Antonio Campanile, Department of Naval Engineering, University “Federico II”, Naples

Masino Mandarino, Department of Naval Engineering, University “Federico II”, Naples

Vincenzo Piscopo, Department of Naval Engineering, University “Federico II”, Naples

DESIGN CHARTS FOR CLAMPED ORTHOTROPIC PLATES

Summary

This paper focuses on the application of orthotropic plate bending theory, particularly of
Schade’s design curves, to the scantlings assessment of deck primary supporting members in
absence of pillars. Preliminarily, a review of Schade’s works is presented, paying attention to
the design curves relative to the plate with all edges clamped, that are almost totally
incomplete. Therefore a numerical solution for the clamped orthotropic plate equation is
obtained. The Ritz method is adopted, expressing the vertical displacement field by a double
cosine trigonometric series, whose coefficients are determined by solving a linear equation
system. Several design curves are proposed for different values of the torsional parameter t
in a range comprised between 0 and 1, as functions of the virtual aspect ratio  comprised
between 1 and 8, in order to estimate the asymptotic behaviour of the orthotropic plate when
   . Finally, a numerical application relative to the deck of a fast ferry is presented.

Key words: orthotropic plate bending; Ritz method; stiffened plate.

XVIII Symposium SORTA 2008 Design charts for clamped orthotropic plates

1. Introduction
Schade, 1942, proposed some practical general design curves, based on the “orthotropic
plate” theory, in order to obtain a rapid, but accurate, dimensioning of plating stiffeners.
Schade considered four types of boundary conditions for the orthotropic partial differential
equation: all edges rigidly supported but not fixed; both short edges clamped, both long edges
supported; both long edges clamped, both short edges supported; all edges clamped. The last
case with all edges clamped was left almost totally incomplete. The few data useful for this
boundary condition were taken from Timoshenko et al., 1959, and Young, 1940, as given for
the isotropic plate only for the torsional coefficient value t =1 and for a range of the virtual
aspect ratio  comprised between 1 and 2.
In this work a numerical solution of the clamped orthotropic plate equation is obtained.
Numerical results are presented in a series of charts similar to those ones given by Schade.

2. A numerical solution of the clamped rectangular orthotropic plate equation

Orthotropic plate theory refers to materials which have different elastic properties along
two orthogonal directions. In order to apply this theory to panels having a finite number of
stiffeners, it is necessary to idealize the structure, assuming that the structural properties of the
stiffeners may be approximated by their average values, which are assumed to be distributed
uniformly over the width and the length of the plate.
The deflection field in bending is governed by the so called Huber’s differential
equation:
4w 4w 4w
DX  2 H 2 2  DY 4  p ( x, y ) (1)
x 4 x y y
where:
 DX is the unit flexural rigidity around the y axis;
 DY is the unit flexural rigidity around the x axis;
 H  D XY  2 D33   t D X DY according to the definition by Schade;
 p is the pressure load over the surface.
It is noticed that the behaviour of the isotropic plate with the same flexural rigidities in
all directions is a special case of the orthotropic plate problem.
Indicating with n the normal external to the plate contour, a numerical solution of the
orthotropic plate equation with the boundary conditions:
w
w=0 and 0 (2)
n
along all edges is presented. Two solution methods are available: the double cosine
series and the Hencky’s method. The second one is well known to converge quickly but does
pose some difficulties with regard to programming due to over/underflow problems in the
evaluation of hyperbolic trigonometric functions with large arguments. The double cosine
series method, instead, is devoid of the over/underflow issue but is known to converge very
slowly.
If a and b are the plate lengths in the x and y directions respectively, the vertical
displacement field may be expressed by means of the following double cosine series whose
terms satisfy the boundary conditions (2):

2
Design charts for clamped orthotropic plates XVIII Symposium SORTA 2008

M N
w ,      1  cos 2m  1  cos 2n wm , n (3)
m 1 n 1

where:
x = a with 0 ≤  ≤ 1 ; y = b with 0 ≤  ≤ 1 (4)
The unknown coefficients wm,n may be determined using the Ritz method, searching for
the minimum of the total energy, which is given by:

1   2w  2w  
2 2 2
2w 2w  2w 
V    D X  2   2 D XY  4 D33    DY  2     wpdA (5)
2 A   x  x 2 y 2  xy   y   A

Applying the generalized integration by parts formula, equation (5) reduces in this case
to:

1   2w   2w  
2 2
2w 2w
V    D X  2   2 H 2  DY  2     wpdA (6)
2 A   x 

x y 2  y   A
which can be re-written in the form:

1  DX 
2 2
V
1 1
 2w  2 H  2 w  2 w DY   2 w 
1 1

ab 2 0 0  a 4
    2   2 2  4  2   dd    wpdd (7)
    a b  2
 2
b     0 0

The stationary point is obtained imposing the following MxN equations system:

0 for m = 1, …, M ; n = 1, …, N (8)
wm,n

So, considering p as uniformly distributed, the generic equation, for m  m and n  n ,

assumes the form:


1 1 D  2w 
2
2 H  2 w  2 w DY   2 w 
2
 
1 1

0 0  a 4
X
 2   2 2     dd  2 p   wdd (9)
wm ,n    a b  2  2 b 4   2   wm ,n
 0 0

Defining the non dimensional vertical displacements:

w wm ,n
 ;  m ,n  (10)
pb 4 DY pb 4 DY
from the equations (3) and (9) it is obtained:
 1  4 N
 M
2 t 2 2 
 
4 4 4
4 4  4  m   2 m    n   2 n   m n  m,n 
1 (11)
 
m,n m,n m ,n m ,n 2
 n 1  m 1 
where  is the virtual side ratio.
Even if the double cosine trigonometric series converges very slowly, adopting
sufficiently high values for M and N, it is possible to obtain a very accurate solution of the
equation (1) with the boundary conditions (2).

3
XVIII Symposium SORTA 2008 Design charts for clamped orthotropic plates

3. Characterization of the behaviour of clamped stiffened plates

The orthotropic plate bending theory can be applied to a plate reinforced by two systems
of parallel beams spaced equal distances apart in the x and y directions. The rigidities DX and
DY of equation (1) can be specialized as follows:
EI eX EI eY
DX   EiX DY   EiY (12)
sX sY
where E is the Young’s modulus, sX (sY) is the distance between girders (transverses),
IeX (IeY) is the moment of inertia, including effective width beX (beY) of plating, of long (short)
repeating stiffeners.
The torsional coefficient t and the virtual side ratio  can be specialized according to
Schade’s works:
i pX i pY a iY
t   4 (13)
i X iY b iX
where ipX (ipY) is the moment of inertia of effective breadth of plating working with long
(short) supporting stiffeners per unit length. In the following rXp (rYp) is the vertical distance
of the associated plating working with long (short) supporting stiffeners from the section
neutral axis, while rXf (rYf) is the distance of the free flange from the section neutral axis.
The meaning of the two parameters is quite clear. In particular, the torsional coefficient t,
which lies between 0 and 1, exists because only the plating is subject to horizontal shear,
while both the plating and stiffeners are subject to bending stress. Obviously t = 1, and
ipX = ipY = iX = iY , represents the isotropic plate case. The virtual side ratio  is the plate side
ratio modified in accordance with the unit stiffnesses in the two directions; as usual, it has
been admitted that  is always equal to or greater than unity. In the next the quantities
represented in the diagrams are presented.
Deflection at center, fig.1: the vertical displacement at the plate center (== 0.5) is the
maximum and is so expressed:
pb 4 M N
wmax  kW where: kW  ,     m,n 1  cos m 1  cos n  (14)
EiY m 1 n 1

Edge bending stress in plating, fig.2: these curves give the bending stress in the plating at
the centers of edges where fixity exists. The stress at the center of such an edge may be
treated as the maximum along that edge. The maximum stresses in the plating in the long and
short directions respectively are:
pb 2 rXp pb 2 rYp
 XpSUP  k XpSUP  ,   YpSUP  kYpSUP  ,  (15)
i X iY iY

where:
1 4 2 M N
k XpSUP  ,   2  m 2 1  cos n  (16.1)
 1  2
m,n
m 1 n 1

4 2 M N
kYpSUP  ,    n 2 1  cos m  (16.2)
1  2
m ,n
m 1 n 1

4
Design charts for clamped orthotropic plates XVIII Symposium SORTA 2008

Edge bending stress in free flanges, fig.3: these curves give the bending stress in the
free flanges at the centers of edges where fixity exists. The stress at the center of such an edge
may be treated as the maximum along that edge. The maximum stresses in the free flanges for
girders and transverses are respectively:
pb 2 rXf pb 2 rYf
 XfSUP   k XfSUP  ,   YfSUP   kYfSUP  ,  (17)
i X iY iY

where:
4 2 M N
k XfSUP  ,     m 2 1  cos n  (18.1)
2 m 1 n 1
m ,n

M N
kYfSUP  ,   4 2   m ,n n 2 1  cos m  (18.2)
m 1 n 1

It is important to note that when    kYfSUP is substantially independent on t and is

equal to 1/12, that is the beam theory value. Furthermore the curves show that for low values
of t the maximum deflections and stresses parallel to the short direction occur at values of 
between 1.5 and 2.0: this indicates that the long beams add to the load taken by the short
beams, instead of helping to support it.
In order to verify the goodness of the method, the following tables shows a comparison
between the values obtained applying the Ritz method and the ones taken from Timoshenko
et al., 1959, for the isotropic plate.

0.0030

0.0028

0.0026
0.0026

0.0024
eta 0
0.0022
eta 0.25
kW

0.0020 eta 0.50

eta 0.75
0.0018
eta 1.00
0.0016

0.0014

0.0012

0.0010
1 1.5 2 2.5 3 3.5 4 4.5  5 5.5 6 6.5 7 7.5 8
Fig. 1 Deflection at center

5
XVIII Symposium SORTA 2008 Design charts for clamped orthotropic plates

0.100

0.095
kYpSUP 0.0916
0.090
eta 0
0.085 eta 0
eta 0.25
eta 0.25
0.080
k values

eta 0.50
eta 0.50
0.075 eta 0.75
eta 0.75
0.070 eta 1.00
eta 1.00

0.065
kXpSUP 0.0627

0.060

0.055
1 1.5 2 2.5 3 3.5 4 4.5  5 5.5 6 6.5 7 7.5 8
Fig. 2 Edge bending stress in plating

0.090

0.085
KYfSUP 0.0833

0.080 eta 0
eta 0
eta 0.25
0.075 eta 0.25
eta 0.50
k values

eta 0.50
0.070
eta 0.75
eta 0.75
0.065 eta 1.00
eta 1.00

0.060
kXfSUP 0.0571

0.055

0.050
1 1.5 2 2.5 3 3.5 4 4.5  5 5.5 6 6.5 7 7.5 8
Fig. 3 Edge bending stress in free flanges

6
Design charts for clamped orthotropic plates XVIII Symposium SORTA 2008

Table 1 Comparison between Timoshenko and actual data

Edge bending moment Edge bending moment

Deflection at center
in short direction in long direction
kW (1-n2) KYpSUP (1-n2) KXpSUP
 Timoshenko Timoshenko Timoshenko
t = 1.00 t = 1.00 t = 1.00
1.00 0.00126 0.00126 0.0513 0.0510 0.0513 0.0510
1.20 0.00172 0.00172 0.0639 0.0636 0.0554 0.0558
1.40 0.00207 0.00207 0.0726 0.0724 0.0568 0.0570
1.60 0.00230 0.00230 0.0780 0.0779 0.0571 0.0571
1.80 0.00245 0.00245 0.0812 0.0811 0.0571 0.0571
2,00 0.00254 0.00253 0.0829 0.0828 0.0571 0.0571
∞ 0.00260 0.00260 0.0833 0.0833 0.0571 0.0571

4. Convergence of the method

In the following, the influence of the number of harmonics on k values is shown.
Particularly, assuming = 5 and = 0.50, M=N has been varied from 5 up to 100, in order
to obtain a number of harmonics comprised between 25 and 10000. If the number of
harmonics is > 4900, i.e. M=N > 70, a good convergence in the assessment of k values, and
then of the proposed curves, is obtained for practical purposes, as it can be appreciated from
fig. 4, 5, 6.

0.092
0.00266

0.091
0.00265
0.090

0.00264 0.089

0.088
0.00263
kYfSUP

0.087
kW

0.00262 0.086

0.085
0.00261
0.084

0.00260 0.083

0.082
0.00259
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000
MxN
Fig. 5 Design charts for clamped orthotropic
MxN
Fig. 4 KW convergence
plates

0.065

0.060

0.055

0.050

0.045
kXfSUP

0.040

0.035

0.030

0.025

0.020
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000
MxN

Fig. 6 KXpSUP convergence

7
XVIII Symposium SORTA 2008 Design charts for clamped orthotropic plates

5. Application to a ro-ro fast ferry deck

The orthotropic plate bending theory has been applied to the garage deck of a fast ferry,
with a displacement of 1420 t, considered in [6]. The deck spans 80 m in length and 16 m in
breadth; its thickness is 8 mm. Both transverse beams and longitudinal girders are T sections
320x150x10x15 mm, 2 m spaced. The material is high-strength steel with ReH = 355 N/mm2.
The maximum equivalent pressure load peq.max acting on transverse beams has been
estimated as 4914 N/m2, according to RINA rules, 2005. To this extent, it has been assumed a
typical vehicle axle load of 1.2 t, 7 vehicles located on the transverse beam and a vertical
acceleration of 0.909 g.
To enter relevant charts, the plate parameters have been assessed:  = 5; t = 0.19.
Because of the large value of the virtual side ratio, as usual for garage decks, the asymptotic
values hold, that is:
k YfSUP  0.0833 k XfSUP  0.0571

Hence, maximum stresses in transverses and girders are derived:

 YfSUP  205 N / mm 2  XfSUP  141N / mm 2
For comparison, a coarse mesh FE analysis has been performed, obtaining stresses close
to the previous ones, i.e. YfSUP.= 214 N/mm2; XfSUP.= 146 N/mm2
Obtained results are relative to a uniform pressure acting on the deck. In this respect, it
must be pointed out that peq.max corresponds to the vehicle axle located directly on a transverse
beam and this coincidence cannot obviously occur at every transverse beam. In other terms,
the actual load is not evenly distributed, at least longitudinally; peq.max has to be considered as
a peak value and corresponding stress values as upper bound values.
For instance, on the basis of longitudinal dimensions of the typical vehicle, the actual
equivalent pressure distribution can be characterized by:
peq.mean = 0.75 peq.max ; COV = 0.26
Then, a lower bound of acting stresses can be evaluated considering a uniform pressure
peq.mean; correspondingly it is obtained:
 YfSUP  161N / mm 2  XfSUP  105 N / mm 2
The significance of these values has been stressed by an independent analysis where the
actual pressure distribution has been taken into account; the following values have been
obtained:
 YfSUP  164 N / mm 2  XfSUP  100 N / mm 2
These values are very close to the previous ones relative to p = peq.mean = const; it must
be concluded that there is a significant stress re-distribution due to the longitudinal rigidities.
Moreover, obtained results suggest the possibility of evaluating acting stresses with reference
to the mean equivalent pressure load evenly distributed over the deck. Obviously, this
simplified approach, and its range of applicability, should be verified extending the analysis
to typical deck configurations and appropriate load distributions.

8
Design charts for clamped orthotropic plates XVIII Symposium SORTA 2008

6. Conclusions
In this work the orthotropic rectangular plate bending equation with all edges clamped
has been solved adopting the Ritz method. Numerical calculations have been systematically
performed varying two non-dimensional parameters, namely the virtual side ratio and the
torsional coefficient. Response non-dimensional parameters, in terms of maximum deflection
and maximum stresses, are given in a series of charts for their easy application. Some
comparisons with well known published data and FEM analyses give a validation to the
method. The method has been applied to the garage deck of a fast ferry, taking into account,
according to the geometrical and mass characteristics of the reference vehicle, two different
pressure loads, representative of the upper and lower bounds.

REFERENCES
[1] Schade H.A., ‘Design Curves for Cross-Stiffened Plating under Uniform Bending Load’, Trans.
SNAME, 49, 1941.
[2] Timoshenko S., Woinowsky-Krieger S., ‘Theory of Plates and Shells’, Mc-Graw-Hill Book
Company, 1959.
[3] Young D., ‘Analysis of Clamped Rectangular Plates’, Journal of Applied Mechanics, Volume 7,
No.4, December 1940.
[4] Fiorenza R., ‘Appunti delle lezioni di Analisi Funzionale’, Gli Strumenti di Coinor, 2005.
[5] RINA, ‘Rules for the Classification of Ships’, 2005.
[6] Campanile A., Mandarino M., Piscopo V., ‘Considerations on dimensioning of garage decks’,
ICMRT 2007.