1Graduate Student Researcher, Dept. of Civil Engineering, University of Washington, WA 98195 2Graduate Student Researcher, Dept. of Civil Engineering, University of Washington, WA 98195 3Professor, Dept. of Civil Engineering, University of Washington, WA 98195 4Associate Professor, Dept. of Civil Engineering, University of Washington, WA 98195 5Associate Professor, Dept. of Civil Engineering, University of Washington, WA 98195 Johnson Molly, Sloat Dan, Roeder Charles W, Lehman Dawn E, Berman Jeffrey W. Seismic Performance of Concentrically Braced Frame Connections. Proceedings of the 10th National Conference in Earthquake Engineering, Earthquake Engineering Research Institute, Anchorage, AK, 2014.

Tenth U.S. National Conference on Earthquake EngineeringFrontiers of Earthquake Engineering July 21-25, 2014 Anchorage, Alaska 10NCEE

SEISMIC PERFORMANCE OF CONCENTRICALLY BRACED FRAME

CONNECTIONS

Molly Johnson1, Dan Sloat2, Charles W. Roeder3, Dawn E. Lehman4, Jeffrey W. Berman5

ABSTRACT Concentrically braced frames (CBFs) are used as lateral-load resisting systems throughout the US. Special concentrically braced frames (SCBFs) are currently used in regions of high seismicity, but older CBFs (referred to as non-seismic concentrically braced frames or NCBFs) in these regions were designed without ductile detailing requirements currently required for SCBF design. The University of Washington conducted two experimental investigations to evaluate NCBF and SCBF connection performance with funding of NEES and AISC. The first phase of the NCBF testing assessed the inventory of NCBF buildings in seismic regions. The survey resulted in the identification and categorization of a variety of deficiencies according to current seismic provisions. Those deficiencies involving connections were used to inform the planned NCBF experimental and analytical studies. The study also aimed to assess the effectiveness of bolted connections in satisfying ductility requirements for SCBFs. For the second phase of the project, full-scale quasi-static cyclic tests were conducted to evaluate both modern and older detailing practices (from the survey), including the brace-to-gusset plate connection, gusset plate-to-frame and beam-to-column connections. These experiments will improve understanding of NCBF systems and reduce their seismic hazard, while also informing new design of SCBFs. The research results offer potential guidance for updating the ASCE 41 provisions.

Seismic Performance of Concentrically Braced Frame Connections

Charles W. Roeder1, Dawn E Lehman2, Jeffrey W. Berman3, Molly Johnson4, Dan Sloat5

ABSTRACT

Concentrically braced frames (CBFs) are broadly used as lateral-load resisting systems in buildings throughout the US. Current practice in regions of high seismicity is the use of special concentrically braced frames (SCBFs). Older buildings in high seismic regions also used CBFs; however, prior to modern seismic codes, braced frames were designed without ductile detailing requirements. They are referred to as non-seismic concentrically braced frames (NCBFs) herein. The University of Washington conducted two concurrent sets of experimental testing to evaluate CBF connections in both NCBFs and SCBFs with the support of NEES and AISC. The first phase of the NCBF testing assessed the infrastructure of NCBF buildings on the west coast. The survey resulted in the identification and categorization of a variety of deficiencies according to current seismic provisions. Those deficiencies involving connections were used to inform the planned NCBF experimental and analytical studies. The study also aimed to assess the effectiveness of bolted connections in satisfying ductility requirements for SCBFs. For the second phase of the project, full-scale quasi-static cyclic tests were conducted which involved the evaluation of both modern and older detailing practices (from the survey), including the brace-to-gusset plate connection, gusset plate-to-frame and beam-to-column connections. These experiments will improve understanding of NCBF systems and the risk that they present, while also informing new design of SCBFs. The presentation will provide the experimental results with an eye towards informing structural engineers and updating the ASCE 41 provisions.

Introduction Concentrically braced frames (CBFs) have been a common choice for lateral load resisting systems in steel buildings for many decades in the United States. A typical CBF consists of diagonal braces attached to beams and columns using gusset plate connections. Their continued popularity makes CBFs a strong candidate for further research to improve their performance and design. Special concentrically braced frames (SCBFs) are a subset of CBFs that are typically utilized in areas with high seismic hazards. As specified in the current AISC Seismic Provisions [1], SCBFs ensure system ductility through performance and detailing requirements. The system’s yield mechanism is the brace tensile yielding and compressive buckling. Brace connections and surrounding framing components are designed to ensure the system can sustain the desired yield mechanism’s deformation and force demands, preventing undesirable limit states. Before publication of the 1988 UBC [2], braces, framing members, and connections of braced frames were designed to calculated seismic lateral loads, without consideration for brace overstrength or component ductility demands. These older CBF systems are identified as non-seismic concentrically braced frames (NCBFs) in this paper, and their seismic performance may be controlled by limit states other than brace yielding and buckling. Many of the possible controlling limit states do not offer significant ductility, and could result in rapid loss of system

lateral resistance and inelastic deformation capacity. Systems with these limitations present a significant concern. Seismic evaluation and retrofit of NCBFs is currently an important engineering task. However, knowledge of the seismic performance of NCBFs is limited because their behavior has not been a major focus of academic research. This paper presents research conducted at the University of Washington to improve the understanding of NCBF performance, and to develop improved strategies for seismic retrofit of NCBF systems. Initially, a survey of older CBFs in high seismic regions in the United States was conducted to investigate the framing and connection configurations common in NCBFs and to evaluate the expected performance of the braces, connecting elements, and framing members. The results of the survey gave direction to a series of experiments investigating the performance of single brace, single bay CBF frames.

Infrastructure Review Fourteen buildings designed prior to 1988 with NCBF lateral load resisting systems in high seismic regions of the US were selected, typical braced frame bays from each building were evaluated using current SCBF design criteria to determine the predicted deficiencies of the brace, connections, and framing members. Characteristics of Buildings Surveyed In the selected structures, the most common used brace cross-section was Hollow Structural Sections (HSS), found in more than 70% of the buildings surveyed. Other brace sections included wide flanges, pipes, and angles. The bracing configuration varied significantly, and most structures used multiple configurations. Chevron or inverted-V configurations were present in 70% of buildings. This brace configuration was especially common in buildings less than 5 stories in height. Single diagonal braces, with an opposing brace in an adjacent bay, were the predominant system for structures over 8 stories. The connections used in these NCBF systems also had substantial variability as illustrated in Fig. 1.

Figure 1. Samples of Existing Connections from Survey

Connection and Frame Analysis

The evaluation of the typical braced bays was based upon the ability to develop the expected brace tensile and compressive capacity of the brace as currently required in SCBF design criteria. For this analysis, the ratios between nominal and expected yield and ultimate

(a) (b) (c) (d)

strengths, Ry and Rt, were applied to establish expected brace yielding and buckling capacities, and other elements were evaluated based upon their nominal properties as required in current SCBF design. Resistance factors were not applied to the capacities for any limit states investigated to permit evaluation of the expected performance of the connections and other elements.

For analysis of the frames, all bays above and below the selected bay that contained braced frames were included. All joints were modeled as pinned, and therefore only truss action was considered. The frames were evaluated against four load cases. Lateral load was applied in the plane of the frame in the left-to-right direction for two load cases, and in the right-to-left direction in the other two. For frames not near the edges of the building, half of the load was applied to each of the beam-column intersection nodes of that floor. For frames at the edges of buildings, the entire lateral load was applied to the interior node of the frame. This load distribution is consistent with the expectation of a load path running from the diaphragm to collecting elements and then into the frame.

The other load case variable was brace bucking. In two load cases, the braces in tension were assumed to carry their expected tensile capacity, and the braces in compression were assumed to carry their expected compressive capacity. In the other load cases, the compression braces were instead assumed to carry 30% of their expected compressive capacity, as is required for the analysis of chevron configuration frames. A uniform gravity dead load of 100 psf was applied to each floor of the structure, and the portion of this load in the tributary area of the columns of the braced frame was applied to the column at that floor. Observed Deficiencies

Framing members rarely had sufficient capacity to develop the brace force for all the applied load cases. Figure 2 shows that over 75% of frames had beams with inadequate flexural capacity. This deficiency was a result of weak floor beams in chevron configurations, which led to a predicted failure in all those systems. The unbalanced load case, with braces at 30% of buckling capacity, created flexural demands that were not considered in braced frame design before 1988. In many cases, the columns were also not adequately sized to develop the capacity of the braces.

Figure 2. Demand-Capacity Ratios for Frame Limit States Connections had several limit states with Demand-Capacity Ratios (DCRs) greater than 1

when compared to the brace expected capacity, with a few failing nearly all of the design checks (see Fig. 3). The lack of net section reinforcement made tensile fracture a common concern, though the DCR values were not significantly over 1. Additionally, a number of weld and bolt fracture limit states were frequently deficient, raising concern over brittle connection failures.

0 20 40 60 80 100

Beam Bending

Column Compression

Percentage of Frames FailingFram

e Li

mit

St

ate

1.5<DCR

1.2<DCR<1.5

1.0<DCR<1.2

Figure 3. Demand-Capacity Ratios for Connection Limit States

Experiments at the University of Washington

The results of the survey and analysis of existing buildings was used as guidance for the development of a series of a test program and for establishing a more comprehensive understanding of the behavior of classes of connections. Experimental Setup

The experiments at the University of Washington consist of a single bay, single diagonal brace configuration (see Fig. 4). Lateral frame displacement is applied cyclically and quasi-statically with gradually increasing magnitudes. Simulated axial load is applied to the columns using post-tensioned steel rods. Out-of-plane restraint is provided at several locations to more accurately model restraint provided by the floor slab and perpendicular structural bays.

Figure 4. Dimensions of NCBF 0 (a) Frame and (b) Connection

0 20 40 60 80 100

Bolt Bearing at Beam-Gusset PlateBolt Bearing at Column-Gusset Plate

Block Shear at Beam-Gusset PlateBolt Shear at Colum-Gusset Plate

Block Shear at Column-Gusset PlateBolt Shear at Beam-Gusset

Weld Fracture at Column-Gusset PlateBrace Block Shear

Whitmore FractureBolt Shear at Brace-Gusset

Block Shear of Gusset Plate at BraceGusset Plate Shear at Column-Gusset

Gusset Plate BucklingGusset Plate Shear at Beam-Gusset

Weld Fracture at Brace-Gusset PlateWeld Fracture at Beam-Gusset Plate

Whitmore YieldingBrace Net Section Fracture

Percentage of Frames Failing

Conn

ecti

on L

imit

Sta

te

1.5<DCR

1.2<DCR<1.5

1.0<DCR<1.2

NCBF 0 – Pilot Test

The first NCBF test [3] was conducted in 2012 (prior to the infrastructure review) as a pilot to demonstrate concerns associated with the performance for braced frames. The connection used a pair of double angles to connect the beam web and the gusset plate to the column flange, shown in Fig. 5. Welds on the specimen complied with AWS E71T-8, a high toughness material less vulnerable than the weld material more commonly used in existing NCBF systems. The connection was designed in accordance with the 1988 UBC [2].

Fig. 5a shows the applied deformation history, and Fig. 5b shows the hysteretic response

with annotations of significant points, described below. Initial buckling of the brace was observed at 0.36% story drift with a compressive load of 182 kips. At 0.44% story drift, cracking initiated in the welds between the gusset plate and the brace. The crack lengths increased in subsequent cycles, resulting in fracture of the welds at 0.52% story drift (Fig. 5c). This test confirmed concerns with NCBF connection performance, particularly ductility.

Figure 5. NCBF 0 (a) Load History (b) Hysteretic Response, and (c) Connection Fracture NCBF 1 – Welded Shared Shear Tab

NCBF 1 was the first test based on the results from the survey. An HSS7x7x1/4 brace was used, which has a significantly higher width-thickness ratio than allowed by current SCBF criteria, but was typical of NCBF structures. The thin gusset plate (3/8”) and short splice length (9”) made gusset plate Whitmore yielding and shear yielding design concerns. No clearance was provided for end rotation of the brace. The beam flange was coped on one side to allow the placement of a single shear tab, which was fillet welded continuously to both the beam web and the gusset plate. The tab was welded to the column flange (Fig. 6a).

Figure 6b shows the hysteretic behavior of NCBF 1 and Fig. 6c shows brace hinging. Initial brace buckling was observed at 0.22% drift. At 0.51% drift in compression, a hinge formed in the brace 1 foot northeast of the brace center. Despite the lack of end clearance for the brace, the connection was capable of sustaining brace out-of-plane deflections as large as 10 inches. The connection suffered minimal damage, despite the design expectation that the connection would not develop the brace capacity. This test confirmed concerns associated with the performance of non-seismically compact braces and indicated that brace compactness may be a larger concern in evaluation and retrofit than connection capacity.

c)

a) b) c) Figure 6. NCBF 1 (a) Connection Detail (b) Load-Drift History (c) Brace Hinge Formation

NCBF 2 – Compact Brace

Because NCBF 1 sustained only minor connection and frame damage aside from the brace fracture, the frame was reused for a similar test. The fractured HSS7x7x1/4 brace was removed, minor weld damage was repaired, and the gusset plates were heat-straightened. The brace was replaced with a more compact HSS5x5x3/8, which meets current SCBF width-thickness ratio requirements. The existing damage to the gusset plates was expected to have a minimal impact on the performance of the frame.

a) b) Figure 7. NCBF 2 (a) Initial Weld Cracking (b) Connection Fracture

Figure 7 provides some photos of this test specimen. NCBF 2 buckled at the same drift

level as NCBF 1, though the smaller radius of gyration of the HSS5x5x3/8 caused a smaller brace buckling load. In subsequent cycles, the brace retained a larger fraction of its compressive capacity than the brace in NCBF 1. This behavior was likely due to the lower width-thickness ratio of the brace walls in NCBF 2. The lack of hinging in the brace increased the axial and rotational demands on the connection, causing weld tearing to initiate at the ends of the gusset plate-to-beam welds. The weld tear lengthened over subsequent cycles, tearing completely in the first tensile cycle at 1.6% applied drift. The tear propagated along the gusset plate-to-shear tab weld, causing complete connection failure. NCBF 3 – Modified Welding Material

NCBF 3 was a similar specimen, which had no prior damage, and was constructed to more completely evaluate the connection performance. The connection details were identical to NCBF 2, except that all welds conformed to AWS E71T-11, which does not have a minimum toughness requirement, and is expected to more accurately represent the weld material used in

existing NCBFs.

Figure 8. NCBF 3 (a) Load-Drift History (b) Weld tearing from weld center (c) Complete weld fracture

The frame reached a maximum drift range of 3.3% (Fig. 8a), which was greater than that achieved by NCBF 2. Since this connection was fundamentally the same as the connection in NCBF 1, this result suggests that higher brace compactness alone can significantly increase the drift capacity of a braced frame, and it further suggests that the performance of NCBF 2 was influenced by damaged induced in testing of NCBF 1. The failure mechanism for this specimen was less desirable, however. The gusset plate-to-shear tab weld tore open from the back side (Fig. 8b). The low toughness weld allowed the cracks to propagate quickly, resulting in complete fracture of the shear tab away from the gusset plate and beam web in the next cycle. This was an undesirable, nonductile failure. SCBF 1 – Bolted Split Shear Tab

The first SCBF (see Fig. 9a) test was a bolted connection funded and designed by AISC with the current AISC Seismic Provisions and used the Uniform Force Method. The specimen included the 2 times the gusset plate thickness fold line clearance, a seismically compact HSS5x5x3/8 brace, net section reinforcing plates on the brace, and demand critical welds. The beam web was bolted to the column flange using a 1” thick plate and the gusset plate was bolted to the column flange with a 3/4” plate. Both connections used A325, 1” diameter bolts.

Fig. 9 shows the hysteretic behavior of SCBF 1 and photographs from the experiment. The maximum compressive force achieved in the brace was 376.6 kip and the maximum tensile force was 171.4 kip. Brace buckling was observed in the second compression cycle at 0.347% drift. At 0.9% drift the frame was able to sustain higher compressive loads again due to frame action. At 2.431% drift a hinge formed at the center of the brace. In the second tension cycle at 2.778% drift the brace fractured. The compact brace achieved over 15” out-of-plane displacement at midspan before fracture.

There was no evidence of bolt hole elongation or notable damage to the bolts, plates, or

other elements of the brace-beam-gusset plate connection. From these observations, it was determined the performance of the braced frame was not affected by bolt ductility. Weld cracking was observed weld cracks in both gusset-to-beam welds during 1.736% applied compression drift cycles, and cracks grew to 2” and 3” lengths for the north and south connections, respectively.

(a) (c) (b)

Figure 9. SCBF 1 (a) Connection detail (b) Load-Drift history (c) Complete brace fracture NCBF 4 – Bolted Shared Shear Tab NCBF 4 was designed to model a popular NCBF bolted connection from the building survey. It consists of a single shared shear plate. It was similar to NCBF 3, but a bolted joint was employed. The prototype connection was extremely deficient in bolt shear, and the test explored the impact of this deficiency on system performance. The gusset plate and shear tab were 3/4” thick plates. The shear tab was connected to the column flange by a complete joint penetration weld. The gusset plate did not have brace end rotation clearance, the brace had no net section reinforcement, and block shear and weld size were issues of concern. The welds were designed to match DCRs found in the building survey and used E71T-11 weld metal without minimum toughness requirements.

Fig. 10 shows the hysteretic behavior of NCBF 4 and photographs from the experiment. NCBF 4 had a surprising performance – it was able to achieve a maximum tensile load of 362 kips and a maximum compressive load of 170.9 kips. It exhibited ductile behavior typical of an SCBF and ultimately failed through brace fracture. The frame experienced an applied drift range of 4.82%. There was visible upward brace buckling at 0.347% applied drift. A hinge formed at the brace midspan at 2.431% applied drift with full brace fracture at 3.125% applied drift. The frame’s ductility was likely due to hinging in the columns, weld fracture, bolt hole elongation, and bolt ductility.

Figure 10. NCBF 4 (a) Connection Detail (b) Load-Drift history (c) Bolt hole elongation in beam web

There was significant deformation of bolt holes, most notable in the beam web, since it was the thinnest component. The largest elongation experienced was 5/16” (Fig. 10b). NCBF 4 suggests systems such as these might be capable of capitalizing on the ductility of other

(a) (b) (c)

(a) (b) (c)

components in the configuration. Future tests will explore this assertion further.

Conclusion

Because there were no seismic requirements prior to the 1988 UBC [2], it is believed that a number of NCBF systems across the US contain deficient connections and that their lateral load resistance may be restricted by limit states other than brace yielding. A number of these vulnerable connections were investigated to determine their performance and examine potential retrofits. Through testing completed to date, the following conclusions can be drawn:

• Not all calculated deficiencies in NCBFs have a negative impact on system performance. As seen in NCBF 4, the bolt shear deficiencies did not cause a brittle failure as anticipated. In fact, the bolts and bolt hole elongation provided inelastic deformation capacity to the frame. NCBF 3 had a high DCR for gusset plate shear, and the thin gusset plate allowed for significant out of plane deformation of the brace.

• Systems with a non-seismically compact brace fractured quickly at relatively small inelastic deformation shortly after brace buckling. NCBFs with this deficiency can benefit from a retrofit with a seismically compact brace as supported by NCBF 3.

• Some NCBF system behavior was more ductile than expected, but the failure modes are not always desirable. NCBF 3 achieved moderate deformation capacity but had abrupt, brittle failure of the weld connecting the shear plate to the beam and gusset plate before the brace could fracture.

• SCBF systems can benefit from bolt hole elongation and bolt deformability.

Additional tests are planned to further investigate and elaborate upon these conclusions, with the goal of more clearly understanding NCBF behavior and SCBF ductility with bolted connections. With new test data, braced frame retrofits can become a reliable and cost effective choice, and CBFs can continue to improve.

Acknowledgements

This research is funded by the National Science Foundation through research grant CMS1208002,”NEESR: Collaborative Developments for Seismic Rehabilitation of Vulnerable Braced Frames,” and by the American Institute of Steel Construction. This financial support is gratefully acknowledged.

References 1. AISC (2010a). “Seismic Provisions for Structural Steel Buildings.” American Institute of Steel Construction,

Chicago, IL.

2. ICBO (1988). “1988 Uniform Building Code.” International Conference of Building Officials, Whittier, CA.

3. Hsiao, P. C., Lehman, D. E., Berman, J. W., Roeder, C. W., and Powell, J. (2012). “Seismic Vulnerability of Older Braced Frames.” Journal of Performance of Constructed Facilities, DOI: 10.1061.

4. AISC (2010b). “Manual of Steel Construction, Load and Resistance Factor Design.” 14th Edition, American Institute of Steel Construction, Chicago, IL.