LINKED COLUMN FRAME STEEL SYSTEM PERFORMANCE VALIDATION USING HYBRID SIMULATION

Transcription

1 10NCEE Tenth U.S. National Conference on Earthquake Engineering Frontiers of Earthquake Engineering July 21-25, 2014 Anchorage, Alaska LINKED COLUMN FRAME STEEL SYSTEM PERFORMANCE VALIDATION USING HYBRID SIMULATION A. Lopes 1, P. Dusicka 2 and J. Berman 3 ABSTRACT The Linked Column Frame (LCF) is a new brace-free lateral structural steel system intended for rapid return to occupancy performance level. LCF is more resilient under a design level earthquake than the conventional approaches. The structural system consists of moment frames for gravity that combines with closely spaced dual columns (LC) interconnected with bolted links for the lateral system. The LC links are sacrificial and intended to be replaced following a design level earthquake. The centerpiece of this work was a unique full scale experiment using hybrid testing; a combination of physical test of a critical sub-system tied to a numerical model of the building frame. This paper outlines the experimental setup, testing and validation of the LCF steel frame system. Hybrid testing allows for full scale study at the system level accounting for the uncertainties via experimental component and having the ability to model more conventional behavior through numerical simulation. The experimental sub-system consisted of a two story LCF frame with a single bay while the remainder of the building was numerically modeled. Two actuators per story were connected to the specimen. The LC links have been designed to be short and plastically shear dominated and the LCF met the design intent of 2.5% inter-story drift limits. For evaluating the LCF response, hybrid testing was performed for ground motion at three different intensities; 50%, 10% and 2% probability of exceedence in 50 years for Seattle, Washington ground motions. The system overall had exhibited three distinct performance levels; linearly elastic, rapid return to occupancy where only the replaceable links would yield, and collapse prevention where the gravity beam components also became damaged. Experimental results demonstrated a viable system under seismic loading, offering a ductile structural system with the ability to rapidly return to occupancy. 1 Graduate Assistant, Dept. of Civil Engineering, Portland State University, Portland, OR 97201, arlindo@pdx.edu 2 Associate Professor, Dept. of Civil Engineering, Portland State University, Portland, OR 97201, dusicka@pdx.edu 3 Associate Professor, Dept. of Civil Engineering, University of Washington, Seattle, WA 98195, jwberman@uw.edu Lopes A., Dusicka, P. and Berman, J. Linked Column Steel Frame System Validation Using Hybrid Testing. Proceedings of the 10 th National Conference in Earthquake Engineering, Earthquake Engineering Research Institute, Anchorage, AK, 2014.

2 Linked Column Frame Steel System Performance Validation Using Hybrid Simulation A. Lopes 1, P. Dusicka 2 and J. Berman 3 ABSTRACT The Linked Column Frame (LCF) is a new brace-free lateral structural steel system intended for rapid return to occupancy performance level. LCF is more resilient under a design level earthquake than the conventional approaches. The structural system consists of moment frames for gravity that combines with closely spaced dual columns (LC) interconnected with bolted links for the lateral system. The LC links are sacrificial and intended to be replaced following a design level earthquake. The centerpiece of this work was a unique full scale experiment using hybrid testing; a combination of physical test of a critical sub-system tied to a numerical model of the building frame. This paper outlines the experimental setup, testing and validation of the LCF steel frame system. Hybrid testing allows for full scale study at the system level accounting for the uncertainties via experimental component and having the ability to model more conventional behavior through numerical simulation. The experimental sub-system consisted of a two story LCF frame with a single bay while the remainder of the building was numerically modeled. Two actuators per story were connected to the specimen. The LC links have been designed to be short and plastically shear dominated and the LCF met the design intent of 2.5% inter-story drift limits. For evaluating the LCF response, hybrid testing was performed for ground motion at three different intensities; 50%, 10% and 2% probability of exceedence in 50 years for Seattle, Washington ground motions. The system overall had exhibited three distinct performance levels; linearly elastic, rapid return to occupancy where only the replaceable links would yield, and collapse prevention where the gravity beam components also became damaged. Experimental results demonstrated a viable system under seismic loading, offering a ductile structural system with the ability to rapidly return to occupancy. Introduction New structural systems are being developed with emphasis on immediate occupancy following a design level earthquake. An example of these includes developments of damage free beam column connection, either through post tensioning [1] or friction based resistance [2]. An alternative approach to immediate occupancy performance level is to design for damage in nongravity members that could be replaced. Such structural systems would be capable of rapid repair that would facilitate return to occupancy following a seismic event. For example, buckling restrained brace frames exhibit this characteristic. A brace-free alternative was developed in structural steel and is referred to as the Linked Column Frame (LCF) system. The LCF aims to 1 Graduate Assistant, Dept. of Civil Engineering, Portland State University, Portland, OR 97201, arlindo@pdx.edu 2 Associate Professor, Dept. of Civil Engineering, Portland State University, Portland, OR 97201, dusicka@pdx.edu 3 Associate Professor, Dept. of Civil Engineering, University of Washington, Seattle, WA 98195,jwberman@uw.edu Lopes A., Dusicka, P. and Berman, J. Linked Column Steel Frame System Validation Using Hybrid Testing. Proceedings of the 10 th National Conference in Earthquake Engineering, Earthquake Engineering Research Institute, Anchorage, AK, 2014.

3 address the rapid return to occupancy design performance while maintaining the architectural advantages of brace-free steel frame construction. Overview of the Linked Column Frame System Brace-free frames are the lateral systems of choice when design constraints need to accommodate large openings. The LCF system is a new lateral load resisting system with the goals of rapid return to occupancy (buildings can be rapidly re-occupied after future large earthquakes) via replacement of sacrificial components. The system consists of moment frames (MF) and linked columns (LC) with replaceable links. The MF provides priority gravity loadcarrying capacity and under earthquake excitation the structure remains elastic. The LC consists of closely spaced columns interconnected with links, which are designed to yield, deform plastically and be replaceable under the design level earthquake. The LCF system's ability to achieve rapid return to occupancy relies on the behavior of the replaceable links. The LCF system also offers architectural advantages of open perimeter bays and occupation versatility in the interior floor layout. Example LCF layout for a 2-story building is shown in Fig. 1. Figure 1. LCF layout for a 2-story building. During the initial development of the system [3], non-moment transferring connections were introduced at all column to foundation locations and in strategic beam to column locations. These idealized pin connections at the base of each column limit yielding at the foundation and thereby minimize damage to the columns that is typical in ductile moment frame designs. The idealized pin connections in the MF beams reduce the lateral stiffness of the gravity moment frame. Complementary numerical and experimental study has been carried out on bolted links that are either shear or moment dominated [4]. Practical connection details for the link components are used in an effort to provide elastic connections that shift plastic strains away from critical welds, thereby avoiding some of the failures observed in past tests on similarly detailed link-to-column eccentrically braced frames. Other experiments on a steel frame with multiple sacrificial metal rods at mid-length of the links that were distributed throughout a story height had also indicated favorable cyclic response [5]. Also, system level numerical model development and analyses of the LCF system have been done considering non-linear static analyses [6] and time history analyses [7] leading to design procedure approaches for the LCF system toward performance level response.

4 Prototype Building Layout The design resistance of the LCF was based upon a prototype building that is a modified version of the 3-story building SAC configuration [8]. The modification is due to physical space laboratory constraints. In elevation, the typical bay width is 24.5 ft., typical story height is 10 ft., and each LC is spaced 3.5 ft. apart. Numerical models were designed per LCF preliminary design procedure described in [6] to determine the section sizes of LCF systems meeting the design intent of 2.5% inter-story drift limits. All steel is assumed to be 50 ksi nominal yield stress and the resulting member sizes are summarized as follows: columns (W14x132), gravity beams (W16x57) and links (W10x45). The LC links have been designed to be short and plastically shear dominated. Three links for the LCF-3L specimen (Fig. 2a) with identical geometric and material properties were fabricated. Shear links were bolted to the columns to facilitate post-earthquake replacement. Gravity beams were bolted to the columns through endplate moment connection at one end and a double-angle shear connection at the other end (Fig. 2b). For 2-story buildings the LCs dominated the stiffness as well as strength of the overall structure. Lateral parameters indicate that rapid return to occupancy performance level can occur over a drift range of 2% [9]. Experimental Test Setup and Instrumentation The analyses thus far have focused on system level numerical model development and analyses of the LCF system. The next step in understanding the behavior of the system was to perform experimental validation of an LCF system. The goals of the tests were to understand how the LCF system components interact together, to monitor the progression of damage in the replaceable links and ultimately validate the rapid return performance based design methodology. The experiments were conducted at the National Science Foundation NEES (George E. Brown Jr. Network for Earthquake Engineering Simulation) node at the University of California, Berkeley. (a) (b) Figure 2. (a) LCF-3L specimen and (b) Moment and shear connections.

5 Actuators for the tests were selected from among those available in the laboratory, considering the loads and displacement targets. Thus, two static MTS T, 216 kips with ± 72 inches stroke were used at the second story level and two dynamic MTS S, 220 kips with ± 20 inches stroke were used at the first story level. With these actuators and also with laboratory constraints, a maximum second story level displacement equal to about 5% drift could be imposed. To distribute the internal forces from the specimen to the strong floor, a heavy built-up floor beam was provided at the top of the strong floor slab, and a series of relatively stiff load transfer beams were provided on the bottom side of the slab. The floor beam and load transfer beams were connected by post-tensioned anchor rods. To avoid out-of-plane movement of the frame, a lateral restraint system was used. The lateral restraint system consisted of T-beam with sliding mechanisms (Fig. 3a), HSS columns, plates that connect T-beams to the gravity beams (Fig. 3b), and plates that connect T-beams to the columns. (a) (b) Figure 3. Parts of the LCF lateral restraint system. The response of the LCF steel system subjected to earthquake loading was measured using a total of 224 data acquisition channels. Three types of strain gages were used, namely, general purpose uniaxial gage, large deformation uniaxial gage, and rosettes. The general purpose uniaxial gages were placed along column flanges. They were also placed on top and bottom flanges of the gravity beams near the double-angle shear connection. The large deformation uniaxial gages were attached on top and bottom flanges of the links and were also attached on top and bottom flanges of the gravity beams near the end-plate moment connection (Fig. 4a). Rosettes were placed on the link web and column web. Additional LVDTs were used to measure the panel zone deformation and link rotations (Fig. 4b). (a) (b) Figure 4. (a) Strain gage installation and (b) LVDT installation.

6 Hybrid Simulation and Earthquake Selection The seismic behavior of steel structures has been the subject of extended research and several experimental tests on such structures were conducted. In most cases, these structures were either subjected to quasi-static cyclic loading [10] in which the dynamic response of the system cannot be captured or were tested on a shaking table [11]. Typically, is very difficult to do a large or full scale test using a shake table due to limitations on table capacity. Seismic tests may also involve hybrid simulation, a combination of physical test of a critical subsystem tied to a numerical model of the building. In a hybrid simulation test, the well understood part of the structure is modeled in a finite element program and the critical sub-system (highly non-linear or numerically hard to model) is built in the laboratory to be tested. In order to understand the behavior of the system during an earthquake, different levels of ground motions were applied to the LCF. Due to differences on lateral parameters between the multistory (2-story, 4-bay) and experimental (2-story, 1-bay) models, hybrid simulation was needed to investigate the overall structure response [9]. To perform the hybrid simulation, the Open System for Earthquake Engineering Simulation (OpenSees) [12] was used as a finite element software to model and analyze the LCF system. The Open Source Framework for Experimental Setup and Control (OpenFresco) [13], was also used as a middleware to connect the finite element analysis software with a control and data acquisition software. A 2-story, 1-bay physical specimen was built in the laboratory and a hybrid model was developed in OpenSees. Two generic experimental elements, two rigid-link trusses and a leaning column were defined. Lump floor masses were transferred from the LCF sub-systems to the hybrid model. For the LCF-3L specimen, p-delta effects were considered through introducing a leaning column while its section properties were obtained from the gravity columns. Fig. 5 shows the model used for the hybrid simulation. Figure 5. LCF hybrid simulation model. Ground motion intensity was selected such that three distinct performance levels were induced in the LCF specimen: linearly elastic (E), rapid return to occupancy (RR1 and RR2) corresponding to a moderate damage state, and collapse prevention (CP1 and CP2)

7 corresponding to a significant damage state. The ground motions were those developed in the SAC project for the Seattle site [14]. This location is considered a medium-high seismic design category of the United States. The ground motion for the linearly elastic intensity was obtained taking 15% of SE-05. Details of ground motions and Response Design Spectra (RDS) are given in Table 1 and illustrated in Fig. 6. Table 1. Details of Seattle ground motions. Name Code Record Magnitude Scale factor SE-05 RR1 Olympia, SE-20 RR2 Vina del Mar, SE-25 CP1 Olympia, SE-29 CP2 Valparaiso, Figure 6. Ground motions used in the hybrid simulation model. LCF Earthquake Response For the hybrid simulation testing a combined ground motion input time history was used. The test was conducted very slowly compared to real time and continued until the end of the last earthquake. The experimental earthquake response of LCF-3L is presented in terms of total base shear versus drift and link shear deformation versus drift hysteresis. Thus, the focus will be set on describing the global behavior of the LCF system and local behavior of links. The application of the ground motions revealed a sequence of yielding events as follows: link 1 at base, link 2 at first story, link 3 at second story, beam 1 at first story, and beam 2 at second story. Fig. 7 (a,c) shows that the structure displaced up to 1.8% drift and 2.8% when subjected to earthquakes RR1 & RR2 and CP1 & CP2, respectively. Fig. 7 (b,d) shows link rotation for all three performance levels. The link rotation demands in link 1 at base are greater than in link 3 at second story which concur with the sequence of yielding events mentioned above. Whitewash on the links began flaking near mid-span, then propagated toward the end plates as shown in Fig. 8. In the collapse

8 prevention performance level the links have larger inelastic demand and are more likely to require replacement. It should be noted that none of the links failed under any ground motions used in the hybrid simulation. (a) (b) (c) (d) Figure 7. Base shear versus drift (a, c) and link shear deformation versus drift (b, d) hysteresis. Figure 8. Progression of damage in replaceable link 1. Fig. 9 (a,c) shows the gravity beam flange strain distribution for the rapid return to occupancy and collapse prevention performance levels. The yielding strain y is shown by horizontal dashed lines for reference and was obtained using Hooke s law. As shown, beam flange yielding does not occur until collapse prevention performance level is achieved. This ensures that no repairs would be necessary and will help to minimize post event repair costs. Fig. 9 (b) shows limited demands on gravity beams and this could indicate less rigorous detailing connections. LCF-3L specimen exhibited three regions within the lateral response; elastic, yielding of links and yielding of links as well as MF beams. Provided the links are replaceable, these correspond to three distinct performance levels; elastic, rapid return to occupancy and collapse prevention.

9 (a) (b) (c) Figure 9. Gravity beam flange strain distribution. Ultimate Cyclic Loading In the framework of the project two full scale experimental specimens could be tested. The LCF is a new lateral system and has never been tested before, hence after the hybrid test was finished; the physical LCF sub-system was tested under an ultimate cyclic loading up to 5% total drift. Lateral loads were applied at first and second story elevations using four servo-controlled hydraulic actuators, two per floor. The displacements of both first and second story were monitored and controlled during the entire test. FEMA 641 testing protocol was used. The loading protocol calls for two targets ( 0 and m ) and a predetermined number of increments (n). For testing purposes, a controlling value of 0 = 0.33 in., m = 5.7 in. and n = 10 were used. The testing was conducted quasi-statically in a displacement-controlled mode. Fig. 10 (a) shows the maximum response in terms of hysteresis loop of the physical LCF sub-system. A maximum second story displacement of 10.8 in. with a 200 kip base shear was obtained. In regards to the shear links, Fig. 10 (b) shows the link rotation demands in link 1 ( =0.07 rad.) are greater than in link 3 ( =0.05 rad.). Web buckling did not start to form in the web of the shear link until 4% drift. With the progression of the cycles, web buckling became more pronounced and a crack started at mid-span followed by a crack between the top flange and web of the link. The cracks kept propagating until 5% drift. The web began to tear and the test was stopped. Fig. 11 shows the behavior of links 1, 2 and 3; and gravity beams 1 and 2 at 5% drift. Shear links shown to be effective in protecting gravity system and participating well past 4% drift and gravity beams had limited damaged.

10 (a) (b) Figure 10. Base shear versus drift (a) and link shear deformation versus drift (b) hysteresis. Figure 11. Photos of links 1, 2, and 3 and gravity beams 1 and 2 at 5% drift of LCF sub-system. Summary and Conclusions A LCF system was investigated experimentally via hybrid simulation to validate the response at a system level. The system experiment was a combination of physical test of a critical subsystem tied to a numerical model of the building. Hybrid simulation allowed for full scale study at the system level accounting for the uncertainties through experimental component and having the ability to model more conventional behavior through numerical simulation. Subsequently, LCF sub-system was tested under an ultimate cyclic loading up to 5% total drift. For the case considered, the following conclusions can be drawn: The structural layout was suitable for entire frame validation using hybrid simulation. The LCF moment frame remained elastic until rapid return to occupancy performance level was achieved, while links yielded and deformed plastically.

11 Shear links shown to be effective in protecting gravity system and participating well past 4% drift. Limited demands on gravity beams could indicate less rigorous detailing connections. LCF-3L specimen exhibited three regions within the lateral response; elastic, yielding of links and yielding of links as well as MF beams. Provided the links are replaceable, these correspond to three distinct performance levels; elastic, rapid return to occupancy and collapse prevention. Acknowledgments This paper is based upon work supported by the National Science Foundation under Grant No with initial support from the American Institute of Steel Construction. The authors thank the staff at for their assistance. The first author s leave from Amazonas State University in Brazil is also gratefully acknowledged. References 1. Ricles JM, Sause R, Garlock MM, and Zhao C. Post Tensioned Seismic-Resistant Connections for Steel Frames. Journal of Structural Engineering 2001; 127 (2): Khoo HH, Clifton GC, Butterworth JW, Mathieson CD, and MacRae GA. Development of the Self-Centering Sliding Hinge Joint. Proceedings of the 9 th Pacific Conference on Earthquake Engineering, 2011, Auckland. 3. Dusicka P and Iwai R. Development of the Linked Column Frame System for Seismic Lateral Loads. Proceedings of the American Society of Civil Engineers Structures Congress, 2007, Long Beach, California. 4. Dusicka P and Lewis G. Investigation of Replaceable Sacrificial Steel Links. Proceedings of the 9 th US and 10 th Canadian Conference on Earthquake Engineering, 2010, Toronto, Canada. 5. Palkopoulou O, Karydakis P and Vayas I. Innovative Bracing System for Seismic Resistant Steel Structures. Proceedings of the International STESSA Conference Behaviour of Steel Structures in Seismic Areas, 2009, Philadelphia, Pennsylvania. 6. Lopes A, Dusicka P and Berman J. Design of the Linked Column Frame Structural System. Proceedings of the International STESSA Conference Behaviour of Steel Structures in Seismic Areas, 2012, Santiago, Chile. 7. Malakoutian M, Berman J and Dusicka P. Seismic Response Evaluation of the Linked Column Frame. Earthquake Engineering and Structural Dynamics 2013; 42: FEMA-355C. State of the Art Report on Systems Performance of Steel Moment Frames Subject to Earthquake Ground Shaking. SAC Joint Venture, 2000, Washington-DC. 9. Lopes A, Dusicka P and Berman J. Linked Column Frame System Analyses Toward Experimental Validation. Proceedings of the American Society of Civil Engineers Structures Congress, 2012, Chicago, Illinois. 10. Roeder C and Popov E. Inelastic Behavior of Eccentrically Braced Steel Frames under Cyclic Loadings. Earthquake Engineering Research Center 1997; Report No. UCB/EERC-77/ Yang M Seismic Behavior of an Eccentrically X-Braced Steel Structure. Earthquake Engineering Research Center 1997; Report No. UCB/EERC-82/ McKenna F. Object-Oriented Finite Element Programming: Frameworks for Analysis, Algorithms and Parallel Computing. Ph.D. Dissertation in Civil Engineering, 1997, University of California, Berkeley. 13. Schellenberg A. Advanced Implementation of Hybrid Simulation. Ph.D. Dissertation in Civil Engineering, 2008, University of California, Berkeley. 14. Somerville P, Smith N, Punyamurthula S., and Sun J. Development of Ground Motion Time History for Phase 2 of the FEMA SAC Steel Project. SAC Joint Venture, 1997, Washington-DC.