Professional Documents
Culture Documents
discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/238741253
CITATIONS READS
8 1,389
3 authors, including:
All content following this page was uploaded by Sang Whan Han on 06 October 2015.
The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original document
and are linked to publications on ResearchGate, letting you access and read them immediately.
Magazine of Concrete Research
doi: 10.1680/macr.2008.00081
Hanyang University
The equivalent frame method (EFM) was developed for two-way slabs as a simple approximate method. Current
design codes (ACI 318-05, Eurocode 2 and BS 8110) permit the EFM for the analysis of two-way slab systems
under gravity loads, as well as lateral loads such as seismic loads. The EFM, however, was developed based on a
semi-empirical formula, derived from the behaviour of two-way slab systems under gravity loads. This study shows
that the EFM is not appropriate in accurately predicting the response of two-way slab systems under lateral loads.
To improve the EFM under lateral loads, this study proposes a modified equivalent frame method (MEFM). To
verify the MEFM, an idealised two-storey flat plate frame is considered and analysed by both the MEFM and
elastic finite-element (FE) analysis for estimating slab moments and lateral drifts. Analysis results obtained from
the conventional EFM and effective beam width method (EBWM) are also compared. The lateral stiffness, based on
the proposed MEFM, was also compared with experimental data.
where K ec is the flexural stiffness of equivalent col- From Equation 5, the flexibility of the equivalent slab
umns, K c is the flexural stiffness of columns and K t is elements can be expressed as follows
the stiffness of the torsional elements 1 1 1
Stiffness of the torsional elements (K t ) is specified (6)
K es K s K t
in ACI 318-05, as follows
X 9Ec C where K es is the stiffness of the effective slab element,
Kt (3)
L2 1 c2 =L2 3 K s is the stiffness of the slab element ( 4Ec I s =L1 ), K t
X is the stiffness of the torsional element, Ec is Youngs
x x3 y
C 1 0:63 (4) modulus of elasticity of concrete, I s is the moment of
y 3 inertia of slab sections, L1 is the slab span length in the
loading direction.
where Ec is Youngs modulus of elasticity for concrete,
C is the torsional constant, L2 is the width of the slab,
c2 is the width of the column parallel to L2 and x and y
are the shorter and longer side of the rectangular cross- Stiffness of torsional element, K t
section of a torsional element respectively.
To determine the stiffness of torsional elements, the
The stiffness of torsional elements in Equation 3 is
distribution of the twisting moment per unit length
derived, assuming that the distribution of the twisting
along the torsional elements should be defined. As
moment per unit length along the torsional element is
mentioned earlier, triangular distribution is assumed for
triangular.4,13 This assumption may only be valid for
the ACIEFM. However, it seems that the stiffness of
slab systems under gravity loads.
torsional elements in two-way slab systems under lat-
eral loads is not the same as the stiffness under gravity
loads because of the difference in the load-transfer
Modified equivalent frame method mechanism.
(MEFM) To show the difference in stiffness of torsional ele-
ments according to types of loads, FE analyses are
The load-transfer mechanism under gravity loads is carried out for a two-storey building with flat plate
different from that under lateral loads. When two-way slabs, as shown in Figure 2. The commercial software
slab systems are subjected to lateral loading, columns SAP200015 was used for the FE analyses. Poissons
displace first, and the forces in columns are transferred ratio () and Youngs modulus (E) are assumed to be 0
to the torsional elements. Subsequently, the torsional and 25 300 MPa, respectively. This study assumes that
elements transfer the forces to the slabs. Thus, under columns in the first storey are clamped to the base.
lateral loads, the flexural deformation of columns is Slabs are modelled using plate element, and columns
restrained by both the stiffness of torsional elements using beam element, which are provided in the
and the flexural stiffness of the slab; it is more appro- SAP2000.15 Plate elements adjacent to the columns are
priate to use equivalent slab elements under lateral rigidly connected to the column elements as used in the
loads rather than to use equivalent column elements prior researches.5,6
(see Figures 1a and 1b). It is noted that, under gravity To estimate the rotational angle along the torsional
loads, slabs transfer the forces to the torsional elements elements, a gravity load of 7.8 kPa is applied to the
first, then the forces of the torsional elements are slabs without lateral loads. Subsequently, removing the
transferred to the columns. gravity loads, lateral loads are applied to the frame, as
This study develops the MEFM for two-way slab shown in Figure 2(b).
systems under lateral loads, which consist of equivalent Figure 3 shows rotational angle along the torsional
slab elements and column elements, shown in Figure elements under gravity loads and lateral loads, and
1b, instead of the equivalent column elements and slab which denotes the rotation angle normalised by the
elements used in the ACIEFM, shown in Figure 1a. maximum rotation angle max . As expected, the distri-
bution pattern of the rotational angles shown in Figure
3a and 3b are different. In the slab under gravity loads,
the rotation angle decreases on approaching the column
Stiffness of equivalent slab element
face (Figure 3a), whereas, under lateral loads, the rota-
The rotation angle (es ) of an equivalent slab element tion angle increases on approaching the column face
can be estimated by summing the rotation angle of slab (Figure 3b). Thus, it is necessary to derive the stiffness
elements (s ) and the average rotation angle of the of the torsional elements under lateral loads which
Magazine of Concrete Research 3
Park et al.
10
L1 L2 6 m
35 m
06 L1
35 m CL Panel
04
Slab thickness 02 m L2
1
Columns 05 m 05 m
2
(a) 02 CL Panel
1 2
00
(a)
10
L1
(b) L2
1
06 2
105 kN
CL Panel
Rigid joint 1 2
Column 35 m
Slabbeam 04
210 kN
35 m
02
6m 6m 6m
00
(c) 00 05 10 15 20 25 30
Distance from the column face: m
Figure 2. Model frame for verification: (a) dimensions; (b)
(b) FEM modelling; (c) EFM modelling
Figure 3. Distribution of rotation angle along the torsional
element: (a) under gravity loads; (b) under lateral loads
reflect the distribution of rotation angle shown in Fig-
x x x
ure 3b. Tx 1 1
The derivation of the stiffness of the torsional ele- x x dx dx xdx x2
0 0 CG 0 L2 CG 2CGL2
ments under lateral loads is summarised in Figure 4. (8)
For comparison, the torsional stiffness under gravity
loads is also depicted in Figure 4. A unit twisting According to the above equation, the maximum rotation
moment (T 1) is applied to the torsional elements, as angle (max ) is obtained at the column face
shown in Figure 4a. Under gravity loads, the moment [x L2 (1 c2 =L2 )]
per unit length of the torsional element may be as-
sumed triangular since the moment in the slab tends to L2 (1 c2 =L2 )2
max (9)
be attracted towards the column (see Figure 4b). How- 8CG
ever, under lateral loads, the moment in the slab is
As shown in Figure 4e, the distribution of rotation
distributed more uniformly along the torsional element;
angle is parabolic. Thus, the average rotation angle
thus, in this study the uniform distribution of the twist-
(t,avg ) is taken as 1/3 of max , which is
ing moment per unit length is assumed, as shown in
Figure 4b. 1 L2 (1 c2 =L2 )2
t:avg (10)
Under lateral loads, the magnitude of a twisting mo- 3 8CG
ment per unit length is 1=L2 . As shown in Figure 4c,
twisting moment T x at x can be calculated by integrat- When Poissons ratio () for concrete is assumed to be
ing the twisting moment per unit length from 0 to x, as zero, the shearing modulus of elasticity for concrete
follows (G) is one-half of Youngs modulus (E): finally, Equa-
x tion 10 can be rewritten as follows
1 1
Tx dx x (7) L2 (1 c2 =L2 )2
0 L2 L2 t:avg (11)
12CE
Twisting moment T x at x produces curvature at x
( x T x =CG). Thus, rotation angle at x is computed Torsional stiffness is calculated by K t T =t. As
by integrating curvature from 0 to x, using the follow- Equation 11 is derived for one-half of the torsional
ing equation element (T 1/2), the torsional stiffness of half of the
4 Magazine of Concrete Research
A modified equivalent frame method for lateral load analysis
Under gravity loads Under lateral loads
T1 T1
L2 c2 L2 c2 L2 c2 L2 c2
1 1 1 1
2 L2 c2 2 L2 2 L2 c2 2 L2
(a)
2
4 x L2
L22 1
L2
x x
(b)
1 1
T T
2 2
2 x2 1 x
Tx Tx
L22 L2
x x
(c)
x x
(d)
x
(e)
Figure 4. Derivation of torsional stiffness Kt : (a) column and torsional element; (b) distribution of twisting moment per unit
length; (c) twisting moment diagram; (d) unit rotation angle; (e) rotation angle
torsional elements can be represented by the following Note that Equation 13 is not the same as Equation 3,
equation derived using two-way slab systems under gravity
loads.
6EC
Kt (12)
L2 (1 c2 =L2 )2
The torsional stiffness of torsional elements is then Verification of proposed MEFM
calculated by summing the stiffness of each torsional To verify the accuracy of the proposed MEFM, this
element as follows study considers an idealised, two-storey, flat plate frame
(Figure 2). Analytical results from the MEFM and
X 6EC
Kt (13) FEM are compared. The results of the ACI-EFM and
L2 (1 c2 =L2 )2 EBWM are also compared. The FEM results are treated
Magazine of Concrete Research 5
Park et al.
as exact results. To model the frame using the ACI- result of approximate methods is greater than that of
EFM and the MEFM, rigid joints are placed at the the FEM analysis.
connections between slabs and columns, as shown in Moment ratio, MR of the EFM, EBWM and MEFM
Figure 2c. The length of the rigid joints is equal to the are 126185%, 110114% and 105111%, respectively.
column depth. One example to calculate slab moments The MEFM most accurately predicts the slab moment
in the flat plate systems using the proposed MEFM is of the FEM whereas the ACIEFM overestimates slab
presented in the appendix. moments significantly. The greatest overestimation
For EBWM, the effective beam width factor in Equa- (185%) occurs at the exterior joints in the frame. This
tions 14 and 15 proposed by Banchik8 is used. Hwang could be attributed to the stiffness of the equivalent
and Moehle2,9 reported that the effective beam width column elements used in the ACIEFM. The stiffness
factor in Equations 14 and 15 is appropriate to use for of the equivalent columns is calculated by combining
simulating experimental results the stiffness of column elements and torsional elements
(Equation 2). As column elements and torsional ele-
c1 1 L1 1
i 5 (for interior frame lines) ments are connected as serial springs, the stiffness of
L2 4 L2 1 2
the equivalent column elements is smaller than the
(14) stiffness of the column element; thus, in the ACI
c1 1 L1 1 EFM, the moment at the slabcolumn joint tends to
e 3 (for exterior frame lines)
L2 8 L2 1 2 concentrated on slabs, compared with moment calcu-
(15) lated by the other methods. This phenomenon is more
significant at the exterior joints since there is only one
slab element connected at the joint.
where c1 is the column dimension in the loading direc-
Drift ratio, DR, of the ACI-EFM, MEFM and
tion, L1 is the slab span length in the loading direction,
EWBM is 282%, 122% and 118%, respectively. MEFM
L2 is the slab span length perpendicular to the loading
and EBWM provide lateral drift ratios close to the
direction and is Poissons ratio.
FEM analysis. However, owing to the smallest stiffness
Moreover, this study compares the analytical results
of the equivalent columns in the ACI-EFM, the largest
of two-way slab systems with different dimensions
lateral drift is produced using this method.
of columns (c1 =c2 0:51:5) and slabs (L1 =L2
0:51:5) obtained from the four different analyses:
finite-element method (FEM), EFM, MEFM and
EBWM. Effect of different dimensions of slabs and
columns
To investigate the influence of different column and
Analytical results of the two-storey flat slab dimensions on analytical results, a two-storey, flat
plate frame (Figure 2) is used. The slab thickness and
plate frame
storey height are assumed to be 0.2 m and 3.5 m,
Table 1 summarises slab moments at the column respectively. Four different types of slab panels (e.g.
faces obtained from different analysis methods. In this 4 m 3 8 m, 4 m 3 6 m, 6 m 3 6 m and 6 m 3 4 m)
table, moment ratio, MR(%), and drift ratio, DR(%), are tested with a fixed column section (0.5 m 3 0.5 m).
denote the ratios of slab moment and drift obtained Furthermore, this study considers slab systems with
from the approximate methods (ACIEFM, EBWM, four different column sections (e.g. 0.4 m 3 0.8 m,
MEFM) to those obtained from FEM, respectively. A 0.4 m 3 0.6 m, 0.5 m 3 0.5 m and 0.6 m 3 0.4 m)
positive ratio value of MR and DR indicates that the with a fixed slab dimension (6 m 3 6 m). Slab
FEM 7000 (100) 6504 (100) 5987 (100) 5987 (100) 0.37 (100)
EBWM 7693 (110) 7220 (111) 6833 (114) 6833 (114) 0.43 (118)
ACI-EFM 12964 (185) 10358 (159) 7534 (126) 7534 (126) 1.03 (282)
Proposed MEFM 7357 (105) 7015 (107) 6650 (111) 6650 (111) 0.45 (122)
MR: %
of span length in the loading direction (L1 ) to the 140
perpendicular direction (L2 ). It is assumed that column
120
section is constant (0.5 m 3 0.5 m) in this calculation.
The fluctuation in MR obtained using the ACI-EFM 100
is very large. The effective beam width method
80
(EWBM) and the proposed MEFM produce slab mo-
ments close to those of the FEM analysis, irrespective (a)
220
of slab aspect ratios (L1 =L2 ). The difference between
FEM
slab moments, obtained from these two methods (pro- 200 MEFM
posed MEFM and the EBWM) and FEM, is less than EBWM
180 ACIEFM
15%.
Figure 6 shows the moment ratio, MR with respect 160
MR: %
to column aspect ratios (c1 =c2 ). Column aspect ratio is 140
the ratio of column depth (c1 ) to column width
(c2 ). The slab dimension is set to 6 m 3 6 m in this 120
calculation. 100
The EWBM and MEFM produce slab moments close
80
to those of the FEM analysis, irrespective of column
aspect ratios. The difference between slab moments 04 06 08 10 12 14 16
obtained from these two methods (MEFM and the Column aspect ratio: c1 /c2
(b)
EBWM) is less than 20%.
Figures 7 and 8 show the drift ratios, DR, with Figure 6. Moment ration (MR) with respect to column aspect
respect to slab aspect ratios and column aspect ratios, ration (c1 /c2 ) (slab dimension: 6 m 3 6 m): (a) exterior
connection; (b) interior connection
220
200
400
180 FEM
350 MEFM
160 EBWM
MR: %
ACIEFM
300
140
250
DR: %
120
200
100
80 150
(a) 100
220
FEM 50
200 MEFM 04 06 08 10 12 14 16
EBWM Slab aspect ratio: L1 /L2
180 ACI-EFM
160 Figure 7. Drift ratio (DR) with respect to slab aspect ratio
(L1 /L2 ) (column dimension 0.5 m 3 0.5 m)
MR: %
140
183
250
DR: %
2
200 Line of symmetry
150
3
100 163 163 244 244 325 163 244 122
50
04 06 08 10 12 14 16 163 163 163 163 244 122 244 122
4
Slab aspect ratio: c1 /c2
(a)
Figure 8. Drift ration (DR) with respect to column aspect
ratio (c1 /c2 ) (column dimension 6 m 3 6 m)
1219 305
81
Table 2. Comparison between the values of lateral stiffness obtained from UCB test and MEFM
( ) denotes the lateral stiffness ratio obtained using each method to that obtained using test / 20, 21
60 60
Verification using NRC slab test
ACIEFM
ACIEFM
50 50
EBWM
EBWM
MEFM
MEFM
7
during this loading stage is compared with that calcu-
lated using the analysis methods considering a stiffness 6 L /L 157 (Unit: mm) L1/L2 064
1 2
5
duce lateral stiffness close to the actual lateral stiffness
2300
obtained at a drift ratio of 0.125%, which is 94% and 4
Slab thk: 286
Measured
95% of the actual lateral stiffness, respectively. In con- 3
MEFM
trast, the lateral stiffness of the ACIEFM is 81% of
2 1820 EBWM
the stiffness obtained from test NS400. NRC slab ACIEFM
Figure 11 shows the relationship between measured 1
00 01 02 03 04 00 01 02 03 04 05
moment and rotation at each connection of the speci-
Deflection: mm
men. The stiffness of each connection is also calculated
using the proposed MEFM and plotted in Figure 11. Figure 12. Comparison of lateral stiffness between MEFM
The connection moment is calculated as the product of and NRC slab tests
10
08
A 1 B 1 C 1 D 1
06
04
02
00
02
Normalised moment: Mu /Mmax
04 UCB test
06 (LAT1)
08 MEFM
10
10
08
A 2 B 2 C 2 D 2
06
04
02
00
02
04
06
08
10
0008 0 0008 0008 0 0008 0008 0 0008 0008 0 0008
0004 0004 0004 0004 0004 0004 0004 0004
Rotation: radian
Figure 11. Comparison of connection stiffness between MEFM and UCB tests
Magazine of Concrete Research 9
Park et al.
Using the approximate methods (MEFM, EBWM, cording to the slab aspect and column aspect ra-
ACIEFM), lateral deflection was also estimated. In tios.
this estimation, stiffness reduction owing to cracks was (d ) Compared with UCB test results in the elastic
not considered since no visual crack was reported in range, the proposed MEFM, FEM and EBWM
the frames during the test.20 provide lateral stiffness values close to those
Figure 12 shows the lateral deflections at each floor obtained from the UCB test at a lateral drift
level obtained from the test as well as the approximate ratio of 1/800 ( 0.125%) which was the load-
methods. The EBWM and the proposed MEFM pro- ing stage LAT1. Moreover, the proposed MEFM
duce lateral deflections close to the measured deflec- with stiffness reduction factor of 1/3 predicts the
tions of the frame having two bays (slab aspect ratio, actual lateral stiffness at a drift ratio of
L1 =L2 1.57). The lateral defections estimated using 1/400 ( 0.25%) which was the loading stage
the ACIEFM overestimate lateral deflection by 15%. NS400.
In the case of the frame having four bays (e) Compared with NRC test results, the proposed
(L1 =L2 0.64), the MEFM and the EBWM accurately MEFM and EBWM provide lateral deflections
predict lateral deflection: the ratio of deflections esti- close to the measured deflections. However, ACI-
mated using the MEFM and the EFM to the actual EFM overestimates deflection by 40%.
deflection is 0.9 and 1.1, respectively. The ACIEFM
overestimates lateral deflection by 40%.
APPENDIX
Conclusion This appendix illustrates an example to calculate slab
This study developed the MEFM, using column moments in the flat plate systems using the MEFM
elements and equivalent slab elements for two-way proposed by this study. For this purpose, a two-storey
slab systems under lateral loads. Equivalent slab ele- frame is considered, which is shown in Figure 2. The
ments consist of slab elements and torsional elements. floor plan and design strip of the building are presented
The stiffness of the torsional elements is derived in Figure 13.
according to the realistic behaviour of torsional ele- Given that H 3.5 m, h 0.2 m, L1 L2 6.0 m,
c1 c2 0p.5 m, f c9 30 MPa (for all members) and
ments under lateral loads. The following conclusions
are made. Ec 4700 f c9 25743 (MPa).
S N
105 kN
210 kN
(a)
Design strip
6m 05 m
02 m
6m
6m
N
6m
(c)
A A
6m 6m
(b)
Figure 13. Prototype building: (a) elevation; (b) floor plan; (c) section AA