This dissertation focuses on the use of supplemental energy-dissipation devices to improve the seismic performance of tall steel buildings. It is divided into two parts. Part 1 focuses on exploring cost-effective retrofit strategies to improve the seismic performance of an existing tall building. The selected building is a 35-story steel moment-resisting frame, with representative details from the early 1970s. Detailed seismic evaluations were conducted in the framework of Performance Based Earthquake Engineering (PBEE), using the scenario-based performance assessment methods. A three-dimensional numerical model capturing the mechanical properties of the most critical structural elements was generated using the program: Open System for Earthquake Engineering Simulation (OpenSees). Seismic evaluation of the selected building was done following ASCE 41-13, FEMA 351 and FEMA P-58, and two hazard levels: basic safety earthquake levels 1 and 2 (BSE-1E and BSE-2E) prescribed by ASCE 41 were used for the assessment. Results predicted that this building failed to meet the recommended performance objectives and had a variety of seismic vulnerabilities, and possible retrofits were needed.

Therefore, a two-level retrofit approach was examined that focused on achieving the collapse prevention limit state under the BSE-2E hazard level. In Level-1, the brittle column splices were fixed everywhere in the building, and the massive concrete cladding was replaced with lightweight substitute in the exterior of the building. Level-2 strategies augmented the Level-1 methods by adding different supplemental energy-dissipation devices. Devices investigated include: fluid viscous dampers (FVDs), viscous wall dampers (VWDs) and buckling restrained braces (BRBs). Among these, the scheme that used FVDs was expected to be the most promising to upgrade the seismic performance of the case-study steel moment frame, and thus was examined first. In this approach, feasible damper locations and overall effective damping ratios were evaluated through a series of preliminary studies, and then a two-phase manual design method was used to refine the distribution and mechanical properties of the dampers. Thorough assessments of the refined design were carried out and the results indicated that the proposed retrofit method of using FVDs could achieve the retrofit goal and provide a cost-effective means of improving the structural behavior and reducing economic losses in a major seismic event for this case-study building.

The study was extended to examine alternative measures to upgrade the case-study building by using either VWDs or BRBs, and compared their relative effectiveness and economy with the scheme using FVDs. The locations and effective damping ratios were kept the same for all three schemes to insure a valid comparison. Results indicated that the proposed schemes of VWDs and BRBs both failed to achieve the targeted performance goal for this structure under a BSE-2E event, and special design considerations were required.

Part 2 of the dissertation focuses on developing an automated tool to streamline the design of FVDs in tall buildings. Aided by the high-performance computers and parallel processors, a large amount of complicated nonlinear response history analysis was conducted to facilitate the automate design procedure. The optimization problem was devised in a simplified PBEE framework under one hazard level each time. Basic optimization ingredients were selected to reflect the target performance goal, and several cases using different objective functions were evaluated.

Two tall buildings: the existing steel moment frame examined before and a newly-designed mega-brace steel frame were selected to rely on the automated procedure to optimally design FVDs. In both cases, the automated procedure turned to be very efficient, help identify design parameters of dampers in selected locations and reduce a great amount of engineering efforts. With only limited number of iterations, optimal design patterns of FVDs in a tall building could be found, which were able to improve the structural performance under different hazard events. The suggested optimal design could meet retrofit goal for the existing tall building, as well as achieve enhanced performance goal for both existing and new tall buildings.

Seismic isolation is utilized in both traditional and industrial structures for its ability to reduce structural responses while effectively protecting sensitive building contents. Although this mature technology has been studied extensively, there is limited research on its application to Nuclear Power Plants (NPPs). Studies have yet to address topics such as sensitivity of equipment and the effects of seismic isolation on floor response spectra (FRS) caused by high frequencies and the inclusion of vertical motion. This paper's studies investigate the relationship between structural response and isolator mechanics, ground motion sensitivity, and the effectiveness of various isolator types, especially nonlinear isolators. Overall trends are identified that provide insight into the interaction between the isolation system and the response at various levels of the NPP. By understanding these relationships and connections, a set of valid recommendations for the selection and design of isolation systems is provided. This work is pursued through analyzing the performance of an APR1400 plant considering regulatory guidelines and utilizing a variety of ground motions, isolator types, and isolator models.

The research in this dissertation describes simulations of the inelastic seismic behavior of steel braced frames including the effects of low-cycle fatigue. Various types of nonlinear behavior are considered: material inelasticity, low-cycle fatigue, and local and global geometric nonlinearities. The effects of suddenly started, quasi-brittle fracture are not considered herein. For steel braced frames, braces, columns, beams, and connections subjected to significant axial loads and deformations, as well as bending moments and shear. Under these complex-loading conditions, a wide variety of behavior mechanisms and failure modes may occur for each type of member and connection. Thus, numerical models that assess the initiation and propagation of failure during cyclic loading need to account for multi-axial states of material nonlinearity, local and global buckling, and the exhaustion of the ability of the material to deform inelastically caused by low-cycle fatigue.

Following a review of existing material models for simulating structural steel deterioration, a series of investigations are conducted using finite element modeling techniques. Finite element methods can directly account for complex states of stress and changes in deformed shape. And material models are critical for constitutive behavior at integration points of the finite element models. However, available material models tend to emphasize behavior associated with ideal ductile response or with failure occurring under monotonic loading conditions (\emph{e.g.,} during metal-forming processes or vehicle collision). These models are not suitable for progressive collapse analysis under cyclic loading where the consequence of this adverse behavior on the subsequent response or integrity of the structure is of interest.

Therefore, a new, numerically efficient continuum damage mechanics material model capable of simulating inelastic behavior and deterioration of mechanical properties because of low-cycle fatigue has been devised and implemented in a finite element software LS-DYNA (LSTC 2007). Computational results obtained with this new material model correlate well with test results for several beam-to-column connections, individual braces, and braced frame subassemblies. These applications of the finite element model to realistic cases involving progressive collapse illustrate the importance of material deterioration and rupture. Unfortunately, the ability of the material model to predict ultimate behavior depends heavily on the material modeling properties specified. Recommendations for characterizing material properties for these types of analysis are developed and presented.

A series of analyses are presented that evaluate and refine several requirements for detailing and analyzing special concentrically braced steel frame buildings, demonstrating that the fatigue life capacity of braces is heavily dependent on width-thickness ratios and deformation histories. Member slenderness ratios are shown to have negligible effect on fatigue life capacity. Therefore, recommendations are presented for developing fatigue life demand or loading protocols for use in numerical and experimental investigations. Next, damage evolution in gusseted beam-to-column connections is evaluated and compared for different connection details, and improved connection details are recommended to reduce the damage accumulation. The position of lateral bracing members for beams in V-type and inverted V-type braced frames are also examined. More appropriate positions and methods to compensate for problems detected for currently recommended lateral bracing member positions are suggested and evaluated. Finally, for low-rise braced frames that respond inelastically during strong earthquake ground shaking, an alterative method to estimate interstory drift demands is suggested based on the Modal Pushover Analysis procedure.

Abstract

Seismic Performance of Reinforced Concrete Bridges Allowed to Uplift During

Multi-Directional Excitation

by

Andres Oscar Espinoza

Doctor of Philosophy in Engineering - Civil and Environmental Engineering

University of California, Berkeley

Professor Stephen A. Mahin, Chair

The behavior of bridges subjected to recent moderate and large earthquakes has led to bridge design detailed for better seismic performance, particularly through wider bridge foundations to handle larger expected design forces. Foundation uplift, which is not employed in conventional bridge design, has been identified as an important mechanism, in conjunction with structural yielding and soil-structure interaction that may dissipate energy during earthquakes. Preventing uplift through wider foundations looks past the technical and economical feasibility of allowing foundation uplift during seismic events. The research presented in this thesis is part of a larger experimental and analytical investigation to develop and validate design methods for bridge piers on shallow foundations allowed to uplift during seismic events.

Several analytical and some experimental studies have been performed to assess rocking and or uplift of shallow foundation systems, however they have evaluated systems with a limited range of footing dimensions and seismic excitations. As such, there is an uncertainty in the information needed to base a performance evaluation and develop design methods. The purpose of this study is to investigate, through experimental and analytical studies, the seismic performance of uplifting bridge piers on shallow foundations when considering different ground motions and footing dimensions. As well as to identify key differences in performance evaluation criteria for conventional and uplifting bridge pier systems.

The experimental study dynamically tested a single reinforced concrete bridge column specimen with three adjustable footing configurations grouped by footing dimension, and tested for various combinations of one, two, and three components of seismic excitation. Groups one and two evaluated uplifting systems where the column was limited to elastic loading levels while group three considered inelastic column loading levels. All test groups remained stable and exhibited some rocking and or uplift during testing. Analytical models were developed and validated using the experimental testing results to predict local and global footing and column response. Reliable estimates of forces and displacements during elastic and inelastic response were achieved. To assess the seismic performance of a range of bridge pier systems allowed to uplift a parametric investigation using the validated analytical models was performed in which the column was modeled per conventional design criteria to ensure adequate strength and flexural ductility. The parameters varied include footing width, ground motion excitation, and elastic or inelastic column response. Response of the uplifting bridge pier systems was found to be sensitive to the structural periods, magnitude of excitation, and footing width.

The special concentrically steel braced frame (SCBF) system is one of the most effective struc-tural systems to resist lateral forces. Because of its effectiveness and straightforward design, many SCBFs are incorporated in structures throughout the world. However, the highly nonlin-ear behavior associated with buckling and non-ductile fracture of braces reduces the ability of the system to dissipate energy resulting in undesirable modes of behavior. While many studies have investigated the cyclic behavior of individual braces or the behavior of subassemblies, the dynamic demands on the structural system under various seismic hazard levels needs additional study for performance-based earthquake engineering.

Archetype buildings of SCBFs and buckling restrained braced frames (BRBFs) were analyzed using the computer program OpenSees (the Open System for Earthquake Engineering Simulation) to improve the understanding of the seismic behavior of braced frame systems, and to assess seismic demands for performance-based design. Numerical models were calibrated using test data determined from testing of conventional buckling braces, buckling restrained braces, and the braced frame specimens. In addition, fiber-based OpenSees models were constructed and compared with results of a sophisticated finite-element model that realistically captured local buckling and local fracture of structural elements. Because the OpenSees models are reasona-bly accurate and efficient, they were chosen to perform set of parametric computer simulations.

The seismic demands of the system and structural elements were computed and interpreted for 3-, 6-, and 16-story SCBFs and BRBFs under various hazard levels. The analysis results show large seismic demands for the 3-story SCBF, which may result in unexpected damage of struc-tural and non-structural elements. The median expected probability of a brace buckling at one or more levels in a 3-story SCBF is more than 50% for an earthquake having a 50% probability of exceedance in 50 years (the service-level event). The possible need to replace braces fol-lowing such frequent events due to brace buckling should be considered in performance-based earthquake engineering assessments. In addition, brace fracture in SCBFs is likely for an earthquake having a 2% probability of exceedance in 50 years (the MCE-level event). Analy-ses show that in general, BRBF models had larger drift demands and residual drifts compared to SCBF systems, because of the BRBF's longer fundamental period. However, the tendency to form a weak story in BRBFs is less than that in SCBFs.

Evaluation of seismic demand parameters were performed for 2-, 3-, 6-, 12-, and 16-story SCBFs and BRBFs, which demonstrated that short-period braced frame systems, especially SCBFs, had higher probability of collapse than longer-period braced frame systems. Substantially improved response was observed by lowering the response reduction factor of the 2-story SCBF building; this reduced the collapse risk at the hazard level of 2% probability of exceedance in 50 years. For long-period (taller) structures, although the collapse probability was lower compared to the short-period structures, weak story behavior was commonly observed in conventionally designed SCBF. A design parameter related to the ratios of story shear demand and capacity under a pushover analysis is proposed to modify member sizes to reduce weak story behavior efficiently. This is demonstrated for a 16-story SCBF building.

Regarding local deformation and force demands, simple methods to estimate out-of-plane buck-ling deformation of braces and column axial force demands are proposed. The investigation of system performance and member behavior provides seismic demands to more accurately assess the socio-economic losses of SCBFs and BRBFs for performance-based earthquake engineering.

This dissertation summarizes both experimental and analytical studies on the seismic response of conventional steel concentrically braced frame systems of the type widely used in North America, and preliminary studies of an innovative hybrid braced frame system: the Strong-Back System. The research work is part of NEES small group project entitled "International Hybrid Simulation of Tomorrow's Braced Frames."

In the experimental phase, a total of four full-scale, one-bay, two-story conventional braced frame specimens with different bracing member section shapes and gusset plate-to-beam connection details were designed and tested at the NEES@Berkeley Laboratory. Three braced frame specimens were tested quasi-statically using the same predefined loading protocol to investigate the inelastic cyclic behavior of code-compliant braced frames at both the global and local level. The last braced frame specimen was nearly identical to one of those tested quasi-statically. However, it was tested using hybrid simulation techniques to examine the sensitivity of inelastic behavior on loading sequence and to relate the behavior observed to different levels of seismic hazard.

Computer models of the test specimens were developed using two different computer software programs. In the software framework OpenSees fiber-based line elements were used to simulate global buckling of members and yielding and low-cycle fatigue failure at sections. The LS-DYNA analysis program was also used to model individual struts and the test specimens using shell elements with adaptive meshing and element erosion features. This program provided enhanced ability to simulate section local buckling, strain concentrations and crack development. The numerical results were compared with test results to assess and refine and the ability of the models to predict braced frame behavior. A series of OpenSees numerical cyclic component simulations were then conducted using the validated modeling approach. Two hundred and forty pin-ended struts with square hollow structural section shape were simulated under cyclic loading to examine the effect of width-to-thickness ratios and member slenderness ratios on the deformation capacity and energy dissipation characteristics of brace members.

The concept of a hybrid system, consisting of a vertical elastic truss or strong-back, and a braced frame that responds inelastically, is proposed herein to mitigate the tendency of weak-story mechanisms to form in conventional steel braced frames. A simple design strategy about member sizing of the proposed Strong-Back System is provided in this study. To assess the ability of the new Strong-Back System to perform well under seismic loading, a series of inelastic analyses were performed considering three six-story hybrid braced frames having different bracing elements, and three six-story conventional brace frames having different brace configurations. Monotonic and cyclic quasi-static inelastic analyses and inelastic time history analyses were carried out. The braced frame system behavior, bracing member force-displacement hysteresis loops, and system residual drifts were the primary response quantities examined. These indicated that the new hybrid system was able to achieve its design goals.

Experimental results show for the same loading history that the braced frame specimen using round hollow structural sections as brace members has the largest deformation capacity among the three types of bracing elements studied. Beams connected to gusset plates at the column formed plastic hinges adjacent to the gusset plate. The gusset plates tend to amplify the rotation demands at these locations and stress concentrations tended to result in early fractures of the plastic hinges that form. To remedy this problem, pinned connection details used in the last two specimens; these proved to prevent failures at these locations under both quasi-static and pseudo-dynamic tests. Failure modes observed near the column to base plate connections in all of the specimens suggest the need for further study. Both OpenSees and LS-DYNA models developed in this study predict the global braced frame behavior with acceptable accuracy. In both models, low-cycle fatigue damage models were needed to achieve an acceptable level of fidelity. Shell element models were able to predict local behavior and the mode of failures with greater but not perfect confidence. OpenSees analysis results show that the proposed hybrid braced frames would perform better than conventional braced frames and that the story deformations are more uniform. Finally, future research targets are briefly discussed at the end of this dissertation.

Steel braced frames are characteristically efficient seismic force-resisting systems. However, multi-story steel braced frames tend to concentrate demands in one or a few stories in response to severe ground shaking. Brace buckling and yielding results in a reduction in story strength and/or stiffness. Unless a mechanism exists to re-distribute the inelastic demands to other stories, demands tend to concentrate in the story where the inelastic response was initiated, indicative of story mechanism behavior.

Research has identified the advantages of using pivoting seismic force-resisting systems, herein termed strongback-braced frames, to mitigate story mechanism behavior. Strongback-braced frames employ an essentially elastic truss, or “strongback”, that provides an explicit mechanism of re-distributing demands to adjacent stories. Yielding and energy dissipation is provided through inelastic actions, or fuses (e.g., through brace yielding/buckling and/or beam plastic hinging). Forces and moments developed in these fuses are transferred vertically to adjacent stories by the flexural stiffness and strength of the strongback. As such, strongback-braced frames are expected to result in more uniform drift distributions, reduced peak inelastic demands, and improved design flexibility compared to conventional seismic force-resisting systems.

Despite being employed successfully in both research and practice, systematic assessment of the strongback’s behavior and practical design methods have not been developed or validated. Since the behavior of strongback systems is not characterized by the formation of story mechanisms, prior studies have found it difficult to proportion the elastic members in the strongback truss and have recognized detailing issues related to large deformation demands induced in the fuses. As such, a series of investigations were aimed at understanding the dynamic behavior and seismic performance of steel strongback-braced frames.

Archetype designs were numerically analyzed to characterize the seismic demands in the strongback elements. A four-story strongback-braced frame was used to benchmark the dynamic behavior observed during nonlinear dynamic analysis. Improved numerical models were calibrated to more realistically simulate the buckling-restrained brace response and to characterize the modeling parameters influencing brace buckling and low-cycle fatigue. The FEMA P695 methodology was used to assess potential design methods based on collapse performance. Extensive parametric studies were carried out on strongback geometries with a range of bracing configurations, ground motion characteristics, and design alternatives.

Higher mode effects were identified as the cause of substantial force amplification in the elastic strongback truss. Unlike typical yielding systems where force demands are limited by the capacity of the fuses in every mode, force demands in the strongback are characterized by a yielding first-mode “pivoting” and elastic higher-mode “bending” force demands. Since the strongback is designed to remain elastic in all modes, it can exhibit significant strength and stiffness in higher mode bending. Under the second and higher modes, the strongback truss remains elastic and continues to accumulate force demands after the fuses have yielded and as the ground shaking intensifies. These force demands in the strongback members can be significantly larger than those estimated per traditional capacity design assuming first mode-only demands.

The addition of a strongback results in improved dynamic response from typical yielding systems, including a more uniform drift profile compared to reference buckling-restrained braced frames. Based on this research, this study proposes recommendations for the design, analysis, and modeling of strongback-braced frames. Simplified static methods to estimate the dynamic demands in the strongback truss are also proposed, including modal pushover and modal enveloping analysis methods.

Seismic isolators, and more specifically, triple friction pendulum (TFP) bearings are ideal earthquake protection technologies for use in performance-based design because they can be designed to achieve multiple performance objectives corresponding to different levels of ground shaking. TFP bearings can limit structure displacement during a design basis (or maximum considered) earthquake while the still effectively isolating the structure under the service level earthquake, reducing seismic demands on the structure and its non-structural components. Furthermore, TFP bearings allow for a gradual transfer of force to the superstructure at ultimate displacement.

This dissertation presents an advanced bidirectional model for the TFP bearing that is based on the kinematic and compatibility relationships of the components of the bearing. To validate the model, experimental characterization tests were conducted on the shake table at the University of California, Berkeley. Two distinct types of experiments were conducted: (a) displacement-controlled tests in which the bearings were cycled through specified orbits and (b) unrestrained tests in which the table replicated ground motions recorded during earthquakes. The bidirectional displacement-controlled tests, which are the first of their kind, generated new knowledge to aid in the development and validation of TFP numerical models. The experimental results provided valuable insight into TFP bearing response and showed that the developed bidirectional model accurately captured the observed behavior of the TFP bearing. The advanced model, which includes translational and rotational degrees of freedom in both horizontal directions, tracks all component displacements as well as the change in bearing height during lateral displacements. The model is general so that no a priori conditions regarding bearing properties are required for the validity of the model. These properties make the advanced TFP model a valuable tool to explore the use TFP technology in novel applications.

To further the use of seismic isolation in more standard applications, simplified methods are needed for design. This dissertation investigates the use of generalized modal response spectrum methods to approximate global responses of isolated buildings without the need for costly non-linear numerical simulations. The advanced nonlinear TFP model is used for evaluation of the simplified methods.

Advanced isolation models are important and necessary for understanding the complex nonlinear dynamic behavior exhibited by seismically isolated structures, and to have confidence that performance goals are achieved under a wide variety of seismic hazards. Such models can, as shown in this dissertation, also be used to assess and improve simplified analysis methods suitable for use in routine design.