Engineering Structures 201 (2019) 109841

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

A unified approach to evaluate axial force-moment interaction curves T

of concrete encased steel composite columns
Binglin Laia, J.Y. Richard Liewa,b, , An Le Hoangc,d, Mingxiang Xionge,a
⁎

a
Department of Civil and Environmental Engineering, National University of Singapore, E1A-07-03, 1 Engineering Drive 2, 117576 Singapore, Singapore
b
School of Civil Engineering, Nanjing Tech University, No. 30 Puzhu Road(S), Nanjing 211816, China
c
Division of Construction Computation, Institute for Computational Science, Ton Duc Thang University, Ho Chi Minh City, Viet Nam
d
Faculty of Civil Engineering, Ton Duc Thang University, Ho Chi Minh City, Viet Nam
e
Protective Structures Centre, School of Civil Engineering, Guangzhou University, Guangzhou, China

ARTICLE INFO ABSTRACT

Keywords: This paper presents a unified approach to evaluate the axial force and moment interaction strength curve of
Axial force moment interaction diagram Concrete Encased Steel (CES) composite columns made of different steel and concrete grades. A database was
Concrete encased steel columns established by collecting the test results of CES composite columns in the literature covering concrete com-
High strength concrete pressive strength ranging from 20 to 104 MPa and steel yield strength from 280 to 913 MPa. A sensitivity study
High strength materials
was then carried out to investigate the effect of different design parameters on the accuracy of the current design
Strain compatibility
method EN 1994-1-1 in predicting the ultimate strength of CES composite columns. These design parameters
Unified approach
include characteristic strength of materials, steel contribution ratio, concrete cover thickness ratio, longitudinal
reinforcement ratio, and volumetric ratio of transverse reinforcement. Analytical study was performed through a
self-compiled computer program which was developed based on materials strain compatibility principle. The
comparison between the analytical results and test results confirms the validity of the proposed method to
provide reasonable prediction of cross-sectional axial-moment interaction curves for CES columns for a wide
range of steel and concrete grades. The existing EC4 method, which is based on plastic design principle, was
found to be un-conservative in predicting the cross section resistance of CES composite columns with high
strength concrete and high strength steel. Further enhancement was made to the proposed method to include
both the strain gradient effect and concrete confinement effect to achieve a better agreement with the test results
reported in literature. Finally, a simplified method was proposed to construct the axial-moment interaction
curves of CES columns, which can be used as a unified approach to design such columns with normal and high
strength steel and concrete materials.

1. Introduction section depending on structural requirements which are related to

strength, stiffness and ductility demand. It has also been experimentally
Concrete Encased Steel (CES) composite columns have been used in confirmed that the existence of steel section can improve the shear
top-down construction of multi-story buildings. Owning to its superior resistance of composite columns [1,2].
load-carrying capacity as compared with conventional Reinforced The cross-sectional capacity of composite columns can be defined in
Concrete (RC) column and bare steel column, it is commonly applied in terms of axial force-bending moment interaction diagram tracing the
the construction of deep basement, underground car park and metro locus of all combinations of axial force and bending moment. Generally,
railway running below the high-rise building. As illustrated in Fig. 1, the interaction diagrams of CES composite columns can be approxi-
CES column can be regarded as a special steel column partially encased mated as polygonal curves by connecting several key points with
by concrete within the steel flanges or fully encased by reinforced straight lines, but most of the modern design codes restrict the appli-
concrete. The surrounding concrete serves as fire and corrosion pro- cation to normal strength materials due to the scarcity of experiment on
tection to the steel section during the entire service life of the building. high strength composite columns [3–7]. Although several attempts
The embedded steel profile can be I-section, cruciform section or twin have been made to investigate the structural response of high strength

⁎
Corresponding author at: Department of Civil and Environmental Engineering, National University of Singapore, E1A-07-03, 1 Engineering Drive 2, 117576
Singapore, Singapore.
E-mail address: ceeljy@nus.edu.sg (J.Y.R. Liew).

https://doi.org/10.1016/j.engstruct.2019.109841
Received 22 March 2019; Received in revised form 25 September 2019; Accepted 20 October 2019
Available online 29 October 2019
0141-0296/ © 2019 Elsevier Ltd. All rights reserved.
B. Lai, et al. Engineering Structures 201 (2019) 109841

Fig. 1. Types of concrete encased steel composite columns.

CES columns, including buckling behavior [8], axial performance compressed concrete fiber at 0.003 [6]. However, no explicit equation
[9,10], beam-column resistance [11,12], seismic resistance [13,14], as is given in the ACI method, and the presence of steel profile pose ad-
well as the performance of partially encased composite columns [15], ditional difficulty for hand calculation in dealing with the complicated
more effort is still needed to gain a better understanding of the struc- stress distribution. In this paper, the strain compatibility hypothesis is
tural behavior of high strength CES columns. taken into consideration by incorporating the strain gradient effect and
As a commonly adopted design approach, EN 1994-1-1 permits the concrete confinement effect which will be discussed in the later part of
design of columns with double symmetrical and uniform cross-sections the paper.
along the column height. In order to account for the limited spread of Besides the design methods available from the codes, some nu-
plastic stress in steel section, a reduction factor αM is incorporated to merical and analytical studies have been conducted to investigate CES
reduce the moment computed based on full plastic stress distribution composite columns that are not covered by existing design codes, such
over the entire section. The value of αM is highly associated with the as the design of non-symmetrical cross-section and neglect of concrete
steel grade, it is taken as 0.9 for S235/S355 steel and 0.8 for S420/S460 confinement effect, etc. Lai et al. [17] performed numerical simulation
steel, and higher steel grade is currently not mentioned, which requires on CES columns with off-centered steel section and developed a mod-
large axial strain to achieve the yield strength [16]. As compared in ified EC4 design approach to predict the cross-section capacity of such
Fig. 2, AISC 360-16 allows for several methods to construct the inter- non-symmetric composite sections. Liu et al. [18] developed a com-
action curve, among which method 1 is based on the modification of puter program capable of generating the axial force-bending moment
pure steel column design resistance, and the degree of conservatism diagrams of composite columns with arbitrary cross-section shapes.
increases with the increasing extend of concrete contribution to the Roik et al. [19,20] pointed out some special problems of unsymmetrical
overall capacity [4]. Method 2 in AISC 360-16 takes into account the cross-sections design and came up with equations for the determination
slenderness effect, and hence the ordinate of diagram moves downward of elastic centroid axis and plastic centroid axis, which are used in
based on the reduction factor calculated in pure compression scenario. elastic and plastic design, respectively. And some additional numerical
In AIJ method [5], the composite action is completely ignored. Steel techniques were employed by Chiorean [21], Papanikolaou [22] and
and reinforced concrete (RC) hold different neutral axis since they are Chen et al. [23]. El-Tawil et al. [24,25] adopted fiber element method
treated as separate component when computing the axial force and for the calculation of interaction diagrams and inelastic moment-cur-
bending moment acting on the section [5]. As illustrated in Fig. 2, AIJ- vature response of composite cross-section, based on which the ACI 318
based interaction diagram of composite section is generated by shifting and AISC (LRFD) strength provision for composite beam-columns were
the N-M curve contributed by the steel section and superimposed it on evaluated. It was evidenced that ACI method models the CES column
the N-M curve for the RC Section. ACI-310-08 adopts the same design behavior more realistically than AISC. Fenollosa et al. [26] developed a
philosophy as RC design, which is achieved through the use of strain formulation tracing the interaction diagrams based on elastic-plastic
compatibility hypothesis and fixing the maximum strain of the most stress distribution, and Rocha [27] presented a methodology to

Fig. 2. Comparison of three different design codes for composite columns: EN1994-1-1 (EC4); AISC 360-16 (AISC); AIJ-2014 (AIJ).

2
B. Lai, et al. Engineering Structures 201 (2019) 109841

construct the N-M diagrams based on the deformation domains of re- results with available test data from literatures; (2) Perform sensitivity
inforced concrete structures. However, all of these methods cover only study to evaluate the accuracy of EC4 and the proposed method with
normal strength CES columns, and their accuracy remains unknown the variation of material strengths, concrete cover ratio, steel con-
when it comes to columns made of high strength materials. tribution ratio, reinforcement ratio and volumetric ratio of transverse
In view of the absence of a unified approach for designing CES reinforcement; and (3) Propose a unified design method to construct
columns covering wide range of different steel and concrete grades, this axial and moment interaction diagrams applicable for both normal
paper proposes a generic section-analysis method that take into account strength and high strength CES columns.
strain gradient effect and concrete confinement effect. Current EC4
method (“EC4” is commonly used term as an abbreviation of “EN 1994- 2. Experimental research on CES columns
1-1”) is carefully examined and finally a simplified design method is
proposed. Thus the contributions from the present research are: (1) A database consisting of 71 test data collected from literatures is
Develop an analytical procedure to construct interaction diagram of used for the calibration of the proposed design method and the com-
CES columns and validate its accuracy by comparing the predicted parison with the existing EC4 method. Table 1 tabulates the geometric

Table 1
Details of test specimens subject to combined axial force and bending moment.
Ref. Specimen B×D Steel section δ Cz/h fc (MPa) fys (MPa) fyr (MPa) ρl (%) ρv (%) Nt (kN) Mt (kNm) Loading type

Mirza [28,55] RHB-1a 240 × 240 96 × 100 × 5.1 × 8.6 0.27 0.75 27.11* 293.4 565 0.57 0.49 950 64.1 A
RHB-2 a 240 × 240 96 × 100 × 5.1 × 8.6 0.27 0.75 27.11* 293.4 565 0.57 0.49 550 63.2
RHB-3 a 240 × 240 96 × 100 × 5.1 × 8.6 0.27 0.75 27.65* 293.4 565 0.57 0.49 570 78.2
RHB-4 240 × 240 96 × 100 × 5.1 × 8.6 0.29 0.75 25.59* 311.2 634 0.57 0.49 307.5 79.8
RHB-4A 240 × 240 96 × 100 × 5.1 × 8.6 0.29 0.75 24.79* 293.4 565 0.57 0.49 154.3 66
RHB-5 240 × 240 96 × 100 × 5.1 × 8.6 0.26 0.75 28.63* 293.4 565 0.57 0.49 95 65.6
RNHB-1b 240 × 240 96 × 100 × 5.1 × 8.6 0.28 0.75 27.47* 311.2 634 0.57 0.39 925 82.2
RNHB-2b 240 × 240 96 × 100 × 5.1 × 8.6 0.28 0.75 27.47* 311.2 634 0.57 0.39 775 76
RNHB-3 240 × 240 96 × 100 × 5.1 × 8.6 0.27 0.75 26.57* 293.4 565 0.57 0.39 540 82.3
RNHB-4 240 × 240 96 × 100 × 5.1 × 8.6 0.27 0.75 27.11* 293.4 565 0.57 0.39 352.5 93.8
RNHB-5 240 × 240 96 × 100 × 5.1 × 8.6 0.26 0.75 28.09* 293.4 565 0.57 0.39 107.5 73.5
RHNB-1b 240 × 240 96 × 100 × 5.1 × 8.6 0.28 0.75 27.47* 311.2 634 0.57 0.39 927 72
RHNB-2b 240 × 240 96 × 100 × 5.1 × 8.6 0.28 0.75 27.47* 311.2 634 0.57 0.39 720 69.9
RHNB-3 240 × 240 96 × 100 × 5.1 × 8.6 0.29 0.75 25.59* 311.2 634 0.57 0.39 540 83
RHNB-4 240 × 240 96 × 100 × 5.1 × 8.6 0.29 0.75 25.59* 311.2 634 0.57 0.39 296 79.9
RHNB-5 240 × 240 96 × 100 × 5.1 × 8.6 0.29 0.75 25.59* 311.2 634 0.57 0.39 100 68.7
Ricles [29,55] 1 406 × 406 W8 × 40 0.31 0.47 32.70 373.7 455.8 3.04 1.26 1490 626 C
2 406 × 406 W8 × 40 0.35 0.47 34.50 373.7 434.4 0.99 1.26 1490 593
3 406 × 406 W8 × 40 0.32 0.47 30.90 373.7 434.4 3.04 1.68 1490 784
4 406 × 406 W8 × 40 0.35 0.47 31.10 373.7 448.2 1.66 0.84 1490 670
5 406 × 406 W8 × 40 0.30 0.47 34.50 373.7 434.4 3.04 1.68 1490 776
6 406 × 406 W8 × 40 0.33 0.47 35.80 373.7 448.2 1.66 0.84 1490 667
7c 406 × 406 W8 × 40 0.22 0.47 62.90 373.7 434.4 3.04 1.68 1490 840
8c 406 × 406 W8 × 40 0.22 0.47 64.50 373.7 434.4 3.04 1.68 1490 832
Kim [11,12] C1 260 × 260 150 × 100 × 17.6 × 17.6 0.49 0.37 94.00 913 525 1.30 2.59 2120 288 B
C2 260 × 260 150 × 100 × 17.6 × 17.6 0.49 0.37 94.00 913 525 1.30 2.59 3752 264
C3 260 × 260 150 × 100 × 17.6 × 17.6 0.49 0.37 94.00 913 525 1.30 1.00 2020 270
C4 260 × 260 150 × 100 × 17.6 × 17.6 0.49 0.37 94.00 913 525 1.30 2.59 2072 281
C10 260 × 260 150 × 150 × 15 × 15 0.47 0.37 104.00 812 512 1.32 2.00 2023 276
C11 260 × 260 150 × 150 × 15 × 15 0.47 0.37 104.00 812 512 1.32 2.00 1986 288
Naka [30,55] 1d 240 × 300 180 × 120 × 4.5 × 12 0.36 0.33 25.50 344.8 461.3 2.41 0.40 1470 197.4 A
2d 240 × 300 180 × 120 × 4.5 × 12 0.36 0.33 25.50 344.8 461.3 2.41 0.40 980 235
3d 240 × 300 180 × 120 × 4.5 × 12 0.36 0.33 25.50 344.8 461.3 2.41 0.40 490 228.4
4d 240 × 300 180 × 120 × 4.5 × 12 0.36 0.33 25.50 344.8 461.3 2.41 0.40 0 214
Wakabayashi SRC 0N-M 210 × 210 150 × 100 × 6 × 9 0.48 0.20 21.50 304.6 367.6 0.76 0.28 0 69.1 A
[31,55] SRC 2N-Me 210 × 210 150 × 100 × 6 × 9 0.43 0.20 26.90 312.3 367.6 0.76 0.28 300 74
SRC 4N-Me 210 × 210 150 × 100 × 6 × 9 0.41 0.20 29.50 312.3 367.6 0.76 0.28 600 69.1
SRC 6N-Me 210 × 210 150 × 100 × 6 × 9 0.43 0.20 27.50 312.3 367.6 0.76 0.28 900 60.3
Zhao [32] SRHC-E1f 160 × 180 100 × 68 × 4.5 × 7.6 0.37 0.40 35.44* 379 358 1.16 0.54 678 42.09 B
SRHC-E2 160 × 180 100 × 68 × 4.5 × 7.6 0.38 0.40 32.96* 379 358 1.16 0.54 820 40.70
SRHC-1 160 × 180 100 × 68 × 4.5 × 7.6 0.33 0.40 42.48* 379 361 1.16 0.54 654 48.17
SRHC-2f 160 × 180 100 × 68 × 4.5 × 7.6 0.36 0.40 36.56* 379 361 1.16 0.54 486 40.06
SHC-E1 160 × 180 100 × 68 × 4.5 × 7.6 0.36 0.40 38.32* 400 350 1.16 0.54 490 52.69
SHC-E2 160 × 180 100 × 68 × 4.5 × 7.6 0.34 0.40 43.28* 400 350 1.16 0.54 282 46.29
SHC-E3 160 × 180 100 × 68 × 4.5 × 7.6 0.37 0.40 38.00* 400 350 1.16 0.54 441 49.52
SHC-E4 160 × 180 100 × 68 × 4.5 × 7.6 0.31 0.40 49.52* 400 350 1.16 0.54 249 43.52
SHC-E5 160 × 180 100 × 68 × 4.5 × 7.6 0.40 0.40 32.64* 400 350 1.16 0.54 330 40.65
SHC-E6 160 × 180 100 × 68 × 4.5 × 7.6 0.35 0.40 41.20* 400 350 1.16 0.54 210 38.37
Yu [33] SRC1.4-40-1 200 × 200 126 × 74 × 5 × 8.4 0.28 0.29 34.08* 280.5 380.3 1.20 0.78 1300 57.60 B
SRC1.8-40-1 200 × 200 126 × 74 × 5 × 8.4 0.28 0.29 34.08* 280.5 380.3 1.20 0.78 1200 56.80
SRC1.4-40-2g 200 × 200 100 × 68 × 4.5 × 7.6 0.25 0.50 34.08* 280.5 360.3 0.82 0.78 865 37.18
SRC1.8-40-2 g 200 × 200 100 × 68 × 4.5 × 7.6 0.25 0.50 34.08* 280.5 360.3 0.82 0.78 811 36.21
SRC1.4-80-2 g 200 × 200 100 × 68 × 4.5 × 7.6 0.25 0.50 34.08* 280.5 360.3 0.82 0.78 538 45.47
SRC1.8-80-2g 200 × 200 100 × 68 × 4.5 × 7.6 0.25 0.50 34.08* 280.5 360.3 0.82 0.78 557 48.64
SRC1.4-100-1 200 × 200 126 × 74 × 5 × 8.4 0.28 0.29 34.08* 280.5 380.3 1.20 0.78 710 75.27
SRC1.8-100-1 200 × 200 126 × 74 × 5 × 8.4 0.28 0.29 34.08* 280.5 380.3 1.20 0.78 654 71.29
(continued on next page)

3
B. Lai, et al. Engineering Structures 201 (2019) 109841

Table 1 (continued)

Ref. Specimen B×D Steel section δ Cz/h fc (MPa) fys (MPa) fyr (MPa) ρl (%) ρv (%) Nt (kN) Mt (kNm) Loading type

Han [34] AH2-E2-80 160 × 160 100 × 100 × 6 × 8 0.56 0.30 21.70 316 280 1.36 0.76 1222 26.88 B
AH2-E4-80 160 × 160 100 × 100 × 6 × 8 0.56 0.30 21.70 316 280 1.36 0.76 840 36.96
AH2-E8-80 160 × 160 100 × 100 × 6 × 8 0.56 0.30 21.70 316 280 1.36 0.76 537 43.75
BH2-E2-80h 160 × 160 100 × 100 × 6 × 8 0.56 0.30 20.60 294 280 1.36 0.76 990 21.42
BH2-E4-80h 160 × 160 100 × 100 × 6 × 8 0.56 0.30 20.60 294 280 1.36 0.76 764 33.52
BH2-E8-80h 160 × 160 100 × 100 × 6 × 8 0.56 0.30 20.60 294 280 1.36 0.76 509 44.28
CH1-E2-80 160 × 160 100 × 100 × 3.2 × 4.5 0.42 0.30 20.60 310 280 1.30 0.76 894 22.42
CH1-E4-80 160 × 160 100 × 100 × 3.2 × 4.5 0.42 0.30 20.60 310 280 1.30 0.76 574 24.79
CH1-E8-80 160 × 160 100 × 100 × 3.2 × 4.5 0.42 0.30 20.60 310 280 1.30 0.76 352 31.68
CH2-E2-80h 160 × 160 100 × 100 × 6 × 8 0.56 0.30 20.60 294 280 1.36 0.76 1003 24.57
CH2-E4-80h 160 × 160 100 × 100 × 6 × 8 0.56 0.30 20.60 294 280 1.36 0.76 698 30.63
CH2-E8-80h 160 × 160 100 × 100 × 6 × 8 0.56 0.30 20.60 294 280 1.36 0.76 486 44.71
DH2-E2-80h 160 × 160 100 × 100 × 6 × 8 0.56 0.30 20.60 294 280 1.36 0.76 864 22.25
DH2-E4-80h 160 × 160 100 × 100 × 6 × 8 0.56 0.30 20.60 294 280 1.36 0.76 624 32.94
DH2-E8-80h 160 × 160 100 × 100 × 6 × 8 0.56 0.30 20.60 294 280 1.36 0.76 489 44.94

Note:
δ, Cz/h, ρl, ρv refer to steel contribution ratio, concrete cover thickness ratio, reinforcement ratio and volumetric ratio of transverse reinforcement.
fc, fys and fyr refer to concrete compressive strength and the yield strength of structural steel and longitudinal reinforcement bars.
Nt and Mt are the ma × imum a × ial force and corresponding bending moment reported in literatures.
*Cubic strength of concrete is reported in literature, and converted to cylinder strength by multiplying 0.8.
a,b,c,d,e,f,g,h
these test data were plotted in Fig. 6 (a), (b), (c), (d), (e), (f), (g) and (h), respectively.

and material properties of the test specimens. Only the specimens with
double symmetric cross-section with I or H-shape steel profiles encased
by concrete are studied. This database covers a wide range of concrete
cylinder strength fc, steel yield strength fys, reinforcement bar yield
strength fyr, concrete cover thickness ratio Cz/h, steel contribution ratio
δ, longitudinal reinforcement ratio ρl, as well as volumetric ratio of
transvers reinforcement ρv. Among these parameters, only δ and ρl fulfill
the requirements in EC4, with 0.2 ≤ δ ≤ 0.9 and ρl ≤ 6%. As for con-
crete cover thickness ratio Cz/h, it can be as high as 0.75, which is
beyond the limit in EC4.
Although high strength CES column test data are available, there is
still a research gap for concrete strength within 60–90 MPa and steel
yield strength within 400–800 MPa. All the tested specimens reported
herein were subjected to a combined action of axial compression and
bending moment. Depending on the different manners bending moment
is applied, the test data are categorized into three types: Type A, Type B,
and Type C. As illustrated in Fig. 3, Type A refers to members subjected
to axial compression accompanied with transverse loading, and Type B
indicates eccentric compression which leads to single curvature
bending. Both Types A and B have their critical sections located at mid-
span or mid-height of the column. However, for Type C, the column is
cantilevered from the fixed base and it is subject to an axial compres-
sion and a horizontal cyclic load at the top causing maximum moment
at the base.
Mirza et al. [28] performed 16 tests on CES beam-columns with Fig. 3. Three different loading types: Type A: Combing compression with
large concrete cover, and the external axial force and bending moment transverse loading; Type B: Eccentric compression; Type C: Cyclic loading.
covers the entire practical range of interaction diagrams. It was found
that both EC4 and ACI 318 method gives fairly good prediction of ul- volumetric ratio of 1.99% and 2.59%. The comparison between test
timate strength, and the bond between concrete and steel section has results and various design codes indicated that ACI-318-08 method
negligible effect on the load-carry capacity. Ricles & Paboojian [29] underestimates the load-carrying capacity while EC4 method over-
tested 8 CES columns with 406 mm × 406 mm dimension and re- predicted the test results indicating that the full plastic capacity of steel
inforced with W80x40 steel section. All specimens were tested under section cannot be achieved when high strength steel is used. Naka et al.
seismic loading and the test results revealed that both ACI and AISC- [30] tested a series of rectangular CES members using combining axial
LRFD methods were conservative in predicting the failure loads, while force and transverse loading condition. The axial force ranging from 0
the AIJ superposition method gives quite good prediction. Besides, it to 1470 kN with 490 kN interval. The test result reveals that AIJ
also concluded that shear studs have small effect on the flexural stiff- method could estimate the flexural resistance of CES columns at first
ness and strength, which was consistent with the research finding by yield, but it underestimated the maximum flexural resistance. Waka-
Mirza [28]. bayashi et al. [31] performed experimental and theoretical study on the
Kim et al. [11,12] carried out experimental investigation on CES moment (M)-curvature (Ф)-axial force (N) relation of CES members. By
beam-columns with high strength concrete 94/104 MPa and ultra-high using the nonlinear constitutive law of concrete and elastic-perfectly
strength steel with yield strength 812–913 MPa. Most of their speci- plastic model of steel section, the analytical M-Ф-N turned out to be in
mens were heavily reinforced with closely-spaced stirrups with the good agreement with test result. A considerable volume of experimental

4
B. Lai, et al. Engineering Structures 201 (2019) 109841

Fig. 4. Determination of N-M interaction based on strip element method.

research was conducted by Zhao et al. [32] and Yu & Lu [33]. All these diagrams from pure compression to pure bending can thus be traced
normal strength composite columns were tested under eccentric com- through the computer-assisted program compiled in MATLAB.
pression with the load eccentricity ratio ranging from 0.167 to 0.833. Several assumptions are made to in strip element method: (1). Steel-
Both the deformability and bearing capacity of CES column improved concrete interface is perfectly bonded without slippage so that strain
with the increase of steel ratio, while concrete strength did not affect compatibility criterion can be satisfied. This assumption was justified
much on the capacity of slender columns. According to the experi- through the experimental research done by Chen et al. [35] and Mirza
mental investigation by Han et al. [34], the N-M curve generated using et al. [28], which implies the adhesion and friction at the interface of
full plastic capacity can conservatively estimate the beam-column steel section and surrounding concrete is able to provide sufficient
strength despite the variation of load eccentricity, slenderness ratio and bond; (2). Plane section remains plane after deformation; (3). Concrete
steel ratio. The CES column specimens from Han et al. [34] were made tensile strength is small and can be neglected; (4). Initial residual
of normal strength materials with concrete compressive strength of stresses of steel are neglected as they do not affect the cross section
20 MPa and steel yield strength of about 300 MPa resistance; (5). Shear deformation is small for CES and is neglected. (6)
Concrete confinement effect is carefully evaluated and included in the
analysis for well-confined composite columns. (7) Failure is assumed
3. Numerical investigation
when the most compressed concrete fibre reaches the crushing strain.

3.1. Strip Element method

3.2. Material models
Sectional analysis is performed using the Strip Element Method
(SEM), which is an efficient numerical integration technique that fol-
For simplicity, elastic-perfectly plastic model is adopted for steel
lows the same approach as ACI method [6]. The nonlinear material
section and longitudinal reinforcement. Strain-hardening of steel is
properties of concrete can be implemented in the program to better
conservatively ignored since the concrete crushing may prevent steel
reflect the structural behaviour than the current design guides. As il-
from entering into strain-hardening stage. Local buckling of steel sec-
lustrated in Fig. 4, the composite section is discretized into a number of
tion is also disregarded as it was effectively prevented or delayed by the
strip elements with equal thickness. Following the assumption of strain
surrounding concrete.
compatibility and linear strain distribution over the entire cross-section,
The model developed by Legeron [36] is selected to model the
the longitudinal stress of each strip element can be obtained by as-
nonlinear behavior of both unconfined and confined concrete. This
signing respective material constitutive relation, and hence the internal
model is applicable for a wide range of concrete strength. In this model,
axial force and bending moment acting on the cross-section can be
the actual stress of transverse reinforcement is determined through a
computed accordingly. Bending moment at the ith strip is the product of
comprehensive regression analysis, and a non-dimensional parameter
axial force and lever arm (di) measured from the geometric centroid
Ie’ is proposed to relate the concrete confining effect with the transverse
axis of ith strip to that of the entire section. By shifting the neutral axis
reinforcement yield strength and concrete strength as below:
position consecutively, numerous combinations of axial force (N) and
bending moment (M) can be generated, and the full range of interaction

Fig. 5. Concrete model used in Strip Element Method: (a). Stress strain curve of unconfined and confined concrete; (b). Different maximum strain used under
different strain gradient.

5
B. Lai, et al. Engineering Structures 201 (2019) 109841

fle subjected to flexural loading, despite the attainment of maximum stress

Ie =
fco (1) at the most compressed fiber, it is still capable to continue resisting the
flexural load as the stress can further develop in the subsequent un-
where fle is the effective confinement stress acting on core concrete, derlying concrete fibers. However, for concrete columns under pure
and fco is the peak stress of unconfined concrete. compression, the entire section is subjected to a uniform compressive
The confined concrete peak stress fcc and corresponding strain εcc stress. When the maximum stress is reached, any further increase of
can be calculated as a function of unconfined strength fco and strain εco compressive strain may lead to brittle crushing of concrete, unless
as follow: sufficient confinement is provided. In view of this, two concrete strains,
fcc εco and εcu, highlighted in Fig. 5(a) were used in the analytical proce-
= 1 + 2.4(Ie )0.7 dure, which are named as peak strain and ultimate strain, respectively.
fco (2)
As illustrated in Fig. 5(b), when neutral axis locates within the com-
cc posite section (Points C, D and E), the ultimate concrete strain, εcu, is
= 1 + 35(Ie )1.2
co (3) used and taken as 0.003 as recommended by ACI. Otherwise a linear
interpolation is used between the ultimate strain εcu and εco (Point B).
The full range stress-strain curve of confined concrete can be for- For the pure compression point (A), εco is used to determine the axial
mulated using Eqs. (4) & (5): capacity. The equation of εco can be found from Legeron [36] as:
k ( c / cc )
c = fcc For c cc co = 0.0005(fco )0.4 (6)
k 1 + ( c / cc ) k (4)

c = fcc exp [k1 ( c

k
cc ) 2] For c > cc (5)
3.3. Calibration with test results
where k, k1 and k2 are parameters controlling the shape of ascending
and descending branch, respectively. The unconfined concrete stress- 3.3.1. Comparison with test data of normal strength materials
strain model can be obtained by taking the confining pressure fle as 0. Selected test data in Table 1 are plotted in Figs. 6 and 7 to calibrate
Details of this model can be found in Legeron [36]. the interaction diagrams generated by the proposed SEM and EC4
It is widely accepted that flexural concrete members behave more method. It shall be noted that concrete confinement effect is ignored
ductile than compression concrete members. When concrete is and its effect will be discussed later. As shown in Fig. 6, for CES

Fig. 6. Comparison of axial force-bending moment interaction diagrams between SEM and EC4 method for normal strength CES columns.

6
B. Lai, et al. Engineering Structures 201 (2019) 109841

Fig. 6. (continued)

columns with normal strength materials, i.e. concrete cylinder strength observed that the EC4 N-M curves match the SEM estimation quite well
between 20 MPa and 60 MPa, and steel yield strength between 280 MPa when the axial force interacts with a large bending moment, revealing
and 400 MPa, most test data fall onto the envelope with minor devia- that the reduction factor αM is able to account for the difference be-
tion, indicating good accuracy of both SEM and EC4 method. It is tween full plastic stress distribution and actual elastic-plastic stress

Fig. 7. Comparison of axial force-bending moment interaction diagrams between SEM and EC4 method for high strength CES columns.

7
B. Lai, et al. Engineering Structures 201 (2019) 109841

Table 2
Variation range of each parameter in parametric study.
Parameters Variation range

concrete compressive strength (MPa) 20.6–104

Steel yield strength (MPa) 280.5–913
Concrete cover thickness ratio 0.2–0.75
Steel contribution ratio 0.22–0.56
Longitudinal reinforcement ratio (%) 0.57–3.04
Stirrup volumetric ratio (%) 0.28–2.59

discussed in the proposed simplified design method.

3.3.2. Comparison with test data of high strength materials

High strength CES columns are currently beyond the scope of EC4
(EN1994-1-1). For comparison purpose the existing EC4 method is still
employed and the reduction factor αM is taken as 0.8 for steel grade
higher than S460. As can be seen in Fig. 7, EC4 overestimates the cross-
section resistance of high strength CES beam-columns with large error,
revealing that the plastic design method inherited in EC 4 method is not
applicable for composite columns made of high strength steel, and
hence the high strength materials cannot be fully utilized unless it sa-
Fig. 8. Stress distribution in the steel section for concrete encased composite
column under pure bending.
tisfies the materials compatibility requirement originally proposed by
Liew et al. [37,38]. However, the proposed SEM generates quite rea-
sonable prediction even though high strength materials are used.
Therefore, without the need of advanced computational methods, strain
compatibility method works well as a feasible technique for con-
structing the interaction diagrams of high strength CES columns.
The stress distribution in the steel section is extracted from the
analytical procedure and graphically illustrated in Fig. 8. For the ease of
comparison, only the pure bending scenario is selected and difference
can be observed for the stress distribution in normal strength and high
strength steel section. For the specimen tested by Wakabayashi [31],
the tip of both top and bottom steel flanges reaches the yield strength at
the maximum resistance and the plastic stress spreads to a large portion
of the tension web, suggesting that reducing the full plastic moment
resistance by 0.9 (αM = 0.9) may serve as a proper way to approximate
the actual flexural strength. However, when it comes to the high
strength steel section with a yield strength of 913 MPa [11], the entire
steel section still remains in elastic state, and the compression flange
stress (270 MPa) is even lower than 1/3 of the yield strength. Therefore,
Fig. 9. Methodology of comparing analytical result with test result. the plastic design method may lead to overestimation despite the use of
reduction factor αM in design.

3.4. Parametric study of SEM and EC4 Method.

To evaluate the two methods in a more comprehensive way, para-

metric study is carried out to investigate the effect of different para-
meters on the accuracy of the proposed Strip Element Method (SEM). A
methodology introduced in [39] is employed here for the comparison
between analytical and test results which include both axial force and
bending moment. As illustrated in Fig. 9, the difference between the test
and analytical result is evaluated using the index termed Rtest and
Ranalysis.
Rtest is taken as the vector originating from the origin of coordinate
and terminating at the test result, and Ranalysis can be obtained in the
same way but the analytical result is taken as the intersection of the
Rtest line and the N-M envelope.
Six parameters are evaluated, i.e. concrete compressive strength fc,
Fig. 10. Definition of concrete cover thickness. steel yield strength fys, concrete cover thickness ratio Cz/h, steel con-
tribution ratio δ, longitudinal reinforcement ratio ρl, and volumetric
ratio of transverse reinforcement ρv. It shall be clarified that concrete
distribution. However, for the portion with large axial force and smaller
over thickness refers to the depth measured from steel section tip to
bending moment, SEM always produce higher prediction than EC4
concrete surface, which can be termed as Cy and Cz along major and
method, but the accuracy of two methods cannot be judged since no
minor axis, respectively. Since all the specimens reported in this paper
relevant test date is available for comparison. This issue will be further
are bending about the major axis, the cover thickness shall satisfy the

8
B. Lai, et al. Engineering Structures 201 (2019) 109841

Fig. 11. Influence of different parameters on the accuracy of EC4 and SEM prediction.

requirement stipulated in EC4 as below: transverse reinforcement ρv is defined as the ratio of stirrups volume
within a link spacing to the volume of concrete core enclosed by the
Cz 0.3h (7)
stirrups. All the parameters selected in this study span a wide range of
where Cz and h are defined in Fig. 10. variation as summarized in Table 2.
As detailed in Table 1, many specimens are encased with large “Ranalysis” values computed from EC4 and Strip Element Method
concrete cover beyond the limit in EC4, especially the ones tested by (SEM) are compared with the test results as shown Fig. 11. It is ob-
Mirza [28]. However, from Fig. 6(a) & (b), it appears that EC4 is still served that SEM can predict the test results with higher accuracy de-
applicable to predict the capacity of such columns. Volumetric ratio of spite the variation of all design parameters discussed above. However,

9
B. Lai, et al. Engineering Structures 201 (2019) 109841

Fig. 12. N-M curved obtained from Strip Element Method (SEM) with and without concrete confinement effect.

EC4 method seems to be sensitive to material strength. For normal

strength materials, EC4 gives conservative estimation with average
value of Ranalysis/Rtest being 0.850, but this ratio increase to 1.336 for
CES columns with high strength concrete (94/104 MPa) and high
strength steel (812/913 MPa), indicating that current EC 4 method is
not safe for designing CES made of high strength materials. As for other
parameters, the data distribution is quite scattered and irregular, and
hence no clear relation can be established between the accuracy of EC4
method with cover thickness ratio, Cz/h, steel contribution ratio, δ,
longitudinal reinforcement ratio, ρl, and lateral reinforcement volu-
metric ratio, ρv. Therefore, it can be concluded that steel and concrete
strengths are the governing parameters affecting EC4 strength predic-
tion, while other parameters do not play a major role in its accuracy.

3.5. Discussion on concrete confinement effect

Concrete confinement exists in concrete well confined by the steel

Fig. 13. Proposal of simplified design method. section and it improves the concrete post peak behavior. Both strength
and ductility of concrete can be enhanced by the confining pressure,
which results from the lateral expansion of concrete and is highly

Table 3
Details of specimens under pure compression.
Ref. Specimen B×D Steel shape fc (MPa) fys (MPa) fyr (MPa) δ ρl (%) ρv (%) Nt (kN) NEC4 (kN) Nt /NEC4

Zhu [9] C-I-M40 200 × 200 100 × 68 × 4.5 × 7.6 93 254 427 0.10 2.50 2.87 3862 3738 1.03
C-I-M60 200 × 200 100 × 68 × 4.5 × 7.6 93 254 427 0.10 2.50 1.91 3789 3738 1.01
C-I-R40 200 × 200 100 × 68 × 4.5 × 7.6 93 254 427 0.10 2.50 1.95 3809 3738 1.02
C-I-R60 200 × 200 100 × 68 × 4.5 × 7.6 93 254 427 0.10 2.50 1.30 3838 3738 1.03
Soliman [50] C4 200 × 200 S.I.B No.10 18.48* 240 400 0.24 1.18 0.72 1050 1040 1.01
C9 200 × 200 S.I.B No.10 18.48* 240 400 0.24 1.18 1.44 1170 1040 1.13
Chen & Yeh [51,52] SRC1 280 × 280 150 × 150 × 7 × 10 29.5 296 350 0.30 3.35 0.68 4220 3809 1.11
SRC2 280 × 280 150 × 150 × 7 × 10 28.1 296 350 0.31 3.35 1.26 4228 3723 1.14
SRC3 280 × 280 150 × 150 × 7 × 10 29.8 296 350 0.30 3.35 2.71 4399 3828 1.15
SRC7 280 × 280 150 × 75 × 5 × 7 28.1 303 350 0.17 3.25 0.68 3788 3142 1.21
SRC8 280 × 280 150 × 75 × 5 × 7 26.4 303 350 0.17 3.25 1.26 3683 3035 1.21
SRC9 280 × 280 150 × 75 × 5 × 7 28.1 303 350 0.17 3.25 0.68 3630 3142 1.16
SRC10 280 × 280 150 × 75 × 5 × 7 29.8 303 350 0.16 3.25 1.26 3893 3250 1.20
Liu [53] CSRC1 400 × 400 150 × 150 × 7 × 10 26.88* 301.7 356.7 0.21 1.57 1.00 5950 5551 1.07
CSRC2 400 × 400 150 × 150 × 7 × 10 35.2* 301.7 356.7 0.18 1.57 1.00 6664 6638 1.00
CSRC3 400 × 400 150 × 150 × 7 × 10 35.2* 301.7 356.7 0.18 1.57 0.50 7009 6638 1.06
CSRC4 400 × 400 175 × 175 × 7.5 × 11 26.88* 296.7 356.7 0.25 1.58 1.00 6517 5830 1.12
CSRC5 400 × 400 175 × 175 × 7.5 × 11 35.2* 296.7 356.7 0.21 1.58 1.00 6771 6909 0.98
Mean 1.09
St.dev 0.07

Note:
*Cubic strength of concrete is reported in literature, and converted to cylinder strength by multiplying 0.8 as recommend in EC2.

10
B. Lai, et al. Engineering Structures 201 (2019) 109841

one being regarded as a result of concrete confinement effect [11,12].

Although most of the current confinement models were derived from
pure compression, it was experimentally verified by Saatcioglu [45]
that the model based on concentric column test can be extended for
columns with strain gradient if the strain-hardening behavior of re-
inforcement and relevant confining parameters are considered. In view
of this, the concrete confinement model developed by Legeron &
Paultre [36] is implemented in the Strip Element Method for further
analysis. In the program, only concrete core is used for calculation since
the concrete cover was delaminated from column as reported by Kim
[11,12]. It is also noteworthy that the additional effect of concrete
confinement in between the steel flanges is neglected here as no cred-
ible model is developed so far. Besides, the confining action from the
steel flanges is quite weak as evidenced by Chicoine et al. [46] ac-
cording to the experimental research on partially encased composite
columns.
Fig. 12 shows the difference between the interaction diagrams
constructed based on unconfined and confined concrete core. The two
curves show good correlation with the first peak loads and second peak
Fig. 14. Evaluation of coefficient, αc. loads, respectively, but they almost coincide with each other at the
portion with high axial force, which is attributed to the ignorance of
associated with the confining pressure and concrete strength [40,41]. It unconfined area when concrete confinement effect is taken into con-
is believed that confinement effect is not significant in the initial sideration. However, in the lower portion of curve with large bending
loading stage and it will be activated after the spalling of the concrete moment and small axis force, prediction of flexural capacity does im-
cover [42,43]. The confined concrete may not always lead to capacity prove because the confinement effect leads to ductility enhancement
improvement since the spalled or loss of concrete cover may possibly allowing for further redistribution of stress in the steel section.
offset the strength enhancement [43,44]. Based on literature review,
only Kim reported two peak loads in the testing of CES beam-columns.
The first peak load corresponds to concrete cover spalling and second 4. A unified method to construct N-M curve

As analyzed above, the strip element method serves as an efficient

numerical technique for tracing the interaction diagrams of composite
columns, and the strain compatibility hypothesis is able to predict the
ultimate capacity of high strength CES beam columns. However, SEM
requires the assistance of computer program to input material non-
linearities and perform the stress integration over large number of strip
elements, which may not be an ideal solution for use in practical design.
In addition, in the absence of relevant test data, the accuracy of SEM for
columns subjected to high axial force and small bending moment re-
mains unknown. To this end, a unified design method is proposed to
predict the cross-sectional capacity of CES columns subject to com-
pression and bending.
The proposed method is simple to use and general enough to cover
concrete strength from 20 to 100 MPa and steel yield strength from 280
to 900 MPa. In this method, four anchor points are generated as plotted
in Fig. 13, including one pure compression point (A) and another three
points (B, C, D) representing three different strain profiles. The pure
bending point “D” (with zero axial force) is not treated as an anchor
point as it involves a time-consuming iteration process, instead, it was
Fig. 15. Determination of effective strength for high strength steel under pure approximated by an adjacent point “D' ” or “D'' ”, representing a key
compression. point locating slightly above or below the pure bending point, respec-
tively.

Fig. 16. Position of neutral axis and the corresponding strain distribution.

11
B. Lai, et al. Engineering Structures 201 (2019) 109841

Fig. 17. Neutral axis at the mid height and the corresponding stress distribution.

Fig. 18. Flow chart to construct N-M diagrams using the proposed design method.

4.1. Pure compression out by Awati & Khadiranaikar [47], or the consideration of premature
spalling failure of concrete cover, especially for high strength concrete
Following the similar formula in EC4, the capacity of CES sections [42–44]. In EN1994-1-1, αc is taken as 0.85 regardless of material
under pure compression is taken as the superposition of plastic re- strength for CES columns. Table 3 summarize the specimen details of
sistance of each material component as shown below: axially compressed CES stub columns reported in literatures. According
to the evaluation on αc as plotted in Fig. 14, αc = 0.85 seems to be
Npl = c fc A c + f ys As + f yr Ar (8) conservative for concrete cylinder strength within the range of
where Npl is the plastic resistance of cross-section. Ac, As, Ar are the area 20–100 MPa. The average ratio of test result to EC4 prediction is 1.09
of concrete, steel section and longitudinal reinforcement bar, respec- with standard deviation 0.07. Therefore, it is proposed that αc = 0.85
tively. can be extended to concrete with cylinder strength up to 100 MPa.
It shall be noted the concrete strength is carefully treated by in- To account for the material compatibility issue, an effective strength
corporating a coefficient αc, which is an important factor to account for of steel is used for computing axial resistance of CES columns. As
the difference between concrete in a cylinder specimen versus a re- plotted in Fig. 15, the effective strength of the steel section, fy , is de-
inforced column, and the variation of concrete strength along the fined as the stress corresponding to the peak strain εco of concrete. For
column length due to non-uniform concrete compaction and curing high strength steel, fy may be in the elastic stage, and thus can be
condition [25]. Another source of αc may be the size effect as pointed formulated as below:

12
B. Lai, et al. Engineering Structures 201 (2019) 109841

Fig. 19. Comparison between SEM and hand-calculation method prediction of interaction diagrams.

fy = min(Es co, f y ) (9) compression part, the concrete ultimate strain, εcu, is used as discussed
forgoing, which can be taken as 0.003 as validated by Khadiranaikar
where Es refers to elastic modulus of steel, which is taken as 210 GPa for et al. [39] and Mertol et al. [54].
structural steel section [48] and 200 GPa [49] for reinforcing steel bars. Fig. 17 gives an example of stress distribution of each material
Therefore, the axial compression resistance of CES composite sec- component when the neutral axis passes through the geometric centroid
tion can be computed as: of CES section (Point “C” in the N-M interaction curve), and it is as-
Npl = 0.85fc Ac + f ys As + fyr Ar sumed the entire steel section remains in elastic stage. The nonlinear
(10)
stress-strain relationship of concrete is represented by a rectangular
The axial resistance obtained from the above equation is plotted as stress block. The effective strength parameter α and the effective height
Point “A” in the N-M interaction curve as shown in Fig. 13. parameter β is calculated using the equation developed by Khadir-
anaikar et al. [39] as below, which is valid for both normal strength and
4.2. Compression and bending. high strength concrete.
= 0.85 1/1000(fc 20)0.85 0.75 (11)
The proposed simplified method uses three additional points “B”,
“C” and “D” to trace the N-M envelop by assuming different positions of = 0.85 1/500(fc 20)0.85 0.67 (12)
neutral axis which corresponds to the respective strain gradient. As il-
lustrated in Fig. 16, these three anchor points corresponds to three For the ease of calculation, the steel section is divided as steel flange
positions of neutral axis, which shift upward from the centroid of and steel web. The strain of steel flange centroid, the tip of steel web
bottom layer reinforcement bars to the centroid axis of CES section, and and reinforcement bar are computed based on the linear strain dis-
the pure bending point is approximated by locating the neutral axis tribution:
away from the section centroid axis with the distance at a quarter of the df × cu
steel web height (at 0.5d from the centroid in Fig. 16). If it is necessary =
f
x (13)
to get the pure bending point, only fine tuning is needed by moving the
neutral axis upward or downward until the desired approximation is dw × cu
=
achieved. Since all the three strain profiles consist of both tension and
w
x (14)

13
B. Lai, et al. Engineering Structures 201 (2019) 109841

=
dr × cu shows that the entire steel section may remain at elastic stage when
r
x (15) the failure load is reached. This observation is especially valid when
where df , dw and dr are the distance between concrete surface to the very high strength steel section such as S800 is used in concrete
centroid of steel flange, tip of steel web, and centroid of reinforcement encased steel composite columns. This suggests that high strength
bars, respectively. X refers to the depth of neutral axis. steel section shall be used in combination with high strength con-
Based on the longitudinal strain calculated from Eqs. (13)–(15), the crete or well-confined concrete to allow economical use of full
corresponding stress can be obtained based on the elastic-perfectly material strengths. Otherwise the steel section remains in elastic or
plastic model, and the axial force and bending moment acting on the partially plastic state, which is the main short coming of the plastic
cross-section can thus be obtained accordingly. For the ease of calcu- design method, such as EC4, in over-predicting the resistance of
lation, the stress on steel web can be separated into elastic and plastic high strength CES composite columns.
portion if any. Details of the calculations are illustrated in the flow 4. A new design method is proposed based on an extensive evaluation
chart shown in Fig. 18. of CES column behavior under different loading conditions. This
The interactional diagrams predicted by SEM and the proposed method can approximate the N-M interaction diagrams with good
method are compared and plotted in Fig. 19. The proposed method accuracy and it serves as a unified method for designing normal
reasonably represents the cross-section N-M capacity, and high accu- strength and high strength CES columns.
racy can be achieved for CES beam columns with normal strength and
high strength materials. Thus, the proposed method functions as a Declaration of Competing Interest
unified approach to generate N-M interaction diagrams for CES columns
with concrete compressive strength within 20–100 MPa and steel yield The authors declared that there is no conflict of interest.
strength ranging from 280 to 900 MPa.
It is noteworthy that the proposed method applies to CES columns Acknowledgement
with double symmetric cross-section. The arbitrary sections inclusive of
off-center steel section, non-symmetric rebar placement, arbitrary The authors would like to acknowledge the support provided by
shape of concrete, and even openings require advanced analysis Building & Construction Authority in Singapore under project grant R-
methods to study the ultimate strength behaviour. Since these irregular 302-000-168-490.
columns are not widely adopted in construction and attract little ex-
perimental research [20], it is excluded in current study. Among these References
arbitrary sections, the off-center steel section may be the most fre-
quently encountered problem in practice. The installation of steel [1] Zhu M, Liu J, Wang Q, Feng X. Experimental research on square steel tubular col-
umns filled with steel-reinforced self-consolidating high-strength concrete under
kingpost could be off-centre in top-down construction. This issue can be axial load. Eng Struct 2010;32(8):2278–86.
addressed using the method proposed by Lai et al. [17]. Alternatively, if [2] Wang Q, Zhao D, Guan P. Experimental study on the strength and ductility of steel
the off-center eccentricity is small, treating the off-center section as a tubular columns filled with steel-reinforced concrete. Eng Struct
2004;26(7):907–15.
symmetrical one by removing the extra asymmetric portion is also [3] EN 1994-1-1. Eurocode 4: design of composite steel and concrete structures-Part 1-
adopted by practitioners, which is also discussed in Ref. [17]. 1. General rules and rules for buildings; 2004.
[4] ANSI/AISC 360-16. Specification for structural steel buildings. Chicago, USA:
American Institute of Steel Construction (AISC); 2016.
5. Conclusion [5] AIJ-2014. AIJ standard for structural calculation of steel reinforced concrete
structures. Architectural Institute of Japan, Japan; 2014 [in Japanese].
Analytical and statistic studies have been conducted to construct the [6] ACI 318-08. Building code requirement for structural concrete (ACI 318–08) and
commentary. USA: American Concrete Institute; 2008.
axial force-moment (N-M) interaction strength surface for the design of [7] JGJ 138-2016. Code for design of composite structures. China: Ministry of Housing
concrete encased steel composite columns made of high strength ma- and Urban-Rural Development of the People's Republic of China; 2016. [in
terials. The accuracy of existing EC4 method was evaluated by con- Chinese].
[8] Lai BL, Liew JYR, Wang TY. Buckling behaviour of high strength concrete encased
ducting sensitivity analyses on key parameters that affect the structural
steel composite columns. J Constr Steel Res 2019;154:27–42.
performance of CES columns. Based on the strain compatibility cri- [9] Zhu WQ, Meng G, Jia JQ. Experimental studies on axial load performance of high-
terion, a computer-assisted program was developed using strip element strength concrete short columns. Proc Inst Civil Eng – Struct Build
2014;167(9):509–19.
method to trace the locus of N-M interaction diagrams. A unified ap-
[10] Lai B, Liew JR, Xiong M. Experimental study on high strength concrete encased steel
proach based on four anchor points was proposed to construct the axial composite short columns. Constr Build Mater 2019;228:116640.
force-bending moment interaction diagrams of both normal strength [11] Kim CS, Park HG, Chung KS, Choi IR. Eccentric axial load testing for concrete-
and high strength CES columns with double symmetric cross-sections. encased steel columns using 800 MPa steel and 100 MPa concrete. J Struct Eng
2012;138(8):1019–31.
The findings reported in this paper support the following conclusions: [12] Kim CS, Park HG, Chung KS, Choi IR. Eccentric axial load capacity of high-strength
steel-concrete composite columns of various sectional shapes. J Struct Eng
1. Based on the analysis of test data from literatures, EC4 method is 2013;140(4):04013091.
[13] Zhu W, Jia J, Gao J, Zhang F. Experimental study on steel reinforced high-strength
applicable for the design of normal strength CES columns with concrete columns under cyclic lateral force and constant axial load. Eng Struct
concrete cylinder strength within 20–65 MPa and steel yield 2016;125:191–204.
strength between 280 and 400 MPa. But the use of EC4 method is [14] Chen CC, Chen CC, Hoang TT. Role of concrete confinement of wide-flange struc-
tural steel shape in steel reinforced concrete columns under cyclic loading. Eng
not conservative for CES columns with high strength concrete (CE60 Struct 2016;110:79–87.
to C100) and high strength steel (S460 to S800). Parametric studies [15] Begum M, Driver RG, Elwi AE. Behaviour of partially encased composite columns
showed that EC4 prediction is not sensitive to steel contribution with high strength concrete. Eng Struct 2013;56:1718–27.
[16] Ma TY, Liu X, Hu YF, Chung KF, Li GQ. Structural behaviour of slender columns of
ratio, concrete cover thickness ratio, longitudinal reinforcement
high strength S690 steel welded H-sections under compression. Eng Struct
ratio and volumetric ratio of transverse reinforcement. 2018;157:75–85.
2. The proposed Strip Element Method (SEM) is able to predict well the [17] Lai B, Liew JY, Li S. Finite element analysis of concrete-encased steel composite
moment-dominant portion of the N-M interaction diagrams for both columns with off-center steel section. In: Proceedings of the 12th international
conference on advances in steel-concrete composite structures. ASCCS 2018; 2018
normal strength and high strength CES columns. The concrete Jun 5. pp. 297–303.
confinement effect can be implemented in SEM and gives good [18] Liu SW, Liu YP, Chan SL. Advanced analysis of hybrid steel and concrete frames:
correlation with test results. Part 1: Cross-section analysis technique and second-order analysis. J Constr Steel
Res 2012;70(3):26–36.
3. Numerical analysis of CES column with high strength steel section [19] Roik K, Bergmann R. Design method for composite columns with unsymmetrical

14
B. Lai, et al. Engineering Structures 201 (2019) 109841

cross-sections. J Constr Steel Res 1990;15:153–68. [37] Liew JYR, Xiong MX. Design guide for concrete filled tubular members with high
[20] Roik K, Bergmann R. Composite columns—design and examples for construction. strength materials – an extension of Eurocode 4 method to C90/105 concrete and
In: Composite and mixed construction. ASCE; 1985. p. 267–78. S550 steel. Res Publ Singapore 2015.
[21] Chiorean CG. A computer method for nonlinear inelastic analysis of 3D composite [38] Liew JYR, Xiong MX, Xiong DX. Design of concrete filled tubular beam-columns
steel–concrete frame structures. Eng Struct 2013;57:125–52. with high strength steel and concrete. Structures 2016;8:213–26.
[22] Papanikolaou VK. Analysis of arbitrary composite sections in biaxial bending and [39] Khadiranaikar RB, Awati MM. Concrete stress distribution factors for high-perfor-
axial load. Comput Struct 2012;98:33–54. mance concrete. J Struct Eng 2011;138(3):402–15.
[23] Chen SF, Teng JG, Chan SL. Design of biaxially loaded short composite columns of [40] Saatcioglu M, Razvi SR. Strength and ductility of confined concrete. J Struct Eng
arbitrary section. J Struct Eng 2001;127(6):678–85. 1992;118(6):1590–607.
[24] El-Tawil S, Deierlein GG. Strength and ductility of concrete encased composite [41] Wang YB, Liew JYR, Lee SC, Xiong DX. Experimental study of ultra-high-strength
columns. J Struct Eng 1999;125(9):1009–19. concrete under triaxial compression. ACI Mater J 2016(1):113.
[25] El-Tawil S, Sanz-Picon CF, Deierlein GG. Evaluation of ACI 318 and AISC (LRFD) [42] Razvi SR, Saatcioglu M. Strength and deformability of confined high-strength
strength provisions for composite beam-columns. J Constr Steel Res concrete columns. ACI Struct J 1994;91(6):678–87.
1995;34(1):103–23. [43] Foster SJ, Liu J, Sheikh SA. Cover spalling in HSC columns loaded in concentric
[26] Fenollosa E, Gil E, Cabrera I, Vercher J. Elastic-plastic formulation for concrete compression. J Struct Eng 1998;124(12):1431–7.
encased sections interaction diagram tracing. Steel Compos Struct [44] Foster SJ. On behavior of high-strength concrete columns: cover spalling, steel fi-
2015;19(4):861–76. bers, and ductility. ACI Struct J 2001;98(4):583–9.
[27] Rocha PAS, Silva KIDA. Construction of the interaction curve of concrete-encased [45] Saatcioglu M, Salamat AH, Razvi SR. Confined columns under eccentric loading. J
composite columns based on the deformation domains of reinforced concrete sec- Struct Eng 1995;121(11):1547–56.
tions. Ibracon Struct Mater J 2015;8:447–66. [46] Chicoine T, Tremblay R, Massicotte B, Ricles JM, Lu LW. Behavior and strength of
[28] Mirza SA, Hyttinen V, Hyttinen E. Physical tests and analyses of composite steel- partially encased composite columns with built-up shapes. J Struct Eng
concrete beam-columns. J Struct Eng 1996;122(11):1317–26. 2002;128(3):279–88.
[29] Ricles JM, Paboojian SD. Seismic performance of steel-encased composite columns. [47] Awati M, Khadiranaikar RB. Behavior of concentrically loaded high performance
J Struct Eng 1994;120(8):2474–94. concrete tied columns. Eng Struct 2012;37:76–87.
[30] Naka T, Morita K, Tachibana M. Strength and hysteretic characteristics of steel- [48] EN 1993-1-1. Eurocode 3: Design of steel structures-Part 1-1. General Rules and
reinforced concrete columns. Trans AIJ 1997;250:47–58. [in Japanese]. rules for buildings; 2005.
[31] Wakabayashi M, Minami K, Komura K. An experiment study on elastic-plastic [49] EN 1992-1-1. Eurocode 2: Design of concrete structures-Part 1-1. General rules and
characteristics of concrete members using an encased H-section subjected to com- rules for buildings; 2004.
bined bending and axial force. Bulletin Disaster Prevent. Res. Inst. Kyoto Univ. [50] Soliman KZ, Arafa AI, Elrakib TM. Review of design codes of concrete encased steel
1971;14A:417–37. [in Japanese]. short columns under axial compression. HSRC J 2013;9:134–43.
[32] Zhao GT, Wang CH, Gao CY, Wang CX. Experiment study on the capacity of SRC [51] Chen CC, Yeh SC. Ultimate strength of concrete encased steel composite columns.
long column subjected to eccentric compression. J. Baotou Univ. Iron Steel Technol. Proceedings of the third national conference on structural engineering. 1996. p.
2006;25(4):384–400. [in Chinese]. 2197–206. [in Chinese].
[33] Yu Q, Lu ZD. Research on the static performance of eccentric steel reinforced [52] Chen CC, Lin NJ. Analytical model for predicting axial capacity and behavior of
concrete column. Build Struct 2009;39(6):34–8. [in Chinese]. concrete encased steel composite stub columns. J Constr Steel Res 2006;62:424–33.
[34] Han DJ, Kim KS. A study on the strength and hysteretic characteristics of steel [53] Liu Y, Guo ZX, Xu PH, Jia LP. Experimental study on axial compression behavior of
reinforced concrete columns-In the case of columns loaded eccentrically. J Arch Inst core steel reinforced concrete columns. J Build Struct 2015;36(4):68–74. [in
Korea 1995;11(4):183–90. [in Korean]. Chinese].
[35] Chen CC, Chen CC. Flexural behavior of steel encased composite beams. J Chin Inst [54] Mertol HC, Rizkalla S, Zia P, Mirmiran A. Characteristics of compressive stress
Civil Hydraulic Eng 2001;13(2):263–75. [in Chinese]. distribution in high-strength concrete. ACI Struct J 2008;105(5):626–33.
[36] Legeron F, Paultre P. Uniaxial confinement model for normal-and high-strength [55] Weng CC, Yen SI. Comparison of concrete-encased composite column strength
concrete columns. J Struct Eng 2003;129(2):241–52. provisions of ACI code and AISC specification. Eng Struct 2002;24:59–72.

15