You are on page 1of 11

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/257801424

Stiffness requirements for transverse stiffeners of rectangular CFT


compression panels

Article  in  International journal of steel structures · June 2013


DOI: 10.1007/s13296-013-2006-5

CITATIONS READS

0 389

3 authors, including:

Kyoung-Chan Lee Junsuk Kang


Korea Railroad Research Institute Seoul National University
34 PUBLICATIONS   77 CITATIONS    23 PUBLICATIONS   178 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Development of Next-Generation Smart Railroad Bridge View project

All content following this page was uploaded by Junsuk Kang on 15 February 2014.

The user has requested enhancement of the downloaded file.


International Journal of Steel Structures
June 2013, Vol 13, No 2, 1-10
DOI 10.1007/s13296-

www.springer.com/journal/13296

Stiffness requirements for transverse stiffeners of


rectangular CFT compression panels
Kyoung Chan Lee1, Junsuk Kang2, and Chai Hong Yoo3*
1
Senior Researcher, Korea Railroad Research Institute, Uiwang, 437-757, Korea,
formerly, Postdoctoral Fellow, Department of Civil Engineering., Auburn Univ., Auburn, AL 36849-5337, USA
2
Assistant Professor, Department of Construction Management and Civil Engineering, Georgia Southern University,
Statesboro, GA 30460-8047, USA
3
Professor Emeritus, Department of Civil Engineering, Auburn University, Auburn, AL 36849-5337, USA

Abstract

This paper presents a more rational calculation of the minimum stiffness required for transverse stiffeners of stiffened
concrete filled tube (CFT) compression panels. As longitudinal stiffeners are essentially compression members, transverse
stiffeners are required to control the effective length. Both the longitudinal stiffener and transverse stiffener subdivide the
compression panel in a grid pattern so the relatively thin plate can carry the induced compressive load in the most efficient
manner. However, the provisions for the required rigidity of transverse stiffeners in the literature are inconsistent in general and
in particular; do not exist in the case of the concrete filled compression panels. Here, the parameters that govern the behavior
of the transverse stiffeners are identified theoretically using the column buckling approximation. In order to calibrate and
quantify the analytical equations developed, incremental nonlinear analyses were performed on a large number of finite-element
models. The numerically collected data were then used to validate the equations developed.

Keywords: Buckling, CFT, Compression panel, Finite element analysis, Steel plates, Stiffener

1. Introduction strength of such a structural member is the local buckling


strength (plate behavior) of the relatively thin steel skin.
Frequently considered in recent years are concrete- Intuitively, an internal stiffening system of longitudinal
filled columns (fully or partially) for their superior energy and transverse stiffeners is envisioned to improve the
dissipation characteristics during an earthquake as shown local buckling strength. As the longitudinal stiffeners are
in Fig. 1. These columns will also improve the local essentially compression members, they need to be braced
buckling strength of the thin skin two and a half times by transverse stiffeners in order to control their effective
over the stiffened columns without the concrete. The thin- length. Both the longitudinal and transverse stiffeners
walled panel can only buckle (bulge) out from the concrete subdivide the compression panel in a grid pattern, allowing
core after it has hardened. The buckling mode shape of a the relatively thin plate to carry the induced compressive
panel with multiple stiffeners resembles a waffle slab. load in the most efficient manner. As it is frequently the
The effective boundary condition of each subpanel thus case, the bridge-building industry proceeds to build fairly
developed is closer to clamped boundaries on all four large concrete filled rectangular (or occasionally trapezoidal)
edges rather than simple boundaries, thereby rendering tubes for bridge piers and cable-stayed bridge pylons
the plate buckling coefficient equal to 10 instead 4 for a without the benefit of the established specification
four-edge simply supported panel (Lee and Yoo, 2012). provisions.
One of the limiting factors controlling the ultimate The impetus of this study is to analytically investigate
the effects of a number of parameters on the elastic and
Note.-Discussion open until November 1, 2013. This manuscript for inelastic strength of concrete filled stiffened compression
this paper was submitted for review and possible publication on panels with regard to the local buckling strength of the
October 19, 2012; approved on March 7, 2012. relatively thin skin. The overall design aspect of composite
© KSSC and Springer 2013 bridge pylons is not addressed here at all. The concrete
*Corresponding author may or may not be designed to partially carry the
Tel: +; Fax: + compressive load of a concrete-filled pylon. In order to
E-mail: ensure the effectiveness of the concrete core, it is
2 Kyoung Chan Lee et al. / International Journal of Steel Structures, 13(2), 000-000, 2013

assumed that the concrete is well confined throughout the 2


load history.
πE t 2
parts of the plate; σcr,p = -------------------⎛----⎞ the elastic critical
2 ⎝ ⎠
Unfortunately, in recent years some confusion has 3( 1 – µ ) w
arisen due to the different rules adopted in the AASHTO plate buckling stress. A complicated iterative process is
specifications guiding the design of transverse flange required to apply design Eq. (2) due to parameters like
stiffeners (AASHTO 2002; AASHTO 2012). Articles NEd, Asl1, Is, and emax that are interdependent or directly
10.39.4.4.1 and 10.39.4.4.2 of the AASHTO (2002) and coupled with the It.
Article 6.11.11.2 of AASHTO (2012) state that the Article 6.11.11.2 of AASHTO (2012) stipulates that the
transverse stiffeners must be proportioned so that the required minimum rigidity for the longitudinal stiffeners
moment of inertia of each stiffener about an axis through should be
the centroid of the section and parallel to its bottom edge
3
is at least equal to Is = ψwtf (4)

3 3f A in which ψ =0.125 k3 for n=1 and ψ =0.120 k3 when n


It = 0.10(n + 1) w ---s ----f (1)
Ea equals 2 where k =buckling coefficient, which shall not
exceed 4. The original equations were proposed by
where w =width of the subpanel between longitudinal Mattock et al. (1967). The impracticality of the suggested
stiffeners or distance from a web to the nearest longitudinal equations was well documented in references (Yoo et al.,
stiffener; Af =area of the compression flange, including 2001; Yoo, 2001; Choi and Yoo, 2005). Recognizing this
longitudinal stiffeners; a =spacing of transverse stiffeners; awkwardness, the latest AASHTO specifications (2012)
E =modulus of elasticity; fs =maximum longitudinal suggest that the number of longitudinal stiffeners should
bending stress in the flange of the panels on either side of preferably be limited to two as a stop-gap measure. A
the transverse stiffeners; and n =number of longitudinal thorough examination is clearly required in order to
stiffeners. As Af and fs are mutually interdependent, there clarify this issue of the minimum stiffness requirement
can be a large number of combinations of these two for flange stiffeners. Recently, Yoo et al. (2001)
parameters that satisfy Eq. (1). Hence, the proper choice numerically identified the minimum stiffness requirement
of the parameters is basically an iterative process in for the longitudinal flange stiffeners and proposed a new
which the designer’s intuition plays a big role. equation that could replace the current design standard,
Similarly, Article 9.2.1(5), Eurocode: EN 1993-1-5 Eq. (4):
(2006) specifies the minimum required stiffness of the
2 3
transverse stiffeners as below: Is = 0.3α ntf w (5)

σ 4 where α =a/w=aspect ratio of a subpanel of a stiffened


It = -----m-⎛--b-⎞ ⎛1 + w0300
---------u⎞ (2)
E π ⎝ ⎝ ⎠ b ⎠ flange, recommended to be taken the next full number.
The term “subpanel” refers to a portion of a steel panel
with bounded by adjacent longitudinal stiffeners spaced at w
and their adjacent transverse stiffeners spaced at a as
σcr, c NEd⎛ 1 1 ⎞ shown in Fig. 1.
- -------- ----- + -----
σm = --------- (3)
σcr, p b ⎝a1 a2⎠ When a stiffened flange (or compression panel) is
designed, it is clearly desirable that the flange strength
2 should attain the local buckling strength of the subpanel
π Eemax
in which u = ----- ------------
-γ ≥ 1.0 ; emax is distance from the (Yoo et al., 2001). This logic is also extended to the
fy 300b MI design of the concrete filled tube (CFT). Lee and Yoo
extreme fiber of the stiffeners to the centroid of the (2012) derived the minimum required rigidity of the
stiffener; γMI =partial factor; b is length of the transverse longitudinal stiffener to accomplish the above objective
stiffener; a1, a2 =spacing of transverse stiffeners; w0 =the as given by
smallest of a1, a2, or b/300; Ned is the largest design
2 3
compressive force of the adjacent panels but not less than Ic = 1.5α ntf w (6)
the largest compressive stress times one half the effective
compressive area of the panel including stiffeners; scr,c = It is noted that the panel can only bulge (buckle) out of
2 the concrete, and the effective boundary condition of the
π EI
-------------s2 =the elastic critical column buckling stress; Asl1 = square panel is fixed along all four edges. Hence, the
Asl1a buckling mode shape of the compression panel resembles
the cross-sectional area of the stiffener and the adjacent a waffle slab as shown later in a series of figures.
Stiffness Requirements for Transverse Stiffeners of Rectangular CFT Compression Panels 3

Figure 1. Concrete partial-filled tube and stiffened compression panel (n =3, ni =1, α =5).

The elastic buckling strength of a square subpanel


3
subjected to a uniformly distributed uniaxial compressive wt
Ir = ----------------------γ
2 0
(8)
force is determined by 12(1 – µ )
2
kπ E ⎛ t ⎞ 2 where γ0 =a function of the aspect ratio, α, squared and
Fcr = ---------------------- ---- (7)
12(1 – µ )⎝w⎠
2 the ratio of the cross-sectional area of the stiffener to the
area of the panel. Since a longitudinal stiffener is
where µ =Poisson’s ratio. The plate buckling coefficient k essentially a compression member, its flexural rigidity to
in Eq. (7) varies from 4 for a square plate simply resist against buckling should have the length parameter
supported on all four edges to 10.07 for a square plate squared as a prominent variable as pointed out by Bleich
clamped on all four edges. Levy (1942) is credited to be in Eq. (8). It should be noted here that Eq. (4) adopted by
the first to come up with the latter value. AASHTO does not have the length parameter.
Bleich (1952) has shown that the required moment of It is noted that the minimum required moment of inertia
inertia of the longitudinal stiffener attached on compression of the longitudinal stiffener on the concrete-filled compression
panels takes the form panel by Eq. (6) is five times as great as that given by Eq.
4 Kyoung Chan Lee et al. / International Journal of Steel Structures, 13(2), 000-000, 2013

(5). The primary function of a longitudinal stiffener here 2. Column Buckling Approximation
is to maintain nodal lines of buckling mode shapes and to
provide stiffnesses for the panels to behave with the boundary As the mechanics of force transfer between the
conditions imposed. Hence, the required minimum stiffness stiffeners and a compression panel that is stiffened both
of the longitudinal stiffener to simulate fixed boundaries longitudinally and transversely is extremely complex, two
given by Eq. (6) is expected to be greater than that given concepts of modeling have emerged, namely the “column
by Eq. (5) for simply supported panels. The elastic behavior” and “plate behavior” approaches alluded to
critical stress of a square plate fixed along all four edges earlier. The early design philosophy for stiffened bottom
given by Eq. (7) is 2.5 times greater than that of a simply flanges in compression of box-girder bridges that prevailed
supported plate. Structural tees are preferred for in European countries was primarily based on “column
longitudinal stiffeners because a tee provides a high ratio behavior,” in which the compression flange is believed to
of stiffness to cross-sectional area (C6.11.11.2, AASHTO behave essentially as a column. In both the “column
2012). Tees also provide a high resistance against torsion. behavior” and “plate behavior” concepts, however, the
However, flat bars are envisioned to be used here as role of a transverse stiffener is assumed to be similar to
longitudinal stiffeners in the CFT as they are embedded that of the lateral bracing of a laterally braced column
into concrete. Therefore, torsional buckling and/or local system.
buckling of these flat bars are not likely to occur, and the The buckling load of a longitudinal stiffener, Pcr, of the
unit cost of flat bars is likely to be less than that of tees. concrete-filled tube is
Transverse stiffeners used as bracing members for
2
longitudinal stiffeners should have sufficient bending stiffness 10.07π EA1⎛ tf ⎞ 2
Pcr = Fcr ⋅ Al = ---------------------------
- ----
2 ⎝ ⎠
(9)
to effectively restrain the longitudinal stiffeners. This 12(1 – µ ) w
requirement is equally applicable to both the modeling
concepts used to describe the behavior of the longitudinal where Al is the cross-sectional area of the longitudinal
stiffeners, namely the “column behavior” approach adopted stiffener. Based on the equilibrium requirements from the
in Eurocode (2006) and the “plate behavior” approach system of longitudinal stiffeners and the transverse
adopted in AASHTO, as illustrated by Yoo et al. (2001). stiffener, Fig. 1, the required stiffness of the transverse
As the aspect ratio of the subpanel is one of the most stiffener, kt, is expressed as (Timoshenko and Gere 1961;
influential parameters determining the minimum required Salmon and Johnson 1996; Yoo 2011)
stiffness of the longitudinal stiffener, it should also be an
equally important parameter defining the required strength βP
kt = ---------cr- (10)
and spacing of adjacent transverse flange stiffeners. However, a
as mentioned earlier, there are no reliable design guides
available that rationally define the minimum stiffness where β is a coefficient that takes into account the effect
requirement and the spacing of transverse stiffeners, of the span number partitioned by the transverse
particularly for the “plate behavior” approach adopted in stiffeners. The β values corresponding to the number of
AASHTO. It is, therefore, the objective of the present spans are given in Table 1, in which ni is the number of
study to investigate the role of transverse stiffeners in the transverse stiffeners. After substituting Eq. (9) into Eq.
load carrying mechanics of longitudinally and transversely (10), the required stiffness kt can be rearranged as
stiffened concrete-filled compression panels both analytically
2
β 10.07π EA-l⎛ tf ⎞ 2
kt = ⎛---⎞ --------------------------
and numerically. For this study, a series of nonlinear
---- (11)
⎝α⎠ 2 ⎝ ⎠
incremental analyses was performed to establish the 12(1 – µ ) w
minimum required rigidity for the transverse stiffeners.
The data collected from the large number of analyses An isolated transverse stiffener is considered to behave
encompassing a wide spectrum of parameters were as a both-end simply supported beam of length b. If a
reduced by a regression process to extract analytical restraining force of Q is assumed at each interface
equations. The validity of the proposed equations was between the longitudinal stiffener and the transverse
then tested using the results of the parametric study for stiffener, the maximum deflection of the transverse
the minimum stiffness requirement of transverse stiffeners. stiffener at the interface is approximated to be (Choi et

Table 1. Values of the Factor β


ni 1 2 3 4 5 6 8 10
Number of Spans 2 3 4 5 6 7 9 11
β 2.000 3.000 3.412 3.623 3.731 3.802 3.876 3.922
Stiffness Requirements for Transverse Stiffeners of Rectangular CFT Compression Panels 5

al., 2007) design parameters on the required moment of inertia, It.


Note that Eq. (17) includes a factor that describes the
3
Qb uncertainty due to any simplifying approximation, including
∆ = --------- (12)
γEIt the validity of the simply supported boundary condition
assumed for the transverse stiffener. This uncertainty is
where It is the moment of inertia of the transverse reflected by the coefficient, η. In order to calibrate and
stiffener. The value of γ can be shown to be 48, 32.4, 20, verify Eq. (17), a series of parametric studies was
16, 13, and 11.4 for 1, 2, 3, 4, 5, and 6 equally spaced performed through numerical analyses. The correlation
longitudinal stiffeners, respectively, within b. It is noted coefficient from regression analysis for Eq. (17) was
that in the derivation of the β value (Timoshenko and found to be greater than 0.9 using the uncertainty factor,
Gere 1961; Salmon and Johnson 1996; Yoo and Lee 2011), η, to be equal to one.
only rigid body equilibrium is considered conservatively
ignoring the small strain energy stored into the bent 3. Numerical Analysis
compression members. Therefore, the boundary conditions
of the compression member do not play a role in the 3.1. Finite element modeling
equilibrium equation. Timoshenko and Gere (1961) and Typical walls of concrete filled tube panels in compression,
Yoo and Lee (2011) show the reasonable nature of such stiffened with longitudinal and transverse stiffeners, were
conservative modeling indicating only a small fraction of modeled using a commercial finite-element code, ABAQUS
additional load to be carried by the bracing members. (2008). The model was assumed to be subjected to a
Taking the required stiffness parameter, kt, for the uniform compressive stress. A concentrated load was
idealized system without considering initial imperfections, applied at each node on the loaded edges of the panel and
as there is already sufficient conservatism built into the stiffeners to simulate a uniform stress distribution. Although
model, the required stiffness for the transverse stiffener there are a variety of finite elements available in
can be expressed as ABAQUS, only the 4-node plate/shell element, S4, is
used to represent the panels and longitudinal stiffeners
Q γEI-t because of its simple yet numerically stable performance.
kt = ---- = -------- (13)
∆ b3 A previous series of analytical investigations on steel
plate buckling (Yoo et al., 2001; Lee and Yoo, 1999)
The required stiffness for the transverse stiffener to showed that a minimum of four square 4-node shell
restrain the out-of-plane deflection is determined from elements were sufficient to model a side of a box tube. In
Eqs. (11) and (13), as below: order to further improve the accuracy of the numerical
solution, each subpanel of the stiffened flange specimens
γEIt ⎛ β ⎞ 10.07π2EAl⎛ tf ⎞ 2 in the present study was divided into 20 square elements.
kt = --------
- ≥ --- --------------------------- ---- (14)
b ⎝α⎠ 12(1 – µ2) ⎝w⎠
3 For simplicity, transverse stiffeners were modeled by a
beam element, B21. A convergence analysis indicated
Therefore, the required minimum moment of inertia of that this mesh refinement yields the critical stress within
a transverse stiffener, It, is 0.5%.
2 3
π β-⎞ 10.07b Al ⎛ t f ⎞ 2
It ≥ ⎛-------
3.2. Elastic buckling
----------------------
- ---- (15)
⎝ γa ⎠ 2 ⎝ ⎠
12(1 – µ ) w The constitutive relationship of the steel is assumed to
be linearly elastic and perfectly plastic. The concrete core
Substituting a =αw and b =(n+1) w into Eq. (15) yields is modeled with a series of springs placed at each node
point capable of resisting only compression. The typical
2
⎧ 10.07A [ ( n + 1 )w ]
3
t 2 ⎫ 10.07π2β(n + 1)3(A t2 )
π β - ------------------------------------------ - ⎛----f ⎞ ⎬ = ---------------------------------------------------
l f stress-strain curve of the spring is shown in Fig. 2. This
-
It ≥ ------------- ⎨
l
γ(αw) ⎩ 2 ⎝ w ⎭ ⎠ 12(αγ)(1 – µ )
2 curve is adopted from BS EN 1992-1-1 (2004). The
12(1 – µ )
numerical values incorporated into the curve are as
(16)
follows: characteristic compressive cylinder strength at
In general, the cross-sectional area of longitudinal 28 days (fck) is 20 MPa (2.9 ksi); mean value of concrete
stiffeners ranges from 0.1 to 0.15 of the subpanel area, cylinder compressive strength, peak stress ( fcm =fck +MPa)
wtf . Conservatively substituting 0.2 wtf for Al, Eq. (16) is 28 MPa (4 ksi); secant modulus of elasticity of concrete
can be simplified to (Ecm) is 28,848 MPa (4,183 ksi); maximum linear stress
(0.4 fcm) is 11.2 MPa (1.6 ksi); corresponding strain (Ecm/
2 3 3
ηπ β(n + 1) wtf (0.4 fcm)) is 0.000388; compressive strain in the concrete
It ≥ ------------------------------------
2
- (17)
6αγ(1 – µ ) at the peak stress (εc1) is 0.002; and ultimate concrete
strain (εcu) is 0.003. The stress-strain relationship on the
Eq. (17) shows the effect of each of the influential compression side is
6 Kyoung Chan Lee et al. / International Journal of Steel Structures, 13(2), 000-000, 2013

Figure 2. Stress-strain curve of nonlinear spring representing


filled-in concrete (1 ksi=6.895 MPa).
Figure 3. Ultimate stress of stiffened CFT vs. slenderness
ratio (Fy =345 MPa, 1 ksi=6.895 MPa.
2
σ k'η – η
-----c- = ------------------------- (18)
fcm 1 + (k' – 2)η
3.4. Critical stress in the inelastic zone
It has been shown (Yoo et al., 2001; Yoo, 2001; Choi
where σc =compressive stress in the concrete; εc =
and Yoo, 2005; Choi et al., 2007) that a CRC (1960) type
compressive strain in the concrete; η =εc /εc1; and k' =
parabolic formula represents the critical stress of the
1.05Ecmεc1/fcm.
compression subpanels better than that given in AASHTO
Since an eigenvalue routine cannot be applied to the
(2002; 2012) in the inelastic range, particularly with
model in the elastic buckling range because of the
initial imperfections. The parabolic equation is
nonlinear nature of the constitutive relationship of the
concrete shown in Fig. 2, the elastic analysis had to be
2Frc 2
carried out on the incremental nonlinear analysis technique. -(F – Frc)⎛w
Fcr = Fy – ---------- ----⎞ (19)
2 y ⎝ tf ⎠
As it will be shown later, there were significant Fy C c
postbuckling strengths involved, particularly in the panels
If the intensity of the residual stress, Frc, is arbitrarily
with high slendernesses.
taken to be 0.5Fy (similar to the current AASHTO
provision for compression members), Eq. (19) reduces to
3.3. Incremental nonlinear analysis
In order to validate the reliability of Eq. (17) in the
1 w 2
inelastic buckling range, a series of incremental nonlinear Fcr = Fy 1 – --------2-⎛----⎞ (20)
⎝ ⎠
analyses was performed on the longitudinally and transversely 2Cc tf
stiffened panel models. The initial geometric imperfections
where Cc is the slenderness ratio taken from Eq. (7) with
applied to the incremental analyses were determined by
k =10 corresponding to Fcr =Fy/2, which is
scaling the elastic buckling mode shapes using the
maximum tolerance criteria for the fabrication specified 2
in the current authorized specifications. The maximum
5π E -
Cc = ------------------------ (21)
2
tolerance for the panel flatness deviation, ∆, due to 3(1 – µ )Fy
welding was therefore taken to be w/100 from Article 3.5
of the Bridge Welding Code (2010). 3.5. Nonlinear analysis results
In a series of hypothetical models, some of the Figure 3 shows the critical stress of the concrete filled
parameters are fixed such that E =200 GPa (29,000 ksi), compression panels with all four sides clamped. As it can
Fy =345 MPa (50 ksi), w =2540 mm (100 in.), tf =31.8 be seen from the figure, the effect of initial imperfections
mm (1.25 in.). Hence, the slenderness ratio, w/tf , is set to is most pronounced in the inelastic buckling zone with
be 80 and the initial imperfection is also set to be 25.4 slenderness ratios between 40 and 80. As it can be seen
mm (1 in.) (∆ =w/100). The thickness of the stiffeners is from Fig. 3, the maximum reduction of the critical stress
also set equal to 31.8 mm (1.25 in.). The lengths of the observed is up to 18 percent due to an initial imperfection,
longitudinal stiffener, a, are 5,080 mm (200 in.), 7,620 ∆ =w/100, from that of the relatively straight panel with
mm (300 in.), and 12,700 mm (500 in.). The aspect ratios ∆ =w/10,000. It has been reported (Yoo et al., 2001; Yoo,
taken are 2, 3, and 5 and the number of longitudinal 2001; Choi and Yoo, 2005) that the reduction of the
stiffeners, n, and the number of transverse stiffeners, ni, critical stress is up to 14 percent due to a similar initial
run is 1, 2, and 3, respectively. imperfection differential of stiffened panels with all four
Stiffness Requirements for Transverse Stiffeners of Rectangular CFT Compression Panels 7

Table 2. Ultimate stresses of CFT compression panels


No. of stiffeners Moment of inertia of stiffener
α Fcr (MPa) by FEM Ratio (%) to Eq. (20)
Long. n Trans. ni Ic (×10 mm4) Eq. (6)
8
It (×108 mm4) Eq. (16)
1 1 4.878 0.175 2 222.3 93
1 1 10.975 0.154 3 224.5 94
1 1 30.486 0.129 5 225.1 94
1 2 4.878 0.262 2 221.2 92
1 2 10.975 0.229 3 217.3 91
1 2 30.486 0.194 5 224.5 94
1 3 4.878 0.300 2 220.1 92
1 3 10.975 0.262 3 220.1 92
1 3 30.486 0.221 5 225.6 94
2 1 6.898 0.989 2 223.9 93
2 1 15.521 0.864 3 226.7 95
2 1 43.113 0.729 5 230.0 96
2 2 6.898 1.484 2 223.9 93
2 2 15.521 1.296 3 221.7 93
2 2 43.113 1.093 5 228.4 95
2 3 6.898 1.682 2 223.4 93
2 3 15.521 1.469 3 225.6 94
2 3 43.113 1.239 5 230.0 96
3 1 8.448 4.037 2 226.7 95
3 1 52.803 2.997 5 233.3 97
3 2 8.448 6.035 2 226.2 94
3 2 52.803 4.579 5 231.7 97

sides simply supported. judgmental. Substituting η =1.5 and µ =0.3 in Eq. (17)
Table 2 summarizes the selected analysis results with and taking only the full number conservatively in the
panels having a slenderness ratio equal to 80. According denominator yields
to Eq. (20), the critical stress is computed to be 239.2
2 3 3
MPa (34.7 ksi). Figure 4 shows the convergence trends of π β(n + 1) wt
It ≥ ---------------------------------f (22)
selected models. As expected the strength of the models 3αγ
that have lower stiffnesses do not attain the ultimate stress
predicted by Eq. (20). At the same time, they do not show Eq. (22) is identical to that derived by Choi et al.
any further increase in strength beyond that expected (2007). Since the moment of inertia of the transverse
when the stiffnesses are increased. This exhibits the nature stiffener is proportional to the cube of the width of the
of the transverse stiffener being a secondary bracing rectangular bar here, only 14.5% increase of the width of
member. the bar will provide a 50% increase of the stiffener. It is
Figure 5 shows pictorially the waffle slab type buckled recommended that flat rectangular bars be used for both
shape of the concrete filled compression panel. When the the longitudinal and transverse stiffeners as these members
stiffness of the transverse stiffener is not sufficiently strong, are embedded into the concrete, there is no likelihood of
it fails to control the effective length of the longitudinal local buckling. A schematic diagram for this concept is
stiffener. When this occurs, a part or the entire panel given in Fig. 1.
buckles out from the concrete face at a substantially Figure 6 shows the stress contour of a system of two
lower load. Examination of Figs. 4 and 5 reveals that the longitudinal (n =2) and two transverse (ni =2) stiffeners.
η factor should be increased at least to 1.5 in order to The aspect ratio of the subpanel is three (α =3). Hence, a
ensure the buckling mode as assumed although the total of nine buckling mode shapes was developed in the
ultimate stresses stabilize earlier as given in Table 2. It is longitudinal direction with clamped boundary conditions
noted that there is a considerable scatter of data in Fig. 4 on all four edges. The longitudinal compressive stress
depending on the α values and the buckling modes shown intensity is used to generate the contour diagram. The
on the left-hand side column are quite slowly converging material properties and dimensions of the model shown in
to the mode shapes listed on the right-hand side column Fig. 6 are: E =200 GPa (29,000 ksi); Fy =345 MPa (50
of Fig. 5. The uncertainty factor of 1.5 is strictly ksi); w =2,540 mm (100 in.); t =31.8 mm (1.25 in.); the
8 Kyoung Chan Lee et al. / International Journal of Steel Structures, 13(2), 000-000, 2013

Figure 4. Convergence trend.

initial imperfection, ∆ =w/100=25.4 mm (1 in.); and the postbuckling strength for the thin panel subjected to
width of the longitudinal stiffener is 527.3 mm (20.76 in.) elastic buckling as shown in Fig. 3, however, this reduction
to yield the minimum value required by Eq. (6). The in the postbuckling strength is non-consequential. In
moment of inertia of the transverse stiffener is 2.593×10−4 general, the postbuckling strength is not recognized in the
m4 (622.8 in.4) for the beam element used in the model. bridge design due to the potential adverse effect of
Hence, the slenderness ratio, w/t, becomes 80 (less than deflection induced fatigue.
Cc =102.4, Eq. (21) warranting inelastic buckling. The
two transverse stiffeners were removed for clarity. Faint 4. Summary and Concluding Remarks
traces of the transverse stiffeners are visible to the one-
third points of the system. This study has examined the minimum required stiffness
of the transverse flange stiffeners on a rigorous theoretical
3.6. Effects of residual stress basis augmented by a series of numerical analyses. For
The effects of residual stress resulting from the welding the design of longitudinally and transversely stiffened
practice were examined thoroughly in two previous compression flanges, AASHTO specifications currently
studies (Lee and Yoo, 2012; Choi and Yoo, 2005). Even suggest either the use of Eq. (1) or simply utilizing the
when the maximum intensity of the tensile residual stress same size as the longitudinal stiffeners, which appear to
was assumed equal to the yield stress of the base metal, lack any theoretical foundation. The present study has
the effect of the residual stress was less than 5% for all shown that the AASHTO design provisions for transverse
inelastic buckling ranges. It became only noticeable (up stiffeners does not adequately estimate the minimum
to 15%) for the extremely thin panel with a width-to- required stiffness. In this study, which approximated the
thickness ratio well over 100. As there exists a substantial behavior of the longitudinal stiffener to buckling of a
Stiffness Requirements for Transverse Stiffeners of Rectangular CFT Compression Panels 9

Figure 5. Buckled shape at critical load.

Figure 6. Stress contours (1 ksi=6.895 MPa).

multiply braced column by transverse stiffeners, the design or regression equation for the minimum required rigidity.
parameters that govern the behavior of the transverse As a result, Eq. (22) is hereby proposed as offering a
stiffeners were identified theoretically. Extensive numerical simple and rational way to determine the minimum
studies were then conducted in order to extract a numerical stiffness required for transverse flange stiffeners.
10 Kyoung Chan Lee et al. / International Journal of Steel Structures, 13(2), 000-000, 2013

The incremental nonlinear analysis results verified that Yoo, C. H., Choi, B. H., and Ford, E. M. (2001). “Stiffness
the proposed equations can be extended to the inelastic requirements for longitudinally stiffened box-girder
buckling range. As demonstrated in a series of previous flanges.” Journal of Structural Engineering, ASCE,
studies, the critical stress in the inelastic range is better 127(6), pp. 705-711.
computed using a CRC (1960) type parabolic formula, Yoo, C. H. (2001). “Design of longitudinal stiffeners on box
Eq. (20). The current AASHTO provisions of the ultimate girder flanges.” International Journal of Steel Structures,
1(1), pp. 15-23.
stresses of the longitudinally and transversely stiffened
Choi, B. H. and Yoo, C. H. (2005). “Strength of stiffened
compression panel do not reflect the detrimental effect of
flanges in horizontally curved box girders.” Journal of
the initial imperfection that is permitted by the Bridge Engineering Mechanics, ASCE, 131(2), pp. 167-176.
Welding Code (2010). The current study showed that the Levy, S. (1942). Bending of Rectangular Plates with Large
ultimate stress of the concrete filled compression panel Deflections. NACA Technical Report No. 737,
could be decreased as much as 18 percent of that of the Washington, DC.
straight panel, particularly for the width-to-thickness ratio Bleich, F. (1952). Buckling strength of metal structures.
of 40-80. With the successful demonstration of the McGraw-Hill, NY.
applicability of Eq. (22) for the elastic buckling, inelastic Timoshenko, S. P. and Gere, J. M. (1961). Theory of elastic
transition, and yield zones, an optimal transverse stiffener stability. 2nd ed., McGraw-Hill, NY.
design process can be applied, and transversely stiffened Salmon, C. G. and Johnson, E. J. (1996). Steel structures:
flanges can be designed with confidence. Design and behavior, 4th ed., HarperCollins, NY.
Yoo, C. H. and Lee, S. C. (2011). Stability of Structures:
Principles and Applications. Elsevier, Burlington, MA.
References Choi, B. H., Kang, Y. J., and Yoo, C. H. (2007). “Stiffness
requirements for transverse stiffeners of compression
Lee, K. C. and Yoo, C. H. (2012). “Longitudinal stiffeners in
panels.” Engineering Structures, 29, pp. 2087-2096.
concrete-filled tubes.” Journal of Structural Engineering,
ABAQUS (2008). ABAQUS User’s manual, ABAQUS Inc.,
ASCE, 138(6), pp. 753-758.
Pawtucket, RI.
AASHTO (2002). Standard specifications for highway
Lee, S. C. and Yoo, C. H. (1999). “Strength of curved I-
bridges. 17th Ed., American
girder web panels under pure shear.”
Association of State Highway and Transportation Officials,
Journal of Structural Engineering, ASCE, 125(8), pp. 847-
Washington, DC.
853.
AASHTO (2012). AASHTO-LRFD bridge design specifications.
BS EN 1992-1-1 (2004). Eurocode 2: Design of concrete
6th Ed., American
structures, British Standard Institute, London, UK.
Association of State Highway and Transportation Officials,
Bridge Welding Code (2010). ANSI/AASHTO/AWS D1.5-
Washington, DC.
2010. A joint publication of the American Association of
BS EN 1993-1-5 (2006). Eurocode 3: Design of steel
State Highway and Transportation Officials and the
structures. British Standard Institute, London, UK.
American Welding Society, Washington, DC.
Mattock, A. H. et al. (1967). Criteria for design of steel
CRC (1960). Guide to design criteria for metal compression
concrete composite box girder highway
members. 2nd Ed., Column Research Council, Wiley, NY.
Bridges. Appendix B, Report of the Committee
(unpublished).

View publication stats

You might also like