Performance-Based Seismic Bridge Design (2013)

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.

3 performance-based seismic design (PBSD) in lieu of the cur- rent force-based prescriptive design procedures. The vision is that an owner, advised by a knowledgeable designer, would be able to establish desired performance outcomes for bridges that are subject to earthquake loading. This would involve establishing performance levels in specific design earthquakes, which combined would comprise performance objectives. For instance, minor cracking, no bearing dam- age, no superstructure damage, and small permanent dis- placements in the substructure resulting from an earthquake with a 1,000-year return period might comprise a perfor- mance objective of “operational” for such an earthquake. PBSD is becoming more prominent for two reasons: 1. The engineering design and research communities have developed new knowledge and tools related to seismic performance, opening the door to improved design. 2. Public expectations of bridge seismic performance may not be in line with target goals of the seismic design specifications, thus providing an opportunity for PBSD to improve the relation between expecta- tions and target goals. This synthesis reviews the current state of practice and knowledge for bridge PBSD. DEFINITION OF PERFORMANCE-BASED SEISMIC DESIGN What Is Performance-Based Seismic Design? PBSD, or performance-based earthquake engineering as some have named it (Krawinkler and Miranda 2004; Moehle and Deierlein 2004), is a design process that attempts to link decision making for facility design rationally and scien- tifically with seismic input, facility response, and potential facility damage. The goal of PBSD is to provide decision makers and stakeholders with data that will enable them to allocate resources for construction based on levels of desired seismic performance. In such a system, performance is expressed in terms of facility loss or facility availability fol- lowing an earthquake. CHAPTER ONE BACKGROUND, OBJECTIVES, AND RESEARCH APPROACH STATEMENT OF PROBLEM Immediately following the 1971 San Fernando earthquake in southern California—a particularly damaging earth- quake for bridges—the Applied Technology Council (ATC) began to develop a rational seismic design procedure for bridges in the United States. Simultaneously, the Califor- nia Department of Transportation (Caltrans) developed an improved method for designing bridges, which augmented the existing AASHTO design specifications. The two efforts culminated in a prescriptive force-based design method that was eventually adopted nationally following the 1989 Loma Prieta earthquake in northern California. Although this force-based method is relatively simple to apply and reason- ably effective for new design, it became evident that more quantitative methods of ensuring adequate performance of bridges were needed. Caltrans subsequently adopted a displacement-based procedure for seismic design, which has as its basis a more rational assessment of displacement demand relative to displacement capacity. This method linked detailing (e.g., reinforcement layout, section con- figuration) more directly with deformation capacity than the force-based method, which is based on prescriptive measures that provide assumed, but not directly checked, adequate behavior. Taking the displacement-based procedure to the next logical level, expected performance may be quantified by linking component actions and deformations (e.g., rotations, strains) first to damage states and then to the likely postearth- quake functionality of a bridge. This linking of engineering parameters to postearthquake performance has opened the door to more meaningful evaluations of a facility’s potential use following an earthquake. Simultaneously, owners have begun to demand more knowledge of infrastructure risk, at least from the perspective of understanding likely seismic performance. Not long ago, it was an accepted maxim that it was not economical to design bridges to remain undamaged in large earthquakes. Today we are beginning to ask, “What exactly does it cost to design a bridge to deliver a limited amount of damage in a given earthquake?” The profession is beginning to be able to link performance, cost, and engi- neering metrics into a meaningful whole. It is therefore widely believed that the next logical step in the development of seismic design of bridges is to adopt

4 makers. However, at a practical level, the process may be sig- nificantly truncated in order to accomplish limited goals with the currently limited data and analytical tools. This synthesis attempts to summarize the current state of practice of bridge PBSD and to lay out a preliminary road map to a compre- hensive process that may someday provide the rational and scientific tools the profession is currently seeking. Figure 1 presents a visualization of the PBSD process, which is adapted to bridges from a figure that Moehle and Deierlein (2004) present in their description of a framework for performance-based earthquake engineering of buildings and that the authors credit to William T. Holmes of Ruther- ford and Chekene. The figure illustrates a simple pushover curve (base shear versus displacement) for a bridge. The pri- mary feature of the figure shown here is the juxtaposition of several elements: • Conceptual bridge damage states in the sketches above the curve • Performance levels (as further described in chapter six of this synthesis) – Fully Operational – Operational The Federal Emergency Management Agency’s FEMA 445 Next-Generation Performance-Based Seismic Design Guidelines: Program Plan for New and Existing Buildings (2006) describes the PBSD process as follows. Performance-based seismic design explicitly evaluates how a building is likely to perform, given the potential hazard it is likely to experience, considering uncertainties inherent in the quantification of potential hazard and uncertainties in assessment of the actual building response. It permits design of new buildings or upgrade of existing buildings with a realistic understanding of the risk of casualties, occupancy interruption, and economic loss that may occur as a result of future earthquakes. It also establishes a vocabulary that facilitates meaningful discussion between stakeholders and design professionals on the development and selection of design options. It provides a framework for determining what level of safety and what level of property protection, at what cost, are acceptable to building owners, tenants, lenders, insurers, regulators and other decision makers based upon the specific needs of a project. PBSD, when implemented at the highest level, should be comprehensive in consideration of outcomes and uncertain- ties from seismic loading and thus would be probabilistically based, providing holistic tools for designers and decision FIGURE 1 Visualization of PBSD (after Moehle and Deierlein 2004).

5 – Life Safety – Collapse • List of damage repair costs related to replacement cost • Potential casualty rate for the bridge • Estimate of loss of use of the bridge This visualization is powerful because it represents the capacity side of seismic structural response in structural performance and potential outcome terms that decision makers could use to evaluate the success of the design when loaded to various levels along the pushover curve. This simple graphic summarizes much of what PBSD attempts to provide. If this graphic were combined with seismic input for the site, then the entire PBSD method would be illustrated in one figure. (Note that the replacement costs, casualty rates, and downtime values in Figure 1 are provided solely as examples and do not represent actual figures.) Indeed, the PBSD process may be broken into four steps, three of which are included in Figure 1. These steps were con- ceived to guide the work of the Pacific Earthquake Engineering Research Center (PEER), as outlined by Moehle and Deierlein: 1. Seismic hazard analysis that quantifies the seismic input at the site in terms of intensity measures (IM), such as spectral acceleration (SA). 2. Structural analysis that relates the seismic input to structural response that is related by engineering demand parameters (EDPs), such as strains, rota- tions, displacements, drifts, or internal forces. 3. Damage analysis that relates the structural response to damage measures (DMs), which describe the con- dition of the structure, such as the occupancy or use definitions: Fully Operational, Operational, Life Safety, and Collapse. 4. Loss analysis that relates damage to some type of decision variable (DV), such as the repair costs, casualty rate, or downtime, as shown in the figure. Of course, availability, in lieu of loss, could be used for loss analyses. When these four steps are considered in the context of cur- rent design practice, it is evident that the first two steps are routinely performed. The third step is usually not considered directly, although it is inherent in the design specifications because preservation of life safety is the underlying principle of the codes. If the design code requirements are followed, then life safety will be preserved. This was the primary rea- son for the original development of design codes, whether driven by safety in the face of fire or safety from collapse. It is at this third step that our current design methodologies begin to wane with respect to PBSD, and it is important to recognize that the designer does not make choices about per- formance. Instead, he or she simply complies with the code requirements and, therefore, tacitly assumes that life safety will be ensured. In such cases, the code, not the designer or the owner, controls the performance. PBSD therefore seeks to go beyond the current level of rigor required by the design codes by having the designer and owner decide what performance is targeted from the structure under earthquake loading. Here single- or mul- tilevel seismic input may be considered, depending on the desired performance that is sought at various levels or inten- sities of strong ground shaking. However, fundamental to all designs is that life safety must be preserved in some prese- lected level of earthquake shaking. Beyond that minimum, the design may be enhanced to ensure the range of structural performance desired. Such enhancements would be selected based on data provided to decision makers who determine resource allocation based on the facility’s postearthquake functional requirements. For bridges, the long-held notion of preservation of life safety for a predetermined earthquake input has served com- munities fairly well. However, bridges can be important life- lines for communities where life safety of people who are not physically on the bridge at the time of an earthquake may be at stake. Thus, the bridge may have a postearthquake role in serving the community by providing emergency vehicle access. Such a role would suggest a higher performance objective than the basic levels included in the design codes. It is important to recognize that the design codes, begin- ning in the wake of the 1971 San Fernando earthquake, began to introduce Importance Factors, which sought to provide enhancements above basic life safety for people on the structure at the time of the earthquake by increasing the earthquake design forces. However, the force-based design methodologies used for both buildings and bridges were unable to rationally deliver the desired performance that is now sought with PBSD. With the emergence of displace- ment-based design procedures, true PBSD is more likely to be achievable. To deliver on the promise of PBSD, all four steps of the approach must be completed. This most likely will occur through development of design methodologies that permit the process to first be completed in deterministic fashion— without full consideration of the uncertainties that exist at each step of the process. Deterministic analysis is currently in use where specific strain or displacement limits are adhered to, as with the current AASHTO Guide Specifications for LRFD Seismic Bridge Design (AASHTO SGS). However, at some future point it should be possible to employ a probabi- listic approach through all four PBSD steps. The PBSD process has been conceptualized in full proba- bilistic form by PEER for both buildings and bridges. Many

6 missing pieces of knowledge and data must be addressed before such a design process is ready for deployment into practice; however, many researchers are striving to fill in those gaps. Their efforts are discussed throughout this report. In its full probabilistic form, the four-step design pro- cess may be summarized as in Figure 2. The figure clearly delineates the four steps, and the important measures and variables are as defined earlier for the overall concept. The measures are given as probabilities: p[IM], p[EDP], and so on. These probabilities depend on conditional probabilities, for example p[EDP|IM], which is read as “the probabil- ity of reaching an EDP given a value of IM.” Thus, to find the probability of annual exceedance of an EDP, p[EDP], one must combine the conditional probability of the EDP, given an IM with the annual probability of exceedance of the IM. Accordingly, this process is built up by successive combination considering the site location (O) and structure design features (D) to yield a DV that can be used to evalu- ate the adequacy of the site design. Subsequent chapters will address each major component: hazard analysis in chapter four, structural analysis and design in chapter five, damage analysis in chapter six, and loss analysis in chapter seven. It should be apparent to users of the current design pro- visions of either buildings or bridges that the profession is able to relate the seismic input probabilistically, but not the remaining three steps. For example, we currently use spectral accelerations that have a preset percent chance of exceedance in a given window of time (e.g., accelerations that have 7% chance of exceedance in 75 years for bridge design). Completing all four steps in a fully probabilistic fashion will take more effort and many refinements and additions to the current design methodologies. STUDY OBJECTIVES This synthesis project gathers data from a number of different but related areas. The current status of bridge seismic design is briefly summarized and includes meth- odologies for smaller bridges and those that have been used for larger, more important structures designed with enhanced performance objectives. The state of knowledge of large-scale laboratory performance, as well as actual bridge performance in earthquakes, is also summarized. Then, the links between measurable behavior in the lab or field and inferred performance are explored, including a limited review of analytical techniques. From this review a status of the profession today, with respect to the technical challenges of PBSD, has been developed. The intent is that this document will feed the next challenge—deciding how to employ PBSD. It is recognized that challenges beyond the technical face the implementation of PBSD. Tools for decision makers need to be developed such that engineers can provide alternatives and costs to allow informed transportation administrators to make decisions regarding the use of enhanced performance. An obvious use of PBSD would be to support the design of corridors or specific bridges that have distinct postearth- quake operability requirements. This implies that only some bridges might be designed using PBSD, particularly in the near future. PBSD could be used to augment the current life-safety minimum standard and provide enhanced perfor- mance only in selected cases. In fact, several agencies have used this approach in the past, and part of the goal of this synthesis project is to document such project-specific crite- ria for ready access by other agencies considering enhanced seismic performance for bridges. Key to this documentation are the decision-making ele- ments that feed the ultimate selection of PBSD for use on a project. Such decisions are best made with informed data of the risk posed to a facility by seismic activity. Therefore, a logical basis for evaluating risk is a probabilistic one. Work is being performed in this area, and eventually PBSD may be probabilistically based. Currently, however, a simpler deter- ministic basis may be the first logical step. Because PBSD departs from the traditional approach of using standards based on the minimum threshold of life safety, an optional transition to PBSD design may be the way forward, par- ticularly until experience with both design and construction costs is developed. FIGURE 2 Underlying probabilistic framework of PBSD (Moehle and Deierlein 2004).

7 The information gathered for this project includes the following. • Potential benefits that an owner might realize by using a performance-based seismic design to achieve enhanced performance over that available with the current design procedures. In other words, why transition to PBSD and where does it make the most sense? How can PBSD improve the profession’s delivery of infrastructure? • Definitions of performance. Data linking engineer- ing demand parameters (e.g., displacements, rotations, strains) with bridge damage and, thus, with bridge system performance are required. The data linking structural behavior to performance must also con- sider nonstructural and operational characteristics. For instance, displacements must be considered when designing utilities supported on bridges, and these util- ities must have performance goals in addition to those defined for the structure. Additionally, permanent dis- placements of the structure, which may or may not be repairable, will play into performance because such displacements can affect the postearthquake operation of the facility. • Status of PBSD research. Does enough information exist to transition to PBSD, or if not, what essential elements are currently missing? Is the information consistent for all types of bridges, including data on different superstructure, substructure, foundation, and abutment types? • Earthquake hazard level. How does earthquake haz- ard level (expressed as either chance of exceedance in a specific number of years or as return period) play into decision making for PBSD? It is well known that the earthquake input (acceleration or displacement) changes at different rates in different parts of the coun- try. The manner in which this input varies will be con- sidered relative to expected performance in different earthquakes. A single minimum level of earthquake hazard generally will not provide equal protection in all areas of the country. For instance, using a single hazard level (e.g., 1,000-year return period) may not provide the same level of protection and performance in the more seismically active western states as it does in the East because more frequent earthquakes could produce more damage in the West. • Performance in smaller earthquakes. Can we improve our designs with PBSD such that damage in smaller earthquakes is correctly anticipated and controlled during design? • A survey of the developments in PBSD from buildings, waterfront/marine (piers and wharves), and bridge perspectives to determine the overall direction in which the earthquake engineering community is mov- ing. This survey will provide an overview and long- range perspective on PBSD. A second objective is to determine how PBSD is and can be used in the nearer term for bridge design. The current design procedures stop short of true PBSD, but it is becoming possible for designers to make choices that affect the likely per- formance of bridges during earthquakes. This trend is seen in major projects that have been accomplished in the past 15 or so years. • Criteria used in previous projects. Project-specific cri- teria for projects where PBSD has been used will be reviewed. Although the scope of the synthesis is new bridge design, criteria for retrofit design will also be considered. The FHWA Retrofit Manual is based more nearly on PBSD than are the current new design pro- visions of AASHTO. Also, the methodology used in building seismic rehabilitation, which is covered in the ASCE Standard 41-06, is reviewed. Both these docu- ments provide a relatively complete methodology for applying performance-based principles. This synthesis primarily deals with the effects of strong ground motion shaking. Secondary effects, such as tsunami/ seiche, ground failure (surface rupture, liquefaction, slope failure, etc.), fire, and flood, are outside the scope of this docu- ment. Regardless, their impact on bridges may be substantial and investigation into their effects is undoubtedly important. RESEARCH APPROACH The research approach was to conduct an extensive literature and practice review on PBSD. This has been an active area of research for the past 20 years or so, and there have been numerous efforts to implement PBSD on unique, special, and/or important projects. The literature review canvassed various practice areas, with a focus on bridge, marine, and building design because much work has been accomplished in these areas. Results from one area often help “cross-pollinate” ideas and think- ing in the other practice areas. Because the approach to seis- mic design—permit, but control damage—is the same for buildings, marine structures, and bridges, reviewing these areas can bear fruit that a limited review of only bridges might not. The building practice area has also published design standards that address PBSD, particularly for reha- bilitation or seismic retrofit projects. The review covers primarily U.S.-based work, but international research and specification-development efforts were also reviewed. Bridge practice review was accomplished through a survey of all 50 states, with a particular focus on states that have higher seismic hazard. A sampling of organization-specific and proj- ect-specific design requirements was collected and reviewed. From the literature and practice review, an overview of the current status of PBSD engineering details and deployment has been assembled. A general direction of the development

8 of the practice is evident. From these, a road map forward for the bridge engineering community, including near-term research needs, has been achieved. REPORT ORGANIZATION The report is divided into 13 chapters, including this first chapter, which covers background, statement of the prob- lem, definition of PBSD, objectives of the study, and the research approach. Chapter two reviews public and engineering expectations of seismic design and discusses the regulatory framework and associated issues and challenges. Chapter three contains the resulting findings of a litera- ture review, including reviews of the bridge, building, and marine structures (piers and wharves) practice areas with emphasis on their respective design specifications. This includes an overview of current bridge seismic design prac- tice as specified by the AASHTO LRFD Bridge Design Specifications, the AASHTO Guide Specification, and the FHWA Retrofit Manual. Chapters four through seven contain the detailed findings from literature review for the four primary areas of PBSD: seismic hazard analysis (chapter four), structural analysis and design techniques (chapter five), damage analysis (chap- ter six), and loss analysis (chapter seven). Chapters eight through ten review the individual orga- nization (chapter eight) and project-specific bridge prac- tices (chapter nine), which are compared and summarized in chapter ten. The review includes descriptions of specific practices of various departments of transportation (DOTs) in the seismic design area, as well as brief descriptions of selected project-specific data that were contributed by vari- ous DOTs. Additional information was generated by a sur- vey questionnaire that was sent to all 50 U.S. DOTs, the results of which are summarized in chapter eleven. Chapter twelve identifies knowledge gaps and explores the information that is needed with respect to prediction of response and damage, as well as the information or method- ologies that decision makers need. Finally, chapter thirteen provides conclusions and suggested research, along with suggested short- and long-term implementation efforts.