Вы находитесь на странице: 1из 12

A rationale for determining the natural period of RC building

frames having infill


Khan Mahmud Amanata, and Ekramul Hoqueb

a
Department of Civil Engineering, BUET, Dhaka, Bangladesh
b
Bangladesh Open University, Gazipur, Bangladesh
Received 7 December 2004;
revised 1 September 2005;
accepted 2 September 2005.
Available online 25 October 2005.

Abstract
Building design codes generally impose some upper limit on the magnitude of the natural period determined from a
rational numerical analysis if the period is longer than that predicted by empirical code equations since the code
equations are derived on the basis of measuring the period of real buildings during an earthquake. In this study, the
fundamental periods of vibration of a series of regular RC framed buildings are studied using 3D FE modeling and
modal eigenvalue analysis including the effects of infill. It has been found that when the models do not include infill,
as is done in conventional analysis, the period given by the analysis is significantly longer than the period predicted
by the code equations justifying the imposition of upper limit on the period by the code. However, when the effect of
infill is included in the models, the time periods determined from eigenvalue analysis were remarkably close to those
predicted by the code formulas. It is also observed that the randomness in the distribution of infill does not cause
much variation of the period if the total amount of infilled panels is the same for all models. It is also observed that
varying amount of infilled panels causes some changes in the determined period. Based on the findings of the study,
some guidelines are suggested for determining the period. The findings of the study have shown us a practical way to
determine the fundamental period of RC frames using rational approaches like modal analysis, and eliminate the
necessity of imposing code limits.

Keywords: Natural period; Frames; Infill; Reinforce concrete

Article Outline

1.
Introduction
2.
Infill in RC structure
3.
Computational modeling
4.
Code equations for time period
5.
Sensitivity analysis
5.1. Effect of column stiffness, beam stiffness, floor height, numbers of floors and number of bays
5.2. Effect of panel span length
5.3. Influence of number of spans
6.
Period with randomly distributed infill
7.
A rationale for estimating period
7.1. The rationale
7.2. Effect of the proposed rationale on base shear
8.
Conclusion
References

1. Introduction
Building codes provide empirical formulas for estimating the fundamental period. These formulas are
developed on the basis of observed periods of real buildings during ground motion and the period is
generally expressed as a function of building height, type (frame or shear wall), etc. Building periods
predicted by these empirical equations are widely used in practice although it has been pointed out by
many[2], [3], [8] and [15] that there is scope for further improvement in these equations. With the wide
availability of high-speed personal computers it is now possible to develop a rigorous finite element (FE)
model of a structure and determine its natural period by means of the exact eigenvalue analysis or by any
rational method like Rayleigh’s method. However, the period obtained by such rational methods has been
generally found to be significantly longer than the observed period of the buildings [4], [2] and [3]. For this
reason, code specifications (BNBC [5], UBC [7], BSLJ [11]) generally put a limit on the period value if it is
obtained by eigenvalue analysis of the FE model. This, in fact, discourages the use of periods obtained
from computational modeling. Conventional FE modeling of reinforced concrete (RC) structures, which are
widely used in strength analysis and design, renders the structures more flexible than they actually are
due to the fact that the effect of secondary components like the infills are not considered in the modeling.
In reality, the additional stiffness contributed by these secondary components increases the overall
stiffness of the building, which eventually leads to shorter time periods as they are observed during
earthquakes.
The objective of this paper is to investigate the natural period of vibrations of RC buildings by means of FE
modeling under various conditions of geometric and other parameters including the effect of regular as
well as randomly distributed infill. The diagonal strut model of infill proposed by Saneinejad and
Hobbs [14] and later enhanced by Madan et al. [10] has been adopted in the present study. Three-
dimensional finite element modeling of a series of some idealized regular shaped building frames is
analyzed, and their time periods are determined by means of modal (eigenvalue) analysis. These periods
are then compared with the code formulas as well as periods obtained from analysis without considering
infill. The study has been conducted under varying conditions of number of floors, floor heights, number of
spans, amount of infilled panels, etc. Infilled panels were distributed over the structure in regular as well
as in a random pattern. Comparisons of the periods obtained for different parametric conditions revealed
the relative accuracy of the different methods and established the importance of incorporating the
structural effect of infill in FE modeling to more accurately reflect the dynamic behavior of RC frames
which is otherwise not possible to obtain in conventional frame modeling.

2. Infill in RC structure
Infills of brick or stone masonry are frequently used in RC framed buildings. Although these are primarily
intended to serve as partitions, their structural contribution in increasing the lateral stiffness of the frame is
long recognized. There are several analytical models of infill available in the literature, which can be
broadly categorized as (a) continuum models such as the models proposed by Lourenco et al. [9] and
Papia [13] and (b) diagonal strut models such as the model proposed by Saneinejad and Hobbs [14]. For
the type of work presented in this paper the diagonal strut model of Saneinejad and Hobbs [14] has been
found to be more suitable. This model has been successfully used by Madan et al. [10] for static
monotonic loading as well as quasi-static cyclic loading. They have also successfully verified the model by
simulating the experimental behavior of tested masonry infill frame sub-assemblage. A brief outline of the
equivalent diagonal strut model is discussed below.
Considering the masonry frame of Fig. 1, the maximum lateral force Vm and the corresponding
displacementum in the infill masonry panel [10] are

(1)

(2)
in which t = thickness of the infill panel; l′ = lateral dimension of the infill
panel; = masonry prism strength; ε′ = corresponding strain; θ = inclination of the diagonal strut; v
= basic shear strength of masonry; and Ad and Ld = area and length of the equivalent diagonal struts
respectively. These quantities can be estimated using the formulations of the “equivalent strut model”
proposed by Saneinejad and Hobbs [14]. The initial stiffness K0 of the infill masonry panel may be
estimated using the following formula [10],

(3)
K0=2(Vm/um).

Full-size image (60K)

Fig. 1. Strength envelope for masonry infill panel (Madan et al. 1997).

The parameters Vm,um,K0,K1, etc., are clearly shown in Fig. 1. The degradation of strut stiffness
from K0 to K1was assumed to be a bilinear curve by Madan et al. [10]. A more rational degradation path
would be a smooth curve shown by the heavy solid line in Fig. 1. In this paper the form of the curve is
suggested as given below,

(4)
where

(5)

3. Computational modeling
In this study common two-noded frame elements having six degrees of freedom per node has been used
for the columns. For beams, similar elements with node offset capabilities has been used to model the
web of T-beams (monolithic beam and slab). The floor slab has been modeled using common four-noded
plate elements. Point mass elements are used to represent the non-structural dead load like floor finish,
partition walls, etc. The infills are modeled as diagonal struts using two-noded truss elements having only
three translational degrees of freedom at each node. The plan and elevation of the reference building are
shown inFig. 2. The reference RC frame has the properties given in Table 1.

Full-size image (63K)

Fig. 2. Plan and elevation of the reference RC frame.

Table 1.

Building parameters under study

Referen
Sl.
Parameters ce Studied values
no.
values

2×104 N
1 Modulus of elasticity of concrete
/mm2

2.4×10−9
2 Density of concrete
ton/mm3

500 mm 300×300, 400×400,


3 Size of column ×500 m 500×500, 600×600,
m 700×700 mm2

350 mm 350×300, 350×400,


4 Size of beam (width × depth) ×500 m 350×500, 350×600,
m 350×700 mm2

5 Number of story 10 4, 7, 10, 13, 16

3500 2500, 3000, 3500, 4000,


6 Height of each story
mm 4500 mm

Number of spans (along the direction


7 4×4 2, 3, 4, 5, 6
of motion and its transverse direction)
Referen
Sl.
Parameters ce Studied values
no.
values

6000 4000, 5000, 6000, 7000,


8 Size of each span
mm 8000 mm

9 Thickness of slab 150 mm

Amount of infilled panels (percentage


10 40% 20%, 40%, 60%, 80%
of total no. of panels)

11 Thickness of infill 250 mm


Full-size table

The time period of a building is basically a function of its mass and stiffness. Thus any structural or
building parameter that changes the stiffness or mass shall have influence on the period. Typical
parameters are building height, floor to floor height, span length of panels (span) in the direction of motion
as well as in the transverse direction (bay), number of panels in the direction of motion and in the
transverse direction, column and beam stiffness, infill effective as bracing, etc. In order to assess the
relative influence of these parameters, a systematic sensitivity analysis has been performed by varying
such parameters one by one while keeping the other parameters unchanged. The rightmost column
of Table 1 lists the parameters which were included in the investigation and their range of variation.

4. Code equations for time period


The empirical formula for the fundamental period of vibration of RC buildings specified in the Uniform
Building Code (UBC) and Bangladesh National Building Code (BNBC) is of the form

(6)
where h is the height of the building above the base. For framed RC buildings the
numerical coefficient Ct=0.03 for UBC when h is in feet and Ct=0.073 in BNBC when h is in meters. Both
codes permit an alternative way to calculate the fundamental period T using the structural properties and
deformational characteristics of the resisting elements in a properly substantiated analysis. The alternative
formula for determining the period is

(7)

The values of fi represent any lateral force distributed approximately in accordance with rational
distribution and the wi’s are the weights of individual floors. The elastic deflections, δi, shall be calculated
using the applied lateral forces, fi. However, Eq. (7) is rarely used in practice. In Indian Standard (IS)
Criteria for Earthquake Resistant design of Structures, the period T is calculated as [6]

(8)
T=0.1n where n is the number of floors above base. The Building Standard Law of Japan (BSLJ) uses
the following formula for the fundamental period, [1]

(9)
T=h(0.02+.01α), where h is the building height and α the ratio of the total height of steel construction to
the height of the building. In the National Building Code (NBC) of Canada the period is calculated by [12]
(10)
T=0.1N, where N is the number of floors above exterior grade. The fundamental period T, calculated
using any of Eqs. (6), (7), (8), (9) and (10) should be smaller than the actualperiod to obtain a
conservative estimate for the base shear. Therefore, the code equations are generally calibrated to give a
period lower than the actual by about 10%–20%.
Code provisions generally permit the use of computational techniques like eigenvalue analysis using FE
modeling, but specify that the resulting period value should not be longer than that estimated from the
empirical equations by a certain factor. The factor specified in UBC [7] is 1.3 for high seismic zones and
1.4 for other zones. In BNBC [5] the factor is 1.2 for all zones. These restrictions are imposed to
safeguard against unreasonable assumptions in the rational analysis, which may lead to unreasonably
long periods and hence unconservative values of base shear. When FE modeling is employed, an RC
building is typically modeled using common frame elements to model the beams and columns and plate
elements are used to model and slabs and shear walls. The structural contribution of the infill is generally
not included. From the designers’ point of view, such modeling and analysis under static load results in a
safer structure since only the primary structural system carries the load. However, such modeling without
considering the structural contribution of infill also renders the structure too flexible, resulting in an
unacceptably longer value of the period. Therefore, conventional modeling is not suitable for determining
the fundamental period necessary for calculating equivalent static earthquake load or doing dynamic
analysis of RC frames having infill.

5. Sensitivity analysis
Earlier studies on the sensitivity of different structural parameters [1] reveals that when infill is considered
in the analysis of building, the nature of the stiffness contribution of structural elements like beams and
columns or the effect of building parameters like the number of spans or floor panel size are significantly
altered. Other parameters commonly established to have effect on the period are the number of floors and
floor height or building height. In each case a complete 3D Fet model of the building is analyzed. Analysis
is performed including the effect of infill by modeling them as diagonal struts [10] as well as without infill.
The time periods obtained from eigenvalue analysis of both cases are then compared with the period
obtained from code equations. These comparisons established the relative importance of incorporating
infills in the FE model. The following subsections discuss some of the results in more details. It may be
mentioned that although the results of Fig. 3a, Fig. 3b and Fig. 3c correspond to the reference building
(Table 1), we can at least have some general idea of the characteristics of RC frames from these figures.

Full-size image (35K)

Fig. 3a. Period vs. column stiffness.


Full-size image (34K)

Fig. 3b. Period vs. span length.

Full-size image (32K)

Fig. 3c. Period vs. number of span.

5.1. Effect of column stiffness, beam stiffness, floor height, numbers of floors and number
of bays
Fig. 3a shows the determined period versus column stiffness (EI/L). For the frame modeled without infill,
the period is highly sensitive to column stiffness, but for the frame with infill it can be observed that the
fundamental period does not depend much on the column stiffness. This is due to the fact that when infills
are active as diagonal bracing, the stiffness induced by them against lateral deflection is much more
pronounced than the effect of column stiffness. The periods obtained from code equations are also plotted
for comparison. It can be observed that the analysis with infill closely agrees with code equations while the
frame without infill grossly overestimates the period.
Similar results are obtained in cases of varying the beam stiffness, number of floors, floor height and
number of bays. For brevity, the graphs are not shown here. In all these cases, it has been observed that
the period obtained without infill is too long compared to code predictions while the period with infill is
close to the code predictions.

5.2. Effect of panel span length


The code equations do not have any provision to incorporate the effect of span lengths (panel length in
the direction of motion) in determining time period. As such, the periods predicted by these empirical
equations are the same for all values of span length studied. The periods obtained from modal analysis of
the infilled frame do show some variation with the changing values of panel size. It can be observed
from Fig. 3b that the period increases with the increase in panel size. However, the periods obtained lie in
close proximity to the values from the code equations, supporting the acceptability of the code equations.
The modal analysis of the frames without infill produces periods significantly longer, justifying the code
provisions of limiting the period obtained from conventional rational analysis.
It can be observed from the figure that at 6000 mm span, the period is about the same as that obtained
form code equations. The period decreases for smaller spans and increases for higher spans. The reason
is that, for longer spans, the stiffness contribution of the infill decreases as formulated by Madan et al.
(1997). Thus the overall stiffness against lateral sway decreases, with the increase in span resulting in
longer periods. It can be observed from Fig. 3b that the period changes by about 8% for each 1 m change
in span from the reference value of 6 m.

5.3. Influence of number of spans


Increasing the number of spans decreases the period of the structure (Fig. 3c). This is observed from both
the analyses with infill as well as the analysis without infill. Code equations are not capable of reflecting
this fact since there is no parameter relevant to span in the code equations. However, the modal analysis
with infill resulted in periods in the close vicinity of the periods estimated by the code equations while
analysis without infill resulted in periods more than 200% longer. It may be mentioned that the reference
building is ten storied. Thus, when there are only two spans, the building becomes relatively slender and
cantilever action comes into effect against lateral sway which is likely to make the structure more flexible,
resulting in longer period. It can be observed from Fig. 3c that for the number of spans being three or two,
the period decreases by about 10% or 20% respectively. For the number of spans being four or higher,
the difference in period given by code equations and analysis with infill is not very significant.

6. Period with randomly distributed infill


It is already apparent from the results presented in the previous section that the building period obtained
by the analytical method is close to the period predicted by codes when infill is present in the FE model. In
all the results presented in previous section, the number of infilled frame panels was fixed at 40%. In
practical cases, the amount of infill will vary from building to building. Also their arrangement or location
will be different from building to building. Therefore, the effect of different amount of infilled panels in the
same FE model is studied. The amounts of infilled panels included in the investigation are 20%, 40%,
60% and 80%. For example, the reference building has ten stories and four bays in each direction. Thus
the number of vertical frames in each direction is five. Considering this, the total number of frame panels
in the building is 400 (4 span × 10 stories × 5 frames × 2 directions). Thus to provide 40% infill, we need
160 panels—80 in each direction. These 80 panels are chosen randomly, and a diagonal strut is provided
in those panels. In order to get an average result such random distribution was made five times for every
percentage amount of infill.
Fig. 4 shows the periods obtained for different random distributions of infilled panels. There are four lines
corresponding to four different amount of infilled panels, i.e. 20%, 40%, 60% and 80%. It is observed that
the variation of period among different random distributions is not very significant. The average of the five
periods corresponding to the five different random distributions is approximately the same as that obtained
when the infilled panels are distributed in a regular pattern. It can thus be said that randomness in the
distribution of infill in the structure does not influence the building period significantly.

Full-size image (33K)

Fig. 4. Period vs. type of infill distribution.

Fig. 5 shows the change in building period with the increase of number of infilled panels. We observe that
the code equations do not show any variation in period since these do not take into account the effect of
infill in calculating the building period. For conventional analysis, the period gets longer with increase of
infill due to added mass from the infill. When infill is incorporated in the FE analysis, the period becomes
shortened with the increase in number of infilled panels due to added stiffness from the equivalent
diagonal struts. It can be observed that for about 50% of infilled panel, the period obtained from numerical
analysis is approximately the same as that predicted by the code (UBC, BNBC). For 20% infill the period
is about 1.25 times the code prediction while for 80% infill the period is about 0.88 times that predicted by
the code.

Full-size image (33K)

Fig. 5. Period vs. amount of infilled panels.

7. A rationale for estimating period

7.1. The rationale


Based on the preceding discussion, a rationale for estimating the period of an RC framed building having
infilled panels can be suggested. Among the different code equations, the BNBC (or UBC) equation for
determining the period has been found to be closest to the numerical results with infill. Although the code
equations are reasonably accurate, further refinement can be made so that the period is estimated more
accurately. This shall result in more accurate determination of base shear which is very important in
earthquake resistant design of buildings. Here we consider the BNBC [5] equation (Eq. (6)) as our
reference code based on which further refinement in determining the period is suggested. The suggestion
made in the following paragraph of course applies only to regular frame structures within the range of
parameters studied and thus should be applied carefully if applied to any real building design.
From the study presented here we can identify the following three parameters which influences the period:
(a) span length, (b) number of spans and (c) amount of infilled panels. Considering these parameters the
period can be refined as

(11)
T=α1α2α3Cth3/4.
Here, α1 is the modification factor for span length, α2 is the modification factor for number of spans
and α3 is the modification factor for amount of infill. The magnitudes of these factors are suggested in the
followingTable 2a, Table 2b and Table 2c.
Table 2a.

Modification factor for span length

Span length (m) Factor α1

4 0.87

5 0.93
Span length (m) Factor α1

6 1.00

7 1.07

8 1.16

Table 2b.

Modification factor for number of spans

Numer of spans Factor α2

2 1.2

3 1.1

4 and above 1.0

Table 2c.

Modification factor for % amount of infill panels

Amount of infilled panels (%) Factor α3

20 1.20

40 1.08

50 1.00

60 0.91

80 0.82

7.2. Effect of the proposed rationale on base shear


The BNBC equation for calculating base shear by the equivalent static force method is ,
2/3
whereV is the base shear, Z is the zone coefficient, I is the importance factor, C=1.25S/T ,W is the dead
weight of building and R is the response modification factor. In the definition of C, S is the site coefficient
and T is the period. Now, if the period given by Eq. (6) is denoted as and the period given by the
proposed Eq. (11) as and the corresponding base shears as and respectively, then we can
have
(12)

Let consider an example building having 7 floors, a panel span of 7 m and 80% infilled panels, where
other parameters have their reference value as shown in Table 1. For this building, the period given by the
code equation is and the period calculated by modal analysis is 0.64 s. The modification
factors as obtained from Table 2a, Table 2b and Table 2c are α1=1.1, α2=1.0 and α3=0.87. Thus can be
calculated as 0.7 s. Thus can be calculated to be 1.1. Thus it can be said that the base shear is
increased by about 10% for this building, which can be said to be reasonable since we have assumed a
higher amount of infill that has resulted in a stiffer building. From safety requirements such an increase is
significant and must be accommodated in the reinforcement design of the building.
Let us consider another building with panel span of 8 m with 30% of infilled panels, where other
parameters are at their reference value. For this building, the period given by the code equation
is and the period calculated by modal analysis is 1.34 s. The modification factors as
obtained from Table 2a, Table 2band Table 2c are α1=1.16, α2=1.0 and α3=1.14 (by interpolation).
Thus can be calculated as 1.38 s. Thus can be calculated to be 0.83. This means that the
building design can be significantly economized if this 17% reduction in base shear is accounted for.

8. Conclusion
A computational investigation has been performed on the fundamental natural period of vibration of RC
framed buildings having infills. Some sensitivity analysis has been performed in the present study which
shows that, in presence of infill, the beam and column stiffness have negligible effect on the period. The
span length of the panel, number of spans in the direction of motion and amount of infill in the structure
are some important parameters in influencing the period which the code equations do not take into
account. This study also shows that randomness in the distribution of infill does not have a significant
effect on the period; instead, it is the total amount of infill that matters.
While sophisticated computer packages are available to model and analyze building structures in a
comprehensive way, code specifications still advocate the use of empirical equations for estimating the
fundamental period of buildings. The approximations made in the idealization of the structure during the
generation of the computer-based FE model are basically the source of error in predicting the period
though the same idealization is generally thought to be comprehensive enough for strength design under
static design loads. In this paper it has been shown by various sensitivity studies that when the effect of
infill is considered in the FE modeling, rational analyses do give reliable results and agree reasonably well
with the code equations. Such modeling techniques may be applied for dynamic time–history analysis or
spectral response analysis of tall buildings where higher modes of vibration are also active and in which
cases conventional analysis without infill may not accurately reflect the dynamic behavior.
Based on the findings of the study some refinement in the code equation to determine the building period
more accurately has been suggested. The suggested modification takes into account the amount of
infilled panel in the building, panel span length etc., which are otherwise not considered by the prevailing
code equations. However, it must be kept in mind that the study presented in this paper is of limited nature
and thus the suggestion applies only to regular frame structures having different building parameters
within the range of the present study. In order to make a generalized suggestion regarding the
determination of building period considering the effect of infill, more study needs to be carried out.

References
[1] K.M. Amanat and E. Hoque, A reappraisal of time period formulas of design codes for framed
reinforced concrete buildings. In: B.H.V. Topping and Z. Bittnar, Editors, Proceedings of the sixth
international conference on computational structures technology, Civil-Comp Press, UK (2002).

[2] R.K. Goel and A.K. Chopra, Period formulas for moment-resisting frame buildings, Journal of
Structural Engineering, ASCE 123 (1997) (11).

[3] R.K. Goel and A.K. Chopra, Period formulas for concrete shear wall buildings, Journal of Structural
Engineering, ASCE 124 (1998) (4).

[4] Hossain M.Z.. Influence of structure parameters on period of frame structures for earthquake resistant
design. MSc thesis. Dhaka (Bangladesh): BUET; 1997.

[5] Housing and Building Research Institute and Bangladesh Standards and Testing Institution.
Bangladesh National Building Code (BNBC). Dhaka, Bangladesh; 1993.

[6] Indian Standards Institution (IS). Criteria for earthquake resistant design of structures. India; 1984.

[7] International Conference of Building Officials. Uniform building code. California: Willier; 1997.

[8] Q. Li, H. Cao and Gi. Li, Analysis of free vibrations of tall buildings, Journal of Engineering
Mechanics 120(1994) (9).

[9] P.B. Lourenco, R. de Brost and J.G. Rots, A plane stress softening plasticity model for orthotropic
materials, International Journal for Numerical Methods in Engineering 40 (1997).

[10] A. Madan, A.M. Reinhorn, J.B. Mander and R.E. Valles, Modeling of masonry infill panels for
structural analysis, Journal of Structural Engineering, ASCE 123 (1997) (10).

[11] Ministry of Construction. The Building Standard Law of Japan (BSLJ). Japan; 1987.

[12] National Research Council. The National Building Code (NBC). Canada; 1995.

[13] M. Papia, Analysis of infilled frames using a coupled finite element and boundary element solution
scheme, International Journal for Numerical Methods in Engineering 26 (1998).

[14] A. Saneinejad and B. Hobbs, Inelastic design of infilled frames, Journal of Structural Engineering,
ASCE121 (1995) (4).

[15] B.S. Smith and E. Crowe, Estimating periods of vibrations of tall buildings, Journal of Structural
Engineering 112 (1986) (5).

Вам также может понравиться