You are on page 1of 10

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/245304622

Analysis of Cracking Effects on Tall Reinforced Concrete Buildings

Article  in  Journal of Structural Engineering · September 2000


DOI: 10.1061/(ASCE)0733-9445(2000)126:9(995)

CITATIONS READS
16 238

3 authors, including:

C.M. Chan Neil Colin Mickleborough


The Hong Kong University of Science and Technology The Hong Kong University of Science and Technology
77 PUBLICATIONS   1,538 CITATIONS    31 PUBLICATIONS   411 CITATIONS   

SEE PROFILE SEE PROFILE

All content following this page was uploaded by C.M. Chan on 29 January 2016.

The user has requested enhancement of the downloaded file.


ANALYSIS OF CRACKING EFFECTS ON TALL REINFORCED
CONCRETE BUILDINGS
By Chun-Man Chan,1 Member, ASCE, Neil C. Mickleborough,2 Fellow, ASCE,
and Feng Ning3

ABSTRACT: The design of tall reinforced concrete buildings must satisfy serviceability criteria for lateral drift.
It is therefore important to accurately assess the lateral deflection of a structure to account for the nonlinear
effect of cracking in concrete. Iterative procedures are necessary for this serviceability analysis of tall reinforced
concrete buildings, because the concrete members that contribute to lateral stiffness have varying degrees of
cracking. Two procedures for the determination of lateral drift in reinforced concrete structures are presented in
this paper. These procedures have been verified from the experimental data of tests on full-size structural
subassemblages. Both procedures initially consider all the concrete members to be uncracked. An initial analysis
then determines the cracked members and their stiffnesses are modified using probability-based effective stiffness
relationships. The redistribution of force and subsequent modifications of the member stiffnesses are applied
iteratively until convergence of the structural response is obtained. These procedures are computationally more
efficient and direct than the general nonlinear finite-element method and are compatible with linear elastic
analysis software that is commonly available in most structural engineering design offices.

INTRODUCTION nadian Standards Association (CSA) 1994] and AS 3600-1994


[Standards Association of Australia (SAA) 1994]. Based on
The response of reinforced concrete members and structural Branson’s effective stiffness model, Polak (1996) and Scanlon
systems to loads has been the subject of much investigation and Murray (1982) developed computational techniques for
since the time reinforced concrete was first used as a dominant the analysis of reinforced concrete slabs. These techniques are
structural material in engineering construction. In the design simple and accurate, and provide very reasonable estimates of
process of tall reinforced concrete buildings, control of lateral slab deflection, but are limited to flexural beams and slab sys-
drift is one of the most important serviceability criteria, not tems.
only to prevent large second-order P-delta effects, but also to Nonlinear finite-element procedures have been developed
avoid the malfunctioning of nonstructural components such as for calculating deflections of reinforced concrete structures.
elevators and doors. Cracking, which is primarily load depen- These procedures can differ in several aspects: the constitutive
dent, reduces the lateral stiffness and substantially increases models adopted, the methods of stiffness evaluation, the iter-
the lateral deflection of a building. To accurately assess the ation techniques, the types of finite elements used, and mesh
lateral deflection, cracked members in these tall buildings need discretization. Several researchers proposed constitutive mod-
to be identified and their effective member stiffnesses deter- els for reinforced concrete following the rules of nonlinear
mined. elasticity and plasticity. These include the softening effects in
In design, the effects of concrete cracking on the stiffness compression and the effects of cracking in tension (Chen 1982;
of the structural frame are generally accounted for by assign- Vecchio and Collins 1986; Massicotte et al. 1990; Task Com-
ing reduced moments of inertia to the members. The gross mittee 1992). There are two approaches with respect to the
moment of inertia of beams is usually reduced to 50% of their types of finite elements and mesh discretization. One is the
uncracked values, whereas the gross moment of inertia of col- microelement approach where the structure is divided into
umns is reduced to 80% (Stafford Smith and Coull 1991). The many small finite elements including 2D and 3D elements
reductions are generally considered constant over the total modeling concrete and bar elements modeling steel, discrete
height of the structures, independent of the history, type, and crack representation. The other method is the macroelement
magnitude of loading, and independent of the reinforcement approach where each finite element represents both concrete
ratios in the members. No consensus can be drawn as to and reinforcement and the local phenomena are incorporated
whether this type of design is conservative or not, because no into a constitutive model that is used to evaluate the stiffness
reliable analysis has been developed to estimate satisfactorily matrix of a macroelement (Chen 1982; Cauvin 1991). For the
the effect of cracking. When the deflection of beams or one- reinforced concrete material, where the stiffness properties
way slabs is required in calculations, an empirical expression vary throughout the section, a layered model has been adopted
for effective moment of inertia developed by Branson (1963) (Vecchio and Emara 1992; Polak and Vecchio 1993) to inte-
is perhaps the most widely used. Various forms of the Branson grate all the layers’ stiffness contributions into the stiffness
formula can be found in the American Concrete Institute (ACI) matrix of a finite element. However most, if not all, of these
Building Code since 1971 (ACI 1995), CAN3-A23.3-M95 [Ca- analyses are expensive and time consuming for large-scale tall
reinforced concrete structures.
1
Asst. Prof., Dept. of Civ. Engrg., Hong Kong Univ. of Sci. and Tech- In this paper, a general probability-based effective stiffness
nol., Hong Kong. model is proposed to determine the relationship between flex-
2
Assoc. Prof., Dept. of Civ. Engrg., Hong Kong Univ. of Sci. and
Technol., Hong Kong. ural stiffness reductions and various moments due to member
3
PhD, Dept. of Civ. Engrg., Hong Kong Univ. of Sci. and Technol., loading. The most significant feature of the proposed model is
Hong Kong. its extensive applicability to members that are subjected to
Note. Associate Editor: David J. Stevens. Discussion open until Feb- various forms of loading. This effective stiffness model can be
ruary 1, 2001. To extend the closing date one month, a written request considered as a general constitutive model at the member
must be filed with the ASCE Manager of Journals. The manuscript for
level. If the force redistribution and ensuing coupled stiffness
this paper was submitted for review and possible publication on Decem-
ber 23, 1998. This paper is part of the Journal of Structural Engineer- reduction among members in structures are properly dealt with
ing, Vol. 126, No. 9, September, 2000. 䉷ASCE, ISSN 0733-9445/00/ by using iterative algorithms, a practical cracking analysis sys-
0009-0995–1003/$8.00 ⫹ $.50 per page. Paper No. 19926. tem can be established. This can be achieved by integrating
JOURNAL OF STRUCTURAL ENGINEERING / SEPTEMBER 2000 / 995
the proposed effective stiffness model and the iterative algo-
rithms with commercial packages of linear finite-element anal-
ysis. The advantage of this system is the ability to explicitly
and quantitatively predict the lateral deflection and stiffness of
tall reinforced concrete building structures under service load
conditions.

PROBABILITY-BASED EFFECTIVE STIFFNESS


MODEL
Fig. 1 shows the effect of cracking of a flexural member.
In the uncracked regions the applied moment M(x) is smaller
than the cracking moment Mcr. This cracking moment can be
determined
FIG. 1. Service Load Moment Diagram; Cracking in Rein-
(␴v ⫹ fr)Iuncr forced Concrete
Mcr = (1)
yt
where fr = flexural tensile strength of concrete = 0.6兹 f ⬘c MPa; The proposed model [(7)] determines the effective stiffness
yt = distance from centroid of gross section to extreme fiber Ie as the ratio of the area of moment diagram segment over
in tension; ␴v = axial compressive stress; and Iuncr = gross which the working moment exceeds the cracking moment Mcr
uncracked moment of inertia and accounts for the contribution to the total area of the moment diagram. The Ie is determined
of reinforcing steel to the stiffness. In the region where the as a function of load, and subsequently the distribution of
applied moment M(x) is larger than or equal to the cracking cracking, represented by the area of the moment diagram,
moment Mcr, flexural tensile cracks propagate in the member whereas in previous practice only the maximum moment was
from the tensile surface toward the neutral axis, thus causing considered. The proposed model was verified by a series of
the moment of inertia of the cracked section to be reduced to experimental tests, which took into consideration different
the value of Icr. The occurrence of cracking at any section is loading configurations (Ning 1998), and by other recent
dependent on the form (both shape and magnitude) of the mo- publications (Al-Zaid et al. 1991). This concept also can be
ment diagram due to external loading. The larger the moment extended from the analysis of the beams and slabs to columns
value at any point, the greater the probability that a crack will and shear walls with dominant flexural cracking behavior
occur. (Mickleborough et al. 1999).
The probability density function determining the occurrence
of the random variable when the moment value is larger than TWO ITERATIVE ANALYTICAL PROCEDURES
certain moment value can be expressed (Mickleborough et al.
Appropriate equations that describe the force-deformation
1999; Ning et al. 1999)
relationships, equilibrium, and deformation compatibility of
M(x) reinforced concrete can be fully satisfied by linear finite-ele-
p (x) = (2) ment methods (LFEM) when the cracking effects are not sig-
S
nificant. Using these methods, internal element forces such as
where M(x) = moment distribution function; and S = total area moment, shear, and axial force can be readily obtained. These
of moment diagram. The probability that the moment value is LFEM analyses have been commonly used in today’s struc-
>Mcr can be obtained by integrating the probability density tural engineering design offices.
function p (x) in the region where M(x) ⱖ Mcr For tall reinforced concrete buildings, lateral drift, which is


significantly affected by cracking effects, becomes one of the
M(x) Scr
P [M(x) ⱖ Mcr] = dx = (3) important design criteria and the cracking effects need to be
M(x)ⱖMcr S S considered in the analyses. The major difficulty of nonlinear
in which Scr = area of moment diagram segment over which crack analyses is that the stiffness of a cracked member varies
the working moment exceeds the cracking moment Mcr. The according to the amount of crack formation occurring in the
probability of having the moment value < Mcr is members. Such stiffness reduction may result in internal force

冕 冕
redistribution, which will further affect the occurrence of
M(x) M(x) cracking. To cope with this difficulty, two iterative procedures
P [M(x) < Mcr] = dx = 1 ⫺ dx
M(x)<Mcr S M(x)ⱖMcr S have been developed, namely, the load incremental procedure
and the direct-effective stiffness procedure. In both procedures
= 1 ⫺ P [M(x) ⱖ Mcr] (4) the cracked members are first identified at a determined service
load level and their element stiffnesses are then reduced ac-
The probability of occurrence of cracked sections Pcr asso- cordingly using this probability-based effective stiffness
ciated with the outcome Icr, the cracked moment of inertia, is model. The effective stiffness model provides the nonlinear
determined by force-deformation relationships for the cracked members. De-
Pcr = P [M(x) ⱖ Mcr] (5) tails of these two iterative procedures are presented as follows.

and the probability of occurrence of uncracked section Puncr Iterative Procedure I—Load Incremental Method
with outcome Iuncr, the gross moment of inertia, is derived by
Procedure I, represented in Fig. 2, provides the history of
Puncr = P [M(x) < Mcr] = 1 ⫺ Pcr (6) nonlinear behavior of reinforced concrete structures due to
Hence, the expected value of the moment of inertia, the so- cracking, by applying the external loads in an incremental
called effective moment of inertia, of the reinforced beam sub- manner. The force redistribution and the cracking moment Mcr,
jected to a certain type of loading is affected by the change of axial force, are updated at every
increment. This in turn determines the stiffness reduction of
Ie = Puncr Iuncr ⫹ Pcr Icr = (1 ⫺ Pcr)Iuncr ⫹ Pcr Icr (7) the reinforced concrete members for the subsequent increment.
996 / JOURNAL OF STRUCTURAL ENGINEERING / SEPTEMBER 2000
accuracy of the calculation. Generally a larger value of ⌬F
may cause a larger error, whereas a smaller ⌬F may result in
smaller errors. A large incremental load step may not detect
the initial cracking sequence of various members that causes
discrepancies at the onset of the cracking. The magnitude of
the errors are dependent on the discrepancies of the actual
effective stiffnesses in the current iteration and the equivalent
effective stiffnesses determined in the previous iteration,
which in turn are dependent on the value of the load incre-
ment ⌬F.
Iterative Procedure II—Direct-Effective Stiffness
Method
Recognizing the fact that the incremental load method may
become cumbersome when a large number of load increments
is necessary, it may be desirable to consider directly the effects
of cracking on reinforced concrete structures under full service
load. To achieve this objective, a direct-iterative technique has
been adopted and is described in Procedure II. A major feature
of Procedure II is that the iterative calculations are conducted
when the structure is subjected to the full, specified service
load.
Once an initial linear finite-element analysis has been con-
ducted under the full service load condition, the cracked mem-
bers can be identified and their effective stiffnesses determined
from their current values of internal moments and forces using
the proposed formulas [(3)–(7)]. Changes in stiffnesses of the
cracked members cause a transfer of some components of the
FIG. 2. Flowchart of Iterative Analytical Procedure I: Load In- internal moments and forces of these members to the other
cremental Method
uncracked members, resulting in the cracking of some of the
otherwise uncracked members. Because Procedure II allows
These coupled variations can be adjusted to trace the physical for changes (either increases or decreases) in the stiffnesses of
propagation of cracking in the reinforced concrete structures members, subsequent reanalysis of the structure is necessary
from iteration to iteration. The steps for Procedure I are de- to account for internal force and moment redistribution. This
scribed as follows: direct-iterative technique is appropriate for tracing the impact
on internal force redistribution from changes in stiffnesses of
1. For n = 1, where n is the incremental number, all the members when the convergence of overall structural response
members of a reinforced concrete structure are consid- and stability of internal element force redistribution is
ered to be initially uncracked. The full lateral service achieved. This method also can simulate the complete inter-
load can be divided into N load increments, such that action process of progressive cracking and internal force re-
each load increment has a value equal to ⌬F. The initial
applied load F (1) = ⌬F.
2. The load F (n) is applied to the structure. The internal
moment M (n) (n)
i , axial force A i , and lateral deflection D
(n)

are computed using linear finite-element software.


3. The element cracking moment (Mcr)i(n) is updated by the
corresponding element axial force A (n) i , represented by
␴v in (1).
4. If M (n)
i ⱖ (Mcr)(n)i , then the ith member is deemed to be
a cracked member; the flexural stiffness (Iuncr)i is then
reduced to the effective flexural stiffness (Ie)i(n) using
(3)–(7).
5. If M (n)
i < (Mcr)(n)
i , there is no cracking in the ith member;
hence, the flexural stiffness maintains the uncracked val-
ues (Iuncr)i.
6. If n > N, the iteration reverts to Step 7; otherwise
F (n⫹1) = F (n) ⫹ ⌬F and n = n ⫹ 1, and the iteration
returns to Step 2 to repeat the incremental load analysis
and to determine the current effective stiffness of the
structure.
7. The iterative procedure is terminated with the final ef-
fective stiffness of the structure and the lateral displace-
ment of the structure determined under the specified lat-
eral load.

A disadvantage of this procedure is that the best value of


the load increment ⌬F is not necessarily easy to determine. FIG. 3. Flowchart of Iterative Analytical Procedure II: Direct-
This increment may become a critical parameter affecting the Effective Stiffness Method

JOURNAL OF STRUCTURAL ENGINEERING / SEPTEMBER 2000 / 997


distribution that occurs in reinforced concrete structures. The Frame Structure and Material Details
details of this direct iterative Procedure II are presented as
follows and the corresponding flowchart is shown in Fig. 3: An experimental program was undertaken to determine the
behavior of a large-scale single-span, two-story reinforced
1. Set n = 1, where n is the iteration number; all members concrete frame, as shown in Fig. 4(a). The frame was fabri-
of a reinforced concrete structure are considered initially cated with a center-to-center span of 3,000 mm. The first-story
to be uncracked under the full service lateral loading height was 1,170 mm (including a steel pin joint height of 170
condition. mm), and the second-story height was 2,000 mm—an overall
2. The applied load is set to the full service lateral load. structural height of 3,670 mm. The section dimensions of the
The internal element moment M (n) i and force A (n)
i and lat- columns were 250-mm wide by 375-mm deep, and the section
eral displacement D(n) are determined using linear finite- dimensions of the beams were 250-mm wide by 350-mm deep.
element software. The frame was connected to the rigid strong floor of the lab-
3. The element cracking moment (Mcr)i(n) is updated from oratory by two steel pin joints. It is assumed in rigid frame
the corresponding element axial force A (n) i according analysis that there are pin joint connections located at an ap-
to (1). proximate inflexion point of the columns of every story. Hence
4. If M (n)
i ⱖ (Mcr)(n)
i , then the ith member is deemed to be
this frame structure can be regarded as a subassemblage in a
cracked and its flexural stiffness that initially equaled tall reinforced concrete rigid frame building.
(Iuncr)i is reduced to the effective flexural stiffness (Ie)i(n)
using (3)–(7).
5. If M (n)
i < (Mcr)(n)
i , cracking has not occurred in the ith
member; hence, the flexural stiffness remains at the un-
cracked value (Iuncr)i.
6. Incrementally n = n ⫹ 1; all current values of stiffness
for every member are updated and the structure reana-
lyzed under the full specified load to update the element
moment M (n) i and force A (n)
i and the lateral displacement
(n)
D .
7. The current lateral displacement of the structure D(n) is
compared with the immediately preceding iteration. If
兩D(n) ⫺ D(n⫺1)兩 ⱕ ε, where ε is a small convergence cri-
terion, then the iterative procedure is terminated with the
final response of the structure, taking into account the
effects of cracking under the specified full service load.
Otherwise, the iteration proceeds to Step 3 to repeat the
next cycle.

Procedure II allows for the continuous modification of ele-


ment stiffnesses that account for the effects of cracking until
convergence of the structural response is achieved. This iter-
ative procedure determines automatically the response of re-
inforced concrete structures, taking into account concrete
cracking; there is, however, no guarantee that convergence of
the solution will always be achieved. Overreductions of stiff-
nesses in some members at one iteration may lead to smaller
redistributions of internal forces for these members, which in
turn would result in excessive increases in the stiffnesses of
these members in the subsequent iteration. Increases of stiff-
nesses in turn attracts the transfer of more internal forces to
these members, thus causing overreductions to occur again. If
these fluctuations in stiffnesses do not diminish rapidly with
increasing iterations, this procedure may incur significant com-
putational time.

Hybrid Method—Combination of Procedures I and II FIG. 4. (a) Layout and Dimensions of Test Frame; (b) Dimen-
sion of Frame in Previous Test (Emara 1990; Vecchio and Emara
Some errors in Procedure I caused by choosing an unac- 1992)
ceptable load increment ⌬F could be avoided by combining
Procedure II with Procedure I, because Procedure II can de- TABLE 1. Detail Properties, Dimensions, and Reinforcement
termine lateral deflection of reinforced concrete structures at of Frame Specimen
any lateral loading level. The errors found in the load-displace- Iuncr Icr
ment path of Procedure I can be eliminated by corrections of b h dc ␳t ␳c 106 106
Procedure II at each incremental load level. Hence the load- Member (mm) (mm) (mm) (%) (%) (mm4) (mm4)
deformation history can be obtained even with large values of (1) (2) (3) (4) (5) (6) (7) (8)
load increment if Procedures I and II are combined.
B1-6 250 350 50 1.26 1.26 1254 585
B2-5 250 350 50 1.26 1.26 1254 585
EXPERIMENTAL VERIFICATION C1-2 250 375 40 1.88 1.13 1601 804
The two proposed iterative procedures have been verified C2-3 250 375 40 1.88 1.13 1601 804
C4-5 250 375 40 1.88 1.13 1601 804
experimentally through a series of the reinforced concrete C5-6 250 375 40 1.88 1.13 1601 804
frame tests.
998 / JOURNAL OF STRUCTURAL ENGINEERING / SEPTEMBER 2000
els. A LFEM, not considering cracking, predicts 46 and 47%,
respectively, of the test values of the lateral deflections of the
second and first stories at the 70% ultimate lateral load level
(i.e., 140 kN). Procedure I gives predictions of 83 and 85%,
respectively, of the experimental values. If 50% ultimate load
is considered, the predictions of the linear elastic method give
55 and 58%, respectively, of the experimental deflections;
however, the predictions of Procedure I, which considers the
cracking effects, give 92 and 95%, respectively, of the test
results. Even within the serviceability loading range, the basic
assumption of linear elastic behavior is not reliable and quite
inaccurate because the first stiffness reduction caused by initial
cracking of structural members usually appears at a very low
lateral load level. In contrast, the proposed Procedure I ac-
counts very well for the cracking effects with stiffness reduc-
FIG. 5. Detail Setup of Frame Test tion. This method not only predicts lateral deflection to a value
of load equal to approximately half of the ultimate capacity
Table 1 presents detailed information of the beams and col- with good accuracy, but it also can give an estimation of
umns of the structure. The columns of the frame were rein- behavior at 70% ultimate load with an acceptable degree of
forced with three No. 20 deformed bars in each main face and accuracy. Procedure I is, however, most applicable in the
two No. 20 bars in the side face at the approximate position serviceability loading range; beyond this range large discrep-
of the neutral axis. No. 10 closed stirrups at 100-mm spacing ancies will occur, as shown in Fig. 6. This is consistent with
were used as shear reinforcement. The beams were designed the test observation that the plastic hinges were forming at the
with an arrangement of three No. 20 deformed bars as bottom ends of the beams and cracking occurred at the joints of the
reinforcement, three No. 20 bars as top reinforcement, and beam column.
similar shear reinforcement. The concrete was tested in accor- The major advantage of Procedure I is that the historical
dance with ASTM C464-94 and had a compressive strength variation in the flexural stiffness reduction for every member
of 29 MPa and a Young’s modulus of 15,680 MPa. The ma- in the reinforced concrete structure can be shown explicitly.
terial properties of the reinforcing steel were a yield stress of This feature can provide design engineers with significant in-
460 MPa and a modulus of elasticity of 200,000 MPa. formation on the consequence of cracking in members and also

Load and Instrumentation Details of Frame Test


The testing setup involved the application of a total axial
vertical load of 200 kN to each column and maintenance of
this load in a force-controlled mode throughout the test by a
device called a gravity loading simulator. This device approx-
imates true gravity load (Yarimci et al. 1967). The load line
of this system is such that when the sidesway is between 0
and 200 mm, the load remains vertical and moves with the
lateral movement of the structure. Because the actual lateral
deflections are generally <100 mm, no horizontal component
of vertical load was induced. Lateral load was then monoton-
ically applied by an hydraulic actuator in a stroke-controlled
mode at a rate of 0.02 kN/s until the ultimate capacity of the
frame was achieved.
To monitor the behavior during testing, the frame model
was extensively instrumented. Twelve displacement transduc-
ers (LVDTs), positioned at key points around the frame, were FIG. 6. Comparison among Analytical Results of Procedure I
and Elastic Method and Experimental Results
used to monitor both lateral deflection and vertical deforma-
tion. Two LVDTs, positioned at half the height of the frame,
were used to monitor the out-of-plane movement. A load cell,
together with the lateral actuator, measured the applied lateral
load, and two load cells monitored the vertical loads. The res-
olution of the LVDTs and the loading system were 0.01 mm
and 0.01 kN, respectively. All experimental data from the ap-
plied loading, horizontal deformations, and vertical deforma-
tions were recorded by a DATA LOG system at regular inter-
vals of 5 s. Fig. 5 shows the detailed setup of this reinforced
concrete frame experiment.

ANALYTICAL INVESTIGATION OF BEHAVIOR OF


FRAME
Procedure I—Incremental Load Analysis
To determine the applicability of Procedure I, Fig. 6 com-
pares the prediction with both the elastic prediction and the FIG. 7. Flexural Stiffness Reduction of Structural Members in
experimental load deflections at the first- and second-story lev- Frame with Increasing Lateral Load

JOURNAL OF STRUCTURAL ENGINEERING / SEPTEMBER 2000 / 999


the initial cracking sequence of various members and results
in more discrepancies at the onset of the cracking. If the over-
all deflection is of interest, the prediction with a 20-kN loading
increment, which equals 10% of the ultimate load, becomes
relatively more efficient and sufficiently accurate after consid-
ering the factors of computational time and optimum accuracy.

Procedure II—Direct-Effective Stiffness Analysis


Procedure II has been applied to calculate the lateral de-
flection of the frame at two lateral load levels, 100 and 140

FIG. 8. Numerical Comparison among Various Loading Steps


in Procedure I

minimize the uncertainty of stiffnesses in the lateral load re-


sisting systems. Fig. 7 illustrates the quantitative analytical re-
sults of the flexural stiffness reductions of the six members
involved in testing the frame. This figure indicates that the
beams at the first and second stories crack first at lateral load
levels of 25 and 35 kN, respectively, and their flexural stiff-
nesses will reduce quite rapidly after the initial cracking. At
the lateral load of 70 kN, both columns on the lateral loading
side, C4-5 and C5-6, start to crack, and the flexural stiffness FIG. 9. Convergent Process of Lateral Deflection with the It-
reduction ratios of the beams at the first and second stories eration Numbers at 100-kN Loading Stage (50% Ultimate Load)
would already have been reduced to 0.54 and 0.60, respec-
tively. A significant proportion of the load carried by the beams
would have been transferred to columns C4-5 and C5-6 and,
as a consequence, cracking occurs in those columns. The two
columns on the opposite side of the lateral loading, C1-2 and
C2-3, crack at very different lateral load levels. The upper
story column cracks at the 75-kN load level, and the lower
story column cracks at the 95-kN load. This occurs because
the axial forces in these columns are larger than those of the
columns on the loading side and maximum axial force occurs
in the column at the lower story on the opposite loading side.
These analytical results are consistent with the test observation
of the cracking history. The initial cracking detected by ob-
servation occurred at a slightly larger value than the analytical
values. From Fig. 7 it is shown that at 50% of the ultimate
lateral load, the beams of the first and second stories have 50
and 54%, respectively, of their gross moment of inertia and
the two columns at the second story have 80 and 84% of their FIG. 10. Convergent Process of Flexural Stiffness of Every
uncracked values. The two columns at the first story have 76 Member with Iteration Numbers at 100-kN Loading Stage (50%
and 96% of the gross moment of inertia. However, when the Ultimate Load)
lateral load reaches 70% of the ultimate load (i.e., at the load
of 140 kN), the stiffnesses of the four columns further reduce
to 74, 64, 60, and 76%, respectively, whereas the two beams
maintain 48–50% of the uncracked values. This case shows
that, although the effect of cracking is usually considered by
assigning a 50% reduced moment of inertia to all the beams
and an 80% reduced moment of inertia to all the columns,
such an assignment will not guarantee a conservative predic-
tion of the lateral deflection. Moreover, the stiffness reductions
of columns are dependent on the values of the vertical loads.
The effects of various loading increments are shown in Fig.
8. Three kinds of loading increment, 5, 20, and 40 kN, equal
to approximately 2.5, 10, and 20%, respectively, of the ulti-
mate load, were selected in the investigation. It is noted that
the prediction with the 5-kN loading increment gives the most
accurate values among the three cases, and the prediction with
the 40-kN loading increment appears to be the least accurate,
especially at the moment when the initial cracking occurs. The FIG. 11. Lateral Deflection Variation with Iteration Numbers at
reason is that a larger incremental load step may not detect 140 kN (70% Ultimate Load)

1000 / JOURNAL OF STRUCTURAL ENGINEERING / SEPTEMBER 2000


FIG. 12. Flexural Stiffness Variation of Members with Iteration FIG. 13. Comparison between Test and Analytical Results
Numbers at 140 kN (70% Ultimate Load) Combination of Procedures I and II (Values in Bracket Stand for
Iteration Numbers)
TABLE 2. Deflections and Stiffnesses Comparison between
Procedure I and Procedure II der any specified service load level and can minimize the un-
100 kN 140 kN certainty in the prediction of lateral stiffness due to cracking for
Load the practical design of tall reinforced concrete buildings.
method Procedure I Procedure II Procedure I Procedure II
(1) (2) (3) (4) (5)
Combination of Procedures I and II
D1st (mm) 7.05 7.16 10.73 10.80
D2nd (mm) 17.05 17.26 25.64 25.79 The possible errors in using large loading increments in Pro-
B1-6 0.54 0.53 0.50 0.50 cedure I can be avoided by combining the two procedures. As
B2-5 0.50 0.50 0.48 0.48
C1-2 0.84 0.82 0.74 0.73
shown in Fig. 13, Procedure I with a 40-kN increment load
C2-3 0.96 0.92 0.76 0.75 was first adopted in the analysis. The results obtained by Pro-
C4-5 0.76 0.73 0.60 0.59 cedure I at every load step form the initial values of Procedure
C5-6 0.80 0.78 0.67 0.66 II, which terminates when the convergence of the overall re-
sponses at the corresponding load levels is achieved. The val-
ues in the brackets are the iteration numbers needed to con-
kN, which are approximately equal to 50 and 70% of the ul- verge from the results of Procedure I to those of Procedure II.
timate load, respectively. Figs. 9 and 10 present the process As previously mentioned, large discrepancies may occur es-
of convergence by variations of deflection and stiffness with pecially at the onset of initial cracking when a 40-kN loading
the number of iterations at the 100-kN load level. Either the increment is chosen. These discrepancies are largely elimi-
deflection or the stiffness converges toward a solution after nated in Procedure II, as illustrated by a reduction from 36 to
five iterations, and agreements between analytical deflection 2% at the load of 80 kN, corresponding to the point where the
of Procedure II and test results at the first- and second-story maximum error appeared previously in Procedure I. This tech-
levels are good with a variation of only 0.5 and 3.6%, respec- nique of a combination of Procedure I and Procedure II to the
tively. Similar results are presented in Figs. 11 and 12, which analysis of structures can satisfy the demand of either load-
give the convergence process at 140-kN load level. There is a deflection history or deflection at a particular service load level
larger discrepancy between the analytical and test deflection with greater computational efficiency.
of 14% occurring at both levels. As compared with the results
of Procedure I in Table 2 at the 100- and 140-kN load levels, Comparison of Results with Other Test Investigations
the discrepancies of the analytical deflections between Proce-
dure II and Procedure I are on the order of 1–1.5%. Procedure Further verification of these procedures has been conducted
II provides equally good and consistent predictions of effective by comparison with results of a previous experimental pro-
member stiffness and lateral deflections. Procedure II is also gram (Emara 1990; Vecchio and Emara 1992). As shown in
computationally more efficient and direct than Procedure I. Fig. 4(b), this frame was designed with a center-to-center span
Because the discrepancy (14%) between the test and ana- of 3,500 mm, a story height of 2,000 mm, and an overall
lytical lateral deflections at 140-kN load level is greater than height of 4,600 mm and was built integrally with a large, heav-
the 100-kN load level (4%), Procedure II also should be lim- ily reinforced concrete base. All frame members were 300-mm
ited to service load conditions. For indeterminate structures, wide by 400-mm deep and reinforced with four No. 20 de-
the internal force distribution is largely dependent on the rel- formed bars as both tensile and compressive steel. The testing
ative stiffness ratios between the members. The relative stiff- sequence involved initial application of a total axial load of
ness ratios, however, vary due to the initiation and propagation 700 kN to each column, which was maintained throughout the
of cracking that occurs in some members and leads to an in- test. The lateral load was then monotonically applied until the
stantaneous redistribution of the internal forces and moments ultimate capacity of the frame was achieved.
in the overall structure. This phenomenon can be noted in the Using load increments of 5 and 20 kN, two theoretical
process of effective member stiffness convergence in Fig. 10. curves are presented in Fig. 14. The numerical results agree
For instance, the numerical values of flexural stiffnesses of C1- well with the test results for applied loads within the 210-kN
2, C2-3, and C5-6 fluctuate from the second to fourth iteration. load level (i.e., 63% of ultimate load), with maximum dis-
This quantitative result of the proposed methods has signifi- crepancies of 12 and 14% with 5- and 20-kN load increments,
cance in the practical design of tall reinforced concrete build- respectively. Fig. 15 presents a comparison of the top deflec-
ings. Both analytical Procedures I and II can estimate accurately tion using Procedure I with 5-kN loading increment, the lay-
the extent of cracking of every reinforced concrete member un- ered method proposed by Vecchio, and the actual test results.
JOURNAL OF STRUCTURAL ENGINEERING / SEPTEMBER 2000 / 1001
discrepancies may be expected in the prediction of lateral de-
flection if the load is beyond the serviceability range.
Fig. 16 shows quantitatively the cracking sequence and the
flexural stiffness reduction of each member with respect to the
lateral applied load. This figure also shows that the two beams
crack first at the 45- and 50-kN lateral load levels, the two
columns at the first story have initial cracks at 90- and 95-kN
load levels, and the two columns at the second story crack
finally at 160- and 165-kN load levels. When the stiffness of
the two beams are reduced to <50% of their uncracked stiff-
ness, the two first-story columns have reduced to 62–67% of
their uncracked stiffness.

CONCLUSIONS

• A probability-based effective stiffness model, which ac-


counts for cracking in the reinforced concrete flexural
FIG. 14. Verification of Procedure I by Previous Test Results members, was applied to the analyses of reinforced con-
(Emara 1990; Vecchio and Emara 1992) crete structures.
• Two iterative analytical procedures were developed to de-
termine the nonlinear lateral deflection performance of re-
inforced concrete structures under service load conditions.
Although the incremental load method of Procedure I is
able to track the history of cracking in members, the di-
rect-iterative approach of Procedure II can calculate the
lateral deflection of reinforced concrete structures at any
specified level of lateral loads. The main advantage of
these iterative methods is being computationally more ef-
ficient and direct than any general nonlinear finite-element
method.
• A comparison of the numerical results of the proposed
analytical procedures and the test data of a full-scale sin-
gle-span, two-story reinforced concrete frame shows good
agreement within the service loading conditions.
• Further verifications using other frame test results indicate
that the proposed procedures can provide an accurate and
efficient prediction of nonlinear behavior of reinforced
FIG. 15. Comparison between Procedure I with Layered concrete frame buildings due to cracking under service
Model and Test Results (Emara 1990; Vecchio and Emara 1992) loads.
• In the range of service loading conditions, the flexural
stiffness reduction due to cracking is a dominant com-
ponent resulting in the nonlinear load-deformation re-
sponse of the reinforced concrete structures. Beyond the
service load range material nonlinearity may become
more significant in the behavior of reinforced concrete
structures.
• The analytical results and experimental observations in-
dicate that the cracks occurring in the beams are respon-
sible for the initial reduction of overall lateral stiffness at
a relatively low lateral load level, generally at 15% of the
ultimate load.
• The case studies indicate that the common practice of as-
signing a 50% reduced moment of inertia to all the beams
and an 80% reduced moment of inertia to all the columns
to account for cracking effects in design procedures does
not always guarantee a conservative prediction of the lat-
FIG. 16. Stiffness Reduction of Members versus Lateral Load eral deflection.
(Emara 1990; Vecchio and Emara 1992)
ACKNOWLEDGMENTS
This work was supported by the Research Grants Council of Hong
The proposed method shows an improved prediction over Vec- Kong under project No. HKUST-543/94E and was based upon the re-
chio and Emara’s layer model (Vecchio and Emara 1992) for search conducted by Feng Ning under the supervision of Neil C. Mick-
service loads up to approximately 63% of the ultimate load leborough and Chun-Man Chan for the degree of doctor of philosophy
(i.e., at the value of approximately 210 kN). When the lateral in the Department of Civil and Structural Engineering at the Hong Kong
load is beyond the serviceability load level, the difference be- University of Science and Technology, Hong Kong.
tween the analytical and experimental results becomes more
significant. The proposed iterative Procedures I and II are, APPENDIX I. REFERENCES
therefore, applicable to predict the lateral deflections of a re- American Concrete Institute (ACI) (1995). ‘‘Building code requirements
inforced concrete building under service conditions, but large for reinforced concrete.’’ ACI 318-95, Farmington Hills, Mich.

1002 / JOURNAL OF STRUCTURAL ENGINEERING / SEPTEMBER 2000


Branson, D. E. (1963). ‘‘Instantaneous and time dependent deflections of Yarimci, E., Yura, J. A., and Lu, L. W. (1967). ‘‘Techniques for testing
simple and continuous reinforced concrete beams.’’ Ala. Hwy. Res. Rep. structures permitted to sway.’’ Experimental Mech., 7(8).
No. 7, Bureau of Public Roads. Al-Zaid, R. Z., Al-Shaikh, A. H., and Abu-hussein, M. M. (1991). ‘‘Effect
Canadian Standards Association (CSA). (1994). ‘‘Design of concrete of loading type on the effective moment of inertia of reinforced con-
structures for buildings.’’ CAN3-A23.3-M94, Toronto. crete beams.’’ ACI Struct. J., 88(2), 184–190.
Cauvin, A. (1991). ‘‘Influence of tension stiffening on behavior or struc-
tures.’’ Proc., IABSE Colloquium, International Association of Bridge APPENDIX II. NOTATION
and Structural Engineers, Zurich, 153–158.
Chen, W. F. (1982). Plasticity in reinforced concrete, McGraw-Hill, New
The following symbols are used in this paper:
York. A (n) = axial force of ith member in nth iteration;
i
Emara, M. B. (1990). ‘‘Shear deformations reinforced concrete frames.’’
Master’s degree thesis, University of Toronto, Toronto.
D(n) = lateral deflection of reinforced concrete structure at
Massicotte, B., Elwi, A. E., and MacGregor, J. G. (1990). ‘‘Tension- nth iteration;
stiffening model for planar reinforced concrete members.’’ J. Struct. F (n) = applied force at nth iteration;
Engrg., ASCE, 116(11), 3039–3058. f ⬘c = ultimate strength of concrete;
Mickleborough, N. C., Ning, F., and Chan, C. M. (1999). ‘‘Prediction of fr = modulus of rupture of concrete;
the stiffness of reinforced concrete shear walls under service loads.’’ (Ie)(n)
i = effective moment of inertia of ith member in nth it-
ACI Struct. J., 96(6), 1018–1026. eration;
Ning, F. (1998). ‘‘Lateral stiffness characteristics of tall reinforced con- Iuncr)(n)
i = uncracked moment of inertia of ith member in nth
crete buildings under service loads.’’ PhD dissertation, Dept. of Civ. iteration;
Engrg., Hong Kong University of Science and Technology, Hong Kong. Icr = cracked moment of inertia;
Ning, F., Mickleborough, N. C., and Chan, C. M. (1999). ‘‘The effective Ie = effective moment of inertia;
stiffness of reinforced concrete flexural members under service load
Ig, Iuncr = uncracked moment of inertia;
conditions.’’ Australian J. Struct. Engrg., 2(2&3), 135–144.
Polak, M. A. (1996). ‘‘Effective stiffness model for reinforced concrete M, M(x) = moment distribution on member;
slabs.’’ J. Struct. Engrg., ASCE, 122(9), 1025–1030. M (n)
i = moment distribution of ith member in nth iteration;
Polak, M. A., and Vecchio, F. J. (1993). ‘‘Nonlinear analysis of reinforced (Mcr)(n)
i = corresponding cracking moment of ith member in nth
concrete shells.’’ J. Struct. Engrg., ASCE, 119(12), 3439–3462. iteration;
Standards Association of Australia (SAA). (1994). ‘‘SAA concrete struc- Mcr = cracking moment in concrete flexural member;
tures code.’’ AS 3600-1994, Sydney, Australia. N = total number of load increments;
Scanlon, A., and Murray, D. W. (1982). ‘‘Practical calculations of two- Pcr = probability of occurrence of cracked sections associ-
way slab deflections.’’ Concrete Int., November, 43–50. ated with outcome Icr;
Stafford Smith B., and Coull, A. (1991). Tall building structures: Analysis Puncr = probability of occurrence of uncracked section with
and design, Wiley, New York. outcome Iuncr;
Task Committee on Finite Element Analysis of Reinforced Concrete
p (x) = probability density function;
Structures of the Structural Division Committee on Concrete and Ma-
sonry Structures. (1982). State-of-the-Art Rep. on Finite Element Anal.
S = total area of moment diagram;
of Reinforced Concrete, ASCE, New York. Scr = area of moment diagram segment over which work-
Vecchio, F. J., and Collins, M. P. (1986). ‘‘The modified compression ing moment exceeds cracked moment Mcr;
field theory for reinforced concrete elements subjected to shear.’’ ACI yt = distance from centroid of gross section to extreme
J., 83(6), 219–231. fiber in tension;
Vecchio, F. J., and Emara, M. B. (1992). ‘‘Shear deformation in reinforced ⌬F = load increment; and
concrete frames.’’ ACI Struct. J., 89(1), 46–56. ␴v = axial compressive stress.

JOURNAL OF STRUCTURAL ENGINEERING / SEPTEMBER 2000 / 1003

View publication stats

You might also like