Você está na página 1de 36

This article was downloaded by: [University of Illinois] On: 19 July 2010 Access details: Access Details: [subscription

number 917353191] Publisher Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 3741 Mortimer Street, London W1T 3JH, UK

Journal of Earthquake Engineering

Publication details, including instructions for authors and subscription information: http://www.informaworld.com/smpp/title~content=t741771161

CALIBRATION OF FORCE REDUCTION FACTORS OF RC BUILDINGS

A. M. Mwafya; A. S. Elnashaib a Department of Civil and Environmental Engineering, Imperial College, London SW7 2BU, UK b Civil and Environmental Engineering Department, University of Illinois at Urbana-Champaign, Urbana, IL 61801-2397, USA aelnash@uic.edu

To cite this Article Mwafy, A. M. and Elnashai, A. S.(2002) 'CALIBRATION OF FORCE REDUCTION FACTORS OF RC

BUILDINGS', Journal of Earthquake Engineering, 6: 2, 239 273 To link to this Article: DOI: 10.1080/13632460209350416 URL: http://dx.doi.org/10.1080/13632460209350416

PLEASE SCROLL DOWN FOR ARTICLE


Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf This article may be used for research, teaching and private study purposes. Any substantial or systematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material.

Journal of Earthquake Engineering, Vol. 6, No. 2 (2002)234-273 @ Imperial College Press

CALIBRATION OF FORCE REDUCTION FACTORS OF RC BUILDINGS

A. M. MWAFY Deparhent of Civil and Environmental Enginewing, Imperial College, Imperial College Road, London S W 7 ZBU, UK

Downloaded By: [University of Illinois] At: 19:21 19 July 2010

A. S. ELNASHAI' Civil and Environmental Engineering Department, University of Illinois at Urbana-Champaign, Urbana, IL 61801 -2397, USA aeInashOuic.edu
Received 10 May 2001 Revised 20 June 2001 Accepted 17 July 2001

A comprehensive study is undertaken to assess and calibrate the force reduction factors (R) adopted in modern seismic codes. Refined expressions are employed t o calculate the R factors "supply" for 12 buildings of various characteristics represent a wide range of medium-rise RC buildings. The "supply" values are then compared with the "designn and "demand" recommended in t h e literature. A comprehensive range of response criteria a t the member and storey levels, including shear as a failure criterion, alongside a detailed modelling approach and an extensively verified analytical tool are utilised. A rigorous technique is employed to evaluate R factors, including inelastic pushover and incremental dynamic collapse analyses employing eight natural and artificial records. In the light of the information obtained from more than 1500 inelastic analyses, it is concluded that including shear and vertical motion in assessment and calculations of R factors is necessary. Force reduction factors adopted by the design code (Eurocode 8) are over-conservative and can be safely increased, particularly for regular frame structures designed to lower PGA and higher ductility levels.
Keywords: Force reduction factor; ductility; overstrength; shear assessment; vertical motion; RC buildings.

1. Introduction

The conventional approach of reducing the seismic forces using a single reduction factor to arrive at the design force level is widely utilised in seismic codes. Forcebased design procedures, which include a h a l check on deformations, are likely to remain as the primary seismic design method for some time since other design alternatives are still in the development phase. Hence, the need for reliable calibration of force reduction factors, which have a central role in conventional design methods, is a pressing objective.
'Corresponding author.

240

A. M. Mwafy 1 3A . S. EInashai

Despite the fact that the force reduction factor serves the same function in all seismic codes, it is denoted different terms and assigned different numerical values. The force reduction factor i s expressed in the form of the behaviour factor (q) in Eurocode 8 [EC8, 19941, the response modification factor (R) in the US codes and guidelines [Uniform Building Code "UBCn, 1997; NEHRP Provisions "FEMA 273", 19971, the force modification factor (R) in the National Building Code of Canada [NBCC, 19951, the structure displacement ductility factor ( p ) and structural performance factor (S,) in the New Zealand Loading Standard [NZS, 19921and the ductility factor (l/D,) in the Japanese Building Standard Law W E , 19921. In the present study, the term "force reduction factor (R)" is adopted since it precisely describes the main function of this parameter in reducing elastic seismic forces to the force level used in design. As mentioned above, the numerical values of the force reduction factor are notably varied between seismic codes. For instance, the EC8 behaviour factor ranges between 1.5 and 5.0 for RC hame structures, whilst for the same systems the response modification factor of the US codes may be as high as 8.0. ECSdesigned buildings are therefore assigned higher force levels than those imposed by US codes. This implies that buildings designed for the requirements of the US codes will be more economical or more vulnerable than similar buildings designed according to EC8 [ATGlS, 19951. However, the difference between the reliability of buildings designed to different seismic codes cannot be assessed based on the force levels used in the design only. Taking into consideration different partial safety factors may reduce the difference between the quantities and the parameters adopted in seismic codes and account should be taken of LLcapacityn, not just udemand". The commentary of the NEHRP provisions FEMA 274,19971 confirms that the values of force reduction factor are empirical. Since no supporting investigations are reported in modem seismic codes, it is likely that its values are based on judgment, experience and observed performance of buildings during past earthquakes. Seismic codes rely on ductile response and unquantilied levels of overstrength of structures to justify the reduction in seismic forces via the R factor. Hence, The accurate evaluation of force reduction factors and investigation of the interrelationships between the parameters influencing it are essential elements of seismic design according to codes. Previous studies [Miranda and Bertero, 1994; Vidic et d.,1994; Borzi and Elnashai, 2000] have mainly focused on evaluation of the force reduction factor "demand", particularly the ductility-dependent component of the force reduction factor. A review of the studies carried out prior to 1994 can be found in Miranda and Bertero [1994].Few studies [e.g. Elnashai and Broderick, 1996 for composite and steel frames] were carried out to estimate the force reduction factor "supply" of buildings. However, these studies were either performed on a limited number of buildings designed to old versions of seismic codes using a simplified modelling and procedure, or concerned structures other than RC buildings.

Downloaded By: [University of Illinois] At: 19:21 19 July 2010

Cdabration of Force Reduction Factors of RC Buildings

241

Downloaded By: [University of Illinois] At: 19:21 19 July 2010

In the present study, refined definitions are employed to evaluate the R factors "supply" of medium-rise RC buildings, aiming to assess the accuracy of the values adopted in modern seismic codes. Since calibration of R factors should be based on a representative sample of adequately designed buildings, twelve structures varying in characteristics are employed here. Different design ground accelerations, ductility levels and vertical regularity are taken into consideration to cover a reasonable range of contemporary RC buildings. A rigorous technique is employed to evaluate the R factors. This includes inelastic static pushover and incremental dynamic collapse analyses employing a diverse range of ground motions. The latter approach involves successive scaling and applying of each of the employed records followed by assessment of the response using a comprehensive range of performance criteria. This enables identifying and comparing the ground motion intensities corresponding to different limit states. Special attention is also given to investigating the effect of shear modelling and employing the vertical ground motion in analysis.
2. Definitions and Procedure

It is now accepted that the force reduction factor accounts for the inherent ductility, overstrength and damping of structures. Early definitions of the force reduction factor proposed in the mid-1980s suggested subdividing R into the three components mentioned above. Thus, where R, is the ductility reduction factor, ad is the overstrength factor and RE is the damping factor. The effect of damping is generally included in the ductility reduction factor (R,). The factor considered in Eq. (1) was included only to account for response reduction provided by supplemental viscous damping devices [ATC-19, 19951, hence it could be excluded horn Eq. (1). Another term was introduced by ATC-34 [I9951to account for redundancy (RR). This factor is intended to quantify the improved reliability of seismic framing systems that use multiple line of vertical seismic framing in each principle direction of a building. The R factor is therefore given by:

Moreover, the overstrength and redundancy are considered as one component, as has been adopted by many investigators including some of the ATC researchers [e.g. Freeman, 19901. This is because the overstrength parameter implicitly accounts for redundancy through redistribution of actions; which leads to higher overstrength. The force reduction factor can be therefore defined as the product of the ductility reduction factor (R,) and the overstrength factor (ad),as shown in Fig. 1. Thus,

242

A. M. Mwafy EJ A. S. Elnashai

Downloaded By: [University of Illinois] At: 19:21 19 July 2010

Fig. .I. Relationship between force reduction factor (R), overstrength (ad), ductility reduction and diplacement ductility factor (p). factor (R*)

A proper calibration of the R factor can be undertaken by evaluating the two components contributing to it. These can be obtained from the force-displacement relationship of the structure, which can be determined either experimentally or analytically. Since the experimental evaluation of the R factor for a broad range of structures and a realistic suite of excitations is extremely costly, the only alternative left is inelastic analysis methods. The capacity envelope of a structure can be obtained from inelastic pushover analysis, subject to the constraints of this technique [Mwafy and Elnashai, 20011. For structures that exhibit a period > 0.5 s, the ductility reduction factor (R,) may be taken equal to.the displacement ductility factor ( p ) . The latter approximation follows the equal displacement assumption proposed by Newmark and Hall [1982]. In this assumption, which is applicable to a wide range of structures and adopted in many seismic design codes, the maximum displacements are considered comparable for elastic and elasto-plastic systems. Multiplying the ductility factor ( p ) and the overstrength factor ( R d ) results in the force reduction factor (R). The aforementioned procedure is simple to apply and requires relatively less computational effort compared with other alternatives since only inelastic static pushover analysis is needed to obtain p and pd. However, it has the drawback of ignoring completely the ground motion dependence of the force reduction factor. Despite this clear deficiency in this approach, it has been employed by other investigators [e.g. Balendra et al., 19991.An alternative that has emerged recently is adaptive pushover that accounts for spectral amplification and period elongation

Calibmtion of Force Reduction Factors of RC Buildings

243

[Elnashai, 20001. Whereas preliminary results have shown that this development opens a whole range of possibilities for the inelastic pushover metbod [Mwafy, 20011, it is not fully established yet. The R factor in all seismic codes serves to reduce the elastic base shear (V,) to the design base shear levet (Vd). For instance, EC8 defines the force reduction factor (behaviour factor q) as the ratio of elastic seismic forces with 5% viscous damping to minimum forces used in design. The elastic and the design forces are obtained from the elastic acceleration spectrum of the site (Sa)el and the spectrum used in design (So)'", respectively. Thus,

Downloaded By: [University of Illinois] At: 19:21 19 July 2010

where (S,)eiand (So)'" are the spectral acceleration ordinates corresponding to the predominant period of the structure. Collapse is normally anticipated under the effect of an earthquake having a spectrum higher than the elastic spectrum, particularly at the period considered. Therefore, the following definition may be employed to evaluate an ultimate value of the force reduction factor for a particular structure under a specific accelerogram:
K,dy

= (~o):'/(sa)j"

(5)

where the subscripts L ' ~ " and "dy" refer to collapse and design yield (the yield level assumed in design), respectively. Moreover, the structure is mainly designed for forces consistent with its yield limit state. Consequently, Elnashai and Broderick [I9961 have employed a definition that utilises the spectral acceleration causing actual yield in the denominator, as given in the following equation:

where the subscript 'Lay" refers to actual yield. By assuming that the response spectra of the design, yield and collapse earthquake have constant dynamic amplification (ratio of peak ground-teresponse acceleration), at least for the period range considered, Eqs. (5) and (6) can be rewritten as follows:

where ag(collapse) ,

(design)

and as (actual yield) are the peak ground accelerations

of the collapse, design and yield earthquake, respectively. ag(desisn yield) is the design PGA divided by the force reduction factor employed in design (&ode). The , ( i s that the former is the yield difference between a, (design yield) and a, d intensity assumed in the design, whilst the latter is the PGA a t first indication of
actual yield. Both definitions given in Eqs. (7) and (8) relate the intensity of loading at collapse to the elastic seismic forces. Equation (7) adopts the assumption that yielding h i s definition is will occur at the design ground acceieration divided by &ode. T straightforwardand less computationally demanding because only the PGA of the

Downloaded By: [University of Illinois] At: 19:21 19 July 2010

earthquake that causes collapse is required. It is more appropriate for assessing existing force reduction factors since it checks the validity of the design by examining the capability of-the structure to resist greater seismic forces than those implied by the design. However, the definition of has the shortcoming of not accounting for the dissimilarity between the design spectrum and the spectral acceleration of the ground motion at yield plnashai and Broderick, 19961. Even though a set of artificially generated ground motions compatible with the elastic response spectrum the disparity still exists, as shown in of the code are utilised to calculate Fig. 2(a). In this figure, the spectrum of an artificial record is scded to the design PGA divided by (the yield assumed in the design) and compared with the design spectrum. It is clear that synthetic records are not compatible with the design spectrum in the short period range and in the period range beyond the corner period (T,). This is because the code decreases'the force reduction factors in the short period range as a result of the limited capability of structures in this range to develop inelastic deformation. EC8 is a h conservative for long period structures; hence Kode values are decreased in this range. This leads to a decrease i n the steepness of the inelastic spectrum beyond the corner period. On the other hand, structures designed to modern seismic codes usually exhibit a considerable level of overstrength [Elnashai and Mwafy, 20011. This leads to a signhcant dierence between the PGA causing first global yield (a, (,t,,~ yield)) = design PGA/%,de). and the yield intensity implied by the design (a, (d,,i, It is observed in many cases investigated in the present study that the acceleration spectrum of the record causing yield ( s , ) : is even higher than both the design (S,)'" and the elastic spectra of the code (&)"I, as shown i n Fig. Z(b). Clearly, the reserve strength results in delaying the yield to-this high level of ground motion. Since the PGA corresponding to yield is more sensitive to the level of overstrength compared with the PGA that causes collapse, it is expected that the definition of R , , will underestimate the force reduction factor, particularly for , which buildings exhibiting high overstrength. It is clear that the definition of &

Fig. 2. Evaluation of the force reduction factor for a 12-storey regular frame building using an artificial record compatible with the code spectrum: (a) definition of R < , J ~ (b) ; definition of &.ay.

Calibmtion of Force Reduction Factors of RC Buildings

245

Downloaded By: [University of Illinois] At: 19:21 19 July 2010

Fig. 3.

Comparison between the ductility reduction factor ( R p )and the definition of ( & ,) .

represents the inherent force reduction factor of a structure of no known design history, is more suitable when recommending R factors for ideal systems. For practically designed and detailed buildings, this definition should be rnodfied to account for overstrength. It is also important to note that there is a clear similarity between , and the ductility-dependent component of the force reduction the dehition of & factor (Rp = V,/V,), as shown from Fig. 3. This emphasises the need to modify Eq. (8) by adding the overstrength factor (ad = actual-to-design strength) to h&. The suggested modification is given by:

The latter modification allows for reserving characteristics of the original definition of a,, in terms of the ground motion dependence of a , ~,o~~a,,,~ and a, (actual yield). This gives some advantages RLay for over &,dy, which ignores this dependence in its denominator. The main shortcomings of the expression of Eq. (9) are its demanding computations and being based on the assumption of constant dynamic amplification. However, it is an effectiveway of evaluating the force reduction factor of a particular structure subjected to a specific earthquake. It is also noteworthy that in another study, Salvitti and Elnashai [I9961 have concluded that the force reduction factors calculated using Eq. (8) are not significantly influenced by the elongation in the period and the dynamic amplification effect. Other methods and alternative definitions may be suggested to predict the force reduction factors. However, it is beyond the scope of the current study to review and discuss all these approaches. Only the two definitions given in Eqs. (7) and (9) are adopted here for evaluation this important factor using inelastic pushover and incremental collapse analysis. Pushover analysis is employed to evaluate the global yield limit state, structural capacity and overstrength. The extensive dynamic collapse analysis is performed by progressively scaling and applying each of the employed set of accelerogram, starting from a relative low intensity, typically the

Perform ioelasric p w b v c r analysis i c c e (righrward and leftward) up w a high drift Limit to achicvc yield in all critical

Dynamic +sir
Downloaded By: [University of Illinois] At: 19:21 19 July 2010

ourpvt

Run thc staric pushover

global pafonnancc. formatim


lmal a d global rwof= lmdcr Ihc

Fig. 4.

Performed analysis procedure for each building-input ground motion combination.

design intensity divided by the Rcode(ag(design yield)), and terminatkg with the intensity at which all yield and collapse definitions are achieved. This rigorous technique allows evaluating the performance of the structure at different levels of excitations. Hence, the peak ground accelerations causing yield and collapse can be identified according to the performance criteria adopted here. Between 15 to 20 analyses are performed for each structure- input combination to identify the response at different limit states. The procedure is surnmarised in Fig. 4.

3 . Ingredients of the Analytical Study


3.1. Buildings

Twelve structures varying in characteristics are employed in the current study as representatives of medium-rise RC buildings. The structures are split into the three groups shown in Table 1. Four buildings designed to different ductility and PGA combinations are employed in each group. This is aimed to compare between

Calibmtion of Force Reduction Factors of RC Buildings

247

Table 1. Characteristics of the buildings considered. Group Structure reference No. of storeys and structural system Design ductility High &storey irregular frame Mechm Medium Low High 12-storey regular frame bkdium Medium Low High &storey regular fram+wall Mdium Medium Low Design PGA (g) Design force red. factor Elastic period (sec)

IF-HO3O IF-MOJO IF-M015 IF-LO15 RF-H030 RF-M030 *-MOl5 RF-LO15 FW-HO30 FW-MOB0 FW-M015 FW-LO15

Downloaded By: [University of Illinois] At: 19:21 19 July 2010

structures designed according to capacity design set of rules but for different ground accelerations and for the same ground acceleration but different capacity design rules, The buildings were designed and detailed in accordance with EC8 [1994],as a typical modern seismic design code applicable to more than one country with various levels of seismicity, soil conditions and construction practice. TabIe 1 shows the elastic force reduction factors used in the design and the observed elastic fundamental period ''Tel,ti," obtained from elastic free vibration analyses. Live loads and loading from floor finishes and partitions are 2.0 kN/m2. A l l buildings are assumed to be founded on medium soil type "B" of EC8 (firm). A characteristic cylinder strength of 25 N/mrn2 for concrete and yield strength of 500 N/mm2 for steel are utilised. More information regarding member cross- section sizes and reinforcements are given elsewhere [Fardis, 19941. The geometric characteristics of the buildings are illustrated in Fig. 5.
3.2. Analytical modelling

The structural analysis program ADAPTIC [Izzuddin and Elnashai,1989)is utilised to perform the inelastic analyses. The program has been developed f o r the inelastic analysis of two- and three-dimensional structures under static and dynamic loading, taking into account the effects of both geometric nonlinearities and material inelasticity. ADAPTIC has the feature of representing the spread of inelasticity w i t h the member cross-section and along the member length through utilising the fibre approach. Further information regarding the program and its validation can be found elsewhere [e.g. Elnashai and Elghazouli, 1993; Elnashai and Izzuddin, 1993; Martinez-Rueda and Elnashai, 19971.

Downloaded By: [University of Illinois] At: 19:21 19 July 2010

g. 5. Plan and sectional elevation of the buildings: (a) group I "eight-storey irregular frame buildingsn; (b) group 2 "12-storey regular frame buildingsn; (c) group 3 "eight-storey regular frame-wall buildingsn.

Refined analytical models are utilised for the 12 buildings investigated here. Detailed description of this approach is given elsewhere [Mwafy, 2001; Elnashai and Mwafy, 2001). Only the main features of the adopted modelling technique are summarised below: Two-dimensional representation is selected in the light of the symmetry of the buildings and the limited si@cance of torsional effects. Internal and external lateral force resisting systems are combined by means of an overlay approach, as shown in Fig. 6 for the frame-wall structural system. The analyses are conducted along one horizontal direction onIy (global X-axis for group 1 and 2, and Z-axis for group 3). This is justified by the fact that critical response criteria were expected to occur earlier in those directions. Horizontal and vertical structural members are modelled using three cubic elastoplastic elements, where the lengths of the elements are determined according to the distribution of longitudinal and transverse reinforcements. Shear spring elements are introduced to represent the shear stihess of the beam-column connection, as shown in Fig. 6(d). In the frame-wall structures, the core wall on each side of the coupling beams is modelled as a "wide-column". The elements are located at the centroid of the core U-shaped cross-section and connected with the end of beams at each storey level using rigid arms, as shown in Fig. 6(c).

Caiibmtion oJ Force Reduction Factors of RC Buddings


Rigid

249

/- elements

Downloaded By: [University of Illinois] At: 19:21 19 July 2010

(a) Internal s t r u c h l ~ svstem

0 )Plan of the frame-wall


s m c d system

Ic) 2-D modellinp of central core

"

I_

d p = le) Overfy technicwe considered


Fig. 6.

.-----_--------------1L-

Cbl'

fike

JO Decomposition of beam T-section into fibres

Modelling of frame-wall structures.

The concrete is represented using a uniaxial constant confinement concrete model [Martinez-Rueda and Elnashai, 1997). The advanced multisurface plasticity model [Elnashai and Izzuddin, 19931 is utilised for modelling the reinforcement bars.

3.3. Input ground motions

Inelastic dynamic analysis is performed using eight input excitations. Four 10second duration artificially-generated records compatible with the EC8 response spectrum for medium soil class (firm) were selected for comparison and calibration with the design. Furthermore, two natural earthquakes were selected in terms of

250

A. M. Mwafy B A. S. Elnashai
Table 2. Ground motions used in time-history analysis.

Earthquake and station


Kobe (Japan),
Kobe University

Date

Ms

PGA ( ' I Horiz. Vert.


0.276 0.319 0.431 0.349

V/N

NO. of input

excitations

17/01/95 18/10/89

7.20 7.17

1.56
1.09

Lorna Prieta (USA),


Saratoga "Aloha Ave."

2
4

Artificial Records,
Art-recl to Art-rec4

the site-to-source distance and the V / H ratio and applied with and without the vertical component of ground motion (four input combinations). This is motivated by the desire to investigate the effect of the vertical ground motion on buildings situated in the vicinity of active faults, which may be significant as concluded in previous studies [e.g. Papazoglou and Elnashai, 19961. For the sake of brevity, only brief results are presented herein illustrating the significance of this effect on the R factors. Complete results of this investigation are given elsewhere [blwafy, 20011. Characteristics of the selected records are given in Table 2. Employing a reliable method to scale the selected records is significant since the assessment methodology requires successive scaling of the records. A refined normalisation approach i s adopted in here, where all records are scaled to possess equal velocity spectrum intensity in the period range considered, whilst the design code elastic spectrum is taken as a reference. For instance, when the input accelerograms are scaled to a PGA of 0.309,the scaliig factors ensure that the velocity spectral areas of the scaled records equal to the area under the velocity spectrum of the code anchored to a PGA of 0.30g.Therefore, the term "intensity" or "PGA" quoted thereafter is not of the original or scaled records but rather multiples of the design ground acceleration. Advantages of this approach were discussed by Mwafy [2001].
'

Downloaded By: [University of Illinois] At: 19:21 19 July 2010

3.4. Performance parameters

Based on the approach adopted in the present study to evaluate the R factors, a number of response criteria are needed to d e b the yield and collapse limit states. Two yield and six failure criteria are utilised here. These are classified into two groups, iocal and global criteria. Local yield is defined when the strain in the main tensile reinforcement exceeds the yield strain of steel. The global yield limit state is defined as the yield displacement of the elasto-plastic idealisation of the real system. On the other hand, two failure criteria on the member level are utilised: exceeding the ultimate curvature (4, = (E,, + ~ , ) / d ) or shear strength. The latter is evaluated for structural members using a realistic ductility- and axial force-sensitive

Calibmtion of Force Redtictima Fuctws of RC Buildings 251

shear strength approach capable of providing an experimentally verifiable estimate of shear supply in RC members [Priestley et al., 1994). To allow for effective cornparison with the design, the shear strength model of the design code is also employed after eliminating safety factors. The adopted globaj failure criteria are: an upper limit of the interstorey drift (ID) ratio equal to 3%, formation of a column hinging mechanism, a drop in the overall lateral resistance by more than 10% and an upper limit of the stability index (0 = ID x storey gravity load/storey shear) equal to 0.3. The adopted performance criteria are implemented in a post-processing program to directly monitor capacities and demands of shear and curvature and apply the performance parameters during the time history analysis. Further information regarding t.he ingredients of the rigorous analytical study can be found elsewhere [Mwafy, 2001; Mwafy and Elnashai, 2001; Elnashai and Mwafy,20011.
Downloaded By: [University of Illinois] At: 19:21 19 July 2010

4. Analytical Results

Inelastic static pushover and incremental dynamic analyses up to collapse are carried out on the 12 buildings investigated here. In total, over 1500 inelastic dynamic analyses were performed to calculate the force reduction factors. However, for the sake of brevity only a summary of these extensive results is presented in subsequent discussion in the form of average results for the artificial and natural records (with and without the vertical ground motion).

4 . 1 . Results at yield

A summary of the average intensities at which local and global yield are observed from dynamic analysis is shown in Table 3. First indication of member yielding is observed at different locations for the three groups of buildings. In the first group, the severance at the ground storey of some intermediate columns causes early yielding in those columns. For the dual structures, first yield is observed in beams followed by a plastic hinge at the base of the core. Unlike the latter two groups, first yield in vertical members of the 12-storey buildings is observed relatively late compared with the first beam yielding. In many cases, it was necessary to apply higher PGA than that causing global yield to achieve yielding in column. This is due to the regularity of this structural system and the rigorous capacity design provisions imposed by the design code. For two buildings designed to the same PGA, yield occurs at a lower level of excitation for the structure designed to higher R factor (higher ductility level). This is confirmed in all cases when applying the gIobal yield criterion. When the local yield criterion is employed, the aforementioned observation is also applied on the first and the third group of buildings where equal cross-section dimensions were used for the' pairs of buildings designed to the same PGA. For the second group, the

Downloaded By: [University of Illinois] At: 19:21 19 July 2010

Table 3. Summary of ground accelerations a t yield limit state.


ag(actual yield) ag(actual y ~ e l d ) ~ ag(actual yield)'

nat. records

nat. records

Overstrength
r o d )

Reference

art. records

(w
(H
"Wd
Id

+ V)
G '

ag(actual yield)'/ ag(deaign yield)

L
G

L
G
G

Stat.

Dyn.

'Average for the artificial records. b~verage for natural remrds (horizontal earthquake component,is solely utilised). =Average for natural records (horizontal and vertical components are applied). d ag(dcsign yield) = ag(dasign)/Rcode *Average for eight ground motions. L & G: Local and global yield intensities, respectively. Stat. & Dyn.: Overstrength evaluated from static pushover and incremental dynamic collapse analysis, respectively.

Calibration of Force Reduction Factors of RC Buildings

253

Downloaded By: [University of Illinois] At: 19:21 19 July 2010

depth of RF-HO3O beams is 65 cm compared with 60 cm for RF-MO30 beams. The increase in the depth of RF-H030 beams, which was made to fulfil local ductility requirements of the design code, causes delaying the first member yielding. and the average observed I* The values of as(desisn yield) (design PGA/KOde) car and global yield intensities for the eight ground motions divided by ag(design yield)/aS(design yield)) i s the yield) are also shown in Table 3. The latter. ratio between the average PGA that causes actual yieldand the intensity at which yield is implied in the design. In a l l cases, this ratio exceeds 2.0, reflecting the high overstrength exhibited by the buildings. It is clear that the first member yielding is observed at high intensity leveIs compared with ag(design It k also noteworthy that the overall structural response generally remains elastic even slightly beyond the &st indication of member yielding due to the insigruficance of one plastic hinge, as shown f r o m Figs. 7. The latter figure depicts the sequence of hinge formation of the irregular frame group of buildings obtained from pushover analysis at the global yield limit state. The above supports the need for the modification of adding the overstrength factor (ad) to the definition of force reduction factor suggested in Eq. (9). It also reinforces the conclusion of Elnashai and Mwafy (20011

Fig. 7. Progress in plastic hinge formation at global yield limit state for the first group of buildings: (a) IF-H030 (top dkp. = 277 mm, top drift ratio = 1.09%); (b) IF-MO3O (top disp. = 289 mm, top drift ratio = 1.13%);( c ) IF-MOl5 (top disp. = 215 mm, top drift ratio = 0.84%);(d) IF-LO15 (top disp. = 263 mm, top drift ratio = 1.03%).

2%

A. M. Mwafy # A . S. Ellurshai

Downloaded By: [University of Illinois] At: 19:21 19 July 2010

regarding the high overstrength of buildings designed to modern seismic codes and the conservatism of the minimum nd factor of 2.0 suggested for low and mid-rise buildings. The overstrength factors ( Q d ) evaluated by Elnashai and hlwafy [ZOO11 for the 12 buildings using inelastic pushover and incremental dynamic collapse analyses are presented in Table 3. It is clear from the comparison between average ag(actuai yield)-t~-ag(design yield) ratios and overstrength factors that employing Rd in the definition of the force reduction factor proposed in Eq. (9) is conservative (ag(actual yield)/~g(designyield) > ad). However, in the third group of buildings the ag(actua~ yield)-to-ag(design yield) ratios are lower than the R d factors evaluated from h i s is mainly due to the sensitivity of wall structures dynamic collapse analyses. T to higher mode effects, which significantly amplify the base shear during timehistory analysis. This represents one of the differences between static pushover and dynamic analysis results, as concluded by Mwafy and Elnashai (20011. Hence, higher overstrength is evaluated from dynamic collapse analysis compared with static pushover analysis. To be on the conservative side in evaluation of the R factors, the fld factors calculated from static pushover analyses are utilised with the yield intensities determined from the local criterion, whilst the R d factors evaluated from dynamic collapse analyses are employed with the intensities obtained using the global criterion. The distribution of plastic hinges and sequence of hinge formation at the global yield shown in Fig. 7 shed light on the suitability of the local yield criterion for evaluation of the R factors, which are mainly overall parameters. Several plastic hinges are formed at the global yield limit state, which clearly represents the actual yield of the structure. The aforementioned observation may not be applicable to the local collapse criteria since member failure may practically cause partial structural collapse. It was therefore decided to categorise the force reduction factors calculated in the present study according to the following classification: Factors evaluated from local response criteria (yield and collapse). Factors evaluated based on the global performance criteria. Factors calculated from first indication of yield and collapse (local or global).
4.2. Results at collapse

The extensive range of collapse parameters explained earlier (two local and four global criteria) is employed to assess the ground accelerations that cause member or structure failure. It should be mentioned that some of these criteria, particularly many of those defining the global collapse limit state, have not been observed up to a high level of PGA. No formation of a column sidesway mechanism has been observed up to a high drift limit, which reflects the success of capacity design provisions in protecting vertical members. This is clear in Fig. 8 from the distribution of plastic hinges at the interstorey drift collapse for the irregular buildings, which has the weakest lateral force resisting system.

Calibration of Force Reduction Factors of RC Buildings

253

Downloaded By: [University of Illinois] At: 19:21 19 July 2010

Fig. 8. Distribution of plastic hinges a t the interstorey drift collapse limit state (ID = 3%) for the first group of buildings: (a) IF-H030 (top disp. = 532 mm); (b) IF-M030 (top disp. = 552 mm); (c) IF-MO15 (top disp. = 476 mm); (d) IF-LO15 (top disp. = 516 mm).

1
Storey

-111

Art-rccl

[2] Kobe (H)

131 Lorna Rieta ( H ) 1

Fig. 9. Maximum interstorey drift (ID) and stability coefficient (8) for a sample building from each group at ID collapse (results of three records from eight ground motions employed).

The capacity envelopes of the 12 buildings obtained from static as well as from incremental dynamic analysis are given elsewhere [Mwafy and Elnashai, 2001). It was observed that no significant reduction in the lateral capacity of the 12 buildings was recorded. Moreover, the observed values of the stability coefficient (9)

Downloaded By: [University of Illinois] At: 19:21 19 July 2010

Table 4.
b

Ground accelerations at collapse limit state.


*g(co~~apse)~

AverageJ
ag(collapse)/ag(doolgn)

Reference LIsh] IF-H030 IF-M030 IF-M015 IF-LOIS RF-HO3O RF-MOJO RF-M015 RF-LO15 Llcr]

"g(co~~apse) art. records nat. records I,[sh] L[cr]

ag(collapae) nat. records (H)

(H + V)
L[sh]

G
Llsh] L[cr]

G
L[cr]

FW-HD3O FW-MO30 FW-MO15 FW-LO15


-

aAverage for the artificial records. b~verage for the natural records (horizontal earthquake component is solely utilised). CAverage lor the natural records (horizontal and vertical components are applied). d~verage for eight ground motions. L[sh]: Local shear failure. L[cr]: Local crushing failure in concrete (curvature ductility demand > supply). G: global failure (interstorey drift > 3%). I-]: No ahear failure occurs up to observing two types of failure; local and global collapse. Bolds denote the PGA corresponding t o first observed local failure. Italics indicate the PGA a t subsequently observed local failure.

Calibration of Force Reductiun Factors of

R C Buildings

257

up to the interstorey drift collapse were well below the limiting value adopted here (0.3). Sample results of the calculated values of B at the ZD collapse are presented hereafter (Fig. 9). The foregoing observations are noted in all structures considered and under all ground motions employed. Therefore, the interstorey drift (ID) criterion is practically the controlling global collapse parameter that was utilised to evaluate the R factors. The two member failure criteria (ultimate curvature ductility and shear capacity) are also employed here since they are observed within a practical level of excitation, particularly the ultimate curvature ductility. As explained above, the shear suppIy-demand ratio is monitored during dynamic analysis by employing two shear strength mode!s, Priestley et aL (19941 and the design code IEC2, 19941. All safety factors were elimhated from the code shear model to be consistent with the assessment objective of this investigation. However, shear failure under relatively low levels of PGA is frequently recorded when employing this model, confirming the over-conservatism of shear strength predicted using code approaches. Hence, it was decided to exclude this model from calculations of the force reduction factors. The collapse intensities from the shear criterion shown in Table 4 are therefore those obtained from applying Priestley et al. (1994) shear strength model only. The ground accelerations that cause first collapse, as shown in Table 4, confirm the importance of including shear as a failure criterion in assessment studies. In frame structures, it is observed that the most susceptible buildings to shear failure are those designed to ductility level "Low". For those buildings, shear failure is the controlling criterion that defines local collapse intensities. This is because capacity design rules are not required by ECS when designing for this level of ductility. Only some provisions are applied to enhance the ductility of columns. EC8 does not also impose supplementary provisions to those specified in EC2 for shear design or for maximum stirrups spacing in critical regions of beams designed to ductility level L L L ~Therefore, ~". these beams are vulnerable to shear failure. A sample of shear assessment results are presented hereafter, whilst complete results of this comprehensive investigation are presented in Mwafy [2001]. For buildings designed to higher levels of ductility, member shear failure is observed at high PGA levels. Hence, ground motions that cause shear failure are only presented in Table 4 when this type of failure is observed at or before collapse from both the ultimate curvature and the ID criteria. Indeed, the structure at this limit should experience extensive damage since it would undergo two types of collapse (local and global), hence shear failure will not be a controlling parameter. Three different values of collapse ground acceleration (a,~,,ll,,,,~) are therefore shown in Table 4 for each building-ground motion combination (based on two local and one global criteria). The first observed local failure is only utilised in calculations of R factors, in addition to the intensity at global collapse. The intensity levels at first local failure (ultimate curvature or shear) are shown in bold in Table 4.

Downloaded By: [University of Illinois] At: 19:21 19 July 2010

258

A. M. Mwafy # A.

S. Elnashai

Downloaded By: [University of Illinois] At: 19:21 19 July 2010

It is clear for moment resisting frames that critical concrete strains are exceeded slightly before the global ID collapse limit. For the hybrid structures, the interstorey drift collapse is attained well after the critical concrete strain, confirming the efficiency of this structural system in controlling deformations. As shown in Tabie 4 for the third group, ID collapse is observed for the 0.309 pair a t about four times the design PGA, and at about six times the design PGA for the 0.159 pair. It is clear that local failure criteria are more critical for this type of construction than global parameters due to the concentration of demands at wall base and in coupling beams. ratio for the eight input ground motions are The average ag~,,ll,,,,~-to-ag~desi~~ also shown in Table 4. This ratio reflects the average margin of safety exhibited by each building under the effect of the eight ground motions. It is clear that the margin of safety increases with the decrease in the design PGA, reflecting the higher contribution of gravity Ioads. The balance between gravity and seismic design scenarios is the main parameter controlling this margin. It is also observed with few exceptions, mainly in the irregular frame buildings, that the safety margin increases with decreasing the ductility level. This implies that the stringent capacity design rules imposed on structures designed to higher ductility levels do not balance adopting higher force reduction factors. This also suggests that buildings designed for lower PGA and detailed for lower ductility levels are more reliable compared with those with higher ductility and designed for comparatively higher PGA.
4.3. Sample time-history collapse analysis results

To allow comparisons between the seismic response of the investigated building design on the member and structure leveIs, sample dynamic analysis results are presented in Figs. 9 to 11. The results are shown for a sample buildings and accelerograms from those employed in the present study. Further results are given elsewhere [Mwafy, 20011.
4.3.1. Global collapse

Figure 9 shows the distribution of the maximum observed hterstorey drift at the global collapse limit state for a sample building from each group when subjected to one artificial record and two natural ground motions employing the longitudinal component. The recorded m w u m stability coefficients (8) at the same limit state are also shown. The observed values of 0, which place a further limitat ion on P-A effects, are well below the limiting value adopted here (0 < 0.3). T h i s implies that second order effects are not significant up to this high level of PGA. The levels of excitation that cause ID collapse are varied a s shown horn Tables 4. It is clear in Fig. 9 that each building-input combination is a distinct case, where the ID limit state is observed at different storey levels. For the irregular frame buildings, the maximum ID is likely to occur in the ground storey due to its increased height compared with other storeys and .the severance of four columns.

CoIibmtion of Force Reduction Factors of RC Buildings

259

For the second group, ID collapse occurs in the intermediate storeys, highlighting the contribution of higher modes. It is also observed that the distribution of the maximum ID along the height i s more uniform for hybrid structures compared with frame buildings. T h i s emphasises the efficiency of dual structural systems in reducing lateral deformations.
4.3.2. Local collapse - maximum curvature ductility

Downloaded By: [University of Illinois] At: 19:21 19 July 2010

Figure 10 illustrates the distribution of the maximum curvature ductility demand (CDD) in horizontal and vertical structural members for a sample building horn each of the three groups. This is presented at an excitation level equal to twice the design PGA to simplify the comparison between structural systems and input ground motions. Due to the rigorous capacity design provisions imposed by EC8, critical concrete strains are mainly attained in beams under the eight input ground motions. For vertical members, maximum CDDs are observed at the ground storey with the exception of the cu$-off columns in the irregular buildings, where

Art-recl

-Kobe

(H)

-Lorna

Pricta (H)

'-

Storcy

1 .
0 1 2 S 4 S 6 I Z 3 ~

piTl
(a) External

system

.,
6 1 2 1 1 6

Fig. 10. Maximum curvature ductility demand (CDD) for a sample building from each group at twice the design PGA (0.609)(results of three records from eight ground motions employed).

260

A. M. Mwafy El A. S. E l w h a i

Downloaded By: [University of Illinois] At: 19:21 19 July 2010

high demands are observed at the second and the eighth storeys [Fig. 10(b)J. The advantage of regular structural systems is clear from the distribution of CDDs in RF-H030 in comparison with IF-H030, where energy is mainly dissipated in beams of the regular structure. The comparison also point towards the need for possible improvements in EC8 to protect cut-off columns in irregular structures. In the dual structural system, high demand is generated in the central core (walls and coupling beams), contrary to the frame buildings where demands are distributed between external and internal lateral force resisting systems. This confirms that coupling beams and walls provide the primary source of energy dissipation in hamewall .structures and support the special provisions imposed by the code on the design of these members. The sample buildings presented in Fig. 10 are those designed to high ductllity level in each group. For this level of ductility, the minimum conventional curvature
(b) RF-LOIS

-F-

SF.s u p p l y :

-(EC2) -(Riutley el al.)


-

SF.s u p p l y

m-

-(EO) -(Priculyn aL)


i !

i i lime (ref)

is

Fig. 11. Results at first observed member shear failure in frame buildings designed to ductility level "Low" for a sample records from the eight ground motions utilised here: (a) IF-L015; (b) RF-LOIS.

Calibmtion of Force Reduction Factors of RC Buildings

261

ductility factors (CCDF)required by EC8 for columns and walls are 13 and 9.8, respectively. It is clear that the observed CDD values at twice the design intensity are well below the minimum supply imposed by the design code. The satisfactory safety margin, expressed by the difference between the CCDF and the CDD values, at this high level of excitation highlights the robustness of vertical members designed to EC8.
4.3.3. Lour1 collapse - ultimate shear strength

Downloaded By: [University of Illinois] At: 19:21 19 July 2010

Results of the shear assessment investigation clearly confirm the si&cance of employing this rigorous collapse parameter hassessment studies. It is concluded that the shear failure criterion is the controlling parameter that define member failure for frame buildings designed to ductility level "Low". Hence, ignoring this criterion in calculations of the force reduction factors will result in inaccurate prediction of these significant factors. Sample results of this investigation are presented in Fig. 11, where the shear supply-demand time-histories for two records are presented at the intensity levels that cause first indication of member shear failure.
5 . Evaluation of Force Reduction Factors

It is shown in Sec. 2 that the force reduction factor "supplyJ' of a specific structure subjected to a specific earthquake record can be evaluated according to Eqs. (7) and (9), which are used to calculate and RiSay , respectively. Tables 5 and 6 show a summary of the average force reduction factors evaluated using the set of artificial and natural records employed here. Design R factors (&ode) and average supply-todesign ratios are also presented. For each structure investigated here, three values of the R factor are presented. These correspond to the R factors calculated using: the local yield and collapse intensities, global intensities and first observed yield and collapse PGA regardless of the classification of the limit state (being local or global). Generally, the average supply-t~designvalues obtained when employing the definition of &,dy are higher than those of RL,dy. The definition of &,d, assumes that yield will occur at as(design yield) (Design PGA/&,d,), which implicitly accounts for overstrength but ignores the ground motion dependence of the yield intensity. The definition of Rilay takes this into consideration by employing the actual yield intensity corrected by the overstrength factor (Rd). The ratio of ag(actud yield)/ag(design yield), which represents the overstrength assumed in the defs generally higher than the actual overstrength Rd, as shown in inition of &,dy, i Table 3. Therefore, average values of are higher than &,,. The evaluated R:,ay and factors show clearly the dependence of this factor on the ground acceleration used in the analysis. This is exemplified by the Ri,, values of the second group of buildings, as shown from Table 5. Differences in excess of 100% afe observed between the factors calculated using the artificial

Downloaded By: [University of Illinois] At: 19:21 19 July 2010

Table 5.
RL,ay %,ay
-

Average force reduction factors (RL,ay) for all records and performance criteria.
K,ay

Ref.
Average for four art. records

Average for two nat. records (H) Average for two nat. records (H V)

FW-H030 FW-MOB0

7.22 4.59

1.76 1.72
2.59

1.76

FW-MO15 FW-LO15
9.64 9.59

20.66 16.04 13.21 10.32

7.22 4.59 9.64 9.59

5.47 4.46 5.39 4.59

8.69 6.81 8.43 7.59

5.47 4.46 5.39 4.59

5.83 4.54 4.95 4.33

8.50 6.68 9.12 7.76

5.83 4.54 4.95 4.33

3.50 2.63 2.63 1.75

3.52

3.61 9.74 3.90 4.89

1.72 2.59 3.62

Rtc.ay = ( a g ( c o l l s P a s ) / ~ g ( a c t u a l yield))% G: Global criteria are used

L:Local criteria are employed to calculate R i , a y


F: First observed yield and collapse are employed to calculate
IZL,,,

Downloaded By: [University of Illinois] At: 19:21 19 July 2010

for all records and performance criteria. Table 6. Average force reduction factors (RLSdy)
%,dY %,dy ';,dy R&dy Rcode

Ref.

(average)/
&ode

Average for four art. records

Average for two nat. records (H)

Average for two nat. records (W + V)

L
G

F
G
G

F
L F

&,dy

yield)

= ug(collapre)/ag(design G: Global criteria are used

L: Local criteria are employed to calculate RCsdy F: First observed collapse is employed to calculate

&,dy

264

A. M. Mwafy E4 A. S. EInahni

Downloaded By: [University of Illinois] At: 19:21 19 July 2010

accelerograrns and those obtained when employing the natural earthquake records. The comparison between the two sets of accelerograms does not show any trend. ,This emphasises the need to employ a diverse range of natural and code spectrumcompatible records if reliable estimation of force reduction factors is required. It is clear in Tables 3 and 4 that yield and collapse may occur under lower PGA when the structure is subjected to horizontal and vertical ground motion. This may lead to reduce the R factors "supply", leading to the adoption of higher seismic forces in design. The R factors "supply" for the 12 buildings investigated here are evaluated using the artificial and the natural records (with and without the vertical vibration), as shown in Tables 5 and 6; It is observed that both definitions and are influenced by the inclusion of vertical earthquake of R factors (q,, load. Lncluding this effect in analysis does not give a clear trend in increasing or decreasing the R factors "supply". However, the mean values of ET,,, and &,dy' are reduced by up to 18% and 15%, respectively. This confirms the significance of including vertical vibration in calculations of R factors. Force reduction factors evaluated using local yield and collapse criteria are controlled by the response of specific structural members, whilst those calculated using global criteria reflect better the overall characteristics of the structures. The factors obtained from the first indication of yield and collapse may be more worthy of consideration since they account for the h s t observed limit state, which may be more sigruficant than subsequently observed yield or collapse. For instance, early failure in a vertical structural member, which is not accounted for when utilising gbbal failure criteria, is more significant compared with ID collapse occurs afterwards. Therefore, the force reduction factors "supplyn evaluated using the first indication of yield and collapse are employed below to compare with the "design" and the "demand" values.
5.1. Comparison with the design force reduction factor

The calculated average R factors "supply" are compared in Fig. 12 with the design force reduction factor spectra. The latter are obtained by dividing the elastic by the inelastic spectra of each of the 12 buildings, resulting in nine design spectra. It is noteworthy that the two buildings designed to ductility level "Medium" in each group share the same R-spectrum, hence nine design spectra are produced Erom the 12 buildings investigated here. The results are classified'in Fig. 1 2 according to the ductility level (one plot for each of the three ductility 1eveI.s employed in the design), hence three force reduction factor spectra corresponding to the three groups of buildings are shown in each of the three plots. The R;,, and &,+.factors are plotted at the inelastic period of the buildings evaluated a t the design intensity levels. These periods are evaluated by Mwafy and Elnashai [2001] uslug Fourier analyses of the roof response time-histories obtained from dynamic analysis (average of eight ground motions). Moreover, to highlight the increase of the supply-to-design ratio due to period elongation, the calculated

Calabmtion of F m e Reduction Factom of RC Buildings

265

-.-.-,R

-----,R

(IF-building) (RF-building)

13-"-.I KGq (at T a


K G . , (at T &

0
0

RC.& (at T-) R G , , (at T

(a) Ductility level 'High'buildings

-f

Downloaded By: [University of Illinois] At: 19:21 19 July 2010

(b) Ductility level 'Medium' buildings


16

=OIR

(c) Ductility level U w 'buildings


I0l

(
RF-LO15
IF-MIS

Fig. 12. Force reduction factors "supply" versus the "designn: (a) ductility level "Highn buildings (R factors are plotted at the elastic and at the inelastic periods]; (b) ductility "Medium";
(c)

ductility "Low".

266

A.

M. Mwafy

4 A.

S. Elnoshi

Downloaded By: [University of Illinois] At: 19:21 19 July 2010

R factors for the high ductility level buildings ~ i 12(a)] ~ are . plotted a t the elastic and inelastic period levels. It is clear that employing the inelastic periods in the comparison leads to increasing the supply-tedesign force reduction factor ratios shown in Tables 5 and 6. High force reduction factors "supply" are observed for the sample buddings employed here in comparison with the values adopted by the design code. This implies that the design R factors can be increased without adverse effects on structural safety. The satisfactory performance of the buildings at the design and twice the design intensities a well as the observed high ground accelerations that cause first indication of collapse support this conclusion. It is aIso clear that buildings designed to lower PGA have higher force reduction factors compared with their higher design ground acceleration counterparts, as shown in Fig. 12(b). This confirms the need used in seismic for including the effect of the design PGA in the definition of codes to obtain more uniform safety margins for structures designed to different levels of ground motions. This is strengthened by the observation shown earlier regarding the relatively high overstrength exhibited by the buildings designed to lower levels of ground motion.
5.2. Comparison with force mduction factor "demand"

It is interesting to compare the force reduction factors "supply" and the "demand" values suggested in several previous studies [e.g. Miranda and Bertero, 1994; Vidic et uL, 1994; Borzi and Elnashai, 20001. Based on a group of 124 ground motions recorded on different types of soil conditions, Miranda and Bertero [1994] have suggested the following expression for the force reduction factors:

where @ = 1

+ 1 2 T1 -pT

2 --exp[-2(ln~-;)'] 5T

forallurivnsites.

(11)

Borzi and Elnashai [2000] have employed the large dataset of Imperial College to derive force reduction factor spectra using 365 records of magnitude M, 2 5.5. It is noteworthy that the majority of the studies reported in the literature are based on bilinear non-degrading SDOF systems with zero or positive strain hardening [Krawinkler and Seneviratna, 19981. The study of Borzi and Elnashai [2000] employed two.hysteretic models, including a more advanced hysteretic hardeningsoftening model [Ozcebe and Saatcioglu, 19891, to derive trilinear force reduction factor spectra. The R factors "demand" suggested in the two studies of Miranda and Bertero [1994] and Borzi and Elnashai (20001 are employed in the comparison with the "supply" evaluated in the current study. It is important to note that the force reduction factor "demand" represents only the ductility-dependent component of R, whilst overstrength is not accounted for.

Calibmtion of Force Reduction Factors of RC Buildings

267

(a) Ducriliry level 'High' buildings


141

Downloaded By: [University of Illinois] At: 19:21 19 July 2010

mlR

(b) Ductility level 'Medium' buildings

10,

(c) Ductility level 'Low' buildings

.---.*-.--*-

p=4

0 0.0

Time (sec)
I

0.5

IO .

I .5

2.0

2.5

Fig. 13. Force reduction factors "supply" versus the "demand" suggested by Miranda and Bertero [I9941and Borzi and Elnashai [2000]: (a) ductility level "Highnbuildings; (b) ductility "Medium"; (c) ductility "Low".

Table 7. Observed overall displacement ductility Usupply" (p = A,,/A,,).


Group
-HO3O bddings
1.92

-M030 buildings
1.91

-1M015 buildings
2.21 2.62 3.11

-Ml5 b d d i n g s
1.96
2.29

IFRFFW-

2.10
2.61

1.98 2.27

2.83

Figure 13 shows this comparison, where the 12 buildings are classified again according to their ductility. The demand spectra are shown for different ductility levels. Although the 12 buildings investigated here were designed to three levels of ductility (High, Medium and Low), the difference between their actual displacement is not significant, as shown from Table 7. The maximum ductility ( p = A,,/A,) and the yield top displacements (A,, and A,) are evaluated from pushover analyses using an inverted triangular load. It is seen that the displacement ductility factors (ductility supply) of the 12 buildings range between 1.9 and-3.1. Maximum values are observed for the FW-builclmgs, whilst minimum factors are noted for the IF-group. The efficiency of the walls is reducing interstorey drift in the dual structures leads to relatively high A,, at first indication of interstorey drift collapse compared with A,, hence p is higher for this group of buildings. For frame structures, the twelve-storey buildings have higher ductility factors compared with the eight-storey group of structures. For the same group of buildings, the effect of the design ductility on the overall displacement ductility factor is insignificant. The above suggests employing the demand spectra of ductility level 2 and 3 only in the comparison with the R factors "suppIy". Notwithstanding comparisons of the force reduction factors "supply" with the code-adopted factors are more sigdicant, from the design point of view, than comparisons of the "supply" with the "demand". The latter can provide codidence in decisions taken to mod^ the R factors of the code. Considering the demand spectra of ductility 2 and 3, it is clear from Fig. 13 that the force reduction factor l l cases. This confirms the conservatism "supplyn is higher than the "demand" in a of the R factors employed in the design of the 12 buildings. Figure 13 also reveals a number of interesting features. It is observed that the buildings designed to higher ductility levels exhibit the highest margins of safety. This is clear when comparing the supply-to-demand ratio for buildings designed to the same PGA but for different ductility leveIs. The latter observation is not clearly noted in the comparison of the force reduction factor L'supply" with the design. Some other observations are common between the two comparisons shown in Figs. 12 and 13. It is confirmed that frame structures have higher safety margins than frame-wall buildings. The latter observation is not noticed in the buildings designed to ductility level "Low" , a s shown from Figs. 12(c) and l3(c). This is due to vulnerability of beams of frame buildings designed to ductility level "Low" to shear failure, as concluded above. For frame structures, the regular buildings exhibit

Downloaded By: [University of Illinois] At: 19:21 19 July 2010

Cdibmtion of Fone Reduction Factors of RC Buildings

269

higher supply-to-demand ratio compared with the irregular structural system. The trend shown in Fig. 13(b) for buildings designed to the same level of ductility but for difTerent ground motions also reinforces the conclusion drawn above regarding higher safety margins of buildings designed to lower PGA levels and the need to include the design PGA in the definition of R factors.
5.3. ~u&e,sted modification to the code R Factors

The lowest observed supply-to-design force reduction factors are those recorded for the dual structures designed to higher PGA levels (FW-HO3O and FW-MO30). The results obtained from the more conservative definition of R:,, show that factors may be increased by 70%. It should be noted that the concentration of demands at the wall base and coupling beams compared with perimeter frames represents a drawback of this structural system. Moreover, Elnashai and Mwafy [2001] investigated the possibility of predicting the inelastic period of buildings using effective flexural stiffness in eigenvalue analysis. It was concluded that reduction in the stiffness of the walls would result in considerable softening in the overall structural response. Due to the sensitivity and dependence of dual systems on walls as the main lateral force resisting element, it is recommended to apply relatively lower increase to Gode than that observed here. An increase in &ode between 1020% is suggested. The force reduction factors of dual structures designed to lower IeveIs of PGA may be increased hrther. In contrast with the latter structural system; strength, stiffness and ductility in frame buildings are uniformly distributed between different lateral force resisting elements. Hence, their &ode values may be safely increased according to the values observed here. However, for those designed to lower levels of ductility, particularly irregular frames, it is recommended first to improve the shear strength of beams and cut-off vertical members. It is also important to note that due to the established firm relationship between the force reduction factor and overstrength [Elnashai and Mwafy, 20011, the suggested increase in &ode values should not be coupled with major modifications in the code provisions that may lead to a reduction in overstrength. Indeed, increasing the design R factors will cause a reduction in overstrength, hence it should be performed gradually and assessment of ensuing structures should be rigorously carried out. For regular frame buildings designed to medium and high ductility levels, an increase between 30-40% may be applied. It may be argued that the conclusion drawn above regarding the available possibilities to increase the force reduction factors of EC8 i s subject to the modelling assumptions, ground motions and adopted performance criteria. However, the o b served high force reduction factors "supply" compared with the values adopted by the code show clearly the over-conservatism of the code and the possibility of increasing he. The recommended increase to EC8 force reduction factors is s u p ported by the fact that other major seismic design codes such as the US codes adopt higher R factors. For instance, the upper limit of R adopted by the NEHRP

Downloaded By: [University of Illinois] At: 19:21 19 July 2010

270

A.

M. Mwafy 8 A. S. Elnashai

provisions (FEMA 273, 19971 is 60% higher than that of EC8. This lends weight to feasibility of the modifications suggested here. F d y , it should be noted that EC8 values were also obtained from analysis, but using less rigorous analysis and assessment approaches.

6. Conclusions

Downloaded By: [University of Illinois] At: 19:21 19 July 2010

Rigorous approaches were adopted to calibrate t+e force reduction factors recommended by modern seismic codes. A wide range of RC buildings and extensive performance criteria including shear have been utilised. Inelastic pushover and s e rim of timehistory analyses up to collapse have been undertaken. Eight natural and artificial records scaled to different levels of velocity spectrum intensity have been employed. The effect of the vertical pound motion on the force reduction factors has been also investigated. The following conclusions, which are applicable to a large class of RC building, are drawn based on this investigation: The importance of including shear as a failure criterion and vertical ground m e tion in seismic assessment and calculations of the force reduction factors is proven in this study. For frame buildings designed to low levels of ductility, the shear failure criterion is the controlling parameter that defines the intensity of ground motion at first indication of member failure. The vulnerability of beams designed to low ductility to this type of failure is a serious issue. In several cases, vertical ground motion reduces the PGA that causes yield or collapse. It also reduces the mean values of the force reduction factors by up to la%, leading to higher seismic design forces. Although dual structural systems are efficient in eliminating the lateral drift, the advantages of utilising structural systems with more uniform st&ess distribution are confirmed here. The relatively high energy dissipated in the coupling beams and the wall base, their vulnerability to shear failure and the lowest R factor supply-tsdesign ratio obtained for this structural system reinforce this conclusion. Ln regular frame structures, plastic hinges and critical concrete strains occur in beams earlier than columns and the demand tends to be distributed uniformly between different lateral force resisting systems. Amongst other global collapse criteria, the interstorey drift is the collapse parameter that controls the response of buildings designed to modern seismic codes. For moment resisting frames, the critical concrete strain is generally exceeded marginally before the global ID collapse, reflecting the importance of utilising both criteria in assessment. Due to the efficiency of the lateral force resisting system of hybrid structures in controlling deformations, the ID collapse criterion is attained at notably high intensity levels. Member ductility criterion is more critical for this type of structural system. High supply-to-design force reduction factors utilising refined definitions of R are observed for all buildings investigated here. T h i s implies that R factors can be increased without adverse effects on structural safety. The satisfactory safety

Calibmtion of Forte Reduction Factors of

RC

Building3

271

Downloaded By: [University of Illinois] At: 19:21 19 July 2010

margins observed at the design and twice the design intensities and the high ground motions required to achieve collapse confirm the reliability of the buildings and support increasing Z&, factors. Buildings designed to higher ductility levels and lower PGA exhibit higher R factors. This implies that the effect of the design PGA should be incorporated in the expression of f i n d e similar to the inclusion of the effect of ductility. The lowest force reduction factors are observed in frame-wall buildings, which indicate a possibility to increase &ode by up to 70%. However, applying relatively lower increase than that observed here is recommended due to the sensitivity and dependence of this structural system on walls. Regular frame buildings designed to ductility class "High" and "Medium" exhibit the highest force reduction factors "supply", hence their kOde vaIues may be safely increased further. For frame structures designed to low ductility levels, particularly irregular systems, shear strength and ductility of beams and cut-off vertical elements should be enhanced and a more modest increase in the force reduction factors than those suggested for regular frame buildings designed to higher levels of ductility is recommended. The established firm relationship between the force reduction and the overstrength factors suggests applying a gradual increase in &,de factors and rigorous assessment of the performance of buildings designed accordingly. It is suggested to increase factors initially by 10-20% for hybrid structures and by 3 W 0 % for regular frame systems designed to medium and high ductility levels.

Whereas significant increase in R factors is recommended above, the suggested margins remain adequately conservative. Adoption of the above proposals would render EC8 a more economic code, without jeopardising the reliability and safety of the ensuing buildings.

References
Applied Technology Coyncil (ATC) [I9951 L'Structural response modification factors," Rep. No. ATC-19, Redwood City, California. Applied Technology Council (ATC) (19951 "A critical review of current approaches to earthquake resistant design," Rep. No. ATC-34, Redwood City, California. Associate Committee on the National Building Code, 1995, "National Building Code of Canada (NBCC)," National Research Council of Canada, Ottawa, Ontario. Balendra, T., Tan, K. and Kong, S. (19991 'Vulnerability of reinforced concrete frames in low seismic region, when designed according to BS 8110," Earthq. Engrg. Struct. Dyna. 28(11), 1361-1381. Borzi, B. and Elnashai, A. S. [2000]"Refined force reduction factors for seismic design," Engq. Struct. 22(10), 1244-1260. Elnashai, A. S. and Elghazouli, A. Y. [I9931 [[Performanceof composite steel/concrete members under earthquake loading, Part I: Analytical model," Earthq. Engq. Struct. Dyn. 22(4), 315-345. Elnashai, A. S. and Izzuddin, B. A. [I9931 "Modelling of material non-linearities in steel structures subjected to transient dynamic loading," EarUlq. Engrg. Struct. Dyn. 22,
509-532.

Elnashai, A. S. and Broderick, B. M. [I9961 "Seismic response of composite frames - II. Calculation of behaviour factors," E m . Struct. l8(9), 707-723. Elnashai, A. S. (2000) "Advanced Inelastic Static (Pushover) Analysis for Seismic Design and Assessment ," The Gemye Penelis Symposium,University of Thessaloniki, Greece. Elnashai, A. S. and Mwafy, A. M. (20011 "Overstrength and force reduction factors of multi-storey RC buildings," accepted for publication in The Structural Design of Tall Buildings, March 2001. ~ u r o c o d e2 [I9941 "Design of concrete structures. Part 1," CEN (Comite Europeen de Normalisation), European Prestandard ENV 1998-1-1, Bruxelles. Eurocode 8 [I9941 "Design provisions for earthquake resistance of structures. Part 1-1, 1-2, and 1-3," CEN (Cormte Europeen de Normalisation), European Pre-standard ENV 1998-1-1, 1-2, and 1-3, Bruxelles. Fardis, M. N. [I9941 "Analysis and Design of 26 Reinforced Concrete Buildings According to Eurocodes 2 & 8,"Configuration -3, 5 and 6, Report on Prenormative Research in Support of Eurocode 8. Federal Emergency Management Agency (FEMA) [I9971 "NEHRP provisions for the seismic rehabilitation of buildings," Rep. FEMA 273 (Guidelines) and 274 (Commentary), Washington, D. C Reeman, S. A. (19901 "On the correlation of code forces to earthquake demands," P r e ceedings 4th U.S.-Japan Workshop on Improvement of Building Structural Design and Construction Practices, ATC-153 Report, Applied Technology Council, Redwood City, California. IAEE (19921 "Earthquake resistant regulations, a world list," International Association for Earthquake Engineering, Tokyo, Japan. Izzuddin, B. A. and Ehashai, A. S. [I9891 "ADAPTIC - A program for static and dynamic analysis of structures by adaptive mesh refinement, user manual," ESEE Report 8917, Imperial College, London, UK. Krawinkler, H. and Seneviratna, G.D.P.K. [I9981 "Pros and cons of a pushover analysis of seismic performance evaluation," Engfg. Struct. 20(4-6), 452464. Martinez-Rueda, J. E. and Elnashai, A. S. (19971 "Confined concrete model under cyclic load," Mat. SCruct. 30(197), 13S147. Miranda, E. and Bertero, V. V. [I9941 "Evaluation of strength reduction factors for earthquakeresistant design," Earthq. Spectra 10(2), 357-379. Mwafy, A. M. [2001] "Seismic performance of codedesigned RC buildings," PhD Thesis, Imperial CoUege, London, UK. Mwafy, A. M. and Elnashai, A. S. [2001] "Static pushover versus dynamic collapse analysis of RC buildings," Engrg. Struct. 23(5), 407-424. Newmark, N. M. and Hall, W. J. [1982] "Earthquake spectra and design," Earthq. Engrg. Res. Institute, BerkeIey, California. Ozcebe, G. and Saatcioglu, M. [l989] "Hysteretic Shear Model for reinforced Concrete Members," J. Struct. Engrg. ASCE 115, 133-148. Papazoglou, A. J. and Elnashai, A. S. [I9961 "Analytical and field evidence of the damaging effect of vertical earthquake ground motion," Earthq. Engrg. Struct. Dyn. 25, 1109-1137. Priestley, M. J. N., Verma, R. and Xiao, Y. (19941 "Seismic shear strength of reinforced concrete columns," ASCE 120(8), pp. 2310-2329. Salvitti, L. M. and Elnashai, A. S. [I9961 "Evaluation of Behaviour Factors for RC Buildings by Nonlinear Dynamic Analysis," Proceedings 11 WCEE, Acapulco, Mexico.

Downloaded By: [University of Illinois] At: 19:21 19 July 2010

Calibrntion of Fone Reduction Factors of R C Buildings

373

Standards New Zealand [1992] "General structural design and design loadings for buildings," NZS'4203:1992, Standards New Zealand, Wellington. Uniform building code [I9971 International Conference of Building officials, Whittier, California. Vidic, T., Fajfar, P. and Fischinger, M. (19941 'Consistent inelastic design spectra: Strength and displacement," Earthq. Engrg. Struct. Dyn. 23(2), 507-521.

Downloaded By: [University of Illinois] At: 19:21 19 July 2010

Você também pode gostar