Professional Documents
Culture Documents
net/publication/258186891
CITATIONS READS
2 39
3 authors, including:
Some of the authors of this publication are also working on these related projects:
configuration of mooring chain resting on surfaces with single or anticlastic curvatures View project
All content following this page was uploaded by Roger Hobbs on 11 March 2016.
The manuscript was received on 6 March 2005 and was accepted after revision for publication on 4 October 2005.
DOI: 10.1243/030932405X30948
Abstract: Using normality theory and a simplified beam model, a theoretical method is built
up to evaluate the relationship between the central deflection of a corrugated panel and
external uniformly distributed pressure load. The interaction between the bending moment
and the membrane force in the panel is included, as well as the influence of connections. This
theoretical model is calibrated against finite element analyses. The results show encouraging
agreement.
Downloaded from sdj.sagepub.com at Imperial College London Library on March 11, 2016
136 Y H Liang, L A Louca, and R E Hobbs
2.2 Material models can be found. From the condition that, in pure
plastic bending, the area below the neutral axis
In general, the stress–strain material model is based
must equal that above the axis, the plastic neutral
on the properties of the carbon steel grade 420, as
axis position (x h ) can be found as
given by BS 4360 [8]. Both nominal and true plastic p p
variables are given in Table 1, in which the ‘true’
A B
1 (2c −a ) sin h t t
properties are used in FEA. x = − p p p ∏xh∏h− (1)
p 2 2h 2 2
In this paper, the rigid-plastic material model is p
used in the theoretical method, which means that
The cross-sectional area A is
the stress does not change with increasing strain p
after reaching the yield stress. On the other hand, in h −t
the parallel FEA, two elastic–plastic material models, A =(a +2c )t +2 p pt (2)
p p p p sin h p
one including strain hardening effects and the other p
not, are used in order to check whether applying while the plastic modulus Z can be obtained as
the rigid-plastic material model in the theoretical pp
analysis is appropriate. Z =2x c t h +a t (1−x )h
pp p p p p p p p p
(x h −t /2)2 [(1−x )h −t /2]2
+ p p p t + p p p t
3 STATIC ANALYSIS OF THE CORRUGATED sin h p sin h p
p p
PANEL (3)
Downloaded from sdj.sagepub.com at Imperial College London Library on March 11, 2016
Static plastic analysis of corrugated steel panels 137
while the ratio of the reduced plastic moment to the leads to the following relationship between n and
p
plastic moment for zero axial force is d , the central deflection of the simplified corrugated
p
beam
m =1+ f n2 +gn (5)
p p p
where de qw /qN Z
d = p =2 p p =−2 pp (2 f n +g) (7)
p dh qw /qM A p
A sin h 2e2t p p p p
e= p p, f =− p
4t Z sin h
p pp p 3.3 Geometric analysis of the corrugated beam
[(a −2c )+2(1−2x )h /sin h ]et
g= p p p p p p In order to evaluate the relationship between the
Z
pp external distributed load and the central deflection
Turning to the plastic collapse mechanism for the of the corrugated beam, the deformed shape of
beam (Fig. 3), since there are two plastic rotations the beam needs to be considered (Fig. 3). From the
(one at A and one at C) corresponding to the plastic equilibrium condition, it is easy to find
bending moment (only half of the span is studied by
A B
16Z s A
considering the symmetry condition), applying the q= pp 0 m + p n d (8)
normality theory ([11] and Appendix 3) gives L2 p 2Z p p
p pp
qN de =qM 2dh (6)
p p p p 3.4 The governing equation for the corrugated
where [M , N ] is the loading vector and (h , e ) is
p p p p beam
the deformation vector corresponding to the loading
vector. Defining w=m − f n2 −gn −1, equation (6) Incorporating the results of sections 3.2 and 3.3
p p p above, the governing equation for the corrugated
beam throughout the entire analysis is
A B
16Z s A
q= pp 0 m + p n d
L2 p 2Z p p
p pp
Summarizing the results for all four cases (with the
derivations for cases 2 to 4 being presented in
Appendix 2), the relationship between m , n , and
p p
d is
p
t
For 0∏y ∏x h − p
p p p 2
4y t
n = p p
p A sin h
p p
m =1+ f n2 +gn
p p p
Z
d =−2 pp (2 f n +g)
Fig. 3 Deformed shape of the corrugated beam p A p
p
Downloaded from sdj.sagepub.com at Imperial College London Library on March 11, 2016
138 Y H Liang, L A Louca, and R E Hobbs
A B
t (1−2x )h t t been carried out using Abaqus/Standard [14]. As
For x h − p ∏y ∏ p p p+ x h − p
p p 2 p sin h c p p 2 mentioned previously, two series of finite element
p p analyses, with different material models, have been
y −j performed. In each of these series, three different
n = p
p i profiles are examined (Fig. 4). For each profile, three
cases (a single corrugated strip case which is marked
m =sn2 +un +v
p p p as ‘single’; a case with three contiguous corrugated
2 strips which is marked as ‘triple’; and a whole panel
d =− (2sn +u)
p A p case which is marked as ‘whole’) have been studied
p
so that the differences between the simplified
A B
(1−2x )h t t t
For p p p + x h − p ∏y ∏x h + p model and the whole panel can be found. Detailed
sin h c p p 2 p p p 2 information on the FEA models is provided in
p p
y −w Table 2.
n = p The simple elastic–plastic model without the strain
p i
hardening effect is applied in the first series, while
m =P n2 +P n +P strain hardening is included in the second series. The
p 1 p 2 p 3
purpose of using two material models, between which
2
d =− (2P n +P ) the only difference is the strain hardening effect, was
p A 1 p 2 to identify the effects of strain hardening, both within
p
t the finite element (FE) model and between FEA and
For x h + p ∏y the theoretical method.
p p 2 p
All of the finite element modelling used the
n =1 Abaqus/Standard four-node doubly curved general-
p
m =0 purpose shell element with reduced integration, S4R.
p The overall mesh size is 0.03 m, which is fine com-
where any new parameters are defined in the pared with the profile size so that the mesh sensitivity
detailed analysis in Appendix 2.
Table 2 Details of the three corrugated panels
3.5 The finite element analysis (FEA)
Profile 1 Profile 2 Profile 3
Previous research has shown that similar problems
can be modelled by FEA, with sufficient accuracy Panel width, B (m) 4 3.64 6.4
p
[12, 13]. Therefore, in order to validate the newly Panel span, L (m) 4 4 4
p
Span/cross-section depth 20 100 5
developed theoretical method, extensive FEA has
Downloaded from sdj.sagepub.com at Imperial College London Library on March 11, 2016
Static plastic analysis of corrugated steel panels 139
can be neglected. In view of the symmetry, it was area close to the built-in end. Shown in Fig. 6, the
only necessary to model half the structure in all the local plastic strain in the first series model in that
cases analysed. Therefore, each FE model has a fixed- area is extremely large so that the area is actually
ended boundary condition at one end and symmetric necking and being torn open. This means that the
boundary conditions on the other three sides, except big difference between the two series is mainly due
in the whole panel case, which has one symmetric to the very large local plastic strain occurring in the
boundary condition at mid-span and fixed-ended area close to the built-in end, while the global strain
boundary conditions along the other three sides. hardening has not developed elsewhere and is not
Considering there would be a large strain in the FE globally significant in this analysis. Further analyses
results, non-linear analysis was included in all the conducted on the same structures but with pin-
FE models and the Riks method used. ended supports show that, if no large local plastic
strain (necking) exists at both ends, results of these
two series are very similar. This confirms that the
3.6 Results and comparisons actual function of the strain hardening effect in this
Both theoretical and FEA results are given below. In analysis is to reduce the very large local plastic strain
order to illustrate those results, comparisons are also and control necking at the supports. On the other
provided. hand, in fact, the current theoretical model can
achieve an equivalent effect of those FE models with
3.6.1 Results for profile 1 strain hardening, since necking is prevented by the
classic plastic model which assumes dimensionless
The relationships between the central deflection plastic hinges.
and the external distributed load are plotted in Thus it is concluded that the FEA with strain
Fig. 5. As can be seen in Fig. 5, the curves for each hardening is a more realistic model, producing
series of cases (single, triple, and whole panel) are better results for use as a benchmark for the new
generally very close. However, comparing the FE theoretical model. Because that model is based on
results between the two series, there is a significant the cross-section, it does not involve any discussions
difference at larger deflections, where the results of localized strains, and it is accordingly not adversely
including strain hardening are much stiffer than affected by, for example, the large hinge rotations at
those excluding it. However, explaining this difference the supports. (Of course, it is recognized that strain
by a global strain hardening effect alone is an over- hardening would increase the load at large deflections
simplification. Choosing the ‘single’ case and looking and rotations, but not to the extent illustrated by the
into the detail in each result, it is found that the differences between the two types of FEA.)
plastic strain is less than 4 per cent in most of the Turning, then, to the comparison between the
area, while large local plastic strains exist in a small theoretical results and the FEA with strain hardening,
Downloaded from sdj.sagepub.com at Imperial College London Library on March 11, 2016
140 Y H Liang, L A Louca, and R E Hobbs
it can be seen that these agree well with each other realistic. The theoretical and FEA results also agree
except at low deflections, where the neglect of the with each other in this case, which proves that the
elastic effect in the theoretical model is obvious. theoretical method is valid. Looking into more detail,
compared with Fig. 5 a significant difference can be
found at low deflections in Fig. 7, which indicates
3.6.2 Results for profile 2
that the elastic stiffness of profile 2 is much lower
The results for the (much smaller) profile 2 have than that of profile 1. This is reasonable since the
similar characteristics to those for profile 1. Within cross-section of profile 2 is much smaller than that
each series, the results are very close to each other, of profile 1, while the spans of both profiles are the
while when a comparison is made between the same, at 4 m. In Fig. 7, the structure appears to stiffen
two series, a significant difference exists at larger up as it yields. This is due to the axial force effect
deflections. Again, following on from the discussion and is obvious even when the central deflection is
above, the FEA results with strain hardening are more small.
Downloaded from sdj.sagepub.com at Imperial College London Library on March 11, 2016
Static plastic analysis of corrugated steel panels 141
3.6.3 Results for profile 3 two cases, which can validly be simplified as one-
dimensional problems, this case is a two-dimensional,
As can be seen in Fig. 8, the agreement between
deep beam, problem.
the theoretical result and FEA is not as good as
In order to clarify the problem, it can be assumed
before, except at large deflections. This simply
that
means that the theoretical method is not really
suitable for this case. Examining the stress diagram s =k s (9)
from Abaqus (Fig. 9), it can be seen that s and s 11 s 22
11 22
are actually of the same order in the middle part of where k is the ratio between s and the longitudinal
s 11
the corrugated beam. Therefore, unlike the former stress s , and all other stresses are zero. Substituting
22
Downloaded from sdj.sagepub.com at Imperial College London Library on March 11, 2016
142 Y H Liang, L A Louca, and R E Hobbs
these assumed conditions into the von Mises criterion 4 THE INFLUENCE OF CONNECTIONS
Downloaded from sdj.sagepub.com at Imperial College London Library on March 11, 2016
Static plastic analysis of corrugated steel panels 143
4.3 Simplification
Parallel load–deflection curves imply the same stiff-
ness, which means that in fact the influence of the
connections is not important in the later part of the
load–deflection curve. From the above comparison
between the results with/without connections, a
simplification of the entire deflection process of the
case with connections can be made by assuming that
the entire deflection procedure can be idealized into
two stages. In the first stage, the effect of connections
is taken into account so that the deformation of con-
nections also leads to deflection, while in the second
stage, the influence of connections is neglected and
only the corrugated panel is considered.
8(M +M )
q = 0e 0p (13)
0 L2
p
where M and M are the maximum plastic bending
0p 0e
moments of the corrugated beam and the endplate
respectively.
Downloaded from sdj.sagepub.com at Imperial College London Library on March 11, 2016
144 Y H Liang, L A Louca, and R E Hobbs
deflection are as shown in Fig. 13: Combining equations (14) to (17) gives
M +M =N h cos h+Rh sin h (14) q
0e 2 p e e M +M =N h cos h+ (L −2h sin h)h sin h
(M −M )+N [d −h (1−cos h)] 0e p p e 2 p e e
1 2 p p e
A B
q L 2 (18)
= p −h sin h (15)
2 2 e
A B
q L 2
N [d −h (1−cos h)]= p −h sin h (19)
p p e e
A B
L 2 2
R=q p −h sin h (16)
2 e
4.4.4 The relationship between d , h, and w
p
4.4.3 The relationship between M and M
1 2 Following the assumptions noted above, plastic
In order to solve the equations above, it is essential deformation only exists at the end A and the mid-
to find a relationship between M and M . As can be span C. From Fig. 14, the rotation of the endplate is h
1 2
seen from Fig. 13(c), a vertical reaction R acts at A, and the plastic rotational angle at C is w. Applying
which means that shear stress exists on the cross- the normality theory, since M is assumed to be equal
1
section at A, while it is clear that there is no shear to M , the plastic rotational angle at A should also
2
stress at the mid-span C due to the symmetric con- equal w. It follows that h equals 2w.
dition. Thus, strictly speaking, M is not the same In Fig. 13, it can be seen that
1
as M , and indeed it is easy but tedious to show that
2
A B
M is somewhat bigger than M . L
1 2 d =h (1−cos h)+ p −h sin h tan w (20)
However, in a simplified model, particularly in the p e 2 e
first stage when q is relatively small, the relationship
between M and M can still be assumed to be Substituting h=2w into equation (20) gives
1 2
AB
M =M =M (17) L h
p 1 2 d = p tan (21)
p 2 2
Downloaded from sdj.sagepub.com at Imperial College London Library on March 11, 2016
Static plastic analysis of corrugated steel panels 145
Downloaded from sdj.sagepub.com at Imperial College London Library on March 11, 2016
146 Y H Liang, L A Louca, and R E Hobbs
detail, it can be seen that the difference between the deflection of the panel and the external distributed
‘bilinear’ curves and the ‘whole’ panel one excluding load, both theoretical and finite element analyses are
the strain hardening effect is less than the other two. applied to three different profiles, and comparisons
This is because, with lateral supports, the ‘whole’ between the theoretical and FEA results are provided.
case is stiffer than the other two. Consequently, the By doing this, it can be found that, in modelling the
plastic strain of the ‘whole’ case is relatively smaller; corrugated panel without connections, the simplified
therefore the effect of necking and tearing in the method works well compared to the more costly
panel is smaller. finite element analysis. Even in an analysis including
connections, the enhanced simplified method can
also provide accurate enough results after making
5 CONCLUSION some further assumptions and simplifications. The
main differences between this simplified method and
This paper has developed an improved approximate the FEA method are due to the different assumptions,
method for the analysis of a corrugated panel. Con- such as neglecting the shear effect and cross-sectional
centrating on the relationship between the central distortion. To summarize, assuming the finite element
Downloaded from sdj.sagepub.com at Imperial College London Library on March 11, 2016
Static plastic analysis of corrugated steel panels 147
analysis is reliable, the simplified method can model 12 Louca, L. A., Boh, J. W., and Choo, Y. S. Design and
the corrugated panel with sufficient accuracy, as far analysis of stainless steel profiled blast barriers.
as the relationship between the central deflection J. Constructional Steel Res., 2004, 60, 1699–1723.
13 Louca, L. A., Boh, J. W., and Choo, Y. S. Response of
and the external distributed load is concerned. In
profiled barriers subject to hydrocarbon explosions.
practice, the method can give fairly good results In Proceedings of the Institution of Civil Engineers on
for shallow and medium depth profiles (normally Structures and buildings, 2004, Vol. 157, pp. 319–331.
depth <400 mm) depending on the cross-section 14 Abaqus user’s manual, Version 6.3, 2002 (ABAQUS,
geometry, while, for deep profiles, further analysis Inc., Providence, Rhode Island).
concerning the shear effect may be needed. In com- 15 Langdon, G. S. and Schleyer, G. K. Inelastic
parison with finite element analysis, the theoretical deformation and failure of profiled stainless steel
blast wall panels, Part 1: analytical modelling con-
model can provide acceptable results efficiently and
siderations. Int. J. Impact Engng, 2005, 41(4),
economically. On the other hand, due to those 371–399.
assumptions and simplifications, it still has many
limitations, which would repay further research.
APPENDIX 1
Notation
REFERENCES a ,b ,c ,h ,t ,h
p p p p p p
geometric configurations of a corrugated
1 Caldwell, J. B. The strength of corrugated plating panel
for ship’s bulkheads. Trans. Royal Instn of Naval
A cross-sectional area of the corrugated panel
Architects, 1955, 97, 495–522. p
e, f, g, i, j, k, r, s, u, v, w, j, Q, P , P , P
2 Paik, J. K., Thayambali, A. K., and Chun, M. S. 1 2 3
Theoretical and experimental study on the ultimate proceeding variables for simplifying the
strength of corrugated bulkheads. J. Ship Res., 1997, equations
41, 301–317. h height of the endplate
3 Rockey, K. C. and Evans, H. R. The behaviour of e
K ratio between the lateral and longitudinal
corrugated flooring systems. Thin walled steel s
stress
structures, their design and use in building. In
L span of the corrugated panel
Symposium at University College of Swansea p
m dimensionless bending moment=M /M
School of Engineering, 11–14th September 1967, p p 0p
pp. 236–257. M bending moment of the corrugated panel
p
4 Ji, H. D., Cui, W. C., and Zhang, S. K. Ultimate M yield bending moment of the endplate
0e
strength analysis of corrugated bulkheads consider- M yield bending moment of the corrugated
ing influence of shear force and adjoining structures. op
panel
J. Constructional Steel Res., 2001, 57, 525–545. M mid-span bending moment of the
5 Schleyer, G. K. and Hsu, S. S. A modelling scheme 1
corrugated panel
for predicting the response of elastic–plastic structure
M bending moment at supports of the
to pulse pressure loading. Int. J. Impact Engng, 2000, 2
24, 759–777. corrugated panel
6 Jones, N. Structural Impact, 1989 (Cambridge n dimensionless axial force=N /N
p p 0p
University Press, Cambridge). N axial force of the corrugated panel
p
7 Alves, M. and Jones, N. Impact failure of beams N yield axial force of the corrugated panel
using damage mechanics. J. Impact Engng, 2002, 27, 0p
q external distributed uniform load
837–861. q collapse external distributed uniform load
8 Specification for weldable structural steels, 1986 BS 0
4360 : 1986 (British Standards Institution, London).
R support reaction force of the corrugated
9 Konishi, H., Yao, T., Shigemi, T., Kitamura, O., and panel
Fujikubo, M. Design of corrugated bulkhead of x dimensionless ratio to determine the
p
bulk carrier against accidental flooding load. In neutral axis
Proceedings of the 7th International Symposium on y distance between the equal area and
Practical design of ships and mobile units (PRADS’ 98), p
neutral axes
The Hague, 1998, pp. 157–163. z distance to determine the upper limit of
10 Onat, E. T. and Prager, W. The influence of axial
the simplified stress due to an axial force
forces on the collapse load of frames. In Proceed-
Z plastic modulus of the corrugated panel
ings of the First Midwestern Conference on Solid pp
mechanics, University of Illinois, 1953, pp. 40–42.
11 Calladine, C. R. Plasticity for engineers, 1985, d central deflection of the corrugated panel
p
pp. 44–46 (Ellis Horwood Limited, Chichester). h endplate rotation
Downloaded from sdj.sagepub.com at Imperial College London Library on March 11, 2016
148 Y H Liang, L A Louca, and R E Hobbs
GA B A B
s yield stress 1 t 2 t
0 v= c x h + p −2c j x h + p
w rotation of the corrugated panel with Z p p p 2 p p p 2
endplates pp
w yield surface expression +c j2+a t (1−x )h +a t j
p p p p p p p p
C A B
2t r
+ p (1−2x ) j+ h
sin h p 2 p
APPENDIX 2 p
A B
h −t
Detailed plastic analysis of the corrugated panel +(1−2x ) p p h
p 2 p
A B
t (1−2x )h t t
Case 2: x h − p ∏y < p p p+ x h − p
CA B A BD
k r
p p 2 p sin h c p p 2 −(kn +r) i+ n + j+
p p p 2 p 2
From Fig. 18, it can be shown that
DH
h −t
y −j −(kn +r) p p
n = p (22) p 2
p i
Again applying the normality theory, d in this stage
z=kn +r (23) p
p is then
where 2Z
d =− pp (2sn +u) (25)
C A BD
A 1 x h −t /2 t p A p
i= p , j=− p p p t − x h − p c p
4c c sin h p p p 2 p
p p p
A B
(1−2x )h t t t
p p p + x h − p ∏y <x h + p
C A BD
ic t c Case 3:
p p p p p
k= p sin h , r= j− x h − p p sin h sin h c 2 2
t p p p 2 t p p p
p p From Fig. 19, it can be shown that
Consequently, the dimensionless bending moment
m can be obtained as y −w
p n = p (26)
p i
m =sn2 +un +v (24)
p p p
where
where
CA B D
1 t t −2x h
G C A BDH c x h − p + p
1 2t k w= p pt
s= c i2− p k i+ c p p p 2 2 sin h p
Z p sin h 2 p p
pp p
z=jn +Q (27)
G A B
1 t p
u= −2c i x h + p +2c ij+a t i
Z p p p 2 p p p where
pp
2c i
C A B
2t k p
+ p (1−2x ) i+ h j=
sin h p 2 p a
p p
C A B D
2c w t 2(1−2x )h t
C A B A BD
r k 2c
− k j+ +r i+ Q= p − p x h − p + p p p
2 2 a a p p 2 a sin h
p p p p
A BDH
h −t It can then be shown that
−k p p
2 m =P n2 +P n +P (28)
p 1 p 2 p 3
Downloaded from sdj.sagepub.com at Imperial College London Library on March 11, 2016
Static plastic analysis of corrugated steel panels 149
with
C A BD
1 j
P = c i2−a j i+
1 Z p p 2
pp
G A B A B
1 t j
P = −2c x h + p i+2c iw+a t i+
2 Z p p p 2 p p p 2
pp
C A B A BD
Q j
−a j w+ +Q i+
p 2 2
Fig. 20 Normality theory of the yield surface
p H
−a (1−x )h j
p p
CA B A B
1 t 2 t
P = c x h + p −2c x h + p w+c w2 APPENDIX 3
3 Z p p p 2 p p p 2 p
pp
The normality theory of the yield surface
A B
Q
+a t (1−x )h +a t w+
p p p p p p 2 Figure 20 shows the geometric interpretation of
the normality theory. Assuming that the axes of the
A B D
Q principal strain are coincident with the axes of
−a Q w+ −(1−x )h a Q
p 2 p p p the principal stress, it can be easily found that the
work done by the external agency is maximized if the
Using the normality theory then gives
loading vector (s ) corresponds to the point at which
ij
2Z the deflection increment (dep ) is normal to the yield
d =− pp (2P n +P ) (29) ij
p A 1 p 2 surface. It can also be expressed as a requirement
p that the direction of the deflection increment vector
t (dep ) should be normal to the loading increment
Case 4: x h + p <y ij
p p 2 p vector (ds ).
ij
From this theory, the general expression in the
In this stage, the membrane force reaches its maxi- analysis above can be written as
mum value, which means that n equals 1, while the
p
non-dimensional bending moment, m , equals 0. qN Sde=qM Sdh (30)
p
Downloaded from sdj.sagepub.com at Imperial College London Library on March 11, 2016