Sei sulla pagina 1di 14

Engineering Structures 56 (2013) 22–35

Contents lists available at SciVerse ScienceDirect

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

Evaluation of embedded concrete-filled tube (CFT)


column-to-foundation connections
Jiho Moon a, Dawn E. Lehman b,⇑, Charles W. Roeder b, Hak-Eun Lee a
a
School of Civil, Environmental & Architectural Engineering, Korea University, 5-1, Anam-dong, Sungbuk-gu, Seoul 136-701, South Korea
b
Department of Civil & Environmental Engineering, University of Washington, Seattle, WA 98195-2700, USA

a r t i c l e i n f o a b s t r a c t

Article history: Concrete-filled tubes (CFTs) are effective structural components. Relative to conventional reinforced con-
Received 12 April 2012 crete components, they have higher strength-to-size efficiency and facilitate rapid construction. Yet they
Revised 15 April 2013 are not used frequently in US construction. The component response is well understood from extensive
Accepted 16 April 2013
experimental and analytical studies that have been conducted on CFT components themselves. There
remains a fundamental challenge of implementing these components in structural systems, which is
caused by the uncertainty of the connection design and performance. Relative to component research,
Keywords:
limited research has focused on the connections of CFT columns. A recent research effort has resulted
Concrete-filled tubes (CFTs)
Embedded connection
in a new embedded connection for a CFT column anchored into a reinforced concrete foundation. The
Composite structure connection is fully capable of transferring combined bending and axial loads and has sufficient deforma-
Minimum embedment depth bility to sustain multiple inelastic deformation cycles under extreme loading. In addition, the connection
Non-linear finite element analysis has unique features that facilitate constructability and rapid construction. However, the connection
design to date is solely based on experimental study, which evaluated a very limited range of design
parameters. A coordinated analytical study, using high-resolution continuum models, was undertaken
to investigate the unstudied parameters and develop appropriate design expressions, which is the subject
of this paper. An analytical model was developed and verified using the test results. The verified model
was then used to conduct a parametric study to enhance the understanding of the experimental behavior
and extend the test databases. Parameters studied included the embedment depth, diameter-to-thick-
ness (D/t) ratio, shear reinforcement ratio, strength ratio of the concrete in the footing and concrete infill,
and the axial load ratio. From the analysis results, the failure mechanisms were evaluated with respect to
the individual parameters. The results showed that an increase in embedment depth, D/t ratio, shear rein-
forcement ratio, the axial load ratio, or the ratio of the concrete strength in the footing relative to the con-
crete infill of CFT column increased the connection strength and could result in ductile yielding of the CFT
column, which is the desired response mode. Finally, the analysis and test results were combined to
develop a refined design equation for the required embedment depth.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction and developed models appropriate for simulating the CFT compo-
nent response [2–5].
Concrete-filled tubes (CFTs) are composite members, which However, only limited studies have focused on the connection
consist of a steel tube and concrete infill. CFTs used as building col- of the CFT component to adjacent components, such as the founda-
umns and bridge piers have several advantages over conventional tion [6,7]. This limitation in the research has resulted in limiting
structural components, including their strength-to-size efficiency the use of CFT components in US construction. The connections
and facilitation of rapid construction. must be practical and economical. They must develop full compos-
The previous researches are heavily focused on the CFT compo- ite flexural strength to achieve the benefits of CFT members as col-
nent response. Extensive experimental studies on CFT component umns or piers. Further, good ductility or inelastic deformation
subjected to pure axial, pure bending, or combined axial-bending capacity is required of these connections in moderate to high seis-
loading have been conducted and the results are summarized by mic zones.
Roeder et al. [1]. Other researchers have studied CFTs analytically A recent testing effort has resulted in an efficient CFT column to
foundation connection, appropriate for both bridge and building
construction [6]. This connection is capable of transferring the full
⇑ Corresponding author. Tel.: +1 206 715 2108. composite resistance of the CFT column to the footing, while
E-mail address: delehman@u.washington.edu (D.E. Lehman).

0141-0296/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.engstruct.2013.04.011
J. Moon et al. / Engineering Structures 56 (2013) 22–35 23

sustaining inelastic deformation demands resulting from extreme was used to investigate the effects of embedment depth, D/t ra-
loading. The proposed connection is shown in Fig. 1a. Typically, tio, shear reinforcement, concrete strength ratio between the
embedded CFT connections use a full base plate. Here, a steel annu- footing and concrete infill of CFT column, and axial load ratio. Fi-
lar ring is welded to the base of the steel tube to provide anchorage nally, a design equation for minimum embedment depth to en-
of the column into the footing and to transfer the column forces sure ultimate strength of the embedded connection was
and moments into the foundation. Two types of construction evaluated.
methods were explored: (1) monolithic and (2) isolated. For the
monolithic connection, the tube is temporarily supported while 2. Finite element model of embedded CFT connection
the footing is cast around it to embed the steel tube and its annular
ring. The isolated connection casts the footing concrete prior to the The analytical study was conducted using the general-purpose
column concrete, thereby separating the reinforcing cage construc- finite element analysis program ABAQUS [12]. Fig. 2 shows an
tion and the structural steel placement. The isolated connection is example of finite element model of the CFT connection, column
fabricated using the following construction sequence: (1) cast the and foundation. Because of symmetry, a half model of the speci-
footing with a void or recess formed by a corrugated steel tube men was used, which saved modeling and computing time. The
at the column location, (2) lower the CFT column into place, (3) 8-node solid (C3D8R), 4-node shell (S4R), and 2-node truss ele-
grout the column into place with fiber reinforced grout, and (4) fill ment (T3D2) were used to model concrete, steel tube, and steel
the column with self-consolidating low-shrinkage concrete. reinforcement, respectively. The interface between concrete and
Extensive experimental studies were conducted for these con- steel tube was separately modeled explicitly using the GAP ele-
nections [8–11]. The results show that the connections have two ment that is provided by ABAQUS [12] to simulate the contact
different possible failure modes depending on the embedment behavior between the concrete and the steel tube. GAP elements
depth, as shown in Fig. 1b. Pullout failure may occur when the allow relative movement between the steel and concrete while
embedment depth is not sufficient, while punching shear failure restricting penetration of one node into an adjacent one. The nor-
might occur with shallow base depth below the CFT column. With mal stresses generated in the GAP element result in shear stress
sufficient connection strength, both of these failure modes can be transfer between the steel and the concrete fill through friction.
suppressed thereby resulting in ductile yielding of the column, Relative movement (slip) between the internal reinforcing bar
which is the desired response mode. Therefore, it is critical to iden- and the concrete was not modeled; the reinforcement was as-
tify all parameters that influence the failure modes of the connec- sumed to be perfectly bonded, which was modeled using the
tion (Brittle pull-out failure or ductile failure of CFT pier or EMBEDED option in ABAQUS [12].
column). The experimental research alone did not provide suffi- Fig. 3a shows the uniaxial stress–strain relationship of concrete
cient study of all pertinent variables. Thus, a companion analytical used in this study. The compressive stress–strain relationship pro-
study was undertaken. posed by Saenz [13] was employed where it is assumed that the
This paper presents non-linear finite element analysis of the stress–stain relationship is linear up to a stress of 0.5fc0 and the
monolithic embedded connection for the CFT column to enhance maximum compressive strength, fc0 , is achieved when compressive
the understanding of the experimental results by demonstrating strain is 0.003. For the tensile stress–strain relationship of con-
the failure modes of the connection and by extending the exist- crete, the curve proposed by Hsu and Mo [14] was adopted. Tensile
ing test database through analysis of the parameters that have stress–strain relationship is linear uppto ffiffiffiffi stress at cracking of
not been tested such as large D/t ratio and high axial load ratio. concrete, fcr, where fcr is defined as 0.31 fc0 MPa [14] and the soft-
The analysis model was successfully verified. The verified model ening relationship is given by Eq. (1):

Fig. 1. Proposed connection and failure modes of footing: (a) monolithic and isolated connections; and (b) possible failure modes of footing.
24 J. Moon et al. / Engineering Structures 56 (2013) 22–35

Fig. 2. Components of the finite element model: (a) concrete infill and footing; (b) steel tube; (c) interface between infill and tube; and (d) reinforcing bar.

Fig. 3. Material models: (a) concrete; and (b) bare and embedded reinforcing bar.

ecr 0:4 embedded reinforcing bar proposed by Hus and Mo [14] was
fc ¼ fcr ð Þ when ec 6 ecr : ð1Þ
ec adopted, and it is given by

In Eq. (1), ecr is strain at cracking of concrete. Inelastic behavior of fr ¼ Er er when f r 6 fyr0 ðaÞ ð2Þ
concrete was modeled using the concrete damaged plasticity model
proposed by Lee and Fenves [15] available in ABAQUS [12]. This fr ¼ ð0:91  2BÞfyr þ ð0:02 þ 0:25BÞEr er when f r > fyr0 ðbÞ
constitutive model uses a non-associated flow rule and is suitable
where
for simulation of the stress triaxially-dependent plastic hardening.
 1:5
The dilation angle of the concrete, w, which is the measured angle 1 fcr
in the pq  plane where p  are the hydrostatic stress tensor
 and q fyr0 ¼ ð0:93  2BÞfyr ; and B ¼ : ð3Þ
q fyr
and Von-Mises equivalent effective stress tensor respectively, is
an important model parameter in the concrete damaged plasticity In Eq. (2), Er is young’s modulus of the reinforcing bar, fr and er
model. The dilation angle, w, was approximated as 20° and 31° are the stress and strain in the reinforcing bar, respectively, and f’yr
for the concrete infill in the tube and the concrete in the footing, is the reduced yield stress of embedded reinforcing bars. Eqs. (2)
respectively, based on the results of the previous research and para- and (3) are applicable for reinforcement ratios, q, larger than
metric studies [4,15] 0.15% [14]. Eqs. (2) and (3) were derived assuming the tensile
A tri-linear stress–strain relationship was used for the steel stress–stain relationship of the concrete can be expressed as Eq.
in the tube. Young’s modulus, Es, was approximated as (1). The choice of the tensile stress–stain relationship of the con-
200,000 MPa; and Poisson’s ratio, ts, was set as 0.3. The plastic pla- crete can affect the behavior of the concrete structure, and the
teau terminates when strain of the steel, es, is equal to 10 times the parametric study was performed for various types of the tensile
yield strain of the steel (10esy) and stress increases up to ultimate stress–strain relationship of the concrete in this study. From the re-
strength of the steel, fu, which is achieved when the ultimate strain sults, combination of Eq. (1) with Eqs. (2) and (3) was adopted. The
of the steel, esu, is 0.1. The measured properties of the steel and details of the results of parametric study are shown in Section 3.4.
concrete were used to establish the material models when
available. 3. Verification of finite element model
The average stress–strain relationship of reinforcing bar embed-
ded in concrete depends on the effective reinforcement ratio, and 3.1. Test overview
can differ from the properties of bare reinforcing bar for low rein-
forcement ratios as shown in Fig. 3b [14]. The primary difference is The model described above was verified using the test results.
the lower effective yield stress below the yield stress of reinforcing Specifically, the results of the monolithically embedded connection
bar, fyr. In this study, the average stress–strain relationship of tests were compared with analytical results. Fig. 4 shows the
J. Moon et al. / Engineering Structures 56 (2013) 22–35 25

dimensions and reinforcement layout for the test specimens. A nodes at the base of the footing were fully restrained, and the x
spiral weld steel tube with a nominal yield strength of 480 MPa direction of the symmetric section was restrained as shown in
was used in Specimens K1–K3. Specimens K1 and K2 are nominally Fig. 5a. The distributed axial load was applied at the top of the
identical, in which Specimen K2 has shear reinforcement and Spec- CFT on the concrete section using the pressure load option in ABA-
imen K1 does not. Specimen K3 has a deeper embedment depth. In QUS [12]. Then, the distributed lateral displacement was applied
contrast to the K-series, the two remaining specimens used a tube on the 1/4 perimeter of steel tube at the top of the CFT as shown
with a nominal steel strength of 345 MPa where spiral weld tube in Fig. 5a. A mesh convergence study was conducted using three
was used for Specimen 1–50 and straight-seam weld tube was different mesh types as shown in Fig. 5b for K1, K2, and K3 speci-
used for Specimen 4–50. mens where the friction coefficient was set as 0.35.
The height of column above the footing was 1828.8 mm (72 in), The results of the mesh convergence study are summarized in
and the footing was 1930.4 mm (76 in.) in the direction of loading Table 2. In Table 2, Mu,FEM and Mu,test represent the maximum flex-
by 1727.2 mm (68 in.) wide for these specimens. The depth of the ural capacity of the CFT column obtained from finite element anal-
footing was 609.6 mm (24 in.). The diameter of tube was 508 mm ysis and test, respectively. The solutions vary a little for Mesh 1–3,
(20 in.) and the D/t ratio was 80 for all specimens, where D is exter- as shown in Table 2 (the maximum difference is 4% for K3 speci-
nal diameter of CFT and t is the thickness of steel tube, respectively. men). The differences of Mu,FEM/Mu,test between Mesh 2 and Mesh
The size of the footing was determined through preliminary 3 were approximately less than 1% for all analysis models. Thus,
analysis conducted by Kingsley [8]. Kingsley [8] conducted the fi- Mesh 3 was adopted for analysis in this study.
nite element analysis with footing depths from 1.2D to 1.8D and The friction coefficient between steel and concrete can range
showed very little effect on the performance of the connection. from 0.3 to 0.6 [17]. To determine reasonable value of the friction
The 1.2D footing depth develops adequate shear and flexural resis- coefficient for the analysis, a parametric study was conducted for
tance and is proportional to footings used in practice. The flexural K1, K2, and K3 specimen by varying the friction coefficient. The re-
reinforcing bar in the direction of loading consisted of No. 6 bars sults of parametric study are summarized in Table 2. Three differ-
spaced at 101.6 mm (4 in.) for all specimens, as shown in Fig. 4. ent values of friction coefficient (0.25, 0.35, and 0.47) were studied.
The flexural reinforcing bar in the direction perpendicular to load- The analysis results with a friction coefficient of 0.35 provided the
ing consisted of No. 4 bars spaced at 228.6 mm (9 in.) for Speci- best results, as shown in Table 2 with an average discrepancy of
mens K1 and K2, while No. 6 bars spaced 101.6 mm (4 in.) for 4.39%. This value was adopted for the analyses.
Specimens K3, 1–50, and 4–50. The nominal area of No. 4 and 6
bars are 129 and 284 mm2 (0.20 and 0.44 in.2), respectively.
Specimen K1 had no shear reinforcement in the footing, while 3.3. Model validation through comparison with previous test results
No. 3 ties were uniformly used for Specimens K2, K3, 1–50, and
4–50 resulting in a 0.27% shear reinforcement ratio, qv, where Fig. 6 compares the moment-drift relationships obtained from
the nominal area of No. 3 bar is 71 mm2 (0.11 in.2). The material tests with those obtained from finite element analysis. Table 3 also
properties and embedment depth of each specimen are summa- compares analytical results with test results for the five specimens.
rized in Table. 1. In Table 1, le represents the embedment depth. The dashed line in Fig. 6 represents the flexural strength of the CFT
The applied axial load was 0.1Po, where Po is the squash load and column calculated from the plastic stress distribution method
is defined as 0.95fc0 Ac + fyAs where fy is yield stress of steel tube, (PSDM), Mu,PSDM [16]. Drift is defined as the lateral displacement
Ac and As are the areas of concrete infill and steel tube, respectively at the top of specimen divided by the column length above the
[16]. footing. It can be seen that the analysis results accurately predict
stiffness and strength. On average, the model simulated the exper-
3.2. Mesh convergence study including determination of friction imental results well. It should be noted that monotonic loading
coefficient was applied in this study, and post-peak behavior might not be
accurate, since post-peak strength loss is frequently caused by ini-
Fig. 5a and b shows boundary condition and mesh geometries tiation of tearing of the steel tube, which are not included in the
that were used for the mesh convergence study, respectively. The model. However, the post-peak behavior has little effect the max-

Fig. 4. Dimensions of test specimens.


26 J. Moon et al. / Engineering Structures 56 (2013) 22–35

Table 1
Properties of verification models.

Test specimen Study parameter le/D fc0 ,CFT (MPa/ksi) fc0 ,FT (MPa/ksi) fy (MPa/ksi) fu (MPa/ksi)

K1 No shear reinforcement 0.6 75.8/11 75.8/11 525.7/76.3 602.2/87.4


K2 Shear reinforcement 0.6 75.8/11 75.8/11 525.7/76.3 602.2/87.4
K3 Embedment depth 0.9 71/10.3 71/10.3 525.7/76.3 602.2/87.4
1–50 Straight welded tube 0.8 59.9/8.7 75.8/11 337.6/49 413.4/60
4–50 Spiral welded tube 0.8 59.3/8.6 74.4/10.8 351.4/51 537.4/78

Fig. 5. Boundary condition and mesh for analysis model: (a) loading and boundary condition; (b) mesh for analysis model.

Table 2
Results of mesh convergence study and effect of friction coefficient.

Test specimen Mesh convergence study (F.C. = 0.35) Effect of friction coefficient
Mu,FEM/Mu,test Mu,FEM/Mu,test Mu,FEM/Mu,test Mu,FEM/Mu,test Mu,FEM/Mu,test Mu,FEM/Mu,test
(Mesh#1) (Mesh#2) (Mesh#3) (F.C.=0.25) (F.C.=0.35) (F.C.=0.47)
K1 0.98 0.97 0.96 0.86 0.96 1.05
K2 1.07 1.05 1.06 0.95 1.06 1.11
K3 0.93 0.97 0.97 0.93 0.97 0.98
Average discrepancy (%) 8.67 4.39 6.01

imum resistance estimate, which is the primary issue for evaluat- strain is positive. The direction of the vector normal to the crack
ing strength of the CFT column-to-footing connection. plane is assumed to be parallel to the direction of the maximum
Specimens K1 and K2 exhibited pullout failure of the CFT col- principal plastic strain [12]. Specimen K2 exhibited similar damage
umn from the footing [8]. For Specimen K1 (le/D = 0.6, and pattern with larger strength than Specimen K1, as a result of the
qv = 0%), tension yielding and local buckling of the tube above additional vertical reinforcement that was placed in the footing.
the footing was not detected during the test. The theoretical pre- The strength predicted by the finite element analysis was 6% larger
dictions also showed that yielding was not developed and Mu,test/ than the resistance measured in the test. The full flexural capacity
Mu,PSDM is less than 1. The difference between the maximum resis- of the CFT column was developed for specimen K3 (le/D = 0.9, and
tance obtained in the test and finite element analysis was 4% for K1 qv = 0.27%), 1–50 and 4–50(le/D = 0.8, and qv = 0.27%). Further-
as shown in Table 3. Fig. 7 compares the crack formation of the more, large deformation demands were sustained without signifi-
footing, which eventually led to pullout failure, in the experiment cant strength deterioration for these specimens. For all three
and the analysis. In both the experiment and the analysis, cracks specimens, local buckling of the tube occurred near the column
radiating diagonally from the column are observed. The analysis base. Very little damage to the footing was observed in either the
assumed that cracking initiated where the tensile equivalent plas- test or the analysis. Local buckling eventually led to tearing of
tic strain is greater than zero, and the maximum principal plastic the tube, which limited the strength and deformability. Fig. 7
J. Moon et al. / Engineering Structures 56 (2013) 22–35 27

Fig. 6. Comparison of predicted and measured base moment vs. drift: (a) Specimen K1 (b) Specimen K2 (c) Specimen K3 (d) Specimen 1–50 and (e) Specimen 4–50.

Table 3
Comparison of analysis results with test results (verification model).

Test specimen Mu,PSDM (kN m) Mu,test (kN m) Mu,FEM (kN m) Mu,test/Mu,PSDM Mu,test/Mu,FEM Failure mode (test) Failure mode (FEM)
K1 1300.74 1135.42 1091.44 0.87 1.04 Footing failure Footing failure
K2 1300.74 1187.74 1257.58 0.91 0.94 Footing failure Footing failure
K3 1288.20 1451.37 1401.98 1.13 1.04 Local buckling Local buckling
1–50 890.52 1074.40 989.33 1.21 1.09 Local buckling Local buckling
4–50 913.23 996.32 1013.49 1.09 0.98 Local buckling Local buckling

Fig. 7. Simulated and observed damage patterns.

compared the shape of the local buckling observed in the test with finite element analysis were 9 and 2% for 1–50 and 4–50 speci-
that predicted in the analysis for K3, and significant similarity is mens, respectively.
noted. The difference between the test and the finite element anal- Taken as a whole, comparison of the simulation and experimen-
ysis was 4% for K3 specimen. The discrepancy between the test and tal results including the predicted response and damage modes
28 J. Moon et al. / Engineering Structures 56 (2013) 22–35

such as cracking in the footing and local buckling of the tube indi- were analyzed to study the impact of D/t ratio, embedment length,
cates that the finite element modeling approach provides an accu- le, shear reinforcement ratio, qv, the concrete compressive strength
rate simulation. ratio between the footing and infill of the CFT, fc0 ,FT/fc0 ,CFT, and axial
load ratio, P/Po. Specimen K3 shown in Fig. 4 served as the refer-
3.4. Effect of reinforcing ratio on the tensile stress–strain relationship ence model. Nominal material strengths were used. As a result,
of concrete the concrete compressive strength of footing, fc0 ,FT, the concrete com-
pressive strength of the infill of CFT column, fc0 ,CFT, yield stress of steel
An accurate simulation of the tensile stress–strain response is tube, fy, ultimate stress of the steel tube, fu, and yield stress of the rebar,
critical to an accurate analysis of some categories of reinforced fyr, were 34.5, 34.5, 344.8, 482.7, and 413.7 MPa (5, 5, 50, 70, and 60 ksi),
concrete components. In particular, sections with low levels of respectively.
reinforcement and low or no axial load exhibit stronger influences The embedment length ratio le/D was selected based on the re-
from the tensile response. Generally, it is assumed that the tensile sults of the results of previous studies [8–11]. Previous results
strength of concrete is approximately between 8% and 10% of the show that the full flexural strength was observed for most test
concrete compressive strength. The tensile response is modeled specimens when embedment depth is larger than 80% of diameter
linearly to the maximum tensile strength of the concrete; the of CFT column. Thus, to evaluate the impact of the embedment
post-peak response is modeled with the tensile stresses decreasing length ratio, le/D was varied from 0.6 to 0.9. For the base models
linearly to zero at a strain equal to 10 times the peak strain, that is in Table 4, the following values of the key parameters were
the strain corresponding to the maximum tensile strength of the studied:
concrete [12].
In this conventional model, stress–strain curve of bare rebar is  D/t ratios of 60, 80, 100, and 120.
often used. This conventional approach was compared with the  Concrete strength ratios, fc0 ,FT/fc0 ,CFT of 0.7, 1, 1.5, and 2.
modeling approach used in this study, which includes the impact  Shear reinforcement ratios, qv, of 0%, 0.15%, 0.27% and 0.5%.
of embedded reinforcing bars (Eqs. (1)–(3)) [14]. Fig. 8 shows the  Axial load ratios, P/Po, of 0.1, 0.2, and 0.3.
effects of tensile stress–strain relationship of the concrete and  Column diameters, D, of 508, 762, and 1524 mm (20, 30, and
stress–stain curve of the reinforcing bar on the analysis results. 60 in.).
The cracking stress, fcr, is assumed to be 9% of fc0 for the conven-
tional method. It was found that both conventional method and It should be noted that the footing dimension and total amount
embedded-bar modeling approach used in this study provided of reinforcement were proportioned to D.
good results for Specimen K3 where minimal to modest footing
cracks occurred. However, the results were improved using the
4.2. Evaluation of different failure mechanisms
embedded-rebar model for Specimen K2, which exhibited more
significant cracking. For this specimen, conventional method over-
Fig. 9 shows base moment-drift relationship for two different
estimated the strength, as shown in Fig. 8b; the embedded model-
failure modes. The x and y axes in Fig. 9(a) denote drift and normal-
ing approach resulted in predictions that agreed well with test
ized flexural strength of the embedded connection, Mu,FEM/Mu,PSDM,
results. The results of embedding low levels of reinforcement in
respectively. The value of Mu,PSDM is the theoretical flexural
concrete changes the overall curve of the bar, as shown in
strength of the CFT column calculated from plastic stress distribu-
Fig. 3b, and results in a larger strain corresponding to the yield
tion method [16]. The model-naming scheme denotes the key vari-
stress, which becomes more significant with decreases in the rein-
ables studied. Results of two models are shown in Fig. 9, where
forcement ratio. As a result, combination of Eq. (1) with Eqs. (2), (3)
Model CFT60-0.6-0-1 had a D/t ratio of 60, embedment depth ratio
provided good results regardless of the failure modes of the CFT
(le/D) of 0.6, shear reinforcing ratio of 0%, and a concrete strength
column-to-foundation connection.
ratio of 1. In contrast, Model CFT80-0.9-0.27-1 has a D/t ratio of
80, embedment depth ratio (le/D) of 0.9, shear reinforcement (qv)
4. Parametric study of 0.27%, and a concrete strength ratio of 1.
The results shown in Fig. 9 indicate two different failure modes
4.1. Variables for parametric study of the CFT connection. Model CFT60-0.6-0-1 has relatively larger
normalized flexural strength of CFT column than that of CFT80-
A series of parametric studies were performed to investigate the 0.9-0.27-1. For Model CFT60-0.6-0-1, Mu,FEM/Mu,PSDM is 0.75 indi-
behavior of the embedded CFT foundation connection. Table 4 pro- cating that the full flexural capacity of the CFT was not developed
vides values for the key variables evaluated. A total of 56 models while Mu,FEM was 13% larger than the theoretical flexural strength

Fig. 8. Effects of tensile stress–strain relationship of the concrete and stress–stain curve of the reinforcing bar on the analysis results: (a) K3 specimen; (b) K2 specimen.
J. Moon et al. / Engineering Structures 56 (2013) 22–35 29

Table 4 cracking initiates but is limited, resulting in ductile response. The


Base models for parametric study. initial crack formed at the tip of the annular ring with an angle
Model D (mm/in) D/t le/D qv (%) fc0 ,FT (Mpa/ksi) P/Po of approximately 45°. However, the cracking was much more lim-
CFT60-0.6-0.27-1 508/20 60 0.6 0.27 34.5/5 0.1
ited than Models CFT60-0.6-0-1, and the deformability of the sys-
CFT60-0.7-0.27-1 508/20 60 0.7 0.27 34.5/5 0.1 tem came from the extensive yielding of the steel tube in tension
CFT60-0.8-0.27-1 508/20 60 0.8 0.27 34.5/5 0.1 and compression, as shown in Fig. 10b. The results show that the
CFT60-0.9-0.27-1 508/20 60 0.9 0.27 34.5/5 0.1 shear capacity of the footing must be considered in the design to
Note: fy = 344.8 MPa (50 ksi); fu = 482.7 MPa (70 ksi); fyr = 413.7 MPa (60 ksi); and achieve full plastic action in the column resulting in maximum
fc0 ,CFT = 34.5 MPa (5 ksi) for all analysis models. drift capacity.

4.3. Evaluation of parameter impact

Mu,PSDM for Model CFT80-0.9-0.27-1, indicating that the model


The impact of each of the study parameters was assessed
achieved the desired response mode of column yielding. The key
through a series of parametric study. Fig. 11 shows the effect of
difference in these two models is the location of the initiation
D/t ratio, where D/t ratio is plotted on the x-axis. Three different
and progression of damage. For Model CFT60-0.6-0-1, the footing
parameters are used to assess the parameter impact and these
is severely damaged and pullout failure occurs, as illustrated in
parameters are plotted on the y-axis. The parameters were the nor-
Fig. 9b. For Model CFT80-0.9-0.27-1, inelastic action is sustained
malized strength ratio (Mu,FEM/Mu,PSDM) (Fig. 11a), the normalized
by the steel tube with yielding, and local buckling of steel tube
uplift of the footing where the uplift is normalized to H, the height
controls the strength of the specimen while mitigating footing
of the footing (Fig. 11b), and normalized out-of-plane deformation
damage because of the increased embedment depth.
(OPD) of the steel tube, where the OPD is normalized to the column
These differences are studied in more depth in Fig. 10. The dis-
diameter (Fig. 11c). Uplift and out-of-plane deformation of the
tribution of stresses and crack pattern of footing (middle section)
steel tube were measured at the interface between the CFT column
at 4% drift are shown in Fig. 10a and b for Models CFT60-0.6-0-1
and the footing in tension side, and at a height of 76.2 mm (3 in.)
and CFT80-0.9-0.27-1, respectively. When the embedment depth
above the footing on the compression side, respectively (see
was shallow without shear reinforcement (Model CFT60-0.6-0-1),
Fig. 9b for illustration of these key points). The following observa-
significant cracking was noted on the tension side of column in
tions are made: (1) The flexural strength ratio, Mu,FEM/Mu,PSDM, is
the footing; this cracking was severe and resulted in lost tensile
not impacted by the D/t ratio for well-anchored connections (e.g.,
resistance. Significant cracking permitted separation of the
le/D = 0.9), (2) The uplift is similarly constant for well-anchored
concrete infill and footing and resulted in increased compression
connections (e.g., le/D = 0.9), (3) The out-of-plane displacement in-
stresses at the bottom of the annular ring and interface between
creases significantly with D/t ratio, regardless of the anchorage
the CFT and footing in compression side, as shown in Fig. 10a. As
depth.
a result, a portion of the steel tube that was in compression had lo-
These figures also illustrate the impact of normalized embed-
cally yielded. In contrast, Model CFT80-0.9-0.27-1, which had ade-
ment depth on the response. Each connected line represents one
quate embedment depth and shear reinforcement, the footing
le/D ratio, as indicated by the symbols and the legends. In contrast

Fig. 9. Simulation results for models CFT60-0.6-0-1 and CFT80-0.9-0.27-1: (a) moment–drift relationships; and (b) damage modes.
30 J. Moon et al. / Engineering Structures 56 (2013) 22–35

Fig. 10. Stress distributions and crack patterns for analysis models: (a) CFT60-0.6-0-1; and (b) CFT80-0.9-0.27-1 (Note: 1 ksi = 6.89 MPa).

Fig. 11. Effect of D/t ratio: (a) Mu,FEM/Mu,PSDM vs. D/t; (b) uplift/H vs. D/t; and (c) O.P.D/D vs. D/t.

to the D/t ratio, the embedment depth has a much more significant with increasing le/D, but this is expected since, for lower le/D ratios,
impact on the response. The flexural strength of the system re- the response of the system is controlled by damage in the footing
duces for lower le/D, while the uplift increases. The OPD decreases and therefore tube buckling is limited.
J. Moon et al. / Engineering Structures 56 (2013) 22–35 31

Fig. 12 shows the effect of concrete compressive strength ratio 5. Design of embedded CFT connection
of the footing and CFT column infill, fc0 ,FT/fc0 ,CFT. Larger fc0 ,FT/fc0 ,CFT ratio
results in a stronger concrete strength of the footing. Although this is An empirically-derived design equation to estimate the mini-
contrary to practice, the results show that a ratio of 1 is required to mum embedment depth, le, was proposed based on a series of test
achieve the flexural strength. Furthermore, Fig. 12 suggests that an in- results [8–11]. The expression was derived using geometry and
crease in fc0 ,FT/fc0 ,CFT ratio reduces the damage of the footing and is helpful equilibrium of the connection corresponding to pullout failure, as
to increase Mu,FEM/Mu,PSDM. This is a similar trend to that identified with depicted in Fig. 16. The tube is embedded a length le. The annular
increasing D/t ratio. Larger compressive capacity of the footing concrete ring has an outer diameter of Do. For the derivation of the design
also reduced the uplift (Fig 12b). The OPD increased with increasing the equation, shear stress angle b1 in the concrete footing is approxi-
strength ratio, since the larger footing strength encourages yielding and mated as 45°. (Note, this angle has been verified by the finite ele-
buckling of the steel tube. ment analysis, for which the crack that initiates on the tension side
The impact of the shear reinforcement on the response is shown of the column forms at an angle of approximately 45°.)
in Fig. 13. In all cases, models without shear reinforcement failed to Equilibrium in vertical direction is given by the following
achieve their flexural strength. For lower values of le/D, the impact equation:
of the shear reinforcement in the footing was more significant. For
example, the models with an le/D = 0.6 showed an increase in T s ¼ V c sin b1 : ð4Þ
strength with additional shear reinforcement; for models with lar- In the expression, Ts and Vc are the tensile force in the steel and
ger le/D ratio, the increase was less significant. shear force in the concrete, respectively. For an incremental angle
The effects of axial load ratio are summarized in Fig. 14. In gen- of Dc, Ts and Vc are given by
eral, Mu,FEM/Mu,PSDM was increased with increasing P/Po when Mu,-
FEM/Mu,PSDM is less than one (lower anchorage lengths with D le
T s ¼ Dc fs t and V c ¼ sc Dc ðD0 þ le cot b1 Þ: ð5Þ
significant impact of the footing response). For well-anchored 2 2 sin b1
models, increasing P/Po beyond 0.2 does not increase the strength
ratio. However, in all cases increasing P/Po reduced the uplift defor- Substituting Eq. (5) into Eq. (4) with b1 = 45°, shear stress of the
mation of the footing. In most cases, increasing P/Po increased the footing, sc, that is needed to develop stress of fs in steel tube can
OPD. be computed using Eq. (6).
In contrast to the expected response, for well-anchored models Dtfs
(le/D = 0.9), the OPD decreased with increasing values of P/Po. This sc ¼ 2
: ð6Þ
ðle þ D0 le Þ
trend is explored further by comparing two models with two dif-
ferent axial load ratios of 0.1 and 0.2 at 4% drift, as shown in It is noted that shear stress of the footing, sc, is the same with the
Fig. 15. The figure plots the maximum principle plastic strain gra- shear stress of the concrete in the footing since the shear reinforcing
dients that represent the cracking in the concrete infill and footing. bar is ignored in Eq. (6). To achieve full flexural strength of the CFT
The figure shows that the compressive region of the concrete infill column, the steel tube must fully yield, ideally reaching its ultimate
increases with an increase in the axial load ratio, suggesting that an strength, fu. Thus, the shear stress capacity, sc, was computed by
increased amount of concrete is engaged to sustain the combined substituting fu for fs in Eq. (6).
demands from tension and axial compressive loading (P). In addi- Significant variability exists in the shear stress capacity of the
tion, the arc of the tube engaged in sustaining the compressive concrete. For ffi reason, Lee [11] provided that sc should be smal-
this
qffiffiffiffiffiffiffi
force is also increased. The combination of these responses results ler than a fc;FT
0
to achieve full flexural capacity of the CFT column,
in decreasing the simulated OPD. However, additional displace- and a varies from 0.42 to 0.57 (MPa unit). However, that proposal
ment demand that may result in significant damage to the concrete was based on limited experimental data. In this study, this shear
infill accelerates buckling of the steel tube. This latter behavior has stress range was reexamined and calibrated by using results of
been observed by others [18]. It is also noted that the tensile parametric study as well as the experimental results. Fig. 17 shows
strains in some cracked region are very high in Fig. 15. This is be- the evaluation of design equation to estimate the minimum
cause that the slope of the tensile stress–strain curve is almost zero embedment depth. In Fig. 17, the x-axis is the normalized shear
qffiffiffiffiffiffiffi

when the tensile strain is large [approximately 0.01 from Eq. (1)]. stress, sc/ fc;FT
0
, where sc is calculated from Eq. (6) with fs equal
Thus, tensile strain could be dramatically increased even if the to fu. As expected, the results show that Mu,FEM/Mu,PSDM decreases
decrease in tensile stress is very small. Finally, the impact of the and uplift increases (equivalent to increased damage sustained
qffiffiffiffiffiffiffi

column diameter (D) was investigated by studying CFT columns by the footing) with increasing sc/ fc;FT 0
. Also, out-of-plane defor-
qffiffiffiffiffiffiffi

with diameters of 508, 762, and 1524 mm (20, 30, and 60 in.). mation of the steel tube increases with decreasing sc/ fc;FT 0
as
There was no impact on the response. shown in Fig. 17c, which also indicates that higher shear stress

Fig. 12. Effect of fc0 ,FT/fc0 ,CFT: (a) Mu,FEM/Mu,PSDM vs. fc0 ,FT/fc0 ,CFT; (b) Uplift/H vs. fc0 ,FT/fc0 ,CFT; and (c) O.P.D/D vs. fc0 ,FT/fc0 ,CFT.
32 J. Moon et al. / Engineering Structures 56 (2013) 22–35

Fig. 13. Effect of qv: (a) Mu,FEM/Mu,PSDM vs. qv; (b) Uplift/H vs. qv; and (c) O.P.D/D vs. qv.

Fig. 14. Effect of P/Po: (a) Mu,FEM/Mu,PSDM vs. P/Po; (b) Uplift/H vs. P/Po; and (c) O.P.D/D vs. P/Po.

Fig. 15. Variation of crack patterns for different axial load ratio (4% drift).

in the footing limits the inelastic action in the tube, and potentially including regions of low and moderate seismicity, as indicated
changes the response mode from ductile yielding of the column to below.
brittle response of the connection. For regions of low and moderate seismicity
From the comparative study, test results of monolithically qffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffi
Dtfu
embedded connection and analysis results without shear reinforc- 2
6 a fc;FT
0 0
¼ 0:6 fc;FT MPa ð7:2 for psi unitÞ ðaÞ ð7Þ
ing bar provide a lower bound on the embedment depth. A linear ðle þ D 0 le Þ
regression was performed using these data (Linear Regression_1 For regions of high seismicity
in Fig. 17a). The value of Mu,FEM/Mu,PSDM was larger than 1 when qffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffi
Eq. (7a) is satisfied. These results are applicable to foundation Dtfu
2
6 a fc;FT
0 0
¼ 0:55 fc;FT MPa ð6:6 for psi unitÞ ðbÞ
designs for which ductility in the tube is not a design requirement, ðle þ D 0 le Þ
J. Moon et al. / Engineering Structures 56 (2013) 22–35 33

Fig. 16. Schematic view of footing resistance to cone pullout failure.

Fig. 17. Evaluation of design equation: (a) Mu,FEM/Mu,PSDM vs. sc/(fc0 ,FT)0.5; (b) Uplift/H vs. sc/(fc0 ,FT)0.5; and (c) O.P.D/D vs. sc/(fc0 ,FT)0.5.

Satisfaction of this minimum does not guarantee significant plastic respectively. Thus, this target strength ratio for high seismic region
action in the CFT column, since CFT column-to-foundation connec- may not be appropriate for the CFT column-to-foundation connec-
tion must be fully yielded and provide the significant inelastic tion that deviates from the profiles of test specimens used in this
deformation capacity. From the observation of the test and finite study, and the care should be taken to apply the proposed equation
element analysis results, it was found that the full strength of the for high seismic region.
CFT column was usually 10–20% larger than theoretical value. Full Scatter is observed in the data presented in Fig. 17. The simula-
plastic action occurred when Mu,test/Mu,PSDM is larger than 1.15. Ta- tions suggest that the shear reinforcement ratio and axial load ratio
ble 5 shows the summary of drift ductility, l, for the test results can reduce the required embedment depth. However, these factors
conducted [9–11], where drift ductility is defined as the ratio of were neglected in the derivation of Eq. (6). Eq. (8) was derived by
maximum amount of deflection (drift) divided by the amount of conducting a regression analysis of the results of parametric study.
deflection (drift) at yield. The yield point of these specimens was
Dtfu
defined as the point where the steel strain equals the yield strain ð1 þ qv Þ0:46 ð1 þ P=Po Þ0:51 sc;m ¼ 2
: ð8Þ
of the material. The maximum deformation was defined as the ðle þ D0 le Þ
point prior to 20% loss in lateral strength. On average, the drift duc- In Eq. (8), sc,m is the modified shear stress of the concrete in the
tility of the CFT column was 8.8. footing that is needed to develop stress of fu in steel tube, and qv
Thus, target strength ratio of CFT column-to-foundation con- is the shear reinforcement ratio in percent (%). In Eq. (8), the value
nection was selected as 1.15Mu,PSDM for more extreme loading. of qv never exceeds 0.3%, and the value of P/Po cannot exceed 0.2
Then, an alternative linear regression was conducted using analysis since little benefit of these parameters were observed for a well-an-
results with shear reinforcing bar and isolated embedded connec- chored CFT-footing connection from the analysis. The shear stress
tion test data (Linear Regression_2 in Fig. 17a), because the scatter range of embedded CFT connection was evaluated by using Eq. (8)
was considerably reduced when these data were used for regres- and the results are shown in Fig. 18. The variability in the prediction
q ffiffiffiffiffiffiffiffi The ratio of Mu,FEM/Mu,PSDM was larger than 1.15 when sc/
sion. was reduced by using Eq. (8). Similar with Eq. (7), the shear stress
0
fc;FT 6 0.55 (MPa unit). This value is recommended for high seis- range was evaluated and the results are given by
mic regions as described in Eq. (7b). It should be noted that target For regions of low and moderate seismicity
strength ratio of CFT column-to-foundation connection was qffiffiffiffiffiffiffiffi
adopted as 1.15Mu,PSDM for high seismic region based on the test re- Dtfu
2
6 a fc;FT
0
sults shown in Table 5 since they show appropriate ductility level ðle þ D 0 le Þ
that show the plastic action. However, these tests have the clear
¼ 0:55ð1 þ qv Þ0:46 ð1 þ P=Po Þ0:51
limit about the details and material properties of tested specimens. qffiffiffiffiffiffiffiffi
For example, D/t ratio and qv were limited to 80% and 0.27%, 0
 fc;FT MPa ð6:6 for psi unitÞ ð9aÞ
34 J. Moon et al. / Engineering Structures 56 (2013) 22–35

Table 5
Drift ductility of the test specimens where Mu,test/Mu,PSDM P 1.15.

Test specimen D (mm/in.) le/D D/t fy (MPa/ksi) fc0 ,CFT (MPa/ksi) fc0 ,CFT/fc0 ,FT qv (%) P/Po Mu,test/Mu,PSDM l
W5 [9] 508/20 0.9 80 520.9/75.6 77.2/11.2 1 0.27 0.1 1.15 7.1
W6 [9] 508/20 0.75 80 520.9/75.6 81.3/11.8 1 0.27 0.1 1.22 8.3
C1 [10] 508/20 0.9 80 520.9/75.6 67.5/9.8 1 0.27 0.1 1.27 6.3
C2 [10] 508/20 0.9 80 520.9/75.6 66.8/9.7 1 0.27 0.1 1.30 7.3
C4 [10] 508/20 0.9 80 520.9/75.6 68.2/9.9 1 0.27 0.2 1.23 7.8
1–50 [11] 508/20 0.8 80 337.6/49 59.9/8.7 0.79 0.27 0.1 1.17 13.5
2–50 [11] 508/20 0.775 80 337.6/49 64.1/9.3 0.82 0.27 0.1 1.17 11.6
5–50 [11] 508/20 0.7 80 351.4/51 67.5/9.8 0.84 0.27 0.1 1.31 9.2
6–50 [11] 508/20 0.6 80 351.4/51 69.6/10.1 0.86 0.27 0.1 1.22 8.2

Fig. 18. Evaluation of modified design equation: (a) Mu,FEM/Mu,PSDM vs. sc,m/(fc0 ,FT)0.5; (b) Uplift/H vs. sc,m/(fc0 ,FT)0.5; and (c) O.P.D/D vs. sc,m/(fc0 ,FT)0.5.

For regions of high seismicity This research extended those studies by conducting non-linear
qffiffiffiffiffiffiffiffi finite element analyses to investigate the impact of salient design
Dtfu
2
6 a fc;FT
0 parameters on the embedded foundation connection for the CFT
ðle þ D0 le Þ columns. The finite element model was verified using measured re-
¼ 0:47ð1 þ qv Þ0:46 ð1 þ P=Po Þ0:51 sponse and observed damage of prior test results, including crack-
qffiffiffiffiffiffiffiffi ing of the footing and local buckling of the steel tube.
0
 fc;FT MPa ð5:7 for psi unitÞ: ð9bÞ The analytical study revealed two different failure modes, both
of which were observed in the testing. For models with shallow
Thus, the minimum embedment depth of the CFT connection can be embedment depths and insufficient shear reinforcement in the
obtained by solving Eq. (9) to include the benefits of shear rein- footing, heavy cracking was observed adjacent to the tension side
forcement and axial load. a in Eq. (9a) is amplified up to 0.68 of the CFT column in the footing and this cracking led to loss of
(MPa unit) by introducing the shear reinforcement ratio of 0.3% strength prior to development of the theoretical flexural strength.
and axial load ratio of 0.2. In the case of a in Eq. (9b), a varies from In contrast, when the embedment depth was sufficiently deep and
0.47 to 0.58 (MPa unit). Lee [11] suggested a is between 0.42 and shear reinforcement were provided, the steel tube fully yielded in
0.57 (MPa unit) and his proposed range for a is similar with that tension and compression; in that case the CFT column controlled
from this study. the response of the system with local buckling and yielding of
ACI 318 code [19] provides the critical failure shear force in Sec- the tube.
tion 11.2.1. Similarly with thus study, theq critical
ffiffiffiffiffiffiffiffi shear stress spec- With a validated model in hand, a parametric study was per-
ified in ACI [19] can be expressed as sc/ fc;FT 0
= Ck where C varies formed to study the effects of D/t ratio, embedment length, shear
from 2 to 4 (in psi unit) and k is 1 for normal concrete. This dimen- reinforcement ratio, concrete compressive strength ratio of the
sionless critical shear stress range was also plotted with analysis footing and infill in the CFT, and axial load ratio. The analysis re-
results and test results, as shown in Figs 17 and 18. From the com- sults showed that: (1) The embedment depth had the most dra-
parative study, it was found that ACI318 [19] give a conservative matic impact on the response of the CFT column-to-footing
value for pullout failure of the CFT column-to-foundation subassemblies, (2) The flexural strength ratio, Mu,FEM/Mu,PSDM, was
connection. not impacted by the D/t ratio for well-anchored connections (e.g.,
le/D = 0.9) while D/t ratio had significant effect on the Mu,FEM/Mu,-
6. Summary and conclusions PSDM for shallow embedment depth (e.g., le/D = 0.6), (3) For models
with lower embedment depths, the impact of the shear reinforcing
Concrete filled tubes offer substantial benefits over conven- bar was more significant, (4) An increase in fc0 ,FT/fc0 ,CFT ratio reduced
tional construction methods because of their inherent strength, the damage of the footing and was helpful to increase Mu,FEM/Mu,-
stiffness, stability all of which contribute to reduced labor and PSDM, and (5) In general, Mu,FEM/Mu,PSDM was increased with increas-
material requirements. However, CFT components are seldom used ing P/Po when Mu,FEM/Mu,PSDM was less than one. However, for
in part because there are few validated constructible connections. well-anchored models, increasing P/Po beyond 0.2 did not increase
Prior experimental studies have developed a feasible CFT-to-foun- the strength ratio.
dation connection. However, the data were limited and the full Two design equations to estimate the minimum embedment
range of design parameters could not be evaluated. depth were derived using the analytical results and previous test
J. Moon et al. / Engineering Structures 56 (2013) 22–35 35

data. From the results, the minimum embedment depth that guar- J Struct Eng ASCE; in press. http://dx.doi.org/10.1061/(ASCE)ST.1943-
541X.0000788.
antees full flexural strength of CFT column from plastic stress dis-
[6] Lehman DE, Roeder CW. Foundation connections for circular concrete-filled
tribution method can be obtained by solving Eq. (7a). For regions of tubes. J Construct Steel Res 2012;78:212–25.
high seismicity, Eq. (7b) is recommended. A modified design equa- [7] Pecce M, Rossi F. The non-linear model of embedded steel-concrete composite
tion was proposed to include benefits of shear reinforcement and column bases. Eng Struct 2013;46:247–63.
[8] Kingsley A. Experimental and analytical investigation of embedded column
axial load ratio as Eq. (9) and the scatter of analysis and test results base connections for concrete filled high strength steel tubes. MS thesis.
was reduced. Seattle, WA: Univ. of Washington; 2005.
[9] Williams TS. Experimental investigation of high strength concrete filled steel
tubes in embedded column base foundation connections. Seattle, WA: Univ. of
Acknowledgements Washington; 2006.
[10] Chronister A. Experimental investigation of high strength concrete filled steel
This work was completed partially with funding provided by tubes in embedded column base foundation connections. MS thesis. Seattle,
WA: Univ. of Washington; 2007.
the Washington State Department of Transportation (WSDOT) [11] Lee JR. Experimental investigation of embedded connections for concrete-
through the project entitled, ‘‘Design of Bridge Foundations with filled steel tube columns subjected to combined axial-flexural loading. MS
Steel Casings’’. Mr. Bijan Khaleghi is the WSDOT Bridge Design thesis. Seattle, WA: Univ. of Washington; 2011.
[12] ABAQUS. ABAQUS analysis user’s manual version 6.9-2 2009. Providence, RI,
Engineer and the coordinator of this project. The advice and finan- USA: Dassault Systemes Simulia Corp.
cial support of the WSDOT is gratefully acknowledged. [13] Saenz LP. Discussion of ‘equation for the stress-strain curve of concrete’ by P.
Desayi, and S. Krishnan. ACI J 1964;61:1229–35.
[14] Hsu TTC, Mo Y-L. Unified theory of concrete structures. 2nd ed. John Wiley &
References
Sons, Inc.; 2010.
[15] Lee J, Fenves GL. Plastic-damage model for cyclic loading of concrete
[1] Roeder CW, Lehman DE, Bishop E. Strength and stiffness of circular concrete structures. J Eng Mech ASCE 1998;124(8):892–900.
filled tubes. J Struct Eng ASCE 2010;136(12):1545–53. [16] AISC. Specifications for structural steel buildings, Chicago, IL; 2008.
[2] Hu HT, Huang CS, Wu MH, Wu YM. Nonlinear analysis of axial loaded concrete- [17] Baltay P, Gjelsvik A. Coefficient of friction for steel on concrete at high normal
filled tube columns with confinement effect. J Struct Eng ASCE stress. J Mater Civ Eng ASCE 1990;2(1):46–9.
2003;129(10):1322–9. [18] Marson J, Bruneau M. Cyclic testing of concrete-filled circular steel bridge piers
[3] Lu H, Han LH, Zhao XL. Analytical behavior of circular concrete-filled thin- having encased fixed-base detail. J Bridge Eng ASCE 2004;9(1):14–23.
walled steel tubes subjected to bending. Thin-Walled Struct 2009;47:346–58. [19] ACI. Building code requirements for structural concrete and commentary,
[4] Moon J, Lehman DE, Roeder CW, Lee H-E. Analytical modeling of bending of ACI318-11 2011, Farmington Hills, MI.
circular concrete-filled steel tubes. Eng Struct 2012;42:349–61.
[5] Moon J, Lehman DE, Roeder CW, Lee H-E. Strength of circular concrete-filled
tubes (CFT) with and without internal reinforcement under combined loading.

Potrebbero piacerti anche