Behavior of Biaxially Loaded Concrete-Encased Composite Columns by Pedro R. Munoz Member, ASCE, and Cheng-Tzu Thomas Hsu/ Fellow, ASCE
Behavior of Biaxially Loaded Concrete-Encased Composite Columns by Pedro R. Munoz Member, ASCE, and Cheng-Tzu Thomas Hsu/ Fellow, ASCE
Behavior of Biaxially Loaded Concrete-Encased Composite Columns by Pedro R. Munoz Member, ASCE, and Cheng-Tzu Thomas Hsu/ Fellow, ASCE
COMPOSITE COLUMNS
By Pedro R. Munoz; Member, ASCE, and Cheng-Tzu Thomas Hsu/ Fellow, ASCE
ABSTRACT: This paper reports the experimental test results of four small-scale concrete-encased I-shape steel
columns subjected to biaxial bending moments and axial compressive load in single curvature. The composite
colu~ s~cimens, on: short and three slender, had a square cross section. The effects of the eccentrically
apphed axIal compressIve force, slenderness of the cross section, different material properties of concrete and
steel, and load-deflection and moment-curvature behavior on the maximum load capacity of a composite column
were examined. The test results were compared with the analytical results of the maximum load capacity obtained
from a numerical analysis. Th~ numerical analysis uses the finite differences method to establish the relationship
Downloaded from ascelibrary.org by Northeastern Univ Library on 09/29/14. Copyright ASCE. For personal use only; all rights reserved.
between curvature and defleCtion. A computer program to perform the numerical analysis is used for the case
of composite columns. The comparative results indicate that the analytical method and computer program used
to model and analyze the composite column specimens can accurately predict the maximum load capacity and
deformation behavior of a pinned-ended biaxially loaded concrete-encased steel column with axial compressive
load in single curvature.
(a) O.._-----,.----r----r------,----+O times decreases due to the second order P-B effect. It appears
o 5 10 15 20 25 that the column in the large deflection range or in its final
DEFLECTION (mm)
load-deformation stage behaves like a member being subjected
to uniaxial bending alone about the weak axis. The type of
MOMENT·CURVATURE CURVE FOR MC1 failure observed for the column specimens at the time of test-
(x and y direction)
14,--------------------,.1.6 ing was typically that of crushing of concrete on the inside
corner of the column (the corner on the concave side of the
............................................................................... 1.4 column) with some noticeable cracking on the tensile faces of
the column on the convex side. The behavior of each one of
the composite column specimens under biaxial bending mo-
ments and axial compressive load can be summarized as fol-
lows.
and stroke values were processed and plotted for study and
comparison with the results obtained from a computer analy-
sis. The lateral displacement of the column specimens at mid-
height points were calculated by d = d; - do, where d; = the
dial gauge reading taken at each stroke increment, and do =
the initial dial gauge reading at the time before the first stroke
increment. The axial strain values used to calculate the cur-
vature of the column specimen about the two major axes, X
and Y, were determined by £ = (L; - Lo)/Lo, where L; = length
of the mechanical strain gauge points at each loading stage,
and Lo = length of the mechanical strain gauge points at the
initial loading stage or zero loading. The curvature for each
column specimen was calculated by determining the slope of
the strain values across the cross section. The linear regression
method was used to obtain the strain distribution values along FIG. 3. Photographs of Column Specimens MC1, MC2, MC3,
the column cross section for each loading stage. Plots of the and MC4 after Experimental Test
troduced by Wang and Hsu (1992), Tsao and Hsu (1993), and
side and the dial gauges had to be reset at a load level right finally by Munoz (1994) to study the behavior of composite
beyond the maximum load. A premature formation of crushing columns. A segmental subdivision of the column length is used
of the concrete near the top bracket started to appear, possibly to determine the complete load-moment-curvature-deflection
due to a misalignment of the top and bottom eccentricities that (P-M-<\>-8) for both short and slender columns. The load-de-
could have created an unsymmetrical pinned-ended condition. formation behavior includes the ascending and descending
branches of the loaded column under study. The column cross
Behavior of Column Specimen MC3 section is divided into a number of small square or rectangular
Specimen MC3 showed hairline cracks starting to appear at areas for which the conditions of equilibrium and strain com-
a load level of about 40% of the maximum load. Signs of patibility must be satisfied at the nodal points using the secant
concrete crushing started to appear on the concave side of the modulus of elasticity for the concrete elements. The second-
specimen at a load level near the maximum load. Dial gauges order effects due to the deformed shape of the composite col-
were reset at the maximum load level stage and final concrete umn under load are included in the analysis. The numerical
crushing near the middle of the column height disabled some technique adopted in the computer method is based on the
of the dial gauges measuring the lateral displacements. The incremental deflection approach, where an assumed deflection
test was then terminated. value is specified at a selected joint in a specified direction,
the corresponding equilibrium loads to the specified deflection
Behavior of Column Specimen MC4 are calculated, and the conditions of strain compatibility and
equilibrium are then satisfied along the column length. The
Specimen MC4 showed hairline cracks starting to appear at procedure successfully carries out the results when the as-
a load level of about 40% of the maximum axial load. Dial sumed deflection values accurately match the computed de-
gauges had to be reset at a load level close to the maximum flection values within certain allowable limits. An iterative
load and later at load levels in the descending branch. No procedure to solve a system of nonlinear equations is used to
strain measurements of the steel bars and the steel shape were obtain the solution of the nonclosed form of the second-order
taken for the column specimens tested in the present experi- equations generated by the finite difference method with ex-
mental investigation. tremely rapid convergence. The iteration process achieves the
The four composite column specimens presented a zone maximum load in the ascending branch and proceeds with the
around the midheight of the specimen where concrete crushed, descending branch up to failure.
indicating a typical compression failure of the column in the
compression side. Fig. 3 shows a photograph of the composite Basic Assumptions
specimens after testing. The results obtained from the experi-
mental tests of the four composite column specimens provide The computer analysis is based on the following assump-
valuable information to help in understanding the load-deflec- tions: plane sections remain plane during and after bending;
tion and moment-curvature behavior of short and slender com- the stress-strain relationships for the composite column mate-
posite columns. rials are known; the tensile and compressive strengths of con-
crete are well defined; strain softening of concrete is consid-
COMPUTER ANALYSIS ered and steel is being modeled as an elastic-inelastic material;
the effect of creep and any tensile stresses due to shrinkage
Several computational methods have been used previously are neglected; perfect bond exists between concrete and steel;
to study the nonlinear behavior of biaxially loaded composite the twisting effects and the axial and shear deformations are
columns. A general method to calculate the ultimate strength negligible; the column does not buckle locally before the ul-
of biaxially loaded composite columns with restrained ends, timate load is achieved; the column segments are considered
based on the well-known Newton-Raphson iterative technique to be straight at zero loading and each segment curvature var-
to solve a system of nonlinear equations was presented by ies linearly along the segment under increasing loading con-
Virdi and Dowling (1976). Morino et al. (1984) used a sim- ditions; the member does not have any initial deflection or
plified approach based on an assumed deflected shape of the curvature before it is subjected to a monotonic loading path;
column in the x- and y-directions, given by cosine functions. and the effect of residual stresses in the structural steel section
Lachance (1982) used a numerical integration procedure to is not included.
determine the deflected shape and maximum load of a built-
up composite column identical to that used by Roderick and METHOD OF ANALYSIS AND
Rogers (1969). Roik and Bergmann (1982) and Roik and FUNDAMENTAL EQUATIONS
Schwalbenhofer (1989) used iterative techniques based on an
incremental force approach and use of the secant modulus of The deflected shape of a biaxially loaded composite column
elasticity to calculate the cross-section stiffness matrix. Met- may be approximated by a series of interconnected linear
wally (1988) presented a numerical method to perform inelas- segments undergoing lateral linear displacements and angu-
1166 / JOURNAL OF STRUCTURAL ENGINEERING / SEPTEMBER 1997
The displacements of the column joints are assumed to be substituting the stress-strain relationship crt == (E.)t£t in terms
small, so that the total curvature in the two major bending of the secant modulus of elasticity we obtain
planes labeled as <l>x and <l>y can be represented in the form of n
second derivatives of the displacements as follows:
P == 2: (E.)t£tat (6)
<1>, == -ilu/ax 2 ; <l>y == -a 2 v/ay 2 (3a,b) t-I
n n
The resultant strain distribution corresponding to the cur-
vatures <l>x, <l>y, and the axial compressive strain ~ is assumed Mx == 2: (E.)tEtatYt;
k-l
My == 2: (E.)tEtatXt
,t_l
(7a,b)
Yk X
{
P(eyP+ v) } == [811
821 8
12
822 8813]
23
{Eo}
<1>, (9)
Latmlties pee, + u) 831 832 833 <l>y
Higbly confined
iilil::~
The moment-curvature-thrust relationship for the overall
slender column becomes a nonlinear system of equations that
L
requires an iterative procedure to obtain a solution of the final
r---I--- Point "5" displacements and internal forces. The second-order partial de-
(columnjoint) rivatives of the joint displacements are expressed in terms of
~=i=+---t-- Original line of the finite difference operators and the curvatures of each col-
actionoflhe umn segment. The computer program provides the ultimate
Axial Load
axial load and bending moments at each iteration cycle and
end (typ.) 0 X the axial deformation and the lateral displacements at each
j=n+1
y~ joint of the segmented column for every load level. The com-
puter program for composite columns was coded in FOR-
(joints) TRAN 77 and may be used to study composite sections with
symmetrically placed steel rebars and structural steel shapes.
The writers introduced a more refined model of the stress-
strain relationship of concrete by considering different types
FIG. 4. Blaxlally Loaded Segmented Column with Restrained of confinement for the concrete surrounding the steel shape
Ends and Typical Composite Column Cross Section and the steel bars.
JOURNAL OF STRUCTURAL ENGINEERING / SEPTEMBER 1997/1167
0 · 0 - - -.....5---....,10~·---1.-5---2O...----2+50
alytical results obtained from the theoretical computer model
for column specimen MC1 are presented in Fig. 5. The com- DEFLECTION (mm)
parative plots of the load-displacement and moment-curvature
curves of the analytical and experimental results obtained for LOAD·DEFLECTION CURVES FOR MC1
specimen MC1 are presented in Figs. 6 and 7, respectively. (y • direction)
The plots of the comparative analytical and experimental load- 7.---------,.--------------.....-30
deflection and moment-curvature curves for the other three 28
..·· · ·· ·..· · · _·..·-- ·26
specimens, MC2, MC3, and MC4, all showed the ascending
part up to the maximum recorded and analytical computed 24
· ..· · ·..·22
value of axial load and a clearly marked descending branch.
With the exception of specimen MC2, for which dial gauge !g, 4 -. _ _..__
AA~.
_ _ _ _ _ _
20
__. 18 ~
reading problems were experienced during testing, the other Expertmental 16 ~
two slender specimens, MC3 and MC4, were observed to pro- ~ 3 ..__ _.._._ _ _ - _ _ _ _. 14 9
vide a pattern of load deflection and moment curvature that .:.I
2 ..._ _-_ -
1 kip - 4.45 kN
_ _.-..__.._-_.._ _..__.__.._. 8
~~
was very accurately predicted by the computer model pre-
....................................._._
1 inch. =25.4 mm. _ - _._-_.__
6
4
LOAD-DEFLECTION CURVE FOR MC1
(x and y direction) 2
O......---.....----...----,----r-----+O
::J
(b)
7.------.--.:r-----------T3O o 5 10 15 20 25
28 DEFLECTION (mm)
.
6 =::~ ~ 28
24 FIG. 6. Comparative Load-Deflection Curves (x and y Direc-
5 _... .. 22 tions) for Column Specimen MC1
! x..clrec1lon 20
g, 4 .. _ _..-.......................... 18 ~
sented by the writers. As it was expected for the three slender
y..clrectlon 16 9
~ 3 14 a specimens, MC2, MC3, and MC4, an increased value of the
eccentricity of the applied axial load produced a decrease in
-J 1 Kip = 4.45 kN ~~ ..;J
2 _ ~ _ 6 the value of the maximum axial load and an increase of the
1 inch. =25.4 mm. 6 maximum lateral displacement at the midheight of the speci-
1 .._ .._._.._ _._ _ _ _._ _ -._-_ _ _ 4 mens.
Analytical 2
<a) 0......- - - - , . - - - - , - - - - - , - - - - - , - - - - - + 0 COMPUTER ANALYSIS OF COMPOSITE COLUMNS
o 5 10 15 20 25
DEFLECTION (mm) TESTED BY VIRDI ET AL. (1973)
Virdi and Dowling (1973) presented the results of ultimate
MOMENT·CURVATURE CURVE FOR MC1 loads and lateral deflections of nine pinned-ended composite
(x and y direction) columns tested under axial loads and biaxial bending in single
14.------------------T1.6 curvature. The nine composite columns A-I, tested by Virdi
and Dowling (1973), had a 254 X 254 mm (10 X 10 in.)
._.._. 1.4 square cross section reinforced with a 152.4 X 152.4 mm X
'?E 23.4 kg/m (6 in. X 6 in. X 15.7 lb/ft) structural steel section
'C encased in 50.8 mm (2 in.) of concrete and four 12.7 mm (1/2
.~ ~ in.) diameter rebars, one at each corner and with a 19.05 mm
g, § (3/4 in.) clear cover. The results of the ultimate load obtained
~
!Zw 6 .__ _.._._-_ ; ..~~~~;~: :
y..dlrectlon
~ ..;..~:.;; ~~:~~._ ..__.._
0.8
0.6 ffi by the writers' computer method are presented in Table 2. In
::Ii Table 2, P test is the ultimate test load by Virdi and Dowling
0 ......._ __ _ _ __._ - __ _.._ -... ::Ii
::Ii 4 (1973), P XY is the analytical ultimate load by Virdi and Dowl-
0.4 0
1 Inch. = 25.4 mm. ing's method (1973), and P NJIT is the analytical ultimate load
2 _.._ _ _._ _.._ __ 0.2 ::Ii
by the writers' computer program.
Analytical
(b)
00 2 3 4 5 6 COMPUTER ANALYSIS OF COMPOSITE COLUMNS
CURVATURE (xO.OOO1/mm) TESTED BY MORINO ET AL. (1984)
FIG. 5. Analytical Load-Deflection and Moment-Curvature Morino et al. (1984) presented the experimental results of
Curves for Column Specimen MC1 biaxially loaded composite columns. Concrete-encased wide
1168/ JOURNAL OF STRUCTURAL ENGINEERING / SEPTEMBER 1997
_ ..~
- -.-- -.•...-.-- -.-
.__.._.._.._._. 1
. 1.2
I~ A4-00
A4-30
A4-45
499.91
513.44
518.83
469.74
487.32
513.49
1.064
1.053
1.01
0.952
0.98
0.862
~ ~ 0.6
! :..
A4-60 524.25 538.81 0.971 1.00
~ A4-90 740.44 573.16 1.299 0.833
.__.__.._._.._._~_~.~~~~~_~.~~~?3 k~:~"2. ::: ~ B4..00
B4-30
371.04
392.62
340.02
359.92
1.092
1.091
0.99
0.944
1 inch. - 25.4 mm. B4-45 389.64 381.19 1.022 1.00
2 - - - . -...-.---'-.-.--.------.-..- 0.2 B4-60 436.41 413.32 1.056 0.971
B4-90 503.56 496.13 1.015 0.98
Downloaded from ascelibrary.org by Northeastern Univ Library on 09/29/14. Copyright ASCE. For personal use only; all rights reserved.
"::=~~~--=- :: j
04-60 219.96 210.04 1.047 0.971
04-90 287.87 275.10 1.046 1.01
A8·00 344.34 329.03 1.046 1.042
&
rr::;.
'is.
6 \.
AneIyac.I
.-.-.-.-..-----.--.-.--.--..--..-...-....
1 c.
~
A8-30
A8-45
A8-60
377.05
378.78
447.67
351.28
367.79
393.69
1.073
1.03
1.136
0.98
0.99
0.917
A8..90 520.83 489.10 1.064 0.99
._......_.. _._~
ffi
~
6 _ ..
.._..__._......._..... ..
...- - -
. .:..
1 inch. = 25.4 mm.
- - -..- ..- ..-..-
..
_ _ ::
-
..__ 0.6 )C
-...... 0.2
B8-00
B8-30
B8-45
B8-60
B8·90
260.19
263.62
294.32
328.45
418.03
256.00
268.56
285.38
313.64
405.13
1.016
0.981
1.031
1.047
1.032
1.053
1.031
1.00
0.98
1.01
2 C8..00 179.60 173.77 1.033 1.042
C8-30 176.49 182.05 0.969 1.075
(b) 0.....---.---..--r--...,....---.,------"T---±70 C8-45 195.36 194.78 1.003 1.00
o 2 3 4 5 6
CURVATURE (xO.OOO1/mm) C8-60 194.11 215.56 0.901 1.099
C8·90 296.19 292.68 1.012 1.053
FIG. 7. Comparative Moment-Curvature Curves (x and y DI- D8·00 139.64 130.65 1.068 1.042
rections) for Column Specimen MC1 D8-30 117.35 136.62 0.859 1.075
D8..45 146.67 145.69 1.007 1.053
D8-60 158.42 160.96 0.984 1.064
TABLE 2. Comparative Results for Specimens Tested by Vlrdl
D8-90 221.83 193.66 1.145 1.075
and Dowling (1973)
mean 1.0550 0.998
P._ Pxy P NJ1T standard deviation 0.0779 0.055
Column (kN) (kN) (kN) P_IPxy P....IPNJ1T
(1 ) (2) (3) (4) (5) (6)
A 1,256.02 1,328.91 1,277.46 0.945146 0.9831 puter model and the experimental and theoretical values ob-
B 647.95 684.37 655.53 0.946775 0.9884 tained by Morino et al. (1984) together with the comparative
C 473.50 458.64 433.63 1.032408 1.0919 ratio of the analytical load to test load are shown in Table 3.
D 927.07 1,070.52 1,049.38 0.865994 0.8835 An excellent agreement between the theoretical and experi-
E 573.19 580.56 556.64 0.987293 1.0298 mental ultimate loads is obtained. Most of the computed loads
F 418.67 405.58 385.58 1.032283 1.0858
G 667.89 526.70 546.37 1.268081 1.2223
are below the experimental maximum load, giving a lower
H 353.88 376.38 347.90 0.965802 1.0172 bound solution for the biaxially loaded column specimens
I 294.07 286.41 258.18 1.026731 1.1389 tested, which is considered conservative and safe from the
mean 0.998097 1.04072 practical point of view and for design purposes.
standard deviation 0.094 0.086
COMPUTER ANALYSIS OF COLUMNS TESTED BY
BRIDGE ET AL. (1978)
flange steel sections were used. The column specimens had a
160 X 160 mm (6.3 X 6.3 in.) concrete square cross sections Built-up composite columns were used in the USA since the
encasing rolled steel H-sections of 100 X 100 X 6 X 8 mm very early 1900s. Experimental tests were carried out by some
(4 X 4 X 0.236 X 0.315 in.) and four comer deformed bars investigators such as Emperger (1907) and Mensch (1917),
of 6 mm (0.236 in.) diameter and rectangular ties of 4 mm among others. The concentric and eccentrically loaded built-
(0.157 in.) diameter evenly spaced at a pitch of 150 mm (6 up sections consisted of two or more rolled steel sections or
in.). The writers' computer program was used to study the channels and angles latticed or battened together encased by
behavior of the four types of column specimens: A4, A8, B4, concrete. The writers' computer method is used to model and
B8, C4, C8, 04, and 08. Three experimental parameters were obtain the maximum axial load capacity of uniaxially and bi-
varied for the tested column specimens: (a) the slenderness axially loaded built-up composite columns such as the ones
ratio; (b) the eccentricity of the applied axial compressive tested by Bridge and Roderick (1978) at the Sydney University
load; and (c) the angle location of the applied load. The results in Australia. All the column specimens had the same cross
of the maximum load capacity obtained from the writers' com- section, consisting of two C3 X 5 steel channels encased in
JOURNAL OF STRUCTURAL ENGINEERING I SEPTEMBER 1997 I 1169
posite beam-columns." Wilhelm Ernst & Sohn Verlag jar Architektur My = bending moment about y-axis;
und technische Wissenschaften. Stahlbau, 58, Berlin, Germany. P = axial load;
Roik, K., Diekmann, C., and Schwalbenhofer, K. (1987). "Composite p. = maximum axial load from computer analysis;
columns with steel fiber reinforced concrete." Bauingenieur, 62(4), PNIIT = maximum axial load from writers' computer program;
179-182. P, = maximum axial load from experimental testing;
Stevens, R. F. (1965). "Encased stanchions." The Struct. Engr., London,
U.K.,43.
P tlteo = maximum axial load from Morino and Taylor's theoret-
ical analysis;
Suzuya, J., and Kawana, H. (1984). "Experimental studies on steel re-
inforced concrete beam-columns." Tohoku Kogyo Daigaku Kiyo, 1(4), Pxy = maximum axial load from Virdi and Dowling analysis;
23-40. r = radius of gyration of cross section;
Talbot, A. N., and Lord, A. R. (1912). "Tests of columns: an investigation Su = element of secant stiffness matrix;
of the value of concrete as reinforcement for structural steel columns." u, v = lateral deflections along x- and y-axis, respectively;
Engrg. Experimental Station Bull. No. 56, University of Illinois, Ur- x, y = coordinates of cross-section point in x-y plane;
bana, III. Xl> Yk = coordinates of cross-section element k in x-y plane;
Taylor, R., Shakir-Khalil, H., and Yee, K. M. (1983). "Some tests on a 8 = lateral deflection of cross section;
new type of composite column." Proc., Inst. of Ciy. Engrs., Part 2, E = strain at point of cross section;
Technical Note 355, 283-296.
Tsao, W. H., and Hsu, C. T. T. (1993). "A non-linear computer analysis
Ec = concrete compressive strain;
of biaxially loaded L-shaped slender reinforced concrete columns. " Int. Ek = strain at element k of cross section;
J. Compo and Struct., 49(4), 579-588. E, = tensile strain of concrete;
Tsui, S. H., and Mirza, M. S. (1969). "Model microconcrete mixes." Ey = tensile strain of steel;
Struct. Concrete Series No. 23, McGill University, Montreal, Canada. Eo = strain at coordinate origin of cross section;
Varghese, P. C. (1961). "Ultimate strength of encased steel members 9 = angular direction of neutral axis;
subjected to combined bending and axial loads." J. Instn. of Ciy. cr = stress at point in cross section;
Engrs., India, XLI(6), Pt. I, 225-237. crk = stress at element k of cross section;
Virdi, K. S., and Dowling, P. J. (1973). "The ultimate strength of com- <I> = curvature;
posite columns in biaxial bending." Proc.• Instn. of Ciy. Engrs., Part
2,251-272.
<1>. = curvature with respect to M,; and
Wang, G., and Hsu, C. T. T. (1992). "Complete biaxial load-deformation <l>y = curvature with respect to My.
behavior of reinforced concrete columns." J. Struct. Engrg., ASCE,
118(9), 2590-2609. Subscripts
Watanabe, Y. (1966). "A study of the strength of axially loaded com-
posite columns made with steel H-sections embedded in concrete." = number of segments along column length;
Trans., Arch. Inst. of Japan (in Japanese). j = number of joints along column length;
k cross-section element number;
APPENDIX II. NOTATION n = number of cross-section elements and column segments;
and
The following symbols are used in this paper: s = joint number along deflected column length.