You are on page 1of 39

Thin-Walled Structures 40 (2002) 45–83

www.elsevier.com/locate/tws

Ultimate strength formulations for stiffened


panels under combined axial load, in-plane
bending and lateral pressure: a benchmark study
Jeom Kee Paik*, Bong Ju Kim
Department of Naval Architecture and Ocean Engineering, Pusan National University,
30 Jangjeon-Dong, Gumjeong-Gu, Busan 609-735, South Korea

Received 14 March 2000; revised 17 July 2001; accepted 30 July 2001

Abstract

This paper develops advanced, yet design-oriented ultimate strength expressions for stiff-
ened panels subject to combined axial load, in-plane bending and lateral pressure. The collapse
patterns of a stiffened panel are classified into six groups. It is considered that the collapse
of the stiffened panel occurs at the lowest value among the various ultimate loads calculated
for each of the collapse patterns. The panel ultimate strengths for all potential collapse modes
are calculated separately, and are then compared to find the minimum value which is then
taken to correspond to the real panel ultimate strength. The post-weld initial imperfections
(initial deflection and residual stress) are included in the developed panel ultimate strength
formulations as parameters of influence. The validity of the developed formula is confirmed
by comparing with the mechanical collapse tests and nonlinear FEA. A comparison of the
present method is also made with theoretical solutions from the Det Norske Veritas classi-
fication society design guideline. Important insights developed are summarized.  2001
Elsevier Science Ltd. All rights reserved.

Keywords: Stiffened panel; Ultimate limit state design; Ultimate strength; Overall collapse; Plate-induced
failure; Stiffener-induced failure; Local buckling of stiffener web; Tripping; Gross yielding

* Corresponding author. Tel: +82-51-510-2429; fax: +82-51-512-8836.


E-mail address: jeompaik@pnu.edu (J. K. Paik).

0263-8231/02/$ - see front matter  2001 Elsevier Science Ltd. All rights reserved.
PII: S 0 2 6 3 - 8 2 3 1 ( 0 1 ) 0 0 0 4 3 - X
46 J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83

1. Introduction

A limit state is formally defined as a condition for which a particular structural


member or an entire structure fails to perform the function that it has been designed
beforehand for. From the special view point of a structural designer, four types of
limit states are usually considered, namely serviceability limit state, ultimate limit
state, fatigue limit state and accidental limit state. In design, these various types of
limit states may be considered against different safety levels, the actual safety level
to be attained for a particular type of limit state being a function of the perceived
consequences to be accounted for and ease of recovery to be incorporated in design.
Of concern in this paper is the ultimate limit state design of steel stiffened panels.
Theoretically, the primary modes of overall failure for a stiffened panel subject
to predominantly compressive loads may be categorized into the following six
groups, namely

앫 Mode I: Overall collapse after overall buckling of the plating and stiffeners as a
unit, see Fig. 1(a),
앫 Mode II: Plate-induced failure by yielding at the corners of plating between stiff-
eners, see Fig. 1(b),
앫 Mode III: Plate-induced failure by yielding of a plate–stiffener combination at
mid-span, see Fig. 1(c),
앫 Mode IV: Stiffener-induced failure by local buckling of stiffener web, see Fig.
1(d),
앫 Mode V: Stiffener-induced failure by lateral–torsional buckling of stiffener, see
Fig. 1(e), and
앫 Mode VI: Gross yielding.

Mode I typically represents the collapse pattern when the stiffeners are relatively
weak. In this case, the stiffeners can buckle together with plating, the overall buckling
behavior remaining elastic. The stiffened panel can normally sustain further loading
even after overall buckling in the elastic regime occurs and the ultimate strength is
eventually reached by formation of a large yield region inside the panel and/or along
the panel edges. In Mode I, the panel behaves as an ‘orthotropic plate’.
The other groups (i.e., Modes II–VI) normally take place when the stiffeners are
relatively strong so that the stiffeners remain straight until the plating between stiff-
eners buckles or even collapses locally. The stiffened panel will eventually reach
the ultimate limit state by failure of stiffeners together with associated plating. It is
noted that the stiffened panel with weak stiffeners where failure of stiffeners occurs
prior to buckling of plating normally follows Mode I, i.e., failure after overall buck-
ling occurs in the elastic regime.
Mode II typically represents the collapse pattern wherein the panel collapses by
yielding at the corners of plating between stiffeners, which is usually termed a plate-
induced failure at ends. This type of collapse can also occur in some cases when
the panel is predominantly subjected to biaxial compressive loads. Mode III indicates
a failure pattern in which the ultimate strength is reached by column or beam-column
J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83 47

Fig. 1. (a) Mode I: Overall collapse after overall buckling of the plating and stiffeners as a unit. (b).
Mode II: Plate-induced failure by yielding at the corners of plating between stiffeners. (c). Mode III:
Plate-induced failure by yielding of plate–stiffener combination at mid-span. (d). Mode IV: Stiffener-
induced failure by local buckling of the stiffener web. (e). Mode V: Stiffener-induced failure by lateral–
torsional buckling of stiffener.

type collapse of the plate–stiffener combination with the associated effective


(reduced) plating. Mode III typically takes place by yielding of the plate–stiffener
combination at mid-span, which is usually termed a plate-induced failure at mid-span.
Modes IV and V failures typically arise when the ratio of stiffener web height to
stiffener web thickness is too large and/or when the type of the stiffener flange is
inadequate to remain straight so that the stiffener web buckles or twists sideways.
Mode IV represents a failure pattern in which the panel collapses by local buckling
48 J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83

of stiffener web, while Mode V can occur when the ultimate strength is reached by
lateral–torsional buckling (also called tripping) of stiffener.
Mode VI typically takes place when the panel slenderness is very small (i.e., the
panel is very stocky or thick) and/or when the panel is predominantly subjected to
the axial tensile loading so that neither local nor overall buckling occurs until the
panel cross section yields entirely.
Calculation of the ultimate strength of the stiffened panel under combined loads
taking into account all of the possible failure modes noted above is not straightfor-
ward, because of the interplay of the various factors previously mentioned such as
geometric/material properties, loading, post-weld initial imperfections (i.e., initial
deflection and residual stress) and boundary conditions. As an approximation, this
paper considers that the collapse of stiffened panels occurs at the lowest value among
the various ultimate loads calculated for each of the above collapse patterns. This
leads to an easier alternative wherein one calculates the ultimate strengths for all
collapse modes mentioned above separately and then compares them to find the
minimum value, which is then taken to correspond to the real panel ultimate strength.
The behavior of steel stiffened panels normally depends on a variety of influential
factors, namely geometric/material properties, loading characteristics, initial imper-
fections, boundary conditions and existing local damage related to corrosion, fatigue
crack and denting. A number of studies on the ultimate strength for stiffened panels
have been undertaken experimentally, numerically and theoretically. Comparisons
between some of these methods have been performed by many investigators [1–4].
Smith [5] presents a series of tests on full scale welded steel grillages subjected
to a combination of axial compression and lateral pressure. Efforts on experimental
investigation for stiffened panel collapse behavior are made by Tanaka and Endo
[6], Hu et al. [7] and Hopperstad et al. [8], among others. Smith and his colleagues
[9–12] provide an extensive contribution to the ultimate strength design for ship
stiffened panels under combined in-plane and lateral pressure loads. The panel ulti-
mate strength formulation based on column or beam-column type collapse pattern
is typically based on the so-called Perry–Robertson formula [13–19]. Local buckling
of stiffener webs is studied by Paik et al. [20]. Many researchers have studied tripping
of stiffeners theoretically, numerically and experimentally. Earlier work that used
classical theory of thin-walled bars has been summarized and expanded by Bleich
[21]. During the 1970s and 1980s, further studies have been undertaken by Faulkner
et al. [22–25] and Adamchak [26]. Hughes [15] has reviewed and summarized some
of these studies. During the 1990s, in addition to the tripping problem under axial
compression alone [27–30], the effect of combined axial compression and lateral
loads has been studied by Hughes and Ma [31,32] and Hu et al. [33].
While the previous studies being useful for prediction of the stiffened panel ulti-
mate strength, most studies are based on one or two collapse patterns. As previously
noted, various failure modes can potentially involve in collapse of a stiffened panel
and the panel ultimate strength formulation should accommodate all these potential
collapse patterns mentioned above. The stiffened panel in ships is generally subjected
to combined in-plane and lateral pressure loads. In-plane loads include axial load
and in-plane bending, which are mainly induced by overall hull girder bending. Lat-
J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83 49

eral pressure is due to water pressure and cargo. These load components are not
always applied simultaneously, but more than one normally exist and interact. Thus
it is of crucial importance to evaluate the panel ultimate strength taking into account
the effect of combined loading. Since the post-weld initial imperfections in the form
of initial deflections and residual stresses exist in ship stiffened panels and can affect
(reduce) significantly the ultimate strength, such welding induced initial imperfec-
tions should be included in the strength calculations as parameters of influence.
The aim of the present paper is to develop an advanced, yet design-oriented
method for evaluating the ultimate limit state of stiffened panels under combined
axial load, in-plane bending and lateral pressure. It is considered that the collapse
of stiffened panels occurs at the lowest value among the various ultimate loads calcu-
lated for each of the six collapse patterns mentioned above. The panel ultimate
strengths for all collapse modes are calculated separately and then compare them to
find the minimum value which is then taken to correspond to the real panel ultimate
strength. The post-weld initial imperfections are included as parameters of influence.
To test the validity of the developed method, verification examples are studied by
comparing with the mechanical collapse tests and nonlinear FEA. A comparison of
the present method is also made with the Det Norske Veritas design guideline for
marine stiffened plate structures.

2. Modeling of steel stiffened panels

Fig. 2 shows a typical steel stiffened plate structure. Its response can be studied
at three levels, namely the entire structure level, the stiffened panel level and the
bare plate level. This paper is primarily concerned with the second level (i.e., the
stiffened panel level) in which a panel is supported by girders along longitudinal

Fig. 2. A typical stiffened plate structure.


50 J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83

edges and by frames (or floors) along transverse edges. In the following, some basic
idealizations made for development of the panel ultimate strength formulations are
described.

2.1. Panel geometry

The length and breadth of the stiffened panel are denoted by L and B, respectively.
The thickness of the plate is t. For generality, the ultimate strength formulations
developed in this paper consider that the panel can have stiffeners in one or both
directions. The numbers of x- or y-stiffeners are nsx or nsy. The cross-sectional area
of a single stiffener in the x or y direction is denoted by Asx = hwxtwx + bfxtfx or
Asy = hwytwy + bfytfy. The spacing of the stiffeners is b = B/(nsx + 1) between x-stiff-
eners or a = L/(nsy + 1) between y-stiffeners.
Fig. 3 shows typical cross-sections of longitudinal or transverse stiffeners. The
moments of inertia of x- or y-stiffeners with attached effective plating are denoted
by Ix and Iy, respectively. The cross-sectional areas of a stiffener together with
attached effective plating are denoted by Ax and Ay for x- and y-stiffeners, respect-
ively.

Fig. 3. Typical cross-section types for longitudinals and transverses.


J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83 51

2.2. Panel material

The material used in stiffened panels of merchant ship structures is usually mild
or high tensile steel. Note that the use of aluminum alloys is increasing in the design
and fabrication of high speed ship structures. The elastic modulus of both plating
and stiffeners is E and the Poisson ratio is n. The panel ultimate strength formulations
are designed so that the yield stress of the plate can be different from that of the
stiffeners. The yield stress of plating and stiffeners is denoted by sop and sos, respect-


b sop
ively. The plate reduced slenderness ratio is denoted by b = .
t E

2.3. Panel boundary condition

It is assumed that the panel edges are simply supported [i.e., with zero deflection
and zero rotational restraints along all (four) edges] with all edges kept straight. For
steel stiffened plate structures, this assumption normally provides some pessimistic
but adequate results.

2.4. Load effects

When a continuous plated structure is subjected to external loads, the load effects
(e.g., stress, displacement) can normally be calculated by linear elastic finite element
method or classical theory of structural mechanics. When a ship hull is under vertical
bending, deck panels are predominantly subjected to longitudinal axial compression
in sagging or longitudinal axial tension in hogging, while bottom panels are subjected
to combined longitudinal axial compression/tension and lateral pressure. Upper side
shells are normally subjected to combined longitudinal axial compression/tension
and longitudinal in-plane bending, while lower side shells are subjected to combined
axial compression/tension, longitudinal in-plane bending and lateral pressure loads.
Therefore, the number of load components considered in the present paper is of three
types, namely longitudinal axial compression/tension, longitudinal in-plane bending
and lateral pressure, as shown in Fig. 4.
The average value of longitudinal stress is denoted by sxav = (sx1 + sx2)/2 and the
longitudinal axis stress at the most highly stressed x-stiffener is denoted by sxM. It
is idealized that the panel is subjected to uniform lateral pressure p; an average value
of lateral pressure is taken for side shell panels of ships. The in-plane bending stress
is determined by sxb = |sx1⫺sxav| = |sx2⫺sxav|.

2.5. Fabrication related initial imperfections

Both plating and stiffeners may have post-weld initial deflections as shown in Fig.
5(a). The maximum initial deflection of plating between stiffeners is denoted by wopl.
The column type initial deflection of stiffeners is denoted by wosx and wosy for x-
and y-stiffeners, respectively. Both plating and stiffeners may also have welding
52 J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83

Fig. 4. A stiffened steel panel under combined longitudinal axial load, longitudinal in-plane bending
and lateral pressure.

Fig. 5. (a) An example pattern of post-weld initial deflections in steel stiffened panels.(b) Idealization
of the distribution of welding induced residual stresses in the plating between stiffeners in the x and y
directions (Tens.=tension, Comp.=compression).

induced residual stresses. Fig. 5(b) shows the typical idealization of welding induced
residual stresses in plating between stiffeners. It is often idealized that the stiffeners
may have uniform ‘equivalent’ compressive residual stress over the stiffener web
cross-section which are denoted by srsx and srsy for x- and y-stiffeners, respectively.
J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83 53

3. Panel ultimate strength formulations

This paper calculates the panel ultimate strengths for various collapse modes sep-
arately. The real panel ultimate strength is then considered to be the lowest value
among the various ultimate loads calculated for each of the collapse patterns. In a
sense, contribution of the present paper is integration of past efforts which have been
separately made for individual collapse patterns. Since a more elaborate description
for panel ultimate strength formulations of all collapse modes considered may be
found in Paik et al. [34], this paper focuses on the interpretation of the basic ideas
how to calculate the ultimate strength for each of various collapse modes. Mode VI
(i.e., failure by gross yielding) is checked as part of each of the other five failure
modes.

3.1. Panel ultimate strength formulation for collapse mode I

When the panel has a number of relatively small stiffeners, the stiffened panel is
replaced by an ‘orthotropic plate’. In this case, the stiffeners are smeared into the
plating. The orthotropic plate theory taking into account large deflection effects [35]
can be used to calculate the membrane stress distribution of a stiffened panel whose
failure follows Mode I. The ‘orthotropic plate’ is considered to have an equivalent
yield stress which is a representative yield stress of both plating and stiffeners, which
may be taken as soeq = (Btsop + nsxAsxsos)/(Bt + nsxAsx).
The effect of in-plane bending moment on the ultimate strength of the orthotropic
plate may be neglected for practical design purposes (that is, it is considered that
the ‘orthotropic plate’ is subjected to combined sxav and p). This is because some
part of the panel is subjected to axial tension while axial compression is applied in
the other part. The tensiled region works toward stabilizing the plate until the ultimate
strength is reached. However, in-plane bending is an important factor in the ultimate
strength calculations for a panel which collapses by failure of the stiffeners.
The influence of post-weld initial deflection is taken into consideration in the
analysis, but the effect of welding induced residual stresses is ignored for Mode I.
This is because when the stiffened panel is subjected to predominantly axial com-
pressive loads, the compressive residual stresses will work toward reducing the
strength but the tensile residual stresses will contribute resistance to the applied loads,
and vice versa for axial tensile loading. The decrease/increase of load carrying
capacity due to welding induced residual stresses may in effect be offset by the
effect of the increase/decrease of the ultimate strength.
The panel ultimate strength following mode I is denoted by sIxu. For the detailed
derivation of the ultimate strength formulation following collapse mode I, the reader
may be referred to Paik et al. [35].

3.2. Panel ultimate strength formulation for collapse mode II

In Mode II, the stiffened panel is considered to collapse if the corners of plating
between the most highly stressed stiffeners yield after plating between stiffeners
54 J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83

buckles or collapses. It is in this case idealized that the plating between stiffeners
is subjected to the axial stress sxM at the most highly stressed stiffener and lateral
pressure p. Also, the post-weld initial imperfections in the form of initial deflection
and residual stress are included as parameters of influence. The panel ultimate
strength formulation for collapse mode II is then given by

冉 冊 冉 冊冉 冊 冉 冊
sxmax 2 sxmax symax
sop

sop sop

symax
sop
2
⫽1 (1)

where sxmax and symax are maximum membrane stresses of plating between stiffeners
in the x or y-direction, respectively, which are expressed as functions of sxM and p
as well as initial imperfections.
The solution of Eq. (1) with regard to sxM is then denoted by sIIxu. The membrane
stress distribution inside the plating under combined loads is analyzed by solving
the governing nonlinear differential equations of large deflection (isotropic) plate
theory. For an elaborate description for the membrane stress distribution inside the
plating, the reader may be referred to Paik et al. [36].
For predominant lateral pressure loading, two upper limits of the ultimate pressure
load for stiffened panels are also typically relevant. The first upper limit is given as
the ultimate lateral pressure load as previously obtained for Mode I overall grillage
collapse. The second upper limit is the ‘critical’ lateral pressure (denoted by pcr) of
plating (between stiffeners) clamped at all edges, as obtained by rigid plastic theory
[37], which is given by
s2op
pcr ⫽ Cp (2)
Eb2
where
Cp ⫽ 12/[√3 ⫹ (b/a)2⫺b/a]2

Thus, the ultimate lateral pressure load should not be greater than the two upper
limits noted above. While Eq. (2) is used for illustrative examples of this paper, the
coefficient, Cp of Eq. (2), is often replaced by Cp = 4[1 + (b/a)2] for more pessimistic
computations, which corresponds to the rigid plastic theory result for a plate simply
supported at all edges. Note that the rigid plastic theory formulae noted above do
not account for the membrane effects and thus they may predict the critical lateral
pressure pessimistically.

3.3. Panel ultimate strength formulation for collapse mode III

For this collapse mode, the beam-column approach using the Perry–Robertson
formula is typically applied. The Perry–Robertson formulation assumes that the stiff-
ener with associated plating will collapse as a ‘beam-column’ when the maximum
compressive stress in the extreme fiber reaches the yield strength of the material.
The two possible collapse modes for the Perry–Robertson formulation are usually
J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83 55

considered depending on the failure of the most highly stressed fiber, i.e., ‘plate-
induced failure’, and ‘stiffener-induced failure’. The former is related to yielding of
the associated plating due to compression while the latter may result from either
yielding of the extreme stiffener fiber (without twisting of stiffener) or tripping of
stiffener (with twisting of stiffener).
When subjected to combined axial compression (sxM) and lateral line load (
q = pb, multiplied by uniform lateral pressure p and the breadth b of plating between
longitudinal stiffeners), the ultimate strength of a plate–stiffener combination can be
calculated using the Perry–Robertson formula. While a more elaborate description
may be found in Hughes [15], the Perry–Robertson ultimate strength expression for
the plate–stiffener combination under combined sxM (compression) and q accounting
for the effect of initial deflection is derived (when compression is taken as positive),
as follows

sIII
xu

sox
⫽ Kr
1
2冋冉
1⫺m ⫹
1⫹h
l2
⫺ 冊 冪冉 1
4
1⫺m ⫹
1 ⫹ h 2 1⫺m
l2
⫺ 2
l 冊 册 (3)

where l=slenderness ratio of longitudinals with attached effective plating,


Axzc Mqmax zc
h= (w + wosx)=initial deflection related parameter, m = =lateral load
Ix qmax sos Ix
related parameter, Mqmax=maximum bending moment due to lateral load alone (at
mid-span) which may be taken as Mqmax = qa2/8, wqmax=maximum deflection due to
lateral load alone (at mid-span) which may be taken as, wqmax = 5qa4/(384EIx),
wosx=column type initial deflection of x-stiffener, zc = zpx + 0.5t for plate-induced
failure and zc = zfx + 0.5tfx for stiffener-induced failure (see Fig. 6), Kr=knock-down
factor due to residual stress which is in the illustrative examples of this paper taken
as Kr = 1.03⫺0.08|srsx/sox|ⱕ1.0 [14] (For a more refined treatment of the residual

Fig. 6. Geometric properties of longitudinals with attached effective plating.


56 J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83

stress effect on beam-column strength, the reader may be referred to Trahair and
Bradford [38]).
It is important to realize that Eq. (3) does not involve local buckling of stiffener
web or tripping of stiffener. Since compression is defined as negative, sIII xu will
eventually take a negative value for sxM in compression. For sxM in tension, the
panel ultimate strength will approximately equal the equivalent yield strength, i.e.,
xu = sos.
sIII
In the course of the present study, it was recognized that the Perry–Robertson
formula predictions based on stiffener-induced failure mode are too pessimistic.
Therefore, we exclude the stiffener-induced failure condition, i.e., yielding at tip of
the stiffener from Mode III panel ultimate strength calculations. The possibility of
the stiffener-induced failure mode due to either local buckling of stiffener web or
lateral–torsional buckling of stiffener is dealt with in Modes IV or V, respectively
later on.
It was also realized that for panels with relatively weak stiffeners the Perry–Rob-
ertson approach can predict an ultimate axial compressive strength which is in some
cases even less than that of the bare plate (i.e., panel without stiffeners), while the
orthotropic plate approach gives the value which is greater than the real panel ulti-
mate strength. In this regard, a lower limit of the panel ultimate strength following
Mode III, given by the weighted average of the bare plate ultimate strength and the
orthotropic plate ultimate strength as long as sxM is compressive, is used:
xu ⫹ teqsxu
tsGB GO

| ||
xu ⱖ
sIII
t ⫹ teq | (4)

where sGB GO
xu and sxu are the ultimate strengths for the bare plate and the orthotropic
plate under sxM plus p, respectively, the latter being calculated from the panel ulti-
mate strength equation for Mode I, and the former being also calculated from the
same equation but removing stiffeners from the entire panel. teq is the equivalent
plate thickness which is taken as teq = t + nsxAsx/B. In this paper, this fictitious col-
lapse pattern is denoted as ‘Mode I (mean)’ for record keeping purposes.

3.4. Panel ultimate strength formulation for collapse mode IV

The web or flange of a stiffener in a stiffened panel can locally buckle, a possibility
that must usually be considered for built-up sections. Such occurrence of local buck-
ling in the stiffener cross-section can sometimes be quite a sudden phenomenon
resulting in subsequent unloading of the stiffened panel, particularly with use of flat
bar stiffeners.
In this case, once such stiffener web buckling occurs, the buckled or collapsed
plating is left with essentially little stiffening and thus overall stiffened panel collapse
may follow immediately. In Mode IV, it is assumed that the panel reaches the ulti-
mate limit state immediately after the local buckling of the stiffener web at its most
highly stressed location.
The local buckling of the stiffener web or flange, and the buckling/collapse of
J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83 57

plating between stiffeners normally interact, and can take place in any order,
depending on the dimensions of plating and stiffener. Clearly, buckling of the stiff-
ener flange or web prior to the inception of buckling in plating between stiffeners is
normally an undesirable failure mode which should be specifically avoided by design.
In most design guidelines from the classification societies, it is normally con-
sidered that the stiffener web or flange should not buckle until the plating between
stiffeners buckles or collapses. More exactly, the local buckling strength of the stiff-
ener web or flange depends on the torsional rigidities of the adjacent members to
which they are attached, among other factors.
One assumption that is implicit in our present description is that the stiffener
flange, where one is present, does not buckle prior to the buckling of the web. Flange
buckling is normally not a concern with rolled sections, but in built-up sections, it
needs to be considered and controlled. This is normally carried out in design by
meeting certain flange slenderness limits. Hence, the Mode IV ultimate strength may
be estimated from the local buckling strength of the stiffener web taking into account
the influence of rotational restraints along the plate–stiffener and stiffener web-flange
junctions. This is, however, likely to be pessimistic, a better approximation being
possibly to take the panel ultimate strength as the average of the applicable plate
and stiffener web ultimate strengths.
The local buckling strength of the stiffener web depends significantly on the tor-
sional rigidities along the plate–stiffener or stiffener web-stiffener flange intersection,
among other factors. While a detailed description may be found in Paik et al. [20],
the elastic buckling strength for the stiffener web considering rotational restraints
between plating and stiffeners and between stiffener web and flange can be given
by considering the effect of residual stress srsx (when compression is taken as
positive), as follows
p2Dwx
xE ⫽ kwx
sW ⫺s (5)
h2wxtwx rsx
where sW xE is the elastic local buckling strength of x-stiffener web and kwx is the
elastic buckling coefficient of x-stiffener web. Expressions for these are given in
Paik et al.[20], established on the basis of curve fits to the more refined solutions
of the governing differential equation applicable to buckling. Dwx = Et3wx/12(1⫺n2)
A stocky stiffener web may buckle in the elastic-plastic regime. To account for
the effect of plasticity, the so-called Johnson–Ostenfeld formula with the yield stress
of the stiffener sos is used, resulting in the collapse strength of x-stiffener web
(denoted by sW xu). When sxM is tensile, sxu = sos is normally taken.
W

The ultimate axial strength of the entire panel by the stiffener web buckling of
the most highly stressed x-stiffener can then be obtained as the weighted average of
the ultimate strength of the plating between stiffeners and the collapse strength of
the x-stiffener web for the entire cross sectional area of the panel, as follows
sLxubt ⫹ sW
xuAsx
xu ⫽
sIV (6)
bt ⫹ Asx
where sLxu=ultimate strength of plating between stiffeners [36].
58 J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83

3.5. Panel ultimate strength formulation for collapse mode V

The lateral–torsional buckling (also called tripping) of stiffeners is a phenomenon


in which the failure of a stiffened panel occurs subsequent to the stiffener twisting
sideways about the edge of the stiffener web attached to the plating. When the tor-
sional rigidity of the stiffener is small or the stiffener flange is weak, this phenom-
enon is more likely to take place.
Like the stiffener web buckling previously described, tripping can be a relatively
sudden phenomenon resulting in subsequent unloading of the stiffened panel. Once
tripping occurs, the buckled or collapsed plating is left with little stiffening and thus
overall collapse may follow. In Mode V, the stiffened panel is considered to collapse
if tripping occurs. If the height of stiffener web increases in comparison to its thick-
ness, the stiffener web is likely to deform and local buckling can in some cases
occur. This type of failure is included in Mode IV. Hence, for purposes of Mode
V, we consider a type of tripping wherein the cross-section of the stiffener web does
not deform locally, consistent with a similar assumption used in ordinary beam-
column theory.
It therefore follows that the tripping strength of the flat bar type stiffener equals
the local buckling of the stiffener web, and such a case is to be treated as part of
Mode IV and not Mode V. The ultimate strength of the stiffened panel is then
approximated as a weighed average of the ultimate strengths of the plating and the
tripping strength of the stiffener. The intention behind the averaging proposed is to
avoid an overly pessimistic estimate of the stiffened panel ultimate strength.
While the rotational restraints between the stiffener web and the attached plating
may generally play an important role in tripping behavior or local buckling of stiff-
ener web (the latter being previously discussed in Mode IV), rotational restraint
effects from the plating may be ignored if the plating between stiffeners buckles
prior to tripping so that the contribution of the attached plating to restrict the rotation
of the stiffener web at the plate–web junction is small and thus it may be considered
that the stiffener and the attached plating are pin-joined. This assumption will argu-
ably result in a lower bound solution of the tripping strength since plate rotational
restraints will always exist to some extent, while the effect of buckled plating may
approximately be incorporated using an effective plate width in calculating the
restraint properties for the plate–stiffener combination.
The elastic tripping strength of angle or tee-type stiffeners under combined axial
compression (sxM) and uniform lateral pressure line load q = pb (i.e., multiplied by
uniform lateral pressure p and the breadth b of plating between longitudinal
stiffeners) in the x direction can be calculated by applying the principle of the mini-
mum potential energy [31]. The effect of plasticity is then incorporated by using the
Johnson–Ostenfeld formula, resulting in the inelastic tripping strength denoted by
sTxu. When sxM is tensile, sTxu = sos may be used. Also, sTxu = 0 is used when the
panel does not have stiffeners in the x direction.
Applying an argument similar to that used for Mode IV, the ultimate longitudinal
axial strength of the stiffened panel subsequent to tripping (denoted by sVxu) is
J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83 59

approximated by the weighted average of local collapse strength of plating and stiff-
ener tripping strength in the x direction, as follows:
sLxubt ⫹ sTxuAsx
sVxu ⫽ (7)
bt ⫹ Asx
where sLxu=ultimate strength of plating between stiffeners [36].

4. Verification examples

To test the validity of the panel ultimate strength formulations, verification


examples are now considered. In these examples, the ultimate strength formulations
are compared with experimental, numerical and theoretical results. Predictions from
the design oriented procedures proposed in this paper were automated using a com-
puter program called ALPS/ULSAP which stands for Analysis of Large Plated
Structures/ULtimate Strength Analysis of Panels.
The nonlinear FEM computes the elastic-plastic large deflection behavior of the
stiffened panel as an assembly of both plating and stiffeners until the ultimate
strength is reached. In the FE computations, a sufficiently fine mesh is taken for
stiffeners as well as plating so that the local buckling or tripping of stiffeners is
automatically considered as those for the mechanical tests. The SPINE solutions are
obtained using a semi-analytical method [39] for elastic-plastic large deflection
analysis of plates and stiffened panels when stiffener tripping is not considered. The
DnV CN 30.1 results are obtained by the design guideline of the Det Norske Veritas
[40]. We have documented at length the mechanical collapse test data used in our
comparisons so that they may be of use to others as well.

4.1. Ultimate strength characteristics of plating

Since local collapse of plating between stiffeners is a primary failure mode of


stiffened panels and grillages, verification examples of the present method to bare
plates (unstiffened plating) are studied. Fig. 7 compares theoretical results of
ALPS/ULSAP and DnV CN 30.1 methods with the mechanical collapse tests and
the nonlinear FEA for long plating with different plate aspect ratios and under longi-
tudinal axial compressive loads. For more details of the test data, Ellinas et al. [41]
may be referred to. In the FEA, two types of the unloaded plate edge condition
are applied; (1) the unloaded plate edges move freely in plane, and (2) they are
kept straight.
The mechanical collapse tests involve various levels of both initial deflections and
residual stresses. For the FE analyses, an ‘average’ level of initial deflections is
considered, while the welding residual stresses are not included. The ALPS/ULSAP
method deals with initial imperfections as direct parameters of influence, while the
DnV CN 30.1 method implicitly considers an ‘average’ level of initial imperfections
in the ultimate strength calculations.
60 J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83

Fig. 7. Comparison of the ALPS/ULSAP and DnV CN 30.1 methods with mechanical test results and
FEA for long plating under uniaxial compression, reference numbers being extracted from Ellinas et
al. [41].

It is seen from Fig. 7 that the ultimate strengths as obtained from the DnV CN
30.1 method correspond well with those from ALPS/ULSAP with initial deflection
of w0 = 0.1b2t and residual stress of 15% yield stress. The FE predictions with the
edge condition (1) are smaller than those with the edge condition (2), as would be
expected. It may be surmised from Fig. 7 that in FE computations, keeping the
unloaded plate edges straight is more relevant for thick or medium thickness plating,
while it is more appropriate for thin plating to take the boundary condition that the
unloaded edges move freely in plane. When considering that most actual steel plating
of marine structures is relatively thick or in medium thickness, i.e., with the plate
reduced slenderness ratio in the range of 1.0–3.0, however, the boundary condition
(2), i.e., that the unloaded plate edges are kept straight, may be more appropriate
for practical design purposes.
Fig. 8 compares the ALPS/ULSAP and DnV CN 30.1 methods with the mechan-
ical collapse test results by Yamamoto et al. [42] for long plating of a/b = 3 under
combined longitudinal axial compression and lateral pressure loads. The SPINE sol-
utions accounting for the membrane effect are also compared. The model uncer-
tainties of the ALPS/ULSAP method against the Yamamoto testing are
mean = 0.967 and COV = 0.064.
J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83 61

Fig. 8. (a) Comparison of the ALPS/ULSAP and DnV CN 30.1 methods with the Yamamoto collapse
test results for plating under combined axial compression and lateral pressure, b = 3.508. (b) Comparison
of the ALPS/ULSAP and DnV CN 30.1 methods with the Yamamoto collapse test results for plating
under combined axial compression and lateral pressure, b = 2.554. (c) Comparison of the ALPS/ULSAP
and DnV CN 30.1 methods with the Yamamoto collapse test results for plating under combined axial
compression and lateral pressure, b = 3.084.

4.2. Ultimate strength characteristics of longitudinally stiffened panels

The ultimate strength characteristics of longitudinal stiffened panels, i.e., with flat-
bar, angle or tee-type stiffeners in the longitudinal direction, are now studied by
comparing the ALPS/ULSAP method with mechanical collapse test results and non-
linear FEA as well as the DnV CN 30.1 design guideline.

4.2.1. The Tanaka and Endo mechanical collapse tests


Tanaka and Endo [6] carried out a series of experimental and numerical FE investi-
gations on the ultimate compressive strength characteristics of longitudinally stiff-
ened panels having three flat bar stiffeners, which were intended to fail by local
62 J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83

buckling or tripping of longitudinal stiffeners (Note that for flat-bar stiffeners, local
buckling of stiffener web is treated to be identical to tripping).
Fig. 9 shows a typical test structure from the Tanaka and Endo study. To account
for the effect of adjacent panels on the collapse behavior of the central panel, three
bay models with two adjacent (dummy) stiffened panels and supported by two trans-
verse frames were employed. The plate thickness of two adjacent plating and stiffener
was 1.2–1.3 times that of the center plate so that collapse of adjacent panels prior
to the collapse of the center panel was prevented. Tables 1–3 represent
geometric/material properties and initial imperfections for the Tanaka and Endo
test structures.
Both the panel ultimate strength formula calculations and the Tanaka and Endo
FEA were carried out only for the longitudinally stiffened panel at the center with
all edges presumed simply supported and kept straight. Tables 4 and 5 indicate the
comparison results of the ultimate strength formulations with the experiments and
nonlinear FE solutions as applicable. The ALPS/ULSAP method predicts that except
for Nos. D0 and D3 all test structures fail by Mode IV (initiated by local buckling
of stiffener web or tripping) which corresponds well to the experiments. Figs. 10
and 11 show correlations between ALPS/ULSAP method, DnV method, experiment
and FEA.
For Nos D1 and D3, the experiment provided the ultimate strengths greater than
the material yield stress. For No. D4, the panel ultimate strength obtained by the
experiment was 99% of yield stress. This may be due to a strain-hardening effect.
It is noted that mean and COV of the Tanaka and Endo FEA against their experiments
is 0.978 and 0.139, respectively. Also, the ALPS/ULSAP method shows
mean = 0.953 and COV = 0.114 against experiments and mean = 0.979 and
COV = 0.060 against FEA, while the DnV CN 30.1 method shows mean = 0.682

Fig. 9. The Tanaka and Endo test structure for longitudinally stiffened panels under uniaxial com-
pression, with two adjacent dummy panels.
Table 1
Mean values of geometric and material properties for the Tanaka and Endo test structuresa

Structure no. a (mm) B (mm) t (mm) nsx (mm) hwx (mm) twx (mm) sop (MPa) sos (MPa) soeq (MPa)

D0 1080 1440 6.15 3 110.0 9.77 234.2 287.1 248.3


D0A 1080 1440 5.65 3 110.0 10.15 249.9 196.0 234.2
D1 1080 1200 5.95 3 110.0 10.19 253.8 250.9 252.9
D2 1080 1560 5.95 3 110.0 10.19 253.8 250.9 253.0
D3 1080 1440 5.95 3 103.5 11.84 253.8 326.3 275.6
D4 1080 1440 5.95 3 118.5 7.98 253.8 284.2 261.4
D4A 1080 1440 5.65 3 118.5 8.08 249.9 274.4 256.3
D10 1080 1200 4.38 3 65 4.38 442.0 442.0 442.0
D11 1080 1200 4.38 3 90 4.38 442.0 442.0 442.0
D12 1080 1440 4.38 3 65 4.38 442.0 442.0 442.0

a
n = 0.3, E = 205.8 GPa
J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83
63
64 J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83

Table 2
Geometric characteristics of the Tanaka and Endo test structuresa

Structure no. b b a l Asx


t k bt

D0 58.54 1.97 33.0 0.39 0.49


D0A 63.72 2.22 32.2 0.32 0.55
D1 50.42 1.77 31.3 0.35 0.63
D2 65.55 2.30 33.0 0.37 0.48
D3 60.50 2.12 33.8 0.43 0.57
D4 60.50 2.12 31.4 0.37 0.44
D4A 63.72 2.22 31.0 0.36 0.47
D10 68.49 3.17 69.7 1.03 0.22
D11 68.49 3.17 46.1 0.68 0.30
D12 82.19 3.81 74.3 1.10 0.18

冪 E =column slenderness ratio


a sos
a
k=radius of gyration of longitudinals with associated plating, l =
kp

Table 3
Initial imperfections for plating and longitudinals for the Tanaka and Endo test structuresa

Structure no. Ao3 (mm) wosx wosy srcx (MPa) srcy (MPa) srsx (MPa) srsy (MPa)
(mm) (mm)

D0 0.101 – – 20.58 – – –
D0A 0.250 – – – – – –
D1 0.143 – – 25.97 – – –
D2 0.288 – – 18.62 – – –
D3 0.312 – – 23.03 – – –
D4 0.119 – – 34.00 – – –
D4A 0.379 – – – – – –
D10 0.515 – – 37.24 – – –
D11 0.503 – – 24.99 – – –
D12 0.523 – – 37.73 – – –

a
Ao3=buckling mode initial deflection for half-wave number m = 3, and the items marked by hyphen
were neither measured nor reported

and COV = 0.120 against experiments and mean = 0.714 and COV = 0.201
against FEA.

4.2.2. Effect of stiffener geometry under uniaxial compression


To investigate the effect of stiffener geometry on the panel ultimate strength under
uniaxial compression, nonlinear FE analyses using the computer program ANSYS
were carried out for longitudinally stiffened panels under uniaxial compressive loads
varying the stiffener web height. The ALPS/ULSAP method is then compared with
nonlinear FE solutions.
The dimensions and material properties of plating in the stiffened panels presented
Table 4
Comparison of the ALPS/ULSAP with the Tanaka and Endo experiment and FEAa

Structure no. Tanaka and Endo ALPS/ULSAP

sxu Mode sxu (sxu)FEA sxu (sxu)ULSAP (sxu)ULSAP Mode


soeq
冉 冊 Exp. soeq
冉 冊 FEA (sxu)Exp. soeq
冉 冊 ULSAP (sxu)Exp. (sxu)FEA

D0 0.931 IV/V 0.910 0.977 0.885 0.951 0.973 II


(0.914) (0.982) (1.004) IV/V
D0A 0.843 IV/V 0.867 1.028 0.839 0.995 0.968 IV/V
D1 1.095 IV/V 0.952 0.869 0.953 0.870 1.001 IV/V
D2 0.900 IV/V 0.842 0.936 0.799 0.888 0.949 IV/V
D3 1.032 IV/V 0.888 0.860 0.811 0.786 0.913 II
(0.869) (0.842) (0.979) IV/V
D4 0.990 IV/V 0.784 0.792 0.831 0.839 1.060 IV/V
D4A 0.875 IV/V 0.758 0.866 0.834 0.953 1.100 IV/V
D10 0.547 IV/V 0.631 1.154 0.586 1.071 0.929 IV/V
D11 0.527 IV/V 0.618 1.173 0.589 1.118 0.953 IV/V
D12 0.510 IV/V 0.571 1.120 0.540 1.059 0.946 IV/V
Mean 0.978 Mean 0.953 0.979
COV 0.139 COV 0.114 0.060
J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83

a
Mode IV/V indicates local buckling of stiffener web or tripping of stiffener, being identical because of flat bar profiles. Collapse mode of the
ALPS/ULSAP method, Mode II indicates plate-induced failure by yielding at the corners of plating between stiffeners. The values of () are given for
comparisons when the ALPS/ULSAP method predicted a collapse mode different from the experiment
65
66 J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83

Table 5
Model uncertainties of the DnV CN 30.1 method against the experiments and FEA for the Tanaka and
Endo test panels under uniaxial compressiona

Structure no. DnV CN 30.1

sxu (sxu)DnV (sxu)DnV Mode


( )
soeq DnV (sxu)Exp. (sxu)FEA

D0 0.685 0.736 0.753 A


D0A 0.679 0.805 0.783 B
D1 0.758 0.692 0.796 B
D2 0.665 0.739 0.790 A
D3 0.644 0.624 0.725 A
D4 0.675 0.682 0.861 A
D4A 0.663 0.758 0.875 A
D10 0.338 0.618 0.536 B
D11 0.337 0.639 0.545 B
D12 0.270 0.529 0.473 B
Mean 0.682 0.714
COV 0.120 0.201

a
Mode A=plate-induced failure, Mode B=stiffener-induced failure

Fig. 10. Correlation of ALPS/ULSAP predictions with experiments or FEA from Tanaka and Endo.

in these examples were selected to be typical of the bottom plating between bottom
longitudinal girders and transverse floors (frames) in large merchant vessel structures.
The values used are B = 3600 mm, a = 2640 mm, t = 21 mm, material yield stress
of plating sop = 352.8 MPa, elastic modulus E = 205.8 GPa and Poisson’s ratio
n = 0.3. The geometric properties of the stiffeners are varied to investigate the
characteristics of panel collapse pattern while the yield stress of stiffeners is taken
J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83 67

Fig. 11. (a) Correlation of ALPS/ULSAP and DnV CN 30.1 predictions with the Tanaka and Endo
experiments. (b) Correlation of ALPS/ULSAP and DnV CN 30.1 predictions with the Tanaka and
Endo FEA.

equal to that of the plating, i.e., sos = sop. The panel has three stiffeners in the
longitudinal direction. The thickness of the stiffener web is 12 mm.
In the FE computations, the following assumptions are made: The boundary con-
ditions for the stiffened panels are all simply supported along the four edges which
are kept straight. The cross section of stiffeners remains plane at panel edges as it
would in a stiffened panel supported by heavy transverse frames and longitudinal
girders. The column type initial deflection of stiffeners is taken as wosx = 0.1b2t,
while initial deflection for plating between stiffeners is assumed to be
wopl = 0.5wosx. A small value of initial local side way deflection of the stiffener web
is assumed so that the convergence is achieved more easily. The welding residual
stress is not considered.
Fig. 12 shows the variations of ultimate compressive strength for longitudinally
stiffened panels as obtained by ALPS/ULSAP. It is evident from Fig. 12 that the
height of stiffener web increases, the panel ultimate strength increases. When the
stiffeners are relatively small, the stiffeners buckle together with plating showing
Mode I failure. When the stiffeners become stiff, the plating between stiffeners
buckles while the stiffeners remain straight, and the ultimate strength is eventually
reached by the column type of collapse involving Mode III. If the height of stiffener
web exceeds a critical value, however, the ultimate strength tends to flatten out or
even decrease. This is because the stiffener web buckles locally or twists sideways
when the height of stiffener web is large.
Fig. 12 also shows the ultimate strength predictions from the DnV design guide-
line. For collapse mode identification of the DnV formulae, Table 5 is referred to.
It is noted that the design formulae from the classification societies normally consider
the stiffened panels that the stiffeners have been properly designed with sufficient
68 J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83

Fig. 12. (a) Ultimate strengths of longitudinally stiffened panels with three flat bars under uniaxial com-
pression by varying the ratio of stiffener web height to web thickness. (b) Ultimate strengths of longitudi-
nally stiffened panels with three angle bars under uniaxial compression by varying the ratio of stiffener
web height to web thickness. (c) Ultimate strengths of longitudinally stiffened panels with three T-bars
under uniaxial compression by varying the ratio of stiffener web height to web thickness.

strength. In that sense, the comparison of the classcification society design guidelines
on the stiffened panels with small stiffeners will not be in fact meaningful.

4.2.3. Effect of stiffener geometry under combined axial compression and lateral
pressure
Some preliminary studies of the effects of lateral pressure and welding induced
residual stresses on the ultimate strength characteristics of longitudinally stiffened
J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83 69

panels are now presented. The lateral pressure at water head of 20 m is considered
to be applied on the same direction as the column type initial deflection of stiffeners.
Fig. 13(a–c) show the variations of the ultimate compressive strength in terms of
the ratios of stiffener web height to web thickness, lateral pressure and welding
residual stresses for the longitudinally stiffened panels with three flat bars, angle
bars or tee-type stiffeners, respectively. When the stiffener web height is small, the
stiffened panels collapse in Modes I or I(Mean), while if the stiffener web height is
larger than a certain value, then the plating between stiffeners buckles locally
between stiffeners and the panel reaches the ultimate strength by beam-column type
collapse initiated by plate-induced failure. As the stiffener web height increases

Fig. 13. (a) Effect of lateral pressure and residual stress on ultimate strength of a longitudinally stiffened
panel with three flat bars. (b) Effect of lateral pressure and residual stress on ultimate strength of a
longitudinally stiffened panel with three angle bars. (c) Effect of lateral pressure and residual stress on
ultimate strength of a longitudinally stiffened panel with three tee-bars.
70 J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83

further, the ultimate strength is reached by local buckling of the stiffener web or
tripping of stiffener.
As apparent, lateral pressure or residual stresses can significantly affect (reduce)
the panel ultimate compressive strength in some cases. The DnV CN 30.1 predictions
are generally very pessimistic. This is because the DnV CN 30.1 method determines
the panel ultimate strengths as the minimum value of the two results from the Perry–
Robertson formula based on either the stiffener-induced failure mode or the plate-
induced failure mode, while in most cases the minimum value of the ultimate strength
is due to the stiffener-induced failure mode. Experimental confirmation by model
tests is in this regard lacking, and further investigations are needed regarding the
degree of effect noted.

4.3. Ultimate strength characteristics of grillages

The ultimate strength characteristics of grillages, i.e., with flat-bar, angle or tee-
type stiffeners in both longitudinal and transverse directions, are studied by compar-
ing the ALPS/ULSAP method with experimental and numerical results as well as
theoretical solutions from the DnV classification society design guideline.

4.3.1. The Smith mechanical collapse tests


Smith [5] carried out a series of collapse tests using a total of 11 full scale welded
steel grillages representing typical warship deck structures under axial compression,
or bottom structures under combined axial compression and lateral pressure. The test
structures include four pairs of nominally identical grillages (Nos 1a, 1b, 2a, 2b, 3a,
3b, 4a, 4b) representing a ship bottom configuration, together with two grillages (Nos
5, 7) representing frigate strength decks and one grillage (No. 6) corresponding a
light superstructure deck. The typical grillage model that Smith tested is shown in
Fig. 14. Smith et al. [10] later computed the ultimate strengths of the Smith test
grillages using the nonlinear finite element method as well.

Fig. 14. The Smith test ‘grillage’ with support members in both directions.
J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83 71

The overall dimensions of each grillage were L = 6096 mm long by B = 3048 mm


wide, excluding the panel ends which are bolted to the test frames along the edges.
Except for Nos 4a and 4b which have both large girders and small stiffeners in the
longitudinal direction, all test grillages have identical T-type longitudinal stiffeners
and identical tee-type transverse frames. Table 6 indicates the geometric properties
of longitudinals and transverses, respectively, where Nos 4a and 4b are represented
by the longitudinally stiffened panel between two adjacent longitudinal girders and
two adjacent transverse frames. The material yield stresses of plating and stiffeners
are indicated in Table 6. Table 7 presents the important ultimate strength related
geometric characteristics for each grillage.
The initial deflections of plating, longitudinals or transverses were measured in
these tests. There was reportedly a high degree of variability associated with the
plate initial deflection measurements, where the COVs of wopl and wosx were in the
range of 0.22–0.63 and 0.29–1.04, respectively. Specifically, it is reportedly observed
that plating and stiffener imperfections for model No. 3b were abnormally large,
with an unfavourable relative stiffener distortion as well. Also, No. 6 representing
a light superstructure deck had serious level of initial imperfections which would be
untypical in a real structure. The welding induced residual stresses of plating were
also measured for selected grillages. The COV of srcx was in the range of 0.12–
0.52. The residual stresses of longitudinals or transverses were neither measured nor
reported. Table 8 summarizes the initial imperfections of plating and stiffeners of
each grillage on the basis of the measurements and insights provided in Smith [5]
and Smith et al. [10]. Based on the measured initial deflection patterns of plating,
Table 8 represents the buckling mode initial deflection component of each grillage
also. It is noted that in the FEA of Smith et al. [10], two types of computations were
tried with different level of initial deflections, namely FEA-1 and FEA-2, the former
being with average initial imperfections and the latter being with actual initial imper-
fections.
The ALPS/ULSAP method treats the entire test grillage as a unit, i.e., that includes
both longitudinals and transverses. The Smith FEA models the test grillage as a
representative two-bay plate–stiffener combination. The DnV CN 30.1 method takes
longitudinally stiffened panel between two adjacent transverse frames as the extent
of calculation.
Tables 9–11 compare the present design equation predictions as proposed, with
DnVCN 30.1 and the actual Smith mechanical test results. Fig. 15 shows the corre-
lation of the theoretical solutions and experimental results. The collapse modes pre-
dicted by ALPS/ULSAP method are also indicated in Table 10. It is seen that most
test models collapsed by tripping of longitudinals (Mode V) as obtained by the
experiments. It is also noted that the ALPS/ULSAP predictions compare reasonably
well with the other data in most cases. Fig. 16 shows the correlations of
ALPS/ULSAP, DnV CN 30.1, experiment and FEA.

4.3.2. Effect of combined axial compression and lateral pressure


Fig. 17 compares the design methods with the Smith collapse test results or SPINE
solutions for the grillages under combined axial compression and lateral pressure,
72

Table 6
Mean values of geometric properties and material yield stresses for the Smith test grillagesa

Grillage L (mm) B (mm) t nsx hwx (mm) twx bfx (mm) tfx nsy hwy (mm) twy bfy (mm) tfy sop sos soeq
no. (mm) (mm) (mm) (mm) (mm) (mm) (MPa) (MPa) (MPa)

1a 6096 3048.0 8.00 4 153.67 7.21 78.99 14.22 4 257.56 9.37 125.48 18.29 249.1 253.7 250.4
1b 6096 3048.0 7.87 4 152.40 7.11 76.20 14.22 4 254.00 9.14 127.00 18.29 252.2 252.4 252.3
2a 6096 3048.0 7.72 9 115.57 5.44 45.97 9.53 3 204.98 8.31 102.62 16.26 261.3 268.9 263.1
2b 6096 3048.0 7.37 9 114.30 5.38 44.70 9.53 3 203.71 8.33 102.62 16.26 259.7 274.9 263.3
3a 6096 3048.0 6.38 9 77.72 4.52 25.91 6.35 3 156.21 6.81 78.99 14.22 250.6 227.9 246.8
3b 6096 3048.0 6.40 9 77.22 4.65 27.94 6.35 3 153.92 6.88 79.25 14.22 252.2 223.3 247.3
4a 1219.2 1016.0 6.43 3 76.71 4.85 27.69 6.35 – – – – – 259.7 223.9 252.2
4b 1219.2 1016.0 6.40 3 76.96 4.55 26.16 6.35 – – – – – 264.3 227.9 257.3
5 6096 3048.0 6.43 4 116.08 5.33 46.23 9.53 3 154.18 6.76 77.22 14.22 247.6 230.9 244.9
6 6096 3048.0 6.32 4 76.20 4.55 27.43 6.35 4 114.55 5.36 46.23 9.53 256.7 241.5 255.2
7 6096 3048.0 6.30 4 115.06 5.16 45.21 9.53 3 153.92 6.65 78.74 14.22 290.1 305.3 303.3

a
Grillage Nos 4a and 4b represent longitudinally stiffened panels between two adjacent longitudinal girders and two adjacent transverse frames,
J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83

n = 0.3, E = 205.8 GPa


J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83 73

Table 7
Other geometric characteristics of the Smith test grillages

Grillage no. b b a l Asx


t k bt

1a 76.2 2.67 21 0.24 0.42


1b 77.4 2.72 21 0.23 0.43
2a 39.5 1.42 36.5 0.42 0.40
2b 41.4 1.48 36 0.42 0.42
3a 47.8 1.68 66 0.70 0.24
3b 47.6 1.68 66 0.70 0.24
4a 39.5 1.41 50 0.54 0.28
4b 39.7 1.43 50 0.53 0.28
5 94.9 3.31 42 0.45 0.24
6 96.4 3.42 68 0.75 0.12
7 96.8 3.65 42 0.52 0.24

as studied above. For collapse mode identifications of the DnV CN 30.1 method,
Table 5 is referred to. The upper limit of the ‘critical’ lateral pressure is also shown
in Fig. 17. The real panel ultimate lateral pressure should then be smaller than the
critical lateral pressure.
While the experiments and SPINE analyses were undertaken for the models of
the entire grillages, i.e., including both longitudinals and transverse frames, two kinds
of modeling with regard to the extent of calculation were considered for the
ALPS/ULSAP ultimate strength predictions, namely one for an entire grillage and
the other for a longitudinally stiffened panel between two adjacent transverse frames.
The real ultimate strength must be taken as the smaller value of the two results of
ALPS/ULSAP models.
It is evident that the latter ALPS/ULSAP calculation model significantly overesti-
mates the ultimate strength compared to the former, particularly when the magnitude
of lateral pressure loads is large. The ultimate strength characteristics of ship grillages
depend on the dimensions of transverse frames as well as those of longitudinals.
When a larger magnitude of lateral pressure loads is applied, the relatively weak
transverse frames in the grillage can fail and do not continue to support the related
panel, and therefore the latter type of modeling, i.e., only for stiffened panel between
two adjacent transverse frames which are assumed to remain straight or not to fail,
may provide quite optimistic ultimate strength predictions of the grillage.
The grillages fail in the overall collapse mode I as the lateral pressure loads are
dominant. The ALPS/ULSAP method with the former type of modeling predicts the
ultimate strength reasonably well for the entire range of lateral pressure considered
in the Smith collapse test data used. As previously noted, the DnV CN 30.1 method
is available only for longitudinally stiffened panels between two transverse frames
which are considered not to fail until the panel reaches the ultimate strength. In fact,
the DnV CN 30.1 method is designed for longitudinally stiffened panels so that it
should not be used when the grillages fail together with transverse frames.
74

Table 8
Initial imperfections of plating, longitudinals and transverses for the Smith test grillagesa

Grillage no. wopl wosx wosy srcx srcy srsx srsy Aom
b a wosx sop srcx sos srsx wopl

1a 0.0060 0.0007 0.7 – – – – 0.1


1b 0.0077 0.0011 – – – – – 0.1
2a 0.0044 0.0025 – 0.48 0.10 – – 0.1
2b 0.0060 0.0010 – 0.33 0.10 – – 1.0
3a 0.0093 0.0028 0.2 0.38 0.10 – – 0.7
3b 0.0150 0.0019 –0.8 0.43 0.10 – – 1.0
4a 0.0081 0.0023 0.5 0.38 0.10 – – 0.8
4b 0.0063 0.0008 0.5 0.41 0.10 – – 0.7
5 0.0100 0.0008 –0.4 0.16 0.10 – – 0.1
6 0.0125 0.0020 0.4 0.31 0.10 – – 1.0
7 0.0094 0.0007 – 0.08 0.10 – – 0.1

a
Source—Smith [5] and Table 3 of Smith et al., Aom=buckling mode initial deflection
J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83
Table 9
Comparison of the Smith FEA with the experiment for ultimate strength of grillagesa

Grillage no. p (MPa) sxu sxu sxu (sxu)FEA⫺1 (sxu)FEA⫺2


soeq
冉 冊 Exp. soeq
冉 冊 FEA⫺1 soeq
冉 冊 FEA⫺2 (sxu)Exp. (sxu)Exp.

1a 0 0.76 0.65 0.69 0.855 0.908


1b 0.103 (15 psi) 0.73 0.57 0.57 0.781 0.781
2a 0.048 (7 psi) 0.91 0.81 0.81 0.890 0.890
2b 0 0.83 0.82 0.82 0.988 0.988
3a 0.021 (3 psi) 0.69 0.69 0.63 1.000 0.913
3b 0 0.61 0.71 0.60 1.164 0.984
4a 0 0.82 0.80 0.75 0.976 0.915
4b 0.055 (8 psi) 0.83 0.73 0.76 0.880 0.916
5 0 0.72 0.51 0.55 0.708 0.764
6 0 0.49 – – – –
7 0 0.65 0.49 0.53 0.754 0.815
Mean 0.900 0.887
COV 0.152 0.087
J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83

a
FEA-1=with average imperfections, FEA-2=with actual imperfections
75
76

Table 10
Comparison of ALPS/ULSAP with the Smith experiments and FEA for ultimate strength of grillagesa

Grillage no. p (MPa) sxu (sxu)ULSAP (sxu)ULSAP (sxu)ULSAP Collapse modes


soeq
冉 冊 ULSAP (sxu)Exp. (sxu)FEA⫺1 (sxu)FEA⫺2
Exp. ULSAP

1a 0 0.76 1.000 1.169 1.101 V V


1b 0.103 0.62 0.849 1.088 1.088 V V
2a 0.048 0.79 0.868 0.975 0.975 III+V I
(0.86) (0.945) (1.062) (1.062) III
(0.90) (0.989) (1.111) (1.111) V
2b 0 0.79 0.952 0.963 0.963 III+V V
3a 0.021 0.69 1.000 1.000 1.095 III+V V
3b 0 0.58 0.951 0.817 0.967 III+V V
4a 0 0.80 0.976 1.000 1.067 III+V V
4b 0.055 0.81 0.976 1.110 1.066 III+V V
5 0 0.52 0.722 1.020 0.945 III+V V
6 0 0.37 0.755 – – I+V V
7 0 0.52 0.800 1.061 0.981 III+V V
Mean 0.895 1.020 1.025
COV 0.113 0.095 0.062
J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83

a
(1) I+V or III+V indicate that the grillage collapsed in Mode I or III together with Mode V. (2) The values of () are given for comparisons when
the ALPS/ULSAP method predicted a collapse mode different from the experiment
Table 11
Model uncertainties of the DnV CN 30.1 method against either the Smith experiments or the Smith FEAa

Grillage no. p (MPa) sxu (sxu)DnV (sxu)DnV (sxu)DnV Mode


soeq
冉 冊 DnV (sxu)Exp. (sxu)FEA⫺1 (sxu)FEA⫺2

1a 0 0.61 0.803 0.938 0.884 A


1b 0.103 0.57 0.781 1.000 1.000 A
2a 0.048 0.79 0.868 0.975 0.975 B
2b 0 0.79 0.952 0.963 0.963 B
3a 0.021 0.54 0.783 0.783 0.857 B
3b 0 0.64 1.049 0.901 1.067 B
4a 0 0.71 0.866 0.888 0.947 B
4b 0.055 0.52 0.627 0.712 0.684 B
5 0 0.50 0.694 0.980 0.909 A
6 0 0.41 0.837 – – B
7 0 0.44 0.677 0.898 0.830 A
Mean 0.812 0.904 0.912
COV 0.151 0.102 0.117
J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83

a
(1) Collapse modes of the DnV CN 30.1 method: Mode A=plate-induced failure, Mode B=stiffener-induced failure, (2) FEA-1=with average imperfec-
tions, FEA-2=with actual imperfections
77
78 J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83

Fig. 15. Correlation of ALPS/ULSAP predictions with the experiments or FEA for the Smith test gril-
lages.

5. Concluding remarks

The aim of the present study has been to develop an advanced, yet design-oriented
method for predicting the ultimate strength of steel stiffened panels under combined
axial load, in-plane bending and lateral pressure. The collapse patterns of a stiffened
panel are classified into six groups, namely overall grillage collapse, yielding at the
corners of plating between stiffeners, yielding of the plate–stiffener combination at
mid-span, local buckling of stiffener web, lateral–torsional buckling (tripping) of
stiffener and gross yielding. The real ultimate strength is then calculated as the lowest
value among the various ultimate strengths calculated for each of the collapse pat-
terns. The validity of the developed method is confirmed by comparing with experi-
mental, numerical and theoretical results.
Some important insights developed in this paper are as follows:

1. Most existing ultimate strength formulations of steel stiffened panels including


structural design guidelines from classification societies are typically designed for
the panels with relatively strong (longitudinal) stiffeners which do not fail prior
to plating between stiffeners. When the stiffeners are weak, these formulae may
not be meaningful since their ultimate strength predictions are too pessimistic;
they are even smaller than the ultimate strength of the bare panel, i.e., without
stiffeners. In this regard, the structural designer should be careful in use of such
design formulae when the stiffeners are relatively weak. It would be necessary to
clearly define the applicable range of the design formulations in terms of the
stiffener dimensions or strengths, while the ALPS/ULSAP method presented in
this paper extends over all dimensions of the stiffeners.
2. The so-called Perry–Robertson formula for a representative plate–stiffener combi-
J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83 79

Fig. 16. (a) Correlation of ALPS/ULSAP and DnV CN 30.1 predictions with the experiments for the
Smith test grillages. (b) Correlation of ALPS/ULSAP and DnV CN 30.1 predictions with the FEA-1
solutions for the Smith test grillages. (c) Correlation of ALPS/ULSAP and DnV CN 30.1 predictions
with the FEA-2 solutions for the Smith test grillages.

nation has typically been used in most existing ultimate strength formulations of
stiffened panels. In the course of the present study, it is realized that the stiffener-
induced failure mode based Perry–Robertson formula generally provides too pessi-
mistic results, while the plate-induced failure mode based Perry–Robertson for-
mula reasonably predicts the panel ultimate strengths in a specific range of stiff-
ener dimensions which follows the beam-column type collapse mode (i.e., Mode
80 J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83

Fig. 17. (a) Comparison of ALPS/ULSAP and DnV CN 30.1 predictions with the Smith test grillage
No. 1b under combined axial compression and lateral pressure. (b) Comparison of ALPS/ULSAP and
DnV CN 30.1 predictions with the Smith test grillage No. 2a under combined axial compression and
lateral pressure. (c) Comparison of ALPS/ULSAP and DnV CN 30.1 predictions with the Smith test
grillage No. 3a under combined axial compression and lateral pressure.

III). In this regard, the ALPS/ULSAP method excludes the stiffener-induced fail-
ure mode in using the Perry–Robertson formula for Mode III.
3. In FE computations for the ultimate strength of plating (between stiffeners) under
axial compression, keeping the unloaded plate edges straight is more relevant for
thick or medium thickness plating, while it is more appropriate for thin plating
so that the unloaded plate edges move freely in plane. When considering that
most actual steel plating of marine structures is relatively thick or in medium
thickness, i.e., with the plate reduced slenderness ratio in the range of 1.0–3.0, it
would be suggested for practical design purposes so that the unloaded plate edges
J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83 81

are kept straight, as long as the support members (stiffeners) do not fail until the
plating collapses.
4. The ultimate strength characteristics of steel grillages depend on the dimensions
of transverse frames as well as those of longitudinals. When transverse frames
are strong enough so that they do not fail prior to longitudinally stiffened panels
between transverse frames, the extent of analysis can be taken only for the longi-
tudinally stiffened panel. When a larger magnitude of lateral pressure loads is
applied, however, the relatively weak transverse frames in the grillage can fail
and do not continue to support the related panel, and this may result in quite
optimistic ultimate strength predictions of the grillage if the longitudinally stiff-
ened panel between transverse frames was taken as the extent of analysis. There-
fore, it would be best to treat an entire steel grillage, i.e., with both longitudinal
stiffeners and transverse frames as a unit by accounting for their interaction. This
may be very important for bottom or lower side shell stiffened grillages of large
merchant vessels.

With a modern powerful desk-top computer, the tendency to focus on development


of structural design methodology has slightly transformed from ‘simplification’ to
‘sophistication’. The ALPS/ULSAP method has been designed to be more sophisti-
cated than previous simplified theoretical methods, and requires much less compu-
tation than conventional nonlinear iterative finite element methods. For many years,
structural researchers have been working towards the goal of reliability-based design
of structures. Also, these calculations must be performed a large number of times.
Therefore, it is not practical to use iterative nonlinear finite element analyses for
these calculations. In contrast, the ALPS/ULSAP method is based on advanced, yet
design-oriented expressions and is therefore computationally very efficient.

Acknowledgements

The present study was undertaken under financial support from the Research Insti-
tute of Marine Systems Engineering of the Seoul National University, Seoul, Korea
(Director: Prof. Hang S. Choi) who is thanked for this support. The authors are
grateful to Dr. A.K. Thayamballi of Chevron Shipping and Dr. E. Steen of Det
Norske Veritas for their valuable comments.

References

[1] Das PK, Garside JE. Structural redundancy for continuous and discrete systems. Ship Structure
Committee, report no. SSC-354, 1991.
[2] Hughes OF, Nikolaidis E, Ayyub B, White G, Hess PE. Uncertainty in strength models for marine
structures. Ship Structure Committee, report no. SSC-375, 1994.
[3] Rigo P, Moan T, Frieze PA, Chryssanthopoulos M. Benchmarking of ultimate strength prediction
for longitudinally stiffened panels. Proceedings of the Sixth International Symposium on Practical
Design of Ship and Mobile Units (PRADS’95), Seoul, September, 1995;2:869–82.
82 J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83

[4] Paik JK, Kim DH. A benchmark study of the ultimate compressive strength formulations for stiffened
panels. J Res Inst Ind Technol Pusan Nat Uni 1997;53:373–405.
[5] Smith CS. Compressive strength of welded steel ship grillages. Trans Roy Inst Naval Archit
1975;117:325–59.
[6] Tanaka Y, Endo H. Ultimate strength of stiffened plates with their stiffeners locally buckled in
compression. J Soc Naval Archit Jpn 1988;164:456–67.
[7] Hu SZ, Chen Q, Pegg N, Zimmerman TJE. Ultimate collapse tests of stiffened plate ship structural
units. Marine Struct 1997;10:587–610.
[8] Hopperstad OS, Langseth M, Tryland T. Ultimate strength of aluminum alloy outstands in com-
pression: experiments and simplified analysis. Thin-Wall Struct 1999;34(4):279–94.
[9] Smith CS, Davidson PC, Chapman JC, Dowling PJ. Strength and stiffness of ships’ plating under
in-plane compression and tension. Trans Roy Inst Naval Archit 1988;130:277–96.
[10] Smith CS, Anderson N, Chapman JC, Davidson PC, Dowling PJ. Strength of stiffened plating under
combined compression and lateral pressure. Trans Roy Inst Naval Archit 1992;134:131–48.
[11] Davidson PC, Chapman JC, Smith CS, Dowling PJ. The design of plate panels subject to in-plane
shear and biaxial compression. Trans Roy Inst Naval Archit 1990;132:267–86.
[12] Davidson PC, Chapman JC, Smith CS, Dowling PJ. The design of plate panels subject to biaxial
compression and lateral pressure. Trans Roy Inst Naval Archit 1992;134:149–54.
[13] Carlsen CA. Simplified collapse analysis of stiffened plates. Norweg Marit Res 1977;4:20–36.
[14] Carlsen CA. A parametric study of collapse of stiffened plates in compression. Struct Eng
1981;58B(2):33–40.
[15] Hughes OF. Ship structural design, a rationally-based, computer-aided optimization approach. New
Jersey: The Society of Naval Architects and Marine Engineers, 1988.
[16] Mikami I, Niwa K. Ultimate compressive strength of orthogonally stiffened steel plates. J Struct
Engng 1996;122(6):674–82.
[17] Schafer BW, Pekoz T. The behavior and design of longitudinally stiffened thin-walled compression
elements. Thin-Wall Struct 1997;27(1):65–78.
[18] Grondin GY, Chen Q, Elwi AE, Cheng JJ. Stiffened steel plates under compression and bending. J
Construct Steel Res 1998;45(2):125–48.
[19] Paik JK, Thayamballi AK, Kim DH. An analytical method for the ultimate compressive strength
and effective plating of stiffened panels. J Construct Steel Res 1999;49:43–68.
[20] Paik JK, Thayamballi AK, Park YI. Local buckling of stiffeners in ship plating. J Ship Res
1998;42(1):56–67.
[21] Bleich F. Buckling strength of metal structures. New York: McGraw-Hill, 1952.
[22] Faulkner D, Adamchak JC, Snyder GJ, Vetter MR. Synthesis of welded grillages to withstand com-
pression and normal loads. Comput Struct 1973;3:221–46.
[23] Faulkner D. A review of effective plating for use in the analysis of stiffened plating in bending and
compression. J Ship Res 1975;19(1):1–17.
[24] Faulkner D. Compression strength welded grillages. In: Evans JH, editor. Ship structural design
concepts, Centreville, MD: Cornell Maritime Press; 1975.
[25] Faulkner D. Toward a better understanding of compression induced tripping. In: Narayanan R, editor.
Steel structures, London: Elsevier; 1987. pp. 159–75.
[26] Adamchak JC. Design equations for tripping of stiffeners under inplane and lateral loads.
DTNSRDC-79/064, Washington DC, October 1979.
[27] Danielson DA, Kihl DP, Hodges DH. Tripping of thin-walled plating stiffeners in axial compression.
Thin-Wall Struct 1990;10(2):121–42.
[28] Danielson DA. Analytical tripping loads for stiffened plates. Int J Solids Struct 1995;32(8–
9):1317–28.
[29] Louca LA, Harding JE. Torsional buckling of outstands in longitudinally stiffened panels. Thin-
Wall Struct 1996;24(3):211–29.
[30] Kesti J, Davies JM. Local and distortional buckling of thin-walled short columns. Thin-Wall Struct
1999;34(2):115–34.
[31] Hughes OF, Ma M. Elastic tripping analysis of asymmetrical stiffeners. Comput Struct
1996;60(3):369–89.
J.K. Paik, B.J. Kim / Thin-Walled Structures 40 (2002) 45–83 83

[32] Hughes OF, Ma M. Inelastic analysis of panel collapse by stiffener buckling. Comput Struct
1996;61(1):107–17.
[33] Hu Y, Chen B, Sun J. Tripping of thin-walled stiffeners in the axially compressed stiffened panel
with lateral pressure. Thin-Wall Struct 2000;37(1):1–26.
[34] Paik JK, Thayamballi AK, Kim BJ, Wang G, Shin YS, Liu D. Ultimate limit state design of ship
stiffened panels and grillages. Presented at the 2001 SNAME annual meeting (to appear in Trans-
actions of the Society of Naval Architects and Marine Engineers, Vol 109), Orlando, October 2001.
[35] Paik JK, Thayamballi AK, Kim BJ. Large deflection orthotropic plate approach to develop utimate
strength equations for stiffened panels under combined biaxial compression/tension and lateral press-
ure. Thin-Wall Struct 2001;39(3):215–46.
[36] Paik JK, Thayamballi AK, Kim BJ. Advanced ultimate strength formulations for ship plating under
combined biaxial compression/tension, edge shear and lateral pressure loads. Marine Technol
2001;38(1):9–25.
[37] Wood RH. Plastic and elastic design of slabs and plates. New York: The Ronald Press, 1961.
[38] Trahair NS, Bradford MA. The behavior and design of steel structures. London: Chapman and
Hall, 1988.
[39] Paik JK, Thayamballi AK, Lee SK, Kang SJ. A semi-analytical method for the elastic-plastic large
deflection analysis of welded steel or aluminum plating under combined in-plane and lateral pressure
loads. Thin-Wall Struct 2001;39(2):125–52.
[40] DnV. Buckling strength analysis. Classification notes no. 30.1, Det Norske Veritas, Oslo, July 1995.
[41] Ellinas CP, Supple WJ, Walker AC. Buckling of offshore structures. A state-of-the-art review. Hous-
ton, TX: Gulf Publishing Co, 1984.
[42] Yamamoto Y, Matsubara N, Murakami T. Buckling strength of rectangular plates subjected to edge
thrusts and lateral pressure (2nd report). J Soc Naval Archit Jpn 1970;127:171–9.

You might also like