ABSTRACT

In the introduction, this paper reviews and discusses briefly the nature of the earthquake (EQ) problem, the occurrence of an EQ disaster and control of seismic risks, and concludes that: although earthquakes are inevitable, their disaster potential can be reduced to acceptable socioeconomic levels through control of the built environment, because this allows us to control the potential sources of EQ hazards, which are consequences of the interaction of seismic activity (which we cannot control) with the vulnerability of the built environment (which we can control); and in spite of the tremendous increase in our knowledge in the field of EQ engineering in the last four decades, statistical studies show that seismic risks in our urban areas is increasing rather than decreasing. Reasons for this increase are identified, and it is shown that one of the main reasons is the large inventory of existing seismically hazardous facilities in our urban areas. Many of these seismically hazardous facilities were constructed when EQ engineering did not exist or was in its infancy. Furthermore, although EQ-resistant requirements in building codes have become more stringent and improved significantly in the last four decades, even current seismic codes are not infallible, and this problem of existing hazardous facilities has been markedly exacerbated by continuous uncontrolled growth of population, urbanization, and high technology industries in our urban areas. The state of the practice and of the art in assessment of the seismic vulnerability of existing facilities is briefly reviewed and discussed.

From the above reviews it is concluded that one of the most effective ways to mitigate the destructive effects of EQs is to improve present methods and develop more reliable methods for seismically repairing and upgrading existing hazardous facilities. To identify the information needed for realizing better methods, the issues implicit in the following questions are analyzed: What went wrong in the past (why so many hazardous facilities?); What is happening at present?; and What are the directions toward short-term and long-term solutions? In this analysis, special attention is paid to the history of seismic repair and upgrading of buildings in the U.S. and particularly to the last ten years, i.e., since the 1985 Michoacán EQ. This review Points out that although the urgent need for upgrading existing seismically hazardous buildings was recognized many years ago, and comprehensive research programs were formulated and proposed in the 1960s to improve the state of the art and of the practice in this area of EQ engineering, with the exception of some important studies conducted in the 1970s (after the 1971 San Fernando EQ) on highway overpasses and bridges very limited research efforts were devoted until 1986. It was the damage observed in Mexico City during the 1985 Michoacán EQ that triggered a tremendous growth of interest in supporting research in the area of repair and seismic upgrading of buildings, and several young Mexican students not only participated as research assistants in the research conducted in different U.S. universities, but also contributed significantly to improve the state of the art in this area.

Objectives. After the above review, the paper takes aim at its main objectives: a brief discussion of the problems involved in the decision to upgrade existing seismically hazardous facilities; formulation of guidelines for the selection of efficient strategies and techniques for the seismic repair and upgrading of such facilities; illustration of the applications of the guidelines and techniques to the upgrading of existing buildings; and discussion of future directions in this area of earthquake risk reduction. However, before addressing each of these main objectives, the importance and role of proper earthquake-resistant construction and the continuous monitoring of the function (use) and maintenance of the entire facility systems are emphasized, and directions toward improvements in these areas are offered.

Discussion of the problems involved in the decision to upgrade an existing facility and in the choice of the retrofitting strategy and techniques identifies at least thirteen factors affecting such decisions and selections. it is clearly noted that although basic concepts and guidelines for seismic upgrading of buildings have been formulated, the upgrading of a given facility is a unique problem requiring a customized solution. It is recommended that in selecting an upgrading strategy the designer should examine the two sides of the design equation (Demands < Supplies), and that a promising methodology for a rational selection is one based on an energy approach, specifically on the use of an energy balance equation: E_I = E_E + E_D, where E_I is the input energy, E_E is the stored elastic energy, and E_D the dissipated energy which in turn can be expressed as E_D = E_Hx + E_Hm, where E_Hx and E_Hm, are the energy dissipated through the viscous and inelastic (plastic) hysteretic behaviors, respectively.

Formulation of guidelines for selecting an efficient upgrading strategy. Based on the use of the above two equations, guidelines are offered for selecting an efficient upgrading strategy. These guidelines are grouped under the following two categories, which are interrelated: decreasing the demands, and improving the dynamic characteristics supplied to the existing building.

Selection and application of appropriate techniques for upgrading. The existing techniques are classified into two groups.- the conventional techniques, and the innovative techniques. Although conventional techniques of stiffening and strengthening existing buildings through the use of reinforced concrete shear walls and steel braces are discussed, and some applications are illustrated and discussed (particularly the proposed retrofitting of an existing six story non-ductile RC framed building infilled with URM walls, built in 1923, using a post-tensioned steel bracing system), this paper emphasizes advances since 1985 in the use of innovative techniques, particularly in the use of protective systems such as base isolation and passive energy dissipating systems.

After a brief discussion of the concept of base isolation and its advantages over the traditional fixed-base approach in earthquake-resistant design and construction, the different base isolation systems used or proposed in the U.S. air summarized, and a list of completed base isolation retrofit projects, as well as projects that are likely to go forward to construction, is offered. The response of some of the existing base-isolated buildings during recent earthquakes (1994 Northridge and 1995 Great Hanshin) is briefly reviewed.

Regarding the use of passive energy dissipation systems for earthquake protection, after the 1985 Michoacán earthquake there has been a considerable growth of interest in the U.S. in dissipating a large part of E_I through the use of specially designed and constructed energy dissipation devices. The main systems already in use or proposed are classified under the following groups: Friction, Metallic, Lead Extrusion, Shape Memory Alloys (SMAS) and Viscous and Viscoelastic. The main advantages and disadvantages of each of these systems are briefly discussed.

The application of metallic yielding steel systems to the upgrading of a two-story RC building is illustrated and discussed. The possible applications of SMAs to improve the already proposed seismic upgrading of the old six-story RC infilled with URM building by using post-tensioned steel braces, and for repair and upgrading of steel special moment-resisting frame (SMRF) buildings damaged during the 1995 Northridge EQs, are also briefly discussed.

After a brief discussion of different types of viscous and viscoelastic systems and of their main features and issues, their application for retrofitting buildings is illustrated and discussed. Viscoelastic dampers were used recently to upgrade an existing 13-story steel-framed building, which was designed in 1972 and completed in 1976. A seven-story RC building, constructed in the 1920s and already seismically upgraded in 1984, recently was upgraded again using viscous dampers.

Finally, the main conclusions that have been drawn from the reviews conducted are presented, and recommendations for research, development, education and implementation needs are formulated.

INTRODUCTION

INTRODUCTORY REMARKS REGARDING THE NATURE OF THE EARTHQUAKE PROBLEM, THE OCCURRENCE OF AN EQ DISASTER, AND CONTROL OF SEISMIC RISKS

The Nature of the EQ Problem: Press [1989], in discussing this, pointed out that "Earthquakes (EQs) are a very Special type of natural hazard in the sense that they are very rare, low-probability events, whose consequences, when they do occur, are very large in terms of destruction and suffering." EQs are natural disasters whose feature is that most human and economic losses are not due to the EQ mechanisms, but to failures of human-made facilities: buildings and lifelines, such as dams, bridges, transportation systems, etc., which supposedly were designed and constructed for the comfort of human beings. Although this is depressing, it is also fortunate and encouraging, because it tells us that in the long run the EQ problem is in principle solvable. With sufficient resources for research and development (R&D), formulation of EQ preparedness programs, and education needed for the use of the results of this R&D, EQs are hazards to which it is in our power to respond effectively. We can reduce their threat over time by as much as we want We can learn where not to build and how to build so that facilities will not fail

The psychological impact on millions of people who experience major EQs is an enormous, complex fear that remains a nightmare to them for many years. Thus, it is important that attempts be made to find the reasons for EQ disasters and to eliminate or reduce the potentially catastrophic consequences of major EQs.

Occurrence of an EQ disaster. Four conditions determine the occurrence of an EQ disaster in an urban area: (1) the magnitude of the EQ, (2) the distance of the source from the urban area, (3) the size and distribution of the population and the economic development (high technology industries), (4) and the degree of EQ preparedness of the urban area. Obviously, the potential of an EQ disaster increases with increases in the magnitude of the EQ, the proximity of the EQ source to the urban area, the size of the population, the economic development, and the poverty of the preparation.

Clearly, EQ hazards depend not only on the seismicity of the region, but also on its population density, economic development and degree of preparedness. In this sense, EQ hazards are becoming more important each year, and therefore is it not surprising that recent statistical studies [Bertero, 1992] show that seismic risk in our urban areas is increasing rather than decreasing. Seismicity remains constant but in spite of tremendous advances in our knowledge in the field of EQ Engineering, we have not counterbalanced the uncontrolled and rapid increases in population, urbanization and economic development of our urban areas with advances in our knowledge and particularly with the improvement of earthquake preparedness. Analysis of what has happened since 1984 shows that we are very far from reducing the seismic risk in our urban areas to socio-economically acceptable levels [Bertero, 1992]. From this analysis it is concluded that: EQs are inevitable, but EQ disaster can be controlled. The fault rupture that originates the significant EQs does not itself kill people or induce great economic losses. What causes most of the injury and economic losses is the interaction of the EQGMs with the built environment. What is needed is to control seismic risks in our urban areas by controlling the built environment, and this should be a main objective of EQ preparedness programs.

Control of seismic risks. To learn how to control seismic risk, it is necessary to define it. According to the glossary of the EQ Engineering Research Institute' Committee on Seismic Risk (1984), seismic risk is "the probability that social or economic consequences of EQs will equal or exceed specified values at a site, at various sites, or in an area, during a specified exposure time." According to Dowrick [1987], seismic risk is an outcome of seismic hazards, as described in the following relationship, which is also illustrated in the flow chart of Fig.1.

From the above considerations, it is clear that to control seismic risk at any given site it is necessary to estimate seismic risk, which requires the following assessments.

• First, estimation of the seismic activity at the site, which requires identification of all sources of EQGMs that could affect the built environment, i.e., that could induce damage. Once the sources of various potential seismic hazards created by the seismic activity have been identified, it is necessary to determine whether the EQs will be single or multi-events, and to estimate their moment magnitudes, recurrence periods and the attenuations of the intensities of the EQGMs with distance.
• Second, prediction of whether the EQ faulting generating the EQGMs could induce any of the following potential seismic hazards at the site or in the surrounding region: surface fault ruptures, tsunamis, seiches, landslides, floods and ground failure.
• Third, prediction of the time history of the six components of the EQGM at the site and at the foundation of each facility.
• Fourth, prediction of whether the predicted EQGMs can induce ground failure, i.e., liquefaction, settlement, subsidence, differential compaction, loss of bearing and shearing strength, lateral spreading, landsliding and/or lurching.
• Fifth, assessment for any given facility of the mechanical behavior (performance) of the whole facility system under the predicted six components of the EQGM at its foundation, estimating the degree of damage and losses, considering the possibility of fire, flood and other consequent or indirect sources of seismic hazards.

Sixth, evaluation of the economic consequences of the losses and the socioeconomic impact on the community. The costs and benefits of seismic upgrading of existing hazardous facilities should be estimated.

Furthermore, the problem of seismic risk reduction will not be solved just by the acquisition of knowledge through research. Research must be accompanied by the needed technological developments, and the needed knowledge and the developments must be implemented in practice.

What is needed is a translation of current engineering and architectural know-how into simplified options that can answer the social, political and economic concerns. This will require not only a multidisciplinary approach, but also a comprehensive education program, not only for owners and users but for all of the different audiences that in one way or another are involved in the implementation of the seismic risk reduction measures, The education program should emphasize the importance of EQ disaster preparedness, including preparing for fires, control of panic, etc.

Until now, most of the emphasis has been on (1) trying to predict EQs based on probabilistic approaches, and (2) gaining knowledge of the mechanical behavior (performance) of different facilities. While these are necessary, they are not sufficient and prediction alone will not solve the problems. What is necessary is to improve the preparedness of the public against EQ disaster. There is an urgent need to coordinate the acquisition, processing, evaluating and synthesizing of the research results already available, and the knowledge gained through lessons learned in past EQs. This integrated knowledge must then be converted into action. This will not be achieved unless multidisciplinary groups of researchers, practicing professionals, users, government officials, etc., develop and ensure the implementation of reliable and suitable policies and strategies for the reduction and control of seismic risks to acceptable levels.

To summarize: the main issue confronting all of us interested in solving the EQ problem is the need to control the seismic risk in our urban and rural areas. The solution is controlling the vulnerability of the built environment, because this allows us to control the potential sources of EQ hazards, which are consequences of the interaction of seismic activity (which we cannot control) with the vulnerability of the built environment (which we can control).

STATE OF THE PRACTICE IN SEISMIC VULNERABILITY ASSESSMENT. Most methods for conducting seismic vulnerability assessments of existing structures are based on estimates of capacity/demand ratios using current seismic code-specified criteria. This approach is not satisfactory because even in present seismic regulations the reliability of the procedures for evaluating seismic demands and capacities is highly questionable. Bertero [1988] and Miranda [1991] have conducted reviews of the available criteria for evaluating the vulnerability of existing structures.

The real problem in assessing the vulnerability of a given facility is in estimating the response to future critical EQGMs. Prediction of seismic response depends on an adequate knowledge of at least (1) the seismic activity at the site, (2) the sources of seismic hazards, which depend on the seismic activity, local soil conditions, and the type, size, shape and detailing of the foundation, superstructure and nonstructural components of the facility and their mechanical characteristics under dynamic excitation; and (3) the desired level of safety and/or the acceptable level of damage. The difficulties involved in reliably assessing the vulnerability were clearly in evidence in the studies assessing the vulnerability of the Cypress Viaduct, of which a large part collapsed during the 1989 Loma Prieta EQ [Miranda and Bertero, 1991]. In spite of the simplicity of the structure, the availability of "as built" drawings and the fact that tests of the structural materials were conducted, it was not possible to estimate the actual strength of the weakest frame because in the real existing structure the detailing and particularly the anchorage length of some main bars were not as specified in the "as built" drawings.

In spite of progress in the above assessments, considerable uncertainties remain. The importance of reliable assessments of seismic activity and hazards and of the vulnerability of the existing facilities cannot be overemphasized. If the expected intensity of EQGMs is greatly overestimated, the costs of new construction and seismic rehabilitation of existing structures may be excessive. On the other hand, if the intensity of EQGMs is seriously underestimated, the result may be costly damage and loss of life, such as that seen in the disasters of Tangshan, Mexico City, Armenia, Loma Prieta, the Philippines, Erzincan (Turkey), Northridge, and Kobe. Similar results attend the gross over- or underestimation of the vulnerability of existing facilities.

In the U.S., until the 1989 Loma Prieta EQ, not only were the assessments of seismic activity, hazards, vulnerability and consequently, risk, based on the use of building code seismic regulations for Earthquake-Resistant Design (EQ-RD) of new buildings, but also the common philosophy for upgrading existing buildings was to bring them into compliance with such regulations. This approach usually results in technically inefficient rehabilitation of buildings, and in some cases, prohibitively uneconomical upgrading solutions. For example, many old concrete buildings are condemned because they have structural systems and/or reinforcement detailing which are not ductile enough to be acceptable under present seismic codes, although they may possess sufficient lateral stiffness and particularly overstrength over the code-required yielding strength (C_y) to remain elastic under the effects of the maximum credible EQGM. The main difficulty in proving that they need not comply with the code is in the evaluation of their actual dynamic characteristics, i.e., in assessment of their real seismic vulnerability. Often, bringing the building up to compliance with building code seismic regulations does not guarantee good seismic performance, particularly when damage control is desired under the expected future major EQGMs.

The basic philosophy of present building code seismic regulations is to protect the public in and about buildings from loss of life and serious injury during major earthquakes. These regulations are not intended to limit damage, maintain functions, or provide for easy repair. The EQs, particularly in the last decade, starting with the 1985 Michoacán, followed by the 1989 Loma Prieta, and recently the 1994 Northridge and 1995 Kobe earthquakes, clearly indicate that seismic codes ought to consider damage control: the level of "acceptable damage" should vary with the function (occupancy category) of the building. The need for code specifications that require damage control is urgent for certain occupancies. A comprehensive approach to the upgrading of existing buildings will require that at least the following three levels of critical upgrading EQGMs be considered: Service Level, Functional or Operational Level ' and Safety Level. Guidelines for establishing these different levels of design EQGMs have been recommended by Miranda [1991] and Bertero and Bertero [1992] and recently recommended by SEAOC Vision 2000 Committee [1995].

From the above discussion, it is clear that good EQ-RC of facilities (engineered and non-engineered) is at present the key element in EQ hazard reduction. One of the most effective ways to mitigate the destructive effects of EQs is to improve present methods and develop more reliable methods for designing, constructing and maintaining and monitoring new structures and particularly seismically upgrading existing hazardous facilities. To find out what information is necessary for the realization of such improvements, it is convenient to analyze the main issues.

MAIN ISSUES. These can be expressed as questions about: what went wrong in the past (why are there so many hazardous facilities?); what is happening at present (where do we stand right now?); and where we should go (what are the directions for short-term and long-term solutions?) Does the increase in seismic risks mean that EQ Engineering in general and our knowledge of how to design and construct EQ-resistant structures and upgrade existing ones has not advanced? No, that is not the case. Although EQ Engineering is a relatively new branch of engineering research, advances in this field have already played a significant role in reducing seismic hazards through improvement of the built environment, making possible the design and construction of EQ-resistant civil engineering structures (highway structures, dams, pipelines, critical facilities, high-rise buildings, etc.) and improving the seismic safety of non-engineered construction.

Rather, the answer can be found in the following circumstances. First, many studies have shown that the greatest threat to life safety arising from moderate-to-severe EQs occurring near urban areas is posed by existing hazardous structures. Many hazardous structures and facilities were constructed when EQ Engineering was in its infancy. Furthermore, although EQ resistance requirements in building codes have become more stringent and improved significantly, even current codes are not infallible. Second, this problem of existing hazardous civil engineering structures has been markedly exacerbated by continuous uncontrolled population growth, increased urbanization, and the development of high-technology industries in our urban areas.

The seismic response of any facility (structure), and therefore the degree of damage that it will suffer, depends on the mechanical status of the whole building system (soil-foundation, superstructure and nonstructural components and contents) when the EQ occurs, i.e., response depends not only on how the building has been designed and constructed, but also on how it has been maintained up to the time that the EQ strikes. Thus, the principal issues that need to be considered in order to improve EQ-RC, and therefore to reduce the seismic risks in our urban areas, are the ones grouped into the following categories.

• Improvement of the EQ-RD of the whole facility system (soil, foundation, superstructure, and nonstructural components and contents).

• Improvement of the construction of the foundation, the superstructure and the nonstructural components (which can become unintentional structural components).

• Improvement of the maintenance and function, together with all alterations, repair and/or upgrading of the entire soil-building system during its service life.

In 1984, during the 8WCEE, while there were some reports by U.S. authors concerning seismic retrofitting guidelines for highway bridges as well as retrofitting of bridges using base isolation concepts, there was only one report on retrofitting of historical buildings, on the other hand, there were several reports by Japanese authors on retrofitting RC buildings. The growth of interest in supporting research in this area started in 1968 as a consequence of the need to repair and strengthened several damaged reinforced concrete buildings [Endo, 1982]. This interest received a boost in 1978 with the observed damage during the Miyagiken-oki EQ and in 1980 as a consequence of the publication of the Earthquake Control Guidelines for Building Installations by the Shizouka Prefecture, which is located in a region that could be severely affected by the predicted Tokai EQ.

It was the damage observed in Mexico City during the 1985 Michoacán EQ that triggered in the U.S. and of course in Mexico a tremendous growth of interest in supporting research in the area of repair and seismic upgrading of existing buildings. Through an informal cooperative agreement among the U.S. National Science Foundation (NSF), the Government of Mexico and the Consejo Nacional de Ciencia y Tecnología (CONACyT), several analytical and experimental research projects on repair and/or upgrading were conducted in Mexico and at different universities in the U.S. (California, Illinois [Lehigh], Michigan and Texas). Several young Mexican engineers participated in the research conducted at the U.S. universities, contributing significantly toward the improvement of the state of the art in this area of EQ Engineering.

The above growth of interest was evident already in the 1986 Third U.S. National Conference on Earthquake Engineering, at which 13 papers discussed different aspects of repair, strengthening, isolation, and retrofit of structures, revealing the significant increase in research activities in this area. This increase in interest in seismic upgrading was even more evident during the 9WCEE in 1988, where 56 papers in the areas of seismic capacity assessment, repair and strengthening of structures were presented, of which eleven papers were written by U.S. engineers. In 1994, at the Fifth U.S. National Conference on EQ Engineering, forty-nine papers were presented in the sessions on damage assessment, repair and strengthening of structures.

MAIN OBJECTIVES OF PAPER. Because seismic upgrading is one of the most effective ways to reduce the seismic risk in our urban areas, the main objectives of this paper are: to present a brief analysis of the problems involved in the decision to upgrade existing seismically hazardous facilities; to illustrate applications of the available guidelines and techniques to the upgrading of existing buildings; to formulate guidelines for the selection of efficient strategies and techniques for the seismic repair and upgrading of such facilities; and to formulate recommendations for research needs in the area of earthquake risk reduction. However, before addressing these main objectives, the importance of proper EQ-resistant construction and continuous monitoring of the function (use) and maintenance of the entire facility systems is emphasized, and directions toward improvements in these areas are offered.

ROLE AND IMPORTANCE OF PROPER EQ-RD, EQ-RC AND CONTINUOUS MONITORING OF THE FUNCTION (USE) AND MAINTENANCE OF FACILITIES

It is well known that while a sound EQ-RD of any given new structure is necessary, it is not sufficient to ensure a satisfactory EQ-resistant facility. Similarly, while the selection of a proper strategy and technique for the EQ-resistant upgrading of an existing hazardous facility is also necessary, it is not sufficient to guarantee a good seismic performance. The seismic response of a facility depends on the state of the whole facility system [soil-foundation-superstructure and nonstructural components and contents (particularly those that become unintentional structural elements)] at the moment that the earthquake shaking occurs. In other words, the seismic response depends not only on how the structure has been designed, or redesigned if upgraded, but also on how it has been constructed, maintained and used (the kind of occupancy or function) up to the time that the EQGMs strike its foundation, and how it interacts with the surrounding facilities during such ground motions. A design can only be effective if the model used to engineer the design can be and is constructed maintained and properly used.

Although the importance of construction, maintenance and proper use in the seismic performance of structures has been recognized, insufficient effort has been made to improve them (e.g., through improving supervision and inspection in the field). Design and construction are interrelated. If good workmanship is to be achieved, the detailing of members and their connections and supports must be simple. Field inspection has revealed that a great deal of damage and failure is due to poor quality control of structural materials and/or poor workmanship - problems that would not have arisen if the building had been properly inspected during construction. A main factor in the failure of several buildings during the 1964 Alaska, 1971 San Fernando, 1985 Mexico, 1986 San Salvador, 1990 Philippines and 1992 Erzincan EQs was the poor quality of concrete and the poor workmanship in the detailing and placement of the reinforcement. Poor workmanship in the connections was the main reason for the failure of many industrialized (prefabricated) buildings during the 1988 Armenia EQ, and of several parking facilities during the 1994 Northridge EQ. In many other cases, damage may be attributed to improper monitoring of the function of the building as in the case of several buildings during the 1985 Mexico EQ. Some of these buildings were built for offices or residences, but were later used to shelter lightweight industries. Similarly, many observed failures of buildings have been due to improper maintenance during their service lives. Inappropriate alteration, repair and upgrading (retrofitting) of the structure and the nonstructural components can lead to severe damage during moderate and major EQs. Several buildings that were repaired and upgraded after the 1971 San Fernando EQ, as well as some recently retrofitted masonry buildings in Los Angeles area, suffered significant damage during the 1994 Northridge EQ. It is necessary to monitor the health of the constructed facilities continuously.

Directions toward improvements of EQ-RC and monitoring and maintenance of buildings. Seismic codes should regulate both the supervision of construction and the monitoring of the function, maintenance and repair and upgrading of the entire building system, and the enforcement of these seismic code regulations should not be lax under any circumstances. Furthermore, there is an urgent need to develop code guidelines and redesign procedures for efficient repair and upgrading of existing hazardous buildings.

SEISMIC UPGRADING OF HAZARDOUS FACILITIES

The decision to upgrade an existing facility and the choice of the retrofitting strategy and techniques depend on many factors, including not only assessments of (1) the seismic activity; (2) the potential sources of seismic hazards; and (3) the main seismic hazards given the vulnerability of the facility's whole system; but also (4) the type, function, age and/or repair of the facility, with particular emphasis on how much of its original capacity to absorb and dissipate energy remains; (5) the required or desired levels of performance (serviceability, continuous operation, life safety) expected of the upgraded facility for different levels of EQGMs and seismic hazards that can occur during the expected service life of the facility; (6) architectural requirements; (7) the need to n-minimize disturbance to the occupants and operation of the facility during the upgrading; (8) selection of the best strategy for rehabilitating the whole facility system; (9) development of alternate rehabilitation schemes; (10) the availability of equipment and expertise for the field work; (11) assessment of the vulnerability of the alternate upgrading schemes to the identified sources of seismic hazard; (12) cost vs. benefits of the upgrading work, and the socioeconomic impact on the community; and (13) selection of the solution that is most efficient technically and economically.

Selection of the Proper Upgrading Strategy. Selection of the most efficient upgrading strategy requires thorough study of the factors discussed above. There are many uncertainties involved in these studies. Although basic concepts and guidelines for seismic upgrading of structures have been formulated [Bertero and Whittaker, 1989; Jirsa and Badoux, 1990; Bertero, 1992], the upgrading of a given facility is a unique problem requiring a customized solution. For two identical buildings with different occupancies or functions, located on sites with different soil conditions, and with different histories of damage (i.e., how much of the original capacity to absorb and dissipate energy remains), the upgrading strategies can be vastly different.

In selecting the upgrading strategy the designer should examine the two sides of the design equation:

DEMANDS < SUPPLIES (2)

This equation shows that the designer can seismically upgrade an existing facility by(1) decreasing the seismic hazard demands; (2) improving the facility's supplied mechanical (dynamic) characteristics; or (3) a combination of (1) and (2), As discussed in detail by Bertero and Terán [1993], a promising methodology for rational selection of upgrading strategies for existing hazardous facilities as well as for EQ-RD of new structures is one based on an energy approach, specifically on the use of an energy balance equation. Although Uang and Bertero [1988] have shown that two different basic energy equations can be derived from the basic equation of a viscous damped Single Degree of Freedom System (SDOFS) subjected to an EQGM, they have also shown that the maximum values of the Energy Input (E_I) are very close in the period range of practical interest for buildings, which is 0.3 to 5.0 s. Therefore, the two basic energy equations can be rewritten as:

where E_E is the stored elastic energy and ED the dissipated energy. A comparison of this equation with the design equation (2) makes it clear that E_I represents the demands, and the summation of E_E + E_D represents the supplies, From analysis of Eqs. 3a and 3b it is clear that a good estimate of the E_I for the critical EQGM is the first step in selecting an efficient upgrading strategy. Next, the designer has to analyze whether it is possible to meet this demand just by keeping the behavior of the structure in the elastic range, or whether it is convenient to try to dissipate E_I as much as possible, i.e., to use large E_D (E_I = E_E + E_D).

As shown in Eq. 3b, there are three ways to increase E_D: one is to increase the hysteric damping energy (E_Hx) by increasing the equivalent viscous damping coefficient, x, another is to increase the plastic hysteretic energy (E_Hm); and the third is to increase both E_x and E_Hm.. It is common practice at present to try to increase E_Hm through inelastic (plastic) behavior alone, which implies damage of the structural members. It has recently been recognized that it is possible to increase E_D significantly with energy dissipation devices such as viscoelastic and viscous dampers (viscoelastic shear, or oil) and hysteretic dampers (friction, yielding metals and lead) [Bertero & Whittaker, 1989; Kelly and Aiken, 1991; Hanson, 1993]. Although the use of dissipation devices based on friction and yielding of metals does improve behavior efficiently at safety level, where some level of damage is tolerable, viscous dampers have the great advantage that they can be used to control the behavior of the upgraded structure under both safety and service levels.

If technically and/or economically it is not possible to balance the required E_I through either E_E alone or E_E + E_D, the designer has the option of attempting to decrease the E_I to the structure with base isolation techniques. Combining base isolation techniques with energy dissipation devices is a very promising strategy both for upgrading existing hazardous structures and for EQ-RD of new structures. (As will be discussed later in more detail, structural systems based on base isolation and passive energy dissipation systems have lately been classified as Seismic Protective Systems).

The energy approach requires the reliable selection, for each of the limit states that need to be considered (service, function, and/or safety), of the critical design EQGM that controls the design, i.e., which has the largest damage potential for the structure. Most of the parameters used to establish design EQs are unreliable for assessing the damage potential of EQGMs. A promising parameter for this assessment is E_I, However, as discussed by Bertero and Uang [1992], E_I alone does not give a clear picture of the E_D that has to be supplied to balance the E_I for any specified acceptable damage, and the additional information given by E_Hx + E_Hm m_{a. (cumulative ductility ratio), NYR (Number of Yielding Reversals) and NEYC}m_{max. (Number of Equivalent Yielding Cycles at} m_{max) spectra is needed. Bertero and Terán [1993] discuss simplifications in the use of energy concepts both for EQ-RD of new structures and for the upgrading of existing seismically hazardous facilities.}

_{GUIDELINES FOR SELECTING EFFICIENT UPGRADING STRATEGY. From analysis of the design equations (2 and 3), it is obvious that three main strategies can generally be considered for upgrading a seismically hazardous facility: (1) decreasing the earthquake demands [EI in Eq. (3)1; (2) improving the facility's supplied mechanical (dynamic) characteristics [EE and/or ED, in Eq. (3)]; and (3) both (1) and (2). It should be noted that strategies (1) and (2) are not independent, because the demands of seismic excitations depend on the mechanical characteristics supplied to the whole facility system. The following main guidelines should help with the selection of efficient upgrading strategies. In applying these guidelines it should be kept in mind that although damage depends on both strength and deformation, damage is more a consequence of deformation than of strength, and thus that to control damage it is desirable to limit deformations, particularly the tangential interstory drift.}

_{1. Decreasing the demands. If the effects of the expected critical EQGMs at the site at the different limit states are reliably defined by their response spectra [EI, Cy (yielding strength) and Sd (displacement), as illustrated in Fig. 2], the demands on the structure depend on the relation between the period of the structure, T, and the predominant period of the EQGM, TGM . The largest response, and thus the largest demands on EI and Cy . usually occur when T/TGM=1. Thus, one main strategy for decreasing demands is:}

_{Fig. 2. Qualitative illustration of the following response spectra: EI (energy input), Cy (yielding strength coefficient), and Sd (displacement) corresponding to firm and soft soil earthquake ground motions and the variations of these spectra when for (1) given a ductility ratio, m, the damping ratio, x, varies from 1% to 20%; and (2) a given x the m varies from 1 to 6.}

_{1a. Shifting the T of the whole facility system. How? By increasing or decreasing T depending on the ratio of T/TGM. If T < TGM and the structure is on soft soil, the solution is usually to decrease T, because, as can be seen from Figs. 2(b), 2(d), 2(f) and 2(h), this decreases the demands. If the structure is on firm soil, with a T close to TGM , and its is very small, then the solution is to increase T with base isolation, because, as can be seen from Figs. 2(a), 2(c), 2(e) and 2(g), this decreases the and EI demands. However, Figs. 2(e) and 2(g) show that the increase in T can lead to a considerable increase in lateral displacement (Sd) demands - but this increase is concentrated in the base isolation elements, which are specially designed for such large displacements. in general, if base isolation cannot be used, T should be decreased by increasing the supplied stiffness and/or decreasing the mass, because both lead to smaller deformations and therefore to better control of damage. In practical applications of this strategy, the extent to which the T of the original structure has to be shifted depends on the reliability with which T and TGM can be predicted.}

_{1b. Another strategy for decreasing the demands is to increase the damping ratio x [Figs. 2(a), 2(b), 2(e) and 2(j)], and/or to lower the Cy., by increasing m [Figs. 2(c), 2(d), 2(g) and 2(h)]. In the case of soft soils and where T < TGM, the increase in u could lead to larger values of input energy [Fig. l(d)]. Comparison of Figs. 2(e) and 2(f) with Figs. 2(g) and 2(h) shows that when it is necessary to control deformations (Sd) at either service or safety level, and where T < TGM, it is better to increase} x than to increase m, because at service level no inelastic deformations are acceptable, and at safety level an increase in m results in an increase in S_d.

2. Improving the dynamic characteristics supplied to the existing structure. The demands depend not only on the dynamic characteristics of the EQGM, but also on the interaction of these characteristics with the dynamic characteristics supplied to the structure. Some guidelines for improving the supplied dynamic characteristics follow. The legitimacy of these guidelines is supported by the plots of Fig. 2.

2a. Decrease the reactive mass. Why? Because it can result in a reduction of EQ inertia forces, and so will reduce the EQ's demand (this is particularly true when T < TG). How? By removing unnecessary weight, reducing the weights of walls, partitions, and so on -- in a multistory building, removing one or more of the upper stories.

2b. Minimize the distance between the centers of mass, stiffness and resistance. Why? To reduce torsional effects. How? By properly modifying the values and/or dispositions of reactive mass, stiffness and yielding resistance of the structural elements.

2c. For a flexible building on deep soft soil deposits, reduce the fundamental period. Why? To avoid resonance problems with long-period ground motions, How? By reducing the building's mass or increasing its stiffness or both. 'Me stiffness can be increased efficiently with steel bracing and/or RC shear walls.

2d. For a building lacking ductility, tie all of the lateral load resisting elements together to maximize the building's yielding and maximum strengths, or increase the damping, or both. Why? It is difficult to improve the ductility of existing structural members, but relatively easy to increase a building's strength, reducing its required ductility. How? By a variety of means, e.g., using prestressed rods to tie the roof and floor slabs to the gravity load supporting elements and/or as bracing elements to enhance the building's lateral stiffness and strength, In the case of building isolated from adjacent buildings, buttresses and outriggers can be used, as can dampers.

2e. For a stiff structure lacking the necessary strength and ductility and that can be subjected to earthquake shaking with high frequency (small TGM), increase the building's fundamental period or increase its damping, or both. Why? To reduce EQ strength demand by avoiding engineering resonance, thus decreasing the dynamic amplification factor. How? By some combination of eliminating infills, isolating (separating) nonstructural elements from the structure, adding dampers, using base isolation systems and using energy dissipation devices.

2f. For a flexible building lacking ductility and located on soft soils and where interstory displacements are the major concern, increase the building's stiffness, strength, and damping. Why? An increase in stiffness will decrease T, which, together with an increase in strength, will reduce deformations and ductility requirements; the increase in damping will reduce the required strength and the deformations. How? Through the use of braces in the form of prestressed rods, concentric braces and/or eccentric braces; or shear walls; or energy dissipation devices. If the T of the building cannot be shifted very much, i.e., the stiffness and/or the reactive mass cannot be changed significantly, the supplied dynamic characteristics can be improved by a change (usually increase) of the: strength (C_y); m; x, or a combination of these. Jirsa and Badoux (1990) formulate redesign strategies based on strength-ductility relationship, A more promising strategy is to increase x by adding viscous dampers.

2g. In selecting the retrofitting strategy, careful consideration should be given to the entire soil-foundation-superstructure-nonstructural component system, and not just to the superstructure. Evaluation of the adequacy of the foundation is an essential step in the selection of the most appropriate upgrading strategy.

SELECTION OF APPROPRIATE TECHNIQUE AND FINAL DESIGN. After the selection of the proper strategy, schemes by which it can be implemented must be developed and analyzed. The final implementation scheme must consider not only the technical aspects and the total cost of the upgrading, but must also minimize disturbance to the function of the building during upgrading. The latter consideration requires upgrading strategies that either (1) involve construction activity only on the external faces of the building, or (2) require only minimal reconstruction work inside the building. Generally, the final upgrading strategy is a compromise among the optimal technical strategy, the strategy that demands the smallest construction costs, and the strategy with minimum disturbance to the building's occupants.

Seismic Upgrading Techniques. Although there are many techniques that have been either already applied in practice or that have been suggested and applied in proposed projects, these techniques will be classified herein into just two main groups: the conventional techniques, and the innovative techniques.

• Conventional Techniques. In the U.S. most of the conventional or traditional techniques for seismically upgrading existing facilities, particularly buildings, consists in stiffening and/or strengthening the structures of such facilities and in most of the cases these have been achieved through the use of structural walls (most of them of RC) and/or braces (most of them of structural steel).

• Innovative Techniques. Since the 1960s there has been a growing interest in searching and developing new EQ-resisting systems for new civil engineering facilities based on controlling their seismic responses. These new systems have lately been called "protective systems," such as base isolation and passive energy dissipating systems. Since the late 1970s and more particularly since the 1985 Michoacán EQ there has also been a growth of interest in using such protective systems for seismically upgrading existing hazardous buildings. The concepts involved in the development of these protective systems and their applications for seismic upgrading are discussed later.

FIELD CHANGES. As a result of non-documented changes and additions to existing buildings, many more field changes are generally required for rehabilitation work on existing buildings than for construction work on new ones, entailing a need for clear, detailed drawings for the retrofitting of a building. The construction must be monitored closely to ensure that the retrofitting details and the intent of the retrofitting strategy are carried out -- thorough inspection is an integral part of any retrofitting project.
 

APPLICATION OF THE CONCEPTS AND GUIDELINES TO THE UPGRADING OF SOME EXISTING BUILDINGS

The author and his research associates have conducted a series of detailed studies on the seismic upgrading of existing hazardous structures [Bertero & Whittaker, 1989; Miranda and Bertero, 1990; Miranda, 1991; Terán et al., 1995]. For a more detailed discussion of these studies, the reader is referred to the above references and to Bertero [1993]. Herein only the application to a few of the buildings studied will be summarized.

In order to select an efficient strategy, the designer must begin by identifying the main response parameters which are unacceptable or inadequate and therefore need to be improved (have their values upgraded). These parameters are usually the following:

• Strength, usually represented by the base shear V = C_SW.
• Deformation: overall displacement (S_d); floor displacement or drift; story drift or interstory drift index (IDI); and tangential IDI.
• Energy dissipation capacity (E_D = E_Hx + E_Hm), usually represented by the damping coefficient, x(E_Hx º E_x) , and by the ductility ratio, m (global or local or both).
• Damage function or damage index, which depends on deformation and energy dissipation.
• Floor acceleration and velocity.

If some damage can be tolerated, then it is possible to dissipate some energy (E_D) through plastic hysteretic behavior, by accepting a certain amount of E_Hm through the use of acceptable p. The designer, then, has at his or her disposal all of the above dynamic characteristics to play with -- V, K, x, T (K and M), and p. However, in order to use these tools it is necessary to have a clear picture of how the values of each of these dynamic characteristics affect the main response parameters, strength, deformation, energy dissipation, damage function or damage index, and acceleration and velocity.

SEISMIC UPGRADING OF AN OLD RC FRAME BUILDING INFILLED WITH UNREINFORCED MASONRY WALLS USING POST-TENSIONED STEEL BRACES

INTRODUCTORY REMARKS. Unreinforced masonry (URM) buildings and framed buildings infilled with URM walls and/or partitions, designed and constructed before the development and flourishing of seismic design, constitute an important part of the vast inventory of high-risk structures in most cities in regions of moderate to high seismicity. There is a need to develop technically and economically efficient retrofitting schemes to upgrade these buildings in such a way that they can have adequate performance during strong EQGMs.

In recent years, several researchers and practitioners have shown that the seismic performance of existing buildings when subjected to strong EQGMs can be enhanced considerably by bracing the buildings with post-tensioned (PT) rods or cables. The use of this upgrading technique yields several advantages, such as versatility, low cost, fast and clean construction, and does not add any significant reactive mass to the existing facility. The implementation of this technique to the upgrading of framed buildings with URM infills will probably yield large economic advantages in the rehabilitation of these buildings. Nevertheless, there are many aspects and issues that need to be studied and resolved before attempting such implementation.

Terán et al. [1995] have reported the studies conducted to shed light on the solutions of the above problems. The main objectives of the studies were: First, to identify, study and discuss relevant issues in the evaluation of the seismic hazards of non-ductile frames infilled with URM walls; second, to investigate the use of PT steel braces to reduce these seismic hazards in framed buildings with URM walls located in regions of high seismic risk in California; third, to study and discuss the issues that need to be considered during the design process to attain efficient (technically and economically) retrofitted facilities using this technique; fourth, to assess the use of this technique by studying the seismic performance of a specific building with non-ductile reinforced concrete (RC) frames and URM infills before and after it has been upgraded with PT braces; and fifth, to offer some conclusions drawn from the study and recommendations regarding the research that is needed to improve the application of such a technique.

USE OF STEEL BRACES IN THE SEISMIC REHABILITATION OF EXISTING BUILDINGS. A review of the literature available on this subject reveals that the interest and the application of this rehabilitation technique have increased significantly in the last ten years. From this review, there is no doubt that the seismic rehabilitation or restructuring of an existing building using steel braces is an attractive option, and many different types of steel structural elements (members) and arrangements of these members can and have been used. Usually it is possible to achieve large increases in the lateral stiffness and strength of an existing building. The use of this technique offers the following advantages:

• Stiffness and deformation capability of the bracing system. A very attractive aspect of the use of steel braces to upgrade an existing building is the wide range of stiffness that can be considered in the design of the bracing system. Once the stiffness of the existing structure is evaluated, a bracing system with adequate stiffness can be developed such that the original system is allowed to resist a portion of the lateral forces induced by EQGM. In some cases, it is important for the existing structure and the braces to reach their ultimate strength simultaneously (i.e., at similar levels of deformation), Designing the bracing system with these characteristics will usually result in efficient EQ-RD, as shown in Figure 4a. In other words, it would not be efficient to reach a level of deformation at which the original elements of the structure start to fail, while the braces still remain far from reaching their ultimate capacity, as shown in Figure 4b. It will not be desirable in every case to accomplish compatibility of stiffness and/or deformation, as in the case where the purpose of the bracing system is to unload the existing elements as much as possible (Figure 4c).

• Loads induced in the foundation. Under normal conditions, it will be possible to distribute the braces in the building and design them such that the loads that the bracing system induces in the foundation are distributed over the whole foundation system. In this way, it is possible to rehabilitate the building without costly modification of the existing foundation.
• Lightness. The weight of steel braces is usually small compared to that of the existing structure and that of other upgrading techniques that involve the addition or resizing of structural elements. Thus, there is a small increase in the structure's weight and reactive mass.
• Other advantages. There are other advantages that, although not important from a structural point of view, can have a considerable influence in the selection of this upgrading technique. Among them, the following can be mentioned: clean and fast construction process, the use of braces to achieve interesting-looking architectural patterns in the structure while allowing sunlight to reach the interior of the building, etc.

• Change of behavior of the original frame members. It is important to study the change in behavior and failure mode of the existing frame members when introducing the braces. In some cases, if the existing elements are not strengthened properly to avoid their premature failure due to this change of behavior, the structure can have a poor seismic performance. The introduction of steel braces into the existing structure usually reduces the lateral deformation of the structure when subjected to EQGM, and thus usually reduces the bending moments at the ends of the existing frame members. This reduction usually occurs simultaneously with an increase in their axial forces, as shown qualitatively in Figure 5 for the case of RC frames.

• Change in dynamic characteristics. There is the need to assess the change in the dynamic characteristics of the building once it is upgraded in order to detect possible changes in its lateral response.
• Connection of braces to existing structure. The connection of the steel braces to the existing structure should be done carefully in order to allow the bracing system to fully develop its lateral stiffness and strength. If the connection fails before the brace, this brace will not be able to develop its maximum strength and/or lateral stiffness.
• Buckling of the steel brace. To achieve a good seismic performance of the rehabilitated structure, it is necessary to avoid buckling of the braces. When a brace suffers global buckling during cyclic loading, it can lose a large percentage of its original strength. Overall buckling of a stiff member can lead to local buckling, and this local buckling under reversals of deformation can lead to premature failure. Also, the unexpected components of deformation produced by the buckling of the brace can induce undesirable stress components that could lead to a premature failure of its connection to the existing structure [Badoux and Jirsa, 1987].

As pointed out above, the key issue for achieving good performance with these rehabilitating steel concentric bracing systems is the prevention of the overall (global) buckling of the braces. Thus, the design of the needed restructuring (retrofitting) usually has to be based on the sole criterion of avoidance of significant nonlinear elastic behavior of the braces and/or bracing system. As will be discussed later, most of the upgrading systems that are based on the use of energy dissipation devices require the use of structural steel bracing systems, which have to remain elastic even under the maximum credible or capable EQ ground motions (MCEQGMs).

Although the use of PT braces has advantages and problems similar to the use of other types of steel braces, there are some aspects peculiar to PT brace behavior.
 
 

Linear elastic behavior of the PT cables. PT braces are usually designed to work in their linear elastic range of behavior. This is to prevent them from yielding in tension and thus from losing their initial prestress. Figure 7a shows the idealized basic axial deformation vs. axial force curve for a rod or cable (such as those used in PT bracing systems) with no prestress. As shown, the rod or cable buckles elastically for very low compressive forces, and can develop its yielding strength under tensile strains. Note that this type of element dissipates energy when it yields, although it does not when it buckles. Figure 7b shows the behavior of the rod or cable under cyclic loading producing yielding and buckling. As shown, all inelastic tensile elongation accumulates with reversals of actions, i.e., the length of the brace increases every time it yields in tension.

Figure 8a shows a counterpart of Figure 7a for a prestressed rod or cable. As shown, both figures are basically the same, with the exception that there is an initial state of stress and strain (produced by the prestress) in the prestressed rod or cable which is accounted for in Figure 8a by shifting the origin of the axial force vs. axial displacement Cartesian axes. As a consequence, the rod or cable can resist axial force under lateral forces that induce, due to a decrease in the initial tension in the rod or cable, shortening in the brace (this can be interpreted as the rod or cable developing a compressive force), as shown in Figure 8a. From this figure, it is clear that if the rod or cable loses its prestress, it loses its capacity to resist axial loads when subjected to compressive strains. Figure 8b shows that if the rod or cable yields, there is a loss of prestress. This is illustrated by following the load path OABC in Fig. 8b. As shown, the rod or cable remains elastic in the OA portion of this path. Once it reaches its yielding strength (point A) it yields and follows AB. As soon as there is a load reversal, the rod or cable unloads and reaches point C, which corresponds to zero axial deformation. Comparison of the location 0 and C shows that there has been a loss of prestress.
 
 

The above observations can be used to understand the consequences that yielding of the PT braces can have on their performance. First, excessive loss of prestress will reduce significantly the ability of the PT braces to resist lateral loads that will shorten them. Second, excessive elongation of a PT brace can result in a decrease of the lateral stiffness of that brace. These two effects are detrimental to the performance of the PT bracing system.

It is also convenient to assess the consequences of the PT braces' elastic behavior on the dynamic response of the structure. For example, if the braces carry the majority of the lateral loads, the structure will respond essentially elastically to the effects of an EQGM. Possible increases in the response of the entire building due to this effect should be carefully assessed.

• Yielding Strength. The PT braces can be fabricated from steels with different yielding strengths, and thus they can easily be designed for a wide range of elastic deformation capacities. Even if the PT braces are designed to remain elastic, a variety of yielding strengths can be used in the design process to enhance the compatibility of strength and deformation between the existing structure and the new bracing system, as shown in Fig. 9 [Rioboó 1989].

• Initial state of stresses in the PT braces. The amount of prestress provided to the PT braces should be designed to prevent their yielding or buckling. Thus, it is necessary to have a good estimate of the maximum axial forces and interstory drifts that can be induced in the PT braces and the upgraded building, respectively, when the building undergoes the design EQGM.

          Figure 10. Energy dissipation in existing buildings upgraded with PT braces.

• Elastic buckling. Due to their low axial stiffness, the PT braces do not buckle inelastically. If they are subjected to net compressive strains, the PT braces just buckle (bend) without developing compressive stresses, but as soon as the loads reverse (to tension) the brace can develop its full tension capacity. This behavior can be repeated through several cycles without degradation of the tensile axial strength of the brace. In some cases, it may be necessary to assess the consequences that the elastic buckling of the braces can have on the seismic performance of the building (i.e., changes in strength and deformation demands in the existing elements that can lead to demands for which they were not designed).

• Whipping of the PT braces. Due to their low axial stiffness,, the PT braces deform in or out of plane when subjected to compressive strains (i.e. when they undergo elastic buckling). The problem is when the buckling occurs out of plane, because even a small axial deformation in the braces can produce large out-of-plane deformations. Thus, it is necessary to provide out-of-plane support to the PT brace to avoid this deformation component, or better, to have a good estimate of the minimum axial force that could be developed in the brace when the structure is subjected to the design EQ, in such a way that, by proper post-tensioning, its buckling can be avoided.

• Initial state of stresses in the existing elements. Due to the initial level of prestress in the PT braces, an initial state of stresses is induced in the existing elements. Thus, the level of prestress to use cannot be determined without studying its effects on the behavior of the existing members. The existing members are subjected to an initial state of compression, which in some cases will enhance their seismic performance (mainly in low-rise buildings). Nevertheless, if the transverse steel of the existing members is poorly detailed, especially in columns, the initial compressive forces can be detrimental to their behavior. In some cases, the existing elements should be upgraded to resist these forces.

• Energy dissipation capacity of the braced building. The fact that the PT braces remain elastic does not mean that the members of the existing structure, rehabilitated by this technique, will exhibit elastic behavior. As shown in Fig. 10, it is possible to achieve controlled energy dissipation in the existing members while the braces remain elastic. It should be emphasized that the braces by themselves do not contribute to the energy dissipation capacity of the upgraded structure, because they are supposed to remain elastic. Nevertheless, the braces may indirectly enhance the energy dissipation capacity of the upgraded structure by enhancing the seismic performance of the existing elements [Miranda and Bertero 1990].

Inelastic behavior of the PT cables. It has been suggested by some researchers that in some cases it is appropriate to use high levels of prestress for the PT cables, such that the braces yield in tension at relatively small drifts. The bracing system is expected to dissipate energy through the braces' hysteretic behavior during the early stages of an extreme event. Pincheira and Jirsa [1992] note that this design criterion can be more effective than using lower levels of initial prestress, and they emphasize the importance of preventing the braces from becoming slack. Figure 11 shows the axial load vs. axial deformation behavior for a rod or cable with a high level of prestress. This figure shows that if the rod or cable yields, there is a loss of prestress. This is illustrated by following the load path OABC in Figs. 11a and 11b, and comparing the location of points 0 and C. Nevertheless, it can be seen that if the initial level of prestress is high and the inelastic deformation demand is small, the remaining prestress is enough to allow the rod to adequately resist axial forces under relative compressive strains, as shown in Figure 11a. As shown in this figure, some plastic hysteretic energy has been dissipated in the process. Figure 11b shows a case in which the inelastic axial deformation of the rod or cable is excessive.

Figure 11. Axial displacement vs. axial load behavior of rod or cable with high prestress

• Economy. Usually, the only materials needed to implement this technique are the braces themselves and their connection. Considering other costs, such as equipment and qualified labor, the total cost of implementing this technique in the field is usually lower than that of other upgrading techniques.

• Need to establish a rational performance criteria that takes into consideration the structural and mechanical characteristics of the URM infills. Before attempting to discuss the use of PT braces in the rehabilitation of framed buildings with URM infills, it is necessary to define the desired performance of the upgraded building when subjected to EQGMs corresponding to the different relevant limit states (service, damageability, safety, etc.). One way of defining the desired performance of the building consists in establishing performance criteria, i.e., defining limits for the value that the global and local response of the building can have in such a way that the response of structural and nonstructural elements can be controlled within a certain acceptable range of behavior. For instance, damage in frame members and URM infills (in-plane) can be controlled by limiting their deformation and energy dissipation demands, while out-of-plane damage control in URM infills and the integrity of the contents of the building can be achieved by limiting the story accelerations in the building.

As illustrated in Fig. 12, URM walls and infills show stable hysteretic behavior without considerable degradation of their resistance and hysteretic energy dissipation capabilities for relatively large drift. Thus, it seems that a reasonable way to enhance the seismic performance of URM infills, and thus of the entire building, consists in controlling their interstory distortions by controlling the global lateral displacement of the building. Note that if the inplane degradation of the mechanical characteristics of the URM infills is kept within reasonable values, the probability of occurrence of an out-of-plane failure due to in-plane effects diminishes considerably.
 
 

In some cases, damage control in URM infills cannot be achieved by only limiting their interstory distortions, given that in some cases the nonlinear cumulative demands are relevant to their behavior (damage). The fundamental period of translation (T) of low-rise buildings tends to be small, especially if they are infilled with URM walls and/or upgraded with a bracing system. In this range of T, the damage produced by nonlinear cumulative demands (such as the demand of hysteretic energy dissipation) is less relevant, in many cases, than that produced by interstory distortion [Terán-Gilmore, 1993]. In these cases, it is reasonable to attempt damage control by focusing on displacement control. For small T, one way to control the displacement of a structure is by decreasing its global ductility demands by increasing its lateral strength [Shimizaki, 1988; Qi and Moehle, 1991]. Nevertheless, once the lateral strength of the system reaches a certain value, a further increase in strength will not significantly affect the displacement response of that system. Thus, for the upgrading of a framed building with URM infills, it seems reasonable to increase adequately the strength and stiffness of the building through the introduction of the PT braces, in such a way that the interstory drifts, and thus damage in the URM infills, can be controlled to acceptable values (which must be defined as part of the performance criteria). Note that this is not the case for structures with larger T and built on soft soils, in which case the nonlinear cumulative demands can be significant and the elastic displacement can be similar to or even considerably larger than the inelastic displacement.

• Proposed performance criteria and philosophy of design for the PT braces. The design of an adequate PT bracing system for the seismic upgrading of a building can be based on different performance criteria. Once a criterion has been established and quantified, different philosophies of design can be used to satisfy it. In this section, one approach to the upgrading of existing infilled framed buildings with PT braces is discussed.

One of the drawbacks of keeping the frame members and PT braces elastic is the probable increase of the strength demand in the building when subjected to ground motion. One way of diminishing such demand is to provide energy dissipating devices to the structure. It should be noted that this is not necessary in the case of infilled frames, given that they have a large natural source of equivalent viscous and hysteretic energy dissipation ill the URM infills. Nevertheless, to use the URM infills as energy dissipators it is necessary to make sure they can provide this dissipation in a stable manner throughout the duration of its response to its critical ground motion. This is achievable by controlling their in-plane deformation by controlling the maximum IDI in the building, within certain limits.

• Non-ductile frame members should not develop battle failure.
• URM infills should not collapse.
• The PT bracing system should not lose stiffness or develop soft stories (prevent PT braces from becoming slack and/or from buckling in compression).

To achieve the above performance criteria, the following philosophy is suggested:

• Keep the PT braces and non-ductile frame members in their elastic range of behavior.
• Control the maximum IDI in the building in such a way as to achieve a stable hysteretic behavior in the URM infills.

• Out-of-plane failure. The upgraded building can have an adequate seismic performance only if out-of-plane failure of the infills is avoided, given that its occurrence can induce sudden and very large stiffness and strength irregularities, and thus unpredictable and large changes in the dynamic properties of the building. Thus, it is important to address again the concept of out-of-plane dynamic stability of the URM infills. It has been noted by researchers interested in the ABK Method [1984] that, if the movement of the whole building is dampened by the yielding and nonlinear behavior of some of its members, the out-of-plane forces are considerably reduced [Langenbach 1990]. In the case of upgrading an infilled frame, the use of PT braces with high stiffness and strength will likely reduce the nonlinear demands (deformation and hysteretic energy dissipation) in the building, which in turn will likely increase the in-plane and out-of-plane lateral forces and accelerations in the building (with respect to those in the unstrengthened building). Nevertheless, given that in the upgraded building the URM infills are supposed to dissipate energy through controlled nonlinear hysteretic behavior, the likely increase of lateral forces and story accelerations may be controlled to acceptable values.

• Efficient (optimal) relative stiffness and limiting deformation. As mentioned before, in order to achieve efficient EQ-RD of the upgraded structure it is important to select an efficient stiffness for the bracing system, and to supply this system with a lateral deformation capability similar to that of the existing structure. The mechanical characteristics of the PT bracing system should be provided in such a way that enough strength and stiffness is added to the upgraded building to achieve adequate control of the interstory drifts and nonlinear (inelastic) cumulative demands, and it should be flexible enough to allow the infills to resist a significant portion of the lateral loads, and thus to allow the infills to be used extensively to dissipate energy. In other words, the PT braces should add enough stiffness to control the maximum IDI in the building, but they should not be so stiff that they minimize the contribution of the URM infills to resisting the effects of the ground motion.

• Stiffness and strength irregularities in existing framed buildings with URM infills. URM infills have been commonly used as nonstructural elements. Given that in the past the contribution of these elements with such high stiffness and strength was usually neglected in the design process, no special consideration was given to their location within the existing frames of a building. Thus, large irregularities in plan and height of stiffness and strength usually exist in this type of building. The PT bracing system should attempt to correct the irregularities created by the infills, both in plan and height.

• Difficulty in assessing the real behavior of framed buildings with URM infills. Brokken and Bertero (1981) have noted that the lateral stiffness and strength of a masonry infill are very sensitive to the quality control of the material, as well as to the quality of their workmanship (including the interfaces of the infill and the existing frame elements). It should be mentioned that infill walls have a wide variety of configurations, depending on whether there are doors, windows, or other holes in the infill. The variability of the properties and geometry of the infills, as well as the large irregularities of strength and stiffness they may produce in the building, have to be considered carefully if a realistic prediction of the behavior of the infilled building needs to be obtained. Given the complexity of the models and our lack of knowledge about how to model adequately the characteristics and irregularities of infilled framed buildings, it is necessary to use simplified models with the corresponding introduction of uncertainty (which is considerably larger than that involved in evaluating the real behavior of regular framed buildings) in the results obtained in their analysis. This uncertainty needs to be assessed carefully, given that a reasonable estimate of the behavior of the upgraded building is essential in order to avoid the loss of prestress by yielding and/or the elastic buckling of the PT braces. Given our current limitations, it would be desirable at least to bound the response of the upgraded building by bounding some of the main structural and dynamic characteristics of the analytical model to be used in the final analysis, and to make the final design of the PT bracing system accordingly.

• Degradation of structural properties of URM infills during an EQGM, and its consequence for the use of elastic analysis to predict the response of a framed building with URM infills. It has been observed experimentally that the cyclic loading of URM infills leads to degradation of stiffness and strength, and that the effective equivalent viscous damping coefficient of the virgin system increases considerably as soon as some cracking develops [Brokken and Bertero 1981]. Therefore, the stiffness, strength and damping properties used to model the URM infills in the building need to be considered carefully according to the expected deformation and cumulative demands on those infills. This is especially true if elastic analyses, as required by current EQ-RD codes, are carried out to analyze the behavior of the building, given that even at small deformation levels the stiffness of the URM infills can decrease considerably from its uncracked stiffness [Brokken and Bertero 1981].

• Initial state of stress in the existing URM infills. The consequences that the initial state of stresses (due to prestressing of the PT braces) has on the seismic performance of the URM infills has to be evaluated. Two possible effects can be mentioned: an initial state of moderate in-plane compression in the infills will usually enhance their ultimate strength, deformation capability and overall stability; while high compressive stresses can be detrimental to their behavior. Because of the large plan area of the URM infills, the increase in compressive stress is not expected to be very large, and thus this initial state would probably enhance the behavior of the existing URM infills. If the initial state of compression enhances the behavior of the infills, it would be desirable where possible to locate the PF braces in the frames where the URM infills are located.

• Yielding of the PF braces. If the inelastic deformation demands in the braces are so large that the braces become slack, their stiffness diminishes, and this decrease in stiffness can be reflected in an increase in the displacement of the building that can induce excessive damage to the URM infills. Heavy damage (or failure) in the URM infills can lead to large irregularities of strength and stiffness throughout the plan and height of the upgraded building, which may produce unpredictable changes in its dynamic characteristics and, very probably, detrimental changes in its behavior. Even if no irregularities are created by the excessive degradation of their mechanical characteristics, if the contribution of the infills to the strength and stiffness of the building is lost unexpectedly as a consequence of the excessive yielding of the bracing system, an important percentage of the lateral load will begin to be carried by the frame. This can produce a large combination of axial forces and moments in the frame elements: axial forces induced by the braces and flexural moments due to the increased lateral deformation of the building. This type of loading, for which the frame elements were not designed, can lead to non-ductile failures in the existing columns. Avoiding excessive yielding of the braces is important for control of the displacement in the URM building to acceptable values so that excessive damage to the infills and other vertical elements is avoided. If the PT braces are allowed to yield, it is necessary to limit this yielding in such a way that it will not be detrimental to the response of the building. As the forces to be induced in the braces depend on the interaction between the dynamic characteristics of the entire building system and those of the EQGM, the importance of having a reasonable estimate of the characteristics and intensity of the EQGMs at the site is emphasized.

PRACTICAL APPLICATIONS. The technique of using PT braces to rehabilitate hazardous non-ductile RC frame buildings infilled with URM walls was applied to an old existing building identified herein as the Pomona Building. This building, illustrated in Fig. 13, is a six-story commercial building built in 1923, having as a structural system RC frames with unreinforced brick infills in all perimeter frames and three internal frames. At the ground level, the building measures 65 feet (E-W direction) by 120 feet (N-S direction) in plan, as shown in Fig. 14. Table 2 shows floor elevations and masses, along with the approximate locations in plan of the centers of mass and rigidity on each floor. The centers of mass corresponding to the second through fifth floors are close to the geometric centroid of the floor diaphragms, while large mass eccentricities exist in the mezzanine and sixth floors. In the mezzanine floor, the center of mass is displaced towards the northwest comer, and in the sixth floor (roof), towards the southwest corner. As shown in Table 2, the distance in the N-S direction between the centers of mass and stiffness in the mezzanine and second floors is very large, while in the third to sixth floors it is small. In the E-W direction, this distance is large for all floors.

The floor system consists of a three-inch thick one-way RC slab supported by RC joists spaced every two feet, while the structural system for gravity and lateral loads is formed by non-ductile RC frames (having beams, girders and columns) infilled with URM walls. Figure 14 shows schematic plan views of the different floors of the building, while Fig. 15 shows schematic elevation views of the four perimeter frames and the notation used for floors and stories. As shown in Fig. 15, there are URM infills in the perimeter frames, which probably contribute significantly to resist the lateral loads induced in the building by EQGMs. As shown, except for three small openings all infills in Frame I are full infills (without openings), while those located in Frame 6 have large openings, creating large stiffness and strength eccentricities in the E-W direction of the building, especially in the mezzanine and second floors (see Table 2). Infills located in the upper stories of Frames A and F show similar characteristics (openings); nevertheless, in the lower stories (ground and mezzanine) the stiffness and strength of Frame A are lower than those of Frame F (by comparing Figs. 15c and 15d, it can be seen that Frame A has a double-height first story and weaker and more flexible URM infills in this story than those in Frame F), creating large strength and stiffness eccentricities in these levels in the N-S direction (see Table 2). This eccentricity (N-S direction) is magnified by the presence of an L-shaped mezzanine (which, as shown schematically in Fig. 16, runs along Frames 1 and F), and a large mass eccentricity at the roof in this direction. It can be concluded that the building has large irregularities of mass, strength and stiffness in plan.
 
 

As shown in Fig. 15, Frames A and 6, which correspond to the two facades facing the streets, show a double-height first level that has considerably fewer infills than the upper levels. As will be discussed in more detail later, this creates a weak and flexible first level, which produces a large irregularity in height, both in strength and stiffness.

The basement of the building is enclosed by a perimeter 12-inch thick concrete wall, which provides a stiff and strong support for the columns of the ground story, except those of Frame 6, because this perimeter wall has an offset of 7 feet 6 inches with respect to the plane defined by Frame 6, as shown in Fig. 17.
 
 

Vulnerability Assessment of the Pomona Building. This was done through 2D and 3D analyses of linear elastic and nonlinear models of the building. Selection of some of the structural parameters (dynamic characteristics) in the development of these 2D and 3D models was guided and calibrated by the results of the analyses of the recorded response of the building during the 1990 Upland and 1992 Landers EQs. The building has been instrumented with sensors located in the basement, at the second floor, and on the roof, as is illustrated in Fig. 18. The dynamic characteristics of the existing buildings are shown in Table 3.
 
 

From analysis of the desired performance of the building in the future and the levels of EQGMs that can be expected at the site, it was decided that the seismic rehabilitation of this Pomona Building should be based on the minimum or basic design performance objectives of the performance-based EQ-RD matrix shown in Fig. 3. Further-more, an analysis of expected EQGMs at the site and their return period indicates that the preliminary design would be controlled by the life safety performance level. From an analysis of the expected EQGMs at the site, the spectra shown in Fig. 19 were selected for the vulnerability assessment of the life safety of the existing building as well as for its rehabilitation. According to the selected design performance objectives, the life safety design criteria summarized in Table 4 were formulated.
 
 

From the results obtained in the analyses conducted on the existing buildings, it was concluded that "the existing building is not satisfying the desired performance of life safety and that under the very rare but possible maximum credible EQ (MCEQ) it could collapse." The estimated maximum (ultimate) story shear strengths of the existing building are shown in Tables 5 and 6 and plotted in the spectra shown in Fig. 19. "The building exhibited insufficient lateral strength and stiffness as well as large irregularities of mass, lateral stiffness, and strength through plan and height. "
 
 

Design of the Post Tensioned Bracing System

Selection of layout of bracing system. An ideal solution is to place the PT braces in the perimeter of the building, However, aside from the structural properties of the bracing system, probably the most important consideration needed to determine its final layout is related to the architectonic and functional integrity of the building. These types of considerations significantly influenced the proposed layout of the PT bracing system. It should be mentioned that it was virtually impossible not to alter the existing distribution of space within the building.
 
 

Figure 20 shows the proposed location in plan of the PT braces, while Figures 21a to 21e show schematically the location in height.

The properties and geometry of what were considered the best two alternatives for the PT bracing system are summarized in Tables 7 to 9 and Figs. 20 and 21. As shown in Fig. 21, the first option (configuration 1) consists in adding braces to four planes of resistance: 23 and 47 in the N-S direction, and C and E in the E-W direction; while the second option (configuration 2) consists in adding to the previous four planes one more plane of resistance in the E-W direction (one more braced bay located in Frame B). An asymmetric distribution of braces in the planes of resistance parallel to the N-S direction was provided to correct the large torsional response of the building when loaded in this direction. Each diagonal in Fig. 21 represents two braces, which were provided in this way as to avoid inducing large load eccentricity to the existing elements. As shown, each brace spans two stories in order to achieve angles that allow for an efficient solution. Five different types of braces were considered, as shown in Tables 7 to 9. As shown, the sizes of the braces in the lower two stories are considerably larger than those in the upper four stories, in such a way as to avoid the existence of a flexible and weak first story. The diameter of the PT-braces ranged from 2 1/8" to 3 1/2", and it was proposed to use galvanized bridge wire with a yielding stress of 150 ksi. Fifty per cent of the available yielding stress is used for post-tensioning.
 
 

Table 7. Characteristics of proposed bracing system for configuration 1, N-S direction

Table 8. Characteristics of proposed bracing system for configuration 1, E-W direction
 

Table 9. Sizes of PT braces

In the proposed upgrading configurations there is a need to strengthen some existing columns in such a way that they can resist adequately the increase in axial loads induced in them by the braces, particularly those columns that support braces in both directions, such as those in the intersection of the following frames: 23 and C, 23 and E, 47 and B, and 47 and E. Figures 21c and 21d illustrate the need to add new beams to frames C and E in such a way that the braces have enough support at the floor levels (i.e., to avoid the buckling of the slab).

Table 10 summarizes the dynamic characteristics of the first translational modes estimated from the stick models corresponding to the E-W and N-S directions.
 
 

Figures 19a and c illustrate the lateral strength of the building upgraded with brace configuration I (obtained by adding the strength of the original building plus that provided by the PT braces) relative to the design strength spectra for the safety limit state. As shown, in the N-S direction, the building is expected to dissipate some energy (p slightly larger than 1). As mentioned before, this energy dissipation should be provided by the URM infills.

The values of the elastic coefficients of distortion (which is defined as the ratio between the maximum-n and the average IDI demands in the building) in plan and height in both directions were reduced considerable by the introduction of the PIT braces: from about three for both directions in the original building to 1.5 and 1.6 in the E-W and N-S directions, respectively.

From Fig. 19a it can be concluded that "the designed bracing system enhances considerably the performance in both directions, and this performance can be considered within the target performance for safety that has been specified in the selected design criteria" (Table 4).

The performance of the rehabilitated building was assessed through a series of 2D and 3D linear elastic and nonlinear analyses. The curves shown in Fig. 22 permit comparison of the performance of the rehabilitated building with that of the original building. It is clear that the lateral stiffness and strength of the Pomona Building are considerably enhanced by upgrading it with PT braces. Figure 23 shows four d_{roof vs. Vb curves, one for each corner of the building upgraded with brace configuration 1, obtained from a 3D pushover analysis in the E-W direction. It is clear from these curves that there is a significant torsional response when this version of the upgraded building is loaded in the E-W direction. As shown, the comers located in the south end of the building (Frame A) have a} d_{roof demand that is roughly twice that corresponding to the corners located in the north end (Frame F). It can be observed that the difference between the} d_{roof demands of the south and north comers tends to increase with increasing} d_roof.

_{To improve the torsional performance of the rehabilitated building, a new layout of PT braces was selected and designed. This new layout was a modification of configuration 2 (shown in Fig. 20), in which the braced bays in Frame E of brace configuration 1 has been translated to frame B. Thus, the modified configuration 1 has the same number of braced bays as configuration 1. Figure 24 shows the results from a 3D pushover analysis in the E-W direction for the modified configuration 2. Comparing the results (curves) presented in this figure with those plotted in Fig. 23 for configuration 1, it can be concluded that the torsional responses of the upgraded building with the modified configuration 2 is reduced considerably. As shown in Fig. 24, the} d_{roof demands in the corners located in the south end of the building (Frame A) are now roughly 1.3 times the} d_{roof demands corresponding to the corners located in the north end (Frame F) and this ratio remains fairly constant with increasing} d_roof.
 
 

_{Concluding Remarks.}

• _{A significant increase in the lateral strength and lateral stiffness of framed buildings with URM infills can be achieved by introducing PT braces into their frames. In the specific case of the Pomona building, the ultimate base shear in both directions of the upgraded building was about three times those corresponding to the existing building, as reflected by increases from 0.15 W to 0.50 W in the E-W direction, and from 0.40 W to 1.20 W in the N-S direction. Large increases in the lateral stiffness were also achieved by introducing PT braces to this building, as reflected by the fact that the lateral stiffness in the upgraded building is about three times that of the existing building in the E-W direction (decrease in the fundamental period of translation from 1.0 sec to around 0.6 sec) in the upgraded building, and about one and a half times that of the existing building in the N-S direction (decrease in the fundamental period of translation from 0.5 sec to 0.4 sec).}

• _{The introduction of PT braces into the existing building allows considerable reduction of its existing strength and stiffness irregularities in plan an height. An indirect way to measure the above irregularities in plan and height consists in using a coefficient of distortion, which is defined as the ratio between the maximum and average interstory drift index demands in the building. An idea of the improvements attained by introducing PT braces into the Pomona Building can be provided by the fact that the coefficient of distortion in both directions was reduced from values of about 3 in the existing building to about 1.5 in the upgraded building.}

• _{The use of PT braces to upgrade non-ductile frame buildings with URM infills may be limitedto low-rise and squat medium-rise buildings. This is because the PT braces are usually designed to resist very high lateral forces, which in turn is a consequence of designing them to remain elastic while keeping the frame members from having significant nonlinear demands (i.e., basically elastic). The large forces introduced by prestressing the braces which can be increased significantly by the deformations that can be generated in the PT braces when the building is subjected to lateral loads can induce significantly large axial forces, particularly in the existing columns that are used at the comers in the selected brace configuration system and the foundation, which can create structural problems that may be very expensive to fix. Among such problems, the need to upgrade the mechanical characteristics of the existing columns and foundation can be mentioned.}

• _{A quantitative measure of the qualitative definition of damage in the elements of the Pomona building was established by setting limits to the maximum IDI in the building. The controlling IDI at safety level for the Pomona building was imposed by the need to achieve stable hysteretic behavior in the URM infills. This limiting value of IDI was equal to 0.005.}

• _{As it is desirable to analyze not just one, but several alternatives to upgrade a building, it is advantageous to establish simple preliminary analysis procedures that allow for the identification of the most promising alternatives. A simplified analysis procedure based on the use of stick and single-degree-of-freedom models to assess the performance of the existing and upgraded building was used successfully for the preliminary design of several alternatives of the bracing system, as well as for the preliminary assessment of the performance of the building upgraded using these alternatives. These simple methods are invaluable tools for the simple and rational EQ-RD of upgrading schemes.}

• _{Given that the PT braces, which provide the majority of lateral stiffness and strength in the upgraded building, and the existing frame members should remain elastic, it was found that using the results obtained from detailed elastic analyses is a reasonable and conservative way of assessing the performance of the upgraded building.}

• _{The design of the PT bracing scheme should not only consider the design of the braces themselves, but the possible failure of the existing elements or the formation of local mechanisms before the braces can reach their ultimate capacities. The upgrading of the existing elements should be carried accordingly.}

• _{In many cases, structural considerations do not determine the final configuration of the PT bracing system. Other constraints (architectonic, space, constructability, etc.) can have as much or more influence in determining such configuration.}

• _{Several issues in the in-plane behavior of URM elements are yet to be understood. Among them, it is necessary to gain a better understanding of the cyclic behavior of URM walls and infills. Special consideration should be given to the way in which degradation of stiffness and strength occurs in the infill as a function of its maximum deformation and cumulative hysteretic energy dissipation demands. These demands can be defined through a damage function or index. Thus, to provide a quantitative measure to the qualitative description of different levels of damage, these damage functions or indices should be determined experimentally to allow the use of multi-limit state performance-based methods in the EQ-RD of upgrading schemes of buildings with URM infills. The possible effect that the maximum deformation demand in one direction has in the strength and stiffness that an URM infill can develop in the opposite direction needs to be clarified.}

• _{The out-of-plane behavior of the URM infills needs to be studied and its effect on the in-phase response should be assessed.}

• _{Another issue that deserves consideration is the in-plane behavior of URM infills with openings.}

• _{Further research needs to be devoted to obtaining simple but reliable mathematical models of URM infills. These models should incorporate a good understanding of the cyclic behavior of URM infills with and without openings. Research should be carried out to quantify the possible reduction of the global response of the building due to the hysteretic energy dissipation provided by the URM infills. The possibility of representing these effects by using an equivalent damping coefficient (}x_{EQ), needs to be assessed.}

• _{The effect that the introduction of PT braces to the original building has on the magnitude of the floor accelerations in the upgraded building still needs to be studied. Also, the possible reduction in their magnitude due to the damping effects created by the hysteretic energy dissipation provided by the URM infills also need to be investigated.}

_{In view of the above apparent weakness in the use of just post-tensioned steel bracing systems, efforts have lately been devoted to searching for ways of overcoming this weakness. Analysis of the available ways of dissipating energy through the use of energy dissipation devices have permitted identification of a possible technical solution for such weakness, and this is the use of the Shape Memory Alloys (SMAs). This is so because these alloys have the ability to "yield" repeatedly and not lose their prestress, as is illustrated in Fig. 25. From analysis of the experimental results obtained at EERC [Sazaki, 1989; ; Aiken et al., 1993; and Clark et al., 1995], it is clear that the use of post-tensioned Nitinol (also written as NiTi, a nickel-titanium alloy) rods can be technically a very attractive solution, because as a consequence of its energy-dissipating capacity it can lead to a great reduction in the main response parameters (acceleration, relative displacement, and interstory drift) of the frame to which it is attached when it is subjected to earthquake ground motions. However, this ideal technical solution, because of its present high cost, will not be an economical solution. A technically and economically feasible solution appears to be the use of post-tensioned high-strength steel rods, together with what can be called a self-balanced prestressed energy dissipation device based on the use of SMAS.}
 
 

_{INNOVATIVE TECHNIQUES FOR THE SEISMIC UPGRADING OF EXISTING FACILITIES}

_{INTRODUCTORY REMARKS. As discussed previously, the significant life and economic losses induced by recent earthquakes clearly indicate a need to control damage to our civil engineering facilities not only under rare major earthquake ground motions but also under the occasional moderate and even the frequent minor earthquake ground motions. For buildings, damage control is needed to protect not only entire structural systems, but also their nonstructural components, equipment, contents and service lines, under the different levels of earthquake ground motions, This protection can be achieved using different types of "Protective Systems." Herein, only the following two types of protective systems will be briefly discussed and illustrated: Base Isolation and Passive Energy Dissipation Systems.}

_{BASE ISOLATION. The notes prepared by Kelly [1994a and 1994b], Aiken [1994b] and Kircher [1994] discuss in detail the concepts of base isolation, linear theory and design, the different types of isolation systems, applications, future directions in base isolation, and the seismic isolation code provisions. In what follows, after summarizing the concept of base isolation and its advantages compared with other protective systems, the different base isolation systems, applications for upgrading buildings and the observed responses are briefly described.}

_{The Concept of Base Isolation. The traditional idea was to prevent ground motions from being transmitted from the building foundation into the superstructure. At present, the basic idea is to control the earthquake energy input into the building and so to reduce its response to acceptable levels of deformation and rate of deformation.}

_{When a building is built on an isolation system, it should have a fundamental frequency that is lower than both its fixed-base frequency and the dominant frequencies of the ground motion. The first mode of the isolated structure then involves deformation only in the isolation system, the structure above being almost rigid. The higher modes that produce deformation in the structure are orthogonal to the first mode, so that if there is high energy in the ground motion at the higher mode frequencies, this energy cannot be imparted to the structure. In this way, the demand on the structural system is reduced, as are the accelerations transmitted to the internal nonstructural components and equipment.}

_{This makes isolation a very attractive approach where protection of expensive sensitive equipment is needed, and it is no surprise that this technology has been used for buildings such as hospitals, computer centers, emergency operations centers and nuclear facilities. The beneficial effects of the isolation system are independent of damping, although some damping is desirable to suppress resonance due to long-period motion, but high levels of damping are not needed in isolation systems for buildings where protection of internal equipment is the goal of the design. High levels of energy dissipation lead to smaller displacements in the isolation system but higher accelerations in the superstructure, and are thus advantageous for bridges, where control of displacements is important and accelerations in the deck are not important.}

_{Advantages of Seismic Isolation}

_{Base Isolation vs. Traditional Fixed-Base Approach to Earthquake-Resistant Design and Construction. In the conventional fixed-base design approach, improvement of the seismic response of a building was attempted by increasing strength (which usually increased the stiffness). Lately, emphasis has shifted away from increase in strength to increase in stiffness, following the recognition that damage is more a consequence of deformations and deformation rates than of strength. However, in buildings located on rock or firm soils the resulting increase in stiffness (directly or through the increase in strength) results in buildings with shorter periods of vibration, which leads to the development of larger accelerations in the structures and in its nonstructural components and contents. In few words, a fixed-base building tends to amplify the ground motion. To minimize this amplification, the structural system must either be extremely rigid or have a large amount of damping. At best, rigidity results in the contents of the building experiencing the peak ground accelerations, which may still be too high for sensitive internal equipment and contents. Incorporating high levels of damping in a structural system means either damage to the system or designing an expensive structural form to mitigate this damage.}

_{Base Isolation vs. Passive Energy Dissipation Approach. As will be discussed later, significant advances have been made in recent years, improving the seismic performance of structures through the use of energy dissipation devices. However, the use of these devices is still in its infancy in the U.S. Usually, the proper use of these devices demands an increase in the stiffness of the structures, which may result in a significant increase in the accelerations developed in the structure and in the contents of the building, which, as discussed above, may lead to problems in the case of sensitive equipment and contents.}

_{Base Isolation Systems. Several types of isolation systems are currently in use, and several others are being developed or investigated. Despite wide variations in the detailing of the base isolators, their system behavior, and the design, the systems can be divided into the following two distinctive basic groups (Chopra 1995).}

• _{First basic group. In the first basic group, the isolation system introduces a layer of low lateral stiffness between the structure and the foundation. With this isolation layer, the structure has a natural period that is much longer than its fixed-base natural period. This lengthening of the period can usually reduce the pseudo-acceleration and hence the earthquake-induced force in the structure, but the deformation is increased; this is the deformation across the isolation system. As noted above, this type of isolation system is effective even if the system is linear and undamped. Damping is beneficial in further reducing the deformation in the isolation system.}

• _{Second basic group. This second type of isolation system uses rollers or sliders between the foundation and the base of the structure. The shear force transmitted to the structure across the isolation interface is limited by keeping the coefficient of friction as low as practical. However, the friction must be sufficiently high to sustain strong winds and small earthquakes without sliding, a requirement that reduces the isolation effect. In this type of isolation system, the sliding displacements are controlled by high-tension springs or laminated rubber bearings, or by using a concave dish for the rollers. These mechanisms provide a restoring force, otherwise unavailable in this type of system, to return the structure to its equilibrium position. The dynamics of structures on rollers or on the slider type of isolation system is complicated because the slip process is intrinsically nonlinear. Table 11 lists the isolation systems used and/or proposed in the U.S.}

_{Theory and Design. Kelly [1994b] and Chopra [1995] discuss elementary linear dynamic analysis that permits insight into the dynamic behavior of isolated buildings.}

_{Civil Applications of Base Isolation. Over the past decade, the use of seismic isolation around the world has proliferated. Kelly [1994a] and Aiken [1994b] discuss this application in different countries.}

Table 12 summarizes and compares the status in 1986 and 1993 of base-isolated buildings worldwide. Only applications of base isolation in the United States and Japan will be summarized herein.
 
 

Base Isolation in the United States. At present, there are many base-isolated building projects in the U.S. Some of these projects are for new structures but a large number are seismic retrofit projects (Table 13). These retrofit projects were stimulated by damage to several historic buildings in San Francisco, Oakland and surrounding areas from the 1989 Loma Prieta EQ. As pointed out by Kelly [1994a], it is interesting to note that to date the use of isolation for new buildings is mainly in Southern California, whereas most isolation projects in Northern California are for retrofit. The photos of Fig. 26 illustrate some of the buildings that have been or are being upgraded in the San Francisco Bay Area using base isolation.
 
 

Base Isolation in Japan. At present the most widespread use of this innovative protective approach is in Japan. The first building project was approved in 1985 and since then the implementation of base isolation in Japan has been rapid. As of January 1993, the total number of base-isolated buildings was 67, and as of January 1995 the number exceeded 75. Most of these isolated buildings are in the Tokyo area and comprise a wide range of building types. Most have reinforced concrete superstructures and are up to six stories. Only a few base-isolated buildings exceed six stories, and these have a composite steel-reinforced concrete structural system. As of January 1993, there were only three base-isolated wooden-framed structures. Roughly half of the buildings have natural rubber (low damping) isolators with additional damping components, such as steel bars or frictional elements. Most of the other half of isolated buildings use lead-rubber bearings and high-damping rubber bearings. Applications of sliding systems are rare. All rubber isolators in Japan up to 1994 were circular and most about 40 to 60 cm in diameter. The smallest isolators in use are 20 cm diameter bearings under a two-story wooden houses and the largest are 1.5 meter diameter bearings. The isolation periods at large displacements cluster around 2.5 seconds, although there are some few isolated buildings with periods greater than 3.0 seconds

Low-Cost Isolation Systems. As pointed out before, the emphasis in most base isolation applications up to this time has been on large structures with sensitive or expensive contents, but there is increasing interest in the possibility of applying the technology to public housing and other buildings such as schools and local health centers in developing countries. Several projects are under way for such applications. A challenge in this type of application is to develop a low-cost isolation system that can be used in conjunction with vernacular methods of construction, such as masonry block and lightly reinforced concrete frames. Demonstration buildings for housing projects have been complete using high-damping rubber bearings as the isolation system in Santiago, Chile [Sarrazin and Moroni, 1992] and in Squillace in Southern Italy [Vestroni et al. 1992]. A demonstration project in Indonesia is currently under construction. Both of these projects have partial support from the United Nations Industrial Development Organization.

Response of Base-Isolated Buildings During Recent Earthquakes

Response to the Ground Motions Generated by the 1994 Northridge Earthquake. When the Northridge earthquake occurred, there were in the greater Los Angeles area five occupied base-isolated buildings and several others in various states of design and construction. Two are supported on elastomeric isolators -- the University of Southern California (USC) Teaching Hospital and the Los Angeles County Fire Command and Control Facility (FCCF) -- while thesprings and viscous dashpots. These three buildings were isolated during their construction, i.e., they were not isolated for seismic upgrading.

The USC Hospital, an eight-story steel-braced frame structure built in 1988, is the base-isolated building closest to the epicenter. The steel-braced frames are supported on 68 lead-rubber isolators and 81 elastomeric isolators. The peak free-filed ground acceleration was 0.49g, and the peak acceleration at the foundation was 0.37g. The peak accelerations in the structure above the isolators were 0.13g and 0.2lg at the base and the roof, respectively, implying amplification ratios of 0.32 and 0.57 relative to the input motion at the foundation level. These ratios are in the expected range for a seismically isolated structure under the level of recorded ground acceleration. No structural damage was observed, and the hospital remained completely functional during and after the EQ. There were no reports of damage to equipment. Pharmacy shelves, files and bookshelves retained their contents. An adjacent new USC building, of the same height but on a fixed base, suffered significant nonstructural damage and virtually all the shelf contents in the pharmacy fell.

The FCCF building has as a structural system a two-story braced-steel frame supported on 32 high-damping rubber isolator. While the recorded response in the north-south direction was as expected for a PGA of 0.18g at the foundation, with amplification ratios of about 0.4 and 0.5 at the first floor and roof, respectively, the recorded response in the east-west direction was unexpected, with spike representing amplification ratios substantially greater than 1.0. A field inspection of the facility revealed that an architectural detail at the east-facing entryway near the north wall of the building may have restrained the relative movement of the ground floor with respect to the foundation. The original tiles were damaged during the 1992 Landers EQ, and the replacement tiles apparently were stronger and provided more lateral restraint at the ground floor demonstrating the importance of proper maintenance and repair of buildings.

The third base-isolated facility consists of two identical three-story braced-steel frame residences in Santa Monica, each supported at its comers by GERB helical springs and viscous dashpots. Additional springs are distributed around the building perimeter. From inspection of these buildings, it appears that this isolation system is more effective vertically than horizontally, because several details limit horizontal movement. Damage was observed as a consequence of this restrained movement.

Response to the Ground Motions Generated by the 1995 Great Hanshin (Kobe) Earthquake. Although there has not been any retrofitted building in which base-isolation was used when this EQ occurred, there are two recently completed buildings in the northern part of Kobe near the city of Sanda that were moderately shaken by the Kobe EQ. One of these is a very large six-story building having a composite steel-reinforced concrete superstructure with a total floor area of 505,000 ft² (47,000 m²) . The other, on a site approximately one-third of a mile (500 m) away, is a three-story reinforced concrete laboratory with a total floor area of 5200 square feet (486 square meters). These structures are supported on relatively stiff soil approximately 22 miles (35 km) northeast of the epicenter of the earthquake, and because of their proximity to one another, it can be assumed that both experienced similar levels of ground motion.

The smaller of the two structures is part of the Technical Research Institute of the Matsumara-Gumi Construction Company. it was completed in March 1994 and is used as a laboratory and conference area. The superstructure is a reinforced concrete space frame with a total height of 42 ft (12.8 in) and is supported on eight high-damping rubber bearings. The fixed-base period of the superstructure is 0.24 second and the target period of the isolation system vanes with displacement from 1.2 seconds at 0.5 inch (1.35 cm) to 2.3 seconds at 8.0 inches (20.3 cm). Adjacent to the isolated structure is a four-story steel moment frame with a rigid foundation, and the two buildings are connected via sliding joints.

The peak ground accelerations observed at the Matsumura-Gumi site were approximately 0.27g in both the longitudinal and transverse directions of the building, with a duration of strong motion of approximately seven seconds. Although digital data is not yet available, Table 14 shows the peak response accelerations, indicating that while there was some attenuation of the input, it was not as great as anticipated for this level of excitation. There was no damage in the isolated structure, but at the roof of the adjacent fixed-base structure there were reports of dropped ceiling tiles and a crack in a ventilation duct. The accelerations at the roof of this steel frame peaked at 0.98g. Based on scratches in the stainless steel sliding joint at the first floor of the isolated building, the relative bearing displacement was almost five inches (12.5 cm), which would imply a vibration period of approximately two seconds, However, the measured period was closer to 1.5 seconds, perhaps because the EQ struck early in the morning when the temperature in the basement of the building was about O° C (32° F), causing a slight stiffening of the rubber isolators.
 

The larger building is the West Japan Postal Services Computer Center, which is owned by the Ministry of Post and Telecommunications and serves as the computer center for all the financial transactions of this Ministry in western Japan. It was occupied in late 1994, and since that time has been closed to the public for security reasons. The lateral force-resisting system in the superstructure consists of braced frames while the isolation system is a hybrid of several different types of devices including 54 lead-rubber bearings, 66 natural rubber bearings, and 44 steel coil dampers. The fixed-base period of the superstructure is 0.68 second, and the target period of the isolation system varies with displacement from 2.8 seconds at 4.7 inches (12 cm) to 3.3 seconds at 9,4 inches (24 cm). Although detailed records of the response of this building to the Kobe EQ are not yet available, peak response quantities have been reported (Table 15). These data indicate that the isolation system was very effective. The input accelerations can be assumed to be equivalent to those at the Matsumura-Gumi site, but the bearings in the Postal Center are not exposed to the environment. The peak displacement of the isolators in this building was reported to be approximately four inches (10 cm).
 
 

USE OF PASSIVE ENERGY DISSIPATION SYSTEMS FOR EARTHQUAKE PROTECTION. Although since the late 1960s there has been a growing interest in the use of mechanical energy dissipation devices to reduce the dynamic response of buildings to extreme load conditions, such as those caused by wind storms, and EQs, it was in the early 1980s when the interest in the development and application of these devices for EQ-RD and EQ-RC became a reality through significant analytical and experimental investigations. This perhaps occurred due to a contemporaneous tremendous growth in the interest of developing and applying energy approach to EQ-RD. From analysis of the basic balanced energy equation (Eq. 3), it is clear that there are two ways of dissipating part of the input energy: one is through the hysteric behavior of the structural material as a consequence of its inelastic (yielding in the case of metallic material) deformation, E_{Hm .}; and the other is through the hysteretic viscous or equivalent viscous damping of the structure, E_Hx . Therefore, from the analysis of this energy equation and the way that it can be obtained, i.e., by integrating the equation of motion of an elastic/inelastic system, it is clear that there is an essential physical difference between the dissipation of energy through yielding of the metals, i.e., E_{Hm .}, and that through equivalent viscous damping, E_Hx . While dissipation of energy due to equivalent viscous damping occurs even in the elastic behavior of the structural material, the dissipation due to plastic deformation (yielding) is not present during the elastic response of the structural material. Thus, from this physical point of view it is unfortunate that in the literature the devices that have been developed and used to achieve these two basic different ways of dissipating energy has been denominated under the general term "passive damping devices".

Numerous pseudo-dynamic and earthquake simulator studies accompanied and augmented by analytical investigations have been conducted in the last decade. The last few years have seen the beginning of implementation of these devices in buildings to control their seismic response, as well as effort devoted to developing code provisions for passive energy dissipating systems by the SEAONC and by the Building Seismic Safety Council (BSSC). In spite of these efforts, as will be discussed in more detail later, although at present there are many proposed projects, to date there are only very few seismic applications of passive energy dissipation devices in buildings in California, and these few have been for seismic upgrading purposes and as a consequence of the observed response during the 1989 Loma Prieta earthquake. However, there have been several other applications to buildings in other countries as will be pointed out later. In this paper, attempts will be made to give a general overview of the main types of passive damping devices, which offer some potential for their use in buildings (new and existing) to control their seismic responses, and that have been the focus of research and development activities in the last decade.

Analysis and Design Issues. The main issues have been discussed by Whittaker [1994].

Types, Behavior and Applications of Energy Dissipation Devices. According to Aiken [1994a], the main types or systems can be classified under the following groups: Friction, Metallic, Lead Extrusion, Shape Memory Alloys, and Viscous and Viscoelastic.

Friction Systems. There are a variety of friction devices which have been proposed for structural energy dissipation. All of the friction systems, except one (the Fluor-Daniel EDR), generate rectangular hysteresis loops characteristic of Coulomb friction (Fig. 27a). Typically, these devices have very good performance characteristics, and their behavior is not significantly affected by load amplitude, frequency, or the number of applied load cycles. The devices differ in the mechanical complexity and in the materials used for the sliding surfaces.

Friction devices made by Sumitomo Metal Industries, Ltd. (Fig. 27b) have been used in two buildings in Japan [Aiken and Kelly, 1990], and a friction device manufactured by Pall Dynamics, Ltd., has been used in three buildings in Canada, one retrofit and two new buildings [Pall et al., 1987, 1991; Vezina et al., 1992]. The Pall device (Fig. 28) is intended to be mounted in X-bracing. Several earthquake simulator studies of multi-story steel frames incorporating Pall devices have been performed [Filiatrault and Cherry, 1987; Aiken et al, 1988], and a design methodology has been developed for friction-damped structures
 
 

Figure 29. Fluor-Daniel EDR hysteresis loops [Richter, 1990]

Issues o importance with friction devices are long-term reliability and maintenance; the potential for introduction of higher frequencies as the devices undergo stick-slip behavior; and possible permanent offsets after an earthquake. The maintenance and protection from deterioration of a device in which the sliding surfaces are required to slip at a specific load during an earthquake, even after decades of nonuse, is essential.

• Metallic Systems. These energy dissipation systems take advantage of the hysteretic behavior of metals when deformed into their inelastic post-elastic range. A wide variety of types of devices have been developed that utilize flexural, shear, or extensional deformation modes into the plastic range. A particularly desirable feature of these systems is their stable behavior, long-term reliability, and generally good resistance to environmental and temperature factors.

Yielding Steel Systems. The ability of mild steel to sustain many cycles of stable yielding behavior has led to the development of a wide variety of devices which utilize this behavior to dissipate seismic energy [Kelly et al., 1972; Skinner et al., 1980]. Many of these devices use mild steel plates with triangular or hourglass shapes [Tyler, 1978; Stiemer et al., 1981] so that the yielding is spread almost uniformly throughout the material. The result is a device which is able to sustain repeated inelastic deformations in a stable manner, avoiding concentrations of yielding and premature failure.

One device that uses X-shaped steel plates is the Bechtel Added Damping and Stiffness (ADAS) device. ADAS elements are an evolution of an earlier use of X-plates, as damping supports for piping systems [Stiemer et al., 1981]. Extensive experimental studies have investigated the behavior of individual ADAS elements and structural systems incorporating ADAS elements [Bergman and Goel, 1987; Whittaker et al., 1991]. The tests showed stable hysteretic performance (Fig. 30).

The principal characteristics which affect the behavior of an ADAS device are its elastic Stiffness, yield strength and yield displacement. ADAS devices are usually mounted as part of a bracing system, which must be substantially stiffer than the surrounding structure. The introduction of such a heavy bracing system into a structure may be prohibitive.
 

• Application of ADAS Devices to Seismic Upgradine of Buildings. ADAS devices have already been installed in a two-story, non-ductile reinforced concrete building in San Francisco and in three buildings in Mexico City.

Upgrading of a Two-Story RC Building in San Francisco. This downtown building suffered after the building suffered structural and nonstructural damage in the 1989 Loma Prieta earthquake. A detailed analysis of the building subsequent to the earthquake by Dames & Moore [Fiero and Perry, 1993] indicated that the it was vulnerable to severe damage in a major EQ in the San Francisco Bay Area. This 1300 m² building (Fig. 3 1) is composed of two stories and a part mezzanine, and is constructed above a one-level underground parking garage that occupies an entire city block. The plan dimensions of the building are 24.7 m by 24.7 m. The building, constructed in 1967 using lightweight reinforced concrete, has a number of seismic deficiencies that include shear-critical columns and link beams, and nonductile detailing in all elements in the lateral force resisting system.

For the retrofit of this building, the design team and their client, Wells Fargo Bank, chose a higher performance criterion than that called for in the Uniform Building Code -- one that included some measure of damage control. The client wished to be able to resume banking operations shortly after a major earthquake. The design team established several design objectives in order to attain this performance goal. These objectives included: limiting the lateral forces in the building during the design-level earthquake so as not to overload the existing foundations; and, limiting the maximum roof deflections during the design-level earthquake to approximately those deflections experienced during the Loma Prieta earthquake.

The design-level strong ground motion at the site was characterized using two response spectra: one site-specific with a peak ground acceleration of 0.39g, the second a generic spectrum for a soft soil site with a peak ground acceleration of 0.459. Acceleration time-history records compatible with both spectra were developed.

• Application of T-ADAS Energy Dissipation Devices. Triangular-plate energy dissipators were originally developed and used as the damping elements in several base isolation applications [Boardman et al., 19831. The triangular-plate concept has been extended to building dampers in the form of the triangular ADAS, or T-ADAS, element [Tsai and Hong, 1992]. Component tests of T-ADAS elements and pseudo-dynamic tests of a two-story steel frame have shown very good results (Fig. 30b). The T-ADAS device embodies a number of desirable features; no rotational restraint is required at the top of the brace connection assemblage, and there is no potential for instability of the triangular plate due to excessive axial load in the device.

• Other Yielding Steel Systems. An energy dissipator for cross-braced structures, which uses mild steel round bars or flat plates as the energy absorbing element, was developed by Tyler [1985]. This concept was applied to several industrial warehouses in New Zealand. Variations on the steel cross-bracing dissipater concept have been developed in Italy [Ciampi, 1991]. A 29-story steel suspension building (with floors hung from the central tower) in Naples, Italy, uses tapered steel devices as dissipators between the core and the suspended floors.

A six-story government building in Wanganui, New Zealand, uses steel-tube energy-absorbing devices in precast concrete cross-braced panels [Matthewson and Davey, 1979]. The devices were designed to yield axially at a given force level.

Several types of mild steel energy dissipators have been developed in Japan [Kajima Corporation, 1991; Kobori et al., 1988]. So-called honeycomb dampers have been incorporated in 15-story and 29-story buildings in Tokyo. Honeycomb dampers are X-plates (either single plates or multiple plates connected side by side) that are loaded in the plane of the X. (This is orthogonal to the loading direction of triangular or ADAS X-plates). Kajima Corporation has also developed two types of omni-directional loading steel dampers, called "Bell" dampers and "Tsudumi" dampers [Kobori et al., 1988]. The Bell damper is a single-tapered steel tube, and the Tsudumi dampers is a double-tapered tube intended to deform in the same manner as an ADAS X-plate but in multiple directions. Bell dampers have been used in a massive 1600-ft ski-slope structure to permit differential movement between four dissimilar parts of the structure under seismic loading while dissipating energy. Both of these applications are located in the Tokyo area.

Another type of joint damper for application between two buildings has been developed [Sakurai et al., 1992]. The device is a short lead tube that is loaded to deform in shear (Fig.32). Experimental investigations and an analytical study have been undertaken.
 
 

• Lead Extrusion Devices (LEDs). The extrusion of lead was identified as an effective mechanism for energy dissipation in the 1970 s [Robinson and Greenbank, 1976]. LED hysteretic behavior is very similar to that of many friction devices: essentially rectangular (Fig. 33). LEDs have been applied to a number of structures, for damping in seismic isolation systems, and as energy dissipators within multi-story buildings. In Wellington, New Zealand, a ten-story cross-braced concrete police station is base isolated, with sleeved-pile flexible elements and LED damping elements [Charleson et al., 1987]. Several seismically-isolated bridges in New Zealand also utilize LEDs [Skinner et al., 1980]. In Japan, LEDs have been incorporated into 17-story and 8-story steel frame buildings [Oiles Corp., 1991]. The devices are connected between precast concrete wall panels and the surrounding structural frame.

LEDs have a number of particularly desirable features: their load-deformation relationship is stable and repeatable, being largely unaffected by the number of loading cycles; they are insensitive to environmental factors; and tests have demonstrated insignificant aging effects [Robinson and Cousins, 1987].

Shape Memory Alloys (SMAs). As mentioned previously, SMAs have the ability to "yield" repeatedly without sustaining any permanent deformation. This is because the material as it deforms undergoes a reversible phase transformation, rather than the intergranular dislocation typical of steel. Thus, the applied load induces a crystal phase transformation, which is reversed when the load is removed (Fig. 25). This has potential for the development of simple devices which are self-centering and which perform repeatedly for a large number of cycles.

Several earthquake simulator studies of structures with SMA energy dissipators have been carried out. At the Earthquake Engineering Research Center (EERC) at the University of California [Aiken et al., 1992; and Clark et al., 1995], a three-story steel model was tested with Nitinol (nickel-titanium) tension devices as part of a cross-bracing system, and at the National Center for Earthquake Engineering Research [Witting and Cozzarelly, 1992], a five-story steel model was tested with copper-zinc-aluminum SMA devices. In this second study, devices with torsion, bending and axial deformation modes were investigated. Typical hysteresis loops from these tests are shown in Fig. 34. Results showed that the SMA dissipators were effective in reducing the seismic responses of the models. At present at EERC application of SMAs to the repair and/or upgrading of the connections of existing steel special moment-resisting frames (SMRFs), as well as to the design of new steel and precast reinforced concrete structures, is under investigation. Although these SMAs are very expensive relative to steel, it is believed that their use in the connections or in the construction of structural fuses to control the response of these SMRFs can be technically and economically very efficient when compared to other current solutions.

• Viscoelastic and Viscous Systems.

• Viscoelastic (VE) Materials have been in use in structural engineering for vibration control for more than 20 years. Mahmoodi [1969] described the characteristics of a double-layer, constrained-layer, VE shear damper. Viscoelastic copolymers developed by 3M Company have been used in a number of structural applications. Double-layer shear dampers of a 3M material were used in the 110-story twin towers of the World Trade Center in New York City, where a total of 10,000 dampers were installed in each tower to damp wind-induced dynamic response [Mahmoodi et al., 1987]. VE damping systems have since been adopted in several other tall buildings for the same purpose [Keel and Mahmoodi, 1987; Mahmoodi and Keel, 1989].

• A Viscous-Damping (VD) Wall System has been developed by Oiles and Sumitomo Construction (Fig. 35). Earthquake simulator tests of a full-scale four-story steel frame with and without VD walls showed very large response reductions -- up to 60 to 75% -- achieved by the walls [Arima et al., 1988]. A four-story reinforced concrete test building with VD walls was constructed in Tsukuba, Japan in 1987. It has since been monitored for earthquake response; observed accelerations are 25% to 70% lower than those of the building without VD walls [Arima et al., 1988]. A VD wall system in a 15-story building now already constructed in Shizuoka City, Japan, will provide between 20% and 32% damping to the building, and achieve response reductions of the order of 75% to 80% [Miyazaki and Mitsusaka, 1992]. Another type of wall damping system has been developed and tested by Kumagi-Gumi Corp. It is a super-plastic and silicone ribber VE shear damper that is included at the top connection of a wall panel to the surrounding frame [Uehara et al., 1991]. Earthquake simulator tests of a 1/2-scale three-story steel frame showed significant response reductions in the VE-damped model-, as large as 50% in story accelerations and 60% in story displacements.

Figure 35, VD wall and hysteresis loops [Miyazaki, 1992]

• An Elastomeric Spring Damper was recently successfully studied experimentally by Pekcan et al. [1995]. This damper, which has a distinctive re-centering characteristic, was used successfully to retrofit a non-ductile previously damaged 1/3-scale model RC building frame Structure.

• Fluid Viscous Dampers, which for many years have been used in the military and aerospace fields, are beginning to emerge in structural engineering. These dampers possess linear viscous behavior, are relatively insensitive to temperature changes, and can be very compact in size in comparison to force capacity and stroke. Experimental and analytical studies of building and bridge structures incorporating fluid viscous dampers made by Taylor Devices, Inc., have recently been performed [Constantinou et al., 1993]. Very large response reductions were achieved by the presence of the devices. The main feature of a pure viscous damper, that the damping force is out-of-phase with the displacement, can be a particularly desirable attribute for passive damping applications to buildings. If dampers are included in the structure in such a way that there is a column axial force component to the damper force (i.e., with a diagonal brace), then the out-of-phase peak damper force means that the peak induced column moments are less than if the peak damper force occurred at peak displacement.

• Applications of Viscoelastic and Viscous Dampers to Seismic Upgrading (Retrofitting) of Buildings.

• Viscoelastic Damper. An existing 13-story steel frame building, located in San Jose, California, has been retrofitted using viscoelastic shear dampers. In a number of recent small and moderate earthquakes, significant nonstructural damage had occurred in this building before its retrofitting. The initial brief from the building owner, the County of Santa Clara, was to formulate a plan to brace and secure nonstructural components in the building.

 

• Viscous Dampers. The Travelers' Hotel in Sacramento, California, is a seven-story reinforced concrete building over a complete basement. The plan dimensions of the building are 25.0 m by 49.4 m. The original building was constructed in the 1920s and a lightly framed penthouse was added in 1984 when the building was seismically upgraded.

• Design Provisions for Passive Energy Dissipation Devices. Design provisions for the implementation of supplemental passive energy dissipation systems have been developed by both the Structural Engineers Association of Northern California (SEAONC) and the Building Seismic Safety Committee (BSSC). The SEAONC group has drafted a document entitled "Tentative Seismic Design Guidelines for Passive Energy Dissipation Systems". The BSSC task group has developed design guidelines loosely based on the SEAONC document for inclusion as an Appendix to the 1994 NEHRP Recommended Provisions for the Development of Seismic Regulations of New Buildings [FEMA, 19941.

The Provisions provide general design guidelines applicable to a wide range of possible energy dissipation devices rather than addressing specific methods of energy dissipation. By remaining general, the Provisions rely on mandatory testing of system hardware to confirm the engineering parameters used in the design and to verify the adequacy of the energy dissipation systems. Some systems may not be capable of demonstrating acceptability by test and consequently would not be permitted.

The performance goals that form the basis of the Provisions are identical to those of conventional structures, that is, to prevent substantial loss of life by partial or complete building collapse in the design-level earthquake. No special protection against structural on nonstructural damage is sought or implied by the Provisions.

Exposure of energy dissipation devices to environmental effects that include wind-induced member actions, aging effects, high- and low-cycle fatigue, changes in operating temperature, exposure to airborne contaminants and moisture, may lead to either degradation or changes in their mechanical characteristics. Such effects must be accounted for in the design process.

SUMMARY, CONCLUSIONS AND FUTURE DIRECTIONS IN UPGRADING BUILDINGS USING PROTECTIVE SYSTEMS

• Base Isolation. It seems clear that the increasing acceptance of base isolation throughout the world will lead to many more applications of this technology. It is also clear that while elastomeric systems will continue to be used, there is a willingness to try other systems. The initial skepticism that was so prevalent when elastomeric systems were first proposed is no longer evident, and the newer approaches currently being developed will benefit from this more receptive climate and lead to the development of systems based on different mechanisms and materials.

• Passive Energy Dissipation Devices. Significant advances in the field of passive energy dissipation for improved structural resistance have been made in recent years. Developments in the research and analysis arena have been paralleled by significant improvements and refinement of available device hardware. Most, if not all, of these systems are now sufficiently well understood for their use in new or retrofit design of buildings.

CONCLUSIONS AND RECOMMENDATIONS

CONCLUSIONS. From the studies summarized above, the following conclusions can be made.

• Proper seismic upgrading of existing hazardous facilities is one of the most effective ways to reduce seismic risk in urban areas.

 
• To achieve proper upgrading requires: (1) reliable evaluation of the seismic activity at the facility site and its surroundings; (2) identification, according to the evaluation of both the seismic activity and the built environment around the facility, of the potential sources of seismic hazards to the facility; and (3) a reliable assessment of the vulnerability of the whole facility system (soil-foundation-superstructure nonstructural components and contents) to the identified sources of seismic hazards.

 
• A reliable assessment of the facility's vulnerability and seismic risk requires not only a proper evaluation of the dynamic characteristics (reactive mass, strength, stiffness, stability, period, damping and energy dissipation capacity) of its whole system, but also its response to the expected seismic excitations and the resulting losses as a fraction of its value.

 
• Present seismic codes in the U.S. not only do not provide any methodology, but do not even give any guidelines to properly identify seismic activity and potential sources of seismic hazards at a facility site.

 
• The codes' methodology and procedures for evaluating the dynamic characteristics and particularly the deformation capacity of the facility's whole system and of the response of this system to the expected critical EQGMs are inadequate.

 
• There are many uncertainties in conducting the seismic vulnerability assessments of an existing entire facility system, even when "as built" drawings exist. This assessment cannot be conducted just in the office by studying drawings and/or design computations for the original facility. It should be conducted through thorough field investigation of the state of the entire facility system.

 
• The importance of reliable assessment of the seismic risks of an existing facility cannot be overemphasized. If it is generally overestimated, rehabilitation costs may be prohibitive; underestimation can result in costly damages and loss of life.

 
• In the U.S., the commonly-used criterion for upgrading of seismically hazardous existing buildings is compliance with building code seismic regulations for EQ-RD of new buildings. This usually results in inefficient rehabilitation of the hazardous buildings, and in some cases in prohibitively uneconomical upgrading solutions.

 
• Although basic concepts and general guidelines for upgrading of structures can be and have been formulated, the proper upgrading of a given facility is a unique problem requiring a customized solution.

 
• Based on the design equation (DEMANDS < SUPPLIES) a designer has at his or her disposal the following three alternatives for upgrading an existing structure: (1) Decrease the seismic demands; (2) improve the supplies; and (3) a combination of (1) and (2).

 
• The use of an energy approach, based on the use of an energy balance equation, is a promising basis for rational selection of an efficient strategy for upgrading an existing hazardous facility.

 
• Although there are two different ways to establish the energy balance equation, the difference between the two resulting values of the input energy, E_I, is negligible in the period range of practical interest.

 
• The understanding and simultaneous examination of the basic design equation and the energy balance equation can guide the designer to a selection of the most efficient strategy and technique for seismically upgrading an existing hazardous structure.

 
• The selection of the most efficient strategy and technique depends on the main objective of the upgrading program, i.e., what are the required levels of performance (serviceability, continuous operation, life safety, impending collapse) desired of the upgraded facility for different levels of potential EQGMs and the associated seismic hazards projected as possible occurrences during the expected service life of the upgraded facility. Thus, the first and perhaps most difficult problem in upgrading an existing facility is the establishment of what can be called the "Redesign or Upgrading EQGM".

 
• For proper seismic upgrading of existing hazardous buildings, it is necessary to establish at least two of the following three levels of Redesign or Upgrading EQGMS: Service; Functional or Operational; and Safety or Survival. A promising and reliable approach for establishing these critical EQGMs is through analysis of the response spectra of the E_I, E_Hx , E_Hm , C_y, S_d or IDI, and DMI of these EQGMs.

 
• The ideal approach for obtaining an efficient upgrading program is first to select an efficient upgrading strategy and techniques, and then to plan the design and estimate no only the initial cost involved in upgrading the existing facility, but also the cost of the losses that can occur during the service life of the facility, considering the following three alternatives of possible expected future performances: (1) The structure will behave according to the requirements of the ideal philosophy of Earthquake-Resistant Design (EQRD), i.e., will realize all of the objectives of the presently generally accepted philosophy of EQRD for new buildings; (2) under the Design-Basic EQGM (DB-EQ), or safety EQ, the building will be not only safe but operational as well; and (3) the structure will remain without any damage even under the Maximum Credible or Capable EQ (MC-EQ). The advantages and total costs of each of these different designs should be explained to the client, and he or she should decide what is affordable.

 
• From analysis of the basic design and balance energy equation, the following three main upgrading strategies can be considered: (1) Decreasing the EQGM demands (E_I); (2) improving the facility's supplied mechanical (dynamic) characteristics (E_E and/or E_D = E_Hx + E_Hm ); and (3) a combination of (1) and (2). It should be recognized that strategies' (1) and (2) are not independent.

 
• The demands can be decreased by shifting the effective T of the structure. If the structure is on soft soil and T is smaller than T_GM , the efficient solution is to decrease T by increasing stiffness and/or decreasing the reactive mass. If the structure is on firm soil (i.e., small T_GM), and its yielding resistance is very small compared to that required, then attempts should be made to increase the effective T by means of base isolation. If base isolation cannot be used economically, it is desirable to decrease T either by increasing the supplied stiffness or by decreasing the mass or both, because both lead to smaller deformation demands. Another strategy to decrease the demands is to lower the demanded S_d (or IDI) and C_Y by increasing x or m _a or both. Decreasing S_d (or IDI and DMI) by increasing x is usually more effective than that obtained by increasing m and m _a.

 
• Several general guidelines are offered for improving the dynamic characteristics supplied to the existing structures, emphasizing that the strategy and technique for achieving such improvements depends not only on the characteristics of the building, but also the interaction of these characteristics with the dynamic characteristics of the established redesign (upgrading) EQGMs.

 
• In selecting the most appropriate retrofitting strategy and technique for improving the supplied dynamic characteristics, careful consideration should be given to the entire soil-foundation-superstructure-nonstructural components and contents system rather than just to the superstructure. Stiffening and strengthening the structure may lead to changes which are significant not only in the demands on the existing foundation but also in the soil-structure effects and on the response of the contents (accelerations).

 
• In the selection of the upgrading technique, the designer should consider not only the design that is the most technically efficient and least expensive in construction costs, but also the one which will require the minimum disturbance to the function (operation) of the building during the upgrading, Generally, the optimum upgrading strategy and technique are compromises between the ideal technical strategy and the least demanding in construction cost and the minimum disturbance to building occupants.

 
• The available techniques for restructuring existing buildings have been classified into two groups: the conventional (traditional) techniques, and the innovative techniques. Because in the U.S. the most-used traditional strategy for seismically upgrading existing facilities has been that of stiffening and particularly strengthening of their structure, the most-used conventional techniques for existing engineered buildings have been through the use of RC shear walls or steel bracing.

 
• Following the traditional strategy of strengthening and stiffening of the building structure, a very attractive and innovative technique which has been developed and applied since the 1985 Michoacán EQ is the use of post-tensioned steel rods or cables. The interest in application of this upgrading technique has increased significantly since 1985.

 
• By proper selection and design of the layout of the post-tensioned (PT) steel braces to keep them and the non-ductile frame members in their linear elastic range of behavior, and to control the maximum IDI in the building to values in the range for which a stable hysteretic behavior of the URM infills can be attained, it is possible to correct the existing weaknesses in translational and torsional stiffness, strength and energy dissipation capacities of an existing building in a manner that is economically and technically efficient.

 
• According to the recognized need to control damage to protect not only the entire structural system of the facility but also the nonstructural components, equipment, contents and service lines of the facility under the different levels of EQGMs, different types of "protective systems" have been developed and have begun to be applied for the seismic upgrading of existing buildings. At present, the most attractive type of protective systems are those based on base-isolation and passive energy dissipation.

 
• In the U.S., after the 1989 Loma Prieta EQ there has been a significant increasing acceptance of base isolation which has lead to several applications in the upgrading of buildings using this innovative technique. There is no doubt that base isolation will play a major role in future prospects for upgrading not only historic buildings but also relatively modem buildings sheltering essential/hazardous facilities and safety-critical facilities, as well as basic facilities. The challenge in applying base isolation to public and other buildings, such as schools and hospitals, in developing countries, is to develop a low-cost isolation system in conjunction with a simplified EQ-RD approach, such as Prescriptive Code Approach) and vernacular construction methods.

 
• There are a large variety of proposed energy dissipation devices, most of which have already been used in the seismic upgrading of existing buildings. These devices have been classified herein under the following groups: Friction, Metallic, Lead Extrusion, Shape Memory Alloys, and Viscous and Viscoelastic.

 
• The friction systems, except for the Fluor-Daniel EDR, generate rectangular hysteresis loops characteristic of Coulomb friction. Typically, these devices have good performance characteristics. The main issues with these devices are long-term reliability and maintenance, and possible permanent offsets after an earthquake,

 
• A wide variety of metallic systems have been developed that use flexural, shear or extensional deformation modes into the plastic range of deformation. A particularly desirable feature of these systems is their stable behavior, long-term reliability, and generally good resistance to environmental and temperature factors. The most-used metallic system is the one based on the use of yielding of mild steel. Many of the yielding steel systems use mild steel plates with triangular or hourglass shapes. Usually, these devices are mounted as part of a bracing or wall system, which must be substantially stiffer than the surrounding structure. The need for adding such stiff and heavy supporting system into the original structure may be prohibited in some cases,

 
• Within the family of passive energy dissipation devices, one class of smart materials which shows particular promise in civil structural applications is the class of metallic alloys known as shape memory alloys (SMAs). There are several characteristics which make them well-suited for such applications. First and most importantly, they have the ability to dissipate energy through repeated cycling without a significant degradation or permanent deformation. Their usable strain range is also much larger than for ordinary metals. SMAs show excellent resistance to environmental effects, such as corrosion, and possess an inherent fail-safe mechanism if strained beyond their working range. Potential disadvantages associated with SMAs include cost and sensitivity to temperature.

 
• Viscoelastic (VE) dampers (VED) have been in use in structural engineering for vibration control, particularly to damp wind-induced dynamic response, for more than two decades. The application of VE shear dampers to the seismic domain and particularly for seismic upgrading of buildings has occurred more recently. The inclusion of supplemental damping in the form of VED has already been proved to be a very cost-effective method for seismic upgrading of existing buildings. To obtain an efficient EQ-RD of a VED system, a number of factors that affect the properties of the VE material mechanical characteristics must be taken into account. The stiffness and damping properties of VE polymers are influenced by the level of shear deformation in the material, temperature and frequency of loading. Furthermore, similarly to the cases of the use of friction and metallic energy dissipation systems, the peak VED force and peak structure displacement occur nearly simultaneously and consequently careful consideration of the increase in the internal forces of the structural members must be made.

 
• A variety of viscous dampers (VD) have been developed recently and used in the EQ-RD and EQ-RC of new buildings, as well as in the seismic upgrading of existing buildings. The main feature and main advantage of VD over VED, as well as over frictional and metallic systems is that the forces in the VD are out of phase with the forces in the existing structural elements.

 

 
 
 
 
 
 
 
 
 

In the U.S., the use of fluid VD, which has been used for many years in the military and aerospace fields, are now being used in seismic structural engineering. They are relatively insensitive to temperature changes, and can be very compact in size in comparison to force capacity and stroke.
 

• A review of the application of the concepts and guidelines to the upgrading of the existing seismically hazardous buildings leads to the following main observations:

• For very flexible low rise RC buildings which are not strong enough and located on very soft soil, an efficient strategy is to decrease the fundamental period by increasing stiffness or reducing reactive mass, or both. Usually the increase in lateral stiffness should be accompanied by an increase in the elastic (yielding) strength. Two efficient conventional (traditional) techniques for increasing stiffness and strength are the use of bracing, usually steel, and RC shear walls. In the design of the bracing or shear walls, the following two approaches should be compared: One approach is to try to keep the response of the upgraded building in its elastic range (i.e., an elastic solution)-, the other approach is to use a protective system based on a passive energy dissipating solution, i.e., an attempt to dissipate part of the input energy through the use of energy dissipation devices.

• For rigid buildings located on firm soils, an efficient strategy is to use a protective system based on base isolation technology, which is a particularly attractive solution in case of historic buildings and essential and safety critical facilities. However, for the upgrading of an old six-story non-ductile RC framed infilled with URM walls, it was found that the elastic solution based on the use of high strength post-tensioned steel cables seems to be the most economical. For upgrading flexible frame buildings it was found that incorporation of energy dissipating devices appears to be the most promising strategy. Although the use of viscoelastic dampers has not only the advantage of improving (reducing) response in the inelastic (i.e., for safety) but also in the elastic range (i.e., for serviceability and functionality) because of their sensitivity to temperature and aging at present it appears that the use of yielding devices (which are only efficient in the inelastic range) appears to be more reliable. Very promising upgrading techniques include the addition of fluid viscous dampers and the use of SMAS.

• Research Needs.

• There is an urgent need to improve the reliability of seismic vulnerability assessment of existing facilities. As this assessment should be based on field inspection of the entire system of these existing facilities and ideally through simple but reliable experiments (tests), there is an urgent need to develop such reliable testing procedures and apply them. It is of utmost importance to monitor continuously the dynamic characteristics of the existing facilities.

• To help in understanding the mechanical behavior of real facilities, and so to improve the vulnerability assessment of real facilities, it is necessary to: (1) properly instrument entire building systems (soil-foundation-superstructure-nonstructural components and contents) having different structural systems (engineered and non-engineered); (2) to develop more efficient and reliable computer programs for the 3D nonlinear dynamic analysis of real facilities; (3) to conduct integrated analytical and experimental investigations of the energy dissipation capacities of the different building structural components as well as their basic subassemblages when these components are subjected to excitations reliable simulating the effects of the response of the building to critical ground motions; (4) to use earthquake simulator facilities to perform integrated analytical and experimental studies on the 3D seismic performance of different types of building systems.

• To improve the necessary seismic upgrading it is highly desirable to undertake integrated analytical and experimental studies to investigate not only the short-term applicability and response of the different techniques, particularly those based on use of new energy dissipating devices, but also their long-term reliability. It is also necessary to investigate whether there is a need to develop system maintenance methods.

• As most of the advances in the use of energy concepts for the needed seismic vulnerability assessment and the necessary upgrading solutions have been achieved through analytical studies conducted on SDOFS, and recognizing that real facilities are coi-nplex MDOFS, there is an urgent need to conduct integrated analytical and experimental studies on the validity of the results obtained from the SDOFS to realistic MDOFS.

• Development Needs.

• It is necessary that the results of the research be used to develop simple and reliable practical methods for applying energy concepts, as well as the derived energy balance equation. It is also necessary to develop low-cost efficient and above all more reliable (particularly in the long term) devices for base isolation and energy dissipation devices (particularly those based on dissipation of energy through viscoelastic and viscous damping) that can be used to control the E_E, E_Hx and E_Hm of the upgraded facility. There is an urgent need to develop simple but reliable code regulations for the upgrading of existing facilities.

• Education Needs.

• It is time that the use of the energy concepts be introduced into the education of our engineering students into the EQ-RD of new structures and the vulnerability assessment and upgrading of existing facilities. Furthermore, the practitioners (professional engineers) in regions of seismic risk should also be educated or at least exposed to the use of energy concepts. This information may be disseminated through short courses.

• Implementation Needs.

• Researchers should work with professional engineers and code officials to develop practical methods of assessing the vulnerability and proper upgrading of existing seismically hazardous facilities through the use of energy concepts, so that such methods can be introduced into the seismic codes.

Most of the studies on seismic vulnerability assessment of existing facilities and on the development of guidelines for seismic upgrading of existing buildings were sponsored by the National Science Foundation. Experiments conducted on the use of the post-tensioned steel rods were supported by a research grant from Mr. Rioboó and the experiments on the ADAS elements by a research grant from Bechtel. All of these supports are gratefully acknowledged. The author wishes to thank his ex-graduate students, I. Aiken, L. J. Alonso, M. Blondet, E. Fierro, J. Guh, E. Miranda, A. Terán-Gilmore and C. L. Thompson and A. Whittaker, who conducted most of the detailed studies to which this paper refers. The author would like to acknowledge also the contributions of Pat Crosby and Nabih Youssef for supplying information on the retrofit buildings. Special thanks are also to Prof. James Kelly and to Dr. Peter Clark for their invaluable assistance in supplying information and for the review of the protective systems. Thanks are also due to B. Young for editing and typing this paper.

REFERENCES

ABK, A Joint Venture (1984). Methodology of mitigation of seismic hazards in existing unreinforced masonry buildings: The Methodology ABK-TR-08. El Segundo, California.

Abrams, D.P. (1992). Strength and behavior of unreinforced masonry elements. Proceedings of the Tenth World Conference on Earthquake Engineering, Vol. 6, pp. 3475-3480; Madrid, Spain.

Aiken, I.D. (1994a). Applications of passive control systems for enhanced seismic protection. Notes for the Short Course on Advances in Earthquake Engineering, May-June 1994, University of California at Berkeley.

Aiken, I.D. (1994b). Seismic isolation -- an overview of applications and hardware. Notes for the Short Course on Advances in Earthquake Engineering, May-June 1994, University of California at Berkeley.

Aiken, I.D. et al. (1993) Analytical and experimental study of a mass damper -- testing of passive energy dissipation systems. Earthquake Spectra, Vol. 9, No. 3, August, Earthquake Engineering Research Institute, Oakland CA.

Aiken, I.D. Nims, D.K. and Kelly, J.M. (1992) Comparative study of four passive energy dissipation systems. Bulletin of the New Zealand National Society for Earthquake Engineering, September, Vol. 25, No. 3.

Aiken, I.D. and Kelly, J.M. (1990). Earthquake simulator testing and analytical studies of two energy-absorbing systems for multistory structures. Report No. UCB/EERC-90/3, October, Earthquake Engineering Research Center (EERC), University of California at Berkeley.

Aiken, I.D., Kelly, J.M. and Pall, A.S. (1988). Seismic response of a nine-story steel frame with friction dai-nped cross-bracing. Report No. UCB/EERC-88/17, Earthquake Engineering Research Center, University of California at Berkeley, November.

Alonso, L.J. (1989). Mechanical characteristics of X-plate energy dissipators. CE299 Report. University of California at Berkeley: Dept. of Civil Engineering.

Arima, F. et al. (1988). A study on buildings with large damping using viscous damping walls. Proceedings, 9WCEE, August, Tokyo/Kyoto, Japan.

Badoux, M., and Jirsa, J.0. (1987). Seismic retrofitting of reinforced concrete frame structures with steel bracing systems. Proceedings of the IABSE Symposium; Paris, France.

Bergman, D.M. and Goel, S.C. (1987). Evaluation of cyclic testing of steel-plate devices for added damping and stiffness. Report UMCE 87-10, November, Civil Engineering Department, University of Michigan.

Bertero, V.V. (1988). Earthquake hazard reduction: technical capability - U.S. perspective. Proceedings, 2nd Japan-U.S. Workshop on Urban Hazards Reduction. Shimizu, Japan, Tokai University, pp. 10-5 1.

Bertero, V.V. & Whittaker, A.S. (1989). Seismic upgrading of existing buildings. Proceedings, 5as Jornadas Chilenas de Sismología Ingeniería Antisísmica. 1:27-46. Santiago: Asociación Chilena de Sismología e Ingeniería Antisísmica.

Bertero, V.V. (1992). Seismic upgrading of existing structures. Proceedings, 10th World Conference on Earthquake Engineering, Madrid, Spain, 9:5105-5106.

Bertero, V.V. (1992). Major issues and future directions in earthquake-resistant design. Proceedings, 10th World Conference on Earthquake Engineering, Madrid, Spain, Post-Cotiference Volume, 37 pp.

Bertero, V.V. and Bertero, R.D. (1992). Tall reinforced concrete buildings: conceptual earthquake-resistant design methodology. Report No. UCB/EERC-92/16, EERC, University of California at Berkeley.

Bertero, V.V. and Uang, C.-M. (1992). Issues and future directions in the use of an energy approach for seismic-resistant design of structures. In Nonlinear seismic analysis and design of reinforced concrete structures, ed. P. Fajfar and H. Krawinkler, Elsevier Applied Science.

Bertero, V.V. (1993) Pathology in earthquake-resistant design facilities and the necessary therapy.

Bertero, V.V. and Terán-Gilmore, A. (1993). Use of energy concepts in earthquake-resistant analysis and design: issues and future directions. VIII SLIS, Primeras Jornadas Andinas de Ingeniería Estructural, pp. 1-39, Mérida, Venezuela.

Bertero, V.V. (1994) Comprehensive earthquake-resistant design approach. SEAOC Vision 2000 Report (I 995).

Boardman, P.R., Wood, B.J. and Carr, A.J. (1983). Union House -- a cross braced structure with energy dissipators. Bulletin of the New Zealand National Society for Earthquake Engineering, June, Vol. 16, No. 2.

Boroschek, R.L. and Mahin, S.A. (1991). Investigation of the seismic response of a lightly-damped torsionally-coupled building. Report No. UCB/EERC-91/18, EERC, University of California at Berkeley.

Bouadi, A. et al. (1995). Eccentrically braced frames for strengthening of a R/C building with short columns. Proceedings of the Fifth U.S. National Conference on Earthquake Engineering, Vol. III, pp. 612-629, EERI, Oakland CA.

Brokken, S.T. and Bertero, V.V. (1981). Studies on effects of infills in seismic resistant R/C construction. Report No. UCB/EERC 81/12, EERC, University of California at Berkeley.

Charieson, A.W., Wright, P.D. and Skinner, R.I. (1987). Wellington Central Police Station, base isolation of an essential facility. Proceedings, Pacific Conference on Earthquake Engineering, August, Vol. 2, New Zealand.

Ciampi, V. (1991). Use of energy dissipating devices, based on yielding of steel, for earthquake protection of structures. International Meeting on Earthquake Protection of Buildings, June, pp. 41/D-58/D, Ancona, Italy.

Clark, P.W. (1995). Experimental and analytical studies of shape memory dampers for structural control. Proceedings, Passive Damping, March, San Diego CA.

Constantinou, M.C. et al. (1993). Fluid viscous dampers in applications of seismic energy dissipation and seismic isolation. ATC-17-1, March, ATC.

Dowrick, O.J. (1992). Earthquake-resistant design for engineers and architects. John Wiley and Sons, Ltd.: New York.

Earthquake Engineering Research Institute, H.C. Shah, Chair (1984). Glossary of terms for probabilistic seismic risk and hazard analysis. Earthquake Spectra: 1: 1: 33-40.

Endo, T. (1982). Retrofitted structure against earthquake and research on them. Proceedings of the Third Seminar on Repair and Retrofit of Structures, University of Michigan, Ann Arbor, May 1982, pp. 28-31.

Fierro, E. and Perry, C. (1993). San Francisco retrofit design using added and stiffness elements. ATC-17-1, March, Applied Technology Council, San Francisco.

Filiatrault, A. and Cherry, S. (1990). A simplified design procedure for friction damped structures. Proceedings, 4th U.S. National Conference on Earthquake Engineering, May, Palm Springs CA.

Filiatrault, A. and Cherry, S. (1987). Performance evaluation of friction damped braced steel frames under simulated earthquake loads. Earthquake Spectra, May, Vol. 3, No. 5, EERI.

Foutch, D.A., Hjelmstad, K.D. and Del Valle Calderon, E. (1989). Investigation of two buildings shaken during the September 19, 1985 Mexico earthquake. EERI Publication 89-02, pp. 161-165, EERI, Oakland CA.

Fujita, S. et al. (1991). Seismic response of steel framed buildings using viscoelastic damper. Transactions, llth International Conference on Structural Mechanics in Reactor Technology, August, Vol. K2, pp. 109-114, Tokyo.

Gergely, P. et al. (1 994). Evaluation and modeling of infilled frames. Technical Report NCEER 94-0004; National Center of Earthquake Engineering Research; State University of New York at Buffalo.

Giachetti, R. et al. (1989). Seismic response of a DMRSF retrofitted with friction slip devices. Proceedings, First International Meeting on Base Isolation and Passive Energy Dissipation, Assisi, Italy.

Grigorian, C.E., Yang, T.-S. and Popov, E.P. (1992). Slotted bolted connection energy dissipators. Report No. UCB/EERC-92/10, July, EERC, University of California at Berkeley.

Guh, T.J. (1989). Seismic upgrading of buildings structures using post-tensioning. Thesis submitted to the University of California at Berkeley in partial fulfillment of the requirements for the degree of Doctor of Philosophy.

Guh, T.J. (1987). Analytical study of the seismic rehabilitation of a five story reinforced concrete waffle slab structure. CE299 Report, Dept. of Civil Engineering, University of California at Berkeley.

Jirsa, T.O. & Badoux, M. (1990). Strategies for seismic redesign of buildings. Proc. 4th US National Conference on EQ Engineering. 3:343-351. Oakland: EERI.

Kajima Corporation (1991). Honeycomb damper systems.

Keel, Ci. and Mahmoodi, P. (1986). Design of viscoelastic dampers for the Columbia Center Building. Building Motion in Wind, ASCE, New York.Kelly, J.M (1994a) Earthquake-resistant design using protective systems. Notes for the Short Course on Advances in Earthquake Engineering, May-June 1994, University of California at Berkeley.

Kelly, J.M. (1994b) "Mechanics and design of elastomeric bearings. Notes for the Short Course on Advances in Earthquake Engineering, May-June 1994, University of California at Berkeley.

Kelly, J.M., Skinner, R.I. and Heine, Aj (1972). Mechanisms of energy absorption in special devices for use in earthquake-resistant structures. Bulletin of the New Zealand National Society for Earthquake Engineering, September, Vol. 5, No. 3.

Kobori, T. et al. (1988). Effect of dynamic tuned connector on reduction of seismic response - application to adjacent office buildings. Proceedings, 9WCEE, Vol. 5, Tokyo

Langenbach, R. (1990). Traditional and contemporary construction practices utilizing unreinforced masonry in seismic areas. Proceedings of the Fourth U.S. National Conference on Earthquake Engineering, Vol. 3, pp. 263-272; Palm Springs, California,

Lin, R.C. et al. (1988). An experimental study of seismic structural response with added viscoelastic dampers. Report NCEER-88-0018, NCEER, State University of New York at Buffalo.

Mander, J.B. et al. (1994). Physical and analytical modeling of brick infilled steel frames. Technical Report NCEER-94-0004; National Center for Earthquake Engineering Research; State University of New York at Buffalo.

McNamara (1995). Seismic damage control with passive energy devices: a case study. Earthquake Spectra, May, Vol. 11, No. 2, pp. 217-232, EERI, Oakland CA.

Mahmoodi, P. and Keel, C.J. (1989). Analysis and design of multi-layer viscoelastic dampers for tall structures. Proceedings, 7th ASCE Structural Congress, May, San Francisco CA.

Mahmoodi, P. et al. (1987). Performance of viscoclastic dampers in World Trade Center Towers. Proceedings, 5th ASCE Structural Congress, Orlando FL.

Mahmoodi, P. (1969). Structural dampers. Journal of the Structural Division, Vol. 95(ST8), ASCE, New York.

Meli, R., et al. (1992). Experimental study on earthquake-resistant design of confined masonry

structures. Proceedings of the Tenth World Conference on Earthquake Engineering, Vol. 6, pp

3469-3474 ; Madrid, Spain.

Miranda, E. (1991). Seismic evaluation and upgrading of existing buildings. Ph.D thesis, 482 pp. University of California at Berkeley, California.

Miranda, E. and Bertero, V.V. (1991) Evaluation of the failure of the Cypress Viaduct in the Loma Prieta earthquake. Bulletin of the Seismological Society of America: 81-. 2070-2086.

Miranda, E. and Bertero, V.V. (1990). Post-tensioning technique for seismic upgrading of existing low-rise buildings. Proceedings of the Fourth U.S. National Conference on Earthquake Engineering, Vol. 3, pp. 393-402; Palm Springs, California.

Miranda, E. and Bertero, V.V. (1989). Performance of low rise buildings in Mexico City. Earthquake Spectra, Volume 5, Number 1, pp. 121-143-, Earthquake Engineering Research Institute; Oakland, California.

Miyazaki, M. and Mitsusaka (1992). Design of a building with 20% or greater damping. Proceedings, 10WCEE, Madrid.

Nims, D.K. et al. (1993). Application of the energy dissipating restraint to buildings. Base Isolation, Passive Energy Dissipation, and Active Control, ATC 17-1, March, Applied Technology Council.

Oiles Corporation (1991). General catalogue of earthquake isolation and vibration damping.

Pall, A.S., Ghorayeb, F. and Pall, R. (1991). Friction dampers for rehabilitation of Ecole Polyvalente at Sorel, Quebec. Proceedings, 6th Canadian Conference on Earthquake Engineering, pp. 389-396, Toronto.

Pall, A.S., Verganelakis, V. and Marsh, C. (1987). Friction dampers for seismic control of Concordia Library Building. Proceedings, 5th Canadian Conference on Earthquake Engineering, Ottawa.

Pekcan, G. et al. (1995). The seismic response of a 1:3 scale model RC structure with elastomeric spring dampers. Earthquake Spectra, May, Vol. II, No. 2, 1995.

Pincheira, J.A. and Jirsa, J.0. (1992). Post-tensioned bracing for seismic retrofit of RC frames. Proceedings of the Tenth World Conference on Earthquake Engineering, Vol. 9, pp. 5199-5204; Madrid, Spain.

Popov, E.P. and Yang, T. (1995). Steel seismic moment resisting connections. May, University of California at Berkeley.

Press, F. (1984). The role of science and engineering in mitigating natural hazards. Keynote address, Proc., 8th WCEE, San Francisco I- 13-24.

Qi, X. and Moehle, J.P. (1991). Displacement design approach for reinforced concrete structures subjected to earthquakes. Report No. UCB/EERC-91/02, Earthquake Engineering Research Center; University of California at Berkeley,

Richter, P.J. et al. (1990). The EDR -- energy dissipating restraint, a new device for mitigation of seismic effects. Proceedings, SEAOC 59th Annual Convention, September, Lake Tahoe.

Rioboó, J.M. (1989). Sistema de rigidización y refuerzo de estructuras mediante cables de presfuerzo. Proceedings of the VIII Congreso Nacional de Ingenieía Sísmica, Vol. III, G74-G82; Acapulco, Mexico.

Robinson, W.H. and Greenbank, L.R. (1976). An extrusion energy absorber suitable for the protection of structures during an earthquake. International Journal of Earthquake Engineering and Structural Dynamics, Vol. 4.

Robinson, W.H. and Cousins, W.J. (1987). Recent developments in lead dampers for base isolation. Pacific Conference on Earthquake Engineering, August, Vol. 2, New Zealand.

Roik, K., Dorka, U. and Dechent, P. (1988). Vibration control of structures under earthquake loading by three-stage friction-grip elements. Earthquake Engineering & Structural Dynamics, May, Vol. 16, No. 4, pp. 501-521.

Sakurai, T. et al. (1992). Application of joint damper to thermal power plant buildings. Proceedings, IOWCEE, Vol. 7, pp. 4149-4154, Madrid.

Sarrazin, M. and Moroni, M. (1992). Design of a base isolation confined masonry building. Proceedings, 10WCEE, Vol. 4, pp. 2505-2508, Madrid.

Sasaki, K. (1989). Experimental evaluation of nitinol for energy dissipation devices. Report No. UCB/EERC-89/20, Earthquake Engineering Research Center, University of California at Berkeley, December.

Schuller, M. et al. (1994). Performance of masonry-infilled R/C frames under in-plane lateral loads: experiments. Technical Report NCEER-94-0004; National Center for Earthquake Engineering Research; State University of New York at Buffalo.

SEAOC (1995). Vision 2000 Report on performance based seismic engineering of buildings. Prepared by the Structural Engineers Association of California, Sacramento, April 3, 1995.

Shimizaki, K. (1988). Strong ground motion drift and base shear coefficient for R/C structures. Proceedings of the Ninth World Conference on Earthquake Engineering, Vol. V, pp. 165-170; Tokyo, Japan.

Skinner, R.I. et al. (1980). Hysteretic dampers for the protection of structures from earthquakes.

Bulletin of the New Zealand National Society for Earthquake Engineering, March, Vol. 13, No.1

Stevenson, A. (1985). Longevity of natural rubber in structural bearings. Plastics and Rubber Processing and Applications, Vol. 5, p. 253.

Stiemer, S.F., Godden, W.G. and Kelly, J.M. (1981). Experimental behavior of a spatial piping system with steel energy absorbers subjected to a simulated differential seismic input. Report No. UCB/EERC-91/09, July, EERC, University of California at Berkeley.

Taylor, A.W., Lin, A.N. and Martin, J.W. (1992). Performance of elastomers in isolation bearings: a literature review. Earthquake Spectra, Vol. 8, No. 2, pp. 279-304.

Terán-Gilmore, A. (1993). On the design of RC buildings located on soft soils using an energy approach. CE299 Report; University of California at Berkeley.

Tsai, K.-C. and Hong, C.-P. (1992). Steel triangular plate energy absorber for earthquake-resistant buildings. Proceedings, First World Conference on Constructional Steel Design, Mexico.

Tyler, R.G. (1985). Tests on a brake lining damper for structures. Bulletin of the New Zealand National Society for Earthquake Engineering, September, Vol. 18, No. 3, pp. 280-285.

Tyler, R.G. (1985). Further notes on steel energy-absorbing elements for braced frameworks. Bulletin of the New Zealand National Society for Earthquake Engineering, Vol. 18, No. 3.

Tyler, R.G. (1978). Further notes on a steel energy-absorbing element for braced frameworks. Bulletin of the New Zealand National Society for Earthquake Engineering, December, Vol. 11, No. 4.

Uehara, K. et al. (1991). Experimental studies on a vibration control wall with viscoelastic material. Transactions, llth International Conference on Structural Mechanics in Reactor Technology, August, Vol. K2, pp. 115-120, Tokyo.

University of Michigan (1980, 1981 and 1982). Proceedings of the first (1980), second (1981) and third (1982) seminar on repair and retrofit of structures. Department of Civil Engineering, University of Michigan, Ann Arbor.

Vestroni, F. et al. (1992). Dynamic behavior of isolated buildings. Proceedings, 10WCEE, Vol. 4, pp. 2473-2478, Madrid.

Vezina, S. et al. (1992). Friction-dampers for aseismic design of Canadian Space Agency. Proceedings, 10WCEE, July, Madrid, Spain.

Warner, J. (1980). Repair and retrofit of buildings: overview of U.S. experiences. Proceedings of the First Seminar on Repair and Retrofit of Structures, Department of Civil Engineering, University of Michigan, Ann Arbor, pp. 41-57.

Whittaker, A.S. (1994). Passive control -- analysis and design issues. Notes for the Short Course on Advances in Earthquake Engineering, May-June 1994, University of California at Berkeley.

Whittaker, A.S. et al. (1991). Seismic testing of steel plate energy dissipation devices. Earthquake Spectra, November, Vol. 7, No. 4. pp. 563-604, EERI, Oakland CA.

Wilson, E.L. (1993). An efficient computational method for base isolation and energy dissipation analysis of structural systems. ATC-17-1, Applied Technology Council, San Francisco CA.

Witting, P.R. and Cozzarelli, F.A. (1992). Shape memory structural dampers: material properties, design and seismic testing. Report No. NCEER-92-0013, May, NCEER, State University of New York at Buffalo.

Yokota, H. et al. (1992). Structural control for seismic load using viscoelastic dampers. Proceeditigs, 10WCEE, Madrid.