You are on page 1of 8

Bending Stiffness Requirement for Closed-Section

Longitudinal Stiffeners of Isotropic Material Plates


under Uniaxial Compression
Byung H. Choi1; Jung J. Kim2; and Tae-Hyung Lee3

Abstract: Longitudinally stiffened plates with closed-section ribs are supposed to be an effective system for axially compressed members.
Downloaded from ascelibrary.org by New York University on 05/17/15. Copyright ASCE. For personal use only; all rights reserved.

However, current design specifications on closed-section ribs, particularly for the minimum required stiffness, are not sufficient, and thus, ex-
cessive designs are quite common because of the absence of suitable design guides. In this paper, simple closed-form formulas for the minimum
required stiffness of compressively loaded isotropic plates braced by U-shaped longitudinal stiffeners (U-ribs) are derived. The effects of the
sectional stiffness of U-ribs on the buckling modes and strengths of stiffened plates are examined by numerical analyses. Three-dimensional
finite-element models of U-rib stiffened plates were obtained, and a series of eigenvalue analyses was conducted. By parametric study, thresh-
olds of the sectional stiffness of U-ribs that may be adopted as the minimum requirement were collected, and these were used for regression
analysis to obtain a simple and practical design equation for determining a suitable sectional stiffness after conducting a comparative study
using a theoretically derived formula. The findings of this study would contribute to improving the optimum design of U-rib-stiffened plates.
DOI: 10.1061/(ASCE)BE.1943-5592.0000689. © 2014 American Society of Civil Engineers.
Author keywords: Closed-section rib; Elastic buckling; Longitudinal stiffener; Stiffened plate; Stiffness requirement.

Introduction AASHTO (2002, 2012) provisions limit the number of longitudinal


stiffeners to two for the application of Eq. (1). Yoo et al. (2001)
Longitudinally stiffened plates have been widely used for thin-walled proposed Eq. (2) for the minimum required moment of inertia of
plate structures because of their structural economy and superior sta- a longitudinal stiffener about an axis parallel to the flange and at the
bility. Longitudinally stiffened compression plate structural members base of the stiffener for straight flanges, which may reasonably re-
generally offer an economical structure by efficiently proportioning place the current design standard based on Eq. (1) (Yoo et al. 2001;
the material to resist induced compressive stresses. The minimum Choi 2002; Choi and Yoo 2005)
required moment of inertia of a longitudinal stiffener for compressive pffiffiffi
flanges is given by AASHTO [2002, Eq. (10-138)] and AASHTO Is ¼ 0:3a2 ntp3 w (2)
[2012, Eq. (6.11.11.2-2)]. It is expressed as
where a 5 a=w 5 aspect ratio of a subpanel of a stiffened flange; and
Is ¼ wtp3 w (1) a 5 spacing of the transverse stiffeners. The term subpanel refers to
a portion of a steel panel bounded by adjacent longitudinal stiffeners
where w 5 0:125k 3 for n 5 1 and w 5 0:07k 3 n4 for n 5 2, 3, 4, or 5; or webs and their adjacent transverse stiffeners.
k 5 buckling coefficient, which should not exceed 4; n 5 number of Longitudinal stiffeners for stiffened plates can be simply clas-
longitudinal stiffeners; w 5 width of the flange between the lon- sified as open-section type or closed-section type based on the cross-
gitudinal stiffeners or the distance from a web to the nearest lon- sectional shape of the stiffener and the plate. Previous study has been
gitudinal stiffener [that is, wL in Fig. 1(a)]; and tp 5 thickness of the conducted on open-section stiffeners to establish design principles,
bottom flange. particularly the use of ordinary structural T-shapes commonly used
This equation was developed by A. H. Mattock et al. (unpublished in practical designs. Current design specifications do not provide a
report, 1967). However, it has been demonstrated that the equation specific rule regarding the minimum required stiffness of closed-
is too conservative for compression flanges that have three or more section stiffeners. However, considerable differences are expected
longitudinal stiffeners (Yoo et al. 2001; Choi 2002; Choi and Yoo in the buckling modes and strengths of stiffened plates with regard
2005). Because of this observed inadequacy of Eq. (1), current to the section types of the longitudinal stiffeners. There are several
reasons for this, which are as follows.
Stiffened panels buckle in the symmetric mode, as shown in
1
Associate Professor, Dept. of Civil and Environmental Engineering, Fig. 1(a), and it has been found that the stiffeners bend along the
Hanbat National Univ., Daejeon 136-702, South Korea. buckled surface of the panel. As previously reported (Yoo et al.
2
Assistant Professor, School of Architecture, Kyungnam Univ., 2001; Choi 2002; Choi and Yoo 2005), symmetric mode buckling
Changwon-si, Gyeongsangnam-do 631-701, South Korea. is likely to occur when the rigidity of the longitudinal stiffener is
3
Associate Professor, Dept. of Civil Engineering, Konkuk Univ., Seoul
relatively small. However, when the rigidity of the longitudinal
143-701, South Korea (corresponding author). E-mail: thlee@konkuk.ac.kr
Note. This manuscript was submitted on February 12, 2014; approved on stiffener is greater than a certain value, the panels buckle in the
August 7, 2014; published online on September 2, 2014. Discussion period antisymmetric mode as shown in Fig. 1(b), which is characterized
open until February 2, 2015; separate discussions must be submitted for by the longitudinal stiffeners remaining straight. Buckling behavior
individual papers. This paper is part of the Journal of Bridge Engineering, similar to that of plates stiffened by open-section stiffeners is also
© ASCE, ISSN 1084-0702/04014092(8)/$25.00. anticipated for those reinforced by closed-section stiffeners, which

© ASCE 04014092-1 J. Bridge Eng.

J. Bridge Eng.
Downloaded from ascelibrary.org by New York University on 05/17/15. Copyright ASCE. For personal use only; all rights reserved.

Fig. 1. Buckling of stiffened plates with longitudinal stiffeners: (a) symmetric mode, with open-section-type stiffeners; (b) antisymmetric mode, with
open-section-type stiffeners; (c) symmetric mode, with closed-section-type stiffeners; (d) antisymmetric mode, with closed-section-type stiffeners;
(e) plane view of stiffened plate model

are shown in Figs. 1(c and d). In Fig. 1, the differences between the than that of the open-section stiffeners as shown in Fig. 1(b). The
buckling modes reinforced by open-section stiffeners and closed- uneven effective width in the local buckling mode shape may offer
section stiffeners are schematically compared. The columnlike buck- a restraining effect, which increases the strength. It is essential to
ling modes of both stiffener types appear to be similar. In contrast, the consider the effect of elastic restraints at the plate-stiffener web joint
difference between the critical modes of the two stiffeners is apparent connections of thin-walled sections in conducting a reliability analysis.
for local plate buckling. The arrangement of the nodal points on the For these reasons, it is important to examine the applicability of
stiffened plate in the case of the closed-section stiffener varies as shown existing design equations to closed-section stiffeners, because some
in Fig. 1(d), and thus there may be uncertainty and disagreement in unusual features may be expected in the buckling mode of plates
determining the number of longitudinal stiffeners that should be used. stiffened by closed-section ribs. As mentioned earlier, there is no
This situation is affected by the ratio of the U-rib width wR to the net reliable design guide to facilitate a reasonable specification of the
spacing wS . That is, as the ratio approaches 1, the resisting effect of the minimum stiffness requirement for closed-section stiffeners. The ob-
stiffeners is exerted not on their centroid but on each connected point, jective of the current study was therefore to investigate the charac-
and twice the number of closed-section stiffeners should therefore be teristics of closed-section stiffeners in the load-carrying mechanics of
adopted for n when applying Eq. (2) to designing the stiffness. The longitudinally stiffened compression panels both analytically and
spacing between the centroid of the longitudinal stiffeners—that is, numerically.
the effective stiffening region resisted by the individual longitudinal For this study, a series of elastic bend-buckling analyses were
stiffener against the columnlike buckling—does not coincide with the performed to establish the minimum required rigidity for closed-
effective width of the local buckling, which is reduced by half the width section stiffeners. First, an approximate closed-form equation was
of the closed-section stiffener, wR . This is different from the assumption theoretically derived. Then, the effects of influential parameters
on which the derivation of Eq. (2) is based; that is, wG and wS should obtained from the theoretical investigation were verified by the
be consistent with each other as shown in Figs. 1(c and d). In addition, results of a parametric study of the minimum stiffness requirement
the local buckling mode shape clearly changes between Figs. 1(b and d). for closed-section stiffeners. The data collected from a large number
This comparison clearly demonstrates that the antisymmetric mode of of analyses encompassing a wide range of parameters were reduced
Fig. 1(d) has a greater number of unevenly spaced half-sine curves by regression to obtain analytical equations. Consequently, this paper

© ASCE 04014092-2 J. Bridge Eng.

J. Bridge Eng.
particularly provides a simple guide for determining the minimum where IP 5 bending moment of inertia per unit length of the plate
stiffness of closed-section ribs for preliminary design. about an axis parallel to the neutral axis of the plate thickness; and
IR 5 bending moment of inertia per unit length of each subplate
that consists of the rib. The other symbols used are defined in
Theoretical Formulation of Required
Fig. 3 in more detail.
Stiffness Equation
By rearranging Eq. (5), Eq. (6) is obtained as follows:

2EIR
Rotational Restraint Stiffness kR ð2Þ uA wT
uD ¼  h  ¼ ð2Þ uA (6)
The local buckling of an orthotropic plate subjected to uniform in- 4EIR 6EIR 2wT þ 3h
plane axial load along simply supported edges and rotationally re- þ
h wT
strained at one of two unloaded edges is briefly described in this
section. Furthermore, formulation of a closed-form expression for By substituting Eq. (6) into Eq. (4), the moment-rotation relation-
rotational restraint stiffness is briefly presented. The plate stiffened ship at the plate edges elastically restrained by the U-shaped ribs
Downloaded from ascelibrary.org by New York University on 05/17/15. Copyright ASCE. For personal use only; all rights reserved.

by closed-section stiffeners is regarded as an elastically restrained can be defined as


plate at the edges of y 5 0 and wS , as shown in Fig. 2. At the connected    
edges, the elastic rotational restraint kR is defined as the required 2EIR wT 6EIP
M0 ¼ uA 22 þ (7)
moment per rotational angle h 2wT þ 3h wR
M0
kR ¼ (3) Because M0 5 kR × f 5 kR × uA , the explicit formula for the rotational
f
restraint stiffness is solved as
For determining the restraint stiffness kR , the U-shaped rib is treated    
M0 2EIR wT 6EIP
as a statically indeterminate frame consisting of four subplates with kR ¼ ¼ 22 þ (8)
unit length on the x-axis and subject to a transverse bending M0 , as uA h 2wT þ 3h wR
shown in Fig. 3. Then, the explicit formula for the moment-rotation
relationship can be simply derived based on the idealized mechanical Elastic Buckling Strength Equations of Stiffened Plates
behavior of Fig. 3. By using the equilibrium condition at Points A
and D of the fictitious frame, Eqs. (4) and (5) are simply derived as The local buckling strength of an orthotropic plate subjected to
uniform in-plane axial load along simply supported edges and
P 2EIR 2EIP rotationally restrained at one of two unloaded edges is given by
MA ¼ 0, 2 M0 þ ð2uA þ uD Þ þ ð2uA þ uB Þ ¼ 0
h wR Qiao and Shan (2005). The approximate formula of critical local
(4) buckling load is obtained by adopting the unique displacement func-
tion chosen by combining harmonic and polynomial buckling de-
P 2EIR 2EIP formation functions and then applying a variational formulation of
MD ¼ 0, ðuA þ 2uD Þ þ ð2uD þ uC Þ ¼ 0 (5) the Ritz method.
h wT
By inserting the isotropic material properties into their explicit
formula, the following approximate solution is simply derived for
the local buckling of a plate of width wS with equal rotational re-
straints along unloaded edges:
 pffiffiffiffiffiffiffiffiffi 
24D 1:871 t 1 t 2 þ t 3
RR
Ncr ¼ 2 (9)
wS t1

where the coefficients of t 1 , t2 , and t 3 are functions of the rotational


restraint stiffness. For a plate stiffened by closed-section U-type
stiffeners, kR , defined in the “Rotational Restraint Stiffness” section,
can be inserted into the restraint stiffness. For the case of symmetric
arrangement of rotational restraints, the coefficients are, respectively,
Fig. 2. Rotationally restrained plate model defined as

kR w S k 2 w2 kR wS k 2 w2
t 1 ¼ 124 þ 22 þ R 2 S, t 2 ¼ 24 þ 14 þ R 2 S,
D D D D
kR w S kR2 w2S
t 3 ¼ 102 þ 18 þ
D D2
Eq. (9) is expressed in a different form by applying the strength
increment factor, FðkR Þ, as

4p2 D
RR
Ncr ¼ FðkR Þ × Ncr
SS
¼ × FðkR Þ (10)
w2S
Fig. 3. Rotational restraint stiffness attributable to stiffening by closed-
section stiffener where NcrSS
5 elastic buckling load of simply supported plates along
the four sides without a rotational restraint along the unloaded edges

© ASCE 04014092-3 J. Bridge Eng.

J. Bridge Eng.
and has the unit of force per unit width. Accordingly, the strength Numerical Analysis
increment factor can be defined as
 pffiffiffiffiffiffiffiffiffiffiffiffi 
6 × 1:871 × t2 =t 1 þ t 3 =t 1 Finite-Element Modeling
FðkR Þ ¼ (11) To calibrate and verify Eq. (14), a series of parametric studies were
p2
performed through finite-element modeling. For the numerical para-
Meanwhile, the general equation for calculating the critical stress metric study, the stiffened models as shown in Fig. 4 were designed by
of the global buckling mode is simply presented as (Choi 2002; considering the geometrical dimensions of practical stiffened plates
Timoshenko and Gere 1961) (Chou et al. 2006), which have two U-ribs for longitudinal stiffeners
  and are simply supported along four sides.

2 2 Pn ip Detailed dimensions of the plate models stiffened with U-ribs


1 þ b þ 2 g sin 2
p2 D i51 nþ1 are presented in Table 1, in which wS is the net spacing between
S
Ncr×G ¼ 2    (12) the U-shaped stiffeners, as mentioned earlier, and thus should be the
b Pn ip
b 1 þ 2 i51 d sin
2 2
effective width of the local plate buckling mode. The three types of
Downloaded from ascelibrary.org by New York University on 05/17/15. Copyright ASCE. For personal use only; all rights reserved.

nþ1
U-ribs are selected with the ratio of the U-rib width, wR , to the net
spacing, wS , and are denoted as L-, M-, and S-type, as listed in
where a=b5b; EIS =ðbDÞ5g; PS =ðbNx Þ5AS =ðbtp Þ5d; D5ðE×tP3 Þ Table 1. The L-type is a common type that comes from the ortho-
=½12ð12m2 Þ; As 5 area of closed-section stiffeners; E 5 modulus of tropic steel deck for long-span bridges where wR and wS are similar
elasticity; and tp 5 thickness of stiffened plate. Derivation with in magnitude to each other, as explained in Chou et al. (2006). The
respect to coefficients used in the assumed deflection shape of the U-rib thickness tu was varied in the range given in Table 1 for the
buckled plate expressed by a double sinusoidal series leads to a parametric analyses. The minimum limit of tu was determined
system of homogeneous linear equations. By employing a simple based on the height and width of the U-ribs, denoted as hu and wR ,
shape function to take only the first term from the system of equa- respectively.
tions, a simplified solution of Eq. (12) can be acquired. Accordingly, Fig. 4 shows the finite-element mesh and the boundary condition
Timoshenko and Gere (1961) verified that the approximation given of a prototype model. In the model, half of the U-shaped stiffeners
by Eq. (12) is fairly accurate for longer plates, such as for b$2, and with the symmetric boundary condition were also installed along
that, for shorter plates, a larger number of terms must be considered both unloaded sides to confer the consistent rotational restraint
to obtain a more exact solution than Eq. (12). effect on each subpanel during the local buckling. The plate
The two terms in Eq. (12) condense, respectively, to the following: models stiffened with longitudinal stiffeners were modeled using
    the commercial finite-element code ABAQUS 6.12, and the stiffened
Pn ip Pn ip
2 g sin2 ¼ ðn þ 1Þg, 2 d sin2 ¼ ðn þ 1Þd plates were modeled using the four-node plane element S4R5.
i51 nþ1 i51 nþ1 The effective subpanel width wS was divided into at least 12
elements, as suggested by Choi (2002) and Choi and Yoo (2005),
As a result, Eq. (12) can be simplified as which provides sufficient stability and convergence for the nu-
merical analysis.

2 2
p2 D 1 þ b þ ðn þ 1Þg
S
Ncr×G ¼ 2 2 ½1
(13)
b b þ þ 1Þdðn

Theoretical Equation for Required Stiffness


Eq. (10) gives the critical strength of stiffened plates buckled in an
antisymmetric mode shape (local plate behavior). Eq. (13) represents
the symmetric mode buckling (column/strut behavior). Because
Eq. (13) has a linear relationship with the function of Is , the critical
stress given by Eq. (13) may be higher or lower depending upon the
value of Is . Furthermore, the buckling coefficient depending upon
the value of k does not vary greatly for panels with longitudinal
stiffeners of significantly higher rigidity because of antisymmetric
mode buckling, and therefore, setting a threshold value of rigidity
may ensure an antisymmetric buckling mode (Yoo et al. 2001; Choi
2002). The equation for the minimum required moment of inertia
of closed-section longitudinal stiffeners can be derived by setting
Eq. (13) equal to Eq. (10) as follows:
( ) Fig. 4. Finite-element modeling with boundary conditions
a2S × tp3 × wL 1

2 2
IsðreqÞ ¼ 4½1 þ ðn þ 1Þ × d × FðkR Þ 2 2 1 þ b
12ð1 2 m2 Þ aS
(14) Table 1. Model Dimension
Type wG (mm) wS (mm) wR (mm) hu (mm) wT (mm) wS =wR
where wL 5 b=ðn 1 1Þ; and aS 5 a=wS . From the theoretically drawn
L 471.6 250 221.6 164 156 1.13
closed-form solution Eq. (14), the main characteristics of influential
M 412.4 250 162.4 105 112 1.54
parameters for the minimum required stiffness for the U-shaped
S 350.0 250 100.0 105 100 2.50
stiffeners may be traced.

© ASCE 04014092-4 J. Bridge Eng.

J. Bridge Eng.
Downloaded from ascelibrary.org by New York University on 05/17/15. Copyright ASCE. For personal use only; all rights reserved.

Fig. 5. Typical buckling mode shapes of L-type, tp 5 19 mm, aS 5 3: Fig. 6. Typical buckling mode shapes of M-type, tp 5 16 mm, aS 5 3:
(a) stiffener buckling (SB); (b) global buckling (symmetric) mode (a) stiffener buckling (SB); (b) global buckling (symmetric) mode
(GB); (c) local plate buckling (antisymmetric) mode (PB) (GB); (c) local plate buckling (antisymmetric) mode (PB)

© ASCE 04014092-5 J. Bridge Eng.

J. Bridge Eng.
Downloaded from ascelibrary.org by New York University on 05/17/15. Copyright ASCE. For personal use only; all rights reserved.

Fig. 7. Typical buckling mode shapes of S-type, tp 5 11 mm, aS 5 5:


(a) stiffener buckling (SB); (b) global buckling (symmetric) mode (GB); Fig. 8. Buckling strength variation along with stiffener rigidity:
(c) local plate buckling (antisymmetric) mode (PB) (a) L-type, tp 5 28 mm, aS 5 3; (b) M-type, tp 5 16 mm, aS 5 3;
(c) S-type, tp 5 2 mm, aS 5 3

The axial uniform compression was loaded at both end sides, the derived formula, Eq. (14). In each model case, an iterative
and the stiffened plates were simply supported along the four sides. process was adopted by bifurcation analysis using the ratio of the
Over 120 model cases were designed by considering characteristic thickness of the U-ribs to the threshold value of the minimum
parameters such as tp , aS , wL , and FðkR Þ, which are obtained from rigidity.

© ASCE 04014092-6 J. Bridge Eng.

J. Bridge Eng.
Analysis Results buckling strength of the simply supported plate of width wS was
increased from NcrSS
by a ratio equal to FðkR Þ. Thus, it is notable that
The finite-element analysis reveals that the buckling modes can be
the plates stiffened by U-ribs may have a more enhanced strength
classified into three types, as shown in Figs. 5–7: the U-rib local
along the U-rib that may be expressed using the strength increment
buckling (UB) mode, the plate local buckling (PB) mode of sub-
factor FðkR Þ once they have sufficient bending stiffness to form
panels, and the global column buckling (GB) mode, respectively.
a nodal line along the attached locations on the plates. The analysis
Figs. 5–7 also demonstrate that these buckling modes generally
results show an excellent correlation with Eq. (10), particularly for
appear regardless of the U-rib types (L-, M-, and S-types). The major
the PB mode of the M- and S-type stiffener cases, because FðkR Þ of
characteristic of the PB mode is that the nodal lines are expected
the FEM results converges to Eq. (11) as tu gets thicker to sufficiently
to be formed on the plate along all lines of the attached U-ribs.
exert the rotational restraint stiffness. As a consequence, it is likely
Conversely, in the GB mode, these nodal lines are not formed at the
to provide a more accurate prediction for a stiffener type with
U-rib locations, and the U-ribs are buckled into an out-of-plane
wS =wR $ 1:5. In addition, it is notable that, because the approxi-
direction resembling columnlike behavior when subjected to com-
mations used in the derivation were more accurate for longer plates,
pressive loading. It is notable from the PB mode that a half-sine curve
as mentioned earlier, the numerical analysis results provide a better
Downloaded from ascelibrary.org by New York University on 05/17/15. Copyright ASCE. For personal use only; all rights reserved.

can be clearly observed in the wR region under the U-ribs, similar to


match with the model cases of aS $ 5. As can be seen from Table 2,
that shown in Fig. 1(d).
The buckling strength Ncr corresponding to the antisymmetric
The elastic buckling stresses Fcr were obtained from the para-
buckling mode (PB) is always greater than that corresponding to
metric numerical analyses, some of which are graphically presented
the UB mode and the GB mode.
in Fig. 8. The critical stresses were increased incrementally in
proportion to the second moment of inertia of the U-ribs, Is . Once the
threshold value was reached, the buckling strength of the stiffened Proposal and Validation of the Design Equation
flange increased only slightly with an extremely reduced rate, as can
be seen in Fig. 8 and Table 2. Above the threshold point, the buckling Because Eq. (14) has been derived simply through theoretical ap-
mode changed to the PB mode. Additionally, the buckling strength proximation methods, it must have distinctive limitations. Thus, re-
of the PB mode was similar to the value of Eq. (10). Hence, the elastic gression analyses were performed on the collected data through a series

Table 2. Eigenvalue Analysis Results


Dimensions Ncr (3103 N=mm) F ðkR Þ
SS
Stiffener type tp (mm) aS tu (mm) Ncr Eq. (10) FEM Eq. (11) FEM Difference (%) Mode
L 14 7 7 31.7 42.3 21.8 1.33 0.69 93.5 UB
14 7 9 31.7 42.8 33.1 1.35 1.04 29.5 GB
14 7 14 31.7 45.0 38.6 1.42 1.22 16.5 PB
18 5 18 67.5 95.7 80.8 1.42 1.20 18.3 PB
26 3 16 203.3 273.5 179.3 1.34 0.88 52.5 GB
26 3 26 203.3 288.3 212.6 1.42 1.05 35.6 PB
24 3 48 159.9 259.2 211.6 1.62 1.32 22.5 PB
M 16 3 7 47.4 65.5 47.0 1.38 0.99 39.4 UB
16 3 10 47.4 66.6 55.9 1.40 1.18 19.1 GB
16 3 15 47.4 69.5 62.1 1.47 1.31 11.8 PB
11 5 5 15.4 21.3 16.0 1.38 1.04 33.1 GB
11 5 9 15.4 22.2 20.9 1.44 1.35 6.3 PB
11 5 10 15.4 22.5 21.4 1.46 1.39 5.0 PB
9 6 4 8.4 11.7 9.6 1.38 1.14 21.8 GB
9 6 5 8.4 11.8 10.8 1.39 1.28 8.8 PB
9 6 15 8.4 13.6 13.5 1.62 1.60 0.7 PB
8 7 4 5.9 8.2 7.1 1.39 1.20 16.0 GB
8 7 6 5.9 8.4 8.1 1.43 1.36 4.7 PB
8 7 10 5.9 9.1 9.1 1.54 1.53 0.3 PB
7 9 4 4.0 5.5 4.6 1.40 1.15 21.2 GB
7 9 8 4.0 6.0 6.0 1.51 1.51 0.2 PB
7 9 20 4.0 6.8 6.7 1.71 1.69 0.9 PB
S 11 5 5 15.4 22.5 17.2 1.46 1.11 31.2 GB
11 5 8 15.4 22.9 21.0 1.49 1.37 8.8 PB
9 6 4 8.4 12.3 10.3 1.46 1.22 19.4 GB
9 6 5 8.4 12.4 11.4 1.47 1.35 8.9 PB
9 6 9 8.4 12.9 12.4 1.53 1.47 3.9 PB
8 7 4 5.9 8.7 7.7 1.47 1.30 12.5 GB
8 7 6 5.9 8.8 8.3 1.49 1.41 5.7 PB
8 7 20 5.9 10.1 9.9 1.70 1.68 1.1 PB
7 9 4 4.0 5.8 5.1 1.47 1.29 14.1 GB
7 9 7 4.0 6.1 5.9 1.53 1.49 2.1 PB
7 9 14 4.0 6.6 6.6 1.66 1.67 20.2 PB

© ASCE 04014092-7 J. Bridge Eng.

J. Bridge Eng.
Table 3. Comparison of Proposed Equations on the Minimum Bending Stiffness, IsðreqÞ
IsðreqÞ (3106 mm4 )
Model aL tp (mm) tu (mm) FEM Eq. (1) Difference (%) Eq. (2) Difference (%) Eq. (14) Difference (%) Eq. (16) Difference (%)
L 3 26 16 94.1 594.1 531.1 30.5 268 38.3 259 100.8 7
5 18 16 100.2 197.1 96.7 28.1 272 55.3 245 119.1 19
7 14 10 64.6 92.8 43.6 25.9 260 43.2 233 82.3 27
M 3 15 8 13.6 99.8 635.9 5.1 262 7.6 244 15.2 12
5 11 8 14.2 39.3 176.6 5.6 261 10.3 228 17.4 22
7 8 5 9.2 15.1 64.2 4.2 254 7.3 221 10.6 15
9 7 6 11.2 10.1 29.4 4.7 258 9.7 213 13.2 18
S 5 11 7 11.5 33.4 189.2 4.8 259 8.6 226 14.6 27
6 9 5 8.4 18.3 116.7 3.8 256 6.6 222 10.2 20
7 8 5 8.5 12.8 50.4 3.6 258 6.6 222 9.7 14
Downloaded from ascelibrary.org by New York University on 05/17/15. Copyright ASCE. For personal use only; all rights reserved.

of parametric numerical analyses to extract a valid design equation to buckling strength of the plates stiffened by the U-ribs. Then, the
replace the explicit formula, Eq. (14), for the required stiffness pre- threshold values for the minimum required stiffness were collected
diction. As a result, nonlinear regression analysis yielded Eq. (15) for through numerical parametric analyses using finite-element mod-
the limiting value of the moment of inertia Is that is required to ensure eling. Regression analyses were then performed on the collected
an antisymmetric mode shape. The regression equation acquired from data through a series of parametric numerical analyses to extract
the Minitab 16 statistical program package is given as follows: a valid design equation. From the results of the comparative study,
the regression equation shows an excellent correlation with the
wL numerically collected data, particularly for the parameter ranges of
IsðreqÞ ¼ × ½1 þ ðn þ 1ÞdFðkR ÞFEM
3ð1 2 m2 Þ wS =wR $ 1:5 and aS $ 5. The proposed design equation corresponds
to the local plate buckling strength (PB mode) and also reflects
 e20:7669 a1:894
L tp3:539 (15)
reasonably the enhanced buckling strength attributable to the ro-
tational restraint effect of the closed-section stiffeners by applying
The coefficient of correlation, the R-squared value, was greater than the strength increment factor, FðkR Þ. For full implementation of the
0.98. Instead of applying Eq. (11), FðkR Þ values determined from the developed method, however, further study on the effect of the ro-
finite-element analysis were considered to obtain the regression tational stiffness of the closed-section stiffeners and the strength
equation, Eq. (15). This is further simplified by adopting an integer increment factor is needed.
form for the exponent coefficients. Because the value of FðkR Þ
calculated from Eq. (11) has a tendency to match well the numer-
ically evaluated value, adopting FðkR Þ from Eq. (11) is expected to Acknowledgments
reflect well the effect attributable to the rotational restraint of closed-
section stiffeners. Thus, the regression equation is expected to be of This work was supported by a National Research Foundation of
the form Korea (NRF) grant funded by the Korean Government (MSIP)
(No. 2013-026539).
a2L tp3:5 wL
IsðreqÞ ¼ × ½1 þ ðn þ 1Þd × FðkR Þ (16) References
6ð1 2 m2 Þ

Table 3 presents a comparison between the proposed design for- AASHTO. (2002). Standard specifications for highway bridges, 17th Ed.,
mulas in Eqs. (1), (2), (14), and (16) regarding the required bending Washington, DC.
AASHTO. (2012). LRFD bridge design specifications, 6th Ed., Washington,
stiffness of the closed-section stiffeners, where the differences in the
DC.
threshold value of Is are also shown. It can be seen that Eq. (1) re- ABAQUS 6.12 [Computer software]. Providence, RI, Dassault Systèmes.
quires excessively larger values for the longitudinal stiffer stiffness Choi, B. H. (2002). “Design requirements for longitudinal stiffeners for
as stated previously. Eq. (2), which is proposed based on the open- horizontally curved box girders.” Ph.D. dissertation, Auburn Univ.,
section longitudinal stiffener, does not match well the threshold Auburn, AL.
value data. Moreover, it is found that Eq. (14), which is derived from Choi, B. H., and Yoo, C. H. (2005). “Strength of stiffened flanges in
this study based on the theoretical approximation methods, has a horizontally curved box girders.” J. Eng. Mech., 10.1061/(ASCE)0733-
tendency to underestimate the bending stiffness requirement, and 9399(2005)131:2(167), 167–176.
that the regression equation Eq. (16) shows the best correlation with Chou, C. C., Uang, C. M., and Seible, F. (2006). “Experimental evaluation
the threshold value of the closed-section stiffener rigidity. The error of compressive behavior of orthotropic steel plates for the new San
Francisco–Oakland Bay Bridge.” J. Bridge Eng., 10.1061/(ASCE)1084-
in this application of Eq. (16) is smaller than the errors for the other
0702(2006)11:2(140), 140–150.
equations and is on the side of caution. Minitab 16 [Computer software]. State College, PA, Minitab.
Qiao, P., and Shan, L. (2005). “Explicit local buckling analysis and design of
fiber–reinforced plastic composite structural shapes.” Compos. Struct.,
Significant Findings and Concluding Remarks
70(4), 468–483.
Timoshenko, S. P., and Gere, J. M. (1961). Theory of elastic stability, 2nd
This study has examined the bending stiffness requirements for the Ed., McGraw Hill, New York.
closed-section longitudinal stiffeners of plates under uniaxial com- Yoo, C. H., Choi, B. H., and Ford, E. M. (2001). “Stiffness requirements for
pression. An explicit formula for the minimum stiffness requirement longitudinally stiffened box-girder flanges.” J. Struct. Eng., 10.1061/
was derived by applying theoretical approximate equations for the (ASCE)0733-9445(2001)127:6(705), 705–711.

© ASCE 04014092-8 J. Bridge Eng.

J. Bridge Eng.

You might also like