Vous êtes sur la page 1sur 8

A multilevel performance based seismic design method for reinforced

concrete structures
Blank line 11 pt
Blank line 11 pt
J.E. Barradas, A.G. Ayala & H. Castellanos
Instituto de Ingeniería, UNAM, México
Blank Line 9 pt

Blank line 11 pt
Blank line 11 pt
ABSTRACT
A method for a multilevel performance-based seismic design of reinforced concrete structures is presented. The
procedure defines a structure that does not surpass the exceedence rates of the performance indices which
characterize multiple design limit states and their corresponding seismic demands generally defined by means of
uniform hazard spectra. To illustrate and validate this method this paper only deals with the serviceability limit
state, for which the structure behaves elastically and the ultimate, for which an acceptable distribution of damage
is also considered. As example the design of a 17 storey reinforced concrete plane frame is described using as
demands response spectra calculated from real records at a site in Mexico City consistent with the seismic
demands associated to the design limit states. The quality of the results obtained is assessed by comparing the
design performance indices with those obtained from nonlinear dynamic analysis using the design earthquake
records, the agreement found was satisfactory.
Blank line 11 pt
Keywords: Multi-level design, performance-based seismic design, displacement based design.
Blank line 11 pt
Blank line 11 pt
1. INTRODUCTION
Blank line 11 pt
Current seismic design regulations stipulate that the performance of a structure must not exceed the
accepted limit states associated with serviceability and ultimate conditions, when subjected to seismic
demands of frequent and rare occurrence, respectively. However, in most cases the analysis and design
of a structure is done considering only a design spectrum representative of seismic demands associated
with the ultimate limit state, checking that for the design obtained the stiffness of the structure also
complies with the conditions of the service limit state by considering as seismic demand a reduced or
scaled spectrum obtained from the ultimate design spectrum, which is an approach that does not
represent correctly the seismic demands associated with that limit state. Furthermore, during this
process no care is generally taken to check whether the structure exceeds its elastic strength demand
represented by this scaling spectrum.
Blank line 11 pt
Recent literature on the performance based seismic design of structures, contains some procedures for
the multilevel seismic design of reinforced concrete structures, e.g., Bertero, 2001 and Bertero and
Bertero, 2002. These procedures design explicitly consider two limit states and their corresponding
seismic demands, however, they make use of theoretical concepts not commonly familiar to structural
engineers which application requires a considerable degree of expertise in the seismic design of
structures. Considering that one of the most important purposes of new design procedures should be
their potential of practical applicability, it is desirable that they make use of concepts known to
practice structural engineers and also that they can be easily implemented using commercial software.
Blank line 11 pt
This paper presents a method for the multilevel seismic design of reinforced concrete frames through
the simultaneous consideration of the requirements of stiffness and strength for serviceability and
ultimate limit states associated to low-intensity earthquakes of frequent occurrence and earthquakes of


 
large intensity and rare occurrence, respectively, G.D.F., 2004. Even though not discussed in this paper
due to length limitations, the design seismic demands are defined by uniform hazard spectra, which
characterize the seismic hazard of the site where the structure is located.
Blank line 11 pt
This method is based on the assumption of an acceptable predefined damage configuration for the
structure occurs under ultimate design conditions; on the use of concepts of basic structural dynamics
to construct a bilinear behavior curve of a reference single degree of freedom, SDOF, system generally
associated to the fundamental mode of the structure. To illustrate and validate this procedure, the
multilevel design of a 17 storey reinforced concrete frame is carried out and its design performances
compared with those obtained from nonlinear step by step analyses.
Blank line 11 pt
Blank line 11 pt
2. DESCRIPTION OF THE PROCEDURE
Blank line 11 pt
In the formulation of the design procedure the following considerations and assumptions are adopted:
Blank line 11 pt
The design procedure presented is derived from a method for the seismic performance evaluation of
structures proposed by Ayala, 2001. This method makes use of a simplified reference SDOF system
defined in terms of the dynamic properties of the fundamental mode of the structure. The behavior of
this system is represented by a bilinear curve in the ADRS spectral space defined in the spectral
pseudo acceleration and the spectral displacement, i.e., Sa. vs Sd, as shown in Fig 1. These variables
are related by the equation: Sa=(2π/T)2 Sd, where T is the period of the system. This equation shows
that the slopes of the initial, elastic, and post yielding branches of the bilinear curve are respectively
defined by the elastic period, T1, and the inelastic period, T2. The coordinates of the breaking point of
this behavior curve are the elastic strength per unit mass associated with the fundamental mode, R1/m,
and its corresponding yield spectral displacement, Sd1, and the end point coordinates are the total
strength per unit mass, R2/m, associated to the same mode and the ultimate spectral displacement, Sd2,
Fig.1. The slope of the post-yielding branch of the system, k2, is expressed as a fraction α of the slope
of its initial branch, k1, i.e., α=k2/k1=(T1/T2)2.
Blank line 11 pt
To control the elements that are allowed to develop damage under design conditions, the distribution
of structural damage is modeled by introducing hinges at the ends of these elements, trying in this way
to represent the formation of plastic hinges. This simplification ignores the post yielding stiffness of
the reinforced concrete sections, i.e., it slightly underestimates the lateral stiffness of the structure and
its strength and overestimates its lateral displacements in the range of nonlinear behavior.
Blank line 11 pt
Based on the foregoing assumptions, the steps followed by the design procedure are:
Blank line 11 pt
1. The preliminary design of structural elements is carried out to generate a lateral stiffness of the
structure that does not allow the exceedence the design interstorey drift, γserv, for the serviceability
limit state (SLS), intended to avoid damage and nonstructural elements under these design
conditions. This preliminary design is obtained iteratively through a series of spectral modal
analysis considering elastic behaviour for the structure and a uniform hazard design spectrum as
service seismic demand. This spectrum is representative of frequent earthquakes at the site, scaled
by a factor defined as the ratio of the interstorey drift which starts structural damage, γdb, and γserv
defined to. This approach has as objectives to limit the interstorey drifts of the structure to the
value given by, γserv when subjected to the unscaled seismic demand, and to obtain the elastic
strength or initial yield of the structure. From the results of the last analysis of this iteration the
fundamental period of the structure, T1, and their spectral coordinates, Sadb, and, Sddb, are used to
build the initial branch of the behaviour curve of the reference SDOF system, Fig.1
Blank line 11 pt


 
Sa
R 2 /m

Sa db ˜ R 1 /m 2
˜ 2

2
1

Sd

Sd db ˜ R 1 /m R 1 /m
˜ 2 = Sd 1 Sdu = 2 µ = Sd 2
1 1

Blank line 10 pt
Figure 1. Bilinear behavior curve of the simplified reference SDOF system
Blank line 11 pt
2. An acceptable damage distribution in the structure is proposed. This distribution must be in
accordance with the performance level for which the structure will be designed (ultimate limit
state, ULS), Fig.2. A criterion to introduce a damage distribution involves the placement of
hinges mainly at the ends of the beams, trying to encourage the formation of strong column-
weak beam mechanisms. Once the damage is introduced in the model a modal analysis is done
to obtain the fundamental period of the damaged structure, T2, and the displacements
associated to the fundamental mode shape. With the values of T1 and T2 the ratio α is
calculated.
Blank line 11 pt
3. A design interstorey drift is proposed for the ULS, γu, and the displacements obtained in the
previous step are amplified so that the maximum interstorey drift be equal to γu. From the
amplified roof displacement and the dynamic properties of the fundamental mode, the ultimate
displacement, Sdu, of the reference SDOF system is obtained. The ratio of this spectral
displacement and the spectral displacement for which structural damage is initiated, Sddb,
defines the displacement ductility, μ, of the reference SDOF system.
Blank line 11 pt

Blank line 10 pt
Figure 2. Damage distribution associated with the ultimate limit state
Blank line 11 pt
4. With the values of μ, α, damping ratio, ξ, of a SDOF system and the seismic records
associated to the ULS, a uniform hazard strength per unit mass spectrum is constructed. In this
spectrum, the ordinate corresponding to the period T1 of the reference SDOF system, R1/m, is
read, Fig.3. If R1/m is approximately equal to Sadb this strength per unit mass will be adequate
for the system to remain elastic when subjected to the seismic demand associated to the scaled
SLS, and to develop the ductility, μ, when subjected to the seismic demand associated to the
ULS.
Blank line 11 pt
If R1/m is different from Sadb there are three possibilities:
Blank line 11 pt


 
a). If R1/m is smaller than Sadb and their difference is moderate, to increase its value a
different damage distribution in accordance with the ultimate limit state should be proposed,
to increase the value of α. Furthermore, a smaller γu may be proposed and the procedure
repeated from the second step.
Blank line 11 pt
200

150
 R1/m ≈ Sadb 
R/m  100

50

T1  
0
0.0 1.0 2.0 3.0 4.0 5.0
Period
Blank line 10 pt
Figure 3. Strength per unit mass spectrum for μ and α, associated to the ultimate limit state
Blank line 11 pt
b). If R1/m is larger than Sadb and their difference is moderate, to reduce its value a different
damage distribution, which might be more severe, should be proposed to reduce the value of
α. Furthermore, a larger γu may be proposed and the procedure repeated from the second step.
Blank line 11 pt
c). If the difference between R1/m and Sadb is large, to this difference and produce ductilities
within the range recommended by the design recommendations, it may be necessary to modify
the elastic period of the structure and repeat the method from the beginning.
Blank line 11 pt
5. The strength per unit mass, R2/m, is obtained by plotting the complete behaviour curve of the
reference SDOF system, Fig 1, or analytically, considering the relationship it has with the
design ductility and α, i.e.:
Blank line 11 pt
R2 / m  R1 / m 1     sv -1  (1)
Blank line 11 pt
6. To obtain the design forces of the structural elements where damage is accepted to occur when
the structure is subjected to the seismic demand associated with the ULS and the partial design
forces for the elements where damage is not allowed to occur, a gravitational load and a
spectral modal analysis of the undamaged structure is carried out, Fig 5a, using as demand the
elastic design spectrum representative of earthquakes associated to the ULS, scaled so that for
the initial fundamental period of the structure, T1, its corresponding pseudo acceleration, Sao,
be equal to R1/m, Fig 4 a.
Blank line 11 pt
It may be noticed that from this analysis, the element forces obtained are such that, their
performance remains elastic. Thus, it is important at this stage to compare these strengths to
those obtained with the last spectral modal analysis of the designed preliminary structure for
the SLS, and assign as design forces the largest of both analyses.
Blank line 11 pt
7. For the complementary part of the design forces of the structural elements in which damage
was not allowed to occur, a second modal spectral analysis is performed to the damaged
structure without gravitational loads, Fig 5b, using the same elastic spectrum but now scaled
so that for the fundamental period of the damaged structure, T2, the corresponding pseudo
acceleration, Sao, be equal to the difference of strengths per unit mass, R2/m- R1/m, Fig 4b.
Blank line 11 pt
8. The design forces of the structural elements in which no damage is allowed, are obtained
adding the element forces obtained from the two previous analyses.
Blank line 11 pt


 
1000 1000

Elastic ultimate  Elastic ultimate 
800 800 spectrum
spectrum
Escaled elastic 

R/m o Sa
T1, Sao Escaled elastic  600
R/m o Sa

600
ultimate  ultimate 
Sao‐ R1/m  spectrum 400
spectrum
400

200 T1, R1/m 200 Sao‐ (R2/m ‐R1/m) 


T2, (R2/m ‐ R1/m) 
0 0
0.0 1.0 2.0 3.0 4.0 5.0 0.0 1.0 2.0 3.0 4.0 5.0
Period Period

a b
Blank line 10 pt
Figure 4. Scaled strength spectra for the design forces of the elements a) with damage
allowed and b) with damaged not allowed
Blank line 11 pt

(a) (b)
Blank line 10 pt
Figure 5. Analyses required to obtain the final design forces of the structural elements
Blank line 11 pt
9. During the design process it is common to find that for two or more structural elements of the
same type and even contiguous damage is accepted for some and is no accepted for the others,
i.e., their design forces and the assigned reinforcing steel are different, something that leads to
a structure impractical to build. Due to this, it is necessary to harmonize the design by
modifying their strength, however, this standardization leads to different design forces and, as
a result, a different damage distribution to that originally proposed, something that, in general,
significantly modifies the structural performance assumed in limit states considered. The
definition of a criterion to have uniform design forces with the least possible impact in the
desired design performance and assumed damage distribution, is an aspect of this performance
based design method that needs further investigation.
Blank line 11 pt
Blank line 11 pt
3. APPLICATION EXAMPLE
Blank line 11 pt
To illustrate the practical implementation of the proposed procedure, the design of the regular
reinforced concrete frame shown in Fig 6 is carried out.
Blank line 11 pt
The design gravity loads considered are those stipulated for office buildings by the Mexican Building
Regulations for the Federal District, G.D.F., 2004. The mechanical properties of the materials used are
those recommended by these standards for the design of ductile reinforced concrete structures.
Blank line 11 pt
The values of the design interstorey drifts are, for the SLS the allowable value specified by the rules
0.004 and for the onset of structural damage, according to (Reyes, 1999), the value of the interstorey
drift when damage starts in beams which are part of a structural system based on ductile reinforced
concrete frames is 0.005


 
Blank line 11 pt

Beams Storey Columns


sections height sections
17
16 0.45 x 0.45
15
14
0.35 x 0.60 13
0.60 x 0.60
12
11
10
9 @ 3.2 0.75 x 0.75

8
7
6
5
0.35x 0.75
4
0.95 x 0.95
3
2
1
4.0

8.0 8.0 8.0


Dimensions in meters
Blank line 10 pt
Figura 6. Geometry of the example 17 storey frame
Blank line 11 pt
To validate the proposed method and show the quality of the results obtained, uniform hazard design
spectra are not used and, for the same sake, element forces are not standardized. This work considers
as seismic demand associated to the ULS, the spectrum of the East-West component of the record of
the 19 September 1985 earthquake recorded at the SCT station in Mexico City and as demand
associated to the SELS, the spectrum of the East-West component of the 25 April 1989 earthquake
recorded at the same station. It is noteworthy that these earthquakes are respectively considered
representative of the seismic movements of rare and frequent occurrence in Mexico City.
Blank line 11 pt
After applying the procedure for the SLS, the sections of the elements for which the lateral stiffness of
the structure complies with the requirements of this limit state are obtained. The fundamental period of
the frame is T1 = 1.77 s.
Blank line 11 pt
After trying some acceptable damage distributions and an acceptable value for γu, the strength R1 /m
readed from the inelastic spectrum built for values of α and μ, and for the period of 1.77 s, is roughly
equal to Sadb. Fig 9 shows the frame with the proposed damage distribution which now has a
fundamental period T2 = 8.54 s and a ratio α = 4.31%. Fig 7 shows the performance curve of the
reference SDFOF system constructed during the application of the method. It is noteworthy that the
value obtained from μ ≈ 2.50, which follows the behaviour curve, is in the range of values specified in
the Mexican standards it was not necessary to modify the initial period of the structure.
Blank line 11 pt
Sa

1371.9 mm/s2
1180.9 mm/s2 0.866 rad/s

13.19 rad/s

Sd
89.5 mm 311.1 mm
Blank line 10 pt
Figure 7. Behavior curve of the reference SDOF system of the designed structure
Blank line 11 pt
Modal spectral analyses are carried out with the scaled spectrums. To obtain the design forces of
beams and columns, the element forces calculated from each of the analyses are added.
Blank line 11 pt
Blank line 11 pt


 
4. VALIDATION OF RESULTS
Blank line 11 pt
To evaluate the design procedure, the parameters or performance indices established for each limit
state and those obtained from nonlinear dynamic analysis are compared, i.e.,:
Blank line 11 pt
 When the structure is subjected to a seismic demand associated with the SLS, the maximum
interstorey drift should be approximately equal to the permissible level for this limit state and
no structural element must exceed its strength.
Blank line 11 pt
 When the structure is subjected to a seismic demand associated with the ULS, the overall
damage distribution and the overall ductility demand should be very similar to those assumed
as target. In addition, the maximum interstorey drift should be approximately equal to the
design value for this limit state. Values in a range of ± 10% are considered acceptable.
Blank line 11 pt
For this example, two nonlinear step by step analyses with seismic demands associated to the two
limits states considered were carried out. These analyses were performed using the program DRAIN-
2DX (Prakash et al., 1993). The results obtained are described in the following paragraphs.
Blank line 11 pt
Fig 8 shows the maximum interstorey drift. It may be observed that when the structure is subjected to
the seismic demand associated with SLS, some interstorey drifts exceed lightly the permissible level
for this limit state, γserv, however, they are within an acceptable range, with no structural element
exceeding its strength. The same figure shows that when the structure is subjected to the seismic
demand associated with the ULS, the maximum interstorey drift was less than the target for this limit
state, γu, but also in this case the difference is within the acceptable limits.
Blank line 11 pt
Storey γserv  γu 
18
16
14
12
10
8
6
4
2
0
0.0000 0.0030 0.0060 0.0090 0.0120 0.0150
Interstorey drift
Before seismic demand asociated to serviceability
Before seismic demand asociated to ultimate limit state

Blank line 10 pt
Figure 8. Interstorey drifts in the example frame
Blank line 11 pt
Fig 9 shows a comparison of the proposed and developed damage distributions in the example frame
due to the seismic demand associated with the ULS. It may be observed that they are very similar
except that the later, hinges are formed in some columns and in all beams of the two highest storeys.
Blank line 11 pt
Blank line 11 pt
5. CONCLUSIONS AND RECOMMENDATIONS
Blank line 11 pt
This paper presented and illustrated the application of a multilevel performance based method for
seismic design of structures, which allows the explicit and simultaneous consideration of two or more
limit states. The validation of results obtained by this design method is done by comparing the target
performances to those of the nonlinear dynamic analyses of the frame carried out considering as
seismic demands for each limit state records of real earthquakes consistent with the demands
recommended by the code. It is evident that in the practical applications of this method uniform hazard


 
design spectra, as those of the code, should be considered and not as in this case spectra derived from
real records. From the analysis of the results obtained it can be concluded that:
Blank line 11 pt
1. The slight underestimation of the lateral stiffness of the structure as a range of nonlinear behavior,
led to a maximum interstorey drift close and somehow smaller than the interstorey drift used for
design. The post yielding stiffness considered in the nonlinear behaviour for the structural elements
was the typical value of 5%. Regarding its effect on the design forces of the structural elements where
the damage is accepted, the average differences are of 15%.
Blank line 11 pt
2. In general, the structure met the performance standards established for each limit state, something
that shows the potential of the proposed design procedure.
Blank line 11 pt

Proposed hinge
Developed hinge
Blank line 10 pt
Figure 9. Comparison of damage distributions, proposed and developed under seismic action
Blank line 11 pt
Blank line 11 pt
ACKNOWLEDGEMENTS
The sponsorship of the National Council for Science and Technology of Mexico the scholarships awarded to the
first and third authors for the completion of their graduate studies and of the project "Development of the
conceptual framework, theoretical models and simplified methods for evaluation and seismic design of structures
based on performance, " in which this work is based is acknowledged.
Blank line 10 pt
Blank line 10 pt
REFERENCIAS
Blank line 10 pt
Ayala A.G. (2001). Evaluation of seismic performance of structures, a new approach. (in Spanish) Revista
Internacional de Métodos Numéricos Para Cálculo y Diseño en Ingeniería. Vol 17: No. 3, 285-303.
Bertero R. (2001). Earthquake Resistant Performance based Design. (in Spanish) PhD thesis. University of
Buenos Aires, Argentina
Bertero R. and Bertero V. (2002). Performance – based seismic engineering: the need for a reliable conceptual
comprehensive approach. Earthquake Engineering and Structural Dynamics. Vol. 31: No. 3, 627–652.
Prakash V., Powell G. and Campell S. (1993). DRAIN-2DX, Base Program Description and User Guide,
version 1.10 Report No. UCB/SEMM-93/17. Departament of Civil Engineering. University of California,
Berkeley CA..
G.D.F. (2004). Construction Code for the Federal District, Complementary technical standards on criteria and
actions for the structural design of buildings and Complementary technical standards for seismic design,
Official Gazzette, Mexico.
Reyes J. C. (1999). The service limit state seismic design of buildings. (in Spanish) Doctoral Thesis. Graduate
Studies Division, Faculty of Engineering, Universidad Nacional Autonoma de Mexico, Mexico City.
México


 

Vous aimerez peut-être aussi