2.krawinkler (1)

Yedinci Ulusal Deprem Mühendisliği Konferansı, 30 Mayıs- 3 Haziran, 2011, İstanbul Seventh National Conference on Earthquake Engineering, 30May- 3 June 2011, Istanbul, Turkey

YÜKSEK BİNALAR İÇİN YENİ DEPREM TASARIMI KILAVUZU VE UYGULAMALARI

RECENT SEISMIC DESIGN GUIDELINES FOR TALL BUILDINGS AND THEIR

IMPLEMENTATION

Helmut KRAWINKLER1

ÖZET

Bu bildiri yüksek binaların performansa göre tasarımına ayrılmıştır. Bildiride özellikle PEER Tall Building Initiative (TBI)’nin bir çalışması olarak kısa bir süre önce geliştirilen tasarım kılavuzuna vurgu yapılmıştır. TBI Tasarım Kılavuzu, Performansa Dayalı Deprem Mühendisliği’nin geleneksel kavramları olarak adlandırılabilecek esaslara dayanmaktadır: Tasarım iki özel performans düzeyinin gerçekleştirilmesi için yapılmakta ve günümüzde Performansa Dayalı Deprem Mühendisliği’nin ana çalışma alanı olan kayıp değerlendirmesi konusunu içermemektedir. Kayıp değerlendirmesi, kılavuzun bir gereği olmanın ötesinde, zorunlu olmayan bir tasarım sonrası çalışma olarak ele alınmaktadır. Kılavuz tarafından hedeflenen iki performans düzeyi, servis düzeyi ve öngörülen en büyük deprem düzeyidir. Bu performans düzeyleri, hasar başlangıcı ve göçme öncesi durumla tam anlamı ile ilişkili olmayıp, göreli olarak sıkça meydana gelen depremlerde aşırı hasara karşı korunmayı ve çok seyrek depremlerde can kaybını önleme bağlamında göçmenin önlenmesini amaçlamaktadır. Tasarım kılavuzu gelişmeye açık yeni yapı sistemlerinin uygulanması için önemli fırsatlar sunmaktadır. Ancak bu tür sistemlerin geleneksel yapı sistemlerine göre hem ekonomik, hem de can güvenliği bakımlarından üstün oldukları gösterilmelidir. Anahtar Kelimeler: yüksek binalar, performansa dayalı deprem mühendisliği, tasarım kılavuzu

ABSTRACT

This paper addresses performance-based design of tall building. The focus is on guidelines recently developed as part of PEER Tall Building Initiative (TBI). The TBI Design Guidelines contain what one may call conventional PBEE concepts, i.e., they use two specific performance levels to accomplish design and do not require loss assessment, which has become the core activity of modern PBEE. Loss asessment is a voluntary post-design activity, but not a requirement of the guidelines. The two performance levels targeted by the guidelines are a service level and a maximum considered earthquake level. These levels are not rigidly tied to damage threshold and collapse onset; they are intended to provide adequate protection against excessive damage under relatively frequent earthquakes and against collapse (a surrogate for casualties) under very rare earhquakes. The guidelines offer the opportunity to implement innovative structural systems, provided it can be demonstrated that these systems indeed present economic and life-safety advantages compared to conventional strutural systems. Keywords: tall buildings, performance-based earthquake engineering, design guidelines

INTRODUCTION It is becoming accepted practice in many countries to separate seismic design of tall buildings from general building code design procedures that are intended to be applied to low to mid-rise buildings. Many reasons can be quoted for this practice, including increasing importance of higher

1 Professor Emeritus, Dept. of Civil and Env. Engineering, Stanford University, Stanford, California, USA. [email protected]

2 Recent Seismic Design Guidelines for Tall Buildings and Implementation

mode effects, significant “dynamic redistribution” in the inelastic range, amplification of P-Delta effects, and the ability to devote more time (analysis effort) to assessment of nonlinear dynamic effects because of the larger engineering budget allocated to tall building design. But perhaps the two most important triggers for more elaborate design and evaluation are that (1) most presently utilized structural systems for tall buildings fall outside the code box (they are not listed as code-approved systems), and (2) recent developments in performance-based earthquake engineering (PBEE) made it feasible to assess non-code-conforming structural systems in a rational manner.

The latter has caused considerable professional excitement because it opens the door to the utilization of innovative systems that should lead to better performance at comparable or lower cost. In this context, cost is not necessarily limited to up-front construction cost; for many educated owners it implies lower life-cycle costs in which the penalty of slightly increased up-front construction cost is weighted against the benefit of reduced losses during the life time of the building. In this context it is critical to recognize that the cost of the structural system is a relatively small percentage of the building value (on the order of 20%), and that earthquake induced losses are dominated by non-structural losses in relatively frequent earthquakes and maybe dominated by structural losses and collapse potential in rare (large return period) seismic events. General concepts of PBEE and loss assessment are not the subject of this paper; the reader is referred to the extensive literature on these subjects (e.g., the Bled 2004 workshop proceedings [Fajfar and Krawinkler, 2004], and an upcoming special issue on building loss assessment of the International Journal on Earthquake Engineering and Structural Dynamics [to be published late in 2011]).

There is no unique way to define tall buildings, but the TBI Design Guidelines (PEER, 2010) define tall buildings as those that have

a fundamental translational period of vibration significantly in excess of 1 second, significant mass participation and lateral response in higher modes of vibration, and a seismic-force-resisting system with a slender aspect ratio such that significant portions of

the lateral drift result from axial deformation of the walls and/or columns as compared to shearing deformation of the frames or walls.

This paper uses the TBI Design Guidelines as a basis for discussion. Many paragraphs are a direct quotation from these guidelines. TBI stands for Tall Building Initiative, a development efforts conducted by the Pacific Earthquake Engineering Research (PEER) Center.

The author believes that the TBI Design Guidelines form a solid basis for design of tall buildings, but the paper contains personal interpretations, i.e., it is the full responsibility of the author and does not necessarily reflect the judgment of the other authors of the TBI Guidelines. İt is also emphasized that the TBI Design Guidelines contain what one may want to call conventional PBEE concepts, i.e., they use two specific performance levels to accomplish design and do not require loss assessment, which has become the core activity of modern PBEE. Loss asessment is a voluntary post-design activity, but not a requirement of the guidelines. Loss assessment test cases are summarized at the end of the paper. The concepts of conventional PBEE as applied to two specific performance levels is illustrated in Figure 1. The two performance levels targeted by the the TBI Design Guidelines are somewhere within the ovals shown in this figure. They are not equivalent to any of the performance levels used in FEMA 356 (or ASCE 41, 2007). For one, it is recognized that there is large uncertainty in defining performance levels and associated deformations, and secondly, the two performance levels are not rigidly tied to damage threshold and collapse onset.

Figure 1. Illustration of performance levels of FEMA 356 (ASCE 41, 2007) and TBI Design Guidelines

H. Krawinkler 3

It is important to point out that the TBI Design Guidelines represent a US West Coast view of the design process. They are written from the perspective of US design practice and economic constraints. Applications to other countries need to consider economic and societal constraint dictated by that country.

The guidelines discussed in this paper are based on research performed during the past one to two decades, and have been preceded by several documents of similar content and objective, such as SFDBI-AB-083 (2007), CTBUH (2008), and LATBSDC (2008). This serves to show that very few things are new, and that present criteria are an accumulation of past knowledge and development. And the author hopes there will be a follow-up containing further improvements to the guidelines discussed in this paper.

The objective of the TBI Guidelines is to provide guidance for design that will result in a building with equivalent (or better) performance capability to that anticipated for buildings that strictly comply with present code criteria.

Seismic design of tall buildings in accordance with these guidelines can offer a number of advantages including

more reliable attainment of intended seismic performance, reduced construction cost, accommodation of architectural features that may not otherwise be attainable, and use of innovative structural systems and materials

Notwithstanding these potential advantages, engineers contemplating building design using these procedures should be aware that implementation requires extensive knowledge of ground shaking hazards, structural materials behavior, and nonlinear dynamic structural response and analysis.

PERFORMANCE OBJECTIVES The TBI Design Guidelines use performance-based principles to achieve, as a minimum, the following objectives:

demonstrate that the structure will be capable of essentially elastic response and limited damage under Service-level Earthquake shaking having a mean return period of 43 years (50% exceedance probability in 30 years), and

demonstrate, with high confidence, that the structure will respond to Maximum Considered Earthquake shaking without loss of gravity-load-carrying capacity; without inelastic straining of important lateral-force resisting elements to a level that will severely degrade their strength; and without experiencing excessive permanent lateral drift or development of global structural instability.

Performance evaluation is carried out at two levels, one associated with “serviceability” and the other associated, implicitly, with safety against collapse. It may be desirable to design some structures to achieve performance superior to that described in the previous two paragraphs. To accomplish this, the guidelines can easily be modified to attain enhanced performance. Such modifications could include:

selection of an alternative, lower probability of exceedance, either for Service-level shaking or “Maximum Considered Earthquake” shaking, or both;

selection of more restrictive acceptance criteria, potentially including lower limiting levels of lateral drift and/or reduced levels of acceptable cyclic straining of ductile elements and larger margins for capacity-protected elements;

design of nonstructural components and systems to withstand shaking more intense or inter story drifts larger than that required by the building code;

design to limit residual displacements as a means of ensuring the structure can be repaired following earthquake ground shaking;


incorporating the use of damage-tolerant structural elements that are capable of withstanding cyclic inelastic deformation without degradation or permanent distortion; and,

incorporating the use of response modification devices including isolation systems, energy dissipation systems, and passive and active control systems to limit structural response.

HAZARD ANALYSIS, AND GROUND MOTION SELECTION AND INPUT Seismic Input The TBI Design Guidelines require determination of two levels of ground motion: a Service-level shaking motion and a Maximum Considered Earthquake shaking motion. Service-level motion is represented by a 2.5%-damped, acceleration response spectrum having a 43-year mean return period. Maximum Considered Earthquake shaking is represented by a 5%-damped acceleration response spectrum conforming to the requirements of ASCE-7 (see also FEMA P750, 2009) and a suite of earthquake ground acceleration records that have been appropriately selected and scaled to be “compatible” with this spectrum. Acceleration response spectra for both levels of ground shaking may be obtained from probabilistic or deterministic seismic hazard analysis. The use of appropriate damping has become a matter of much discussion. The Guidelines recommendation is to use at most 2.5% damping in all analyses. The use of 5%-damped elastic acceleration response spectra to represent Maximum Considered Earthquake shaking is only for convenience to allow comparison with spectra that are generated in compliance with the procedures of ASCE 7. Data discussed in ATC (2010) indicate that first mode damping in many tall buildings is less than 2% because of the lack of damping between foundation and soil; see data reported in Satake et al. (2003). Probabilistic Seismic Hazard Analysis Such hazard analysis should be performed for both levels of ground shaking using appropriate contemporary models for the description of regional seismic sources and ground motion prediction equations. Recent developments in those topics and the use of the models should be properly implemented in the probabilistic seismic hazard analysis code being used. The mechanics of probabilistic seismic hazard analysis is described in many publications (e.g., Stewart et al., 2001, McGuire, 2004). When conducting probabilistic seismic hazard analysis, proper consideration should be given to epistemic (modeling) uncertainties in the input source and ground motion prediction models and in the associated parameter values by including weighted alternatives in the analysis.

The following outcomes of probabilistic seismic hazard analysis should be reported: 1) mean ground-motion hazard curves at key structural periods including 0.2 seconds, 1.0 second, 2 seconds and the fundamental period of the structure; 2) uniform hazard spectra associated with the Service-level and Maximum Considered Earthquake shaking levels; and; 3) percentage contributions to the ground-motion hazard at the key structural periods for each hazard level. These contributions are a function of the seismic source, earthquake magnitude and source-to-site distance, and are evaluated from deaggregation of the seismic hazard. Uniform hazard spectra should be computed over a range of periods extending sufficiently beyond the building’s fundamental period to capture the effective (lengthened) building period during response to Maximum Considered Earthquake shaking.

Probabilistic seismic hazard analysis results for any location in the U.S. can be obtained using the USGS seismic hazard tool (http://earthquake.usgs.gov/research/hazmaps/index.php). The USGS site is well maintained and is kept current with respect to source models and ground motion predictive equations. When the building code or other seismic regulations call for a “site-specific analysis” of ground motion, a site specific probabilistic seismic hazard analysis is required in lieu of the USGS web site.

H. Krawinkler 5

Deterministic Seismic Hazard Analysis Deterministic seismic hazard analysis has the same components as probabilistic seismic hazard analysis (source model, ground motion predictive equations). The difference is that the range of possible results at each stage of the analysis is not considered in deterministic seismic hazard analysis. A single earthquake is considered with a prescribed magnitude and location. A single percentile-level of ground motion is taken from the ground motion predictive equation (e.g., 50 %-tile or median motion). The selections made at the various stages of deterministic seismic hazard analysis appear to be arbitrary and it is often difficult to know a priori whether the choices made lead to conservative or unconservative estimates of ground motion. Nevertheless, the ASCE 7 standard requires the use of deterministic seismic hazard analysis to provide a deterministic cap on ground motion in regions near major active faults to limit ground motion to levels deemed “reasonable” for seismic design. Selection and Modification of Accelerograms The TBI Design Guidelines recommend that the following three steps be employed in this process:

1. Identify the types of earthquakes that control the ground motion hazard. Where MCE shaking is controlled by probabilistic seismic hazard analysis, deaggregate the ground-motion hazard for the MCE spectral accelerations at the structural natural periods of interest, and use the results as the basis for selecting representative accelerograms for response history analysis. The structural natural periods of interest should include, as a minimum, the first three translational periods of structural response in each of the structure’s two principal orthogonal response directions.

2. Select a representative set of at least seven pairs of accelerograms recorded during past earthquakes that are compatible with the controlling events and site condition. Each accelerogram set selected must consist of at least two horizontal components, and in rare cases, the vertical component may also be included. If multiple magnitude-distance combinations contribute significantly to the hazard, then select records from each contributing earthquake as part of the total number of records. It needs to be recognized that the use of only seven pairs of accelerograms will make it impossible to do a reasonable statistical evaluation of response. There is considerable uncertainty in estimating a median, and there is unacceptable uncertainty in estimating a measure of dispersion.

3. Modify these motions in some manner to achieve a match with the target spectrum, either using spectral matching or amplitude scaling. If only seven pair of ground motions is used, the issue of spectral matching versus amplitude scaling is almost irrelevant because no relevant information on dispersion is obtained one way or the other. The Guidelines recommend matching records either to the uniform hazard spectrum or conditional mean spectrum. If the conditional mean spectrum approach (Baker, 2011) is used, it is recommended to use a suite of conditional mean spectra, each matched to one of the key periods of the structure. Use of conditional mean spectra for only the fundamental period is not recommended for tall buildings. The record selection process is still a much debated issue on which no consensus has been reached. It is recognized that selecting records that are matched to a uniform hazard spectrum is conservative (results in too high a demand for inelastic structural systems), whereas selecting records that are matched to a single conditional mean spectrum is generally unconservative because it pays insufficient attention to higher modes.

Soil-Foundation-Structure Interaction These effects should be considered when important. Tall buildings generally have subterranean levels to provide space for parking and other facilities. The most common foundation type is mats, although pile systems are used as well, particularly for tall buildings without subterranean levels. A schematic illustration of a building with subterranean levels is shown in Figure 2(a). Spatial variations of ground motion cause motions on foundation slabs (uFIM) to differ from free-field motions (denoted ug in Figure 2(a)), which is referred to as a kinematic interaction effect.


Figure 2. Schematic illustration of tall building with subterranean levels and simple models for analysis in which soil-foundation interaction effects are either neglected (b) or included in an approximate manner (c)

(from PEER 2010)

For Service-level response analysis it is deemed acceptable to extend the analytical models to the structure’s base, as shown in Figure 2(b). The subterranean levels should be included in the structural model, using appropriate element stiffness and capacities for structural members such as walls, columns, and slabs. Soil springs need not be included in the model. Motion should be applied at the base of the structure and can consist either of free-field motion (ug) or the foundation input motion (uFIM), which is modified for kinematic interaction effects.

Maximum Considered Earthquake response analysis should include, if practical, springs and dashpots representing foundation-soil interaction along basement walls and below the base slab, as shown in Figure 2(c). Ground motions should be input to the model via a rigid “bathtub” surrounding the subterranean portions of the structure. Input motion can consist either of free-field motion (ug) or the foundation input motion (uFIM), which is modified for kinematic interaction effects. If the above procedure is not practical for MCE analysis, use of option (b) in Figure 2(b) is deemed acceptable. If option (b) is used, since the soil springs are not included in the model, the mass of the subterranean levels may be modified. One option is to include the mass of the core tower below the grade, and exclude the mass of other extended elements in the subterranean levels.

SERVICE LEVEL EVALUATION Performance Objective The TBI Design Guidelines permit some, very limited, structural damage when a tall building is subjected to Service-level earthquake shaking. If not repaired, this damage should not affect the ability of the structure to survive future Maximum Considered Earthquake ground motions. However, repair may be necessary for cosmetic purposes and to avoid compromising the building’s long term integrity for fire resistance, moisture intrusion and corrosion. It is desirable that tall buildings remain operable immediately after a service-level event. Such performance is achievable with minor structural damage that does not affect either immediate or long term performance of the building and therefore does not compromise safety associated with continued building use. Analysis Method As a minimum, Service-level evaluation shall include a response spectrum analysis. When demand to capacity ratios determined from such analysis exceed acceptable levels, nonlinear response history analysis may be used to demonstrate acceptable performance. The analytical model should be three-dimensional, and seismic input should be applied in two orthogonal horizontal directions simultaneously (and also in the vertical direction if deemed necessary). Torsion caused by inherent

H. Krawinkler 7

eccentricities resulting from the distribution of mass and stiffness should be included. Accidental eccentricities need not be considered. Global Modeling Issues Floor diaphragms should be included in the mathematical model using realistic stiffness properties. Regardless of the relative rigidity or flexibility of floor diaphragms, floors with significant force transfer (e.g., podium effect) should be explicitly included in the mathematical model. Diaphragm chord and drag forces should be established in a manner consistent with the floor characteristics, geometry, and well established principles of structural mechanics. The analytical model of the structure should include the entire building including the subterranean levels (i.e., floors, columns, walls, including the basement walls). Guidance for incorporating podium effects and diaphragm forces are provided in ATC (2010). Component Modeling for Analysis Structural models should incorporate realistic estimates of stiffness and strength considering the anticipated level of excitation and damage. Expected, as opposed to nominal or specified, properties should be used when computing stiffness properties in all analysis cases and when computing strength properties for nonlinear response history analysis. In lieu of detailed justification, values provided in Tables 1 and 2 may be used for expected material strengths and estimates of component stiffness, respectively.

Table 1. Expected Material Strengths (from PEER 2010)

Material Expected Strength Structural Steel Hot-rolled structural shapes and bars ASTM A36/A36M 1.5Fy

ASTM A572/A572M Grade 42 (290) 1.3 Fy

ASTM A992/A992M 1.1 Fy

All other grades 1.1 Fy

Hollow Structural Sections ASTM A500, A501, A618 and A847 1.3 Fy

Steel Pipe ASTM A53/A53M 1.4 Fy

Plates 1.1 Fy

All other Products 1.1 Fy

Reinforcing Steel 1.17 times specified Fy

Concrete 1.3 times specified f’c

Table 2. Effective Component Stiffness Values (from PEER 2010)

Component Flexural Rigidity

Shear Rigidity

Axial Rigidity

Structural steel Beams, Columns and Braces ESI GSA ESA Composite Concrete Metal Deck Floors 0.5EcIg GcAg EcAg R/C Beams – nonprestressed 0.5EcIg GcAg EcAg R/C Beams – prestressed ECIg GcAg EcAg R/C Columns 0.5EcIg GcAg EcAg R/C Walls 0.75EcIg GcAg EcAg R/C Slabs and Flat Plates 0.5EcIg GcAg EcAg Notes: Ec shall be computed per ACI 318, using expected material strength per Table 1

Gc shall be computed as 2 1cE , where shall be taken as having a value of 0.2


Acceptance Criteria The TBI Design Guidelines make a clear distinction between acceptance criteria for response spectrum analysis and nonlinear response history analysis. If a response spectrum analysis is performed, individual component force actions are permitted to have demand-capacity ratios as high as 1.5. But capacities in this case have to be computed from design quantities based on nominal material properties and incorporating -factors. If nonlinear response history analysis is performed, deformation demands should not exceed a value at which sustained damage requires repair, for reasons of strength deterioration or permanent deformation, as demonstrated by appropriate laboratory testing. For actions that are not considered as ductile, the force demands should not exceed nominal capacities (i.e., using specified material strengths) multiplied with applicable code resistance factors. In all cases, the mean story drift should not exceed 0.5% of story height in any story. Note of Caution for Strength Design The TBI Design Guidelines do not include a design level earthquake evaluation. Engineers can use service-level earthquake shaking, together with wind demands, to set the structure’s strength in preliminary design, which is later confirmed for adequacy as part of the MCE shaking evaluation. In regions of relatively high seismicity such as Los Angeles, San Francisco and Seattle, service-level shaking will result in required building strength that is of the same order as the strength required using the prescriptive building code procedures [in US seismic codes, strength design for tall buildings is usually controlled by a code minimum base shear]. However, in cities with lower seismicity [examples in the US include Portland, Oregon; Sacramento, California; and, Salt Lake City, Utah], service-level shaking may result in substantially less required strength than would be obtained from presently employed building codes. Engineers designing buildings in such locations should be aware of this and consider that service-level strength requirements may not result in a building of adequate strength. The TBI Design Guidelines do not demand consideration of a minimum base shear for long period structures.

MAXIMUM CONSIDERED RESPONSE EVALUATION Performance Objective The objective of this evaluation is to provide adequate safety against collapse. This objective is implicitly achieved by using nonlinear response history analysis to evaluate the structure’s response to a relatively small suite of ground motions that represent MCE shaking as defined previously. This response evaluation does not provide a quantifiable margin against (or a probability of) collapse, but is intended to demonstrate that collapse under the selected ground motions does not occur, i.e., the structure maintains stability, and forces and deformations are within acceptable limits. As noted in the commentary to the 2009 NEHRP Provisions (FEMA P750, 2009), for Occupancy (Risk) Category II structures, the target conditional probability of collapse is intended to be 10% or less, with lower acceptable collapse probabilities for structures in higher Occupancy Categories.

The technical capability exists to predict the probability of collapse as a function of ground motion intensity (FEMA P695, 2009, Zareian and Krawinkler, 2007), however the process of collapse prediction is complex and is based on the presumption that force-deformation characteristics of all important structural components can be modeled for the full range of deformations associated with inelastic behavior leading to collapse. At this time insufficient knowledge exists to model such behavior with confidence for all types of structural components that might be utilized in tall buildings, and tools available to engineers do not permit such evaluations within the resources and time constraints available on most design projects. Until such knowledge is developed and available tools improve to a level that will permit practical

H. Krawinkler 9

implementation of rigorous collapse probability evaluation, the stability evaluation recommended herein is the preferred method for providing adequate safety against collapse. Analysis Method Response prediction should be based on nonlinear response history analysis (NRHA) using a three-dimensional model of the structure including subterranean levels. Soil-foundation-structure interaction effects may be included as discussed previously. Ground motion shall be introduced at the base mat or through soil springs as described previously, and applied in two orthogonal horizontal directions simultaneously (and also in the vertical direction if deemed necessary). Torsion caused by inherent eccentricities resulting from the distribution of mass and stiffness should be included. Accidental eccentricities need not be considered.

The need for NRHA in performance assessment of tall buildings has been demonstrated in many publications. Issues such as P-Delta effects, and moment and shear force amplification due to higher mode effects and inelastic redistribution cannot be evaluated by means of elastic modal analysis or nonlinear static (pushover) analysis). A simple illustration is provided in Figure 3, which shows median story overturning moments and shear forces for a 16-story shear wall structure obtained from NRHA, compared to results obtained from a pushover analysis.

Figure 3. Story overturning moment and shear force amplification in a 16-story shear wall structure

comapred to results obtained from pushover analysis

Global Modeling Issues All global modeling issues summarized under Service Level Evaluation apply here, with special emphasis that the three-dimensional model of the structural system should comprise all components and force and deformation characteristics that significantly affect the prediction of seismic demands at the Maximum Considered Response level. P-Delta effects must be incorporated in the analytical model, whether or not elastic concepts indicate that such effects are important. Component Modeling for Analysis Hysteretic models should adequately account for all important phenomena affecting response and demand prediction as the structure approaches collapse including: (a) monotonic response beyond the point at which maximum strength is attained; (b) hysteretic properties characterizing component behavior without the effect of cyclic deterioration; and, (c) cyclic deterioration characteristics. Simplified component models may be used if hysteretic models that incorporate conditions (a) to (c) cannot be developed and incorporated with confidence in the analysis tool. But such simplified models have to provide realistic estimates of component strength and low estimates of component deformation capacities.

The guideline recommendations are founded on simplified models of component behavior derived from phenomenological models based on experimental data that have been collected in databases over the last 10+ years (e.g., Haselton and Deierlein, 2007, Lignos and Krawinkler,

Maximum Overturning moment Profile (SW)N = 16, T1 = 1.6 sec., y = 0.250, Stiff & Str = Uniform, = 0.05

p = 0.03, pc/p = 1, = 20, Mc/My = 1.1

0

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

0.00 0.20 0.40 0.60 0.80 1.00 1.20

Maximum Story OTM / (2/3HyW)

Flo

or

ID

Pushover, NEHRP Pattern k = 2

Median of response history analysis @ Sa(T1) = 0.625

Maximum Story Shear Profile (SW)N = 16, T1 = 1.6 sec., y = 0.250, Stiff & Str = Uniform, = 0.05

p = 0.03, pc/p = 1, = 20, Mc/My = 1.1

0

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

0.00 0.50 1.00 1.50 2.00 2.50

Maximum Story Shear / (yW)

Sto

ry ID

Pushover, NEHRP Pattern k = 2

Median of response history analysis @ Sa(T1) = 0.625


2009). Experimental data show clearly the deterioration characteristics inherent in inelastic component behavior, which are reflected in the cyclic response and in a marked difference between the monotonic loading (backbone) curve and a cyclic envelope curve obtained by connecting the peaks of individual cycles, as is illustrated in Figure 4.

Figure 4. Monotonic and cyclic responses of identical steel beam specimens (from ATC 2010)

Deformation capacities for simplified models may be taken equal to the corresponding Collapse Prevention values for primary elements published in ASCE 41 (with Supplement 1) for nonlinear response procedures, or may be based on analytical models validated by experimental evidence. When applicable, the ASCE 41 component force versus deformation curves may be used as modified backbone curves, with the exception that the drop in resistance following the point of peak strength should not be as rapid as indicated in the ASCE 41 curves. For many components of new structures, ASCE 41 values might be very conservative (low) or might not be applicable.

Alternatively, the modeling options presented in ATC (2010) may be employed. Four options are presented in Section 2.2.5 of ATC (2010), covering the range from explicit consideration of all important cyclic deterioration effects to disregard of all deterioration effects. The more simplified the model, the larger is the “penalty” on the deformation capacities inherent in the model. This penalty applies to the capping deformation (the deformation associated with the peak (capping) strength) and to the ultimate deformation capacity, u. The latter is the deformation beyond which the confidence in the model prediction becomes so low that it has to be assumed that the component strength drops to zero. The choice of an appropriate component modeling option and of the basic hysteresis model used to represent the cyclic response of structural components should be justified and become part of the analysis documentation. Section 2.2.5 of ATC (2010) proposes the following four options for component analytical models.

Option 1 – explicit incorporation of cyclic deterioration in analytical model: This option explicitly incorporates all important modes of strength and stiffness deterioration in the analytical model, by using the monotonic backbone curve as a reference boundary surface that moves “inward” (towards the origin) as a function of the loading history. This is the preferred option. There are many component models that account explicitly for cyclic deterioration. Most are based on a backbone curve similar to that shown in Figure 5, a set of cyclic deterioration rules, and a set of rules that control the basic shape of the hysteresis diagram (e.g., Ibarra et al., 2005).

H. Krawinkler 11

Figure 5. Parameters of the initial backbone curve of the Ibarra-Krawinkler model (from ATC 2010)

Option 2 – use of a cyclic envelope curve as a modified backbone curve; cyclic deterioration is not considered explicitly. If the cyclic envelope curve (the curve enveloping the cyclic force-deformation response of a component) is known (e.g., from a cyclic test that follows a generally accepted loading protocol) then it is acceptable to use this envelope curve as the modified backbone curve for analytical modeling and ignore additional deterioration - provided that no credit is given in the analysis to undefined strength characteristics beyond the bounds established by the cyclic envelope curve, i.e., the ultimate deformation u in any analysis should be limited to the maximum deformation recorded in the cyclic test. When using this approximation, one must make sure to include the negative tangent stiffness portion of the cyclic envelope curve as part of the modified backbone curve of the analytical model.

Option 3 – use of factors for modification of backbone curve; cyclic deterioration is not considered explicitly: If only the monotonic backbone curve is known (or predicted from analysis) and cyclic deterioration is not incorporated in the analytical model, then the shape of the backbone curve must be modified to account approximately for cyclic deterioration effects. Numerical values of the modification factors might depend on material, configuration, detailing of the structural component, and loading history. Recommended values for the parameters defining the modified backbone curve are provided in ATC (2010).

Option 4 – no deterioration in analytical component model: If the post-capping (negative tangent stiffness) portion of the modified backbone curve of option 2 or 3 is not incorporated in the analytical model (i.e., a non-deteriorating model is employed), then the ultimate deformation of the component should be limited to the deformation associated with 80% of the strength cap on the descending branch of the modified backbone curve as obtained from option 2 or 3. No credit should be given in analysis to undefined strength characteristics beyond this deformation limit.

Figure 6 illustrates the four options for a typical experimental cyclic loading history and a peak-oriented hysteresis model. Several equivalent points of equal peak displacement for the four options are identified with symbols. The differences appear to be small, but primarily because the illustrations are for a symmetric and step-wise increasing loading history, which is typical for experimental studies but not for response at the MCE shaking level. As intended, the larger the simplification the more the inelastic deformation capacity is being reduced. This is most evident in Figures 6(c) and (d), in which the attainment of the estimated u severely limits the inelastic deformation capacity.

F

Fc

Fy

Fr

y c r u

Ke

p pc


Figure 6. Illustration of implementation of four options for analytical component modeling(from ATC 2010)

Acceptance Criteria The TBI Guidelines make the following recommendations for acceptance at the MCE level.

Force controlled actions at the component level should fulfill the following criterion (force controlled implies that the component does not have dependable inelastic deformation capacity):

Fu ≤ Fn,e (1) where Fu is 1.5 times the mean demand obtained from NRHA (there are exceptions to the 1.5 factor); Fn,e is the nominal strength from applicable material codes but based on expected material properties; and is the resistance (strength reduction) factor. If failure in a force controlled action does not result in structural instability or potentially life-threatening damage (such as shear failure in coupling beams of shear walls), then Fu is the mean demand and may be taken as unity.

Deformation controlled actions at the component level: if the ultimate deformation capacity (u, see Figure 6) is exceeded in any of the NRHAs, the strength associated with this mode of deformation should be assumed as zero and the effects of component failure on related strength quantities and the stability of the structure should be evaluated.

The guidelines provide also global acceptance criteria at the MCE level. The mean transient story drift ratio in each story should be less than 0.03, and the maximum transient story drift ratio in any analysis is limited to 0.045. The mean residual story drift ratio in each story should be less than 0.01, and the maximum residual story drift ratio in any analysis is limited to 0.015 unless justification is provided and accepted by a peer review. In any nonlinear response history analysis, the loss in story strength associated with maximum deformations should not exceed 20% of the initial strength.

Monotonicbackbone curveMonotonicbackbone curve

u

Modifiedbackbone curveModifiedbackbone curve

(a) Option 1 – with cyclic deterioration (b) Option 2 – modified backbone curve = cyclic envelope curve

u


Fc

0.8Fc

u


(c) Option 3 - modified backbone curve = (d) Option 4 – no strength deterioration factored montonic backbone curve

H. Krawinkler 13

PEER REVIEW

Because of the complexity of the analyses used to demonstrate building performance (which explicitly include nonlinear response effects), most US building departments have initiated a requirement for independent peer review when designs are submitted for permit under the alternative means and methods clause. The composition of the peer review panel should typically be jointly determined by the owner/design team and the building department. Members of the peer review team may be added as appropriate to fully address the special features of the proposed project that are not evident at initiation of the process. The peer reviewers should jointly possess expertise in geotechnical engineering and seismic hazards, seismic performance of structures as a whole, as well as knowledge of the design and behavior of structures with elements of the type employed and structural design of tall buildings.

IMPLEMENTATION EXAMPLES The TBI Design Guidelines are being tested on a series of tall building designs that have been executed for three types of tall building structural systems. Three designs were performed for each structural system: a code design according to IBC 2006, denoted as Design A, a design executed in accordance with the LATBSDC (2008) criteria, Design B, and a design in accordance with the TBI Design Guidelines, Design C. The building systems considered are a 42-story core-wall hotel tower (Bldg#1), a 42-story dual-system building (Bldg#2), and a 40-story buckling-restrained braced (BRB) steel-framed office building (Bldg#3).

The 42-story core wall building (Bldg#1) is a representative residential or hotel tower in the West Coast of the United States. In this building, a central elevator and a stair core are surrounded by hotel rooms. A detailed description of the design of the structure can be found in Fry et al. (2009). The 42-story dual system (Bldg#2) has a penthouse and four subterranean levels. The lateral system consists of a concrete shear wall core and two four-bay concrete special moment resisting frames (SMF) in each direction, Ghodsi et al. (2009). The 40-story BRB office building (Bldg#3) has buckling-restrained Chevron braces. The columns in the braced bays are made up of built-up square-box columns infilled with high strength concrete to resist the large compression force demands. For Design 3C three levels of outriggers were used in one direction to reduce overturning moments and drift. Design details are presented in Dutta and Hamburger (2009).

The 3-D models of these buildings were analyzed, using the Perform 3D software, at five different performance levels (SLE-25 [25 years return period], SLE-43 [43 years return period], DBE [475 years return period], MCE [2475 years return period], and OVE [4975 years return period]). The ground motions were selected based on uniform hazard spectra at the Civic Center in Los Angeles. 15 pairs of records are selected at each performance level such that the average spectrum of the selected records closely matches the target uniform hazard spectrum (UHS) primarily at long periods (1 to 7.5s period range) at the selected site. Information on the spectra and ground motions can be found at the PEER website http://peer.berkeley.edu/tbi/tasks/task-10.

The purpose of the test designs and analyses is to assess performance and losses obtained from conventional designs and the new TBI Design Guidelines. Loss assessment implies estimation of structural and nonstructural damage and associated repair costs. Methods for accomplishing loss assessment have been developed as part of the PEER Center research program (e.g., Krawinkler and Miranda, 2004) and more recently as part of the FEMA sponsored ATC-58 project (ATC, 2009). Basis for this loss assessment are damage fragility and consequence functions, information on engineering demand parameters, such as peak story drifts and peak floor accelerations, and collapse probability represented by the collapse fragility curve. Usually, direct economic losses are determined separately for structural components and for drift and acceleration sensitive nonstructural components and subsystems, and then are aggregated into total losses. Analysis results available for peak story drifts for the test structure designs C (TBI Design Guidelines) indicate that the losses should be smallest for the BRB structural system, see Figure 7. But total loss depends also on acceleration dependent nonstructural damage, and often is dominated by this type of damage. At the time of writing, comprehensive data on peak floor accelerations,


collapse fragility, and damage fragility curves and consequence functions implemented in the loss assessment process are not available to the author, and therefore no definite judgment can be passed on the advantages of various structural systems.

Figure 7. Medians of maximum story drift ratios for the three TBI Guideline designs at the five performance levels (results not to same scale); left = core wall system, center = RC dual system, right = steel BRB system

Results of loss assessment are still under review, and the reader is advised to consult the PEER TBI website (http://peer.berkeley.edu/tbi) for a final assessment of losses for the various structural systems and design procedures. As a preliminary conclusion it can be stated that implementation of the TBI Design Guidelines did lead to a clear reduction in losses at specific hazard levels and on an annualized basis (expected annual loss).

CONCLUSIONS The author believes that the TBI Design Guidelines form a solid basis for design of tall buildings. These guidelines contain what one may want to call conventional PBEE concepts, i.e., they use two specific performance levels to accomplish design and do not require loss assessment, which has become the core activity of modern PBEE. Loss asessment is a voluntary post-design activity, but not a requirement of the guidelines. The two performance levels targeted by the TBI Design Guidelines are a service level and a maximum considered earthquake level. These levels are not rigidly tied to damage threshold and collapse onset; they are intended to provide adequate protection against excessive damage under relatively frequent earthquakes and against collapse (a surrogate for casualties) under very rare earhquakes. Perhaps most important, the TBI Guidelines offer the opportunity to implement innovative structural systems, provided it can be demonstrated that these systems indeed present economic and life-safety advantages compared to conventional strutural systems.

ACKNOWLEDGEMENTS The TBI Design Guidelines were developed under the auspices of the Pacific Earthquake Engineering Research (PEER) Center’s Tall Building Initiative. Funding for writing the guidelines was provided by a grant from the Charles Pankow Foundation. The opinions expressed in this paper are those of the author and may not necessarily reflect those of the sponsor or the co-authors of these guidelines.

H. Krawinkler 15

REFERENCES ASCE 41 (2007). Seismic Rehabilitation of Existing Buildings (ASCE/SEI 41/06 plus Supplement 1),

American Society of Civil Engineers, Reston, VA, 416 pp. ATC (2010). “ATC-72-1: Interim Guidelines on Modeling and Acceptance Criteria for Seismic Design and

Analysis of Tall Buildings,” ATC-72-1, Applied Technology Council, Redwood City, California. ATC (2009). Guidelines for Seismic Performance Assessment of Buildings, ATC-58 50% Draft, Applied

Technology Council, Redwood City, California. Baker, J.W. (2011). "Conditional Mean Spectrum: Tool for ground motion selection," Journal of Structural

Engineering, 137(3), 322-331. CTBUH (2008). Recommendations for the Seismic Design of High-Rise Buildings, Michael Willford,

Andrew Whittaker & Ron Klemencic, Council for Tall Buildings and Urban Habitat. Dutta, A., and Hamburger, R.O. (2009). “Case study of a 40-story buckling-restrained braced frame building

located in Los Angeles,” The Structural Design of Tall and Special Buildings, 19(1-2), 77-93. Fajfar, P, and Krawinkler, H. (2004). Editors, Performance-Based Seismic Design – Concepts and

Implementation, PEER Special Publication, PEER 2004/05, September 2004, ISBN 0-9762060-0-5. FEMA P695 (2009). “Quantification of Building Seismic Performance Factors,” Federal Emergency

Management Agency, Washington, D.C. FEMA P750 (2009). “NEHRP Recommended Provisions for Seismic Regulation of Buildings,” Federal

Emergency Management Agency, Washington, D.C. Fry, J.A., Hooper, J.D., and Klemencic, R (2009). “Core wall case study design for Pacific Earthquake

Engineering Research/ California seismic safety commission,” The Structural Design of Tall and Special Buildings, 19(1-2), 61-75.

Ghodsi, T., Ruiz, J.F., Massie, C., and Chen, Y. (2009). “PEER/ SSC tall building design case history no. 2,” The Structural Design of Tall and Special Buildings, 19(1-2), 197-256.

Haselton, C.B., and Deierlein G.G. (2007). Assessing Seismic Collapse Safety of Modern Reinforced Concrete Moment Frame Buildings, PEER Report 2007/08, Pacific Earthquake Engineering Research Center, University of California, Berkeley, California.

Ibarra L.F., Medina R.A., and Krawinkler, H. (2005). “Hysteretic models that incorporate strength and stiffness deterioration,” Earthquake Engineering and Structural Dynamics, Vol. 34, No. 12, pp. 1489-1511.

Krawinkler, H., and Miranda, E. (2004). “Performance-based Earthquake Engineering,” Chapter 9 of Earthquake Engineering: from engineering seismology to performance-based engineering, V.V. Bertero and Y. Bozorgnia, Editors, CRC Press, pp. 9.1-9.59.

LATBSDC (2008), An alternative Procedure for Seismic Analysis and Design of Tall Buildings located in the Los Angeles Region, Los Angeles Tall Buildings Structural Design Council, 2008 Edition.

Lignos D.G., and Krawinkler H. (2009). Sidesway Collapse of Deteriorating Structural Systems under Seismic Excitations, Report No. TR 172, John A. Blume Earthquake Engineering Center, Department of Civil Engineering, Stanford University, Stanford, California.

McGuire, R.K. (2004). “Seismic Hazard and Risk Analysis,”: Earthquake Engineering Research Institute Monograph MNO-10.

PEER (2010). Guidelines for Performance-Based Seismic Design of Tall Buildings, Pacific Earthquake Engineering Center Report No. 2010/05.

http://peer.berkeley.edu/publications/peer_reports/reports_2010/web_PEER2010_05_GUIDELINES.pdf Satake, N., Suda, K, Arakawa, T., Sasaki, A., and Tamura, Y. (2003). “Damping evaluation using full-scale

data of buildings in Japan,” Journal of Structural Engineering, American Society of Civil Engineers, Vol. 129, No. 4, pp. 470-477.

SFDBI-AB-083 (2007), Seismic design and review procedures for new tall buildings, Administrative Bulletin, San Francisco Department of Building Inspection.

Stewart, J.P., Chiou, S.-J., Bray, J.D., Somerville, P.G., Graves, R.W., and N.A. Abrahamson (2001). “Ground motion evaluation procedures for performance based design,” Rpt. No. PEER-2001/09, Pacific Earthquake Engineering Research Center, University of California, Berkeley.

Zareian, F., Krawinkler, H. (2007) “Assessment of probability of collapse and design for collapse safety”, Earthquake Engineering and Structural Dynamics, 36(13), 1901-1914.

2.krawinkler (1)

Documents