International Journal of Steel Structures

December 2010, Vol 10, No 4, 385-393

www.ijoss.org

Test and Numerical Simulation of Partially Encased Composite

Columns Subject to Axial and Cyclic Horizontal Loads
Yiyi Chen1,*, Tuo Wang2, Jing Yang2,3, and Xianzhong Zhao1
1
State Key Laboratory for Disaster Reduction in Civil Engineering, Shanghai, China 1239 Siping Road, Shanghai, 200092, China
2
School of Civil Engineering, Tongji University, Shanghai, China 1239 Siping Road, Shanghai, 200092, China
3
North-Western China Architectural Design Institute, Xian, China 291 Xiqi Road, Xi’an, 710003, China

Abstract

Partial encased composite (PEC) members with thin steel plates and three types of constructional detailing are studied. Six
pieces of PEC column specimens were tested under constant vertical load and cyclic horizontal loads. A numerical FE model
is established. In the model, the in-filled concrete is decomposed as transverse and longitudinal springs, considering the
functions and the properties of the concrete in two directions. This model simplifies the numerical computation with high
efficiency, and acquires a satisfied agreement with test results in both capacity and deformability of the PEC member. The
parametrical study by the FE model is performed. By tests and numerical analysis, it is found that though local buckling of
thin plate of H steel is a crucial factor, when the compressively axial load does not surpass the encased steel capacity, the PEC
could behave quite well under cyclic horizontal loads. No distinguished difference in three kind details was observed, so that
PEC member could be built in a simple way.

Keywords: partial encased composite member, hysteretic property, cyclic test, numerical model, construction detail

1. Introduction experimental and analytical research was carried out on

PEC columns subjected to eccentrically axial load or
Partially encased composite (PEC) members have been flexural load (e.g. Brent et al., 2006; Oh et al., 2006;
put into practice in past decades. The PEC member takes Zhao et al., 2007; Mahbuba et al., 2007). However, there
advantage of prefabrication, simple installation of formwork, are still some unclear aspects remained. One of them is
efficient compressive load capacity and fire resistance seismic performance of PEC columns, because the
inherent in a concrete column (Vincent et al., 2000). previous experiments were mainly focused on monotonic
The steel of PEC members used in European practice is behavior. The other relates to the hysteretic behavior of
usually with hot-rolled wide flange H steel, thus not PEC column in which the partial encased steel is with
prone to local buckling. In multi-story building structures, large width-thickness ratio.
however, partially encased steel with thinner plate can be In this paper, the experimental research for PEC
utilized for PEC members, with the merits of further columns under pseudo-static loading is reported. A
material saving, and structural weight reduction. Thin special FE method using separate springs to decompose
plate with large width-thickness ratio, however, shall the mechanical functions of concrete is developed to
affect the structural behavior of PEC members. make the complicated steel-concrete composite model
A series of tests on PEC columns under monotonic easy. Based on both experimental study and numerical
axial loading were performed, along with a proposed analysis, the characters of PEC column with thin plate
design equation accounting for local buckling of flange encased steel are evaluated.
(Tremblay et al., 1998; Tremblay et al., 2000; Chicoine et
al., 2002; Chinoine et al., 2003). More recently, the 2. Outline of Experimental Program
2.1. Test specimen
Note.-Discussion open until May 1, 2011. This manuscript for this Six PEC column specimens were designed and
paper was submitted for review and possible publication on February fabricated. The constructional details are illustrated in
9, 2010; approved on December 17, 2010. Fig. 1.
*Corresponding author The height of all six PEC columns is 1,500 mm from
Tel: +86-21-6598-2926; Fax: +86-21-6598-4976 column bottom to the point where horizontal load is
E-mail: [email protected] applied. The partially encased H steel is with the width of
386 Yiyi Chen et al. / International Journal of Steel Structures, 10(4), 385-393, 2010

welded to the steel web. The spacing of the stirrups is set

the same as 100 mm along the column to meet the
requirements of Chinese National Standard (GB 50010-
2002, 2002) for design of concrete structures.

2.2. Material mechanical properties

The characteristic value of the compression strength of
concrete is 33.8 MPa on average by measurement. The
measured mechanical properties of steel plate and bars
are listed in Table 1.

2.3. Loading procedure and test setup

Each specimen is subjected to vertically constant
Figure 1. Cross-sections of specimens. compressive load at first to indicated value and then
cyclic horizontal load. The specimens are divided into
240 mm and the section height of 240 mm (referring as b two groups: one is bent around the major axis, which is
and h in Fig. 1 respectively). The steel plate with perpendicular to the web of steel, and the other is bent
identically nominal thickness of 6 mm is used for the web around the minor axis, denoted by S and W respectively,
and flanges. The effective width-thickness ratio of the in specimen mark referring to Table 2. During the loading
flange, b/t fy ⁄ 235 , is 45.2, slightly greater than 44, process, the vertical actuator is able to move smoothly
which is the maximum value specified by EN 1994-1-1 along a rail following the horizontal displacement at top
(Eurocode 4,2004). of specimen. The axially compressive load is about 0.2N0
Transverse links are set between the steel flanges, and except for specimen C1-S to 0.6N0, where N0 is the
the links are made of ribbed steel bars with nominal axially yielding force of the column defined by Eq. (1).
diameter of 10 mm, welded to steel flanges at every
N0=Asfy+Acfck+Arfyr (1)
120 mm interval except for the column bottom where
transverse links spacing 50 mm are arranged in the extent where As is the cross-sectional area of the H-steel, fy the
of 250 mm. The transverse links are set back from the yield strength of steel plates, Ac the area of concrete and
steel flange tips so that there is 30 mm of clear concrete fck the characteristic value of the compression strength of
cover between the links and the periphery of concrete. concrete, Ar, fyr the gross area of longitudinal rebars and
Different detailings are adopted in the specimens. One the corresponding yield strength. By the results of
is longitudinal rebar which is placed in type II and III material mechanical property tests, N0 is about 3360kN
with nominal diameter of 16 mm, but is absent in type I. for specimen type I and 3100 kN for type II and III.
The existence of longitudinal rebar shall increase the The testing assembly is shown in Fig. 2. In addition,
axially compression capacity of specimen about 9% by normal loading procedure referring to Chinese Specification
assumption of complete plasticity. Another parameter of of Testing Methods for Earthquake Resistant Building
detailing is the stirrups. In type III, U-shaped stirrups of (JGJ 101-96, 1996) is shown in Fig. 3.
8 mm diameter are added. The legs of the stirrups are

Table 1. Mechanical properties of steel plate and rebars

Thick or diameter Yield strength fy Ultimate strength fu Ratio of fy/fu Elongation
Element
(mm) (MPa) (MPa) (%) (%)
Steel plate 6 301 416 72 26
Transverse link 10 367 571 64 26
Longitudinal rebar 16 381 557 68 33
Stirrup 8 466 608 76 -

Table 2. Test specimen and loading condition

Test groups Specimen mark Detailing type Axial ratio Horizontal Loading Feature
C1-S I 0.6
Loading in minor axis but bending around
Group 1 C2-S II 0.2
major axis
C3-S III 0.2
C4-W I 0.2
Loading in major axis but bending around
Group 2 C5-W II 0.2
minor axis
C6-W III 0.2
 Test and Numerical Simulation of Partially Encased Composite Columns Subject to Axial and Cyclic Horizontal Loads 387

3. Experimental Observation and Results

3.1. Failure phenomenon
For major-axis bending columns, local buckling of the
steel flanges along with crushing of concrete and the weld
failure of transverse link in the region of local buckling
were observed. The same phenomenon was observed in
minor-axis bending columns. The local buckling of flange
plates in all of specimens occurred at first at the height of
250 mm to 370 mm from column bottom, where is the
position for the link spacing to change from 50 mm to
120 mm. Some specimens experienced local buckling
near the bottom of column with the test proceeded, though
Figure 2. Loading setup and modeling of PEC specimens. the local buckling occurred after lateral displacement
reached 50 mm and developed slowly.
For major-axis bending columns, concrete crushing
occurred almost simultaneously on both concrete sides
separated by steel web. For minor-axis bending columns,
concrete cracks first spread at the locations corresponding
to a transverse link and then two or three main cracks
developed gradually, becoming wider and deeper, until
failure took place.

3.2. Hysteretic loops of specimens

The loading-displacement curves of specimens are shown
in Fig. 4 and Fig. 5, in which the vertical coordinate is
horizontal load, and horizontal coordinate is lateral
displacement at column top. Specimen C1-S seems quite
different from the other five specimens due to its high
axial load ratio up to 0.6. This compressive load surpasses
Figure 3. Horizontal loading procedure. the fully axial yielding capacity of encased steel, thus the

Figure 4. Hysteretic curves for specimens C1-S, C2-S and C3-S.

Figure 5. Hysteretic curves for specimens C4-W, C5-W and C6-W.

388 Yiyi Chen et al. / International Journal of Steel Structures, 10(4), 385-393, 2010

Figure 6. Skeleton curves for specimens.

specimen exhibited brittle, accompanied with sudden bending direction. From Fig. 4 through Fig. 6, it can be
crushing of concrete and severe local buckling of steel concluded that the constructional detailing, referring to
plates occurring simultaneously. the arrangement of longitude rebars and stirrups, has no
Specimen C2-S and C3-S exhibited identical behavior, significant influence on the load-carrying capacity and
referring Fig. 4b and 4c. After the initial elastic performance, ductility of specimens.
the curves deviated from linearity when lateral displacement The reason for it may be the existence of plates of H
was near to about 10 mm. The ultimate loads reached steel and transverse links that are predominant factor
around 140 kN for C2-S and 130 kN for C3-S, respectively, determining the behavior of the specimens. In the
corresponding to the lateral displacement about 30 mm specimen, the ratio of H steel to gross section area of PEC
for both specimens. The ultimate was after yielding of H- is 7.5%, but only 1.4% for that of longitudinal rebars.
steel but before local buckling of the steel plate. The local Moreover, longitudinal rebars are close to neutral axis
buckling of steel flange was observed when the displacement comparing with the plates of H steel, so little contribution
reached about 50 mm for C2-S and 35 mm for C3-S. of the rebars to the moment capacity can be expected. As
Transverse link weld failed at the same time with local far as transverse links, the arrangement is so compact
buckling and the loss of confining effect on concrete. It especially near the column bottom, so that the function of
can be judged that excessive deformation of buckled constraining concrete has been achieved. The stirrups
flange increased the concentrated stress in welds, and the play a smaller role in the ductility because of little
defect of weld was the possible reason to result in the encircling of concrete.
weld failure. The behavior remained relatively stable up
to a displacement of 60 mm, with the load-carrying 4. Model for Numerical Simulation
capacity fell to about 85% of the ultimate load. The
failure of these columns was ductile and progressive. 4.1. Constitution of analysis model
Other three specimens bent around minor axis, C4-W, For the supplement of experimental study, numerical
C5-W and C6-W, developed approximately the same simulation is necessary, so a mechanical model is
capacity. These three specimens exhibited lower capacity constituted. As a composite structural member, PEC
and more pronounced post-peak strength degradation. By column actually behaves complexly in interaction of two
Fig. 5, the peak loads were 72 kN for C4-W, 78 kN for materials: confining effect on concrete, more or less, due
C5-W and 71 kN for C6-W (average of positive and to steel plates, vertical rebar and links, as well as the
negative ultimate), corresponding to the lateral displacement restraint effect against local deformation of steel plates
of 30, 25, and 25 mm respectively. In the subsequent and rebar due to existence of concrete. In order that a
cycles, the resistance dropped gradually. Local buckling number of parametrical analysis could be performed
occurred at a displacement of 35 mm for both C4-W and efficiently, a simple model is expected. The following
C5-W while 25 mm for C6-W. The load-carrying capacity assumptions are adopted to establish the model.
fell to about 85% of the ultimate load at displacement of (1) The confining effect on concrete is neglect because
35, 50, and 40 mm for the three specimens respectively. it has no significant influence by the observation in
The observations from Fig. 4 and Fig. 5 indicate that tests.
the major axis bending specimens could maintain a much (2) The function of concrete to prevent local buckling
higher capacity than minor axis bending specimens. of steel plate is embodied as transverse spring
However, the behaviors show no relation with the which is compressive only.
specimen details in the identical group of specimen (3) Vertical spring possessing both tensile and compressive
classified in Table 2. capacity is employed to simulate the function of
concrete resisting compressive load and moment.
3.3. Influence of constructional detailing (4) Transverse links are set as tension only springs in
Skeleton curves shown in Fig. 6 are derived from the the model. Vertical steel rebars are set as tension
hysteretic curves of specimens and grouped with the and compression springs before crush of concrete
 Test and Numerical Simulation of Partially Encased Composite Columns Subject to Axial and Cyclic Horizontal Loads 389

Figure 7. The model of numerical analysis.

is exhibited well by the simulation, in spite of the difficulty

to describe the failure process precisely. Another three
specimens were bent around minor axis. The ultimate in
each specimen matches the test data satisfactorily. The
degrading of resistance and the pinch phenomena with
the large deformation are also reflected.
In fact, the mechanism of concrete in PEC column is
three-dimensional. The simple model proposed by authors
Figure 8. Strain-stress relation of concrete. here separates its longitudinal and transverse mechanical
function in order to simplify the constitution of model
and make the analysis efficient. The comparison in Fig. 9
in the model. After the concrete reaches the shows the concept of function separating element feasible.
compressive strength, vertical steel rebars are set as The reason why the combination of uniaxial function can
tension only springs, neglecting the little compression reach so satisfied simulation for integral behavior of PEC
capacity remained. column can be explained as that the restraint preventing
As shown in Fig. 7, the model is composed of shell H-steel from local buckling only induces little inner force
elements for steel plates and several types of springs in horizontal direction, and it does not affect the property
instead of steel rebars, links and concrete. The middle of concrete in vertical direction which is the main stress
point of concrete transverse spring should be unmoved so direction.
that an imaginary stub is fixed to the web as shown in
Fig. 7b. The constitutional relationship of steel in model 5. Parametrical Analysis
is bilinear considering the kinematic hardening effect,
while that of concrete adopts one recommended in The factors affecting the nonlinear behavior of PEC
Chinese design code (GB 50010-2002, 2002) as shown in columns include the width-thickness ratio of H steel
Fig. 8. Non-linearity in geometry and material are both plates, the area and strength ratio of H-steel, longitudinal
taken into account in analysis model. rebar and concrete, the geometrical parameter of steel
The FE model was established and the simulation was links such as diameter and interval distance, the
executed by ANSYS code. constructional details, the ratio of axial load, as well as
the bending direction. By numerical analysis, the effects
4.2. Comparison of the numerical analysis with test of those factors on PEC column capacity and
results deformability were studied parametrically on the base of
In simulation of cyclic loading condition, the constant the monotonic loading analysis. Four of most important
axial load as set in test is first applied, then a given factors, axial load ratio, the interval distance and diameter
displacement routine same as in test is imposed. Figure 9 of links, and the width-thickness ratio of H steel plates
shows the comparison of numerical simulation with the are demonstrated.
test curves.
For the simulation of specimen bent around major axis, 5.1. Axial load ratio
the ultimate, the initial stiffness and unloading stiffness in Take the PEC column as the identical geometrical and
each loop, are quite near the test results in the case of mechanical parameters as in the test specimen type II
specimen C2-S and C3-S, thus the total dissipated energy (referring Fig. 1 and Table 2. Fig. 10 shows how the
also accords with that of test loops well. The specimen horizontal resistance and deformability are influenced by
C1-S was crushed suddenly due to excessive compression the ratio of axial loads. In the figure, the vertical
load as mentioned above, but the feature of soon collapse coordinate is the horizontal load and the horizontal
390 Yiyi Chen et al. / International Journal of Steel Structures, 10(4), 385-393, 2010

Figure 9. Simulation of hysteretic behavior of PEC columns.

Figure 10. Comparison of effects of axial loads.

coordinate is the lateral displacement. The figure with 5.2. Interval distance of steel link
curve notes the ratio of axial load to yielding force As observed above, local buckling behavior of thin
defined by Eq. (1). By the curves, the PEC column walled steel deteriorates the capacity and deformability of
behavior under axial ratio of 0.4 is quite different from the PEC column members. Steel links play an important
that of 0.2. Since the axial load ratio 0.4 is corresponding role in prevention from local buckling of flange plates of
to the full yield capacity of steel by its gross section area H-steel. The closer the interval, the better the loading
in this case, once the concrete stress is beyond its capacity and deformability. However, what interval is
ultimate, the longitudinal rebars shall buckle, and then efficient and economic? Suppose the link diameter keeps
total axially compressive load shall be shifted to H steel the same as in test, with the axial load ratio 0.2 and the
only. The thin wall of the steel plates shall buckle too constructional detail type II. The analysis results for
immediately under both axial compression and bending specimens bending around major axis are illustrated in
stress, and a brittle failure shall occur. It is what we Fig. 11. The analyzed specimen in Fig. 11a with the steel
observed in test referring Fig. 4a, because the axial load geometry same as in tests, while in Fig. 11b the thickness
reached 1.44 times of yield capacity of H steel only in of steel plate reduced to 4 mm, but kept the same width.
specimen C1-S! Considering the post-buckling capacity In the former, the interval change only affects the behavior
of pure H-steel against axially compressive load, the after ultimate, implying that concrete in-filled could
value is around 990 kN, equal to 0.32 times of N0. provide enough restraint against local buckling of flange
Therefore, it can be judged that the PEC member tested before ultimate. In the latter, however, large distance of
under the ratio of axial load not greater than 0.3 shall links shall cause early local buckling, followed with
exhibit satisfied behavior, both for loading capacity and lower capacity. The result shows that if flange width-
ductility. thickness ratio is over the limitation by EN 1994-1-1
 Test and Numerical Simulation of Partially Encased Composite Columns Subject to Axial and Cyclic Horizontal Loads 391

Figure 11. Comparison of transverse link intervals.

Figure 12. Comparison of transverse link diameter.

(Eurocode 4, 2004), the local buckling shall degrade Table 3. Specified width-thickness ratio of H steel plates
ultimate. The steel link functions not only for the
Effective
prevention of early local buckling of steel in case of large Sectional dimensions Steel section
Width-thickness
width-thickness ratio, but also for the post ultimate (b×h×tweb×tflange by mm) ratio
Ratio of flange
behavior.
240×240×6×8 34 9.2%
240×240×6×6 45 7.5%
5.3. Diameter of steel link
240×240×5×5 54 6.3%
Figure 12 shows the effect due to the change of steel
link diameter. The letter d in the figure represents the 240×240×4×4 68 5.0%
diameter of steel link, in millimeter. The obvious difference
can be observed with or without transverse steel link; ratio to whole section changes simultaneously. With the
however no noticeable difference can be seen with the increase of width-thickness ratio of flange, the steel
change of steel link diameter. section occupies small ratio in the entire gross section. To
the loading capacity, the steel ratio plays a crucial role.
5.4. Width-thickness ratio of H steel plates Section strength was computed by strip method according
Large width-thickness ratio of H steel plates may to completely plastic theory ignoring the interaction
induce a premature local buckling, but the large ratio is between encased steel and concrete as well as the effect
favorable if load capacity and deformability expected can of buckling of steel. The results indicate that the ultimates
be obtained in composite structural members. In PEC of specimens shown in Fig. 13 are around 1.09 times of
column, the width-thickness ratio mainly refers to flange the sectional capacity for bending about major axis and
of H steel, and its effect on column performance is 1.05 times for minor axis, both with small errors among
investigated with the parameter scope as shown in Table different specimens. Thus, it can be concluded that the
3. Here, the yield strength of H steel in tests is adopted. ultimate is rarely affected by flange width-thickness ratio
The horizontal load versus lateral displacement curves in the scope of Table 3.
are shown in Fig. 13. It should be noticed that according Then the vertical coordinate of each curve in Fig. 13 is
to Table 3 the change in width-thickness ratio is divided by its corresponding ultimate. Figure 14 shows
implemented by changing the thickness of plates, so steel the graphs. It is revealed the following features, (1) Large
392 Yiyi Chen et al. / International Journal of Steel Structures, 10(4), 385-393, 2010

Figure 13. Comparison of width-thickness ratio of H steel plates.

Figure 14. Comparison of numerical and theoretical ultimate load-carrying capacity.

width-thickness ratio makes the non-linear behavior the complicated simulation for the uncertain interface
appear early, especially when bending about minor axis; behaviors of two materials. The comparison of the
(2) With the increase of flange width-thickness ratio, the numerical analysis and test results shows the validity of
displacement corresponding ultimate exhibits is a little the model.
extension; (3) The post-ultimate behavior, however, is Parametrical study is performed by using the model.
worse in case of large width-thickness ratio. The research shows that axial compression ratio on the
PEC column should be under the encased steel capacity
6. Conclusion when thin plate is adopted for partially encased steel, and
interval of links is a key factor for preventing local
Tests on six column subjected axial compression and buckling of steel with large width-thickness ratio, and for
cyclic horizontal loads were carried out on partially benefiting the post-buckling capacity of steel. The study
encased composite columns with thin-walled H-shaped also indicates the rational scope of width-thickness ratio
steel plates. The experimental study reveals that ductile of encased steel flange.
failure mode and comparatively favorable energy dissipation
ability can be expected if the width-thickness ratio does Acknowledgment
not exceed the limitation by Eurocode 4, and the axial
compression load does not surpass the encased capacity. This research is supported by Chinese National Science
By analysis followed, it is understood that the H steel and Foundation (No.90815029).
transverse links play more important role in resistance
and ductility of PEC members than longitudinal rebars
References
and stirrups, so simple construction detail is possible. The
results of the pseudo-static loading tests give a positive Brent, S. P. and Robert, G. D. (2006). “Behavior of partially
reference for evaluating seismic behavior of the PEC encased composite columns made with high performance
column. concrete.” Structural Engineering Report No. 262,
A simple numerical model is proposed, featured with Department of Civil & Environmental Engineering,
function decomposition springs for concrete, which can University of Alberta, Canand.
reflect main functions of steel and concrete, while avoid Chicoine, T., Tremblay, R., Massicotte, B., Ricles, J. M., and
 Test and Numerical Simulation of Partially Encased Composite Columns Subject to Axial and Cyclic Horizontal Loads 393

Lu, L.W. (2002). “Behavior and strength of partially Tremblay, R., Massicotte, B., Filion, I., and Naranda, R.
encased composite columns with built-up shapes.” (1998). “Experimental study on the behavior of partially
Journal of Structural Engineering, 128 (3), pp. 279-288. encased composite columns made with light welded H
Chicoine, T., Massicotte, B., and Tremblay, R. (2003). steel shapes under compressive axial loads.” Proc. 1998
“Long-term behavior and strength of partially encased Annual Technical Session and Meeting, Structural
composite columns made with built-up steel shapes.” Stability Research Council, Atlanta, pp. 195-204.
Journal of Structural Engineering, 129 (2), pp. 141-150. Tremblay, R., Chicoine, T., Massicotte, B., Ricles, J. M., and
Eurocode 4 (2004). Design of composite steel and concrete Lu, L.W. (2000). “Compressive strength of large scale
structures, Part 1-1: General rules and rules for partially-encased composite stub columns.” Proc. 2000
buildings. (EN 1994-1-1). Annual Technical Session and Meeting, Structural
GB 50010 (2002). Code for Design of Concrete Structures. Stability Research Council, Memphis, pp. 262-271.
China Architecture & Building Press, Beijing (in Tremblay, R., Chicoine, T., and Massicotte, B. (2000).
Chinese). “Design equation for the axial capacity of partially
JGJ 101-96 (1996). Specification of Testing Methods for encased non-compact columns.” Proc. Composite
Earthquake Resistant Building. China Architecture & Construction in Steel and Concrete IV, United
Building Press, Beijing (in Chinese). Engineering Foundation, ASCE, Canada, pp. 506-517.
Mahbuba, B., Robert, G. D., and Alaa, E. E. (2007). “Finite- Vincent, R. B. (2000). “Design and application of partially
element modeling of partially encased composite encased non-compact composite columns for high-rise
columns using the dynamic explicit method.” Journal of buildings.” Proc. Composite Construction in Steel and
Structural Engineering, 133 (3), pp. 326-334. Concrete IV, United Engineering Foundation, ASCE,
Oh, M. H., Ju, Y. K., Kim, M. H., and Kim, S. D. (2006). Canada, pp. 854-864.
“Structural performance of steel-concrete composite Zhao, G. T. and Gao, Z. J. (2007). “Study on the behavior of
column subjected to axial and flexural loading.” Journal partially encased composite columns subjected to axial
of Asian Architecture and Building Engineering, 5(1), pp. loads.” Journal of Harbin Institute of Technology, 39(2),
153-160. pp. 176-178.