-scale columns under cyclic loading to study the seismic behavior of column–pile shaft pocket connections. The details of the two models (DS-1 and DS-2) were the same except for the amount of spiral in the column–shaft transition region. The longitudinal reinforcement in both shafts constituted 30 bundled No. 3 (Ø10 mm) bars. The volumetric transverse steel ratio for DS-1 [0.148-in. (3.8-mm)-diameter wire at 3-in. (76.2-mm) pitch] was twice the ratio for DS-2, so as to increase the likelihood of failure in the DS-2 connection. The extended portion of the columns had an octagonal shape with a roughened surface and was embedded inside the shaft pocket with headed bars. The precast columns were embedded into the pile shaft pockets with a length of 1.4Dc. The pile shafts

58 Proposed AASHTO Seismic Specifications for ABC Column Connections (a) Bent detail (b) Footing pocket (c) Damage to column base Source: Motaref et al. (2011). Figure 2-56. Bent with column–footing pocket connections.

State-of-the-Art Literature Review 59 and the pockets were cast at the same time around the embedded portion of the column. The damage of the column–shaft connections is shown in Figure 2-58, a and b. The failure mode of the connection with the higher transverse steel ratio (DS-1) was the plastic hinging of the column. There was also some damage in the pile shaft. In contrast, the other model failed as a result of the prying action of the column in the shaft. It was also concluded that the column longitudinal bars should be headed to ensure hinging in the column. Figure 2-58c compares the force-displacement hysteresis curves of the two columns. Both columns exhib- ited approximately the same lateral load capacity, but DS-1 achieved a higher drift capacity as compared with that of DS-2 (8.2% versus 6.7%). Three 0.42-scale column–spread footing pocket connections were tested by Haraldsson et al. (2013) under quasi-static cyclic loading. The test specimens were as follows: 1. A precast column with a CIP footing pocket connection (SF-1); 2. A column with the same overall geometry as that of SF-1 but with less footing reinforcement, so as to facilitate construction (SF-2); and 3. A pocket connection in a shallow footing (SF-3). The SF-1 and SF2 columns were embedded 1.1Dc into the footing (Figure 2-59). The embed- ment length of SF-3 in the footing was reduced to 0.5Dc. The embedded region of the columns in the footings had an octagonal cross section with a roughened surface. The footings were cast after footing reinforcements were placed around the column. The column longitudinal bars were straight and were terminated with headed anchors. The failure mode for the first two models was buckling and fracture of longitudinal bars with no damage to the connection, whereas SF-3 failed as a result of failure in the connection, buckling of column longitudinal bars, and punching of the footing through the spread footing (Figure 2-59). The measured strains in the SF-1 and SF-2 footing reinforcements were far below the yield strain, which indicates that the connections remained essentially elastic. All three test models showed a drift ratio capacity of 7% or more. The force-displacement hysteretic loops of SF1 and SF2 were stable up to a 6% drift ratio, whereas SF3 exhibited strength degradation at about a 2.5% drift ratio. After the cyclic (a) Force-displacement hysteresis (b) Interior pile-to-cap damage Source: Ziehl et al. (2011). -30 -25 -20 -15 -10 -10 -8 -6 -4 -2 0 2 4 6 8 10 Displacement (in.) L at er al F o rc e (k ip s) -5 0 5 10 15 20 25 Figure 2-57. Pile cap pocket connection.

60 Proposed AASHTO Seismic Specifications for ABC Column Connections (a) DS-1 pile shaft–column damage (b) DS-2 pile shaft–column damage (c) Force-displacement hysteresis Source: Tran (2012). Figure 2-58. Pile shaft pocket connection.

State-of-the-Art Literature Review 61 loading tests, SF-1 and SF-2 were loaded axially to failure. In both columns, the concrete in the plastic hinge failed in compression with no damage to the pocket connection. A quarter-scale four-span bridge model incorporating precast columns with footing pocket connections in one of the bents was tested on shake tables by Kavianipour and Saiidi (2013) (Figure 2-60). This bent consisted of two precast concrete-filled carbon FRP tubes and a pre- cast spread footing. The columns were connected to the cap beam and footing by pipe-pin and pocket connections, respectively. The pockets were formed with octagonal openings in the footing. The columns were embedded to a length of 1.5Dc into the footing pockets and the gap was filled with high-strength grout. A photograph of the footing pocket during construc- tion is shown in Figure 2-60b. The test results showed that damage was limited to spalling of the footing surface concrete around the columns; the FRP tubes remained undamaged (Figure 2-60c). Figure 2-60d shows the cumulative force-displacement hysteretic response of the precast bent in the transverse direction. The two-column bent achieved maximum drift ratios of 6.2% and 7.58% in the transverse and longitudinal directions, respectively, corresponding (a) Specimen SF-1 in footing (b) Specimen SF-1 plastic hinge damage (c) Specimen SF-1 force-displacement hysteresis Source: Haraldsson et al. (2013). Note: rft = reinforcement. (d) Specimen SF-3 force-displacement hysteresis Full shear friction rft Flexural rft in column slot Figure 2-59. Column–footing pocket/socket connection tests.

62 Proposed AASHTO Seismic Specifications for ABC Column Connections (a) Bent detail Source: Kavianipour and Saiidi (2013). (b) Footing pocket (c) Column base damage (d) Force-displacement hysteresis displacement (in) Fo rc e (k ip s) Figure 2-60. Bent with column–footing pocket connections.

State-of-the-Art Literature Review 63 to displacement ductility values of 9.36 and 8.36 without failure. The tests had to be terminated at this point to avoid overloading the shake table system. The maximum resultant residual displacement was minimal (0.91%). Larosche et al. (2014b) tested four full-scale exterior pile-to–bent cap pocket connections under slow cyclic loading at the University of South Carolina to investigate the performance of connections between precast prestressed square piles and CIP pile caps. The connection details of the four models were as follows: • One reference model representing the current design practice in South Carolina, which uses an embedment length of one pile width into the bent cap to form a moment connection (EB-18-1); • Two connections intended to behave as moment connections (EB-26-1 and EB-22-1); and • One model designed to act as a hinge connection by using No. 6 dowels combined with an embedment length of 0.11 times the pile width to act as a shear key (EB-2-1). Of the two moment connections, one model (EB-26-1) relied on substantial increases in the reinforcement within the bent cap relative to the reference connection and used an embedment length of 1.44 times the pile width. The second model (EB-22-1) increased the bent cap overhang dimension relative to the reference connection and used an embedment length of 1.22 times the pile width. All the test models used square concrete piles 18 in. (457 mm) wide and 192 in. (4.88 m) long that were prestressed with strands in a circular configuration. The cap beams were cast around the extended portion of the piles. Variable axial tension and compression were applied to the piles by an inclined actuator connected to the interior end of the bent cap. For these tests, the damage to the pile cap connection is shown in Figure 2-61. The test results showed that increasing the embedment length of the pile led to improved plastic hinge develop- ment. The reference model (EB-18-1) failed prior to yielding as a result of prying action of the pocket. The damage to this connection was dominated by a large peripheral crack outside the pile–bent cap interface. Because Model EB-2-1 had been intentionally designed to behave as a hinge, the plastic hinge within the pile did not develop and no damage to either element of the connection was observed. Instead, the damage was limited to the dowel bars embedded in the cap beam. Both moment connections failed as a result of the hinging of the piles below the pile cap soffit. It was concluded that increasing the cap overhang leads to similar performance as that of the heavily reinforced model and could be more efficient in terms of cost and materials. Furthermore, the test results indicated that increasing the pile embedment length to at least 1.3 times the pile width allows for plastic hinge to form in the pile rather than the cap beam. Model EB-2-1 achieved a drift ratio capacity of 5.1%. The pile with the heavily reinforced con- nection (EB-26-1) showed lower plastic moment capacity, plastic hinge length, and drift ratio capacity as compared with the model with the extended cap (EB-22-1). Model EB-26-1 obtained an average drift ratio of 3.2% relative to the corresponding value of 3.7% for EB-22-2. A full-scale three-pile bent model was also tested by Larosche et al. (2014a) under an earth- quake displacement history to investigate the performance of different details for pocket connections between precast prestressed piles and CIP pile caps. The test model is shown in Figure 2-62a. The test model incorporated an exterior heavily reinforced pile (Pile A), a pile con- nection with an extended overhang (Pile C), and a pile satisfying the minimum South Carolina DOT requirements (Pile B). Square concrete piles 18 in. (457 mm) wide and 185 in. (4.88 m) long were used for all piles. The piles were prestressed with strands and embedded into the cap beam to a depth of 1.22 times the pile width (1.22Dc). The cap beam was cast around the extended portion of the piles. Damage patterns for the connections are shown in Figure 2-62, b–d. The test results showed that the cap beam and connections remained elastic and that the failure was the result of formation of plastic hinge in the piles adjacent to the cap beam. Piles A and B exhibited distributed damage along the plastic hinge region of the piles. However,

64 Proposed AASHTO Seismic Specifications for ABC Column Connections the damage to Pile C was more concentrated to the connection between the pile and the cap. Damage to the cap beam was limited to some minor spalling of concrete in the cap beam about the perimeter of the exterior connections. It was concluded that the proposed modification to the exterior connections could be successful in preventing cracking of the cap beam. The bent drift ratio capacity was 5.9% under three times the design level earthquake record, which cor- responded to a displacement ductility of 8.8. Mehrsoroush and Saiidi (2014) tested a one-third-scale, two-column bent model under quasi-static cyclic loading at the University of Nevada, Reno, to investigate the seismic per- formance of a new column–cap beam pocket connection and a new pipe-pin connection for (a) Reference pile (EB-18-1) (b) Pile with 0.11Dc (EB-2-1) (c) Pile with 1.44Dc (EB-26-1) Source: Larosche et al. (2014b). (d) Pile with 1.22Dc (EB-22-1) Figure 2-61. Damage to pile cap pocket connections after testing.

State-of-the-Art Literature Review 65 Source: Larosche et al. (2014a). (b) Damage to Pile A (c) Damage to Pile B (d) Damage to Pile C (a) Multiple-pile bent detail Figure 2-62. Multiple-pile bent with cap pocket connections.

66 Proposed AASHTO Seismic Specifications for ABC Column Connections column bases they had developed. The bent detail is shown in Figure 2-63a. The cap beam was a precast member intended for adoption in ABC. The left column (ABC-Col) was a precast member that used ECC in the plastic hinge region as well as the embedded part of the column in the pocket. The right column (CIP-Col) was a CIP RC column. The new pocket connections aimed at providing a moment connection between the columns and the precast cap beam. The pocket connection was innovative because, unlike previously developed cap beam pocket con- nections, there were no cap beam bottom bars passing through the pocket. Furthermore, the entire column was precast. These features substantially simplified the construction and made the details more attractive for ABC. The pockets were made by using corrugated steel pipes to enhance shear transfer between the column and the pocket. Both columns were embedded into (c) Damage to cap beam Source: Mehrsoroush and Saiidi (2014). (a) Bent detail (b) Cap beam pocket detail 1143 [45] 25 40 [1 00 ] Precast Cap Beam 51 [2] 63 5 [2 5] 63 5 [2 5] 1143 [45] 3048 [120] 6. 4 [0 .2 5] 88 9 [3 5] #3 Spiral @ 51 [2] 25 [1] Clear Cover 12 #6 81 3 [3 2] 81 3 [3 2] 5334 [210] 41 72 [1 64 .2 5] 2032 [80] Ø5 08 [Ø 20 ] 10 16 [4 0] Pocket Connection Pipe Pin Connection C IP C ol um n Pr ec as t Pe de st al AB C C ol um n C on ve nt io na l C on cr et e EC C Grouting Duct Ø=102 [4] L=152 [6] Bearing Plate 406x406x38 [16x16x1.5] Threaded Rod Ø=38 [1.5] Corrugated Steel Pipe O.D.=635 [25] I.D.=610 [24] #3 Mesh 61 0 [2 4] 51 [2] #4 Spiral @ 38 [1.5] 2 #6 610 [24] 152 [6] 152 [6] 63 5 [2 5] #6 51 [2] 81 3 [3 2] 914 [36] #6 Bundled Bars 96 5 [3 8] 25 [1 ] #3 Mesh 102x102 [4x4] CL High-Strength Grout Column Cap Beam Figure 2-63. Two-column bent with cap pocket connections.

State-of-the-Art Literature Review 67 the pockets to a depth of 1.2Dc. The space between the column and the pocket was filled with high-strength, nonshrinkage grout. A threaded rod was embedded at the top of the column and anchored on the top of the cap beam that passed through the grouting duct to help transfer the column tensile forces. A reinforcing mesh was provided in the slab on top of the pocket to enhance the punching shear resistance of the upper part of the cap beam covering the pocket under axial forces of the column. A confining spiral was also provided around the full height of each pocket to increase the splitting resistance of concrete caused by bearing of the column against the pocket face. The detailing of the pocket connections was the same for both columns (Figure 2-63b). The design of the cap beam was based on the column overstrength moment, which was 1.2 times the column plastic moment. The columns were not subjected to external axial forces, to allow for the development of the tensile forces in the base pipe pins under over- turning moments. The cap beam was post-tensioned with a 400-kip (1780-kN) force to hold the actuator loading plate in place. The test results indicated that the cap beam remained elastic, as the measured strains in the reinforcement were well below the associated yield strains. Figure 2-63c shows the damage to the pocket connections at 12% drift ratio. No sign of damage to the CIP column pocket connection was observed; however, the column longitudinal bars slipped out of the ECC pocket because of the low bond strength of ECC and the steel pipe. The enclosed area within the CIP-Col hysteretic loops was larger in comparison with that of ABC-Col, indicating larger energy dissipation. After 4% drift ratio, strength degradation and extensive pinching were observed in ABC-Col as a result of slippage of the column bars from ECC in the embedded part of the column. CIP-Col reached a drift ratio capacity of 10%, corresponding to a displacement ductility of 9. It was concluded that pocket connections constructed by using corrugated steel pipes with a column embedment length of 1.2Dc performed well in forming the plastic hinge in the columns. The low maximum strains measured on the top threaded rods revealed that the column tensile force was trans- ferred to the cap beam essentially through the corrugated steel pipes; thus, the top rods were not needed. The hoop strains were small within the top portion of both pockets but increased near the column–cap beam interfaces, indicating that the lower portion of the spirals was effective in providing confining pressures around the pocket; the upper spirals were not needed. The maximum strain in the reinforcing mesh, which was provided to enhance the punching shear resistance of the slab on top of the cap beam pockets, was very small, which suggests that the mesh can be eliminated. Mashal et al. (2014) tested a half-scale, two-column bent under cyclic loading in which pocket connections were utilized at the column base (Figure 2-64). The diameter of the (a) Test setup (b) Hoop bars around pocket (c) Column installation Source: Mashal et al. (2014). Figure 2-64. Two-column bent with footing pocket connections.

68 Proposed AASHTO Seismic Specifications for ABC Column Connections columns was 20 in. (500 mm). The column height from the footing surface to the actuator centerline was 114.7 in. (2915 mm). The embedment length of columns inside the footing pockets was 20 in. (or 1.0Dc). Three hoop bars were placed around the footing pockets at both the top and bottom layers of the footing reinforcement to confine the pockets (Figure 2-64b). The test results showed that full plastic moment was developed at the column ends and was maintained up to a 3.4% drift ratio (equivalent to a displacement ductility capacity of 4.2), at which the test was stopped. Thonstad et al. (2016) tested on shake tables a quarter-scale model supported on three two-column bents. The precast post-tensioned columns were connected to the footings that used wet pocket connections in which the footings were cast after the placement of the precast columns (Figure 2-65). The columns were hybrid, since both reinforcing steel bars and post- tensioning tendons were incorporated in all column connections. Two sizes of reinforcing steel bars were used as the column longitudinal reinforcement: six No. 3 (Ø10 mm) and six No. 4 (Ø13 mm) bars. Four unbonded post-tensioning strands 3⁄8 in. (9.5 mm) in diameter were utilized in the columns. Steel jackets were incorporated at the both ends of the columns to reduce the damage during rocking. The columns were embedded 22 in. (1.8Dc) into the footings. The bridge was tested under several motions, and it was reported that many column longitudinal bars fractured under high-amplitude earthquake motions, which indicates that the column bars were fully anchored in the pocket connections. The peak drift ratio capacity of the bridge was more than 10%. Furthermore, the bridge residual drifts were insignificant because of the rocking mechanism. Mohebbi et al. (2015) tested on a shake table a one-third-scale hybrid rocking square column incorporating UHPC in the plastic hinge region. For the first time, CFRP tendons were utilized to post-tension a column (Figure 2-66a). The column was considered to be a hybrid member because of the combination of reinforcing steel bars and post-tensioning tendons. The column section was a square with a side dimension of 20 in. (508 mm). The column height from the footing surface to the centerline of the loading link was 80 in. (2,032 mm). The precast column was first post-tensioned then connected to the footing by using a pocket connection (Figure 2-66b). The column embedment length into the pocket was 1.0Dc (Dc being the column side dimension). The 1.5-in. (38-mm) gap between the pocket and the column was filled with UHPC. Full plastic moment in the column above the footing surface was developed under (a) Precast columns (b) Test setup (c) Column base damage at design level earthquake Source: Thonstad et al. (2016). Figure 2-65. Two-column bent, two-span bridge with footing pocket connections.

State-of-the-Art Literature Review 69 multirun uniaxial shake table tests. The apparent damage to the plastic hinge was minimal (Figure 2-67c) because of the superior mechanical properties of UHPC, and the residual dis- placements were insignificant because of the post-tensioning. Mehraein and Saiidi (2016) tested two 0.27-scale, two-column bents on a shake table. The columns were connected at the base to 30-in. (762-mm)-high, 22-in. (559-mm)-diameter pile shafts using pipe-pin connection in one specimen and rebar hinge connection in the other specimen. The bents were labeled BPSA (bent with pipe-pin column shaft connections for ABC) and BRSA (bent with rebar pin column shaft connections for ABC). In both test models, columns were connected to precast cap beams by means of pocket connections. The BPSA bent detail is shown in Figure 2-67a. The pockets were fabricated in the cap beams by using corrugated steel pipes. Furthermore, steel spirals were incorporated around each pipe within the lower one-third of the pocket height. The detailing of the precast cap beams was the same for both test models (Figure 2-67, b and c). One column per each bent was CIP and the other column was precast. The prefabricated columns were hollow. Fourteen No. 4 (Ø13 mm) column longitudinal bars were extended into the cap beam pockets to a length of 32db, which was equivalent to 1.0Dc., The cap beam longitudinal bars did not pass through the pockets. Rather, they were bundled to the sides. The hollow column and pockets were filled with self-consolidating concrete. The columns were not subjected to external axial forces, to allow for the development of tensile forces in the base connections under overturning moments. The cap beams were designed on the basis of the column overstrength moment, which was 1.2 times the column plastic moment. The BPSA spec- imen was tested first under several scaled ground motions of the 1994 Northridge earthquake recorded at the Sylmar station with increasing amplitudes until failure. In Run 3, which was 72% of the design level earthquake, the column drift demand was 4%, and the peak measured cap beam longitudinal bar strains were 925 µe (42% of the yield strain) and 1,550 µe (70% of the yield strain) for the bottom and top layers of reinforcement, respectively. No yielding of bars in the cap beam was observed up to Run 3, but the cap beam–load-cell connection failed during Run 3. After Run 3, the cap beam was repaired and post-tensioned with a 400-kip (1,780-kN) force. Upon post-tensioning, the estimated compressive strain in the cap beam longitudinal bars was 150 µe. The bent with the post-tensioned cap beams was subsequently tested under stronger motions (85% to 200% of the design level earthquake) until failure. The peak measured cap beam longitudinal bar strains were less than 410 µe (18% of the yield strain), which confirmed the capacity-protected behavior of the cap beam. On the basis of these observations, the cap (a) Carbon fiber–reinforced polymer tendon (b) Footing pocket (c) Damage after 200% design level earthquake Source: Mohebbi et al. (2015). Figure 2-66. Post-tensioned column with footing pocket connection.

70 Proposed AASHTO Seismic Specifications for ABC Column Connections (a) BPSA bent detail (b) Cap beam pocket connections (c) Cap beam sections Source: Mehraein and Saiidi (2016). Figure 2-67. Two-column bent cap pocket connections. beam of the BRSA specimen was post-tensioned prior to testing, and a similar loading protocol was applied. The peak measured cap beam longitudinal strain was 150 µe (7% of the yield strain) in shake table tests, which indicated linear elastic behavior of the cap beam. 2.4.3 Summary of Pocket/Socket Connection Tests A summary of the findings from the bridge column pocket connection tests is presented in Table 2-14. The table includes the substructure connection location, the geometry and detail- ing of each test specimen, the pocket embedment length, and the overall seismic performance of the test model. 2.4.4 Pocket/Socket Connection Field Application Pocket connections have been used in a few states to connect precast substructure elements. The Texas DOT has utilized precast cap beams in several projects. In fact, Texas was the first state to

State-of-the-Art Literature Review 71 (continued on next page) Reference Connection Column Geometry Embedded Length Remarks Matsumoto et al. (2001)(a) Column–cap beam No. of columns: one Scale factor: 100% Section: circular Diameter: 30 in. (762 mm) Longitudinal bars: 12 No. 9 (Ø29 mm) Transverse bars: No. 3 (Ø10 mm) spirals at 4 in. (102 mm) Filler: grout 0.5Dc Only four column longitudinal bars were extended into the pocket. There was no reference column. Minor damage to the column and cap beam was reported up to the column yielding. Zhu et al. (2006) Column– footing No. of columns: three Scale factor: 17% Section: circular Diameter: 12 in. (305 mm) Longitudinal bars: eight No. 6 (Ø19 mm) Transverse bars: 0.2-in. thick CFFT tube Filler: grout and concrete 1.0Dc All CFFT models performed better than the reference model in terms of initial stiffness, strength, ductility, and energy dissipation capacity as a result of the confinement provided by the CFFT tubes. Restrepo et al. (2011) Column–cap beam No. of columns: two Scale factor: 42% Section: circular Diameter: 20 in. (508 mm) Longitudinal bars: 16 No. 5 (Ø16 mm) Transverse bars: No. 3 (Ø10 mm) spirals at 2 in. (51 mm) Filler: concrete 35db or 1.2Dc 27% and 13% lower drift capacities compared to that of the reference column were reported for Columns CPFD and CPLD, respectively. Plastic hinge formed in columns. Cap beam yielded in both tests. Motaref et al. (2011) Column– footing No. of columns: one two-column bent Scale factor: 30% Section: circular Diameter: 14 in. (356 mm) and 14.57 in. (370 mm) Longitudinal bars: eight No. 6 and seven No. 3 Transverse bars: 0.27-in. thick GFFT tube and No. 3 spirals at 2 in. Filler: grout 1.5Dc Full plastic moment was developed in both columns. A drift ratio capacity of 11% was achieved. Ziehl et al. (2011) Pile to pile cap No. of piles: two Scale factor: 100% Section: square Diameter: 18 in. (457 mm) Longitudinal tendons: Nine 0.5-in. tendons Transverse bars: W6 spirals at 1.0 in. (25 mm) Filler: concrete 1.44Dc The connection performed well. Large displacement capacities were reported. Tran (2012) Column– pile shaft No. of columns: two Scale factor: 28% Section: circular Diameter: 20 in. (508 mm) Longitudinal bars: 10 No. 5 (Ø16 mm) Transverse bars: 0.24-in. wire spirals at 1.25 in. (32 mm) Filler: pile concrete 1.4Dc Plastic hinge formed in the column when the pile was heavily confined. The second specimen failed as a result of prying action in the pile. Table 2-14. Summary of seismic performance of column test models with pocket/socket connections.

72 Proposed AASHTO Seismic Specifications for ABC Column Connections Reference Connection Column Geometry Embedded Length Remarks Haraldsson et al. (2013) Column– footing No. of columns: three Scale factor: 42% Section: circular Diameter: 20 in. (508 mm) Longitudinal bars: 8 No. 6 (Ø19 mm) Transverse bars: 0.24-in. wire spirals at 1.25 in. (32 mm) Filler: footing concrete 1.1Dc and 0.5Dc The failure mode for the two columns with 1.1Dc embedment length was buckling and fracture of longitudinal bars, whereas the column with shorter embedment length failed as a result of failure in the connection. Kavianipour and Saiidi (2013) Column– footing No. of columns: one two-column bent in a four-span bridge Scale factor: 25% Section: circular Diameter: 14.75 in. (375 mm) Longitudinal bars: 7 No. 3 (Ø10 mm) Transverse bars: 0.27-in.-thick CFFT tube Filler: grout 1.5Dc The two-column bent with pocket connections exhibited a large drift ratio capacity with no apparent damage to the connection. Larosche et al. (2014b) Pile to pile cap No. of piles: four Scale factor: 100% Section: square Diameter: 18 in. (457 mm) Longitudinal tendons: nine 0.5-in. tendons Transverse bars: 0.27-in. wire spirals at 3.0 in. (25 mm) Filler: concrete 0.11Dc, 1.0Dc, 1.22Dc, and 1.44Dc The pile with 0.11Dc embedment length acted as a pin connection. The pile with 1.0Dc embedment length failed prior to yielding as a result of prying action in the pocket. The piles with the larger embedment lengths showed high displacement ductility capacities (8.8 and 14). Larosche et al. (2014a) Pile to pile cap No. of piles: one three-pile bent Scale factor: 100% Section: square Diameter: 18 in. (457 mm) Longitudinal tendons: nine 0.5-in. tendons Transverse bars: 0.27-in. wire spirals at 3.0 in. (25 mm) Filler: concrete 1.22Dc The bent drift ratio capacity was 5.9% under three times the design level earthquake, which corresponded to a displacement ductility of 8.8. Mehrsoroush and Saiidi (2014) Column–cap beam No. of columns: one two-column bent Scale factor: 33% Section: circular Diameter: 20 in. (508 mm) Longitudinal bars: 12 No. 6 (Ø19 mm) Transverse bars: No. 3 (Ø10 mm) spirals at 2 in. (51 mm) Filler: grout 1.2Dc The column cast with conventional concrete in the pocket exhibited 12% drift ratio capacity with no damage to the connection. However, the column cast with ECC in the pocket failed in the pocket region at 4% drift ratio. Mashal et al. (2014) Column– footing No. of columns: one two-column bent Scale factor: 50% Section: circular Diameter: 20 in. (500 mm) Longitudinal bars: eight No. 5 (Ø16 mm) Transverse bars: No. 3 (Ø10 mm) hoops at 3.0 in. (75 mm) Filler: grout 1.0Dc The connection performed well up to a displacement ductility capacity of 4.2, at which the test was stopped. Table 2-14. (Continued).

State-of-the-Art Literature Review 73 Reference Connection Column Geometry Embedded Length Remarks Thonstad et al. (2016) Column– footing No. of columns: three two-column bent bridge Scale factor: 25% Section: octagonal Diameter: 14 in. (305 mm) Longitudinal bars: six No. 3 (Ø10 mm) and six No. 4 (Ø13 mm) Transverse bars: wire spirals and steel jackets Filler: footing concrete 1.8Dc Many column longitudinal bars fractured, which indicates that bars were fully anchored in the pocket connections. The bridge showed more than 10% drift ratio capacity with minimal residual displacements. Mohebbi et al. (2015) Column– footing No. of columns: one Scale factor: 33% Section: Square Diameter: 20 in. (508 mm) Longitudinal bars: 24 No. 4 (Ø13 mm) Transverse: No. 3 (Ø10 mm) ties at 2.75 in. (70 mm) Filler: UHPC 1.0Dc Full plastic moment was developed in the column. The hybrid rocking column exhibited 7% drift ratio capacity with no damage of the pocket connection. Mehraein and Saiidi (2016) Column–cap beam No. of columns: two two-column bents Scale factor: 27% Section: circular Diameter: 16 in. (406 mm) Longitudinal bars: 14 No. 4 (Ø13 mm) Transverse bars: 0.23-in. wire spirals at 2.0 in. (51 mm) Filler: self-consolidating concrete 32db or 1.0Dc The bents performed well before and after cap beam post-tensioning. Plastic hinges developed in the columns with no damage of the pocket connections. Note: db = column/pile longitudinal bar diameter; Dc = either the column diameter or the dimension of the column’s largest side. aThis was not a column. It was an RC stub with four bars and was not subjected to cyclic loads that represent earthquakes. Table 2-14. (Continued). use prefabricated bent caps in the United States (Roddenberry and Servos 2012). Other states that used substructure pocket connections are Florida, Iowa, Louisiana, Minnesota, Washington, and South Carolina. Photographs and details of some of the connections are shown in Figure 2-68. 2.5 Pipe-Pin Connections 2.5.1 Introduction Pipe-pin hinges were first developed by the Caltrans to connect columns to cap beams with a moment-free mechanism (Figure 2-69a). A pipe-pin connection consists of a steel pipe that is anchored in the column and a steel can installed in the column’s adjoining member. A small gap between the steel pipe and the can enables the extended segment to freely rotate inside the can and prevents double curvature bending of the protruded segment inside the can. This type of connection is designed to transfer: (1) compressive axial load through bearing, and (2) shear, either through contact of the protruded pipe and the adjoining member’s steel can or through friction force at the interface with the adjoining member. No axial tension or moment is transferred with this connection. Contribution of the friction force to the shear strength may decrease over time because of the abrasion of concrete under cycles of temperature movement. To enhance the rotational capacity of the hinge, a circular hinge throat is provided at the interface with the adjoining member by using a steel ring plate or a circular concrete-bearing area. Under earthquake loading, the top pipe-pin connection is intended to perform as a two- way flexural hinge that transfers only shear and axial loads.

(a) Column–cap beam, Texas (b) Column–cap beam, Texas (f) Steel pile–pile cap, Iowa (c) Pile–pile cap, Florida (d) Column–footing, Washington State (e) Pile–pile cap, Minnesota Figure 2-68. Field application of pocket connections.

State-of-the-Art Literature Review 75 (a) Top pipe pin to transfer shear and compressive axial loads (b) Base pipe pin to transfer shear and any axial loads RC Bent Cap RC Column Threaded Rod Steel Can Inner Spiral Ring Plate Can Spiral Inner Pipe Hinge Gap CIP Footing Anchorage System PT Strand Ring Plate Outer Pipe Shear Stud Inner Pipe Anchorage System RC Column Figure 2-69. Pipe-pin connections. Because of the simplicity of pipe pins, engineers have also been interested in adapting them to link bridge columns to footings or pile shafts, although column base pipe pins are yet to be utilized in the field because research data for these connections have only recently become available. In contrast to top pipe pins, which are composed of a steel pipe and a steel can, base pipe pins must be composed of two steel pipes: one pipe is embedded in the footing (or pile shaft) (inner pipe) and the other in the column (outer pipe). The inner pipe protrudes from the footing (or pile shaft) and extends into the outer pipe (Figure 2-69b). Column shear forces are transferred to the footing (or pile shaft) through contact of the pipes as well as friction at the column–footing interface. The uplift force is resisted by a tension member (post-tensioning strands or high-strength threaded rod) that is anchored at the ends in the column and the foot- ing (or pile shaft). To enhance the rotational capacity of the hinge, the column is placed over a steel or rubber ring plate on the footing (or pile shaft) to form a gap between the column edge and the footing. The column compressive forces are transferred to the footing (or pile shaft) through bearing on the ring plate. The outer pipe force is transferred to the column through welded studs on the surface of the pipe. 2.5.2 Previous Studies Steel pipes were first used in the rehabilitation of structures as shear keys to transfer forces between walls and columns in lieu of interface dowels. Frosch (1999) tested four specimens under slow cyclic shear loading to investigate the pipe connection between the precast infill wall panels and existing building frame elements. Details of the test models are shown in Figure 2-70a. The test variables were the pipe embedment length, the pipe diameter, and the wall thickness. The study demonstrated that failure of the specimens consisted of interface sliding along with shear or flexural yielding of the steel pipe (Figure 2-70, b and c). Also, local- ized bearing failure occurred in the wall adjacent to the pipe at the interface owing to the dis- tribution of variable bearing stress along the pipe length. Frosch concluded that to eliminate the possibility of failure at the panel interface and to induce shear yielding in the pipe instead of flexural yielding, the pipe must have sufficient embedment length.

76 Proposed AASHTO Seismic Specifications for ABC Column Connections (a) Details of pipe shear lug (b) Flexural yielding (c) Shear yielding Source: Frosch (1999). Figure 2-70. Wall–column pipe connections. Pipe shear keys were used in aerial guideway structures in the San Francisco Bay Area Rapid Transit (BART) to connect the girders to the closure pour (horizontal shear key) and the closure pour to the cap beams (vertical shear key). Four proof tests were conducted by Restrepo and Panagiotou (2005) as a part of the BART earthquake safety program. Two tests were performed on the horizontal shear keys and two on the vertical shear keys using round and square steel pipes. The specimens failed as a result of local bearing failure of concrete against the pipe, which caused substantial plastic deformation of the connections. Although none of the specimens failed in shear, an empirical equation was developed to estimate the shear capacity of the pipe shear keys. To investigate the overall response of pipe-pin connections, Doyle and Saiidi (2008) performed slow cyclic loading tests of two 0.3-scale specimens representing Caltrans pipe-pin details to failure. Details of the test models are presented in Figure 2-71, a and b. The first model was tested under pure shear loading (PS1) and the second model under combined flexure, shear, and axial loads (PF1) to represent an actual bridge column loading. PS1 was composed of two blocks with the hinge located at the interface. Model PF1 had the same pin connection detail as that of Model PS1, but the pin was placed at the top of a column. The hinge throat in both test models was formed through a circular elevated concrete bearing area around the pipe. The test model was composed of a footing, column, and loading head, which was connected to the column by using the pipe-pin detail. Model PF1 was tested by using a double actuator con- nected to the loading head to restrain rotations. The test models were subjected to an axial load of 100 kips (444.8 kN), which corresponds to an axial load index of 5.87% for the column during the course of experiments. The axial load index is defined as the ratio of the column dead load to the gross cross-sectional area multiplied by the specified compressive strength of concrete. Figure 2-71, c and d, shows the damage to the test models. Under pure shear loading (Model PS1), the steel pipe fractured and the concrete failed at the interface. In contrast, under flexural loading (Model PF1), the capacity was controlled by the failure of concrete against the pipe with- out pipe fracture. Both test models showed considerable flexural yielding of the pipe; however, the lateral load capacity and the stiffness of Model PS1 was higher. The drift ratio capacity for Model PF1 was 10%. A simple design method was developed to estimate the ultimate capacity of pipe-pin connections used at the top of columns.

State-of-the-Art Literature Review 77 Source: Doyle and Saiidi (2008). (a) Detail of PS1 (b) Detail of PF1 (c) PS1 pipe shear-off (d) PF1 pipe flexural bending Figure 2-71. Test specimens with top pipe-pin connections. Zaghi and Saiidi (2010 and 2011) conducted monotonic testing of three pairs of 0.28-scale push- off specimens to evaluate the bearing strength of concrete against steel pipes. The test variables were the pipe diameter [2.5 and 3.5 in. (64 and 89 mm, respectively)] and the confining spiral around the pipe (inner pipe). The specimens were built in pairs to determine any scatter in the data. The loading protocol consisted of a pull loading to crack the specimen and then pushing to failure (Figure 2-72). The test results confirmed that the inner spiral slightly increased the push load capacity. The push loading capacity was much larger than the pull loading capacity, which was attributed to the difference in load-resisting mechanisms under the two loading types. Under the pull loading, tensile cracking of concrete controlled the capacity, but under the push loading, the bearing failure of concrete governed the behavior. It was concluded that, depending on the geometry, concrete failure may occur because of either bearing failure or splitting failure.

78 Proposed AASHTO Seismic Specifications for ABC Column Connections (a) Push loading b) Pull loading Source: Zaghi and Saiidi (2011). Figure 2-72. Pipe-pin push-off tests. The failure mode when the pipe was embedded in a large body of concrete was due to bearing fail- ure of the concrete against the pipe. The tests further showed that the average bearing strength of concrete against the pipe may be taken as twice the concrete uniaxial compressive strength, owing to the confining effect of the concrete surrounding the pipe. To determine the shear capacity of concrete-filled steel pipes, three pairs of concrete-filled pipes 12 in. (305 mm) long were tested by Zaghi and Saiidi (2011) under pure shear in a double shear configuration. The test parameters were the diameter and thickness of the steel pipes. The first pair of the pipes had a 4-in. (102-mm) diameter and 0.226-in. (5.7-mm) thickness. The second pair of test specimens employed steel pipes with a 4-in. (102-mm) diameter and 0.318-in. (8.1-mm) thickness. The corresponding dimensions for the third pair were 2.88 in. (73 mm) and 0.276 in. (5.16 mm), respectively. The clear shear span of all specimens was half of the outer diameter of the pipe. On the basis of the test results, an empirical design equation was developed to calculate the shear strength of infilled pipes. On the basis of different possible limit states, Zaghi et al. (2012) proposed a method for determining the lateral load capacity of column–cap beam pipe-pin connections. To vali- date the design method, a 0.2-scale two-column bridge bent was constructed and tested on a shake table. One of the columns was conventional RC and the other was a concrete-filled GFRP tube. The columns were connected to a two-piece cap beam at the top and a spread footing at the base by means of pipe-pin and moment connections, respectively. Details of the pipe-pin connections are illustrated in Figure 2-73, a and b. The hinge throat in both columns was formed through a circular elevated concrete bearing at the column–cap beam interfaces. Each column was subjected to an external axial load of 50 kips (178 kN) prior to testing, which resulted in an axial load index of 6.5%. Figure 2-73, c and d shows the post-test damage of the top hinge region. During the test, the pipes remained perfectly straight with no sign of yielding, the edge of the RC column spalled because of the hinge gap closure, and the bearing areas were ground because of the large number of load cycles. The bent displacement was the combination of the column flexural displacements and the horizontal slippage at the pipe-pin hinges. Both columns exhibited wide and stable hysteretic loops; however, the lateral load capacity and the displacement ductility capacity of the GFRP column were higher than those of the RC column. The displacement ductility of the GFRP and RC columns was 10.5 and 12.2, respectively. The pipe-pin connections performed as moment-free hinges while transferring shear and axial loads. The measured strains confirmed that the pipes remained essentially elastic, as is desired for a

State-of-the-Art Literature Review 79 (c) Reinforced concrete column pipe-pin damage (d) Carbon fiber–reinforced polymer column pipe-pin damage Source: Zaghi et al. (2012). (b) Pipe pin after casting(a) Pipe pin before casting (a) (b) (c) (d) Figure 2-73. Two-column bent with cap beam pipe-pin connections. capacity-protected member. The proposed design guideline was validated on the basis of the results of the two-column bent test and comprehensive finite element analytical studies. Motaref et al. (2011) performed shake table testing of a 0.3-scale precast two-column bridge bent model using pipe-pin connections at the column–cap beam joints. Details of the two-column bent model as well as pipe-pin connections were previously shown in Figure 2-56. This study investigated the feasibility and performance of top pipe pins in precast bridge pier construction. The cap beam was precast by incorporating steel cans to accommodate column pipes. The test results confirmed that pipe-pin connections can be effectively used in ABC. Biaxial shake table tests of a quarter-scale four-span bridge were performed by Kavianipour and Saiidi (2013) at the University of Nevada, Reno, as discussed in Section 2.4, “Pocket and

80 Proposed AASHTO Seismic Specifications for ABC Column Connections Socket Connections” (Figure 2-60). Two out of three bents (Bent 1 and Bent 3) of the bridge were constructed with pipe-pin connections between the columns and cap beams to act as hinges and thereby reduce demands. The cap beams were fully precast, which allowed quick installation suitable for ABC (Figure 2-74). The general geometry of the bents and the pipe-pin details is shown in Figure 2-60. The pipe-pin connections performed well under large-amplitude motions by preventing moment transfer between the cap beams and columns. Column base pipe-pin connections were first developed by Mehrsoroush and Saiidi (2014). The seismic performance of the connections was investigated in a quasi-static cyclic test of a one-third-scale, two-column bent model in which pipe-pin and pocket connections were utilized to connect columns to the footing and cap beam, respectively (Figure 2-63). The test model was composed of six segments: a precast concrete pedestal that formed the lower part of the precast column, a precast ECC-concrete column, a conventional CIP RC column, a pre- cast cap beam, and two single footings. The pipe-pin connection of the precast column was incorporated in a concrete pedestal (Figure 2-75). The uplift force was resisted by a 1.5-in. Source: Kavianipour and Saiidi (2013). (a) Step 1 (b) Step 2 (c) Step 3 Figure 2-74. Installation of a precast cap beam with pipe-pin connections. (a) Pipe pin in pedestal (b) Precast column damage at 10% drift (c) CIP column damage at 10% drift Source: Mehrsoroush and Saiidi (2014). Figure 2-75. Precast column with pipe-pin connection.

State-of-the-Art Literature Review 81 (38-mm)-diameter high-strength threaded rod anchored by square bearing plates at the ends in the column and the footing. To enhance the rotational capacity of the column at the base, a circular steel ring plate 0.25 in. (6.4 mm) thick was placed at the column–footing interface, thereby forming a hinge gap. The shear force was transferred through the contact of the pipes as well as the friction at the adjoining member interface. The test results indicated that the pro- posed pipe-pin connection successfully resisted the force and deformation demands even under 12% drift ratio cycles. The pipe-pin connection remained essentially elastic. The maximum measured strains in the outer pipes, spirals, and longitudinal bars at the pipe-pin connections remained below their yield strains. As shown in Figure 2-75, the damage to the columns during the tests was limited to some minor spalling of concrete at the column edges that resulted from the closure of the hinge gap during large base rotations. The base gap was closed when 5% and 6% drifts in push and pull loadings, respectively, were reached. Furthermore, the results revealed that the inner pipe embedded in the footing needs to be designed for shear forces that exceed the column design shear resulting from friction developing in reverse direction under high base rotations. The pipe-pin connections transferred some level of moment to the footings, owing to the eccentricity between the hinge compressive force acting at the rod center line and the con- tact force of the pipe at the top face of the inner pipe. Both pipe-pin connections resisted tensile loading that exceeded the yield force of the rods. Mehraein and Saiidi (2016) developed a new version of pipe-pin connections to link column bases to pile shafts. They conducted shake table tests of a 0.27-scale two-column bent incorporating pocket connections to link the columns to a precast cap beam as well as pipe- pin connections to link the columns to the pile shafts (BPSA). The two-column bent model details were presented earlier (Figure 2-67). One of the columns was CIP and the other was precast (referred to as ABC column). The ABC column was fabricated as a hollow precast element placed on the top of the pile shafts around the base pipe pin. The precast hollow col- umn was filled with self-consolidating concrete after the placement of the precast cap beam over the columns. The pipe pins were designed to improve rotational capacity while facili- tating construction. Details of the base pipe pins are depicted in Figure 2-76. The pipe pin was composed of two steel pipes to carry shear. A high-strength threaded rod was placed to resist the tensile force to the pile shaft. To enhance the rotational capacity, the lower pipe was machined at the protruded end. An elastomeric pad 0.25 in. (6.4 mm) thick was used between the column and pile shaft to provide a hinge gap. The force transfer mechanism of the base pipe-pin connection was the same as that developed by Mehrsoroush and Saiidi (2014), (a) Cast-in-place column (b) Accelerated bridge construction column Source: Mehraein and Saiidi (2016). Figure 2-76. Two-column bent with base pipe-pin connections.

82 Proposed AASHTO Seismic Specifications for ABC Column Connections except that the tensile force was resisted through two end plates installed on the pipe ends as well as welded studs on each pipe surface. The new pipe-pin detail showed excellent per- formance during the test. The 0.25-in. (6.4-mm)-thick gap did not close even when the bent was subjected to drifts higher than 8%. No damage was detected in the pipe-pin connections after the test. Varela and Saiidi (2016) tested a quarter-scale two-span bridge incorporating a novel deconstructible bent detailing. One of the objectives of this research was to develop a new concept that allows bridge columns and other components to be disassembled for reutiliza- tion or recycling at the end of life of the structure. The two-span bridge was made of three two-column bents. Each modular column of the bents was connected to a spread footing with a detachable plastic hinge and to a precast bent cap by using pipe-pin connections. The con- stituent elements of the modular columns are shown in Figure 2-77. The bridge was tested several times on shake tables with no damage of the columns in the plastic hinge regions and no damage of the pipe-pin connections. Source: Varela and Saiidi (2016). Note: CuAlMn = copper–aluminum–manganese. Figure 2-77. Cast-in-place column: modular two-column bent with top pipe-pin connections.

State-of-the-Art Literature Review 83 2.5.3 Summary of Pipe-Pin Connection Tests A summary of the seismic performance of bridge columns incorporating either top or base pipe-pin connections is presented in Table 2-15. 2.5.4 Pipe-Pin Connection Field Application The top pipe pin connections have been used in practice in several bridge columns including approach ramps of the replacement of the San Francisco–Oakland Bay Bridge (Figure 2-78). However, the concept of base pipe-pin connections is new and has not been deployed in practice. 2.6 Analytical Studies of ABC Column Connections 2.6.1 Introduction Many of the previous studies on the seismic performance of ABC connections have included analytical studies, often in conjunction with testing of physical models. The general purpose Reference Connection Column Geometry Remarks Doyle and Saiidi (2008) Column–cap beam No. of columns: one Scale factor: 30% Section: circular Diameter: 22 in. (559 mm) Longitudinal bars: 12 No. 11 (Ø36 mm) Transverse bars: No. 3 (Ø10 mm) spirals at 1.5 in. (38 mm) Concrete surrounding the pipe failed as a result of shear in pipe without pipe fracture. Zaghi and Saiidi (2011) Column–cap beam No. of columns: one two-column bent Scale factor: 20% Section: circular Diameter: 14 in. (356 mm) Longitudinal bars: 20 No. 4 and eight No. 4 Transverse bars: spiral and FRP tube The pipe-pin connections performed as moment-free hinges and remained elastic during the entire testing. Motaref et al. (2011) Column–cap beam As presented in Table 2-14 The pipe-pin connections performed well and were able to transfer the shear and axial loads. Kavianipour and Saiidi (2013) Column–cap beam As presented in Table 2-14 The pipe-pin connections performed well and were able to transfer the shear and axial loads. Mehrsoroush and Saiidi (2014) Column– footing As presented in Table 2-14 The base pipe-pin connections performed well, but the surrounding concrete failed in compression in high drift levels as a result of closure of the pipe-pin connection gap. Mehraein and Saiidi (2016) Column–pile shaft As presented in Table 2-14 No damage was detected in the pipe-pin connections after the test. Varela and Saiidi (2016) Column–cap beam No. of columns: three two-column bents in a two-span bridge Scale factor: 25% Section: circular Diameter: variable Longitudinal bars: variable Transverse bars: spiral and FRP tube The pipe-pin connections performed as expected and successfully transferred the shear and axial loads. Table 2-15. Summary of seismic performance of column test models with pipe-pin connections.

84 Proposed AASHTO Seismic Specifications for ABC Column Connections (a) Column top (b) Ramp (c) Superstructure reinforcement Figure 2-78. Field application of column top pipe-pin connections. of the analytical studies has been to evaluate the adequacy of modeling methods in duplicating experimental results and to identify necessary refinements to improve estimation of the actual behavior. Proven analytical models have also been used to develop a better understanding of the ABC connection stresses and deformations. Calibrated analytical models are frequently used by investigators to conduct parametric studies to identify trends and to generate information for eventual use in developing design guidelines. This section presents a summary of past analytical studies on ABC column connections. 2.6.2 Mechanically Spliced Columns The effect of GC length and location on the seismic performance of bridge columns was first investigated by Haber (2014), who used analytical models that had led to close estimates of large-scale test data. The variables were the GC length (Lsp = 4db, 8db, 12db, 16db, and 20db), the location of the coupler in the plastic hinge region [the pedestal height from 2 in. (0.04Dc) to 48 in. (1.0Dc)], and the column aspect ratio (L/Dc = 3, 4.5, and 6) (Figure 2-79a). The column diameter (Dc) was the same in all cases. The reference nonspliced columns were designed for a target displacement ductility of 7. The pedestal was assumed to be CIP with bonded bars in all cases. Each coupler was modeled using a generic stress–strain model devel- oped by Haber et al. (2015) in which the coupler stress–strain relationship is calculated using elastic and plastic scale factors (Figure 2-79b). Figure 2-79c shows the displacement ductility capacity of GC columns versus the coupler location. It can be seen that columns with the longest GCs installed immediately above the footing surface exhibit the lowest displacement ductility capacity (21% lower than that of the reference nonspliced column). The graph also indicates that shifting the GCs from the footing surface will significantly enhance the displace- ment ductility capacity of the column. For example, columns with GCs shifted by 0.5Dc and an aspect ratio of 4.5 are expected to exhibit the same displacement ductility capacity as that of nonspliced columns. Tazarv and Saiidi (2015a) developed a generic material model for mechanical bar couplers in which the stress–strain of the coupler region is modified on the basis of a coupler rigid length factor that they proposed. The coupler region was defined as the coupler length (Lsp) plus 1.0 db away from each end of the coupler. The main assumption in the new model is that a portion of a mechanical bar coupler (βLsp) is rigid, and thus does not contribute to the overall elongation of the splice (Figure 2-80a). The value β was defined as the coupler rigid length factor. Therefore, for the same tensile force, the coupler region axial deformation is lower (Figure 2-80b) on the basis of this approach, which results in lower strain in the coupler region (esp) as compared with the strain of the connecting reinforcing bar (es):

State-of-the-Art Literature Review 85 (a) Column details (b) Individual coupler model 0 0.25 0.5 0.75 1 5 5.5 6 6.5 7 7.5 0 8 16 24 32 40 48 Column Diameters, D yticapa C ytilitcu Dtne mecalpsi D Splice Location H (in) 16 8 4 Behavior Without Splice Lsp / db 0 0.25 0.5 0.75 1 5 5.5 6 6.5 7 7.5 0 8 16 24 32 40 48 Column Diameters, D yticapa C ytilitcu Dtne mecalpsi D Splice Location H (in) 16 8 4 No Splice Lsp / db 0 0.25 0.5 0.75 1 5 5.5 6 6.5 7 7.5 0 8 16 24 32 40 48 Column Diameters, D yticapa C ytilitcu Dtne mecalpsi D Splice Location H (in) 16 8 4 No Splice Lsp / db AR = 3 AR = 4.5 AR = 6 (c) Effect of splice length on displacement ductility capacity D = 48 in (1.22 m) Cover = 2.875 in (73 mm) No. 6 (D19) Hoop Spacing "s" Varies With AR No. 14 (D43) Bars Typically Cross Section HPed Lsp Mechanical Splice Footing L D L = 288 in (7.32 m) s = 2.5 in (63.5 mm) AR = 6.0 L = 216 in (5.49 m) s = 2.625 in (66.7 mm) AR = 4.5 L = 144 in (3.66 m) s = 3.125 in (79.4 mm) AR = 3.0 Center of Superstructure Mass or Point of Contraflexure Precast Column Element CIP Pedestal Source: Haber (2014); Haber et al. (2015). Note: AR = aspect ratio. Proposed Figure 2-79. Effect of grouted couplers on bridge column ductility. (2-7) sp cr sp cr L L Ls ε ε = − β where Lcr is the length of the coupler region. Overall, the stress–strain relationship of any type of mechanical bar splice can be determined by knowing only the coupler rigid length factor (β). The condition in which β = 0 is similar to a nonspliced connection in which the stress–strain of the coupler region is the same as the reinforcing bar stress–strain. An extensive parametric study was carried out using the proposed coupler model to inves- tigate coupler effects on the seismic performance of bridge columns. Twelve reference CIP

86 Proposed AASHTO Seismic Specifications for ABC Column Connections Source: Tazarv and Saiidi (2015a). Column Section Diam. 4' [1.22] 22-#9 [22-Ø29] Cover: 2 in. [51 mm] 4'-0" [1.22] Hsp Lsp Elem. 1 Elem. 2 E le m . 3 L Footing C ou pl er s Figure 2-81. Analytical modeling detail for mechanically spliced columns. (a) Coupler region (b) Coupler stress–strain model Source: Tazarv and Saiidi (2015a). C ou pl er R eg io n Lsp db a.db B ar R eg io n a.db ß.Lsp B ar R eg io n R ig id L en gt h Lcr St re ss Strain Bar Region Coupler Region Figure 2-80. Generic stress–strain model for mechanical bar splices. RC columns with no coupler were designed according to the AASHTO Guide Specifications for LRFD Seismic Bridge Design (AASHTO 2014) to achieve displacement ductility capacities of 3, 5, and 7. Two axial load indexes of 5% and 10%, and three column aspect ratios of 4, 6, and 8 were included in the analysis. The axial load index is defined as the ratio of the column axial load to the product of the specified compressive strength of column concrete and the column gross section area. The aspect ratio is the ratio of the column height to the column diameter. Figure 2-81 shows the general modeling method of mechanically spliced columns. The variables in this parametric study were coupler length (Lsp), pedestal height (Hsp), coupler rigid length factor (β), and the vertical distance of couplers (Ssp) in the case of columns with a pair of couplers on each bar. Ssp is the center-to-center distance of couplers. Three coupler lengths (Lsp = 5db, 10db, and 15db), four pedestal heights (Hsp = 5db, 10db, 20db, and 30db), three rigid length factors (β = 0.25, 0.50, 0.75), and two vertical coupler spaces (Ssp = 2Lsp and 4Lsp) were included in the analysis to address all practical combinations of these parameters.

State-of-the-Art Literature Review 87 Overall, 560 analyses were carried out in the study. Figure 2-82 shows a sample result. The spliced column ductility (µsp) was normalized to its counterpart CIP column ductility (µCIP). Therefore, a normalized ductility ratio of 1 or greater indicated that the coupler had no adverse effect on the displacement ductility capacity. It was found that larger couplers, couplers closer to the column–footing interface, and more rigid couplers (e.g., GCs) significantly reduced the displacement ductility capacity of the spliced column. In other words, coupler length (Lsp), ped- estal height (Hsp), and the coupler rigid length factor (β) are the most critical parameters that affect the displacement ductility capacity of mechanically spliced columns, but the effect of other parameters, such as the axial load index or the aspect ratio, is minimal. Furthermore, the results showed that mechanical bar splices can reduce the displacement ductility capacity of the spliced column by up to 40% when very rigid, very long couplers are installed immediately above the column–footing interface. Shifting the couplers from the column–footing interface by one-half the column diameter (0.5Dc) significantly improved the displacement ductility capacity of the spliced columns and made them comparable to CIP columns. Another finding was that the coupler effect was more profound on columns with higher ductility. For example, the displacement ductility of a spliced column is expected to be 95% of the CIP column displacement ductility when the CIP displacement ductility is 3. However, this ratio is 85% when the CIP displacement ductility is 7. It was also found that the ductility of columns with two couplers on a reinforcing bar was approximately the same as the ductility of single-level coupler columns when the couplers are vertically spaced at least 2Lsp on center (or 1.0Lsp face to face). Finally, a simple design equation was developed on the basis of the parametric study to determine the displacement ductility of a spliced column relative to that of a column with no splices: 1 0.18 (2-8) sp CIP sp sp 0.1H L ( )µ µ = − β     β where µsp is the spliced column displacement ductility and µCIP is the nonspliced CIP column displacement ductility. Hsp can be taken as 0.1 in. (2.5 mm) for columns in which the couplers are installed immediately above or below the column–adjoining member interface. The equation was verified through comparison with data from three large-scale column tests. 2.6.3 Grouted Duct Connections No analytical studies regarding the seismic performance of columns with different GD con- nections (e.g., different embedment lengths, duct fillers, and duct properties) were found in the literature. However, the bond behavior of bars in GDs was the subject of a few analytical studies (Raynor et al. 2002; Tazarv and Saiidi 2014). Since this type of connection is moment resisting, previous studies simply assumed a fixed boundary condition in analyses and investigated only the overall column behavior. 2.6.4 Pocket and Socket Connections Pocket connections are considered viable for ABC utilizing concrete members. No analyti- cal studies investigating RC column pocket connections were found in the literature. How- ever, pocket connections have been extensively utilized in steel structures. A summary of analytical studies on the performance of steel columns embedded in footings is presented here for completeness.

Source: Tazarv and Saiidi (2015a). (a) Ductility = 3, a = 0.25 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.5 0.6 0.7 0.8 0.9 1 1.1 0 5 10 15 20 25 30 35 Hsp/Dc D uc ti lit y R at io ( l s p/ l R C ) Hsp/db 5 10 15 Target µCIP = 3.0 Axial Load Index = 5% Coupler Rigid Length Factor (a) = 0.25 Lsp /db Average Deviation = 2% (b) Ductility = 3, a = 0.5 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.5 0.6 0.7 0.8 0.9 1 1.1 0 5 10 15 20 25 30 35 Hsp/Dc D uc ti lit y R at io (l sp / l C IP ) Hsp/db 5 10 15 Target µCIP = 3.0 Axial Load Index = 5% Coupler Rigid Length Factor (a) = 0.5 Lsp /db Average Deviation = 2% (c) Ductility = 3, a = 0.75 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.5 0.6 0.7 0.8 0.9 1 1.1 0 5 10 15 20 25 30 35 Hsp/Dc D uc ti lit y R at io ( l s p/ l C IP ) Hsp/db 5 10 15 Target µCIP = 3.0 Axial Load Index = 5% Coupler Rigid Length Factor (a) = 0.75 Lsp /db Average Deviation = 2% (d) Ductility = 5, a = 0.25 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.5 0.6 0.7 0.8 0.9 1 1.1 0 5 10 15 20 25 30 35 Hsp/Dc D uc ti lit y R at io (l sp /l C IP ) Hsp/db 5 10 15 Target µCIP = 5.0 Axial Load Index = 5% Coupler Rigid Length Factor (a) = 0.25 Lsp /db Average Deviation = 3% (e) Ductility = 5, a = 0.5 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.5 0.6 0.7 0.8 0.9 1 1.1 0 5 10 15 20 25 30 35 Hsp/Dc D uc ti lit y R at io ( l s p/ l C IP ) Hsp/db 5 10 15 Target µCIP = 5.0 Axial Load Index = 5% Coupler Rigid Length Factor (a) = 0.5 Lsp /db Average Deviation = 6% (f) Ductility = 5, a = 0.75 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.5 0.6 0.7 0.8 0.9 1 1.1 0 5 10 15 20 25 30 35 Hsp/Dc D uc ti lit y R at io ( l s p/ l C IP ) Hsp/db 5 10 15 Target µCIP = 5.0 Axial Load Index = 5% Coupler Rigid Length Factor (a) = 0.75 Lsp /db Average Deviation = 8% (g) Ductility = 7, a = 0.25 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.5 0.6 0.7 0.8 0.9 1 1.1 0 5 10 15 20 25 30 35 Hsp/Dc D uc ti lit y R at io ( l s p/ l C IP ) Hsp/db 5 10 15 Target µCIP = 7.0 Axial Load Index = 5% Coupler Rigid Length Factor (a) = 0.25 Lsp /db Average Deviation = 3% (h) Ductility = 7, a = 0.5 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.5 0.6 0.7 0.8 0.9 1 1.1 0 5 10 15 20 25 30 35 Hsp/Dc D uc ti lit y R at io ( l s p/ l C IP ) Hsp/db 5 10 15 Target µCIP = 7.0 Axial Load Index = 5% Coupler Rigid Length Factor (a) = 0.5 Lsp /db Average Deviation = 5% (i) Ductility = 7, a = 0.75 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.5 0.6 0.7 0.8 0.9 1 1.1 0 5 10 15 20 25 30 35 Hsp/Dc D uc ti lit y R at io ( l s p/ l C IP ) Hsp/db 5 10 15 Target µCIP = 7.0 Axial Load Index = 5% Coupler Rigid Length Factor (a) = 0.75 Lsp /db Average Deviation = 8% Figure 2-82. Effect of coupler length on ductility of columns with the aspect ratio = 6 and the axial load index = 5% (sample).

State-of-the-Art Literature Review 89 Pertold et al. (2000) tested six full-scale steel columns connected to the footings by pocket connections. Three column tests studied the bond behavior between the steel columns and the concrete footing, and the remaining tests were focused on the punching capacity of the connection. A finite element model of the connection was subsequently developed and verified through the test data. The effect of column embedment length (one, two, or three times the steel profile depth) was studied (Figure 2-83). It was reported that full column moment capacity could be developed when the embedment length was greater than the diameter/side dimension of the column section. Furthermore, a simple design equation was presented to estimate the required steel column embedment length in this type of connection. Pecce and Rossi (2013) investigated the effect of the embedment length of concrete-encased steel columns (composite sections) on the column’s overall behavior. An experimental inves- tigation was carried out first; then the test data were used to calibrate three-dimensional finite element models. The test data confirmed that columns sufficiently embedded into the footing exhibited larger displacement capacities as compared with columns connected to footings by bolted base plates, which is the conventional steel column base connection. Subsequently, parametric study was carried out to determine proper column embedment lengths for dif- ferent steel sections. Figure 2-84 shows force-displacement relationships of columns with a Source: Pecce and Rossi (2013). Figure 2-84. Effect of embedment length of composite steel–concrete column on force-displacement. Source: Pertold et al. (2000). Figure 2-83. Effect of steel column embedment length.

90 Proposed AASHTO Seismic Specifications for ABC Column Connections European steel section HEB 300 (height of section = 300 mm) but with different embedment lengths. It can be seen that the columns with an embedment length of 1.5 of the profile height exhibited large moment and displacement capacities, while the columns with shorter embed- ment lengths failed as a result of connection failure. 2.6.5 Pipe-Pin Connections Zaghi and Saiidi (2010) analytically investigated the effect of different top pipe-pin connec- tion parameters, such as the friction coefficient, the pipe embedment length, the pipe thickness, and the ring plate thickness, on the connection performance. Overall, 93 detailed finite element analyses were carried out (Figure 2-85). Then the findings were compiled as a design guideline for this type of connection. Mehrsoroush and Saiidi (2014) conducted a parametric study to investigate the behavior of base pipe-pin connections. Many parameters, such as the outer pipe height, the number of studs, the hinge gap thickness, and different tensile members, were studied by using detailed finite element models (Figure 2-86). It was found that the height of outer pipe has a negligible effect on the connection performance, that the shear studs should be adequately distributed along the length of the pipe, and that increasing the thickness of the hinge gap substantially reduces the tensile forces in the tensile element. The findings of the parametric study as well as the experimental program were integrated into a design guideline to facilitate field application of this connection type. 2.6.6 Summary of Analytical Studies There are only a few analytical studies that investigated the performance of ABC column connections. The studies available typically focused on the overall behavior of columns. Nonetheless, useful information was obtained from these studies and summarized in this section. Source: Zaghi and Saiidi (2010). Figure 2-85. Detailed finite element modeling of top pipe-pin connections.

State-of-the-Art Literature Review 91 Source: Mehrsoroush and Saiidi (2014). Figure 2-86. Detailed finite element modeling of base pipe-pin connections. 2.7 Issues to Address Before Field Deployment of Precast Bridge Columns The previous sections of this chapter have presented a state-of-the-art literature review of four types of connections. Three of these—mechanical bar coupler connections, GD connections, and pocket and socket connections—were selected by the project panel for further investigation and development of guidelines. This section presents a summary of the knowledge gaps identified for these three precast connections, the seismic performance of the connections, and issues related to the constructability, maintenance, and the speed of construction. 2.7.1 Knowledge Gaps 2.7.1.1 Knowledge Gaps for Mechanical Bar Splices Knowledge gaps for bar coupler connections can be categorized as follows: 1. Gaps regarding the performance of mechanical bar couplers, 2. Gaps regarding the seismic performance of mechanically spliced columns, and 3. Specifications, testing, and construction requirements of mechanical bar splices. Mechanical Bar Splices. A summary of the performance of five coupler types is presented in Tables 2-2 (SSCs), 2-4 (HCs), 2-6 (grouted sleeve couplers), 2-8 (TCs), and 2-10 (swaged couplers). The available test data are not conclusive and are sometimes contradictory. Knowl- edge gaps identified for mechanical bar splices are as follows: 1. A lack of a unified testing protocol for bar couplers. California Test 670 (California Depart- ment of Transportation 2004) for couplers may be adopted but needs modifications to allow different coupler sizes and to include strain data on the coupler region and bar region in the testing. Cyclic and dynamic loading protocols should be established.

92 Proposed AASHTO Seismic Specifications for ABC Column Connections 2. Because the current AASHTO and Caltrans design documents ban couplers in plastic hinges of columns in moderate- and high-seismic zones (Table 2-1), the current acceptance criteria for couplers should be revised, because the current criteria are not intended for plastic hinge use. Tazarv and Saiidi (2015a) developed the following acceptance criteria for bar couplers to be used in column connections subject to earthquake loading: – The length of the bar coupler shall not exceed 15db (db is the smaller of the spliced bar diameters), and – The spliced bar(s) shall fracture outside the coupler region at a minimum of 1.0db away from each end of the coupler. These or other acceptance criteria are yet to be included in design codes. 3. A material model is needed for bar couplers. As was presented, Tazarv and Saiidi (2015a) pro- posed a generic material model for couplers. On the basis of this model, the full stress–strain relationship of the coupler can be established when the coupler rigid length factor (β) is known. This model or other material models for couplers is yet to be included in design codes. 4. A comprehensive monotonic, cyclic, and dynamic tensile testing method for mechanical bar splices that uses a unified loading protocol (Gap No. 1) and new acceptance criteria (Gap No. 2) are needed to conclude the suitability of different coupler types for precast column connections and to establish the material model properties of couplers (Gap No. 3). The literature search clearly shows that only a limited number of available coupler types are suitable for precast column connections in moderate- and high-seismic zones. Coupler types such as compression-only couplers, tension-only couplers, very long couplers, and some of the SSCs are not acceptable for column connections, as test data have shown that their cyclic performance is not satisfactory. Mechanically Spliced Bridge Columns. Tables 2-3 (SSCs), 2-5 (HCs), 2-7 (grouted sleeve couplers), 2-9 (TCs), and 2-11 (swaged couplers) present the seismic performance of mechani- cally spliced column test models with different coupler types. It can be concluded that the coupler length, location, and rigidity and the number of couplers on a single bar are the most important parameters that affect the seismic performance of columns. Twenty-five mechanically spliced columns incorporating different coupler types have been tested. The available column test data are not sufficient to establish with confidence the effect of the aforementioned parameters on the seismic performance of columns with couplers. More than 550 analyses were performed by Tazarv and Saiidi (2015a) to analytically establish the effect of these parameters. Therefore, although some of the identified gaps have been addressed by ana- lytical studies, a larger pool of experimental and analytical data is needed to reliably quantify the coupler effect on the seismic performance of columns. For example, a few studies have proposed debonding a portion of column longitudinal bars near couplers to improve the column’s seis- mic performance, especially in the case of GC columns. However, the effect of the debonded bar length and location on the column performance has not been quantified, either experimentally or analytically. 2.7.1.2 Knowledge Gaps on Grouted Duct Connections The knowledge gaps for GD column connections can be categorized at two levels: (1) the connection level, where gaps exist regarding embedment length design equations, and (2) the component level, where the gap is the lack of test data on the seismic performance of columns with different GD connection detailing. Design Embedment Length. Figure 2-87 shows all possible modes of failure for a bar anchored in a GD connection under tensile loading. The bar fractures outside the connection if the embedment length is sufficient (Figure 2-87b). In contrast, the bar pulls out when the embedment length is insufficient (Figure 2-87c). The pullout may be associated with a part

State-of-the-Art Literature Review 93 of the grout also failing if the grout is not sufficiently strong (Figure 2-87d). A different failure mode occurs if the bond at the grout–duct interface or the duct–concrete interface is weak and leads to the duct pulling out (Figure 2-87e). When the concrete around the duct is not sufficiently strong to resist the ultimate bar forces, it fails in a conical shape (Figure 2-87f). Another mode of failure could be expected when the bar or the duct pulls out because of the splitting failure of the grout or the concrete, respectively (Figure 2-87g). This is an unlikely mode of failure for GD connections, because confinement provided by the duct and the relatively large concrete cover on the duct would prevent this failure mode. The modes of failure for GD connections indicate that two bond surfaces and bond average strengths should be considered: (1) bar bond strength, which is defined as the ratio of the peak bar tensile force to the surface area of the bar (conventional bond strength), and (2) duct bond strength, which is defined as the ratio of the peak bar tensile force to the surface area of the duct (Tazarv and Saiidi 2014). Table 2-12 summarizes all available test data regarding duct connection bond behavior. It can be concluded that the available design equations for duct connections filled with conventional grouts are not reliable, since the equations do not include the duct bond strength. In the studies leading to those equations, the effects of the grout strength, the concrete strength, and the duct diameter were excluded. The study by Tazarv and Saiidi (2014) investigated the effect of the duct size on the bond behavior of UHPC-filled duct connections. It was found that the duct diameter is an important parameter that has to be included in design equations. The effect of duct diameter on the bond strength of duct connections filled with normal-strength grouts was not studied and is yet to be investigated. Furthermore, the effect of the bundling of bars in normal-strength GD connections is also unknown and has to be studied. Cyclic loading usually affects the bond behavior of an anchored bar adversely. Test data on the bond strength of GD connections under cyclic actions are currently limited. In summary, the main knowledge gap regarding normal-strength GD connections is the lack of a comprehensive design equation that includes the simultaneous effect of the duct and bar bond strengths as well as the effects of the duct wall thickness, bundled bars, and the duct group versus single ducts. The experimental program should include monotonic and cyclic loading regimes. (a) (b) (c) (d) (e) (f) (g) Bar Duct Pullout, Concrete Conical Failure Plan View Duct/Bar Pullout, Splitting Failure Duct Pullout, Surrounding Concrete Failure Bar Pullout, Surrounding Grout Failure Bar Fracture D uc t Concrete Bar Pullout, Grout Mass Failure Pullout Specimen Source: Tazarv and Saiidi (2014). Figure 2-87. Modes of failure in grouted duct connections.

94 Proposed AASHTO Seismic Specifications for ABC Column Connections Columns with Grouted Duct Connections. Table 2-13 presents a summary of the seismic performance of column test models that incorporate grout-filled duct connections. Ducts are usu- ally installed in the members to which the columns are connected. Because the connection is outside the column plastic hinge, the effect of this type of connection on the column behavior is minimal if column longitudinal bars are sufficiently anchored. Therefore, the main knowledge gaps regard- ing GD columns are those mentioned for the connection itself. With respect to column test data, past research has addressed the seismic behavior of precast columns connected to cap beams using normal-strength GDs. However, there are no test data on the seismic behavior of precast columns connected to footings by utilizing normal-strength grouted duct connections. This is especially important because (1) the depth of the footing or pile cap is usually shallower than the cap beam depth, which limits the available anchorage zone, and (2) the column plastic hinge rotation and damage at the column base are usually higher than those in the column top plastic hinge region because of the higher rigidity of footings as compared with cap beams. When UHPC is used as a filler material in GD connections, a part of the column longitudinal bars is debonded to avoid strain concentration, since UHPC significantly increases the bond strength. When GD connections are used in the column–pile shaft or column–integral cap beam connec- tions, the minimum size of the pile shafts and cap beams is critical to fully transfer biaxial loads. 2.7.1.3 Knowledge Gaps on Pocket and Socket Connections A summary of the seismic performance of column test models with pocket/socket connec- tions is presented in Table 2-14. The bulk of the available test data on pocket connections is more extensive than the data for other ABC column connections. The available data can help identify the important parameters that affect the connection and column behavior. The knowl- edge gaps regarding pocket connections are (1) the minimum dimensions of column adjoining members, so as to resist uniaxial and biaxial loading, and (2) the tensile capacity of the con- nection. Specifically, the effect of flexural loading resulting from the transverse loading of the bridge combined with torsion and flexure resulting from the longitudinal loading of the bridge is not known. Resisting this combination could require cap beam widths that exceed those in CIP construction. The tensile capacity of the connection could be important for multicolumn bents because of the frame action and when the vertical component of the earthquake is strong. 2.7.2 Constructability The ease of construction is an important parameter for widespread acceptance of a new concept or detail for ABC. It is known that constructability can be addressed through proper detailing; early communication between designers, contractors, and fabricators; and extra care. This section summaries the constructability issues for the selected precast column connections. 2.7.2.1 Mechanically Spliced Columns Five common types of couplers are shown in Figure 2-1. Each coupler type is unique with respect to physical features and installation procedure. For example, in an SSC, bars are inserted into a sleeve, and then the screws are torqued to a specified level. The size of the sleeve, the number of screws, and the torque magnitude vary, depending on the manufacturer. In a grouted sleeve coupler, the bars are anchored by filling the sleeve with nonshrink, high early strength grout. Other couplers have their own installation procedure. The manufacturer guidelines must be followed to ensure correct installation of splices. Clear cover for sections with couplers shall be based on design codes, such as the AASHTO LRFD or AASHTO seismic guide specifications (AASHTO 2013, 2014, respectively). Thus the clear cover on the mechanically spliced bars is generally thicker than that in typical CIP

State-of-the-Art Literature Review 95 members. The minimum and maximum clear distances between the mechanical splices are recommended to be the same as those specified for reinforcing steel bars. That is, large-diameter couplers may be staggered to meet the bar distance requirements. Another case for staggering is when a special tool that is needed to complete the splice. For example, the press used in swaged couplers is bulky and would not fit between two couplers to complete the splice. Table 2-16 presents tools that are needed to complete different mechanical bar splices. Table 2-16 also presents other construction considerations that must be reviewed by the engineer prior to field application. The end of the spliced bars shall be prepared in HCs and TCs. Heading can be done off- or on-site, but in situ threading of bars for repair is generally hard to do and may be impractical. The type, size, and grade of screws in SSCs must be verified and specified by the manufacturer. Grout in the GCs shall be provided by the coupler manufacturer only, and conventional grout or grout from other coupler manufacturers should not be used; otherwise, the manufacturer’s warranty could be voided. Some of the coupler types, such as HCs or TCs, impose very tight constructional tolerances. 2.7.2.2 Grouted Duct Connections The main construction challenge for GD connections is the alignment of column longitudi- nal bars with respect to the ducts placed in the column adjoining members. Therefore, the duct size should be sufficiently large to provide proper bond strength as well as good tolerance. Furthermore, the flow of filler grout is a critical parameter in which only high-flow grouts should be used. Coarse aggregate–based grout should not be used because of its low flow. Gravity treme-tube technique can be used to fill the ducts without trapping the air. Proper sealing is needed when the ducts are used in cap beams. 2.7.2.3 Pocket and Socket Connections As shown in Figure 2-51, there are several configurations for columns with pocket connections. For example, columns can be either partially precast, in which the column bars are extended into the pockets, or fully precast, in which columns are extended into the pockets. Footings can be cast after the placement of columns or can be built with a pocket to form the connection after the footing concrete hardens. Cap beam longitudinal bars can be grouped around the pocket or can pass through the pocket. The former option is significantly easier to construct because it avoids interference between the column and the cap beam bars. Column transverse reinforce- ment may extend into the pocket, or the corrugated pipe can be designed to resist joint shear. In connections with partially precast columns and cap beam bars passing through the joint, the corrugated pipe eliminates the need for extension of the column transverse bars into the pocket and thus alleviates congestion of reinforcement. Item/Coupler Shear Screw Headed Bar Grouted Sleeve Threaded Swaged Bar end preparation Not needed Heading Not needed Threading Not needed Special equipment Wrench or nut runner Wrench, heading machine Grout pump Die and tap Press machine Additional material/piece Screw No need Grout/sealing No need No need Tolerance and alignment Loose Tight Loose Tight Loose Field erection speed for precasting Very fast Fast Very fast Fast Fast Time to complete one splice 1 min. 24 hours 5 min. 5 min. 5 min. Table 2-16. Constructability of mechanical bar couplers.

96 Proposed AASHTO Seismic Specifications for ABC Column Connections Depending on the pocket and column detailing, the construction sequence may be slightly different. Providing sufficient tolerance for the installation of precast cap beams is particularly critical in multicolumn bents. When fully precast columns are used in the pocket connection, the ceiling of the pocket must be sloped slightly to avoid air entrapment, and a high-flow grout must be used to fill the gap between the column and the pipe. 2.7.3 Maintenance In general, the maintenance of precast members may be more involved than the mainte- nance of conventional CIP members, owing to cold joints between prefabricated members. Nevertheless, the maintenance of the ABC column connections discussed in this document is expected to be the same as that for conventional CIP columns, as long as the minimum cover is provided for mechanical bar couplers, corrugated ducts, and corrugated pipes. 2.7.4 Speed of Construction and Cost Studies by Marsh et al. (2011) and Tazarv and Saiidi (2015a, 2015b) have shown that bents with ABC column connections can be built two to four times faster than CIP bents. The over- all cost of bridges built with precast columns (and other precast elements) is expected to be lower than that of conventional CIP bridges. Inevitably, the speed of construction is expected to increase as designers and contractors gain more experience in ABC. 2.7.5 Summary Knowledge gaps that must be addressed for the successful development of comprehen- sive design and construction guidelines for three type of precast column connections were reviewed in the previous sections. The main gaps were found to be the performance of either the connection itself or column–footing and column–cap beams connections. Issues related to constructability, maintenance, speed, and costs were briefly discussed. Table 2-17 presents the critical parameters that affect the performance of each precast column connection as well as the performance of columns incorporating different ABC connections spaG egdelwonK sretemaraP lacitirC epyT noitcennoC Mechanical bar splices Coupler properties (e.g., rigid length factor) and strain rate effect Acceptance criteria with monotonic, cyclic, and dynamic loading protocols, coupler material model, and coupler rigidity Mechanically spliced column connections Coupler length, coupler location, coupler rigid length factor, bar debonding effect, and plastic hinge length Effect of coupler length, coupler location, coupler rigidity, and bar debonding on plastic hinge length and column ductility Grouted duct design equation Bar size, bar embedment length, duct material, duct size, grout material, bar bundling, bar eccentricity in ducts, cyclic bond strength, and number of ducts Design equations that include all of the critical parameters Grouted duct column connections Design embedment length and minimum size of integral cap beam cross sections and pile shafts for biaxial loading Design embedment length, minimum size of integral cap beam cross sections, footings, and pile shafts to resist biaxial loading Pocket/socket connections Design embedment length, minimum size of column adjoining members, biaxial loading, pocket diameter, column section (circular or rectangular), and tensile load transfer mechanism and capacity Minimum size of integral cap beam cross sections, footings, and pile shafts to resist biaxial loading, and tensile load transfer mechanism and capacity Table 2-17. Knowledge gaps regarding column connections in accelerated bridge construction.

State-of-the-Art Literature Review 97 to footings and cap beams. The table also includes the critical parameters that need further investigation for the development of guidelines. 2.8 References AASHTO. (2013). AASHTO LRFD Bridge Design Specifications. American Association of State Highway and Transportation Officials, Washington, D.C. AASHTO. (2014). AASHTO Guide Specifications for LRFD Seismic Bridge Design. American Association of State Highway and Transportation Officials, Washington, D.C. ACI Committee 318. (2014). ACI 318: Building Code Requirements for Reinforced Concrete. American Concrete Institute, Detroit, Mich. ACI Committee 439. (2007). ACI 439.3R-07: Types of Mechanical Splices for Reinforcing Bars. American Concrete Institute, Detroit, Mich. Alam, M. S., Youssef, M. A., and Nehdi, M. L. (2010). Exploratory Investigation on Mechanical Anchors for Connecting SMA Bars to Steel or FRP Bars. Materials and Structures, Vol. 43, pp. 91–107. Ameli, M. J., Parks, J. E., Brown, D. N., and Pantelides, C. P. (2015). Seismic Evaluation of Grouted Splice Sleeve Connections for Reinforced Precast Concrete Column-to-Cap Beam Joints in Accelerated Bridge Construction. PCI Journal, Vol. 60, No. 2, pp. 80–103. Banks, G. A., Parrish, M., and Spry, C. W. (2015). Replacing the Boeing North Bridge. PCI Journal, May–June, pp. 29–38. Belleri, A., and Riva, P. (2012). Seismic Performance and Retrofit of Precast Concrete Grouted Sleeve Connections. PCI Journal, Vol. 57, pp. 97–109. Brenes, F. J., Wood, S. L., and Kreger, M. E. (2006). Anchorage Requirements for Grouted Vertical-Duct Connectors in Precast Bent Cap Systems. FHWA/TX-06/0-4176-1. Center for Transportation Research, University of Texas at Austin. California Department of Transportation. (2004). Method of Tests for Mechanical and Welded Reinforcing Steel Splices. California Test 670. Division of Engineering Services, Sacramento. California Department of Transportation. (2013). Seismic Design Criteria (SDC). Ver. 1.7. Sacramento. California Department of Transportation. (2014). Memo to Designers 20-9. Sacramento. Cruz-Noguez, C. A., and Saiidi, M. S. (2012). Shake Table Studies of a 4-Span Bridge Model with Advanced Materials. Journal of Structural Engineering, Vol. 138, No. 2, pp. 183–192. Culmo, M. (2009). Connection Details for Prefabricated Bridge Elements and Systems. FHWA-IF-09-010. FHWA, Office of Bridge Technology, HIBT-10. Culmo, M. P. (2011). Accelerated Bridge Construction: Experience in Design, Fabrication and Erection of Prefabricated Bridge Elements and Systems. FHWA-HIF-12-013. FHWA, Office of Bridge Technology, HIBT-10. Doyle, K., and Saiidi, M. (2008). Seismic Response of Telescopic Pipe Pin Connections. CCEER-08-01. Center for Civil Engineering Earthquake Research, Department of Civil Engineering, University of Nevada, Reno. Florida Department of Transportation. (2019). EDC—Prefabricated Bridge Elements & Systems. https:// www.fdot.gov/structures/edc/. Frosch, R. J. (1999). Shear Transfer Between Concrete Elements Using Steel Pipe Connection. ACI Structural Journal, Vol. 96. No. 6, pp. 1003–1008. Galvis, F., Correal, J. F., Betancour, N., and Yamin, L. (2015). Characterization of the Seismic Behavior of a Column-Foundation Connection for Accelerated Bridge Construction. VII Congreso Nacional de Ingeniería Sísmica, Bogotá, Colombia. Haber, Z. B. (2014). Precast Columns with Mechanically Spliced Connections for Accelerated Bridge Construction in Seismic Zones. 30th U.S.–Japan Bridge Engineering Workshop. Haber, Z. B., Saiidi, M. S., and Sanders, D. H. (2013). Precast Column-Footing Connections for Accelerated Bridge Construction in Seismic Zones. UNR/CCEER-13-08. Center for Civil Engineering Earthquake Research, Department of Civil and Environmental Engineering, University of Nevada, Reno. Haber, Z. B., Saiidi, M. S., and Sanders, D. H. (2014). Seismic Performance of Precast Columns with Mechanically Spliced Column-Footing Connections. ACI Structural Journal, Vol. 111, No. 3, pp. 639–650. Haber, Z. B., Saiidi, M. S., and Sanders, D. H. (2015). Behavior and Simplified Modeling of Mechanical Reinforcing Bar Splices. ACI Structural Journal, Vol. 112, No. 2, pp. 179–188. Haraldsson, O. S., Janes, T. M., Eberhard, M. O. and Stanton, J. F. (2013). Seismic Resistance of Socket Connection Between Footing and Precast Column. Journal of Bridge Engineering, Vol. 18, No. 9, pp. 910–919. Hillis, D., and Saiidi, M. S. (2009). Design, Construction, and Nonlinear Dynamic Analysis of Three Bridge Bents Used in a Bridge System Test. CCEER-09-03. Center for Civil Engineering Earthquake Research, Department of Civil and Environmental Engineering, University of Nevada, Reno.

98 Proposed AASHTO Seismic Specifications for ABC Column Connections Huaco, G. (2013). Procedures to Rehabilitate Extremely Damaged Concrete Members Using Innovative Materials and Devices. PhD dissertation. University of Texas at Austin. Huaco, G., and Jirsa, J. (2012). Performance of Damaged Column Retrofitted with Innovative Materials and Devices. 15th World Conference on Earthquake Engineering. Jansson, P. O. (2008). Evaluation of Grout-Filled Mechanical Splices for Precast Concrete Construction. R-1512. Michigan Department of Transportation, Lansing. Kapur, J., Bardow, A., Dekelbab, W., Keever, M., Saiidi, M. S., Sletten, J., Tobias, D., and Yen, W. P. (2013). Best Practices Regarding Performance of ABC Connections in Bridges Subjected to Multihazard and Extreme Events. Scan Team Report, NCHRP Project 20-68A, Scan 11-02. Scan Management, Arora and Associates, P.C., Lawrenceville, N.J. Kavianipour, F., and Saiidi, M. S. (2013). Experimental and Analytical Seismic Studies of a Four-Span Bridge System with Composite Piers. CCEER-13-17. Center for Civil Engineering Earthquake Research, Department of Civil and Environmental Engineering, University of Nevada, Reno. Khaleghi, B., Schultz, E., Seguirant, S., Marsh, L., Haraldsson, O., Eberhard, M. and Stanton, J. (2012). Accelerated Bridge Construction in Washington State: From Research to Practice. PCI Journal, Vol. 57, No. 4, pp. 34–49. Larosche, A., Cukrov, M., Sanders, D., and Ziehl, P. (2014a). Prestressed Pile to Bent Cap Connections: Seismic Performance of a Full-Scale Three-Pile Specimen. Journal of Bridge Engineering, Vol. 19, No. 3. Larosche, A., Ziehl, P., ElBatanouny, M., and Caicedo, J. (2014b). Plain Pile Embedment for Exterior Bent Cap Connections in Seismic Regions. Journal of Bridge Engineering, Vol. 19, No. 4. Lehman, D. E., Gookin, S. E., Nacamuli, A. M. and Moehle, J. P. (2001). Repair of Earthquake-Damaged Bridge Columns. ACI Structural Journal, Vol. 98, No. 2, pp. 233–242. Lloyd, W. R. (2001). Qualification of the Bar-Lock Rebar Coupler for Use in Nuclear Safety–Related Applications Mechanical Testing Program and Performance Analysis. INEEL/EXT-02-01387. Idaho National Engineering and Environmental Laboratory Materials Department, Idaho Falls. Marsh, M. L., Wernli, M., Garrett, B. E., Stanton, J. F., Eberhard, M. O., and Weinert, M. D. (2011). NCHRP Report 698: Application of Accelerated Bridge Construction Connections in Moderate-to-High Seismic Regions. Transportation Research Board of the National Academies, Washington, D.C. Mashal, M., Palermo, A., and Chegini, Z. (2014). Quasi-static Cyclic Tests of Half-Scale Fully Precast Bridge Bents Incorporating Emulative and Posttensioned Low Damage Solutions. Second European Conference on Earthquake Engineering and Seismology, Istanbul, Turkey. Matsumoto, E. E., Waggoner, M. C., Sumen, G. and Kreger, M. E. (2001). Development of a Precast Bent Cap System. FHWA/TX-0-1748-2. Center for Transportation Research, University of Texas at Austin. Mehraein, M., and Saiidi, M. S. (2016). Seismic Performance of Bridge Column-Pile-Shaft Pin Connections for Application in Accelerated Bridge Construction. CCEER-16-01. Center for Civil Engineering Earthquake Research, Department of Civil and Environmental Engineering, University of Nevada, Reno. Mehrsoroush, A., and Saiidi, M. S. (2014). Experimental and Analytical Seismic Studies of Bridge Piers with Innovative Pipe Pin Column-Footing Connections and Precast Cap Beams. CCEER-14-07. Center for Civil Engineering Earthquake Research, Department of Civil and Environmental Engineering, University of Nevada, Reno. Mohebbi, A., Saiidi, M. S., and Itani, A. (2015). Seismic Evaluation of a Precast PT/UHPC Bridge Column with Pocket Connection and Precast Footing. 2015 National Accelerated Bridge Construction Conference, Miami, FL. Motaref, S., Saiidi, M. S., and Sanders, D. (2011). Seismic Response of Precast Bridge Columns with Energy- Dissipating Joints. CCEER-11-01. Center for Civil Engineering Earthquake Research, Department of Civil and Environmental Engineering, University of Nevada, Reno. Nakashoji, B., and Saiidi, M.S. (2014). Seismic Performance of Square Nickel–Titanium Reinforced ECC Columns with Headed Couplers. CCEER-14-05. Center for Civil Engineering Earthquake Research, Department of Civil and Environmental Engineering, University of Nevada, Reno. Noureddine, I. (1996). Plastic Energy Absorption Capacity of #18 Reinforcing Bar Splices Under Monotonic Loading. MS thesis. California State University, Sacramento. O’Brien, M., Saiidi, M. S., and Sadrossadat-Zadeh, M. (2006). A Study of Concrete Bridge Columns Using Innova- tive Materials Subjected to Cyclic Loading. CCEER-06-04. Center for Civil Engineering Earthquake Research, Department of Civil Engineering, University of Nevada, Reno. Ou, Y. (2007). Precast Segmental Post-Tensioned Concrete Bridge Columns for Seismic Regions. PhD dissertation. State University of New York at Buffalo. Pang, J. B. K., Steuck, K. P., Cohagen, L., Stanton, J. F., and Eberhard, M. O. (2008). Rapidly Constructible Large-Bar Precast Bridge-Bent Seismic Connection. WA-RD 684.2. University of Washington, Seattle. Pantelides, C. P., Ameli, M. J., Parks, J. E., and Brown, D. N. (2014). Seismic Evaluation of Grouted Splice Sleeve Connections for Precast RC Bridge Piers in ABC. UT-14.09. Utah Department of Transportation, Salt Lake City.

State-of-the-Art Literature Review 99 Pecce, M., and Rossi, F. (2013). The Nonlinear Model of Embedded Steel–Concrete Composite Column Bases. Engineering Structures, Vol. 46, pp. 247–263. Pertold, J., Xiao, R. Y., and Wald, F. (2000). Embedded Steel Column Bases I: Experiments and Numerical Simula- tion. Journal of Constructional Steel Research, Vol. 56, pp. 253–270. Popa, V., Papurcu, A., Cotofana, D., and Pascu, R. (2015). Experimental Testing on Emulative Connections for Precast Columns Using Grouted Corrugated Steel Sleeves. Bulletin of Earthquake Engineering, Vol. 13, No. 8, pp. 2429–2447. Raynor, D. J., Lehman, D. E., and Stanton, J. F. (2002). Bond-Slip Response of Reinforcing Bars Grouted in Ducts. ACI Structural Journal, Vol. 99, No. 5, pp. 568–576. Restrepo, J. I., and Panagiotou, M. (2005). BART Aerial Guideway Shear Key Tests. Department of Structural Engineering, University of California, San Diego. Restrepo, J. I., Tobolski, M. J., and Matsumoto, E. E. (2011). NCHRP Report 681: Development of a Precast Bent Cap System for Seismic Regions. Transportation Research Board of the National Academies, Washington, D.C. Roddenberry, M. D., and Servos, J. (2012). Prefabricated/Precast Bridge Elements and Systems (PBES) for Off-System Bridges. FDOT-BDK83-977-13. Florida Department of Transportation, Tallahassee. Rowell, S. P., Grey, C. E., Woodson, S. C., and Hager, K. P. (2009). High Strain-Rate Testing of Mechanical Couplers. ERDC TR-09-8. U.S. Army Corps of Engineers, Engineer Research and Development Center, Port Hueneme, Calif. Saiidi, M. S., and Wang H. (2006). Exploratory Study of Seismic Response of Concrete Columns with Shape Memory Alloys Reinforcement. ACI Structural Journal, Vol. 103, No. 3, pp. 436–443. Saiidi, M. S., O’Brien, M., and Sadrossadat-Zadeh, M. (2009). Cyclic Response of Concrete Bridge Columns Using Superelastic Nitinol and Bendable Concrete. ACI Structural Journal, Vol. 106, No. 1, pp. 69–77. Saiidi, M. S., Tazarv, M., Varela, S., Bennion, S., Marsh, M. L., Ghorbani, I., Murphy, T. P. (2017). NCHRP Research Report 864: Seismic Evaluation of Bridge Columns with Energy Dissipating Mechanisms. Volume 1: Research Overview and Volume 2: Guidelines. Transportation Research Board, Washington, D.C. Sritharan, S., Ingham, J., Priestley, M., and Seible, F. (1999). Design and Performance of Bridge Cap Beam/ Column Using Headed Reinforcement and Mechanical Couplers. In Developments of Seismic Steel Reinforcement Products & Systems (J. D. Carroll and R. H. Reiterman, eds.), ACI Special Publication 184, pp. 7–22. Steuck, K. P., Eberhard, M. O., and Stanton, J. F. (2009). Anchorage of Large-Diameter Reinforcing Bars in Ducts. ACI Structural Journal, Vol. 106, No. 4, pp. 506–513. Tazarv, M., and Saiidi, M. S. (2014). Next Generation of Bridge Columns for Accelerated Bridge Construction in High Seismic Zones. CCEER-14-06. Center for Civil Engineering Earthquake Research, Department of Civil and Environmental Engineering, University of Nevada, Reno. Tazarv, M., and Saiidi, M. S. (2015a). Design and Construction of Bridge Columns Incorporating Mechanical Bar Splices in Plastic Hinge Zones. CCEER-15-07. Center for Civil Engineering Earthquake Research, Department of Civil and Environmental Engineering, University of Nevada, Reno. Tazarv, M., and Saiidi, M. S. (2015b). Design and Construction of Precast Bent Caps with Pocket Connections for High Seismic Regions. CCEER-15-06. Center for Civil Engineering Earthquake Research, Department of Civil and Environmental Engineering, University of Nevada, Reno. Tazarv, M., and Saiidi, M.S. (2015c). Low-Damage Precast Columns for Accelerated Bridge Construction in High Seismic Zones. Journal of Bridge Engineering, Vol. 21, No. 3. Tazarv, M., and Saiidi, M. S. (2015d). UHPC-Filled Duct Connections for Accelerated Bridge Construction of RC Columns in High Seismic Zones. Engineering Structures, Vol. 99, pp. 413–422. Thonstad, T., Mantawy, I., Stanton, J. F., Eberhard, M. O., and Sanders, D. H. (2016). Shaking Table Performance of a New Bridge System with Pre-Tensioned, Rocking Columns. Journal of Bridge Engineering, Vol. 21, No. 4. Tran, H. V. (2012). Drilled Shaft Socket Connections for Precast Columns in Seismic Regions. M.S. thesis. Depart- ment of Civil and Environmental Engineering, University of Washington, Seattle. Utah Department of Transportation. (2010). Precast Substructure Elements. http://dot.utah.gov. Varela, S., and Saiidi, M. S. (2014). Dynamic Performance of Novel Bridge Columns with Shape Memory Alloy and ECC. International Journal of Bridge Engineering, Vol. 2, No. 3, pp. 29–58. Varela, S., and Saiidi, M. S. (2016). Resilient Earthquake-Resistant Bridges Designed for Disassembly. CCEER-16-02. Center for Civil Engineering Earthquake Research, Department of Civil and Environmental Engineering, University of Nevada, Reno. Yang, Y., Sneed, L., Saiidi, M. S., and Belarbi, A. (2014). Repair of Earthquake-Damaged Bridge Columns with Interlocking Spirals and Fractured Bars. 14-2179. California Department of Transportation, Sacramento. Yang, Y., Sneed. L. H., Morgan, A., Saiidi, M. S., and Belarbi, A. (2015). Repair of RC Bridge Columns with Interlocking Spirals and Fractured Longitudinal Bars: An Experimental Study. Construction and Building Materials, Vol. 78, pp. 405–420.

100 Proposed AASHTO Seismic Specifications for ABC Column Connections Zaghi, A. E., and Saiidi, M. (2010). Seismic Design of Pipe–Pin Connections in Concrete Bridges. CCEER-10-01. Center for Civil Engineering Earthquake Research, Department of Civil and Environmental Engineering, University of Nevada, Reno. Zaghi, A. E. and Saiidi, M. S. (2011). Bearing and Shear Failure of Pipe–Pin Hinges Subjected to Earthquakes. Journal of Bridge Engineering, Vol. 16, No. 3, pp. 340–350. Zaghi, A. E., Saiidi, M. S., and Mirmiran, A. (2012). Shake Table Response and Analysis of a Concrete-Filled FRP Tube Bridge Column. Composite Structures, Vol. 94, pp. 1564–1574. Zhu, Z., Ahmad, I., and Mirmiran, A. (2006). Seismic Performance of Concrete-Filled FRP Tube Columns for Bridge Substructure. Journal of Bridge Engineering, Vol. 11, No. 3, pp. 359–370. Ziehl, P. H., Caicedo, J. M., Rizos, D., Mays, T., Larosche, A., ElBatanouny, M., and Mustain, B. (2011). Testing of Connections Between Prestressed Concrete Piles and Precast Concrete Bent Caps. Federal Project No. BR26 (011). South Carolina Department of Transportation, Columbia.