Theory of Plates and Shells

This page
intentionally left
blank
THEORY OF PLATES
AND
SHELLS
This page
intentionally left
blank
THEORY OF PLATES
AND
SHELLS
S.S. Bhavikatti
Emeritus Professor
Department of Civil Engineering
BVB College of Engineering and Technology, Hubli
(Formerly Professor and Dean NITK–Surathkal,
SDM College of Engg. and Tech.–Dharwad, and
Principal, Rao Bahadur Y. Mahabaleshwarappa Engg. College, Bellary)
Publishing Globally
NEW AGE INTERNATIONAL (P) LIMITED, PUBLISHERS

New Delhi • Bangalore • Chennai • Cochin • Guwahati • Hyderabad
• Kolkata • Lucknow • Mumbai
Visit us at www.newagepublishers.com
Copyright © 2012, New Age International (P) Ltd., Publishers
Published by New Age International (P) Ltd., Publishers
All rights reserved.

No part of this ebook may be reproduced in any form, by photostat, microfilm, xerography,
or any other means, or incorporated into any information retrieval system, electronic or
mechanical, without the written permission of the publisher. All inquiries should be
emailed to rights@newagepublishers.com
ISBN (13) : 978-81-224-3492-7
PUBLISHING FOR ONE WORLD

NEW AGE INTERNATIONAL (P) LIMITED, PUBLISHERS
4835/24, Ansari Road, Daryaganj, New Delhi - 110002
Visit us at www.newagepublishers.com
Preface
Plate is a flat surface having considerably large dimensions compared to its thickness with supports
along few edges and is subjected to transverse load. For civil engineer common example of plate is a
slab. At undergraduate level students are taught design of slab by approximate methods or by using
moment co-efficients given in the code, without going through how they are obtained. At the post-
graduate level theory of plates is taught to structural engineering students to understand actual
load transfer in plate by elastic analysis. It involves forming and solving fourth order differential
equations.
Shells are curved plates. The analysis of shells involves additional complexity. A design engineer
should understand mechanism of load transfer and internal forces developed in the shells. Shells are to
cover large area free of columns and architects prefer them for their aesthetic appeal. A structural
engineer has to learn theory of shells to design economical shell structures with more confidence.
In this book theory of plates and shells is explained. The analysis is restricted to classical method
only. Finite element method is the numerical method suitable for the analysis. Author has covered the
shell analysis by finite element analysis in his seperate book. A number of commercial packages are
available for the analysis by finite element method. But it is necessary that design engineer should have
basic knowledge of load transfer and internal forces that develop, which is possible by going through
classical theory. For validating the results obtained by finite element packages, classical theory for
commonly found standard cases is essential.
The book is essentially based on the lecture notes of the author taught to students of M.Tech.
(Industrial Structure) at NITK Surathkal and M.Tech. (Structural Engineering) at BVB College of
Engineering for over 40 years. The author hopes that the book will be quite useful as a textbook for
M.Tech. students to gain confidence in taking design of plates and shells.
S.S. BHAVIKATTI
This page
intentionally left
blank
Acknowledgement
The author sincerely thanks Dr. A.S. Arya, former Professor of Civil Engineering and Director of
School of Earthquake Engineering, IIT Roorkee for introducing him to the subject theory of plates and
shells and guiding him for M.Tech. thesis on the analysis of continuous cylindrical shell. He is thankful
to NITK Surathkal and BVB College of Engineering and Technology for giving him opportunity to
teach this subject to M.Tech. students for several batchestime. Thanks are also due to a number of
M.Tech. students who took their dissertation work on the analysis and design of shells and to Smt. T.K.
Lakshmy for taking up Ph D work on optimum design of shells and folded plates.
S.S. BHAVIKATTI
This page
intentionally left
blank
Contents
Preface v
Acknowledgement vii
1. Introduction to Plates 1–4

1.1 Coordinate Systems 2
1.2 Stresses on an Element 3
1.3 Types of Theories of Plates 3
1.3.1 Theory of Thin Plates with Small Deflections 3
1.3.2 Theory of Thin Plates with Large Deflections 4
1.3.3 Theory of Thick Plates 4
Questions 4
2. Pure Bending of Plates 5–22

2.1 Slope in Slightly Bent Plate 5
2.2 Curvature of Slightly Bent Plate 7
2.3 Principal Curvature 10
2.4 Displacement—Strain Relations 12
2.5 Strains in Terms of w 14
2.6 Strain-stress Relations 15
2.7 Expressions for Moments 17
2.8 Moment in any Direction 19
2.9 Principal Moments 21
Questions 22
3. Small Deflections of Laterally Loaded Plates 23–30

3.1 Stress Resultants on a Typical Plate Element 23
3.2 Equations of Equilibrium 24
3.3 Expressions for Vertical Shears 26
3.4 Boundary Conditions 27
3.5 Corner Reaction 29
Questions 30
x CONTENTS
4. Fourier Series of Loadings 31–42

4.1 Fourier Series of Loads on Beams 31
4.2 Fourier Series for UDL on Beams 32
4.3 Fourier Equivalent for UDL over a Small Length 33
4.4 Fourier Equivalent for a Concentrated Load 34
4.5 Fourier Equivalent for Hydrostatic Load 34
4.6 Fourier Equivalent for Triangular Load with Peak Value at Mid-Span 36
4.7 Double Fourier Series Expressions for Loads 37
4.8 Double Fourier Form for UDL 37
4.9 Double Fourier Form for UDL over Small Area 38
4.10 Double Fourier Form for Concentrated Load 39
4.11 Double Fourier Equivalent for Hydrostatic Load 40
4.12 Double Fourier Equivalent for Triangular Load with
Peak Value at Mid-Span 41
Questions 42
5. Navier’s Solution for Rectangular Plates 43–46

5.1 Navier Solution for Rectangular Plate Subjected to UDL 43
5.2 Navier Solution for Any Loading 46
Questions 46
6. Levy’s Solution for Rectangular Plate Analysis 47–54

6.1 Analysis of Rectangular Plate Subjected to UDL by Levy’s Method 47
6.2 Physical Meaning of Particular and Complementary
(Homogeneous) Integrals 53
6.3 Convergence Study 53
6.4 Comparison Between Navier’s Solution and Levy’s Solution 54
Question 54
7. Rectangular Plates with Various Edge Conditions 55–72

7.1 Simply Supported Plate Subject to Moments Along y = 0 and y = b 55
7.2 Rectangular Plate with Two Opposite Edges Simply Supported and the
Other Two Edges Fixed 61
7.3 Rectangular Plate with All Four Edges Fixed 65
7.4 Rectangular Plate with One Edge Simply Supported, Other Edges Fixed 66
7.5 Rectangular Plate with Two Adjacent Edges Simply Supported and
Other Two Fixed 67
7.6 Rectangular Plate with Two Opposite Edges Simply Supported,
One Edge Fixed and Another Free 68
7.7 Rectangular Plates with Three Edges Built in and Fourth Edge Free 69
7.8 Rectangular Plate Continuous in One Direction 70
7.9 Rectangular Plates Continuous in Both Directions 70
Questions 72
CONTENTS xi
8. Circular Plates Bent Axi-symmetrically 73–91

8.1 Co-ordinates and Element 73
8.2 Axi-symmetric Bending of Plate 75
8.3 Cases of Practical Importance 90
Questions 91
9. Plates of Other Shapes 92–105

9.1 Elliptic Plate with Clamped Edges and Subjected to UDL 92
9.2 Comparison of Result with Strip Method Analysis 97
9.3 Elliptic Plate Subject to Uniformly Varying Load in x-Direction 97
9.4 Equilateral Triangular Plate Subject to Uniform Edge Moment 99
9.5 Equilateral Triangular Plate Simply Supported Along its Edges
and Subjected to UDL 104
Questions 105
10. Energy Method 106–119

10.1 Expression for Total Energy 106
First term in strain energy V 112
First term in strain energy 113
Second term in strain energy 113
Third term in strain energy 113
Questions 119
11. Finite Difference Method 120–141

11.1 Finite Difference Method vs. Classical Method 120
11.2 Finite Difference Form for Differential Equations 121
11.3 Finite Difference Form for Plate Equation 124
11.4 Expressions for Stress Resultants 127
11.5 Applying Boundary Conditions 129
11.6 Modification of Finite Difference Plate Equation to Apply it on
A Free Edge 138
Questions 141
12. A Folded Plate Roofs 142–157

12.1 Advantages and Disadvantages of Folded Plates Over Shells 143
12.2 Historical Note 143
12.3 Assumptions 143
12.4 Behaviour of Folded Plate Roofs 143
12.4.1 Slab Action 143
12.4.2 Plate Action 144
12.5 Whitney’s Method 145
12.6 Simpsons Method 152
12.7 Stress Distribution Procedure 155
Questions 157
xii CONTENTS
13. Introduction to Shells 158–163

13.1 Classification of Shells 160
13.2 Advantages and Disadvantages of Shells 160
13.3 Efficiency in the Use of Materials 160
13.4 Historical Note 162
13.5 A Brief Literature Review 163
13.6 Analysis and Design 163
Questions 163
14. Introduction to Cylindrical Shell Roof 164–181

14.1 Parts of a Single Cylindrical Shell Roof 164
14.2 Types of Cylindrical Shell Roofs 164
14.3 Design Criteria 164
14.4 Analysis 167
14.5 Beam Theory 168
14.5.1 Beam Action 168
14.5.2 Arch Action 170
Questions 181
15. Membrane Analysis of Cylindrical Shell Roofs 182–196

15.2 Cylindrical Shells with Parabolic, Catenary and
Cycloid Directrices 189
15.3 Cylindrical Shell with Semi-elliptic Directrix 194
15.4 Comments on Membrane Theory 196
Questions 196
16. Bending Theory of Cylindrical Shell Roofs 197–227

16.1 A Typical Shell Element 197
16.3 Deformation in Shell 200
16.4 Strain Displacement Relations 202
16.5 Relations Between Stress Resultants and Displacements 204
16.6 Simplified Stress Resultants and Displacement Relations 208
16.7 DKJ Theory 208
16.8 Determination of Particular Integral for Dead Load 213
16.9 Homogeneous Solution 215
16.10 Edge Conditions 220
16.11 Edge Beam Theory 221
16.12 Summary of Calculations 223
16.13 Statical Checks 224
16.14 Long and Short Shells 226
Questions 227
CONTENTS xiii
17. Analysis of Symmetrically Loaded Cylindrical

Water Tanks and Pipes 228–244
17.1 Forces on An Element 228
17.3 Stress Resultants 230
17.4 Analysis of Long Circular Pipes 240
17.5 Analysis of Water Tanks as Combination of Membrane
Solution and Edge Perturbations M0, Q0 242
Questions 244
18. Membrane Theory for Shells of Revolutions 245–266

18.1 Geometry of Shell of Revolution 245
18.3 Equations of Equilibrium for Axi-symmetrically Loaded Shells 251
18.4 Solution of Equations of Equilibrium 252
18.5 Membrane Analysis of Conical Shells 257
18.6 Rotational Hyperboloid of One Sheet 259
Questions 266
19. Membrane Theory for Shells of Translation 267–294

19.1 Assumptions 267
19.2 Coordinate Axes 267
19.3 Monge’s Notations 267
19.4 Properties of the Element 268
19.5 Membrane Analysis 269
19.6 Puchers Stress Function 273
19.7 Synclastic, Developable and Anticlastic Shells 274
19.8 Membrane Theory of Synclastic Shells 274
19.9 Membrane Theory of Anticlastic Shells 279
19.9.1 Hyperbolic Paraboloid 279
19.9.2 Membrane Analysis of Conoid 289
Questions 294
References and Bibliography 295–297

This page
intentionally left
blank
Chapter
1
Introduction to Plates
Plate is a flat surface having considerably large dimensions as compared to its thickness. Common
examples of plates in civil engineering are
1. Slab in a building.
2. Base slab and wall of water tanks.
3. Stem of retaining wall.
A plate may have different shapes e.g. rectangular, triangular, elliptic, circular etc. as shown in
Fig. 1.1.
Fig. 1.1 Shapes of plates

A plate may have edge conditions like free, simply supported, fixed or elastically supported as
shown in Fig. 1.2.
Beam
Fig. 1.2 Edge conditions

2 THEORY OF PLATES AND SHELLS
In this chapter the coordinate system selected is clearly explained first and then various forces to be
considered on an element of plate are explained and sign conventions are made clear. At the end a brief
introduction is given to different theories available for the analysis of plates.
1.1 COORDINATE SYSTEMS

In the analysis of plates, cartesian coordinate system with right hand rule is used. According to this
when thumb, index finger and middle finger are stretched to show three mutually perpendicular direc-
tions, thumb indicates x-coordinate direction, index finger shows y-coordinate direction and middle
finger indicates z-coordinate direction. Figure 1.3 shows different orientation of x, y, z directions. The
equations derived for plate analysis with cartesian coordinate system with right hand rule hold good for
all these orientations. The commonly used orientation is that shown in Fig. 1.3(a), since slabs are
usually subjected to downward loads and the analyst is interested in downward deflections.
x
o x
y
z
z (b)
( a)
z y z
x y
(c ) (d )
x
Fig. 1.3 Different orientation of coordinates with right hand rule
For the analysis of circular plates, polar coordinate system shown in Fig. 1.4 may be used advanta-
geously.
Reference axis
q
dq
r
z is downward direction
dr
Fig. 1.4 Polar coordinates for circular plate

INTRODUCTION TO PLATES 3
1.2 STRESSES ON AN ELEMENT

In cartesian system an element of size dx × dy × dz is selected at a point (x, y), distance z below the
middle surface [Refer Fig. 1.5(a)]. The stresses acting on the element are shown in Fig. 1.5(b), in their
positive senses. Note that the sign convention used is that a stress on positive face in positive direction
or on negative face in negative direction is positive stress. It means the direct tensile stress is positive.
For shear stresses, the positive senses are as shown in Fig. 1.5(b).
sz
dx tzy
dy dz
tzx
sx
tyx t
xy
txz
y sy tyz
z
(a) An element at point (x, y, z) (b) Stresses on the element
Fig. 1.5 Stresses on an element
1.3 TYPES OF THEORIES OF PLATES

The theories that are available for the analysis of plates are
1. Thin plates with small deflections.
2. Thin plates with large deflections and
3. Thick plates.
1.3.1 Theory of Thin Plates with Small Deflections
1
This theory is satisfactory for plates with thickness less than th of its lateral dimension and having
20
1
deflection less than th of its thickness. In this theory the following three assumptions are made:
5
1. Points on the plate lying initially on a normal to the middle surface of the plate remain on the
normal to the middle surface of the plate even after bending.
2. The normal stresses in the direction transversal to the plate can be neglected i.e. Take σz, τxz,
τyz = 0.
3. There is no deformation in the middle surface of the plate. This plane remains neutral during
bending.
Assumption 1 means shear deformations are neglected. This assumption is generally satisfactory, but
in some cases e.g. in case of holes in the plate, the effect of shear becomes considerable and hence
corrections to the theory of thin plates are to be applied.
Assumption 2 is valid for thin plates, since the stresses are zero in z-direction at top and bottom of
plates, as they are free edges. There may be small variation inside the plate at any depth z, but it is
negligible.
Assumption 3 holds good if the plate is thin. However in actual structure when the plate bends, small
forces may develop in the middle surface. This inplane stress in the middle of plate reduces the bending
moment at any other point. Hence neglecting this force is an assumption on safer side.
1.3.2 Theory of Thin Plates with Large Deflections
If the deflections are not small in comparison with its thickness, strains and stresses are introduced in
the middle surface of the plate. These stresses are to be considered in deriving equilibrium equations.
Inclusion of these stresses results into non-linear equations. This is called geometric nonlinearity. When
this non-linearity is considered, the solution becomes more complicated.
1.3.3 Theory of Thick Plates
The first two theories discussed above become unrealistic in the case of plates of larger thicknesses,
especially in the case of highly concentrated loads. In such cases thick plate theory should be used. This
theory considers analysis as a three dimensional problem of elasticity. The analysis becomes lengthy
and more complicated. Till today the problems are solved only for a few particular cases.
QUESTIONS
1. Draw an element of plate in Cartesian system and show the stresses acting on it in their positive senses. Make
the sign convention clear.
2. Briefly write on the following theories of plates to bring out differences among them.
(a) Thin plates with small deflections.
(b) Thin plates with large deflections.
(c) Thick plates.
Chapter
2
Pure Bending of Plates
As the title suggests, in this theory stress resultants produced due to bending moments only are consid-
ered. In other words, deformation of the membrane due to external loads is ignored. Naturally, in this
type of bending, middle surface remains neutral surface. In this chapter, some of the properties of bent
surface are discussed and expressions are derived for stresses and moments in terms of single unknown
deflection ‘w’.
2.1 SLOPE IN SLIGHTLY BENT PLATE

Figure 2.1(a) shows the plan view of an element and Fig. 2.1(b) shows sectional view of slightly bent
plate.
o x
dx
m (x, y)a n
a dy
a¢
Fig. 2.1 (a) Plan view of element
middle surface
before deformation
m n
w w + dw
middle surface
qx after deformation
Fig. 2.1 (b) Sectional view of element

Consider an element of size dx × dy at point (x, y) in the middle surface of the plate. Figure 2.1(b)
shows the middle surface of the plate cut by plane mn parallel to xz plane. Then,
∂w
Slope along x-axis = q x = ...eqn. 2.1
∂x
Similarly if a plane parallel to yz plane is considered,
∂w
Slope along y-axis = q y = ...eqn. 2.2
∂y
Let aa′ make an angle α with x-axis (Refer Fig. 2.1(a)). The difference between the deflections at
a and a′ is due to slopes in x and y directions. Let it be ‘dw’.
Then, dw = θx dx + θy dy
∂w ∂w
= dx + dy ...eqn. 2.3
∂x ∂y
Slope along aa′ which is in ‘n’ direction is given by
dw ∂w dx ∂w dy
= +
∂n ∂x ∂n ∂y dn
∂w ∂w
= cos a + sin a. ...eqn. 2.4
∂x ∂y
Let the maximum slope be at an angle α to x-axis. Hence
∂ Ê dw ˆ
Á ˜ =0
∂a Ë dn ¯ a=a1
∂w ∂w
∂x
( - sin a1 ) +
∂y
cos a1 = 0
∂w ∂y
or tan a1 = ...eqn. 2.5
∂w ∂x
Putting eqn. 2.4 to zero, we get the direction of zero slope. Let it be α2. Then
∂w ∂w
0= cos a 2 + sin a 2
∂x ∂y
∂w ∂x
∴ tan a 2 = - ...eqn. 2.6
∂w ∂y
From eqns. 2.5 and 2.6, we get,
tan a1 ◊ tan a 2 = -1.
α1) are at right
α2) and the direction of maximum slope (α
It means the direction of zero slope (α
angles to each other.
PURE BENDING OF PLATES 7
Expression for Maximum Slope:

The value of maximum slope
∂w
=
∂n a=a1
∂w ∂w
= cos a1 + sin a1
∂x ∂y
Ê ∂w ∂w ˆ
=Á + ◊ tan a1 ˜ cos a1
Ë ∂x ∂y ¯
Ê ∂w ∂w ∂w ∂ y ˆ 1
=Á +
Ë ∂x ∂y ∂w ∂x ˜¯ sec a1
2 2
Ê ∂w ˆ Ê ∂w ˆ
ÁË ˜ +Á ˜
∂x ¯ Ë ∂y ¯ 1
=
∂ w ∂x 1 + tan 2 a1
2 2
ÁË ˜¯ + Á ˜
∂x Ë ∂y ¯ 1
=
∂w ∂x 2
Ê ∂w ∂y ˆ
1+ Á
Ë ∂w ∂x ˜¯
2 2
Ê ∂w ˆ Ê ∂w ˆ ∂w
ÁË ˜¯ + Á ˜
∂x Ë ∂y ¯ ∂x
=
∂w ∂x 2 2
ÁË ˜¯ + Á ˜
∂x Ë ∂y ¯
2 2
= Á ˜ +Á ˜ ...eqn. 2.7
Ë ∂x ¯ Ë ∂y ¯
2.2 CURVATURE OF SLIGHTLY BENT PLATE

The curvature of a bent surface is numerically equal to the rate of change of slope. If the curvature is
considered positive when it is a sagging surface (Refer Fig. 2.2), the curvature in x-direction
1 ∂ Ê ∂w ˆ ∂2w ...eqn. 2.8

= =- Á ˜ =- 2
rx ∂x Ë ∂ x ¯ ∂x
1
where is radius of curvature.
rx
Similarly, curvature in y-direction

1 ∂2 w q1
= =- 2 . ...eqn. 2.9
ry ∂y q2
Fig. 2.2 Positive sense of curvature
Curvature in any direction

Consider a direction n which makes angle α with x-axis. Then from the definition of curvature,
1 ∂2w
=- 2
rn ∂n
∂w ∂w ∂w
From eqn. 2.3, = cos a + sin a
∂n ∂x ∂y
∂ ∂ ∂
i.e. = cos a + sin a
∂n ∂x ∂y
1 ∂2 w ∂ Ê ∂w ˆ
∴ =- 2 =- Á ˜
rn ∂n ∂ n Ë ∂n ¯
Ê ∂ ∂ ˆ Ê ∂w ∂w ˆ
= - Á cos a + sin a ˜ ÁË ∂x cos a + ∂y sin a ˜¯
Ë ∂x ∂y ¯
È ∂2 w ∂2 w ∂2w ∂2w ˘
= - Í 2 cos2 a + 2 sin 2 a + cos a sin a + sin a ◊ cos a ˙
ÍÎ ∂x ∂y ∂x∂y ∂x ∂ y ˙˚
1 ∂2 w
Noting that, =- 2
rx ∂x
1 ∂2 w
and =- 2
ry ∂y
1 ∂2 w
and taking = , we get
rxy ∂x∂y
1 1 1 2∂ 2 w
= cos 2 a + sin 2 a - sin a cos a
rn rx ry ∂x∂y
1 1 + cos2 a 1 1 - cos2 a 1
= + - sin 2 a
rx 2 ry 2 rxy
1 Ê 1 1 ˆ Ê 1 1 ˆ cos 2 a 1 ...eqn. 2.10

= + + - - sin 2 a
2 ÁË rx ry ˜¯ ÁË rx ry ˜¯ 2 rxy
If ‘t’ is the direction at right angles to ‘n’ direction, the direction ‘t’ is at α + 90° to n-direction. Hence
1
can be obtained by changing α to α + 90 in eqn. 2.9. Thus
rt
1 1Ê 1 1ˆ 1Ê 1 1ˆ 1
= + + - cos (2 a + 90 ) - sin 2 (a + 90)
Á ˜ Á ˜
rt 2 Ë rx ry ¯ 2 Ë rx ry ¯ rxy
1Ê 1 1ˆ 1Ê 1 1ˆ 1
= + - - cos2 a + sin 2 a
2 ÁË rx ry ˜¯ 2 ÁË rx ry ˜¯ rxy ...eqn. 2.11
Adding eqns. 2.10 and 2.11, we get

1 1 1 1
+ = + ...eqn. 2.12
rn rt rx ry
Hence we can conclude, the sum of curvatures in any two mutually perpendicular directions in
a slightly bent plate is constant.
Twist of the surface w.r.t. n and t directions:
It is given by,
1 ∂2w ∂ Ê ∂w ˆ
= = Á ˜
rnt ∂n∂t ∂n Ë ∂t ¯
Now from eqn. 2.4,
∂w ∂w ∂w
= cos a + sin a
∂n ∂x ∂y
∂ ∂ ∂
∴ = cos a + sin a
∂n ∂x ∂y
∂
Since t is the direction at right angles to n, we get from the above expression by changing α to
∂t
α + 90.
∂ ∂ ∂
i.e. = cos (a + 90 ) + sin (a + 90)
∂t ∂x ∂y
∂ ∂
=- sin a + cos a
∂x ∂y
1 ∂ Ê ∂w ˆ
∴ = Á ˜
rnt ∂n Ë ∂t ¯
Ê ∂ ∂ ˆ Ê ∂w ∂w ˆ
= Á cos a + sin a ˜ Á - sin a + cos a ˜
Ë ∂x ∂y ¯ Ë ∂x ∂y ¯
∂2 w ∂2w ∂2 w ∂2w
=- sin a cos a + sin a cos a + cos 2 a - sin 2 a
∂x 2 ∂y 2 ∂x∂y ∂x∂y
Ê 1 1 ˆ sin 2a 1 ( 2
= - + cos a - sin 2 a )
ÁË r r ˜¯ 2 r
x y xy
Ê 1 1 ˆ sin 2a 1
= - + cos2 a
ÁË r ry ˜¯ 2 rxy ...eqn. 2.13
x
2.3 PRINCIPAL CURVATURE

1
The two mutually perpendicular directions ‘n’ and ‘t’ with respect to which twist of the surface =0,
rnt
is called the direction of principal curvatures. Hence from eqn. 2.13, we get the direction of principal
curvature ‘α’ as
1
2¥
rxy
tan 2 a = - ...eqn. 2.14
Ê 1 1ˆ
-
ÁË rx ry ˜¯
It can be shown that in the direction of principal curvatures, the curvature is maximum/minimum.
For this proof, differentiate eqn. 2.10 with respect to α. It gives,
1Ê 1 1ˆ
- ( - sin 2a ) ¥ 2 - 1 2 cos2a = 0
2 ÁË rx ry ˜¯ rxy
1
2¥
rxy
tan 2 a = - ,
i.e. Ê 1 1ˆ
-
ÁË rx ry ˜¯
which is same as eqn. 2.14.
Thus we find the planes of principal curvatures are the planes of extreme curvatures also.
Magnitude of Principal Curvatures
For such planes,
1
2
rxy 1/2
tan 2 a = - 2 4
1 1 1 +
2
- r xy
rx ry 1 - ry
rx 2
-
rxy
Referring to Fig. 2.3
2a 1
1 1
-
rx ry
Fig. 2.3
1 1 1
-2 -
rxy rx ry
sin 2 a = 12
and cos 2 a = 12
ÈÊ 1 1 ˆ 2 4 ˘ ÈÊ 1 1 ˆ 2 4 ˘
Í - + 2 ˙ Í - + 2 ˙
ÍÎ ÁË rx ry ˜¯ rxy ˙˚ ÍÎ ÁË rx ry ˜¯ rxy ˙˚
1
Substituting these values in eqn. 2.10 and noting this as , we get
r1
1 1 Ê 1ˆ
- -2
ÁË rxy ˜¯
1 1Ê 1 1ˆ Ê 1 1ˆ 1 rx ry 1
= + + - -
Á ˜ Á ˜
r1 2 Ë rx ry ¯ Ë rx ry ¯ 2 È 12 12
4 ˘ ÈÊ 1 1 ˆ 2 4 ˘
2 rxy
Ê 1 1ˆ
Í - + 2 ˙ Í - + 2 ˙
Ár ry ˜¯ Ár ry ˜¯
ÎÍ Ë x rxy ˙˚ ÎÍ Ë x rxy ˙˚
2
Ê 1 1ˆ 1
- +4 2
ÁË rx ry ˜¯
1Ê 1 1ˆ 1 rxy
= + +
Á ˜
2 Ë rx ry ¯ 2 È 12
4 ˘
2
Ê 1 1ˆ
Í - + 2 ˙
ÍÎ ÁË rx ry ˜¯ rxy ˙˚
12
1 Ê 1 1 ˆ 1 ÈÊ 1 1 ˆ 4 ˘
2
= + + Í - + 2 ˙
2 ÁË rx ry ˜¯ 2 Í ÁË rx ry ˜¯ rxy ˙˚
...eqn. 2.15(a)
Î
1/2
If we take (Ref. Fig. 2.4) 4
2
2
1 + r xy
Ê 1 1ˆ 1 - ry
- -
ÁË rx ry ˜¯ rx 2
cos 2 a 2 = 12
rxy
ÈÊ 1 1 ˆ 2 4 ˘
Í - + 2 ˙ 2a2
ÍÎ ÁË rx ry ˜¯ rxy ˙˚ -
1
-
1
rx ry
2 rxy Fig. 2.4

and sin 2a 2 = 12
ÈÊ 1 1 ˆ 2 4 ˘
Í - + 2 ˙
ÍÎ ÁË rx ry ˜¯ rxy ˙˚
we get,
12
1 1 Ê 1 1 ˆ 1 ÈÊ 1 1 ˆ 4 ˘
2
= - - Í - + 2 ˙
r2 2 ÁË rx ry ˜¯ 2 Í ÁË rx ry ˜¯ rxy ˙˚
...eqn. 2.15(b)
Î
It may be noted that 2α1 and 2α2 differ by 180° i.e. α1 and α2 differ by 90°.
The equations for principal curvatures are similar to those obtained for finding principal stresses.
Hence Mohr’s circle can be used to determine principal curvatures also. Figure 2.5 shows Mohr’s circle
for principal curvatures.
Twist
1
rxy
1, 1
rx rxy
Curvature
1 1 1
r2 2a r1 r
1, 1
ry rxy
Fig. 2.5 Mohr’s circle for curvatures
2.4 DISPLACEMENT—STRAIN RELATIONS

Let,
u — displacement in x-direction
v — displacement in y-direction and
w — displacement in z-direction.
u,v and w are considered positive when they are in positive directions of the coordinates x, y and z.
¶u
u+ dx
u ¶x
a b
v f2
a¢
¶v
v+ dx
¶x
b¢
f1
c
d
¶u¢
v+ dy
¶y
c¢
¶u
u+ dy
¶y
d¢
Fig. 2.6 Deflected shape of element

Consider an element of size dx × dy as shown in Fig. 2.6. Note that the original element abcd is
deflected to the position a′b′c′d′. In the figure a′b′c′d′ is shown such that the deflections increase in the
increasing directions of coordinates. Thus
(u, v) are deflections at a′.
Ê ∂u ∂v ˆ
ÁË u + dx, v + dx ˜ are deflections at b′.
∂x ∂x ¯
Ê ∂u ∂v ˆ
ÁË u + ∂y dy, v + ∂y dy˜¯ are deflections at c′.
∂u ∂u
∴ Change in length of the element in x-direction = u + dx - u = dx.
∂x ∂x
∴ Strain in x-direction
Change in length in x -direction
ex =
Original length in x -direction
∂u
dx
∂ x ∂u
= =
dx ∂x
Similarly, strain in y-direction
∂v
v+ ◊ dy - v
∂y ∂v
ey = =
dy ∂y
Shearing strain
g xy = f1 + f 2
∂v ∂u
◊ dx ◊ dy
∂ x ∂y
= +
dx dy
∂u ∂v
= +
∂y ∂x
Thus strains are given by the expressions
∂u
ex =
∂x
∂v
ey =
∂y
∂u ∂v
and g xy = + ...eqn. 2.16
∂y ∂x
2.5 STRAINS IN TERMS OF w

Let a point at distance z from middle surface be displaced as shown in Fig. 2.7.
rotated normal
z qx qx
u rotated middle surface
Fig. 2.7 Displacement u in terms of w
From the figure, it is clear that

∂w
u = - zq x = - z
∂x
∂w
Similarly, v = - zq y = - z
∂y
∂u ∂2 w
∴ ex = = -z 2
∂x ∂x
∂v ∂2 w
ey = = -z 2
∂y ∂y
∂u ∂v
and g xy = +
∂y ∂x
∂2 w ∂2 w
= -z -z
∂x∂y ∂x∂y
∂2w
= -2 z
∂x∂y
Thus,
∂2 w
e x = -z
∂x 2
∂2w
e y = -z
∂y 2
∂2 w
and g xy = -2 z ...eqn. 2.17
∂x∂y
The above expressions for strains εx and εy may be derived considering the
curvatures also.
Referring to Fig. 2.8,
qx
Final length - Original length
ex =
Original length
rn
=
(rx + z ) q - rx q
rx q
z ∂2w
= = -z 2 z
rx ∂x
∂2w
Similarly e y = -z .
∂y 2 Fig. 2.8 Strain εx from
curvature consideration
2.6 STRAIN-STRESS RELATIONS

Consider the element shown in Fig. 2.9, which is subjected to stresses σx, σy and τxy. In the figure
stresses are shown in their positive senses.
Taking moment equilibrium condition about z-axis passing through A, we get
τxy dy h dx − τyx h dx dy = 0, where h is thickness of plate.
∴ τxy = τyx ...(1)
sy
A
B
sx sx
txy
D tyx C
sy
Fig. 2.9 Stresses on an element

From theory of elasticity, we know
( )
1
ex = s - ms y ...(2)
E x
( )
1
and ey = s - ms x ...(3)
E y
where E - Young’s modulus
and µ - Poisson’s ratio.
( )
1
From eqn. 3, me y = m s y - ms x ...(4)
E
Adding eqns. (2) and (4), we get
e x + me y =
1
E
(
1 - m2 s x )
Ez Ê ∂ 2 w ∂2 w ˆ
(e x + me y ) = - 1 - m 2 ÁË ∂x 2
E
or sx = +m ˜
1 - m2 ∂y 2 ¯
Ez Ê ∂ 2 w ∂ 2 w ˆ
Similarly, sy = - Ám + ˜
1 - m 2 Ë ∂l 2 ∂y 2 ¯
t xy = G g xy
E
where G = Modulus of rigidity =
2 (1 + m )
E
∴ t xy = g
2 (1 + m ) xy
E ∂2w
=- 2z
2 (1 + m ) ∂x∂y
Ez ∂2w
=- ◊
(1 + m ) ∂x∂y
Ez (1 - m ) ∂ 2 w
=-
1 - m 2 ∂x ∂ y
Thus,
Ez Ê ∂ 2 w ∂2 w ˆ
sx = - Á + m ˜
1 - m 2 Ë ∂x 2 ∂y 2 ¯
Ez Ê ∂ 2 w ∂ 2 w ˆ
sy = - Ám + ˜ ...eqn. 2.18
1 - m 2 Ë ∂x 2 ∂y 2 ¯
(1 - m ) ∂
2
Ez w
and t xy = -
1- m 2 ∂x∂y
2.7 EXPRESSIONS FOR MOMENTS

Let,
Mx = Moment per unit length acting in x-direction.
My = Moment per unit length acting in y-direction.
and Mxy = Twisting moment per unit length w.r.t. x − y directions.
Sign convention used: Positive forces in positive side of z-coordinate produce +ve moments.
It amounts to taking sagging bending moments as positive moments. These moments may be repre-
sented by any one of the way shown in Fig. 2.10.
Note carefully:
Mxy is moment per unit length on face x in y-direction.
Myx is moment per unit length on face y in x-direction.
Myx
My
Mx Mxy
Mxy Mx
My
Myx
(a)
Myx
My
Mx
Mx
Mxy
Mxy
My
Myx
( b)
Myx
My
Mx
Mxy
Mxy
Mx
My
Myx
(c )
Fig. 2.10 Sign convention for moments
The twisting moments are positive when they are as shown in Fig. 2.10 (Positive shear acting in
positive direction of z produces positive twisting moment).
h2
Now, Mx = Ú s z z ¥ 1 ¥ dz
-h 2
Ez Ê ∂ 2 w ∂2w ˆ
h2
= Ú - Á
1 - m 2 Ë ∂x 2
+ m ˜ z dz
∂y 2 ¯
-h 2
h2
Ê ∂2 w
E ∂2 w ˆ È z3 ˘
=- Á + m ˜Í ˙
1 - m 2 Ë ∂x 2 ∂y 2 ¯ Î 3 ˚ - h 2
Eh 3 Ê ∂2 w ∂2w ˆ
=- Á + m ˜
(
12 1 - m 2 ) Ë ∂x 2 ∂y 2 ¯
Ê ∂2 w ∂2w ˆ
= -D Á 2 + m 2 ˜
Ë ∂x ∂y ¯
Eh 3
where D =
( )
is flexural rigidity of plate.
12 1 - m 2
Ê ∂2 w ∂2 w ˆ
Similarly My = -D Á m 2 + 2 ˜
Ë ∂x ∂y ¯
h2
Now, M xy = Ú t xy z ¥ 1 ¥ dz
-h 2
- Ez (1 - m ) ∂ 2 w
h2
= Ú 1 - m 2 ∂x ∂y
z dz
-h 2
Eh 3 (1 - m ) ∂ 2 w
=-
(
12 1 - m 2 ∂x∂y )
∂2w
= - D (1 - m ) .
∂x∂y
Thus,
Ê ∂2w ∂2w ˆ
Mx = -D Á 2 + m 2 ˜
Ë ∂x ∂y ¯
Ê ∂2 w ∂2 w ˆ
My = -D Á m 2 + 2 ˜ ...eqn. 2.19
Ë ∂x ∂y ¯
∂2w
M xy = - D (1 - m ) .
∂x∂y
2.8 MOMENT IN ANY DIRECTION

Let ‘n’ be the direction making angle α in clockwise direction to x-direction as shown in Fig. 2.11.
Consider the element of size dx × dy and thickness h. Now we have to find expression for moment in
n-direction in terms of known moments Mx, My and Mxy.
0 x
dx
a dy
Fig. 2.11 Element considered

Now consider the stresses acting on the triangular element as shown in Fig. 2.12.
dx
sy
txy x
- a
a 90
dy sx
a
dt
sn
t nt
y
Fig. 2.12 Triangular element with stresses on its sides
Noting that τxy = τyx,

∑ Forces in n direction = 0, gives
σn h dt = σx h dy cos α + σy h dx sin α + τxy h dy sin α + τxy h dx cos α
Throughout dividing by h.dt and noting that
dx dy
= sin a and = cos a, we get
dt dt
s n = s x ◊ cos2 a + s y sin 2 a + t xy cos a sin a + t xy sin a ◊ cos a
= s x cos2 a + s y sin 2 a + 2 t xy sin a ◊ cos a
Ê 1 + cos 2 a ˆ Ê 1 - cos 2 a ˆ
= sx Á ˜¯ + s y ÁË ˜¯ + t xy sin 2 a
Ë 2 2
sx + sy sx - sy
= + cos 2a + t xy sin 2 a
2 2
From equation of equilibrium
∑Ft = 0, we get
t nt hdt = -s x hdy sin a + s y hdx cos a
-t xy hdx sin a + t xy hdy cos a
∴ t nt = -s x cos a sin a + s y sin a ◊ cos a + t xy (cos 2 a - sin 2 a )
-s x + s y
= sin 2 a + t xy cos 2 a
2
h2
∴ Mn = Ú s n ¥ 1 ¥ dz ¥ z
-h 2
h2
Ê sx + sy sx - sy ˆ
= Ú ÁË
2
+
2
cos 2 a + t xy sin 2 a ˜ z dz
¯
-h 2
Mx + My Mx - My
= + cos 2a + M xy sin 2a ...eqn. 2.20
2 2
h2
Similarly, M nt = Ú t xy ¥ 1 ¥ dz ¥ z
-h 2
-Mx + My
= sin 2 a + M xy cos 2a ...eqn. 2.21
2
2.9 PRINCIPAL MOMENTS

The planes on which twisting moment is zero are known as principal planes of moment. From eqn.
2.21, if α1 is the direction of principal planes, we get
-Mx + My
0= sin 2 a1 + M xy cos 2 a1
2
2 M xy
i.e. tan 2 a1 = ...eqn. 2.22
Mx - My
It can be easily proved that moments on principal planes have extreme values. For moment Mn to
have extreme value, necessary condition is,
∂M n
=0
da a=a ¢
Mx - My
i.e. ( -2 sin 2a ¢ ) + M xy 2 cos 2a ¢ = 0
2
2 M xy
i.e. tan 2 a ¢ = which is similar to eqn. 2.22.
Mx - My
Thus α1 = α′ i.e. the moments on principal planes are extreme values. It can be shown that the
magnitude of principal moments are
12
Mx + My ÈÊ M - M ˆ 2 ˘
± ÍÁ 2 ˙
x y
M1,2 = ˜ + M xy
2 ÎÍ Ë 2 ¯ ˚˙
The expressions are similar to the expressions for principal stresses. Hence for moments also Mohr’s
circle can be drawn. Figure 2.13 shows Mohr’s circle for moments.
Twisting moment
Mxy, max
(Mx, Mxy)
2a1
0 Moment
M1 M2
(My, Mxy)
Mxy, max
Fig. 2.13 Mohr’s circle for moments
QUESTIONS
1. Prove that in a slightly bent plate under pure bending, the directions of maximum slope and zero slope are at
right angles to each other. Find the expression for maximum slope.
2. Prove that the sum of curvatures in any two mutually perpendicular directions in a slightly bent plate is
constant.
3. Derive the expression for curvature in a slightly bent plate in any direction under pure moment and then
determine the expressions for
(a) principal curvature direction.
(b) values of principal curvatures.
4. Show that planes of principal curvatures are the planes of extreme curvatures also.
5. Derive the expressions for strains in a plate in terms of single displacement ‘w’.
6. State the strain-stress relations and strain-displacement relations in terms of single displacement ‘w’. Hence
establish the expressions for stresses and moments in terms of ‘w’.
7. Derive the expression for moment in any direction and then determine:
(a) Principal planes for moment.
(b) Value of principal moments.
8. At a point in a plate, the moments are as shown below:
Mx = 90 kn-m My = 50 kn-m Mxy = 30 kn-m
Determine
(a) Direction of principal planes.
(b) Maximum/Minimum moments.
(c) Maximum twisting moment and its direction.
Chapter
3
Small Deflections of Laterally
Loaded Plates
Plates are usually subjected to lateral loads and bend in both directions. The bending moment and shear
forces vary from point to point. Hence, the moments and shear forces on negative face and positive face
of an element will not be same. However, the element will be in equilibrium under the action of these
stress resultants and the load on it. In this chapter, the equilibrium equation is derived and the boundary
conditions to be considered are discussed.
3.1 STRESS RESULTANTS ON A TYPICAL PLATE ELEMENT

Figure 3.1 shows a typical element and the stress resultant on it. Stress resultants on negative faces (at
x and y sections) are noted without superscripts while the stress resultants on positive faces (at x + dx
and y + dy) are shown with ‘+’ sign as superscript.
a
0
y x
x dx
dy
b
y
Fig. 3.1 (a) Position of element
Consider the moments in x-direction.

on negative face moment = Mx
on positive face moment = Mx+
∂M x
If the rate of change of moment in x-direction is , then
∂x
∂M x
M x+ = M x + dx
∂x
Qy
Myx
Qx
My
Mx
+
Mx +
Mxy
+
Mxy My
+
Qx
+
+
Myx Qy
Fig. 3.1 (b) Moments on element (c) Vertical shears on element
Similarly the relations of stress resultants on positive faces with negative faces may be written. Thus,
∂M x ∂M y ¸
M x+ = M x + dx M y+ = M y + dy Ô
∂x ∂y Ô
+
∂M xy +
∂ M yx Ô
M xy = M xy + dx M yx = M yx + dy ˝ ...eqn. 3.1
∂x ∂y Ô
∂Q x ∂Q y Ô
Qx+ = Q x + dx +
Qy = Q y + dy Ô
∂x ∂y ˛
Note that all stress resultants are per unit length.
Apart from these stress resultants, load of intensity q per unit area is acting on the element in the
downward direction. Hence, total downward load on the element.
= q dx dy ...eqn. 3.2
3.2 EQUATIONS OF EQUILIBRIUM

Three independent equations of equilibrium can be written for the element—one considering the forces
in z-direction and two moment equilibrium for the moments in x-direction and y-direction
Consider ∑Fz = 0. It gives
Qx+ dy − Qx dy + Qy+ dx − Qy dy + q dx dy = 0
Ê ∂Q x ˆ Ê ∂Q y ˆ
i.e. ÁË Q x + dx ˜ dy - Q x dy + Á Q y +
¯
dy ˜ dx - Q y dy + q dx dy = 0
∂x Ë ∂y ¯
Simplifying and then dividing throughout by dx dy, we get
∂Qx ∂Qy
+ +q =0 ...eqn. 3.3
∂x ∂y
Equilibrium equation,
∑Mx = 0, gives
SMALL DEFLECTIONS OF LATERALLY LOADED PLATES 25
dx dx
M x+ dy - M x dy + M yx
+
dx - M yx dx - Qx+ dy - Q x dy =0
2 2
Ê ∂M x ˆ Ê ∂M yx ˆ
ÁË M x + dx ˜ dy - M x dy + Á M yx +
¯
dy ˜ dx - M yx dx
∂x Ë ∂y ¯
Ê ∂Q x ˆ dx dx
- Á Qx + dx ˜ dy ◊ - Q x dy =0
Ë ∂x ¯ 2 2
Simplifying and neglecting small quantity of higher order, we get
∂M x ∂M yx
+ - Qx = 0
∂x ∂y
∂M x ∂M yx
i.e. + = Qx ...eqn. 3.4
∂x ∂y
Similarly moment equilibrium in y-direction, gives
∂M xy ∂M y
+ = Qy ...eqn. 3.5
∂x ∂y
Substituting the values of Qx and Qy as shown in Eqns. 3.4 and 3.5 in equation 3.3, we get
∂2 M x ∂ 2 M yx ∂ 2 M xy ∂2 M y
+ + + +q=0
∂x 2 ∂x∂y ∂x∂y ∂y 2
∂2 M x ∂ 2 M xy ∂ 2 M y
i.e. +2 + = -q
∂x ∂x∂y ∂y 2
But we know (Refer eqn. 2.20)
Ê ∂2w ∂2w ˆ
Mx = -D Á 2 + m 2 ˜
Ë ∂x ∂y ¯
Ê ∂2 w ∂2 w ˆ
My = -D Á m 2 + 2 ˜
Ë ∂x ∂y ¯
∂2w
and M xy = - D (1 - m )
∂x∂y
Hence, equation of equilibrium is
Ê ∂4w ∂4w ˆ ∂4w Ê ∂4w ∂4w ˆ
- D Á 4 + m 2 2 ˜ - 2 D (1 - m ) 2 2 - D Á m 2 2 + 2 ˜ = - q
Ë ∂x ∂x ∂y ¯ ∂x ∂y Ë ∂x ∂y ∂y ¯
Ê ∂4w ∂4w ∂4w ˆ

i.e. - D Á 4 + 2 2 2 + 4 ˜ = - q.
Ë ∂x ∂ x ∂y ∂y ¯
∂4w ∂4w ∂4w q

i.e. +2 + =
∂x 4
∂x ∂ y
2 2
∂y 4 D
Ê ∂2 ∂2 ˆ Ê ∂2 w ∂2 w ˆ q
i.e. Á 2 + ˜Á + ˜=
Ë ∂x ∂y 2 ¯ Ë ∂x 2 ∂y 2 ¯ D
∂2 ∂2
Denoting + by ∇ we get
∂x 2 ∂y 2
— 2 (— 2 w ) =
q
D
q
or —4w = ...eqn. 3.6
D
Equation 3.6 is known as Equation of Plates or Lagrange Equation for Plates.
3.3 EXPRESSIONS FOR VERTICAL SHEARS

From equation 3.4,
∂M x ∂M xy
Qx = +
∂x ∂y
Ê ∂2w ∂2w ˆ
But Mx = -D Á 2 + m 2 ˜
Ë ∂x ∂y ¯
∂2w
and M xy = - D (1 - m )
∂x∂y
Ê ∂3w ∂3w ˆ ∂3 w
∴ Qx = - D Á 3 + m ˜ - D (1 - m )
Ë ∂x ∂x∂y 2 ¯ ∂x ∂ y 2
Ê ∂3 w ∂3 w ˆ
= -D Á 3 + ˜
Ë ∂x ∂ x ∂y 2 ¯
∂ Ê ∂2w ∂2w ˆ
= -D + m
∂x ËÁ ∂x 2 ˜
∂y 2 ¯
∂ ( 2
= -D — w)
∂x
Similarly
∂ ( 2
Qy = - D — w)
∂y
∂ ( 2
Thus Qx = - D — w)
∂x
∂ ( 2
Qy = - D — w) ...eqn. 3.7
∂y
3.4 BOUNDARY CONDITIONS

(a) Fixed Edge: If the edge x = a is fixed (Ref. Fig. 3.2)
∂w
w = 0 and =0 ...eqn. 3.8
∂x
(b) If edge x = a is simply supported.
w x =a = 0
and Mx x=a
=0
Ê ∂2 w ∂2 w ˆ
i.e. -D Á 2 + m 2 ˜ = 0
Ë ∂x ∂y ¯
∂2 w
But along x = a, = 0, since the edge is supported throughout.
∂y 2
Hence, the boundary condition is
∂2 w
= 0.
∂x 2 x =a
Thus, the boundary conditions are
∂2 w
w x = a = 0 and =0 ...eqn. 3.9
∂x 2
(c) Free edge at x = a
If the edge x = a is free,
Mx = 0, Mxy = 0 and Qx = 0.
The above three boundary conditions were expressed by Poisson. But Kelvin felt that there is some-
thing wrong, since when at all other edges two conditions are found, how there can be three edge
conditions in this case.
Kelvin and Tait pointed out that the last two conditions are not independent. They can be combined
to give a single realistic condition.
Referring to Figure 3.2, twisting moment on an elemental length dy is Mxy dy. This moment may be
replaced by two vertical shears of magnitude Mxy separated by dy. This change will not alter the be-
haviour of the plate. Kelvin and Tail pointed out that the actual boundary condition is at any point on
free edge, vertical shear plus vertical shear due to replacement of twisting moment must be zero. Now,
vertical force at any point due to Mxy
+
= M xy - M xy
∂M xy ∂M xy
= M xy + dy - M xy = dy, downward.
∂y ∂y
Mxy
+
Mxy
Mxy
+
Mxy
Fig. 3.2 Contribution of twisting moment to vertical shear
∴ Total vertical shear

∂M xy
Vx dy = Qx dy + dy
∂y
∂M xy
∴ Vx = Q x +
∂y
È ∂ Ê ∂2 w ∂2 w ˆ ˘ ∂3w
= - D Í Á 2 + 2 ˜ ˙ - D (1 - m )
ÍÎ ∂x Ë ∂x ∂y ¯ ˙˚ ∂x∂ y 2
È ∂3w ∂3 w ˘ ∂3w
= -D Í 3 + ˙ - D (1 - m )
ÍÎ ∂x ∂x∂y 2 ˙˚ ∂ x∂ y 2
È ∂3w ∂3w ˘
= - D Í 3 + (2 - m ) ˙
ÎÍ ∂x ∂x∂y 2 ˚˙
The real boundary condition is
Vx = 0
Thus, at free edge the boundary conditions are, M x x=a
= 0 and Vx x=a
=0 ...eqn. 3.10
(d) If edge x = a is elastically supported.
Figure 3.4 shows this case, in which edge x = a is supported elastically by a beam. Let flexural
rigidity of beam be B and torsional rigidity be G.
We know, for the beam,
∂4w
B = Load intensity
∂y 4
One boundary condition is,

deflection of beam = deflection of plate
∂4w
B =V x =a
∂y 4 x =a
È ∂3 w ∂3 w ˘
= - D Í 3 + (2 - m ) ˙
ÎÍ ∂x ∂x∂y 2 ˚˙ x = a
The second boundary condition may be written by considering the torsional rotation of elemental
length of beam. Referring to Figure 3.3,
Let MTb be twisting moment in the beam. Twisting of the
beam is due to moment M x x =a
in the plate. From moment
MTb
equilibrium condition, we get,
+
M Tb dy - MTb dy = M x dy
∂M Tb
i.e. M Tb dy + dy + MTb dy = Mxdy
∂y Mx
∂M Tb
= -Mx
∂y
∂ Ê ∂2w ˆ Ê ∂2w ∂2w ˆ dy

Á G◊ ˜ = DÁ 2 + m 2 ˜,
∂y Ë ∂x∂y ¯ Ë ∂x ∂y ¯
since w is same for both beam and plate.
∂3 w Ê ∂2w ∂2 w ˆ
i.e. G = DÁ 2 + m 2 ˜
∂x∂y Ë ∂x ∂y ¯
Thus, for plate with elastic support on beam, the boundary +
MTb
conditions are,
∂4w ∂3 w ∂3 w
B = + (2 - m ) Fig. 3.3 Torsional rotation of edge beam
∂y 4 ∂x 3 ∂x ∂ y 2 at x = a
...eqn. 3.11
∂3w Ê ∂3w ∂2w ˆ
and G = D Á + m ˜
∂x∂y 2 Ë ∂x 2 ∂y 2 ¯
3.5 CORNER REACTION

Consider the corner x = a and y = b in which the edges x = a and y = b are discontinuous. If we resort
to replacing twisting moment in the last element along the edge x = a, we find there is an upward force
∂M xy
of , corner. Similarly if we replace twisting moment in the last element along y = b, we find there
∂y
∂M yx ∂M xy
is an upward force of , corner. Thus, at corner x = a, y = b, there is net upward force of
∂x ∂y
∂M yx
corner plus corner. Since Mxy = Myx, we can conclude at the corner there is net upward force of
∂x
∂M xy
2 . Hence, at discontinuous corners, lifting takes place. To take care of this phenomenon, in
∂x
R.C.C. slabs corner reinforcements are provided.
a
x
2Mxy, corner
¶Mxy , corner
y ¶y
¶Myxz
, corner
¶x
Fig. 3.4 Corner reaction
QUESTIONS
1. Derive the biharmonic equation for slightly bent thin plate.

2. Discuss the boundary conditions at free edge.
3. Explain why there is tendency of uplift at corner if two adjacent edges of a plate are discontinuous.
Chapter
4
Fourier Series of Loadings
It is difficult to solve the biharmonic equation of the plate, if the loading is considered as it is. However
satisfactory solution can be obtained easily if loading is expressed in a series of equivalent sinusoidal
loading form. Such loading is called Fourier series loading. Though the loading is expressed in the
infinite series, first term itself gives 95 percentage of deflection and 90 percentage of moment. Hence,
Fourier series solutions are used in the analysis of plates and shells.
4.1 FOURIER SERIES OF LOADS ON BEAMS

A given loading is considered as a sum of series of sinusoidal loadings. The peak values of sinusoidal
loadings are so selected that if a number of such terms are added original loading is obtained. The
Fourier equivalent terms for a given loading are given by,
(a) Given load
(b) Ist term
(c ) IInd term
(d ) IIIrd term
(e) IVth term
(f ) Sum of I, II, III and IVth terms

Fig. 4.1
•
mpx
f ( x ) = Â a m sin where
1 L
...eqn. 4.1
2 m px
a m = Ú f ( x )sin dx.
L L
where L is span of the beam.

Figure 4.1 shows the addition of such loadings to get original loading f(x).
To get reasonably good loading, it may be necessary to take 20–25 terms in Fourier series. But
deflections can be obtained reasonably well (within 5 percent) with the first term itself. Similarly
bending moments can be obtained reasonably well with 3 terms. Hence, this approach has been suc-
cessfully used for the analysis of plates.
4.2 FOURIER SERIES FOR UDL ON BEAMS

Expression for equivalent Fourier series is derived in this article for uniformly distributed load acting
on a beam of span L (Refer Figure 4.2).
q0 /unit length
Fig. 4.2 UDL over entire span
In general for any loading q(x), Fourier equivalent is

•
m px
q ( x) = Â am sin L
m =1
L
2 mpx
L Ú0
where am = q ( x )sin dx ...eqn. 4.2
L
In case of uniformly distributed load,

q(x) = q0
L
2 mpx
∴ am =
L0Ú q0 sin
L
dx
L
2 L È mpx ˘
= Í - q0 cos
L mp Î L ˙˚ 0
2 q0
= [ - cos mp + 1]
mp
4 q0
∴ For odd values of m, am = and for even values of m, am = 0
mp
•
4 q0 m px
\ q ( x) = Â sin ...eqn. (4.3)
m =1,3, ... m p L
FOURIER SERIES OF LOADINGS 33
4.3 FOURIER EQUIVALENT FOR UDL OVER A SMALL LENGTH

Let load intensity be q over a length u with its centre of gravity at ξ acting on a beam of length L as
shown in Figure 4.3. It is required to find Fourier equivalent load for this.
In general
•
m px
q ( x) = Â am sin L
m =1
u q0 /unit length
Fig. 4.3 UDL over a small length
L
2 mpx
where am =
L0Ú q ( x )sin
L
dx
In this case,
u
q ( x) = 0 for x = 0 to x -
2
u u
= q0 for x = x - to x +
2 2
u
=0 for x = x + to L .
2
2È ˘
x- u 2 x+ u 2 L
mpx
∴ am = Í
LÍ Ú 0 dx + Ú sin
4
dx + Ú 0 dx ˙
˙˚
Î 0 x-u 2 x+ u 2
x+ u 2
2 È L mpx ˘
= ◊ q0 Í - cos
L Î mp L ˙˚ x-u 2
2 L È mp ( x + u 2 ) mp ( x - u 2 ) ˘
= q0 Í - cos + cos ˙
L mp Î L L ˚
2 q0 mpx mpu
= ◊ 2sin ◊ sin
mp L 2L
4 q0 mpx mpu
= sin ◊ sin
mp L 2L
•
4 q0 mpx m pu mpx
q ( x) = Â sin sin sin ...eqn. 4.4
m =1 mp L 2L L
4.4 FOURIER EQUIVALENT FOR A CONCENTRATED LOAD

Figure 4.4 shows a concentrated load P at distance ξ from left support acting on a beam of span L.
P
Fig. 4.4 Concentrated Load P
Its Fourier equivalent may be obtained from expression 4.4 by treating it as the case in which total
load remains the same while u tends to zero. Thus, we have to substitute
P = q0u
and u → 0.
m pu m pu
As u Æ 0, sin =
2L 2L
•
4q mpx mpu m px
∴ q (u ) = Â mp0 sin L 2L
sin
L
m =1
•
2P mpx mpx
i.e. q ( x) = Â sin ◊ sin ...eqn. 4.7
m =1 L L L
4.5 FOURIER EQUIVALENT FOR HYDROSTATIC LOAD

Figure 4.5 shows hydrostatic load varying from 0 intensity at x = 0 to q0 at x = L
qx
L q0
Fig. 4.5 Hydrostatic Load on a beam

Its Fourier equivalent is

•
m px
q ( x) = Â am sin a
m =1
L
2 mpx
where am =
L0Ú q ( x )sin
a
dx
x
In this case q ( x ) = q0
L
L
2 q0 x mpx
L Ú0 L
∴ am = sin dx
a
L
2 q0 mpx
=
L 2 Ú x sin a
dx
0
Noting that if u and v are the functions of x

du
Ú uv dx = u Ú v dx - Ú Ú v dx dx
m px
and taking u = x and v = sin , we get
a
2 q0 È m px mpx ˘
2 Í Ú
am = x sin dx - Ú Ú sin ¥ 1 ¥ dx ˙
L Î L L ˚
L
2q È L Ê mpx ˆ L2 Ê mpx ˆ ˘
= 20 Í x ÁË - cos ˜¯ - 2 2 ÁË - sin ˜˙
L Î mp L m p L ¯ ˚0
2 q0 È L2 L2 ˘
= Í - cos m p + (sin mp )˙
L Î mp
2
m p
2 2
˚
2 q0
= ( - cos mp ) , since sin mπ = 0 for all values of m
mp
2 q0
= ( -1)m +1 , since for odd values, −cos mπ = 1
mp
and for even values, − cos mπ = −1.
∴ Fourier equivalent for hydrostatic load is
•
2q m +1 m px
q( x ) = Â mp0 (-1) sin
L
. ...eqn. 4.8
m =1
4.6 FOURIER EQUIVALENT FOR TRIANGULAR LOAD WITH PEAK VALUE AT

MID-SPAN
This type of loading is shown in Fig. 4.6.
q0
2q0 x
L
Fig. 4.6 Triangular load with peak value at mid span
m px
The general form of Fourier expression is q ( x ) = Â am sin
L
L
2 mpx
L Ú0
where am = q ( x )sin .
L
Now, for x = 0 to L/2,
2q0 x
q ( x) =
L
Due to symmetry, it is obvious that
L L 2
Ú q ( x ) dx = 2 Ú q ( x ) dx.
0 0
L 2
2 2 q0 x mpx
∴ am =
L
¥2 Ú L
sin
L
dx
0
8 q0 mpx
=
L2 Ú x sin L
dx.
m px m px mpx
Noting that Ú x sin L
dx = x Ú sin
L
dx - Ú Ú sin
L
dx
L 2
È L Ê mpx ˆ L2 mpx ˘
=Íx ÁË - cos ˜¯ + 2 2 sin ˙
Î mp L m p L ˚0
È L2 mp L2 mp ˘
= Í- cos + 2 2 sin ˙
Î 2 mp 2 m p 2 ˚
L2 mp mp
= sin , since cos =0 ...eqn. 4.8(a)
m p2 2
2 2
∴
8 q0 mp
am = sin
m p 2 2 2
•
8q p mpx
∴ q ( x) = Â m 2 p0 2 sin 2 ◊ sin L
...eqn. 4.8(b)
m =1
4.7 DOUBLE FOURIER SERIES EXPRESSIONS FOR LOADS

The loads acting on a plate of size a × b (Refer Fig. 4.7) are to be expressed in Fourier series involving
sine functions in x as well as in y directions.
It is required to find
0 a
• •
m px npy x
q ( x, y ) = Â Â amn sin sin
m =1 n =1 a b
•
n py
g m ( y ) = Â a mn sin
b
Let
n =1 b
•
mpx
∴ q ( x, y ) = Â gm ( y )sin a
m =1
y
a
2 mpx Fig. 4.7 Typical plate of size a × b.
∴ gm ( y ) =
a0Ú q ( x, y )sin
a
dx
•
n py
Since, gm ( y ) = Â amn sin
n =1 b
b
2 npy
amn =
b0Ú gm ( y )sin
b
dy
b a
2 2 mpx npy
= Ú Ú q ( x, y )sin ◊ sin dx dy
b0a0 a b
ab
4 mpx npy
i.e. amn = Ú Ú
ab 0 0
q ( x, y )sin
a
◊ sin
b
dx dy ...eqn. 4.9
4.8 DOUBLE FOURIER FORM FOR UDL

For uniformly distributed load,
q(x, y) = q0
ab
4 mpx npy
∴ a mn = Ú Ú
ab 0 0
q0 sin
a
◊ sin
b
◊ dx dy
a b
4 q0 È a m px ˘ È b n py ˘
= ÍÎ - mp cos a ˙˚ ÍÎ - np cos b ˙˚
ab 0 0
4 q0
=[ - cos mp + 1] [ - cos np + 1]
mnp 2
∴ For even values of m or n, amn = 0.
4 q0 16 q0
For odd values of m and n, a mn = ¥2¥2 =
mnp 2
mnp 2
16 q0 mpx npy
∴ q ( x , y ) = ÂÂ sin ◊ sin ...eqn. 4.10
mnp 2 a b
4.9 DOUBLE FOURIER FORM FOR UDL OVER SMALL AREA

Figure 4.8 shows a plate of size a × b subject to udl q0 over a small area u × v with its centre of gravity
at (ξ, η). Now, it is required to find Fourier equivalent for this load.
a
x
h u
q0 v
b x
Fig. 4.8 UDL over a small area
For this load,

q(x, y) = 0 for x = 0 to ξ − u/2 and for x = ξ + u/2 to a
= 0 for y = 0 to η − v/2 and for y = η + v/2 to b
= q0 for x = ξ − u/2 to ξ + u/2 and y = η − v/2 to η + v/2.
ab
4 mpx npy
∴ amn = Ú Ú
ab 0 0
q ( x , y )sin
a
sin
b
dx dy
x+u 2 h+ v 2
4 mpx npy
ab x-Úu 2 Ú
= q0 sin sin dx dy
h- v 2
a b
x+u 2 h+ v 2
4 q0 È a mpx ˘ È b npy ˘
= ÍÎ - mp cos a ˙˚ ÍÎ - np cos b ˙˚
ab x-u 2 h- v 2
4 q0 È mp (x + u 2) mp ( x - u 2 ) ˘
= Í cos - cos ˙
mnp 2 Î a a ˚
È np ( h + v 2 ) np ( h - v 2 ) ˘
Í cos - cos ˙
Î b b ˚
4 q0 mpx mpu npn npv
= ◊ 2sin ◊ sin ◊ 2 sin ◊ sin
mnp 2 a 2a b 2b
• •
16 q mpu mpx n pv n pn mpx n py
∴ q ( x, y ) = Â Â mnp02 sin 2a
◊ sin
a
◊ sin
2b
sin
b
◊ sin
a
◊ sin
b
...eqn. 4.11
m =1 n =1
4.10 DOUBLE FOURIER FORM FOR CONCENTRATED LOAD

Let P be the concentrated load acting at (ξ, η) on plate of size a × b as shown in Fig. 4.9.
a
0 x
P
b x
Fig. 4.9 Concentrated Load P at (ξ, η)
Double Fourier expression for this load can be derived from eqn. 4.11 by letting
P = q0 uv
while letting u and v to tend to zero.
Since, u and v tend to zero,
mpu mpu
sin =
2a 2a
npv npv
and sin =
2b 2b
• •
16 q m pu n pv mpx n pn m px n py
∴ q ( x, y ) = Â Â mnp02 ◊ ◊
2a 2b
◊ sin
a
◊ sin
b
◊ sin
a
◊ sin
b
.
m =1 n =1
• •
4P mpx n pn mpx n py
i.e. q ( x, y ) = Â Â ab sin a
◊ sin
b
◊ sin
a
◊ sin
b
...eqn. 4.12
m =1 n =1
4.11 DOUBLE FOURIER EQUIVALENT FOR HYDROSTATIC LOAD

Let loading vary linearly from zero to q0 in x-direction and be constant in y-direction (Refer Fig. 4.10).
q0 x
a
q0
0 q0 x
x a
x
a
y
Fig. 4.10 Hydrostatic load on plate
m px n py
q ( x , y ) = ÂÂ a mn sin ◊ sin
a b
ab
4 mpx npy
ab Ú0 Ú0
where amn = q ( x, y )sin ◊ sin dx dy
a b
ab
4 x mpx npy
ab Ú0 Ú0 0 a
= q sin ◊ sin dx dy
a b
From eqn. 4.8, we know
a
2 q0 x mpx 2q
Ú
a0 a
sin
a
dx = 0 ( -1) m +1 .
mp
b
2 2 q0 npy
∴ a mn = ( -1) m +1 Ú sin dy
b mp 0
b
m +1
4 q0 ( -1)
b
È b npy ˘
= ÍÎ - np cos b ˙˚
mpb 0
m +1
4 q 0 ( -1)
= [ - cos np + 1]
mnp 2
= 0 for even values of m
m +1
8q0 ( -1)
= for odd values of n.
mnp 2
m +1
• •
8q0 ( -1) m px n py
∴ q ( x, y ) = Â Â mnp 2
sin
a
◊ sin
b
. ...eqn. 4.13
m =1 n =1,3,...
4.12 DOUBLE FOURIER EQUIVALENT FOR TRIANGULAR LOAD WITH PEAK

VALUE AT MID-SPAN
Figure 4.11 shows the loading.
q0
2q0 x
L
0 2q0 x
x a
x
Fig. 4.11 Triangular load with peak value at mid-span
In this case
2 q0 x a
q ( x, y ) = for x = 0 to
a 2
and the loading is symmetric
a a2
∴ Ú q ( x, y) dx = 2 Ú q ( x, y ) dx
0 0
ab
4 mpx npy
∴ amn = Ú Ú
ab 0 0
q ( x, y )sin
a
◊ sin
b
dx dy
a 2b
4 2 q0 x mpx npy
=
ab
¥2 ÚÚ a
sin
a
◊ sin
b
dx dy
0 0
a 2b
16 q0 mpx npy
=
a b 2 Ú Ú x sin a
◊ sin
b
dx dy
0 0
From eqn. 4.8(a), we know
mpx a2 mp
Ú x sin a
= 2 2 sin
m p 2
b
16 q0 mp npy
∴ amn =
m2p2b
sin Ú
2 0
sin
b
dy
b
16 q0 mp È b n py ˘
= sin Í - cos
m2p2b 2 Î np b ˙˚ 0
16 q0 mp
= sin ◊ [ - cos np + 1]
m 2 np 3 2
= 0 for even values of n.
32 q0 mpx
= sin for odd values of n
m np 2 3 2
• •
32 q0 mp mpx npy
∴ q ( x, y ) = Â Â m np
2 3
sin
2
sin
a
◊ sin
b
. ...eqn. 4.14
m =1 n =1,3,...
QUESTIONS
1. Derive single Fourier series equivalent for the following type of loads on beam of span L:
(a) Uniformly distributed load over entire span.
(b) Uniformly distributed load over a small length u with its centre of gravity at x = ξ.
(c) Concentrated load P at x = ξ.
(d) Hydrostatic load of maximum intensity q0 at x = L and zero intensity at x = 0.
(e) Triangular load of intensity q0 at x = L/2.
2. Derive double Fourier series expressions for the following loads:
(a) Uniformly distributed load over entire plate.
(b) Concentrated load P at (ξ, η).
Chapter
5
Navier’s Solution for
Rectangular Plates
A French engineer Navier presented a plate theory as back as in 1820. His theory holds good only for
rectangular plates, simply supported along all the four edges. His paper was the beginning of looking
for trigonometric series solutions for the analysis of plates. Navier’s method is explained in this chapter
for a simply supported rectangular plate subject to uniformly distributed load over entire plate. It is then
extended for the plates subject to any other load.
5.1 NAVIER SOLUTION FOR RECTANGULAR PLATE SUBJECTED TO UDL

The uniformly distributed load q0 may be expressed in its equivalent Fourier series load as,
• •
16q0 mpx npy
q ( x, y ) = Â Â mnp 2
sin
a
◊ sin
b
.
m =1,3,... n =1,3,...
In this problem, the solution for w should satisfy the equation,

q ∂4w ∂4w ∂4w q
—4w = i.e. +2 + = ...eqn. 5.1
D ∂x 4
∂x ∂y
2 2
∂y 4 D
and the boundary conditions
w=0 at x = 0 and x=a
w=0 at y = 0 and y=b
Mx = 0 at x = 0 and x = a, and
My = 0 at y = 0 and y = b.
• •
mpx npy
Hence, let w= Â Â Cmn sin a
◊ sin
b
...eqn. 5.2
m =1 n =1
This form of deflection w satisfies all boundary conditions. But it has to satisfy the plate equation
also. Now,
• •
∂4w m4p4 m px n py
∂x 4
= Â Â Cmn a 4
sin
a
◊ sin
b
m =1 n =1
• •
∂4w m 2 p 2 n2 p 2 m px n py
2 = 2Â Â Cmn ◊ 2 sin ◊ sin
∂x 2 ∂ y 2 m =1 n =1 a 2
b a b
• •
∂4w m px n4p4 npy
and = Â Â Cmn
∂y 4 m =1 n =1 b4
sin
a
◊ sin
b
Hence, if equivalent Fourier loading is considered, the plate equation reduces to
• • È m4p4 2m2 n2 p 4 n4p4 ˘ mpx npy
Â Â Cmn ÍÎ a4
+
a2 b2
+ 4 ˙
b ˚
sin
a
◊ sin
b
m =1 n =1
• •
1 16 q0 mpx npy
= Â
D m =1, 3, ...
Â mnp 2
sin
a
◊ sin
b
n =1,3,...
Comparing term by term, we get
Cmn = 0 for even values of m or n.
For odd values of m and n, we get
È m 4 p 4 2m2 n2 p 4 n4 p 4 ˘ 16 q0
C mn Í 4 + + 4 ˙=
Î a 2 2
a b b ˚ mnp 2 D
p4 È 4 2 2 a
2
n4a4 ˘ 16 q0
i.e. C mn Í m + 2 m n + 4 ˙
=
a Î
4
b 2
b ˚ mnp 2 D
16q0 a 4 1
∴ C mn =
mnp D Ê 6
a2 2 ˆ
2
ÁË m + 2 n ˜¯
2
b
• •
16 q0 a 4 1 1 mpx n py
∴ w=
p D
6 Â Â mn Ê 2
sin ◊ sin ...eqn. 5.3
m =1,3,... n =1, 3,... a2 2 ˆ a b
ÁË m 2
+ n ˜
b2 ¯
Ê ∂2w ∂2w ˆ
∴ Mx = -D Á 2 + m 2 ˜
Ë ∂x ∂y ¯
Ï m2p2 n2 p2 ¸
• • -Ì 2 + m 2 ˝
16 q0 a 4
1 Ó a b ˛ mpx n py
= -D
p6 D
Â Â mn Ê 2
sin sin
m =1,3,... n =1,3,... a 2ˆ
2 a b
ÁË m 2
+ n ˜
b2 ¯
a2
m 2 + mn 2
16 q0 a 2 • •
1 b 2 sin mpx ◊ sin npy
i.e. Mx = Â Â
p 4 m =1,3,... n =1,3,... mn Ê 2
...eqn. 5.4
a2 2 ˆ a b
ÁË m + 2 n ˜¯
2
b
a2
• • mm + n 2 2
2
b sin npx ◊ sin npy
2
16 q0 a 1
Similarly, My =
p 4 Â Â 2
...eqn. 5.5
m =1,3,.. n =1,3,... mn Ê 2 2 a
2ˆ a b
ÁË m + n ˜
b2 ¯
NAVIER’S SOLUTION FOR RECTANGULAR PLATES 45
Thus, expressions for all other stress resultants (Mxy , Qx, Qy , Vx, etc.) may be assembled easily.
For square plate a = b,
• •
16 q0 a 4 1 1 m px n py
w= Â Â sin sin ...eqn. 5.6
p D n =1,3,... mn ( m 2 + n 2 )
6 2 a a
m =1,3,...
• •
16 q0 a 2 1 m 2 + mn 2 mpx npy
Mx =
p4
Â Â mn (m 2 + n2 )2 sin a ◊ sin a ...eqn. 5.7
m =1,3,... n =1,3,...
• •
16 q0 a 2 1 mm 2 + n 2 mpx npy
My =
p4
Â Â sin ◊ sin ...eqn. 4.8
n =1, 3,... mn ( m + n )
2 2 2 a a
m =1,3,...
• •
16 q0 a 4 1 1 mp np
wcentre = Â Â sin sin
p D n =1,3,... mn ( m 2 + n 2 )
6 2 2 2
m =1,3,...
16 q0 a 4 È 1 1 1 1 1 ˘
= Í - - + + 2 - ....˙
p D Î 1 ¥ 1(1 + 1)
6 2
3 ¥ 1(3 + 1)
2 2
1 ¥ 3(1 + 3 )
2 2
3 ¥ 3 (3 + 3 )
2 2
˚
16 q0 a 4 È 1 1 1 1 ˘
= ÍÎ 4 - 3 ¥ 100 - 3 ¥ 100 + 9 ¥ 324 - ....˙˚
p D
6
q0 a 4
wcentre, exact = 0.00406
D
16 1 q0 a 4 q a4
The first term gives wcentre = ¥ = 0.0041606 0
p6 4 D D
0.0041606
Thus, first term gives ¥ 100 = 102.48 percent of total deflection
0.00406
In other words, first term gives only 2.48% erroneous result and it is on safer side.
Consider moment at centre of square plate taking µ = 0.
• •
16 q0 a 2 1 m2 mp np
Mx = My = Â Â mn 2 2 2 sin 2 ◊ sin 2 .
p4 m =1,3,... n =1,3,... (m + n )
1 1 16q0 a 2 1 16 q0 a 2
First term = ◊ = ◊
1 ¥ 1 (1 + 1)2 p4 4 p4
1 1 ¥ 9 16 q0 a 2 3 16 q0 a 2
Second term = - ◊ = -
3 ¥ 1 (32 + 1) 2 p4 100 p 4
1 1 16 q0 a 2 1 16 q0 a 2
Third term = - = -
1 ¥ 3 (1 ¥ 32 ) 2 p 4 300 p 4
52 16 q0 a 2 5 16 q0 a 2
Fourth term = =
5 ¥ 26 2 p4 3380 p 4
16q0 a 2
1 16 q0 a 2
12
Fifth term = =
5 ¥ 26 2 p 4 3380 p 4
It may be observed that convergence is not that fast as deflection w. Hence, to get results within
1.1 percent errors only, we have to take at least 3 terms.
5.2 NAVIER SOLUTION FOR ANY LOADING

Navier solution for simply supported rectangular plate subject to any loading can be easily found, if the
load is expressed in the double Fourier form as
• •
mpx npy
q ( x, y ) = Â Â amn sin a
◊ sin
b
m =1 n =1
If we select
• •
mpx npy
w= Â Â Cmn sin a
◊ sin
b
m =1 n =1
all boundary conditions are satisfied. Then from plate equation we get
amn a 4
C mn = 2
Ê a2 ˆ
Dp 4 Á m 2 + n 2 2 ˜
Ë b ¯
• •
amn a 4 mpx n py
w= Â Â 2
sin ◊ sin
∴ m =1,3,... n =1, 3,.., Ê a2 ˆ a b
p 4 D Á m2 + n2 2 ˜
Ë b ¯
Thus, if load is expressed in double Fourier form solution is readily available. Depending upon the
loading, rate of convergence may slightly vary.
QUESTIONS
1. Derive the expression for deflection in a slightly bent simply supported rectangular plate subject to uniformly
distributed load over entire plate. Use Navier’s method. Determine the expressions for moments Mx, My, Mxy
and shears Qx, Qy and Vx and Vy.
2. The Fourier equivalent load on a plate of size a × y is
• •
mp x npy
q ( x, y ) = Â Â amn sin a
◊ sin
b
.
m =1 n =1
Derive the expressions for the following, if the plate is simply supported along all edges.
w, Mx, My, Mxy, Qx, Qy, Vx and Vy.
Chapter
6
Levy’s Solution for Rectangular
Plate Analysis
Navier’s method is slow because it consists of summation of terms in two directions. Levy suggested
a method in which summation of the Fourier terms in only one direction is required and the boundary
conditions at the other two opposite edges are satisfied by closed form function. As series solution is
in only one direction, it is converging faster than Navier’s solution. This method is directly applicable
for a plate with at least two opposite edges simply supported and the boundary conditions on other two
opposite edges are any. The analysis is little bit lengthy but with only a few terms sufficiently accurate
results can be obtained.
In this chapter, first application of Levy’s method to simply supported rectangular plate is explained.
Then its extension to rectangular plate with various end conditions is explained.
6.1 ANALYSIS OF RECTANGULAR PLATE SUBJECTED TO UDL BY

LEVY’S METHOD
The Fourier equivalent load in x-directions corresponding to uniformly distributed load is
•
4 q0 mpx
q= Â m p
sin
a
m =1,3,...
Let the deflection function for a plate of size a × b be,

•
m px
w= Â Ym sin a
m =1
where Ym is a function of y only.
It satisfies the boundary conditions at x = 0 and at x = a (i.e. w = 0; Mx = 0 at x = 0 and x = a).
Now we have to select Ym such that it satisfies the boundary conditions at y = 0 and y = b and also
q
the plate equation — 4w =
D
∂4 w m4p4 mpx
Now, = Â Ym sin
∂x 4
a 4
a
∂4w m2p2 m px
2 = Â -2Ym¢¢ sin
∂x ∂y
2 2
a 2 a
∂4w m px
= Â Ym¢¢¢¢ sin
∂y 4 a
∂ 2Ym ∂ 4Ym
where Ym¢¢ = and Ym¢¢¢¢ = .
∂y 2 ∂y 4
∴ The plate equation for this case is
• Ê m4p4 m2p2 ˆ mpx 1 • 4 q0 mpx
Â ÁË a 4 Ym - 2 a 2 Ym + Ym ˜¯ sin a D Â mp sin a
¢¢ ¢¢¢¢ =
m =1 m =1,3,...
Comparing term by term, even terms on left hand side vanish and for odd terms, we get;
m4p4 m2 p2
1 4q0
Ym - 2 Ym¢¢ + Ym¢¢¢ =...eqn. 6.1
a 4
a D mp 2
When the above equation is solved there will be four constants of integration and we have got four
boundary conditions i.e. two at y = 0 and two at y = b. Hence, the problem can be solved.
Solution of the equation:
It consists of particular solution and complementary solution.
Particular solution:
The equation 6.1 can be written as
2
Ê m2p2 ˆ 4 q0
ÁË 2 - ∂ ˜¯ Ym = mpD
2
a
-2
4 q0 Ê m 2 p 2 ˆ
∴ Ym , particular = Á - ∂2 ˜
m pD Ë a 2 ¯
-2 -2
4 q0 Ê m 2 p 2 ˆ Ê a2 ˆ
= - ∂2 ˜
mpD ËÁ a 2 ¯˜ ÁË 1
m p
2 2 ¯
4 q0 a 4 Ê a2 ˆ
= 5 5 Á
1 + 2 ∂ 2 + ...˜
m p DË m p
2 2 ¯
4 q0 a 4 È Ê a2 ˆ ˘
= Í1 + ∂ 2
ÁË 2 2 2 ˜¯ ˙
m 5 p 5 D ÎÍ m p ˚˙
4q0 a 4
=
m 5p 5 D
LEVY’S SOLUTION FOR RECTANGULAR PLATE ANALYSIS 49
Complementary Integral:
2
Ê m2p2 ˆ
ÁË 2 - ∂ ˜¯ Ym = 0 .
2
a
The repeated roots are
m2 p2
- ∂2 = 0
a2
mp
∂=± .
a
Hence, the complementary integral is
mp - mp m py - m py
y y
Ym = H1e a + H2 e a + H 3 ye a + H 4 ye a
Since, H1, H2, H3 and H4 are arbitrary constants to be determined from boundary conditions, let us
select some other arbitrary constants Am, Bm, Cm and Dm such that,
Am + Cm A - Cm
H1 = , H2 = m
2 2
Bm + Dm - Bm + Dm
H3 = mp and H 4 = ◊ mp
2 2
mpy - mpy
A + Cm A - Cm
Hence, Ym = m e a + m e a
2 2
mpy - mpy
Bm + Dm mpy - Bm + Dm mpy
+ e a + ◊e a
2 a 2 a
Ê m py - mpy ˆ m py - mpy
Áe a +e a ˜ e a -e a
= Am Á ˜¯ + C m
Ë 2 2
mpy - mpy Ê mpy - mpy ˆ

mpy e a -e a mpy Á e a +e a ˜
+ Bm ◊ + Dm ÁË ˜¯
a 2 a 2
m py m py m py
Thus, ym = Am cosh + Bm sinh
a a a
mpy mpy mpy
+ C m sinh + Dm cosh ...eqn. 6.2
a a a
It is convenient to select the origin of the coordinate system x, y at the point of symmetry as shown
in Fig. 6.1.
b/2
0 x
b/2
Fig. 6.1
It may be observed that,

mpy - m py
mpy e a +e a
cosh =
a 2
- m py m py
+e mpy
{ }
a a
cosh - mp ( - y ) =
e
and = cosh .
2 a
m py
Thus, cosh is a symmetric function.
a
cosh mpy sinh mpy

a a
mpy mpy
a a
mpy mpy
(a) Variation of cosh (b) Variation of sinh
a a
Fig. 6.2
Similarly,
mpy mpy
-
mpy e a -e a
sinh =
a 2
- m py mpy
mp ( - y ) e a -e a mpy
and sin = = - sinh .
a 2 a
Ê m py ˆ m py
Hence, sinh Á
Ë a ˜¯
is antisymmetric function. Figure 6.2 shows the variation of cosh and
a
mpy
sinh .
a
m py m py mpy mpy
Similarly, it may be observed that sinh is symmetric term and cosh is anti-
a a a a
symmetric term.
Hence, when symmetric load acts, there cannot be antisymmetric terms in deflection.
i.e. For symmetric loading, homogeneous solution is
mpy m py m py
Ym = Am cosh
+ Bm sinh ...eqn. 6.3
a a a
Similarly for antisymmetric loading, homogeneous solution is
m py m py m py mpy
Ym = C m sinh + Dm cosh ...eqn. 6.4
a a a a
The arbitrary constants are to be determined using boundary conditions.
Thus we have
Y = Yparticular + Yhomogeneous
For udl, since load is symmetric
4 q0 a 4 mpy mpy mpy
Ym = + Am cosh + Bm sinh .
m p D
5 5 a a a
Redefining the arbitrary constants, let us take
q0 a 4 È 4 mpy mpy mpy ˘
Ym = Í + Am cosh + Bm sinh
D Îm p 5 5 a a a ˙˚
•
mpx
and hence w= Â Fmt - ym sin
a
m =1,3,...
The boundary conditions are

at y = ±b/2, w = 0 ...(1)
and My = 0
Ê ∂2 w ∂2 w ˆ
i.e. -D Á 2 + 2 ˜ =0
Ë ∂x ∂y ¯ y ± b
2
∂2 w
but = 0, since it is supported all along y = ±b/2.
∂x 2
Hence, the second boundary condition is
∂2 w
-D =0 ...(2)
∂y 2 y=±
b
2
From first boundary condition for all odd values of m,
4 mpb mpb mpb
+ Am cosh + Bm sinh =0
m p
5 5
2a 2a 2a
mpb
Substituting αm for , the first boundary condition is
2a
4
+ Am cosh a m + Bm a m sinh a m = 0 ...(3)
m p
5 5
dYm È mp mpy mp mpy mpy mp mpy ˘ q0 a 4

Ym¢ = = Í Am sinh + Bm sinh + Bm cosh
a ˙˚ D
Now,
dy Î a a a a a a
q0 a 4 m p È mpy mpy mpy ˘

= ( Am + Bm )sinh + Bm cosh
D a ÍÎ a a a ˙˚
d 2Ym q0 a 4 m p È mp m py Ê mp mpy mpy mp mpy ˆ ˘

Ym¢¢ = = ◊ Í ( Am + Bm ) cosh + Bm Á ◊ cosh + sinh ˜˙
dy 2 D a Î a a Ë a a a a a ¯˚
q0 a 4 m 2 p 2 È mpy mpy mpy ˘

= 2 Í m
( A + 2 Bm )cosh + Bm sinh
D a Î a a a ˙˚
∴ From boundary condition (2), we get
mpb m pb mpb
( Am + 2 Bm )cosh
+ Bm sinh =0
a a a
i.e. (Am + 2Bm) cosh αm + Bm αm sinh αm = 0 ...(4)
Subtracting eqn. (3) from eqn. (4), we get
4
2 Bm cosh a m - =0
m p
5 5
2
or Bm =
m p cosh a m
5 5
Substituting this value of Bm in eqn. (3), we get

4 2
+ Am cosh a m + a m sinh a m = 0
m p
5 5
m p cosh a m
5 5
2
∴ Am cosh a m = - ÈÎ 2 + a m tanh a m ˘˚
m p
5 5
2
∴ Am = - ÈÎ 2 + a m tanh a m ˘˚
m p cosh a m
5 5
•
q0 a 4 È 4 2 (2 + a m tanh a m ) m py 2 mpy mpy ˘ m px
∴ w= Â Í 5 5 - cosh + sinh ˙ sin
ÍÎ m p m p cosh a m m p cosh a m a
5 5 5 5
m =1,3,... D a a ˙˚ a
•
4 q0 a 4 È 2 + a m tanh a m mpy 1 mpy mpy ˘ mpx
Thus, w = Â m 5p 5 D Í1 - 2 cosh a cosh a + 2 cosh a a sinh a ˙ sin a
m =1,3,... Î m m ˚
...eqn. 6.5
The above expression for deflected surface is fast converging and hence only a few terms give
sufficiently accurate results. The main reason for fast convergence in one of the series is already
summed up.
6.2 PHYSICAL MEANING OF PARTICULAR AND COMPLEMENTARY

(HOMOGENEOUS) INTEGRALS
Consider a beam of span ‘a’ subject to uniformly distributed load q0 over its entire span. Hence its
deflection is given by
∂4w
D = q0
∂x 4
Expressing the load in Fourier form,
•
∂4w 4 q0 mpx
D
∂x 4
= Â m p
sin
a
m =1,3,...
•
4 q0 mpx
∴ w= Â m p D
5 5
◊ sin
a
m =1,3,...
This is same as the first term in deflection expression for plate which is due to particular integral.
Hence, physical meaning of particular integral is, it gives deflection of the plate which has free bound-
ary conditions at edges y = ±b/2. In other words, particular solution is due to cylindrical bending of the
plate. Hence, the complimentary integral gives effect of boundary conditions. Thus, for plate
w = w1 + w2
where w1 – deflection of strip unaffected by boundary conditions at y = ±b/2
and w2 – effect of boundary conditions at y = ±b/2. (Complimentary integral contribution).
6.3 CONVERGENCE STUDY

a
Maximum deflection occurs at middle of plate i.e. at x = and y = 0.
2
wcentre = wcentre of strip +wcentre due to complimentary integral
We know,
5 q0 a 4
wcentre of strip =
384 D
•
5 q0 a 4 2 + a m tanh a m mp
∴ wcentre = - Â sin
384 D m =1,3,... 2 cosh a m 2
For a square plate,
mpb mp
a = b \ am = =
2a 2
5 q0 a 4 4 q0 a 4
∴ w= - 5 [0.68562 - 0.00025 + ...]
384 D p D
q0 a 4
= 0.00406
D
It may be observed that second term in the series is less than 0.04 percentage of first term in the
series. Hence, the first term itself gives more than 99.96 percentage accuracy.
On the same lines moment expressions also may be studied. Study of bending moment values for
b
different ratio shows:
a
b
1. For > 3, the calculation for a plate may be replaced by those for strip.
a
b
2. If = 2, the bending moment at middle of the plate is over estimated by 18.6 percent by strip
a
b
theory. Hence, many codes recommend design of slabs treating them as one way slabs if > 2. Note
a
that by this approximation error is on safer side.
6.4 COMPARISON BETWEEN NAVIER’S SOLUTION AND LEVY’S SOLUTION

From study of the above two methods the following points may be observed.
1. Navier’s solutions are simple while Levy’s solutions are lengthy.
2. To get satisfactory results more terms in series should be considered when Navier’s method is
used while Levy’s method gives satisfactory results with few terms only.
3. Navier’s solution is applicable for plate with all four edges simply supported whereas Levy’s
method holds good for a plate with two opposite edges simply supported and other two having
any conditions.
4. Levy’s method can be extended for the analysis of plates with any edge conditions whereas it is
not possible to use Navier’s solutions for plate with boundary conditions other than all four edges
simply supported.
QUESTION
1. Derive the expressions for deflection for a simply supported rectangular plate of size a × b. Use Levy’s
method.
Chapter
7
Rectangular Plates with
Various Edge Conditions
In general, we come across plates with various edge conditions. Some edges may be free, some fixed
and some other clamped. It is possible to extend plate theory to all possible combinations of edge
conditions. In this chapter, method of analysing plates with various combination of edge conditions is
discussed. First, a case of a plate with all edges simply supported and subject to moments f1(x) and f2(x)
along edges y = 0 and y(b) is presented. Then using this solution with other solutions is discussed for
getting solution for plates with different edge conditions.
7.1 SIMPLY SUPPORTED PLATE SUBJECT TO MOMENTS ALONG y = 0 AND

y=b
Figure 7.1 shows a rectangular plate of size a × b simply supported along all four edges and subjected
to moments f1(x) along y = −b/2 edge and f2(x) along the edge y = b/2.
f1(x)
b/2
0 x
b/2 f2(x)
Fig. 7.1 Plate subjected to edge moments
For this case the deflection function w should satisfy the plate equation
∂4w ∂4w ∂4w q
+2 + = =0 ...eqn. 7.1
∂x 4
∂x ∂y
2 2
∂y 4 D
and also the following boundary conditions.

∂2 w
w = 0 and = 0 for the edges x = 0 and x = a
∂x 2
w = 0 at y = ±b/2 ...eqn. 7.2
Ê ∂2 w ˆ
-D Á 2 ˜ = f1 ( x ) ...eqn. 7.3
Ë ∂y ¯ y =- b
2
Ê ∂2 w ˆ
and -D Á 2 ˜ = f2 ( x ) ...eqn. 7.4
Ë ∂y ¯ y = b
2
In this case as q(x, y) = 0, particular integral is zero. Hence, Levy’s solution is
mpx
w = Â Ym sin
a
mpy m py m py m py mpy mpy
where Ym = Am cosh + Bm sinh + C m sinh + Dm cosh
a a a a a a
If f1(x) and f2(x) are not the same, it is possible to split the case into a symmetric case of loading
f1 ( x ) + f2 ( x ) f ( x ) - f2 ( x )
and an anti-symmetric case of loading 1 as shown in Figure 7.2. For example,
2 2
if f1(x) is a uniform moment of 80 kN-m and f2(x) = 60 kN-m it may be split into a symmetric uniform
80 + 60 80 - 60
moment of = 70 kN-m and an anti-symmetric uniform moment of = 10 kN-m. Hence
2 2
m px
it is better to study the analysis for a symmetric moment Â E m sin and for an anti-symmetric
a
m px
moment Â E m¢ sin .
a
80 kN-m 70 kN-m 10 kN-m
60 kN-m 70 kN-m
10 kN-m
General case Symmetric case Anti-symmetric case
Fig. 7.2
RECTANGULAR PLATES WITH VARIOUS EDGE CONDITIONS 57
(a) Symmetric Moment Case

Figure 7.3 shows this case.
å Em sin mpx
b/2
a
0 x
å E m sin mpx
b/2 a
Fig. 7.3 Plate with symmetric edge moments
Since, the loading is symmetric, the anti-symmetric terms in deflection expression must vanish i.e.
Cm = Dm = 0.
Ê m py mpy m py ˆ mpx
∴ w = Â Á Am cosh + Bm sinh ˜ sin
Ë a a a ¯ a
The boundary condition w = 0 at y = ±b/2 gives
mpb mpb m pb
0 = Am cosh + Bm sinh
2a 2a 2a
m pb
Substituting αm for , we can write
2a
0 = Am cosh a m + Bm a m sinh a m ...(1)
∂w mp Ê mpy mpy mpy mpy ˆ mpx

Now =Â ÁË Am sinh + Bm sinh + Bm cosh ˜¯ sin
∂y a a a a a a
∂2 w m2p2 Ê m py m py mpy m py mpy ˆ m px

=Â 2 Á
Am cosh + Bm cosh + Bm cosh + Bm ◊ sinh ˜¯ sin
∂y 2 a Ë a a a a a a
m2 p2 Ê mpy mpy mpy mpy ˆ mpx

=Â 2 Á
Am cosh + 2 Bm cosh + Bm sinh ˜ sin
a Ë a a a a ¯ a
Hence, boundary condition
m px
M y = Â E m sin
a
at y = ±b/2 means
∂2 w mpx
-D = Â E m sin
∂y 2
y=±
b a
2
Comparing term by term,
m2 p2
-D ( Am cosh a m + 2 Bm cosh a m + Bm a m sinh a m ) = Em
a2
a2
or Am cosh a m + 2 Bm cosh a m + Bm a m sinh a m = - Em ...(2)
m2 p2 D
Subtracting equation (1) from equation (2), we get
a2
2 Bm cosh a m = - Em
m2 p2 D
a 2 Em
∴ Bm = -
2 m 2 p 2 D cosh a m
From eqn. (1)
Am = - Bm a m tanh a m
Substituting the value of Bm , we get
a 2 Em
Am = a m tanh a m
2 m 2 p 2 D cosh a m
•
a2 E Ê mpy mpy mpy ˆ mpx
∴ w=
2p 2
D
Â m 2 cosh
m
a
ÁË a m tanh a m cosh
a
-
a
sinh
a
˜¯ sin
a
...eqn. 7.5
m =1 m
Particular Cases:
If moment is uniform along edges y = ±b/2, f1(x) = M0,
•
4 M0 mpx
M0 = Â sin
m =1,3,... mp a
4 M0
i.e. Em =
mp
•
2 M0 a 2 Ê 1 mpy mpy mpy ˆ mpx
∴ w= Â
p D m =1,3,... m cosh a m
3 ÁË a m tanh a m cosh
3 a
-
a
sinh
a
˜¯ sin
a
Along middle line i.e. at y = 0,
•
2 M0 a 2 1 mpx
w y =0 =
p D
3 Â m cosh a m
3
◊ a m tanh a m sin
a
m =1,3,...
If a is very large compared to b,

tanh a m = a m and cosh a m = 1.
2 M0 a 2 1 mpx
w y =0 =
p D
3 Â m3 ◊ a 2m sinh a
2 M0 a 2 1 m 2 p 2 b2 mpx
=
p D
3 Â m3 ◊ 4a 2
sin
a
M0b2 1 m px
=
2 pD
Â m sin a
(b) Anti-symmetric Case
Consider the anti-symmetric case shown in Fig. 7.4 in which
b/2
0 x
b/2
mπx
Fig. 7.4 Plate with anti-symmetric edge moment ∑E m′ sin
a
m px
f1 ( x ) = - f 2 ( x ) = Â E m¢ sin
a
Since, it is anti-symmetric case, symmetric terms in general solution must vanish.
i.e. Am = Bm = 0.
•
Ê m py m py m py ˆ mpx
Hence, w= Â ÁË Cm sinh a
+ Dm
a
cosh
a
˜¯ sin
a
m =1
From boundary condition, w y = ± b 2 = 0, we get

m pb m pb mpb
C m sin + Dm ◊ cosh =0
2a 2a 2a
i.e. C m sinh a m + Dm a m cosh a m = 0 ...(3)

•
∂w È mp Ê m py mpy m py m py ˆ ˘ m px
Now = ÂÍ ÁË C m cosh + Dm cosh + Dm ◊ sinh ˜ ˙ sin
∂x m =1 Î a a a a a ¯˚ a
•
∂2 w m2p2 È m py mpy m py mpy ˘ m px
2 Í( m
∴ = Â C + Dm ) sinh + Dm sinh + Dm ◊ cosh ˙ sin
∂x 2 m =1 a Î a a a a ˚ a
•
m2p2 È m py m py m py ˘ mpx
2 Í( m
= Â C + 2 Dm ) sinh + Dm cosh ˙ sin
m =1 a Î a a a ˚ a
∂2 w mpx
From boundary condition - D = Â E m¢ sin
∂y 2
y=±
b a
2
m2 p2 È mpb mpb mpb ˘

2 Í( m
-D C + 2 Dm ) sinh + Dm cosh = Em¢ .
a Î 2 a 2 a 2 a ˙˚
a2
i.e. (Cm + 2 Dm ) sinh a m + Dm a m cosh a m = - E m¢
m2p2 D
Subtracting equation (3) from equation (4) we get
a 2 E m¢
Dm = -
2 m 2 p 2 D sinh a m
Hence, from equation (3), we get
a 2 Em¢ a m
Cm = ◊ coth a m
2 m 2 p 2 D sinh a m
•
2
∴ w= a Â E m¢ È mpy mpy mpy ˘ mpy
ÍÎ a m coth a m sinh a - a cosh a ˙˚ sin a
...eqn. 7.6
2 p D m =1 m sinh a m
2 2
The following differences in equation 7.5 and 7.6 may be noted:

In symmetric case In anti-symmetric case
tanh αm → coth αm
cosh αm → sinh αm
m py m py
cosh → sinh
a a
m py m py
and sinh → cosh .
a a
(c) If Moments at y = −b/2 and y = b/2 are Different

If the moments applied along edges y = −b/2 and y = b/2 are different, they may be split into a
symmetric case and an anti-symmetric case as shown in Fig. 7.2. Let,
m px m px
symmetric moment be E m sin and anti-symmetric moment be E m¢ sin . Then obviously
a a
solution for such case is
mpx
• sin È Em
a2 Ê mpy mpy mpy ˆ
w= Â a
Í ÁË a m tanh a m cosh - sinh
a ¯
˜
2p Î cosh a m
2 2 a a
Dm =1 m
E m¢ Ê mpy mpy mpy ˆ ˘

+ ÁË a m coth a m sinh - cosh ˜˙ ...eqn. 7.7
sinh a m a a a ¯˚
If the bending moment is acting only along the edge y = −b/2, moment along edge y = b/2 is zero.
f1 f1
Then symmetric moment is and anti-symmetric moment is also .
2 2
7.2 RECTANGULAR PLATE WITH TWO OPPOSITE EDGES SIMPLY SUPPORTED

AND THE OTHER TWO EDGES FIXED
Figure 7.5 shows a typical rectangular plate considered. The plate is subjected to udl q0.
b/2
q0
0 x
b/2
Fig. 7.5 Plate with edges y = ±b/2 fixed
This problem can be solved by Levy’s method applying boundary conditions at y = ±b/2 as w = 0
∂w
and = 0. However, here it is solved by superposition of an appropriate edge moment case on the
∂y
simply supported case. The value of moment to be applied at the edges is arrived from the consideration
that when the two cases are combined slope at fixed edges should be zero.
Figure 7.6 shows how given problem can be split into two cases. The solution for case I and case II
clubbed with the condition that
∂w1 ∂w2 b
+ = 0 at y = ±
∂y ∂y 2
gives the solution for the given case. Thus, the deflection for the given case is
w = w1 + w2.
q0 +
Case I Case II
Fig. 7.6 Splitting the case shown in Figure 7.5 into two cases
From equation 6.5, we know for case I,

•
4 q0 a 4 È 2 + a m tanh a m mpy 1 mpy mpy ˘ mpx
w1 = Â m 5 p 5 D Í1 - 2 cosh a cosh a + 2 cosh a a sinh a ˙ sin a
m =1,3... Î m m ˚
•
∂w1 4q0 a 4 mp È 2 + a m tanh a m mpy
∴ = Â Í - sinh
∂y m =1,3... m p D a Î
5 5 2 cosh a m a
1 mpy 1 mpy m py ˘ mpy

+ sinh + cosh ˙ sin
2 cosh a m a 2 cosh a m a a ˚ a
∂w1 4 q0 a 3 1 È (2 + a m tan a m )
•
1 1 ˘ mpy
∴
∂y b
= Â4 Í
p D m =1,3... m Î
4
-
2
tanh a m + tanh a m + a m ˙ sin
2 2 ˚ a
y=±
2
•
2 q0 a 3 mpy
ÈÎ a m - tanh a m (1 + a m tanh a m )˘˚ sin
1
= 4 Â
p D m =1,3... m 4
a
w2 for this case is given by equation 7.6. Thus,
•
a2 Em È mpy mpy mpy ˘ mpx
w2 = Â
2 p D m =1,3... m cosh a m
2 2 ÍÎ a m tanh a m cosh a - a sinh a ˙˚ sin a
•
∂w2 a2 Em mp È mpy
∴
∂y
= 2 Â ◊
2 p D m =1, 3,... m cosh a m a Î
2 Í a m tanh a m sinh
a
m py m py mpy ˘ m px
- sinh - ◊ cosh ˙ sin
a a a ˚ a
•
∂w2 a Em mpx
∴
∂y
= Â
2 pD m =1,3… m
È a m tanh a m tanh a m - tanh a m - a m ˚˘ sin
Î a
y =± b 2
•
Em mpx
ÈÎ tanh a m (a m tanh a m - 1) - a m ˘˚ sin
a
= Â
2 pD m =1,3,... m a
The condition to be satisfied is
∂w1 ∂ w2
+ =0
∂y y =±
b ∂y y =±
b
2 2
∂w1 ∂ w2
i.e. =-
∂y y =±
b ∂y y=±
b
2 2
•
2 q0 a 3 mpx
ÈÎ a m - tanh a m (1 + a m tanh a m )˘˚ sin
1
i.e. Â
p D m =1,3,... mn
4
a
•
Em mpx
È tanh a m (a m tanh a m - a m ) ˘˚ sin
a
=- Â
2 pD m =1,3,... m Î a
Comparing term by term and rearranging we get for all odd values of m
4 q0 a 2 a m - tanh a m (1 + a m tanh a m )
Em =
p 3 m 3 a m - tanh a m (a m tanh a m - 1)
Hence, the bending moment along fixed edges are
•
mpx
M y=±b 2 = Â Em sin
a
m =1, 3,...
4 q0 a 2 1 a m - tanh a m (1 + tanh a m ) mpx

=
p 3 Â m3 a - tanh a m (a m tanh a m - 1)
sin
a
m
For a square plate maximum value at midspan may be obtained by substituting

m pb m p
a = b i.e., a m = = .
2a 2
and x=a 2
mp mp Ê mp mp ˆ
• - tanh Á1 + tanh ˜
4 q0 a 2 1 2 Ë 2 ¯ mp
M max =
p3
Â m 3 mp 2
mp Ê mp
2
mp ˆ
sin
2
m =1,3,... - tanh - 1˜
Á tanh
2 2 Ë 2 2 ¯
p
Noting that tanh = 0.9175
2
3p
tanh = 0.99984
2
and for all other higher values of m,
mp
tanh = 1.0, we get
2
4 q0 a 2 È 1 1 1 ˘
M max = Í - 0.57236 + 0.03692 - 3 + 3 - 3 + ...˙
p Î
3
5 7 9 ˚
= -0.06973 qa 2
Note that the first term itself gives about 6% extra moment. Thus, even if only one term of the series
is considered there is only 6% overestimation.
Deflection study
Consider deflection due to edge moments only.
mpx
• E m sin
a2 È mpy mpy mpy ˘
w2 = 2 Â a
2 p D m =1,3,... m cosh a m
2 ÍÎ a m tanh a m cosh a - a sinh a ˙˚
At centre of plate y = 0 and x = a 2 .

mp
• Em sin
a2
∴ w2, centre = 2 Â 2 È a tanh a - 0 ˘
2 p D m =1,3,... m cosh a m
2 Î m m ˚
Substituting the value of Em, we get

mp
a2 •
4 q0 a 2 a m - tanh a m (1 + a m tanh a m )
sin
w2, centre = Â
2 p 2 D m =1,3,...
2
p m m cosh a m a m - tanh a m (a m tanh a m - 1)
3 3 2
¥ a m tanh a m
mp
2 a m tanh a m ¥ a m - tanh a m (1 + a m tanh a m )
2 q0 a 4 • sin
=
p5D
Â m 5
cosh a m a m - tanh a m (a m tanh a m - 1)
m =1,3,...
m -1
mp
Noting that sin = ( -1) 2 , we can write
2
m -1
2 q0 a 4 • ( -1) a m tanh a m a m - tanh a m (a m tanh a m + 1)
Â
2
w2, centre = ¥
p D
5
m =1,3,... m 5
cosh a m a m - tanh a m (a m tanh a m - 1)
mp
For square plate: a = b i.e. a m =
2
Hence, we get
2 q0 a 2
w2, centre = ( - 0.328 + 0.000378 - ......)
p5D
qa 4
= - 0.00214
D
For plate with udl,
q0 a 4
w1, centre = 0.00406
D
q0 a 4
∴ w = w1 + w2 = (0.00406 - 0.00214 )
D
q0 a 4
= 0.00192
D
Note that second term of the series contributes very little. Thus, it is a fast converging series.
Using total solution (w = w1 + w2) it is possible to assemble all stress resultants.
7.3 RECTANGULAR PLATE WITH ALL FOUR EDGES FIXED

Figure 7.7 shows a rectangular plate with all four edges fixed and subjected to uniformly distributed
load.
a/2 a/2
b/2
0 x
q0
b/2
Fig. 7.7 Fixed plate subject to udl q0

qo x
x
x
y (w1) y (w2) y (w 3)
Fig. 7.8 Equivalent of fixed plate
The given case may be split into the following three cases (Refer Fig. 7.8).
Case I: A simply supported plate subject to udl.
Case II: A simply supported plate subject to symmetric moment along y = ±b/2.
Case III: A simply supported plate subject to symmetric moment along y = ±a/2.
Thus,
w = w1 + w2 + w3.
For first two cases, we have already solution available. To have convenient coordinates, it is better
a
to keep origin at centre of the plate. It may be achieved by replacing x by x - in the solutions already
2
found for case I and case II. Then to get the solution for case III, interchange x and y in the expression
for w2. Thus, expression for w3 is also obtained. Then the boundary conditions to be satisfied are
Ê ∂w1 ∂w2 ∂w3 ˆ
ÁË ∂y + ∂y + ∂y ˜¯ =0
y= ± b 2
Ê ∂w1 ∂w2 ∂w3 ˆ

ÁË + + ˜ =0
and ∂x ∂x ∂x ¯ x = ± a 2
From these conditions functions Em′ and Em″ may be found.
w = w1 + w2 + w3
After finding w, stress resultants may be assembled.
7.4 RECTANGULAR PLATE WITH ONE EDGE SIMPLY SUPPORTED,

OTHER EDGES FIXED
The plate is ABCD as shown in Figure 7.9. For this first analysis may be made for a plate of size
a × 2b fixed along all edges subject to anti-symmetric loading.
This case is shown in Figure 7.10. Now along edge AB, due to anti-symmetry
w = 0 and My = 0.
Thus, it satisfies required boundary condition along AB. Hence, the required solution is obtained for
the case considered.
a
A B
q0
b
D C
Fig. 7.9 Plate with three edges fixed and one simply supported
b q0
0
x
b -q0
Fig. 7.10 Equivalent of plate shown in Figure 7.9
7.5 RECTANGULAR PLATE WITH TWO ADJACENT EDGES SIMPLY SUPPORTED

AND OTHER TWO FIXED
This case is shown in Figure 7.11. This case may be solved from the analysis of a fixed plate subject
to anti-symmetric load of size 2a × 2b as shown in Figure 7.12. It satisfies the required boundary
conditions along edges AB and AC.
b q0
Fig. 7.11 Plate with two adjacent sides fixed and other two simply supported
2a
q0 -q0
2b
-q0 q0
Fig. 7.12 Equivalent of plate shown in Figure 7.11
7.6 RECTANGULAR PLATE WITH TWO OPPOSITE EDGES SIMPLY SUPPORTED,

ONE EDGE FIXED AND ANOTHER FREE
This case is shown in Figure 7.13.
Fig. 7.13 Plate with two opposite edges simply supported, one edge fixed and other free
For this case

w = w1 + w2
where w1 is particular solution and w2 is complementary solution.
•
4 q0 a 4 1 mpx
We know w1 =
p D5 Â m 5
sin
a
m =1,3,...
m px
and w2 = Â Ym sin
a
q0 a 4 È mpy mpy mpy mpy mpy mpy ˘
Ym = Í Am cosh + Bm sinh + Cm sinh + Dm cosh
a ˙˚
where
D Î a a a a a
The boundary conditions to be satisfied are
At x = 0 and x = a
∂2 w
w = 0 and =0
∂x 2
m πx
Hence, by taking series in the form of sin the boundary conditions are satisfied.
a
The following boundary conditions help to get the constants Am, Bm, Cm and Dm.
w = 0 at y = 0
∂w
= 0 at y = 0.
∂y
Ê ∂2 w ∂2 w ˆ
Á m 2 + 2 ˜ = 0 at y = b
Ë ∂x ∂y ¯
∂3 w ∂3 w
+ (2 - m ) = 0 at y = b
∂y 3 ∂x 2 ∂ y
Hence, the total solution is obtained.
7.7 RECTANGULAR PLATES WITH THREE EDGES BUILT IN AND FOURTH EDGE FREE
Figure 7.14 shows this case. Such cases commonly appear in the design of water tanks and counterfort
retaining walls.
q q
Fig. 7.14 Plate with three edges Fig. 7.15 Equivalent for plate shown in Figure 7.14
fixed, fourth free
The analysis may be treated as the analysis of a plate of size a × b with three edges fixed and fourth
edge free, subject to given loading (Ref. Figure 7.15) plus the solution for a plate with edges x = 0 and
x = a subject to symmetric moment with the conditions slopes at x = 0 or x = a in x direction are zero.
7.8 RECTANGULAR PLATE CONTINUOUS IN ONE DIRECTION

Consider the analysis of continuous plate shown in Figure 7.16. The end plate 1, may be treated as a
mpx
simply supported plate subject to moment Â E m1 sin along the continuous edge. The intermediate
a
mpx
plate may be treated as a simply supported plate subject to moment Â E m1 sin at one edge and the
a
m px
moment Â E m2 sin at the other edge. The third plate is treated as a simply supported plate with
a
m px
moment Â E m sin at the continuous edge.
a
a1 a21 a3
b x1 x2 x3
y1 y2 y3
Fig. 7.16 Continuous plate in one direction
To get Em and Em the conditions to be satisfied are

1 2
∂w1 ∂w 2
=
∂y1 x1 =
a1 ∂ y2 x2 =-
a2 ...(1)
2 2
∂w 2 ∂w3
= .
and ∂ y2 x2 =
a2 ∂y 3 x3 =-
a3
2 2
It may be noted that if a1 = a2 = a3 = a it reduces to problem with only one unknown Em.
One can find deflections first and then assemble coefficients for moments.
7.9 RECTANGULAR PLATES CONTINUOUS IN BOTH DIRECTIONS

This case is shown in Fig. 7.17. Slabs are usually simply supported along outer edges and continuous
over beams or walls and are subjected to downward load. Any panel in the slab may be approximated
as one of the six panels shown in Fig. 7.18. For these cases method of finding edge moment and stress
resultants have been already discussed.
Fig. 7.17 Rectangular plate continuous in both directions
0 0
0
I II III
0 0 0
IV V VI
Fig. 7.18 Standard Cases
At common edges moments may slightly differ, but average moment may be approximately taken as
final moment.
To get design moment at centre of the plate we know the panel should be loaded with live load and
the adjacent panel should not be loaded. The solution for this case is obtained as an appropriate
standard case with q0 + p/2 load plus a simply supported plate with p/2 load where q0 is dead load and
p is live load.
The second part considered is appropriate since if we consider a load of p/2 applied in the checker
bound form as shown in Figure 7.19, each panel behaves as a simply supported case. Hence, final
moment at centre of span
= Moment at centre of span in standard case with a load of (q0 + p/2)

+ Moment in simply supported slab with load of p/2.
+ - + -
- + - +
+ - + -
Fig. 7.19 Checker bound loading
QUESTIONS
1. Analyse a simply supported rectangular plate of size a × b if it is subjected to

(a) Uniform symmetric moment at y = ±b/2.
(b) Uniform anti-symmetric moment at y = ±b/2.
2. Discuss the extension of Levy’s method for the analysis of a fixed plate subject to udl q0.
3. Discuss the analysis of continuous plate to get maximum moment at mid-spans. Assume continuity is in both
directions and the plate is loaded with dead load q0 and a live load of p per unit area.
Chapter
8
Circular Plates Bent
Axi-symmetrically
Circular plates are commonly used as base slabs of circular water tanks and as footings for circular
columns. Usually they are subjected to uniform loads and have axi-symmetric edge conditions. Hence,
the bending is axi-symmetric. In this article after deriving equation of equilibrium for axi-symmetric
bending of circular plate, number of standard cases are analysed and at the end it is pointed out how
several cases can be analysed by suitably combining standard cases.
8.1 CO-ORDINATES AND ELEMENT

For the analysis of circular plates polar coordinate system is more convenient. Figure 8.1 shows a
typical circular plate of radius ‘a’ with polar coordinates for an element of size rdθ × dr ; located at
distance r from centre of plate and making angle θ with a reference axis. Downward direction is
positive z-coordinate direction.
0
q
r dr
dq
Fig. 8.1 Circular plate and polar coordinates
Figure 8.2(a) shows the moments Mr, Mθ and Mrθ acting on the element and Figure 8.2(b) shows the
shear forces and the load intensity qrθ acting on the element, where
Mr = radial moment per unit length
Mθ = Circumferential moment per unit length

Mrθ = Mθr = Twisting moment per unit length
Qr = radial shear per unit length
and Qθ = Circumferential shear per unit length.
Mq Mqr
dq Qr Qq
Mr
Mrq
+
Mr
qrq
+
Mq
+
Mrq
+
Mqr
+
Qq
+
Qr
Fig. 8.2 (a) Moments on the element Fig. 8.2 (b) Forces on the element
All the stress resultants have been shown with their positive senses. The sign convention used is, a
stress resultant is positive when it is acting on positive face in positive direction of coordinate or when
it is acting on negative face in negative direction. It may be noted that forces on positive faces differ
from these on negative faces and their relation is obviously as given below:
∂M r
M r+ = M r + dr
∂r
∂M q
Mq+ = Mq + ◊ dq
∂q
∂Mqr
M q+r = M qr + ◊ dr
∂r ...eqn. 8.1
∂M r+q
M r+q = M r+q + dq
∂q
∂Qr
Qr+ = Qr + dr
∂r
∂Qq
Qq+ = Qq + ◊ dq
∂q
It may also be noted that qrθ is load intensity. Hence, total down load = qrθ dθrdr = qrθ rdrdθ
CIRCULAR PLATES BENT AXI-SYMMETRICALLY 75
8.2 AXI-SYMMETRIC BENDING OF PLATE

If the boundary conditions and loads are axi-symmetric the plate bending also should be axi-symmetric.
In other words, the condition of a strip represents the condition of the plate. Hence, in such bending
twisting moment is zero and there is no variation of any force or moment with respect to θ. It means
Mθr, Mrθ and Qθ are zero. Thus in axi-symmetric bending a typical element is subjected to Mr, Mθ and
Qr only.
B
B¢ f
f + df df
df
B
0 A
z
Fig. 8.3 Axi-symmetrically bent plate
Figure 8.3 shows shape of axi-symmetrically bent plate. The radii of curvatures of points at distance
r meet at B and those at points r + dr meet at B′. Since, bending is axi-symmetric B and B′ are on a
vertical line through centre of the plate. Let slope of bent plate at radius r (i.e. at A) be φ and that at
r + dr (i.e. at A′) be φ + dφ. Hence, radius of curvature AB makes angle φ with vertical through centre
O and A′B′ makes angle φ + dφ with vertical through O as shown in the figure.
∂w
Slope at A = - , since as r increases w decreases.
∂r
∂w
i.e. f=- ...eqn. 8.2
∂r
Curvature of the middle surface in the diametral section rz is,
1 1 df ∂2w
= = =- 2 ...eqn. 8.3
AC rn dr ∂r
The second principal curvature is in the θ-direction. It may be observed that the normals such as AB
for all points at radial distance r form a conical surface with B as apex. Hence, AB is the radius of
principal curvature in θ-direction. Thus,
rθ = AB.
From Figure 8.3, it is clear that,
r = AB ⋅ φ = rθφ
∴ Second principal curvature is
1 f 1 ∂w
= =- ...eqn. 8.4
rq r r ∂r
Hence, the expression for Mr is given by
Ê1 1 ˆ
Mr = D Á + m˜
Ë rn rq ¯
Ê ∂ 2 w m ∂w ˆ
= -D Á 2 +
r ∂r ˜¯
...eqn. 8.5
Ë ∂r
Similarly
Êm 1ˆ
Mq = D Á + ˜
Ë rn rq ¯
Ê 1 ∂w ∂2w ˆ
= -D Á +m 2 ˜ ...eqn. 8.6
Ë r ∂r ∂r ¯
Equilibrium of the element

As stated earlier in axi-symmetrical bending any element is subjected to Mr , Mθ, Qr and applied load
intensity. Figure 8.4(a) shows the section and Figure 8.4(b) shows the plan of the element at (r, θ).
qr
q +
Mr Mr
0
h r
qr+
r dr
a
z (a)
Mq
+
Mr Mr
Mq
(b)
Fig. 8.4 (a) Sectional view (b) Plan view

∂M r
Now, Mr+ = Mr + dr
∂r
∂M q
Mq+ = Mq + ◊ dq
∂q
= Mθ since, there is no variations w.r.t. θ
∂Qr
Qr+ = Qr + dr
∂r
From equilibrium condition ∑Mr = 0, we get
Ê ∂M r ˆ dq
ÁË M r + dr ˜ (r + dr ) dq - M r rdq - 2 ( M q dr ) sin
∂r ¯ 2
Ê ∂Qr ˆ dr dr
- Á Qr + dr ˜ (r + dr ) d q - Qr rd q =0
Ë ∂r ¯ 2 2
dq dq
Since, dθ is small quantity we can take sin =
2 2
Neglecting small quantity of higher order, we get
∂M r
M r dr dq + r dr dq - M q dr dq - Qr r dr dq = 0
∂r
Throughout dividing by r dr dθ, we get
∂M r M r M q
+ - - Qr = 0
∂r r r
Ê ∂ 2 w m ∂w ˆ
Mr = - D Á 2 +
r ∂r ¯˜
Substituting
Ë ∂r
Ê ∂ 2 w 1 ∂w ˆ
and M q = - D Á m 2 +
r r ˜¯
, we get
Ë ∂r
Ê ∂ 2 w m ∂ 2 w m ∂ w ˆ D Ê ∂ 2 w m ∂w ˆ 1 Ê ∂ 2 w 1 ∂w ˆ
-D Á 3 + - 2 ˜ - Á 2 + ˜ - (- D ) Á m 2 + = Qr
Ë ∂r r ∂r 2
r ∂r ¯ r Ë ∂r r ∂r ¯ r Ë ∂r r ∂r ˜¯
∂3 w m ∂ 2 w m ∂w 1 ∂ 2 w m ∂w ∂ 2 w 1 ∂w Q
+ - + + - m - 2 =- r
∂r 3 r ∂r 2 r 2 ∂r r ∂r 2 r 2 ∂r ∂r 2
r ∂ r D
∂3 w 1 ∂ 2 w 1 ∂w Q
+ - 2 =- r
∂r 3 r ∂r 2
r ∂r D
∂ È 1 ∂ Ê ∂w ˆ ˘ Qr
i.e. Í ÁË r ˜¯ ˙ = - ...eqn. 8.7
∂r Î r ∂r ∂r ˚ D
The above equation is known as plate equation for axisymmetric plate bending.
Example 8.1. Analyse a circular plate of radius ‘a’ supported throughout at its outer edge and subjected
to uniform moment M.
Fig. 8.5 Example 8.1
Solution. The plate equation is

∂ È 1 ∂ Ê ∂w ˆ ˘ Qr
Í ÁË r ˜¯ ˙ = -
∂r Î r ∂r ∂r ˚ D
In this case, Qr = 0.
∂ È 1 ∂ Ê ∂w ˆ ˘
∴ Í Ár ˜ ˙ = 0.
∂r Î r ∂r Ë ∂r ¯ ˚
1 ∂ Ê ∂w ˆ
∴ Ár ˜ = C1 , where C1 is a constant.
r ∂r Ë ∂ r ¯
∂ Ê ∂w ˆ
i.e. Ár ˜ = C1r
∂r Ë ∂ r ¯
Integrating both sides again, we get
∂w r2
r = C1 + C 2 , where C2 is a constant.
∂r 2
∂w C1r C2
∴ = +
∂r 2 r
C1r 2 r
Integrating both sides once again, we get w = + C 2 log + C3 , where C3 is a constant. Note
4 a
that in the above form instead of C2log r it is written as C2log r/a, since,
C2log r/a = C2 log r − C2 log a
and C2 log a is a constant. Hence, it just alters the arbitrary constant C3 which is yet to be determined.
The form log r/a is convenient to apply boundary conditions.
To determine the three arbitrary constants C1, C2 and C3 the following three boundary conditions are
available.
∂w
= 0 at r = 0 ...(1)
∂r
w = 0 at r = a ...(2)
and Mr = M at r = a. ...(3)
From boundary condition 1, we get
C2
0=
0
This is possible if and only if C2 = 0
∴ C2 = 0
From boundary condition (2), we get
C1a 2
0= + C3 , since C 2 = 0.
4
C1a 2
∴ C3 = -
4
C1r 2 C1a 2
Hence, w= -
4 4
=- (a - r 2 )
C1 2
4
From boundary condition (3),
Mr r =a
=M
Ê ∂ 2 w m ∂w ˆ
-D Á 2 + =M
Ë ∂r r ∂r ˜¯ r = a
C1 È m ˘
D Í -2 + ( -2r )˙ = M
4 Î r ˚
DC1
-
2
[1 + m] = M
2M
∴ C1 = -
D (1 + m )
¥ (a 2 - r 2 )
2M 1
∴ w=
D (1 + m ) 4
w=
M (a 2 - r 2 )
2 D (1 + m )
i.e. ...eqn. 8.8
∴ At any point,
Ê ∂ 2 w m ∂w ˆ
Mr = - D Á 2 +
Ë ∂r r ∂r ¯˜
È ( ) ˘
= -D Í
M
( -2) + m M -2r ˙
Î 2 D (1 + m ) r 2 D (1 + m ) ˚
[1 + m]
M
=
1+m
=M
(m + 1) = M .
M
Similarly, Mq =
1+ m
Thus in this case at any point
Mr = Mq = M . ...eqn. 8.9
Example 8.2. Analyse a circular plate of radius ‘a’ carrying udl q, if its outer edge is having fixed
support.
Solution. Figure 8.6 shows such plate.
a
q
[Note - Qr value will be -ve]

Qr Qr
Fig. 8.6 Shear is fixed plate subject to udl ‘q’

The shearing stress Qr is given by

2πr Qr + πr2q = 0
qr
Qr = - [Note: It is upward at outer edge.]
2
∴ From plate equation, we get
∂ È 1 ∂ Ê ∂w ˆ ˘ -Qr qr
Í Ár ˜˙ = =
∂r Î r ∂r Ë ∂r ¯ ˚ D 2D
1 ∂ Ê ∂w ˆ qr 2
∴ Ár ˜= + C1
r ∂r Ë ∂r ¯ 4 D
∂ Ê ∂w ˆ qr 3
i.e. Ár ˜= + C1r
∂r Ë ∂r ¯ 4 D
∂w qr 4 C1r 2
∴ r = + + C2
∂r 16 D 2
∂w qr 3 C1r C 2
i.e. = + +
∂r 16 D 2 r
qr 4 C1r 2
∴ w= + + C2 log r a + C3
64 D 4
Boundary conditions to be satisfied are,
∂w
= 0 at r = 0 ...(1)
∂r
∂w
= 0 at r = a ...(2)
∂r
and w = 0 at r = a ...(3)
From boundary condition 1, we get
C2
0=
0
This is possible only when C2 = 0.
From boundary condition 2,
qa 3 C1a
0= + , since C2 = 0.
16 D 2
qa 2
∴ C1 = -
8D
qr 4 qa 2 r 2
∴ w= - + C3
64 D 8 D 4
Hence, from boundary condition (3),

qa 4 qa 4
0= - + C3
64 D 32 D
qa 4
∴ C3 =
64 D
qr 4 qa 2 r 2 qa 4
Thus, w= - +
64 D 32 D 64 D
q ( 2
r - a2 )
2
i.e. w= ...eqn. 8.10
64 D
∴ Maximum deflection is at centre of the plate, where r = 0.
qa 4
wmax =
64 D
∂w qr 3 qa 2 r qr ( 2
Now, = - = r - a2 )
∂r 16 D 16 D 16 D
∂2 w 3qr 2 qa 2
and = - =
q
(3r 2 - a 2 )
∂r 2 16 D 16 D 16 D
Hence,
Ê ∂ 2 w m ∂w ˆ
Mr = - D Á 2 +
Ë ∂r r ∂r ¯˜
È q ( 2 m q( 2 ˘
= -D Í 3r - a 2 ) + r - a 2 )˙
Î 16 D 16 D ˚
2
qa
(1 + m)
16 Mq
Mr
2
qa
m
16
2
qa
8
Fig. 8.7 Variation of Mr and Mθ

q È
= Î -3r + a - mr + ma ˘˚
2 2 2 2
16
a (1 + m ) - r 2 (3 + m )˘˚
q È 2
i.e. Mr = ...eqn. 8.11
16 Î
a (1 + m ) - r 2 (1 + 3m ) ˘˚
q È 2
Similarly Mq =
16 Î
Variation of Mr and Mθ are as shown in Figure 8.7.
Example 8.3. Analyse a simply supported circular plate subject to udl ‘q’.
Solution. This plate may be analysed by solving plate equation with the boundary conditions w = 0 and
∂2 w ∂w
= 0 at r = a, apart from the condition = 0 at r = 0.
∂r 2 ∂r
However, here it is solved by superposing solutions for a fixed plate subject to udl q and a plate
supported at outer edges subjected to edge moment (Refer Figure 8.8).
q/unit area
Fig. 8.8. Example 8.3
- qa 2
The end moment in fixed plate is . If the two cases together are to represent given case, the
8
condition to be satisfied is
qa 2
0=M-
8
qa 2
i.e. M= .
8
Hence, for the given case,
w = w1 + w2
=
q
(a 2 - r 2 )2 + M (a 2 - r 2 )
64 D 2 D (1 + m )
q ( 2 qa 2 (a 2 - r 2 )
a - r2 ) +
2
=
64 D 8 ¥ 2 D (1 + m )
q ( 2 È 4a2 ˘
= a - r 2 ) Ía2 - r 2 + ˙
64 D Î 1+ m˚
q ( 2 È5 + m 2 ˘
i.e. w= a - r2 )Í a - r2 ˙ ...eqn. 8.12
64 D Î 1 + m ˚
Similarly,
M r = M r1 + M r2
a (1 + m ) - r 2 (3 + m ) ˘˚ + M
q È 2
=
16 Î
qa 2
a (1 + m ) - r 2 (3 + m ) ˘˚ +
q È 2
= Î
16 8
= ÈÎ a 2 (1 + m + 2 ) - r 2 (3 + m ) ˘˚
q
16
= (3 + m ) ( a 2 - r 2 )
q
...eqn. 8.13(a)
16
M q = M q1 + M q2
qa 2
a (1 + m ) - r 2 (1 + 3m )˘˚ +
q È 2
= Î
16 8
a (1 + m + 2 ) - r 2 (1 + 3m )˘˚
q È 2
=
16 Î
(3 + m ) a 2 - r 2 (1 + 3m )˘˚
q È
= Î ...eqn. 8.13(b)
16
Note:
1. If µ = 0,
wcentre in simply supported plate
q 2
= a ¥5
64 D
= 5 times deflection in a fixed plate.
2. Mr, centre in simply supported plate
3 2
= qa
16
= 3 × Mr , centre in fixed plate.
Example 8.4. A circular plate of radius ‘a’ is having a concentric hole of radius b and is simply
supported along its outer periphery. Analyse the plate if it is subjected to
(a) Uniform moments M1 and M2 as shown in Figure 8.9.
(b) Uniform load Q0 along its inner edge.
Solution.
(a) Uniform Moments M1 and M2
Moment M1 is acting along inner periphery and M2 along the outer periphery as shown in Figure 8.9.
M2 M1 M1 M2
b
a
Fig. 8.9 Example 8.4(a)
In this case, there is no load. Hence, Qr = 0

Hence, plate equation reduces to
∂ È 1 ∂ Ê ∂w ˆ ˘
Í Ár ˜˙ = 0
∂r Î r ∂r Ë ∂r ¯ ˚
C1r 2 r
∴ w= + C2 log + C3
4 a
The constants of integrations C1, C2 and C3 are to be determined by using boundary conditions.
Mr = M2 at r = a ...(1)
Mr = M1 at r = b ...(2)
and w=0 at r = a ...(3)
Ê ∂ 2 w m ∂w ˆ
Mr = - D Á 2 +
r ∂r ¯˜
Now,
Ë ∂r
∂w C1r C2
= +
∂r 2 r
∂2w C1 C2
and = -
∂r 2
2 r2
ÈC C mC mC ˘
∴ Mr = - D Í 1 - 22 + 1 + 2 2 ˙
Î2 r 2 r ˚
ÈC C ˘
i.e. Mr = - D Í 1 (1 + m ) - 22 (1 - m ) ˙
Î2 r ˚
ÈC C ˘
M 2 = - D Í 1 (1 + m ) - 22 (1 - m )˙ ...(4)
Î 2 a ˚
ÈC C ˘
M1 = - D Í 1 (1 + m ) - 22 (1 - m )˙ ...(5)
Î 2 b ˚
È1 1 ˘
∴ M 2 - M1 = - DC2 (1 - m ) Í 2 - 2 ˙
Îb a ˚
È a2 - b2 ˘
= - D C 2 (1 - m ) Í 2 2 ˙
Î a b ˚
C2 = -
( M 2 - M1 ) a 2 b 2
∴
D (1 - m ) (a 2 - b 2 )
Substituting it in equation (5), we get
ÈC
M1 = - D Í 1 (1 + m ) +
( M 2 - M1 ) a 2 b2 ¥ 1 1 - m ˘˙
( )
ÍÎ 2 D (1 - m ) ( a 2 - b 2 ) b ˙˚
2
ÈC
= - D Í (1 + m ) +
1 ( M 2 - M1 ) a 2 ˘
˙
∴
ÍÎ 2 D (a 2 - b 2 ) ˙˚
C1
D (1 + m ) = -
( M 2 - M1 ) a 2
- M1
∴
2 (a 2 - b2 )
È M a 2 - M1a 2 + M1 a 2 - M1b 2 ˘
= -Í 2 ˙
Î a2 - b2 ˚
M 2 a 2 - M1b 2
=-
a2 - b2
C1 = -
(
2 M 2 a 2 - M1b 2 )
D (1 + m ) ( a - b
2 2 )
C1a 2
0= + C3
4
C1a 2 M 2 a 2 - M1b 2
C3 = - =
2 D (1 + m ) (a 2 - b 2 )
or 4
Thus,
C1r 2 r
w= + C 2 log + C3
4 a
C1 =
(
-2 M 2 a 2 - M1b 2 )
D (1 + m ) ( a - b )
where 2 2
C2 = -
( M 2 - M1 ) a 2 b 2
D (1 - m ) (a 2 - b 2 )
M 2 a 2 - M1b 2
C3 =
2 D (1 + m ) (a 2 - b 2 )
and
Note: If M2 = 0
2 M1 b 2 M1 a 2 b 2
C1 = , C2 =
D (1 + m ) (a 2 - b 2 ) D (1 - m ) (a 2 - b 2 )
M1a 2 b 2
C3 = -
2 D (1 + m ) ( a 2 - b 2 )
and
2 M1b 2 r2 M1 a 2 b 2 r M1a 2 b 2
∴ w= ¥ + log -
D (1 + m ) (a 2 - b 2 ) 4 D (1 - m ) a 2 b 2 a 2 D (1 + m ) (a 2 - b 2 )
M1b 2 (r 2 - a 2 ) M1 a 2 b 2 r
= + log
2 D (1 + m ) ( a - b ) D (1 - m ) a b
2 2 2 2 a
M1b 2 (a 2 - r 2 ) M1 a 2 b 2 r
i.e. w=- + log ...eqn. 8.14
2 D (1 + m ) (a - b ) D (1 - m ) a b
2 2 2 2
a
∂w M1a 2 b 2 È r 1- m 1˘
= +
∂r D (1 - m ) (a - b ) ÍÎ a 2 1 + m r ˙˚
2 2
...eqn. 8.15
(b) Uniform Load Q0 Along Inner Edge

Figure 8.10 shows this case
Q0
Fig. 8.10 Example 8.4(b)

Solution: If P is total load applied,

P = 2πbQ0
At radius r shearing force per unit length of circumference is
P
Qr = -
2 pr
∂ È 1 ∂ Ê ∂w ˆ ˘ P
∴ Í ÁË r ˜¯ ˙ =
∂r Î r ∂r ∂r ˚ 2 prD
1 ∂ Ê ∂w ˆ P r
∴ Ár ˜= log + C1
r ∂ r Ë ∂ r ¯ 2 pD a
∂ Ê ∂w ˆ Pr r
i.e. ÁË r ˜¯ = log + C1r
∂r ∂r 2 pD a
∂w P È r2 r r 2 1 ˘ C1r 2
∴ r = Í log - Ú ¥ dr ˙ + + C2
∂ r 2 pD Î 2 a 2 r ˚ 2
P Èr2 r r 2 ˘ C1r 2
= Í log - ˙+ + C2
2 pD Î 2 a 4˚ 2
∂w P Èr r r˘ Cr C
∴ = Í log - ˙ + 1 + 2
∂ r 2 pD Î 2 a 4˚ 2 r
P È r2 r r2 1 r2 ˘ C r2 r
∴ w= Í log - Ú dr - ˙ + 1 + C 2 log + C3
2 pD Î 4 a 4 r 8˚ 2 a
P È r2 r r 2 r 2 ˘ C1r 2 r
= Í log - - ˙+ + C 2 log + C3
2 pD Î 4 a 8 8˚ 2 a
Pr 2 È r ˘ C1r
2
r
= Í log - 1˙ + + C2 log + C3
8 pD Î a ˚ 4 a
Mr = 0 at r = b ...(1)
Mr = 0 at r = a ...(2)
and w = 0 at r = a. ...(3)
Ê ∂ 2 w m ∂w ˆ
Mr = - D Á 2 +
r ∂r ˜¯
Now,
Ë ∂r
Mr P È1 r 1 1˘ C C mP È 1 r 1 ˘ mC1 mC 2
∴ = Í log + - ˙ + 1 - 22 + Í log - ˙ + + 2
- D 2 pD Î 2 a 2 4˚ 2 r 2 pD Î 2 a 4˚ 2 r

P Ê b 1 ˆ mP Ê b 1 ˆ C1 C
0= Á log + ˜¯ + Á log - ˜¯ + (1 + m ) - 22 (1 - m ) ...(4)
4 pD Ë a 2 4 pD Ë a 2 2 b
1 mP Ê 1 ˆ C1 C
0=
P
¥ + ÁË - ˜¯ + (1 + m ) - 22 (1 - m ) ...(5)
4 pD 2 4 pD 2 2 a
Subtracting (5) from (4), we get
Ê 1 1ˆ
log - C2 (1 - m ) Á 2 - 2 ˜
P b P b
0= log + m
4 pD a 4 pD a Ëb a ¯
(1 + m ) log b - C2 (1 - m ) a 2- b2
2 2
P
=
4 pD a a b
P 1 + m a2b2 b
∴ C2 = log ...(6)
4 pD 1 - m a - b
2 2 a
Substituting it in eqn. 4, we get
P Ê b 1 ˆ mP Ê b 1 ˆ C1 2
0= ÁË log + ˜¯ + ÁË log - ˜¯ + (1 + m ) - P (1 + m ) 2 a 2 log b
4 pD a 2 4 pD a 2 2 4 pD a -b a
C1 P È (1 + m ) a 2 log b + 1 1 - m ˘
∴ - ( )
1 + m = Í log
b
( ) 2 2
1 + m - ( )˙
2 4 pD Î a a -b a 2 ˚
P È1 a2 - b2 - a2 ˘
( (
b
= Í 1 - m ) + log 1 + m ) ˙
4 pD ÎÍ 2 a (a 2 - b 2 ) ˚˙
P È1 - m 2b2 b˘
∴ C1 = - Í - 2 log ˙ ...(7)
4 pD Î 1 + m a - b 2 a˚
2
Pa 2
0= (0 - 1) + C1a + C3
8 pD 4
a 2 Pa 2
∴ C3 = -C1 +
4 8 pD
Pa 2 È 1 - m 1 b2 b˘
= Í1 + ¥ - 2 log ˙
8 pD Î 1 + m 2 a - b 2 a˚
Particular case
b
As b approaches zero, it becomes a plate with central concentrated load. In this case, b 2 log = 0. Hence,
a
P 1- m Pa 2 Ê 1 1- mˆ
C1 = - , C2 = 0 and C3 = Á 1+
4 pD 1 + m 8 pD Ë 2 1 + m ˜¯
Pr 2 Ê r ˆ P 1 - m r 2 Pa 2 Ê 1 1- mˆ
∴ w= ÁË log - 1˜¯ - + Á 1+
8 pD a 4 pD 1 + m 4 8 pD Ë 2 1 + m ¯˜
P È 2 r ˘ Pr 2 Ê 1 - m 1 ˆ Pa 2 È 1 1 - m ˘
= r log + - 1 - + 1+
8 pD ÍÎ a ˙˚ 8 pD ÁË 1 + m 2 ˜¯ 8 pD ÍÎ 2 1 + m ˙˚
P È 2 2 2 (1 + m ) + 1 - m
˘
Ír log + ( a - r )
r
= 2
˙
8 pD ÎÍ a 2 (1 + m ) ˚˙
P È 2 3+m ( 2 ˘
a - r 2 )˙
r
= r log +
8 pD ÍÎ a 2 (1 + m ) ˚
8.3 CASES OF PRACTICAL IMPORTANCE

By suitable superposition of above cases many cases of practical importance can be solved. Some of
these cases are shown in Figure 8.11.
P
Fig. 8.11 Cases of Practical Importance

For example, consider a simply supported circular plate subjected to a concentric ring load of
intensity P having radius ‘b’ (case I).
This may be looked as combination of the following three cases:
(i) A simply supported plate with concentric hole subject to ring load P.
(ii) A simply supported plate with a concentric hole subjected to uniform moment M1 along inner
edge.
(iii) A circular plate of radius ‘b’ subject to uniform moment M1 along outer edge.
The condition to be satisfied is
∂w1 ∂w2 ∂w3
+ = at r = b.
∂r ∂r ∂r
The above condition helps in finding M1.
The above three cases are shown in Figure 8.12.
P
2a
(i)
M1
(ii)
M1
2b
(iii)
Fig. 8.12
QUESTIONS
1. Derive the equation of equilibrium for the axi-symmetrically bent circular plate.
2. Discuss the method of analysing a circular plate subject to a ring load P. Assume outer edge of the plate is
fixed.
Chapter
9
Plates of Other Shapes
The solution for a plate of any shape can be obtained, if a deflection function in single unknown can
be found to satisfy the boundary conditions. Then using the plate equation, it is possible to find the
unknown parameter. In this chapter, the analysis of the following shaped plates is presented:
1. Elliptic Plates with Fixed Edges subject to
(a) udl q0.
(b) Linearly varying load
2. Equilateral Triangular Plate subject to
(a) Pure moment along simply supported edge
(b) udl q0.
9.1 ELLIPTIC PLATE WITH CLAMPED EDGES AND SUBJECTED TO UDL

Figure 9.1 shows the typical elliptic plate fixed along its outer periphery and subjected to load intensity
q0. Let its length along major axis be 2a and along minor axis 2b. With the origin of cartesian coordinates
at centre of plate, the equation of the boundary line is
b
q0
a a
Fig. 9.1 Elliptic plate with clamped edges

PLATES OF OTHER SHAPES 93
x2 y2
+ - 1.0 = 0 ...eqn. 9.1
a2 b2
Along this boundary,
w=0
∂w
and = 0. ...(1)
∂n
∂w ∂w
i.e. cos q + sin q = 0 ...(2)
∂x ∂y
The deflection function should be such that the boundary conditions 1 and 2 are satisfied.
To satisfy the boundary condition w = 0 along the edge it is necessary that w should have a term
Ê x2 y2 ˆ
ÁË 2 + 2 - 1˜¯ as a multiplying factor. The second boundary condition shows that even after first
a b
Ê x2 y2 ˆ
differentiation w.r.t. x and y we should be left with a term Á 2 + 2 - 1˜ . Hence, the deflection func-
Ëa b ¯
2
Ê x 2 y2 ˆ Ê x2 y2 ˆ
tion ‘w’, should contain the term Á 2 + 2 - 1˜ instead only Á 2 + 2 - 1˜ .
Ëa b ¯ Ëa b ¯
Apart from satisfying boundary conditions, the following plate equation also should be satisfied.
∂4w ∂4w ∂4w q
i.e. +2 + = ...eqn. 9.2
∂x 4
∂x ∂ y
2 2
∂y 4 D
The above equation shows that, the deflection function w should not contain any term of degree
higher than four so that left hand side also becomes constant as right hand side is. So, the power of term
Ê x2 y2 ˆ
ÁË 2 + 2 - 1¯˜ should not be more than two. Hence, let
a b
2
Ê x2 y2 ˆ
w = C Á 2 + 2 - 1˜ ...eqn. 9.3
Ëa b ¯
It satisfies the boundary condition that w = 0 at any point on the boundary. Now
∂w ∂w ∂w
= cos q + sin q
∂n ∂x ∂y
Ê x 2 y2 ˆ 2x Ê x 2 y2 ˆ 2y
= 2C Á 2 + 2 - 1˜ 2 cos q + 2C Á 2 + 2 - 1˜ 2 sin q
Ëa b ¯a Ëa b ¯b
x2 y2 ∂w
Since, at any point on the boundary + - 1 = 0, the boundary condition = 0 is satisfied.
a 2
b 2 ∂x
From plate equation,

∂4w ∂4w ∂4w q
+2 + =
∂x 4
∂x ∂ y
2 2
∂y 4 D
Now,
∂w È x2 y2 ˘ 2x
= 2C Í 2 + 2 - 1˙ 2
∂x Îa b ˚a
∂2 w 4C È 3 x 2 y 2 ˘
∴ = Í + - 1˙
∂x 2 a2 Î a2 b2 ˚
∂3 w 4C ¥ 3 ¥ 2 x 24 x
= =
∂x 3
a 4
a4
∂4w 24
=
∂x 4
a4
∂4w 24
Similarly, =
∂y 4 b4
∂4w 4C Ê 2 ˆ 8C
and = 2 Á ˜ = 2 2
∂x ∂ y
2 2
a Ëb ¯ a b
2
Hence, from plate equation,
È 24 2 ¥ 8 24 ˘ q
CÍ 4 + 2 2 + 4 ˙ =
Îa a b b ˚ D
q 1
C=
D 24 16 24
or + +
a 4 a2b2 b4
Thus,
2
q0 Ê x 2 y2 ˆ
w= ÁË 2 + 2 - 1˜¯
Ê 24 16 24 ˆ a b
DÁ 4 + 2 2 + 4 ˜
Ëa a b b ¯
2
Ê x 2 y2
q0 a 4 ˆ
or w= ÁË 2 + 2 - 1˜¯ ...eqn. 9.4
Ê a 2
a ˆ a
4
b
D Á 24 + 16 2 + 24 4 ˜
Ë b b ¯
q0 a 4
∴ wcentre =
Ê a2 a4 ˆ
D Á 24 + 16 2 + 24 4 ˜
Ë b b ¯
Particular Cases
If a = b, it is a clamped circular plate.
q0 a 4 q a4
wcentre = = 0 .
D (24 + 16 + 24 ) 64 D
Same as we got in Chapter 8.
If a = 2b,
q0 a 4 q a4
wcentre = = 0
D (24 + 16 ¥ 4 + 24 ¥ 16 ) 472 D
Expressions for Moments
Ê ∂2w ∂2w ˆ
Mx = -D Á 2 + m 2 ˜
Ë ∂x ∂y ¯
ÈÊ 3 x 2 y2 1ˆ Ê x2 3 y2 1 ˆ˘
= -4CD Í Á 4 + 2 2 - 2 ˜ + m Á 2 2 + 4 - 2 ˜ ˙ ..eqn. 9.5(a)
ÎÍ Ë a a b a ¯ Ëa b b b ¯ ˚˙
È Ê 3x2 y2 1 ˆ Ê x2 3y2 1 ˆ˘
M y = -4CD Í m Á 4 + 2 - 2 ˜ + Á 2 2 + 4 - 2 ˜ ˙ ...eqn. 9.5(b)
ÎÍ Ë a b a ¯ Ëa b b b ¯ ˚˙
Particular Cases
q 1 qa 4
(a) If a = b, it is a circular plate. In this case C = =
D 24 16 24 64
4
+ 4 + 4
a a a
∴ M at x = 0, y = 0
qa 4 D È 1 Ê 1 ˆ˘
= -4 ¥ Í - 2 + m ÁË - 2 ˜¯ ˙
64 D Î a a ˚
qa 4
= (1 + m ) , same as we got in Chapter 8.
16
(b) If a = 2b
q 1 q 1
C= =
D 24 16 24 D 472
+ +
a4 a 2
Ê aˆ
4
a2 ◊ ÁË ˜¯
4 2
qa 4 È 1 1 ˘
∴ M x , centre = -4 ¥ Í - 2 -m ˙
472 Í a
Î (a 2)2 ˚˙
qa 2
=
118
[1 + 4m]
qa 2
M y , centre =
118
[4 + m ]
Moment at the end of major axis (x = a, y = b)
qa 4 È Ê 3 1ˆ Ê 1 1 ˆ˘
M x = -4 Í ÁË 2 - 2 ˜¯ + m Á 2 - ˙
472 Í a
Î a Ë (b ) (b )2 ¯˜ ˚˙
qa 2
=- .
59
Moment at the end of minor axis, i.e. at x = 0, y = b,
qa 4 È Ê 1 1ˆ 3 1 ˘
M y = -4 Ím Á - 2 + 2 - 2˙
472 Í Ë ( a 2) 2
a ˜¯ b b ˙
Î ˚
qa 4 È m ( -3) 2 ˘
=- Í 2 + ˙
118 Í a
Î ( a 2)2 ˙˚
qa 2
118
=-[8 - 3m ]
The variations of moment along major and minor axes are shown in Figure 9.2, if µ = 0.
-
b 2
4qa
118
a a
+
b
-
2
8qa
118
2
qa
- -
59
+
2
qa
118
Fig. 9.2 Variations of moments along major and minor axes
9.2 COMPARISON OF RESULT WITH STRIP METHOD ANALYSIS

Let intensity of load taken by long strip be q1 and that by short strip be q2. Then,
q1 + q2 = q0 ...eqn. 9.6
To ensure central deflection of the two strip same, it should be,
q1a4 = q2b4
a4
∴ q2 = q1
b4
In particular case of a = 2b,
q2 = 16q1
Substituting it in eqn. 9.5, we get
q1 + 16q1 = q0
q0 16
or q1 = and hence q2 = q0
17 17
q1 (2 a )
4
16 q0 ◊ a 4 q a4
∴ wcentre = = = 0
384 D 17 384 D 408 D
Ê q a4 ˆ
Thus, deflection estimated in strip method is more than actual Á 0 ˜ . The overestimation is 15.7
Ë 472 ¯
percent.
Similarly central moment in case of strip method
q1 (2 a )
2
q0 q a2
= = ¥ 4a2 = 0
24 17 ¥ 24 102
Actual value as found by plate theory is
qa 2
(1 + 4 m )
118
If µ is taken as zero, as is made in strip method, we find there is overestimation of moment in strip
Ê 118 ˆ
method to an extent ÁË - 1˜ ¥ 100 = 15.7 percent.
102 ¯
9.3 ELLIPTIC PLATE SUBJECT TO UNIFORMLY VARYING LOAD IN x-DIRECTION

Figure 9.3 shows a typical fixed elliptic plate subject to a load uniformly varying load form q0 at
x = −a to −q0 at x = a.
2
Ê x 2 y2 ˆ
As discussed in the act 9.1, the deflection function should contain the term Á 2 + 2 - 1˜ , so that
Ëa b ¯
boundary conditions are satisfied.
Loading at any point is
x
q = q0 .
a
a a
q0
q0
Fig. 9.3 Elliptic Plate subject to Uniformly Varying Load
Hence, to satisfy plate equation try the function

2
Ê x 2 y2 ˆ
w = cx Á 2 + 2 - 1˜
Ëa b ¯
ÈÊ 2 ˆ
2
Ê x2 y2 ˆ 2x ˘
∂w x y2
= 2C Á 2 + 2 - 1˜ + x ¥ 2 Á 2 + 2 - 1˜ 2 ˙
Í
∂x ÎÍ Ë a b ¯ Ëa b ¯ a ˚˙
∂2 w È Ê x 2 y2 ˆ 2x 4 ÏÔ Ê x 2 y 2 ˆ ¸Ô 2x ˘
= 2C Í 2 Á + - 1˜¯ 2 + 2 Ì2 x Á 2 + 2 - 1˜ ˝ + x 2 ˙
∂x 2 ÍÎ Ë a 2 b 2 a a ÓÔ Ë a b ¯ Ô˛ a ˙˚
4C È Ê x 2 y 2 ˆ 2 x2 ˘
= Í 3 x Á + - 1˜¯ + 2 ˙
a 2 ÎÍ Ë a 2 b 2 a ˚˙
∂2 w 4C È Ê x 2 y 2 ˆ 2x 2 ˘
= Í 3 Á + - 1˜ + 3 x + 2 x ˙
∂x 3 a 2 ÎÍ Ë a 2 b 2 ¯ a 2 a 2 ˚˙
4C È Ê x 2 y 2 ˆ 12 x 2 ˘
= Í 3 Á + - 1˜¯ + 2 ˙
a 2 ÍÎ Ë a 2 b 2 a ˙˚
∂4 w 4C È 2 x 24 x ˘ 4C 30 x 120 x 2
= Í3 + 2 ˙= 2 ¥ 2 =
∂x 4 a2 Î a2 a ˚ a a a4
∂4w ∂ 2 Ê ∂ 2 w ˆ 4C È 2 ˘ 24C
= 2 Á ˜ = 2 Í3 x ◊ 2 ˙ = 2 2
∂x 2 ∂y 2 ∂y Ë ∂x ¯ a Î
2
b ˚ a b
∂w Ê x2 y2 ˆ 2y
= Cx 2 Á 2 + 2 - 1˜ 2
∂y Ëa b ¯b
∂2 w 4C È x 2 y 2 2y ˘
= x Í 2 + 2 -1+ y 2 ˙
∂y 2 b Îa
2
b b ˚
∂2w 4Cx È x 2 3 y 2 ˘
i.e. = 2 Í 2
+ 2 - 1˙
∂y 2 b Îa b ˚
∂3 w 4Cx 6y 24Cxy
∴ = ¥ =
∂y 3
b 2
b 2
b4
∂4w 24Cx
∴ =
∂y 4
b4
∴ The plate equation is,
È 120 x 48 24 x ˘ q x
CÍ 4 + 2 2 x + 4 ˙ = 0
Î a a b b ˚ aD
q0
C=
∴ È 120 48 24 ˘
aD Í 4 + 2 2 + 4 ˙
Îa a b b ˚
2
Ê x2 y2 ˆ
q0 x Á 2 + 2 - 1˜
Ëa b ¯
∴ w= ...eqn. 9.7
È 120 48 24 ˘
aD Í 4 + 2 2 + 4 ˙
Îa a b b ˚
9.4 EQUILATERAL TRIANGULAR PLATE SUBJECT TO UNIFORM EDGE MOMENT

Figure 9.4 shows the typical plate. Let the uniform edge moment be M. The origin is taken at centroid
of the plate. The coordinates x and y are as shown in the figure. If we take perpendicular distance
2a
CD = a, the sides of equilateral triangles are (Ref. Figure 9.4).
3
0 C
D x
a/3 2a/3
2a
3
A y
Fig. 9.4 Equilateral Triangular Plate Subject to edge moment
a
Equation for line AB is x+ =0 ...(1)
3
x y
Equation for line BC is - -1= 0 ...(2)
2a 3 2a
3 3
x y
and equation for line AC is + -1= 0 ...(3)
2a 3 2 a
3 3
w = 0 at edges ...(1)
and Mx = M at edges ...(2)
The plate equation to be satisfied is
∂4w ∂4w ∂4w q q

+2 + = = 0, since =0 ...(3)
∂x 4
∂ x ∂y
2 2
∂y 4 D D
The plate equation indicates that no term in w should be of degree higher than three and boundary
condition 1, indicates that expression for w must include the equations of boundary lines. Hence, let
us try
Ê aˆ Ê x y ˆÊ x y ˆ
w = C1 Á x + ˜ Á - - 1˜ Á + - 1˜
Ë 3 ¯ Ë 2a 3 2a 3 3 ¯ Ë 2a 3 2a 3 3 ¯
a ˆ ÈÊ x y2 ˘
2
Ê ˆ
= C1 Á x + ˜ Í - 1 - ˙
Ë 3 ¯ ÎÍ ÁË 2 a 3 ˜¯ 4 a 2 27 ˚˙
Ê a ˆ È x2 x y2 ˘
= C1 Á x + ˜¯ Í 2 - +1- 2 ˙
Ë 3 ÍÎ 4 a 9 a 3 4 a 27 ˙˚
Ê a ˆ 9 È 2 4 xa 4 a 2 ˘
= C1 Á x + ˜¯ 2 Í x - + - 3 y2 ˙
Ë 3 4a Î 3 9 ˚
9 È 3 4 ax 2 4 a 2 x ax 2 4 xa 2 4 a 3 ˘
= C1 Í x - + - 3 xy 2
+ - + - ay 2
˙
4a2 Î 3 9 3 9 27 ˚
9C1 È 3 4 3˘
= 2 Í
x - 3 xy 2 - ax 2 - ay 2 + a
4a Î 27 ˙˚
=
9C1 È 3
2 Í
(
x - 3 xy 2 - a x 2 + y 2 + a)
4 3˘
27 ˙˚
4a Î
∴ ∂4w ∂4w ∂4w

= 0, = 0 and =0
∂x 4 ∂x 2 ∂y 2 ∂y 4
Hence, the plate equation is satisfied.
a
w = 0 along the edges is satisfied, since, at any point on edges either x + = 0 or
3
x y x y
- - 1 = 0 or + -1 = 0 .
2a 3 2a 3 3 2a 3 2a 3 3
Another boundary condition to be satisfied is
Mn = M
È ∂2 M n ∂2 Mn ˘
i.e. -D Í + m ˙=M
Î ∂n 2 ∂t 2 ˚
∂2 Mn ∂2 Mn
Since, = 0, we get - D =M
∂t 2 ∂n 2
M Ê ∂ ∂ ˆ Ê ∂w ∂w ˆ
or - = Á cos a + sin a ˜ Á cos a + sin a ˜
D Ë ∂x ∂y ¯ Ë ∂x ∂y ¯
9C È
Î(6 x - 2 a ) cos a + 2 ( -3 ¥ 2 y ) sin a cos a + ( -3 x ¥ 2 - 2 a ) sin a ˚
2 ˘
= 2
4a2
9C È ( 2
= 2 Î
6 x cos a - sin 2 a ) - 2 a (cos 2 a + sin 2 a ) - 12 y sin a cos a ˘˚
4a
[6 x cos2a - 2 a - 6 y sin 2a]

9C ...eqn. 9.8
= 2
4a
a
Along the boundary AB, n-direction is x-direction i.e. a = 0 and x = - .
3
M 9C
= 2 ÈÎ6 ( - a 3) - 2 a ˘˚ = -
9C
∴ -
D 4a a
Ma
∴ C= ...(a)
9C
Similarly along boundary BC, n-direction makes angle 60° with x-axis. Hence, equation 9.8 reduces to
M 9C È 3˘
- = Í -3 x - 2 a - 6 y ˙
D 4a2 Î 2 ˚
9C È x y ˘
= ¥ 2a Í -
2
- - 1˙
4a Î 2a 3 2a 3 3 ˚
9C ¥ 2 a ( -2 ) x y
= 2
since + =1
4a 2a 3 2a 3 3
9C
=- ...(b)
a
Ma
or C= , same as in (a)
9D
Similarly, for line AC also we get
Ma
C= .
9D
Ma
Hence, if C = the boundary condition Mx = M along all edges is satisfied. Hence,
9D
w=
9C È 3
2 Í
( )
x - 3 xy 2 - a x 2 + y 2 +
4 3˘
a
27 ˙˚
4a Î
i.e. w=
M È 3
4 aD ÍÎ
( )
x - 3 xy 2 - a x 2 + y 2 +
4 3˘
a
27 ˙˚
...eqn. 9.9
Variation of Moments Along Line CD
∂2 w M
= [6 x - 2 a]
∂x 2
4 aD
∂2 w M
= [ -6 x - 2 a ]
∂y 2 4 aD
∂2 w
[-6 y]
M
=
∂x∂y 4 aD
Ê ∂2 w ∂2w ˆ
∴ Mx = -D Á 2 + m 2 ˜
Ë ∂x ∂y ¯
M
=- ÈÎ 6 x - 2 a + m ( -6 x - 2 a ) ˘˚
4a
È ˘
ÍÎ -3 (1 - m ) a + (1 + m )˙˚
M x
=
2
È ˘
ÍÎ 3 (1 - m ) a + 1 + m ˙˚
M x
My =
2
∂2w
and M xy = - D (1 - m )
∂x∂y
= 3 M (1 - m )
y
.
2a
∴ The variation of moments along DC is as shown in Figure 9.5.
D C
M (-1 + 3m)
2
(a) Variation of M x
M (3 - m)
mM 2
D C
( b) Variation of M y
D C
(c ) Variation of Mxy
Fig. 9.5 Variation of moments along DC

9.5 EQUILATERAL TRIANGULAR PLATE SIMPLY SUPPORTED ALONG ITS EDGES

AND SUBJECTED TO UDL
Figure 9.6 shows the case considered. In this the deflection function has to satisfy the following
conditions:
w = 0 along all boundaries ...(1)
Mx = 0 along all boundaries ...(2)
∂4w ∂4w ∂4w
and +2 + =0 ...(3)
∂x 4
∂ x ∂y
2 2
∂y 4
2a q0
x
3 a 2a C
3 3
A
y
Fig. 9.6 Equilateral triangular plate subjected to UDL
From the first condition it is obvious that there should be the term
Ê aˆ Ê x y ˆÊ x y ˆ
ÁË x + ˜¯ Á + -1
˜ Á
- -1
˜
3 2a 3 2 a 2a 3 2 a
Á ˜Á ˜
Ë 3 3 ¯Ë 3 3 ¯
( 4 3
i.e. a term x 3 - 3 xy 2 - a x 2 + y 2 +
27
)
a . The other two conditions suggest that there should be some
more terms. Krieger suggested the following function.

È
(
w = C Í x 3 - 3 xy 2 - a x 2 + y 2 +
Î
4 3˘È 2
) 4 ˘
a ˙ Í x + y2 - a2 ˙
27 ˚ Î 9 ˚
4 2
It may be noted that the term x 2 + y 2 - a is the equation of circle passing through the corners of
9
the plate.
Mx = 0 means
Ê ∂ ∂ ˆ Ê ∂w ∂w ˆ
ÁË ∂x cos a + ∂y sin a ˜¯ ÁË ∂x cos a + ∂y sin a ˜¯ = 0
∂2 w ∂2 w ∂2w
i.e. cos 2 a + 2 cos a sin a + 2 sin 2 a = 0.
∂x 2 ∂ x ∂y ∂y
It may be verified that Mx = 0 for all three sides.

Now,
∂4 w
= C [5 ¥ 4 ¥ 3 ¥ 2 x - 4 ¥ 3 ¥ 2 ¥ 1 a ] = C [120 x - 24 a ]
∂x 4
∂4w
= C [ -3 ¥ 3 ¥ 2 ¥ 2 x + 3 ¥ 2 ¥ 2 x - a ¥ 2 ¥ 2 ¥ 2 ]
∂x 2 ∂ y 2
= C [ -24 x - 8 a ]
∂4w
= C [ -3 x ¥ 4 ¥ 3 ¥ 2 ¥ 1 - a ¥ 4 ¥ 3 ¥ 2 ¥ 1]
∂y 4
= C [ -72 x - 24 a]
∴ — 4w = q D gives
q
C ÎÈ120 x - 24 a - 2 (24 x + 8 a ) - 72 x - 24 a ˚˘ =
D
q
C ( -64 a ) =
D
q
or C=-
64 aD
w=
q È 3
64 D ÍÎ
( )
x - 3 xy 2 - a x 2 + y 2 +
4 3˘È4 2
a
27 ˙˚ ÍÎ 9
˘
a - x2 - y2 ˙
˚
...eqn. (9.10)
QUESTIONS
1. Derive the expression for deflection in case of a fixed elliptic plate subject to udl. Show that strip method over
estimates the deflection by 15.7%, if a = 2b.
2. Determine the expression for deflection in a fixed elliptic plate subject to a load varying linearly from q0 at
x = −a to −q0 at x = a.
3. Determine the displacement function for an equilateral plate supported along its all edges and subjected to
uniform moment on the edges.
Chapter
10
Energy Method
As structure undergoes deformation due to applied load, the work done by the load is stored as strain
energy. Corresponding to each type strain there is strain energy stored in the structure. The different
strain energies to be considered are
1. Flexural strain energy.
2. Torsional strain energy.
3. Shear strain energy.
Shear strain energy is very small compared to other two. To make calculations simple this is
neglected.
While undergoing deformation strain energy is stored while potential energy is lost. But equilibrium
is reached with minimum total energy. The principal of minimum energy is used in the analysis. For this
a function is assumed so as to satisfy the boundary conditions and minimization of total energy with
respect to unknown parameters in the deflection function is used to determine the unknown parameters.
In this chapter, first expression for total energy is derived. Then a few plate problems are solved to
illustrate how energy method can be used to analyse the plate. This method is one of the approximate
methods, since, there can be a number of deflection functions which satisfy the required boundary
conditions. However, any function satisfying boundary conditions gives reasonably good results.
10.1 EXPRESSION FOR TOTAL ENERGY

Figure 10.1 shows an element of plate subject to the moments Mx, My, Mxy and Myx. Strain energy due
to all these moments are to be assembled.
(a) Strain Energy due to Mx:
1
dv1 = ¥ Moment × Change of angle in elemental length
2
1 Ê ∂2 w ˆ
= M x dy ¥ Á - 2 ˜ dx
2 Ë ∂x ¯
1 ∂2w
= - M x 2 dx dy
2 ∂x
(b) Strain Energy due to My:
1 ∂2w
Similary dv2 = - M y 2 dx dy
2 ∂y
ENERGY METHOD 107
Myx
My
Mx
+
Mx
+
Mxy
Mxy +
My
+
Myx
Fig. 10.1 Element of plate subject to various moments
(c) Strain Energy due to Mxy:

1
dv3 = ¥ Twisting moment ¥ Change in twist
2
1 Ê ∂2w ˆ
= M xy dy ◊ Á -
Ë ∂x∂y ˜¯
dx
2
1 ∂2w
= - M xy dx dy
2 ∂x∂y
(d) Strain Energy due to Myx:
Similarly,
1 Ê ∂2w ˆ
dv4 = M yx dx Á -
Ë ∂x∂y ˜¯
dy
2
1 ∂2 w
= - M yx dx dy
2 ∂x∂y
∴ Total strain energy in the element = dv1 + dv2 + dv3 + dv4
1È ∂2w ∂2 w ∂2w ∂2w ˘

= - Í M x 2 + M y 2 + M xy + M yx ˙
2 ÍÎ ∂x ∂y ∂x∂y ∂x∂y ˙˚
Since, Mxy = Myx, we get
1È ∂2 w ∂2 w ∂2w ˘
Total S.E. in the element = - Í M x 2 + M y 2 + 2 M xy ˙
2 ÍÎ ∂x ∂y ∂x∂y ˙˚
Ê ∂2w ∂2w ˆ
Substituting Mx = -D Á 2 + m 2 ˜
Ë ∂x ∂y ¯
Ê ∂2 w ∂2 w ˆ
My = -D Á m 2 + 2 ˜
Ë ∂x ∂y ¯
∂2w
and M xy = - D (1 - m )
∂x∂y
we get
D È ∂2 w ∂2 w ∂2 w ∂2 w ∂2 w ∂2 w
Total S. E. in the element = - Í 2 ◊ 2 +m 2 + m ◊
2 ÍÎ ∂x ∂x ∂x ∂y 2 ∂x 2 ∂ y 2
∂2w ∂2w ∂2 w ∂2 w ˘
+ ◊ + 2 (1 - m ) ◊ ˙
∂y 2 ∂y 2 ∂x ∂y ∂x ∂y ˙˚
∂w
Representing by ′
∂x
and ∂w
by .
∂y
and total strain energy in the element by dV, we get,
+ 2 (1 - m ) w ¢◊2 ˘˚ dx dy
DÈ 2
dV = w ¢¢ + w
2 + 2 mw ¢¢w
2Î
+ 2 (1 - m ) w ¢◊2 ˘˚ dx dy
DÈ
= ( w ¢¢ + w
)2 - 2 w ¢¢w
+ 2 mw ¢¢w
2Î
DÈ
Î ( w ¢¢ + w
= - w ¢◊2 ) ˘˚ dx dy
) 2 - 2(1 - m )( w ¢¢w
2
Hence, total strain energy in the plate is
D
V=
2 Ú Ú ( w ¢¢ + w - w ¢◊2 ) dx dy
)2 - 2(1 - m )( w ¢¢w ...eqn. 10.1
Potential energy lost may be represented as
U = - Ú Ú qw dx dy
∴ Total energy of the plate
I=U+V
D È
- w ¢◊2 ) ˘˚ dx dy - Ú Ú q ◊ w dx dy
2 ÚÚ
= Î( w ¢¢ + w
) 2 - 2(1 - m )( w ¢¢w ...eqn. 10.2
Example 10.1. By energy method analyse a simply supported plate of size a × b subject to uniformly
distributed load q0 over its entire surface.
ENERGY METHOD 109
0 x
q0 b
Fig. 10.2 Simply supported plate subject to udl
Solution. Figure 10.2 shows such plate. For such plate the following deflection function is suitable,
since, the boundary conditions are easily satisfied:
mpx n py
w = ÂÂ a mn sin sin
a b
In the above expression amn is the unknown parameter. Total energy is given by expression 8-2.
First term in total strain energy:
Ê m 2 p 2 n 2 p 2 ˆ 2 mpx 2 npy
ÚÚ ( w ¢¢ +
w ) 2
dx dy = Ú Ú mn Á 2 + 2 ˜ sin
a 2
sin
Ë a b ¯ a b
Noting the following:
a
mpx 2a
Ú sin a
dx =
mp
, for odd values of m
0
b
npy 2b
Ú sin b
dy =
np
, for odd values of n
0
a
m px a
Ú sin dx =
2
0
a 2
b
npy b
Ú sin dy =
2
0
b 2
m ¢px
a
mpx
Ú sin a
◊ sin
a
dx = 0
0
n ¢py
b
npy
Ú sin b
◊ sin
b
dy = 0
0
ab
mpx npy ab
Ú Ú sin dy =
2
dx sin 2
00
a b 4
If sin θ is replaced by cos θ in the above expressions and the limits changed from −a/2 to a/2, exactly
same results are obtained.
Hence,
2
D 2 Ê m 2 p 2 n 2 p 2 ˆ ab
First term = amn
2
ÂÂ ÁË 2 + 2 ˜¯ 4
a b
2 (1 - m ) Ú Ú ( w ¢¢w
D
Second term = - ) dx dy
2
m 2 p 2 n2 p 2 mpx 2 npy
= - D (1 - m ) Ú Ú amn
2
2 2
sin 2 sin dx dy
a b a b
m 2 p 2 n 2 p 2 ab
= - D (1 - m ) ÂÂ amn
2
◊ 2
a2 b 4
Similarly third term
2 (1 - m ) Ú Ú w ¢◊2 dx dy
D
2
m 2 p 2 n 2 p 2 ab
= D (1 - m ) ÂÂ amn
2
a2 b2 4
Since, 2nd and 3rd terms are equal but opposite, strain energy is given by
2
p 4 a 4 ab D Ê m2 n2 ˆ 2
V=
8
ÂÂ ÁË a 2 + b2 ˜¯ amn
Potential energy
U = - Ú Ú q w dx dy
Ê mpx npy ˆ
= - Ú Ú q Á ÂÂ a mn sin ◊ sin ˜ dx dy
Ë a b ¯
• •
2 a 2b
= -q Â Â amn ◊
m p np
m =1,3,... n =1,3,...
• •
4 qab amn
=-
p 2 Â Â
m =1,3,... n =1,3,... mn
∴ Total Energy
I = V +U
ENERGY METHOD 111
2
p 4 a 4 ab Ê m2 n2 ˆ 4 qab • •
am
= D ÂÂ amn
2
Á + ˜ - Â Â
8 Ëp 2
b ¯
2
p m =1,3,... n =1,3,... mn
2
∂I
∴ = 0, gives
∂amn
For even terms amn = 0 and for odd terms
2
p 4 ab Ê m 2 n2 ˆ 4 q ab 1
D 2 amn Á 2 + 2 ˜ - 02 =0
8 Ëa b ¯ p mn
16 q0 16 q0 a 4
amn = =
∴ Ê m2 n2 ˆ
2
Ê
2
a2 ˆ
p D mn Á 2 + 2 ˜
6
p D mn Á m 2 + 2 n 2 ˜
6
Ëa b ¯ Ë b ¯
This is same as Navier’s solution.
Example 10.2. Analyse a fixed plate subject to uniformly distributed load q0 (Ref. Figure 10.3)
a/2 a/2
0
b/2
0 x
b/2 q0
Fig. 10.3 Fixed plate subject to udl
Solution. In this case, the boundary conditions to be satisfied are

w = 0 at all edges ...(1)
∂w -a a
= 0 at x = and x = ...(2)
∂x 2 2
∂w -b b
= 0 at y = and y = ...(3)
∂y 2 2
Let us select
È 2 px ˘ È 2 py ˘
w = C Í1 - cos 1 + cos
Î a ˙˚ ÍÎ b ˙˚
For simplicity series term is not selected. The function satisfies the boundary conditions w = 0 at all
edges.
∂w 2p Ê 2 px ˆ Ê 2 py ˆ
=C ÁË - sin ˜¯ ÁË1 + cos ˜
∂x a a b ¯
∂w Ê 2 px ˆ 2 p Ê 2 px ˆ
= C Á1 + cos ˜ Á - sin ˜
∂y Ë a ¯ b Ë b ¯
∂w a
Hence, = 0 at x = ±
∂x 2
∂w b
and = 0 at y = ± .
∂y 2
Thus boundary conditions (2) and (3) are also satisfied.
∂2 w 4p2 Ê 2 py ˆ 2 px
= -C 2 Á
1 + cos ˜¯ cos
∂x 2 a Ë b a
∂2 w 4p2 Ê 2 px ˆ 2 py
and = -C 2 Á
1 + cos ˜¯ cos
∂y 2 b Ë a b
It may be observed that these curvatures are not zero at x = ±a/2 and y = ±b/2. Apart from these at
both opposite edges curvatures are the same. Hence, moment conditions are also satisfied.
However, it may be observed that
∂3w a
= 0 at x = ±
∂x 3
2
∂3 w b
and = 0 at y = ± .
∂y 3 2
Thus, shear conditions are not satisfied. Still such a function gives sufficiently good results and a
designer can use them.
∂2 w 4p2 2 px 2 py
Now, w ¢◊ = =C sin ◊ sin
∂x∂y ab a b
First term in strain energy V

D
(w ¢¢ + w )2 dx dy
2 ÚÚ
=
2
D È 4p2 2 px Ê 2 py ˆ 4 p 2 Ê 2 px ˆ 2 py ˘
= Ú Ú C 2 Í 2 cos ÁË 1 + cos ˜¯ + 2 ÁË 1 + cos ˜¯ cos ˙ dx dy
2 Î a a b b a b ˚
ENERGY METHOD 113
a2 b2 b2
2 px a ab
Noting that Ú Ú cos 2
a
dx dy = Ú dy =
2 2
-a 2 - b 2 -b 2
a2
2 px
Ú cos
a
dx = 0 since m is even.
-a 2
a2 b2 b2
2 px 2 py a 2 py ab
and Ú Ú cos2
a
cos 2
b
dx dy = Ú cos 2
2 b
dy =
4
-a 2 -b 2 -b 2
First term in strain energy
DC 2 È 16 p 4 Ê ab ab ˆ 16 p 4 Ê ab ab ˆ 16 p 4 ab ˘
= Í 4 ÁË +0+ ˜ + 4 Á +0+ ˜ +2 2 2 ˙
2 Î a 2 4¯ b Ë 2 4¯ a b 4 ˚
DC 2 È Ê 3ˆ 1 Ê 3ˆ 1 ab ˘
= ◊ 16 p 4 Í ab Á ˜ ¥ 4 + ab Á ˜ ¥ 4 + 2 2 ˙
2 Î Ë 4 ¯ a Ë 4 ¯ b a b ˚
DC 2 4 È 3 b 2 3 a 2 ˘
=4 p Í ◊ 2 + + 1˙
ab Î2 a 2 b2 ˚
4 DC 2 p 4 È 3 Ê a2 b2 ˆ ˘
= Í1 + Á 2 + 2 ˜ ˙
ab ÍÎ 2 Ë b a ¯ ˙˚
Second term in strain energy

ab
2 (1 - m ) Ú Ú w ¢¢w
D
- dx dy
2 00
ab
16 p 4 2 px 2 py Ê 2 py ˆ Ê 2 px ˆ
= - D (1 - m ) C 2 Ú Ú cos cos ÁË1 + cos ˜¯ ÁË1 + cos ˜ dx dy
2 2
a b 00
a b b a ¯
16 p 4 ab
= - D (1 - m ) C 2
a2b2 4
Third term in strain energy
2 (1 - m )
D
2 ÚÚ
w ¢◊2 dx dy
16 p 4 2 px 2 py
= D (1 - m ) C 2 2 2 Ú Ú sin
2
◊ sin 2 dx dy
a b a b
16 p 4 a b
= D (1 - m ) C 2
◊
a 2 b2 2 2
Noting that second and third terms are equal but opposite in sign, we get
V = first term only
È 3 Ê a2 b2 ˆ ˘
4 DC 2 p 4
= Í1 + Á 2 + 2 ˜ ˙
abÍÎ 2 Ë b a ¯ ˙˚
U = Potential Energy
= - Ú Ú q w dx dy
+a 2 b 2
Ê 2 px ˆ Ê 2 py ˆ
=- Ú Ú q0C Á1 + cos
Ë ˜ Á1 + cos
a ¯Ë b ¯
˜ dx dy
- a 2 -b 2
= - q0 Cab
∴ Total Energy I = U + V
4 DC 2 p 4 È 3 Ê a 2 b 2 ˆ ˘
= - q0Cab + Í1 + + ˙
ab ÎÍ 2 ÁË b 2 a 2 ˜¯ ˚˙
∂I
∴ = 0 gives
∂C
8 DC p 4 È 3 Ê a2 b2 ˆ ˘
- q0 ab + Í1 + Á 2 + 2 ˜ ˙
ab ÍÎ 2 Ë b a ¯ ˙˚
q0 a 2 b 2
or C=
È 3 Ê a2 b2 ˆ ˘
8 Dp 4 Í1 + Á 2 + 2 ˜ ˙
Î 2Ëb a ¯˚
Ê 2 px ˆ Ê 2 py ˆ
q0 a 2 b 2 Á1 + cos ˜ cos Á1 + ˜
Ë a ¯ Ë b ¯
∴ w=
È 3 Ê a 2 b2 ˆ ˘
8 DC p 4 Í1 + Á 2 + 2 ˜ ˙
Î 2Ëb a ¯˚
Maximum deflection occurs at middle of the plate i.e. when x = 0 and y = 0.
4 q0 a 2 b 2
wmax =
∴ È 3 Ê a2 b2 ˆ ˘
8 Dp 4 Í1 + Á 2 + 2 ˜ ˙
Î 2Ëb a ¯˚
For square plate a = b,
ENERGY METHOD 115
4 q0 a 2 b 2 q a2b2
wmax = = 0 4
È 3 ˘ 8 Dp
8 Dp 4 Í1 + (1 + 1)˙
Î 2 ˚
q0 a 4
= 0.00128
D
q0 a 4
Exact value is 0.00126 .
D
Note: The following function also satisfies all boundary conditions and hence it may be tried.
2 2 2 2
Ê xˆ Ê xˆ Ê yˆ Ê yˆ
w = C Á ˜ Á1 - ˜ Á ˜ Á1 - ˜
Ë a¯ Ë a¯ Ë b¯ Ë b¯
q0 a 4
in which origin is at the corner of plate. It gives wmax = 0.00133 for a square plate.
D
Example 10.3. By energy method determine expressions for w, Mx, My, Mxy, Qx, Qy, Vx and Vy in the
plate shown in Fig. 10.4, if it is loaded with uniformly distributed load of intensity q0.
a
0 x
y Free edge
Solution. In this, the boundary conditions to be satisfied are

∂2 w
w = 0 and = 0 at x = 0 and x = a ...(1)
∂x 2
∂2 w
w = 0 and = 0 at y = 0 ...(2)
∂y 2
My = 0 and Vy = 0 at y = b. ...(3)
These boundary conditions are satisfied if we select

px
w = Cy sin
a
p2 px
∴ w ¢¢ = -C 2
y sin
a a
= 0
w
∂2 w p px
and w ¢◊ = = C cos
∂x∂y a a
ab
∴ V=
D È
Ú Ú Î (w ¢¢ + w )2 - (w ¢¢w - w ¢◊2 ) 2 (1 - m )˘˚ dx dy
2 00
D È C 2 p 4 2 2 px px ˘
ab
p2
= Ú Ú Í 4 y sin + 2 (1 - m ) C 2 2 cos 2 ˙ dx dy
2 00Î a a a a ˚
D 2 p2
ab
È p2 px px ˘
= C 2 Ú Ú ÍÎ a 2 y
2
sin 2 + 2 (1 - m ) cos 2 ˙ dx dy
2 a 00
a a ˚
È b ˘
D p 2 2 Í p 2 Ï y3 ¸ a
(
ab ˙
= C Ì ˝
ÍÎ a 2 Ó 3 ˛0 2
◊ + 2 1 - m )
2 a2 2 ˙˚
D p 2 2 È p 2b3 a ab ˘
= C Í 2 + 2 (1 - m ) ˙
2 a 2
Î 3a 2 2 ˚
D 2 2 b È p2 b2 ˘
= p C Í + (1 - m )˙
2 aÎ 6 a 2
˚
Potential Energy
ab
U = - Ú Ú q w dx dy
00
ab
px
= - Ú Ú q0Cy sin dx dy
00
a
b
È y2 ˘ 2a
= - q0C Í ˙
Î 2 ˚0 p
b2a
= - q0C
p
ENERGY METHOD 117
∴ I =U +V
b2 a D 2 2 b È p 2 b2 ˘
= - q0 C + p C Í + 1 - m ˙
p 2 a Î 6 a2 ˚
dI
= 0, gives
dC
q0 b 2 a b È p2 b2 ˘
- + CDp 2 Í + 1 - m˙
p aÎ 6 a 2
˚
q0 b 2 a a
∴ C=
p È p b
2 2 ˘
Dp 2 b Í + (1 - m )˙
Î 6 a 2
˚
q0 ba 2
=
È p 2 b2 ˘
Dp 3 Í + (1 - m ) ˙
Î 6 a 2
˚
px
∴ w = Cy sin
a
q0 ba 2 px
= y sin
Èp b 2 ˘ 2 a
Dp 3 Í + (1 - m )˙
Î 6 a 2
˚
Ê ∂2w ∂2w ˆ
∴ Mx = -D Á 2 + m 2 ˜
Ë ∂x ∂y ¯
p2 px
= DC 2
y sin
a a
q0 b px
= y sin
Èp b 2 2 ˘ a
pÍ + (1 - m ) ˙
Î 6 a 2
˚
Ê ∂2 w ∂2 w ˆ
My = -D Á m 2 + 2 ˜
Ë ∂x ∂y ¯
mq 0 b px
=- y sin
Èp b 2 ˘
2 a
pÍ + (1 - m )˙
Î 6 a 2
˚
∂2w
M xy = - D (1 - m )
∂x∂y
p px
= - D (1 - m ) C cos
a a
=-
(1 - m ) qba
cos
px
p 2
È p2b2 ˘ a
Í 2 + (1 - m )˙
Î 6a ˚
∂2w ∂2 w
—2 w = +
∂x 2 ∂y 2
px
q0 by sin
=- a
È p 2 b2 ˘
Dp Í + (1 - m ) ˙
Î 6 a 2
˚
∂ ( 2
∴ Qx = - D — w)
∂x
p px
q0 ◊ by ◊ cos
= a a
È p 2 b2 ˘
pÍ + (1 - m )˙
Î 6 a 2
˚
b
q0 ◊ y
a px
= ◊ cos
Èp b
2 2 ˘ a
Í ◊ 2 + (1 - m ) ˙
Î 6 a ˚
∂ ( 2
Qy = - D — w)
∂y
q0 b px
= sin
È p2 b2 ˘ a
pÍ + (1 - m ) ˙
Î 6 a2 ˚
∂M xy
Vx = Q x +
∂y
b
q0 y
a px
= cos
p b
2 2 a
◊ + (1 - m )
6 a2
ENERGY METHOD 119
∂M xy
Vy = Q y +
∂x
px
q0 b ◊ sin q0 b (1 - m )
a px
= + sin
Èp b
2 2 ˘ Èp b2 2 ˘ a
p Í ◊ 2 + (1 - m )˙ pÍ + (1 - m ) ˙
Î 6 a ˚ Î 6 a 2
˚
q0 b ( 2 - m ) px
= sin
Èp b
2 2 ˘ a
pÍ + (1 - m ) ˙
Î 6 a 2
˚
QUESTIONS
1. Derive the expression for total strain energy in a plate.

2. Determine the expression for deflection in the plate shown in Figure 10.5. Use energy method.
È px Ê py ˆ ˘
Í Hint : Try the function w = c sin ËÁ1 - cos ¯˜ ˙
Î a 2b ˚
a
0 x
Fig. 10.5
Chapter
11
Finite Difference Method
In this method, differential equations are replaced by finite difference equations. The plate is divided
into a grid. The deflection of intersection points of the grid lines are taken as unknowns. The plate
equation, moments, shear etc. are expressed in terms of differences in deflections of neighbouring
points. Using plate equation one equation is formed at each grid point. If any point falls outside the
plate, its deflection is replaced by those of points inside for which boundary conditions are made use.
The set of equations formed are solved to get deflection of each grid point.
In this chapter, first finite difference method Vs. classical method is discussed. Then finite difference
expressions are derived for plate equations and stress resultants. After explaining how to apply boundary
conditions two standard problems are solved and formation of equations for some more cases are
presented.
11.1 FINITE DIFFERENCE METHOD Vs. CLASSICAL METHOD

1. In classical method, exact equations are formed and exact solutions are obtained, whereas in finite
difference method, exact equations are formed but solved approximately.
2. Using classical method solutions are obtained for few standard cases whereas solutions can be
obtained for all problems by finite difference method.
3. Whenever the complexities are encountered, classical method makes the drastic approximations
and then looks for the solution. Various complexities encountered may be grouped into the
following:
(a) Shape
(b) Boundary conditions
(c) Loading.
(a) Irregular shape (b) Irregularboundary conditions (c ) Irregular loading
Fig. 11.1 Complexities in plate analysis

FINITE DIFFERENCE METHOD 121
Figure 11.1 shows typical complexities in the plate analysis. To get the solution in these cases
rectangular shapes, same boundary conditions along a side and regular equivalent loads are assumed.
In finite difference method, no such assumptions are made. The problem is treated as it is.
4. If material property is not isotropic, solutions become difficult in classical method. Only a few
simple cases have been solved successfully. Finite difference method can handle plates with
anisotropic properties without any additional difficulty.
Thus, classical method is good for standard cases while finite difference method is good for the
problems with complexities in shape, boundary conditions and loading.
11.2 FINITE DIFFERENCE FORM FOR DIFFERENTIAL EQUATIONS

Let ‘O’ be the point under consideration (Fig. 11.2) and ‘h’ be the mesh length. The forward points are
referred as 1, 2, 3… and the backward points as −1, −2, −3... . Thus, the deflection of point ‘O’ is
w0 and at points 1, 2 and 3 the defections are w1, w2, w3. Similarly, deflections at points −1, −2, −3 are
∂w
w−1, w−2, w−3. The slope at point O is approximated as
∂x
Ê ∂w ˆ
ÁË ˜¯ = Average slope at ‘O’ towards its left and right side
∂x 0
1 È w0 - w-1 w1 - w0 ˘
= +
2 ÍÎ h h ˙
˚
w-3
-3
h w-2
-2
(c) Cross section along y-axis

w-1
h
-1
w0
h
0
-3 -2 -1 1 2 3 4 5
w1
h h h
h 1
w2
2
w3
( a) Plan y-direction
x-direction
w- 3 w-2 w-1 w0 w1 w2 w3 w4 w5
(b) Cross section along x-axis
Fig. 11.2
( w - w -1 )
1
=
2h 1
∂w
This finite difference form of may be remembered in the form
∂x
¶w 1
= -1 0 1 …eqn. (11.1)
¶x 0 2h w
in which double circled point refers to the point under consideration.

Ê ∂2w ˆ
Similarly, Á 2 ˜ at point ‘O’ may be approximated as,
Ë ∂x ¯
∂2 w ∂ Ê ∂w ˆ
=Á ˜
∂x 2 ∂x Ë ∂x ¯
= rate of change of slope between points h/2 and −h/2.
1 È w1 - w0 w0 - w -1 ˘
= -
h ÍÎ h h ˙
˚
w-1 - 2 w0 + w1
=
h2
1
= 2 1 -2 1 …eqn. (11.2)
h w
∂2 w
It may be noted that could have been taken as rate of change of slope between the points h and
∂x 2
−h also. But in that case, as the points are far away approximation is more. Since, the slopes at middle
of the range 0 to 1 and −1 to 0 could be expressed in terms of deflections at grid points, equation 11.2
is more preferred form. Similarly
Ê ∂3 w ˆ ∂ Ê ∂2 w ˆ
ÁË 3 ˜¯ = ∂x ÁË 2 ˜¯
∂x 0 ∂x
1 ÈÊ ∂2 w ˆ Ê ∂2 w ˆ ˘
= Í - ˙
2 h ÎÍ ËÁ ∂x 2 ¯˜ 1 ËÁ ∂x 2 ¯˜ -1 ˚˙
Ê ∂3w ˆ 1 È w0 - 2 w1 + w2 w-2 - 2 w-1 + w0 ˘

ÁË 3 ˜¯ = Í - ˙
∂x 0 2 h Î h2 h2 ˚
1
= ÈÎ- w-2 + 2 w-1 - 2 w1 + w2 ˘˚
2h3
1
= 3
-1 2 0 -2 1 …eqn. (11.3)
2h w
The above expression may be derived conveniently by the following pattern operation also.
Ê ∂3 w ˆ ∂ Ê ∂2 w ˆ
ÁË 3 ˜¯ = ∂x ÁË 2 ˜¯
∂x 0 ∂x
1 1
= -1 0 1 2
1 -2 1
2h h w
1
= 3 -1 2 -1
2h
+ 0 0 0
+ 1 -2 1
w
1
= 3 -1 2 0 -2 1
2h w
Note in the above operation, first ‘O’ times second pattern is placed (second row). Then −1 times
second pattern is placed at one step backward (refer first row) and then 1 times second pattern is placed
a step forward (refer third row).
∂4w
Using pattern operation techniques may be derived as shown below:
∂x 4
Ê ∂4 w ˆ ∂2 Ê ∂2w ˆ
ÁË 4 ˜¯ = 2 ÁË 2 ˜¯
∂x 0 ∂x ∂x
1 1
= 1 -2 1 2 1 -2 1
h
2
h w
1
= 4 1 -2 1
h
+ -2 4 -2
+ 1 -2 1
w
1
= 4
1 -4 6 -4 1 …eqn. (11.4)
h w
It may be observed that odd differentiations are having anti-symmetric patterns and even
differentiation terms are having symmetric pattern.
Similarly, the differentiations w.r.t. y may be expressed by turning the patterns of differentiations
w.r.t. x by 90°. Thus,
-1 1
2
¶w 1 ¶w 1
= 0 = 2 -2
¶y 0 2h ¶y
2
h
1 1
w w
-1 1
2 -4
3 4
¶w = 1 and ¶ w4 = 4
1
3 3 0 6
¶y 2h ¶y h
-2 -4
1 1
w w
11.3 FINITE DIFFERENCE FORM FOR PLATE EQUATION

We know,
2 2
¶w = 1 and ¶ w2 = 2
1
2 1 -2 1 -2
h
2
¶x h ¶y
1
w
∂2w ∂2w
∴ —2 w = +
∂x 2 ∂y 2
1 1
1 1 1
= 1 -2 1 + -2 = 2 1 -4 1
h
2
h
2 h
w
1 1
w w
∴ — 4 w = — 2 (— 2 w )
1 1
= 12 1 -4 1
1
2 1 -4 1
h h
1 1
w
-4
1+1 -4 1+1
1 -4 1 -4
1 -4
1+16+1 1
h
4
1 -4
-4
1+1 -4 1+1
1
w
2 -8 2
1 1 -8 20 -8 1
= 4
h
2 -8 2
1
w
Hence, the plate equation is,
2 -8 2
4
qh
1 -8 20 -8 1 = …eqn. (11.5)
D
2 -8 2
1
w
11.4 EXPRESSIONS FOR STRESS RESULTANTS
Ê ∂2 w ∂2w ˆ
Mx = -D Á 2 + m 2 ˜
Ë ∂x ∂y ¯
-D 1 -2 1 + m -2
= 2
h
1
w
= -D2
1 -2-2m 1
h
m
w
Similarly, My = -D2
m -2-2m m
h
1
w
∂2w ∂ Ê ∂w ˆ
M xy = - D (1 - m ) = - D (1 - m ) Á
∂x ∂ y ∂x Ë ∂y ˜¯
-1
= -D (1 - m) 1 -1 0 1 1 0
2h 2h
1
w
1 0 -1
- D(1 - m)
= 2 0 0 0
4h
-1 0 1
w
It may be easily seen that,

∂ ( 2
Qx = - D — w)
∂x
-1 0 1
-D
= 3
-1 4 0 -4 1
2h
-1 0 1
w
∂
Vx = Q x +
∂y
(M xy )
-2 0 2-m
+m
-D 6 -6
= 3
-1 -2m 0 +2m 1
2h
-2 0 2-m
+m
w
By rotating Qx and Vx patterns by 90°, we get finite difference patterns for Qy and Vy.
11.5 APPLYING BOUNDARY CONDITIONS

While forming the plate equations at points close to boundary, many points fall outside the plate. The
deflections of such imaginary points should be replaced by the deflections of the points on the plate.
For this boundary conditions are to be used. In this article, the expressions for the deflections of
imaginary points in terms of real points are derived for the following boundaries:
(a) Simply supported edge
(b) Fixed edge
(c) Free edge
(a) Simply supported edge:
1 0 3
Edge under consideration
Fig. 11.3
Consider the point ‘O’ shown in Fig. 11.3. Since, it is on a simply supported edge
Mx|0 = 0
-D -2
2 1 -2m 1 = 0.
h
m
w
Since, displacements are zero along simply supported edge, we get

w1 + w3 = 0
w3
i.e. w1 = −w3 w1
i.e. displacement of imaginary point is equal to negative of that of image point.
(b) Fixed edge:

∂w
If edge under consideration (Refer Fig. 11.3) is fixed, the boundary conditions are w = 0 and = 0,
∂x
along the edge.
∂w
=0
∂x 0
1
i.e. - -1 0 1 =0
2h
w
∴ −w1 + w3 = 0
w-1 w1
i.e. w1 = w3
i.e. in this case, displacement of imaginary point is equal to displacement of image point.
(c) Free edge:
At free edge the boundary conditions are
Mx = 0 and Vx = 0
Mx = 0 gives 1 -2-2m 1 =0
w
m
-m
i.e. w1 = 2+2m -1
-m
Vx = 0 gives
-2 0 2-m
+m
-1 6 0 -6
-2m +2m 1 =0
-2 0 2-m
+m w
2-m 0 -2
+m
wimaginary = -6 0 6-2m -1
+2m
w
2-m 0 -2
+m
Note carefully that double circled point is under consideration.

Example 11.1 A fixed plate of size 4h × 4h is subjected to uniformly distributed load q0 over its entire
surface. Taking grid size as h × h determine.
(i) Deflection at the centre of the plate
(ii) Moment at the centre of the plate take µ = 0.3
Solution: Making use of symmetry, there are only three unknown displacements w1, w2 and w3 as
shown in Fig. 11.4. Displacements are zero along the support.
a = 4h
3 2 3
a = 4h
2 1 2
3 2 3
Since, slope is zero at fixed edges, we get
-1 0 1 =0
w
∴ w−1 = w1
i.e. displacement of imaginary point is equal to displacement of image point. Keeping these points in
mind, plate equations can be written for the points 1, 2 and 3. The plate equation is
2 -8 2
4
qh
1 -8 20 -8 1 =
D
2 -8 2 w
1
Hence, the plate equation at point 1 is

q0 h 4
20w1 − 4 × 8w2 + 4 × 2w3 =
D
q0 h 4
i.e. 20w1 −32w2 + 8w3 = …(1)
D
The plate equation at point 2 is,
qh 4
20w2 − 8w3 − 8w3 − 8w1 + 2w2 + 2w2 + w2 + w2 =
D
qh 4
i.e. −8w1 + 26w2 − 16w3 = …(2)
D
The plate equation for point 3 is,
q0 h 4
20w3 − 8w2 − 8w2 + 2w1 + w3 + w3 + w3 + w3 =
D
q0 h 4
i.e. 2w1 − 16w2 + 24w3 = …(3)
D
Solving above three simultaneous equations, we get
q0 h 4
w1 = 0.4607
D
q0 h 4
w2 = 0.3090
D
q0 h 4
and w3 = 0.2093
D
q0 h 4 q0 a 4
Thus, deflection at centre = 0.4607 = 0.0018
D D
Moment at centre
= -D2 1 -2-2m 1
h
m
w
ÈÎ( -2 - 2 m ) w1 + mw2 + w2 + mw2 + w2 ˘˚

D
=-
h2
D q0 h 4
=- 2
[ -2.6 ¥ 0.4607 + 2.6 ¥ 0.3090]
h D
= 0.3944 q0h2 = 0.02465 q0a4
Example 11.2 Analyse a simply supported plate of size 4h × 4h which is subjected to udl q0. Take grid
size h × h and determine.
(a) Central deflection
(b) Moment at the centre if µ = 0.3.
Solution. Due to symmetry there are only three unknown grid point displacements as shown in
Fig. 11.5. It is to be noted that
1. Displacements of points on boundary = 0
2. Moment about boundary = 0.
3 2 3
2 1 2
3 2 3
i.e. -D 0 = 0
2
1 1
h
m
w
Since, displacements are zero about boundary line this condition gives displacement of an imaginary
point is equal to negative of displacement of image point.
Keeping these two points in mind, plate equation is written for the three grid points.
q0 h 4
For first point, 20w1 − 4 × 8w2 + 4 × 2w1 = …(1)
D
q0 h 4
For second point, 20w2 − 8w3 − 8w1 − 8w3 + 2w2 + 2w2 + (−w2) + w2 =
D
q0 h 4
i.e. − 8w1 + 24w2 − 16w3 = …(2)
D
For point three,
q0 h 4
20w3 − 8w2 − 8w2 + 2w1 − w3 − w3 + w3 + w3 =
D
q0 h 4
i.e. 2w1 − 16w2 + 20w3 = …(3)
D
Solving simultaneous equations 1, 2 and 3, we get
q0 h 4
w1 = 1.0313
D
q0 h 4
w2 = 0.75
D
q0 h 4
and w3 = 0.5469
D
q0 h 4
Thus, deflection at centre = w1 = 1.0313
D
q0 a 4
= 0.0645
D
Moment at centre, if µ = 0.3 is given by
= - 2D 1 -2-2m 1
h
w
ÈÎ( -2 - 2 m ) w1 + mw2 + w2 + mw2 + w2 ˘˚

D
=-
h2
D
=- ÎÈ-2.6 w1 + 2.6 w2 ˚˘
h2
= [ -2.6 ¥ 1.0313 + 2.6 ¥ 0.75] q0 h 2
= 0.7314 q0 h 2 = 0.0457 q0 a 4
Example 11.3 Formulate finite difference equations for the plate shown in Fig. 11.6. Take grid size
h × h.
1 2
3 4 5
6 7 8
Solution. In this problem, deflections of all interior grid points are different. Hence, eight separate
equations are to be written. Noting that all edges are fixed
w = 0 at points on edges
and deflection of imaginary point = Deflection of image point.
Keeping these points in mind the eight equations are written.
For point 1,
q0 h 4
20w1 − 8w2 − 8w3 + 2w4 + w1 + w1 + w6 =
D
q0 h 4
22w1 − 8w2 − 8w3 + w6 = …(1)
D
For point 2,
q0 h 4
20w2 − 8w4 − 8w1 + 2w5 + 2w3 + w2 + w7 =
D
q0 h 4
i.e. −8w1 + 21w2 + 2w3 − 8w4 + 2w5 + w7 = …(2)
D
For point 3,
q0 h 4
20w3 − 8w1 − 8w4 − 8w6 + 2w2 + 2w7 + 2w5 + w3 =
D
q0 h 4
i.e. −8w1 + 2w2 + 21w3 − 8w4 + 2w5 − 8w6 + 2w7 = …(3)
D
For point 4,
q0 h 4
20w4 − 8w2 − 8w5 − 8w7 − 8w3 + 2w1 + 2w8 + 2w6 =
D
q0 h 4
i.e. 2w1 − 8w2 − 8w3 + 20w4 − 8w5 + 2w6 − 8w7 + 2w8 = …(4)
D
For point 5,
q0 h 4
20w5 − 8w8 − 8w4 + 2w2 + 2w7 + w5 + w3 =
D
q0 h 4
i.e. 2w2 + w3 − 8w4 + 20w5 + 2w7 − 8w8 = …(5)
D
For point 6,
q0 h 4
20w6 − 8w3 − 8w7 + 2w4 + w1 + w8 + w6 + w6 =
D
q0 h 4
i.e. w1 − 8w3 + 2w4 + 22w6 − 8w7 + w8 = …(6)
D
For point 7,
q0 h 4
20w7 − 8w4 − 8w8 − 8w6 + 2w3 + 2w5 + w2 + w7 =
D
q0 h 4
w2 + 2w3 − 8w4 + 2w5 − 8w6 + 21w7 − 8w8 = ...(7)
D
For point 8,
q0 h 4
20w8 − 8w5 − 8w7 + 2w4 + w8 + w8 + w6 =
D
q0 h 4
2w4 − 8w5 + w6 − 8w7 + 22w8 = …(8)
D
11.6 MODIFICATION OF FINITE DIFFERENCE PLATE EQUATION TO APPLY IT ON

A FREE EDGE
Fig. 11.7 Modifying plate equation so as to apply at point ‘O’
The plate equation for any point is
2 -8 2
4
1 -8 20 -8 1 q0h
= 4
D
w
2 -8 2
When this equation is to be applied at point, four points fall away from the plate, one at distance 2h
and three at distance h from free edge of the plate. The plate equation can be modified making use of
the boundary conditions that
Vx = 0
and Mx = 0 at free edges.

Using the boundary condition Vx = 0
-2 + m 0 2-m
-1 6 - 2m 0 -6 + 2 m 1 = 0.
w
-2 + m 0 2-m
-2 + m 0 2-m
i.e. w= 6 - 2m 0 -6 + 2 m 1
-2 + m 0 2-m
Substituting it in plate equation, we get
-2 + m 0 2-m
2 -8 2
6 - 2m 0 -6 + 2 m 1 q0h
4
-8 20 -8 1 =
D
w
2 -8 2
-2 + m 0 2-m
1
i.e.
1
m -8 4-m
4
qh
-2 - 2 m 20 -14 + 2 m 2 =
D
w
m -8 4-m
1
From boundary condition Mx = 0, we get
-m
Displacement of imaginary point = 2 + 2m 1

w
-m
Applying this condition to the three points which are at distance ‘h’ from free edge on fictitious plate
we get
2
-m 1
m(2 + 2m)
-m
-8 4-m
-m(-2 - 2m)
2
-m 2 + 2m
4
qh
20 -14 + 2 m 2 =
D
(2 + 2m)(-2 - 2m)
2
-m
-m(-2 - 2m) -8 4-m w
-m
m(2 + 2m)
2
-m 1
2
1-m
4 - 2m
2
-8 + 4m + 4m
4
qh
16 - 8m - 6m
2
-12 + 4m 2 =
D
w
2
-8 + 4m + 4m 4 - 2m
2
1-m
QUESTIONS
1. Derive the plate equation in finite difference form.

2. Modify the plate equation so as to apply it on a boundary with free edge.
3. Write plate equations in finite difference form for the plate shown in Fig. 11.8.
[Hint: Note due to symmetry w1 = w7, w2 = w8, w3 = w9]
B C
1 2 3
4 5 6
7 8 9
A D
Fig. 11.8 Question No. 3

4. In the plate shown in Fig. 11.8, if the edge is simply supported, write plate equations.
[Hint: Note due to anti-symmetry w1 = w9, w2 = w6, w4 = w8]
Chapter
12
A Folded Plate Roofs
Folded plate roof is a structure which is composed of a number of plates monolithic along their
common longitudinal edges. It may be looked as a plate folded at several longitudinal lines. The roof
unit is supported at ends by thin but deep frames. Such support may be treated as simple support
(Fig. 12.1). The end frames in turn are supported by columns. Figure 12.2 shows commonly used folded
plate roofs.
(a) Prismatic
(b) Vee-type
L
(c ) Trough type
End frame
(d ) North light
Fig. 12.1 A typical folded plate roof Fig. 12.2 Types of folded plate roofs
Folded plate roofs are known as hipped plates, prismatic shells and faltwerke. By giving folds to
the plate, bending moment is reduced in the plate and considerable load is transferred as membrane
compressions. Hence, they can be used economically to cover a column free span of 20 to 25m.
However, folded plates are not as efficient as shells in transferring loads by membrane compression.
A FOLDED PLATE ROOFS 143
12.1 ADVANTAGES AND DISADVANTAGES OF FOLDED PLATES OVER SHELLS

The advantages of folded plates over shells are:
1. Form work required is relatively simple.
2. Movable form can be employed.
3. Design involves simpler calculations.
The disadvantages of folded plates over shells are:
1. Folded plates consume a little more material than shells.
2. Form work can be removed after 7 days of concreting while in case of shells it may be removed
after 3 days only.
12.2 HISTORICAL NOTE

Ehler of Germany published first paper on folded plate bunkers in 1930. He assumed the joints at folds
as hinged. In 1934, Gruber put forward a rigorous theory which took into account the rigidity of the
joints and the relative displacement between them. In the course of next few years, many European
engineers made contributions to the subject—noteworthy among them being by Cramer Ohlig and
Girkman. The methods developed were based on theory of elasticity. Hence, they involve differential
and algebraic equations.
After world war II, American took interest in folded plate analysis. In 1947, Winter and Pei published
stress distribution procedure. It ignored relative displacements of the joints. In 1954, Gaafer published
a simple method for analysis, which took into account joint displacements also. Later on Simpsons
improved it and now that method is known as Simpson’s method. Girkmans method, improved by
Whitney is known as Whitney’s method.
12.3 ASSUMPTIONS
The following assumptions are made in the analysis of folded plates:
1. It consists of rectangular plates each being of uniform thickness.
2. The structure is monolithic and the joints are rigid.
3. The material is elastic, homogeneous and isotropic.
4. The length of each plate is more than twice its width.
5. In all plates, plane section remains plane even after deformation.
12.4 BEHAVIOUR OF FOLDED PLATE ROOFS

Load transfer in folded plate roofs consists of slab action and plate action.
12.4.1 Slab Action
Each slab may be assumed to bend as simply supported slab between the adjacent folds. It results into
the reactions at joints (folds). But actually the slab is continuous over folds. Hence, there are end
moments in transverse direction. Taking joint moments as unknowns, joint-reactions may be found in
terms of unknown moments. Thus, total reactions at joints if they are supported externally are the
algebraic sum of reactions due to load plus those due to transverse moments. Figure 12.3 shows slab
action of a typical folded plate roof.
bn+1
M n+1
bn qn+1
Mn
-1
qn (n+1) Rn+1
Mn Rn
qn
(n)
Rn-1
qn-1
bn - width of nth plate
)
-1
qn - Angle made by nth plate with horizontal

(n
Fig. 12.3
12.4.2 Plate Action

Actually there is no external supports at folds. Hence, the joint reactions found due to slab action
are reversed and applied as joint force. This joint loads get distributed as plate load as shown in
Figure 12.4.
Rn Rn+1
Rn-1
P²n P¢(n+1)
P¢n
P²(n-1)
P²(n-1)
P¢(n-2)
Fig. 12.4
Pn
P n-1
Pn+
1
n
n-1
n-2
Tn-1
2
Tn
n-
P
Tn-2
Fig. 12.5
Assuming each plate is simply supported on end frames, bending moment in the plate due to plate
loads can be found. This plate bending results into discontinuity at joints. Since, the joints are rigid, this
discontinuity is not possible. This has happened because longitudinal shears developed at joints are
ignored. Let the longitudinal shears developed be as shown in Fig. 12.5.
Taking longitudinal shears also as unknowns, equations are developed to find stresses at edges of
each plate. Using compatibility conditions, unknown transverse moments and longitudinal shears are
determined.
There are two different methods for the analysis of plate. They are known as Whitney’s method and
Simpson’s method. In Whitney’s method simultaneous equations are solved to get transverse moments
and longitudinal shears whereas in Simpson’s method effect of transverse moments is taken care by
moment distribution procedure while the effect of longitudinal shears is taken care by a stress distribu-
tion procedure. These methods are explained in this chapter.
12.5 WHITNEY’S METHOD

At each joint, there are two unknowns—a transverse moment and a longitudinal shear. Hence, if there
are N number of joints (i.e. N + 1 number of plates), there are 2N number of unknowns. These unknowns
at joints n − 1, n and n + 1 are shown in Fig. 12.6.
The transverse moments at joint 1 and joint N may be taken as those due to cantilever action of 1st
and n + 1th plate. Hence, two of the 2N unknowns are found. Thus, number of unknown reduce to
2N−2. In Whitney’s method, N-number of equations are obtained by equating stresses in adjoining
plates at common edges.
Another compatibility condition to be satisfied is at any joint the angle between the two adjacent
plates should not change. This condition may be applied from joint 2 to joint N−1 to get N−2 equations.
Thus, the compatibility of stresses and transverse rotations gives 2N−2 equations. Hence, 2N−2
unknowns, namely, M2, M3, ..., MN−1 and T1, T2, ..., TN can be determined.
Tn Tn+1
Tn+2
Tn–1 (n+1) (n+2)

Mn+1
Mn (n+3)
(n) Mn+2 Tn+3
Mn – 1
1)
(n– Mn+3
Fig. 12.6 Unknowns in Whitney’s method
The step by step procedure of analysis of folded plates by Whitney’s method is given below:
Step. 1: Express the loading in the Fourier form:
4 m px
q ( x) = Â q0 sin
mp a
If only first term is considered,
4 q0 px px
q ( x) =
sin = q0 sin ...eqn. 12.1
p L L
Step 2: Assuming plates are simply supported along the folds, find reactions (Refer Fig. 12.7)
qn+1 n +1
n qn+2
qn
+
Rn+1 Rn+1
¢+
n -1 R¢n Rn
¢+
¢
Fig. 12.7 Load and reactions at mid span due to loads for slab action
qn bn + qn +1 bn +1 px ...eqn. 12.2
qn = sin
2 L
Step 3: To be compatible with sinusoidal loading transverse moments should vary in the sinusoidal
form. Then the reaction at joint n due to transverse moments (Refer Fig. 12.8) are
È M - M n -1 M n - M n+1 ˘ px
Rn¢¢ = Í n + sin ...eqn. 12.3
Î bn cos f n bn +1 cos f n +1 ˙˚ L
Step 4: Calculate total reaction at the imaginary support along the folds.
Rn = Rn¢ + Rn¢¢ ...eqn. 12.4
Mn+1
Mn
Mn-1
Mn - Mn+1
bn+1 cosfn+1
Mn-Mn-1
bn cos fn
Fig. 12.8 Reactions due to transverse moment at mid span at joint n
Step 5: Actually there is no external support at folds. Hence, the joint reaction Rn acts as joint load in
the downward direction. These joint loads are transferred to the plates meeting at joints causing each
plate to bend as a beam spanning between the end traverses.
px
Rn sin
L
a
(a) (b)
gn q0 – fn
Pn¢ q0 + fn+1
Rn
Pn¢ n ²
Pn+1
fn+1
(c) b
gn
²
fn Pn+1
(n)
C
fn+1
( a) R n and its components at joint n (b) Vector diagram
Fig. 12.9 Finding components of Rn
Referring to Fig. 12.9, if ab represents the joint load Rn, its components in the direction of nth and
nth plate are obtained by traversing from a to b in the directions of nth plate and (n+1)th plate, ac
represents the component in nth plate while cb represents the component in (n+1)th plate. From triangle
of forces abc, we get
px
Rn sin Rn px
Pn¢ = ac =
sin g n
(
L sin 90 + f
n +1 ) =
sin g n
cos f n +1 ◊ sin
L
px
Rn sin Rn px
Pn¢¢+1 = cb =
sin g n
( n)
L sin 90 - f =
sin g n
cos f n ◊ sin
L
Similarly, if we resolve the joint load Rn−1 at joint n − 1, the load Pn″ is given by
px
Rn-1 sin
Pn¢¢ = L cos f
n -1
sin rn-1
∴ Total force in nth plate from n to n − 1 direction is (Ref. Fig. 12.5)
È R cos f n +1 Rn -1 ˘ px
Pn = Pn¢ - Pn¢¢ = Í n - ◊ cos f n -1 ˙ sin ...eqn. 12.5
Î sin g n sin g n -1 ˚ L
where γn is the angle of deviation of (n + 1)th plate from the direction of nth plate. From equation 12.5,
plate loads in all plates may be found.
Step 6: Since, the longitudinal variation of plate loads are in sinusoidal form, the plate moment also
should be in the sinusoidal form. Thus,
px L2
¢ = Pn
M pn sin ...eqn. 12.6
p L 2
Step 7: The moment due to the longitudinal shears Tn−1 and Tn at the centre of the plate (Refer
Fig. 12.10) is
b px
¢¢ = (Tn -1 + Tn ) n ◊ sin
M pn ...eqn. 12.7
2 L
te
pla
n+ 1)th
(
t e
pla
nth
T n+1
Tn
T n-1
Fig. 12.10
px
It may be noted that Mpn to be in sin form, Tn also should vary in sinusoidal form.
L
Step 8: Total plate moment is
È P L2 b ˘ px
M pn = Í n 2 + (Tn -1 + Tn ) n ˙ sin ...eqn. 12.8
Î p 2 ˚ L
Step 9: Writing tensile stress as positive, the fibre stress in nth plate at common edge n is,
6 È Pn L2 b ˘ px Tn - Tn -1 px
= 2 Í
- 2 + (Tn -1 + Tn ) n ˙ sin + sin ...eqn. 12.9
bn hn Î p 2˚ L bn hn L
where first part is due to flexure while second part is due to direct force (Tn+1 − Tn).
Similarly, the tensile stress at the same fibre calculated from (n + 1)th plate is
È L2 bn +1 ˘ px Tn +1 - Tn px
( )
6
= Í Pn +1 - Tn + Tn +1 ˙ sin + sin ...eqn. 12.10
bn +1hn +1 Î
2
p 2 2 ˚ L b h
n +1 n +1 L
To maintain stress compatibility, equation 12.9 should be equal to equation 12.10. Hence,
6 È L2 bn ˘ Tn - Tn -1
Í - Pn 2 + (Tn -1 + Tn ) ˙+
bn hn Î
2
p 2˚ bn hn
È L2 bn +1 ˘ Tn +1 - Tn
n +1 2 - (Tn + Tn +1 )
6
= Í P ˙+ ...eqn. 12.11
bn +1hn +1 Î
2
p 2 ˚ bn +1hn +1
Applying the above compatibility condition from joint 1 to joint N, N equations can be assembled.
Equation 12.11 is known as Equation of Three shears.
Step 10: Rotation of nth plate and (n+1)th plate due to various causes are,
(a) Due to slab action, for applied loads
1 q n bn3 px
a n , n -1 = - cos f n sin
24 EI n¢ L
1 q n +1bn3+1 px
a n , n +1 = cos f n +1 sin
24 EI n¢ +1 L
1 3
where I n¢ = h , moment of inertia per unit length of plate.
12 n
(b) Due to slab action, for transverse moments
bn px
b n , n -1 = È 2 M + M n -1 ˚˘ sin
6 EI n¢ Î n L
bn +1 px
b n , n +1 = - È 2 M n + M n +1 ˚˘ sin
Î
¢
6 EI n +1 L
(c) Slope due to plate deflection:
n+1 gn+1
gn
n
e
n+1
d k
1 g n-1
-
n n
b
f
g
a
-1 c
n
Fig. 12.11
Referring to Fig. 12.11, let ∆ be the free deflection of plates. Then,

∆n−1 = n − 1, a
∆n= n −1, b = n, d
∆n+1 = n, e
Since, the joints are rigid, common edges cannot be at two different points, in other words, free
expansion of plates is not possible. Hence, joint n−1 moves to the point c and the joint n moves to the
point f, which are located by dropping perpendiculars from a, b and e, d respectively. Line cf shows the
position of the plate after deflection.
(i) To find bc:

From triangle n − 1, b, g, we get
n - 1, b Dn
n - 1, g = =
cos g n -1 cos g n -1
∴ a, g = n − 1, a — n − 1, g
Dn
= D n -1 -
cos g n -1
ag 1 È Dn ˘
gc = = Í D n -1 -
∴ sin g n -1 sin g n -1 Î cos g n -1 ˙˚
∴ bc = bg + gc
1 È Dn ˘
= D n tan g n -1 + Í D n -1 -
sin g n -1 Î cos g n -1 ˙˚
(ii) To find df:
en D
nk = = n +1
cos g n cos g n
∴ dk = dn − nk
D n +1
= Dn -
cos g n
∴ df = dk cot γn
È D ˘
= Í D n - n +1 ˙ cot g n
Î cos g n ˚
Now, rotation of nth plate is given by
( df - bc )
1
qn =
bn
1 ÈÊ D n +1 ˆ 1 Ê Dn ˆ ˘
= ÍÁ D n - ˜ cot g n - D n tan g n -1 - Á D n -1 - ˙
bn ÎÍ Ë cos g n ¯ sin g n -1 Ë cos g n -1 ˜¯ ˚˙
1 È Ê 1 ˆ D n -1 D ˘
= Í D n Á cot g n - tan g n -1 + ˜ - - n +1 cot g n ˙
bn Î Ë sin g n -1 cos g n -1 ¯ sin g n -1 cos g n ˚
1 È Ê 1 - sin 2 g n -1 ˆ D n -1 D ˘
= Í D n Á cot g n + ˜- - n +1 ˙
bn ÎÍ Ë sin g n -1 cos g n -1 ¯ sin g n -1 sin g n ˚˙
È D n -1 D ˘
Í D n (cot g n + cot g n -1 ) - sin g
1
= - n +1 ˙
bn Î n -1 sin g n˚
Similarly,
1 È Dn D n+2 ˘
q n +1 = Í D n +1 (cot g n +1 + cot g n ) - -
bn +1 Î sin g n sin g n +1 ˙˚
Since, joints are rigid, the following compatibility condition should be satisfied:
αn + βn + θn = αn+1 + βn+1 + θn+1 ...eqn. 12.12
Equation 12.12 may be applied from joint 2 to joint N − 1 to get N − 2 equations.
Step 11: Thus, from equations 12.11 and 12.12, we get 2N − 2 equations in 2N − 2 unknowns. Solving
simultaneous equations the unknowns M2, M3, ..., MN−1, T1, T2, ..., TN can be found.
Step 12: After knowing transverse moments, reinforcement required in transverse direction may be
px
found. Spacing may be increased towards end frames, since, these moments vary in the form sin .
L
Knowing longitudinal stresses from equation 12.9, longitudinal reinforcement may be decided. From
plate action, end shear may be found and the diagonal shear reinforcement decided. Figure 12.1 shows
typical reinforcements in a tough type folded plate.
Note: Many times instead of bending transverse steel up and down, two layers of transverse steel are
provided longitudinal steels are located so as to support transverse steel.
Shear
reinforcement End frame
4/4
L/2
Transverse
steel
Longitudinal
steel
CL
Fig. 12.12 R.C. Details of a folded plate roof

12.6 SIMPSONS METHOD

In Whitney’s method, there is need for solving 2N − 2 simultaneous equations to get transverse moments
and longitudinal shears at folds. In Simpson’s method, solution of simultaneous equations is avoided by
going through moment distribution and stress distribution procedure. Stress distribution procedure is
similar to moment distribution. It is used to ensure the continuity of stresses calculated for plate action
neglected the effect of longitudinal stresses. This technique is in two parts—first part consists in finding
stress continuity when the joints do not deflect from their original positions and the second part allowing
joint deflections. The method is outlined below with reference to the folded plate shown in Fig. 12.13.
1 2
0 b3
h3
5
3 4
Fig. 12.13
Step 1: Consider a transverse section of unit length at mid span. Assuming that the joints do not deflect
arrive at joint moments by moment distribution. Calculate the reactions at the joints and apply forces
equal and opposite to these as joint loads. Resolve these joint loads into plate loads. Calculate the
bending stresses caused by the plate loads, assuming each plate to be free to bend independently. These
stresses may be referred as free edge stresses.
Next establish stress compatibility at the common edges of adjacent plates by stress distribution. The
resulting stresses in the plates are those which develop if the joints do not deflect. This solution is
referred as no rotation solution. The solution up to this point may be referred as Winter and Pei solution,
Winter and Pei are the researchers who stopped the analysis at this stage.
Step 2: The effect of joint displacements is to be accounted by considering the rotations of plates 2, 3
and 4. The first and the last plates are treated as cantilevers.
Let joint 2 deflect by an arbitrary amount ∆20 below the level of joint 1 (Fig. 12.14). As a result of
3 EJ 2 D 20 3 EJ 2 y 20
it fixed end moment developed at joint 2 is = , where J2 is moment of inertia of the
b22 b2
Ê 1 3ˆ D 20
plate ÁË = h ˜¯ per unit length and y 20 = . As ∆20 is
12 b2 1
arbitrary, ψ20 is an arbitrary rotation of plate 2. Let ψ20 be D20
such that the magnitude of the moment introduced is, say 2
300 units. The arbitrary rotations and the actual rotations

of the plate are clearly related by an unknown constant, b2
say K2, such that ψ2 = K2ψ20. The arbitrary moment of
300 at joint 2 is next distributed by the moment distribution Fig. 12.14
procedure. The resulting joint moments and reactions are found. Forces equal and opposite to the
reactions are applied at the joints and resolved along the plates as plate loads. The free edge stresses
caused by these loads are next determined. The stress compatibility at the common edges is realized by
stress distribution. The resulting stresses shall be referred as case II solution corresponding to an
arbitrary rotation of plate 2.
Step 3: Now consider the effect of an arbitrary rotation of plate 3 (Ref. Fig. 2.15). As before,
ψ3 = k3 ψ30.
6 EJ 3 D 30 6 EJ 3 y 30
The moment introduced at joint 2, is =
b32 b3
Let the arbitrary rotation ψ30 be such that the magnitude of the moment induced is 600 units.
Distribute the moments of 600 units each at joint 2 and 3 by moment distribution procedure. Arrive
at the reactions at the joints and apply forces opposite to these at the joints. Resolve these forces into
plate loads and compute the free edge stresses. Correct them by stress distribution to secure stress
compatibility. The resulting stresses are referred as case III solution.
b3
D30
Fig. 12.15
Step 4: The case IV solution correspond to an arbitrary rotation ψ40 of plate 4. It is worked out in the
same manner as the case III solution. Again, we note that
ψ4 = k4 ψ40 ...eqn. 12.13
Step 5: The plate deflections ∆n are next worked out. ∆n consists
of the deflection corresponding to the no rotation solution plus fn fb
k2 times the deflection due to case II solution, plus k3 times the n
deflection resulting from the case III solution, plus k4 times the
deflection corresponding to the case IV solution.
If edge stresses are known, the corresponding deflections may
be calculated as explained below:
Let the stresses at n−1 and nth joint be fn−1 and fn respectively
as shown in Fig. 12.16. From the figure, it is clear that,
fn -1 - fn n+1
bending stress = fb = fn-1
2
Fig. 12.16
Noting that moment varies in sinusoidal form,

∂ 2 (d ) px
EI = - M sin
∂y 2 L
L2 M px
d= sin
p EI
2
L
i.e. mid span deflection, due to moment is
L2 M
d=
p 2 EI
M f 2f I
But = b or M = b , h is depth of plate.
I h2 hn
L2 2 fb I
∴ d= ◊
p 2 EIhn
2 L2 1 È f n -1 - f n ˘
=
p 2 EIhn ÍÎ 2 ˙˚
L2
= ÎÈ f n -1 - f n ˚˘ ...eqn. 12.14a
p 2 Ehn
If closed form solution is used,
5 L2
d= ( f - fn )
48 Ehn n -1
...eqn. 12.14b
Total deflection,
∆n = δ0 + k2 δ20 + k3 δ30 + k4 δ40
Step 6: From the results of step 5, the plate rotations may be calculated using the formula given below:
È D n +1 D n -1 ˘
Í D n (cot g n + cot g n -1 ) - sin g - sin g ˙
1
yn =
bn Î n n -1 ˚
1 È D n +1 Dn ˘
and y n +1 = D (cot g n +1 + cot g n ) - - ...eqn. 12.15
bn +1 ÍÎ n +1 sin g n +1 sin g n ˙˚
Equating ψ2, ψ3 and ψ4 calculated in Step 6 to k2ψ20, k3ψ30 and k4ψ40 a set of three linear simultaneous
equations in unknowns k2, k3 and k4 are obtained. Solve the equations to get k2, k3 and k4.
Step 7: Compute the edge stresses as
f n = f n0 + k 2 f n2 + k 3 f n3 + k 4 f n4 ...eqn. 12.16
Then the shell may be designed.
12.7 STRESS DISTRIBUTION PROCEDURE

To ensure compatibility of stresses at common edges free edge stress distribution is to be carried out.
The procedure of stress distribution is similar to moment distribution method with appropriate carry
over factor and appropriate distribution factor. In this article, the carry over factor and distribution
factors to be used are derived.
The continuity of stresses along edge n common to the plates n and n + 1 is ensured by the appli-
cation of edge shears in longitudinal direction. When plate n and n + 1 are regarded as bending
independently, free edge stresses fn and fn+1 develop, which are different from each other.
The application of the longitudinal shears at the edges has the effect of correcting the values of fn and
fn+1 so that they become equal. To correct those stresses, it is not necessary to solve for edge shears. The
correction can be by the stress distribution procedure. Figure 12.17 shows nth and n + 1th plate from
one end to mid span. Let Tn be longitudinal edge shear at edge n.
n+1
n + 1th plate
End
frame n Tn
nth plate
n -1
Fig. 12.17
Due to longitudinal shear Tn, stress in plate n at edge n is,

Tn bn 1 bn T
= ¥ ¥ + n , where In = moment of inertia of ith plate
2 I n 2 hn bn
Tn bn 12 b T
= ¥ 3
¥ n + n
2 hn bn 2 hn bn
4Tn 4T
= = n where An = bnhn, cross-sectional area of nth plate ...eqn. 12.17
bn hn An
Similarly, stress at edge n − 1 in nth plate
bn 12 b T
= -Tn ◊ ¥ 3
¥ n + n
2 hn bn 2 bn hn
2Tn
=- . ...eqn. 12.18
An
Stress in n + 1th plate at edge ‘n’
bn +1 12 Tn
= -Tn ¥ 3
-
2 hn +1bn +1 hn +1bn +1
4Tn 4T
=- =- 1 ...eqn. 12.19
hn +1bn +1 An +1
Stress in n + 1th plate at edge n + 1,
Tn bn +1 12 Tn
= ¥ 3
-
2 hn +1bn +1 hn +1bn +1
2Tn 2Tn
= = ...eqn. 12.20
hn +1bn +1 An +1
If free edge stresses at joint n in plate n is fn and at joint n in n + 1 plate is fn+1 the actual stresses
are, in plate n,
4Tn
fn +
An
4Tn
and in plate n + 1, f n +1 -
An +1
But, the effect of Tn forces are to ensure stress compatibility. Hence,
4Tn 4T
fn + = f n +1 - n
An An +1
Ê 1 1 ˆ
∴ 4Tn Á + = f n +1 - f n
Ë An An +1 ˜¯
An +1 + An
i.e. 4Tn = fn +1 - f n
An An +1
4Tn An +1
or An
= ( f - fn )
An +1 + An n +1
4Tn An +1
Thus, the correction to edge stress at n in plate n is
An
= ( f - fn )
An +1 + An n +1
Similarly, the correction to edge stress at n in plate n + 1 is
4Tn An
-
A +1
=- (f - f )
An + An +1 n +1 n
...eqn. 12.21
An +1
Thus, we conclude, if the correction to stress at joint n is (fn+1 − fn), distribute of it to edge
An + An +1
An
n of plate and - of it to edge n of (n+1)th plate. In other words, the distribution factor is
An + An +1
An +1 An
for nth edge of plate n and A + A for nth edge of plate n + 1.
An + An +1 n n +1
4Tn
From equations 12.17 and 12.18, it is clear that due to Tn stress at nth edge of n plate is while
An
2Tn
in the same plate at other (n − 1)th edge is - . From eqn. 12.19, we find the stress at edge n in
An
-4Tn 2Tn
(n +1)th plate due to Tn is while that at other edge (n + 1)th is . It shows that if corrections
An +1 An +1
1
are made for stresses at edge n, - of corrected stress should be carried over to other edge. Thus, the
2
carry over factor is −Y2.
Thus, in the stress distribution procedure,
An +1 An
Distribution factors are and i.e. distribution factor is inversely proportional to
An + An +1 An + An +1
1
cross-sectional area and, carry over factor is - .
2
QUESTIONS
1. Discuss the merits and demertis of folded plate roof over shell roofs.
2. Explain briefly structural behaviour of folded plate roofs.
3. List the assumptions made in the analysis of folded plate roofs.
4. Give step by step procedure of analysis of folded plate roof by Whitneys method. Give the recurring equations
also.
5. Write short note on Simpson’s method of folded plate analysis.
6. Derive the equations of three shears used in folded plate analysis.
7. Derive carry over factor and stress distribution factors to be used in folded plate analysis.
Chapter
13
Introduction to Shells
A shell is a thin curved surface the thickness of which is small compared to the radius and the other two
dimensions. Shells are used for roofing large column free areas. Figure 13.1 shows some of the
commonly used shell roofs.
(a) Cylindrical
(b) Butterfly
(c ) North light
(e) Spherical dome

(d ) Conical shell
(g) Ellipsoid
(f ) Paraboloid
Fig. 13.1 (Contd.)

INTRODUCTION TO SHELLS 159
(h) Elliptic paraboloid ( i ) Hyperbolic paraboloid
(k ) Conoid
( j ) Hyperboloid of revolution
(m) Funicular
( l ) Corrugated
Fig. 13.1
13.1 CLASSIFICATION OF SHELLS

Shells are stressed skin structures i.e. the structures in which load transfer is through the skin of
structure. In shells, because of their curved shape, bending is reduced to the great extent and load is
transferred as compressive force through the membrane of the structure. IS: 2210 gives the classifica-
tions of shell on the basis of their shape and the method of generating such shapes. Table 13.1 gives the
classification of the shell.
Some of the terms used in classification of shells are defined below:
(i) Shells of Revolution: The surface generated when a plane curve is rotated about an axis is known
as shell of revolution.
(ii) Shells of Translation: Shells which are obtained by moving one curve parallel to itself along
another curve, the planes of the two curves being at right angles to each other.
(iii) Ruled Surfaces: The surfaces which can be generated entirely by straight lines are known as ruled
surfaces. The surface is said to be singly ruled if at every point on the shell surface a single
straight line entirely lying on the surface can be drawn and it is doubly ruled if at every point two
straight lines lying entirely on the surface can be drawn.
(iv) Gauss Curvatures: Gauss curvature is the product of the two principal curvatures at any point on
the surface. Thus,
1 1 1
= ¥
R R1 R2
where R is Gauss curvature, R1 and R2 are the principal curvatures.
(v) Singly Curved: A shell is said to be singly curved if its one principal curvature is zero (hence,
Gauss curvature is zero). It is also known as developable shell, since, such surfaces can be
developed from bending plane surface.
(vi) Doubly Curved: A shell is said to be doubly curved if its Gauss curvature is definite. Such shells
are non-developable.
(vii) Synclastic shell: A shell is said to be synclastic if its Gauss curvature is positive.
(viii) Anticlastic shell: A shell is said to be anticlastic if its Gauss curvature is negative.
13.2 ADVANTAGES AND DISADVANTAGES OF SHELLS

Advantages of shell structures are:
(i) Good from the aesthetic point of view
(ii) Material consumption is less
(iii) Large column free areas can be covered.
(iv) Formwork can be removed early.
Disadvantages of shell structures are:
(i) Formwork is costly.
(ii) Analysis is complicated.
Comments on shell structures: Shell structures become economical if a number of units are to be cast
so that repetitive use of formwork is possible. If from aesthetic point of view and from the requirement
of large column free area, if an architect prefers shell roof, a structural engineer should design such roofs.
13.3 EFFICIENCY IN THE USE OF MATERIALS

Shells are an example of strength through form as opposed to strength through mass. In a R.C. beam/
slab material is subjected to maximum stress only at extreme fibre, that too at the section subject to
(Contd. on p. 162)
INTRODUCTION
Table 13.1 Classification of shells
Shells
TO
SHELLS
Singly curved Doubly curved
(developable) (non-developable)
Gauss
curvature = 0 Synclastic Anticlastic Other special
Gauss curvature +ve Gauss curvature -ve types
Shells of Shells of Ruled Alternately Partially Miscel-

Shells of Shells of Ruled Shells of Shells of
revolution translation surfaces synclastic synclastic laneous
revolution translation surfaces revolution translation
anticlastic and anticlastic
Conical Cylindrical Conical Spherical Elliptic Hyper- Hyper- Hyper- Corrugated Funicular
shells shells and domes, paraboloid, boloid bolic boloids of shells shells
including cylindrical paraboloids circular of para- revolution,
north light ellipsoids paraboloid revolution boloids hyperbolic
and butterfly paraboloid
conoid
161
maximum moment. At all other portion the material is under utilised. In a column most of the material
is fully stressed and hence, the material is utilised to maximum extent. That is why we commonly find
a column of size 230 × 450 mm support 3 to 4 beams of size 230 × 450 mm. In shells attempt is made
to reduce flexure and transfer the load as membrane compression. Hence, in shells material utilisation
is improved. It results into use of thinner sections and hence reduction in dead weight. By this means
a minimum of material is used to the maximum structural advantage. Shells of double curvatures are
among the most efficient of known structural forms. Most shells occurring in nature are doubly curved
shells. Shells of eggs, nuts and human skull are common examples.
Thus, in shells strength is due to shape.
13.4 HISTORICAL NOTE

An examination of some of the old places of worship in India and abroad show that shell structures have
been in usage even though the theoretical knowledge of their structural behaviour was then meagre. But
the modern thin shells are a far cry from the massive masonry domes of the middle ages. A comparison
of the relative weights of 17th century Peter Cathedral at Rome and a modern workshop building at
Jena, Germany brings home the difference. Both these domes cover an area of 39.6 m diameter.
St. Peter Cathedral at Rome: (Ref. Fig. 13.2)
1.2 m 1.5 m
27.6 m
39.6 m
3m
Fig. 13.2 Cathedral at St. Peter, Rome
Total weight = 11000 tonnes

i.e. = 88.42 t/m
It requires heavy foundation.
Now at Schott Workshop, Jena, the dome built is shown in Fig. 13.3.
3.9 m
60 mm
thick
7.8 m
39.6 m
Fig. 13.3 Schott workshop at Jena, Germany

INTRODUCTION TO SHELLS 163
Weight including ring beam = 364 tonnes.

∴ Weight per metre of periphery = 2.93 t/m
One brick wall can take this load.
The above improvement was possible because of the following:
1. R.C.C. is better material compared to masonry.
2. Now there is better understanding of analysis and design.
13.5 A BRIEF LITERATURE REVIEW

The thin reinforced concrete shell had its beginning in Germany in 1920s. Two German engineers—
Finsterwalder and Dishinger—were the first two to develop a theoretical analysis applicable to R.C.
Cylindrical shells around the year 1930. An early American contribution was a paper by Scharer (1936),
which attempted a simplified analysis. Until 1940s, cylindrical shell analysis and construction domi-
nated the research field.
It was only in the mid 1930s and early 1940s that the designers began looking for other forms of
shells for roofs. Around 1935, two French engineers, namely, Aimond and Raffaile, published studies
on the properties and potentialities of the hyperbolic paraboloid. However, it was the Mexican architect
Felix Candola who promoted and popularised the hyperbolic paraboloid for roofing factories, churches,
clubs etc.
The conoid had its beginning in France. The one sheet hyperboloid appears to have been first
employed for roofing in Germany in the form of Silberkuhl shells. 1950s and 1960s have witnessed an
unprecedented spurf at activities in the field of thin shells. Many new forms are continuously experi-
mented upon and added to the growing vocabulary of the shells. With the advent of computers and
development of analysis packages, now a days the designers are more confident in going for the shell
roofs. They have even tried to optimize shell roof design and have given guidelines for choosing
dimensions of various type of shell roof.
13.6 ANALYSIS AND DESIGN

In the analysis of shell roofs, the structure is regarded as homogeneous and isotropic. However, in the
design of reinforcement, the concrete is assumed to be cracked and steel is provided to take care of the
full direct and diagonal tension.
Shell roofs of complex shapes do not always lend themselves to calculation by analytical means
using classical theory. Hence, there is growing trends towards the use of experimental investigations
and the use of numerical method like finite element analysis.
To account for the secondary effects like shrinkage and temperature, exact calculations are not
available. In tropical countries, the stresses due to temperature changes is sever. The secret of avoiding
cracks due to secondary effects lies in the provision of closely spaced small diameter reinforcement.
QUESTIONS
1. Explain the following terms with neat sketches:

(i) Shells of revolution (ii) Shells of translation (iii) Ruled surfaces.
2. What is Gauss curvature? How do you classify shells based on Gauss curvature values?
3. Give the Indian standard classification of shells.
Chapter
14
Introduction to Cylindrical
Shell Roof
Cylindrical shell roof is commonly adopted for covering large column free areas in the factories. As
stated in the previous chapter, cylindrical shell is generated when a straight line moves over a curved
surface, maintaining its position at right angle to the curve and moving parallel to itself in the projected
plane. In this chapter, the various parts and types of cylindrical shell roof are explained and design
criteria is presented. A preliminary design procedure, namely, Lundgreens beam theory is presented.
14.1 PARTS OF A SINGLE CYLINDRICAL SHELL ROOF

Figure 14.1 shows a typical single barrel shell and its various parts.
The straight line generating the surface is known as generator and the plane curve that guides the
generator is known as the directrix. The directices that are usually employed are the arc of a circle, the
semiellipse, the parabola, the cycloid or the catenary.
A cylindrical shell may or may not be provided with edge beams. The supporting members at the two
ends of a shell are known as traverses or end frames. The traverse may be a solid diaphragm, a tied arch,
a trussed arch or a rigid frame. It is usually assumed that the shell is simply supported on the traverses.
Hence, the traverse should be rigid in its plane but flexible out of plane. This is achieved to a great
extent if the frame is deep and thin. The distance between the two adjacent traverses is known as span
of the shell and the projection of the directrix is known as chord width.
14.2 TYPES OF CYLINDRICAL SHELL ROOFS

The shell roof may be a single barrel as shown in Fig. 14.2 (a) or it may be a multiple barrel as shown
in Fig. 14.2(b). Usually shells are supported on two end frames. But sometimes they may be built
monolithically over more than two end frames as shown in Fig. 14.2(c). Such shells are known as
continuous shells.
14.3 DESIGN CRITERIA

Indian standard code IS: 2210 recommends the following:
1. Grade of Concrete: M:20
2. Maximum size of the aggregates is to be restricted to 12 mm to 20 mm depending upon the
thickness of the shell.
3. If chord width is 8 m to 12 m, maximum span is restricted to 30 m.
INTRODUCTION TO CYLINDRICAL SHELL ROOF 165
End frame
Directrix
Generator
an
sp
L=
Edge beam
End frame
Chord width
(a) Parts of cylindrical shell roof
Elliptic
Circular
Cycloid
Parabolic
(b) Different types of directories
Fig. 14.1
If chord width used is large (say upto 30 m), span should be reduced to 8 to 12 m.
For spans more than 30 m, prestressing of edge beams is necessary.
4. Height of shell may be kept 1/12 to 1/6th span—the larger figure applicable to smaller span.
However, in case of shells without edge beam, depth shall not be less than 1/10th span.
In case of short shells depth should be at least 1/8th of chord width.
5. Semicentral angle may be kept between 30° to 40°. Restricting it to 40° has the following
advantages:
(a) Wind load effect may be ignored.
(b) During construction back form is not required for concreting.
(a) Single barrel
( b) Multiple barrel shell
End frame
Intermediate frame
End frame
( c) Continuous shell
Fig. 14.2
6. Thickness of shell should be at least 50 mm to avoid the problems of leakage. Usually thickness
selected for long shells is 75 mm while for short shells (more chord width and less span) thickness
of 60 mm may be used.
Near the ends the shell is thickened to 30 percent extra. The distance of thickening from the edge
is 0.38 rt to 0.78 rt where ‘r’ is the radius of the shell directrix and ‘t’ is the thickness of shell.
The need for edge thickening are:
(a) Moments in the shells are larger at the ends.
(b) At the edges 3 layers of reinforcement are to be provided namely—longitudinal, transverse
and shear. The three layers can be suitably accommodated with required cover only by making
edges thicker.
7. Edge beam depth is automatically fixed once we select overall depth and semicentral angle.
Thickness of the edge beam shall be 2 to 3 times the thickness of the shell.
Example 14.1. Fix up the overall dimensions for a circular cylindrical shell of span 25 m and chord
width 10 m.
Solution:
Span L = 25 m
Chord width = 10 m.
Let the semicentral angle be 40°.
10 2
Then, radius of the shell = R = = 7.78 m
sin 40∞
∴ Rise of shell = R - R cos 40∞ = 7.78 (1 - cos 40∞)
= 1.82 m.
Overall depth of shell should be
1 1
= th to th span.
12 6
25 25
= to m.
12 6
Let us select it as 3.4 m.
∴ Edge Beam Depth = 3.4 − 1.82 = 1.58 m.
Let it be 1.6 m.
Let thickness of shell = 75 mm.
∴ Width of edge beam = 2 to 3 times 75 mm. 75 mm
Let us select width of edge beam = 200 mm.

600 mm
Edge thickening 3.4 m 100 mm
10 m
Thickness = 30% extra 1.58
= 75 + 0.3 of 75 = 97.6 mm
Let us use 100 mm thickness of edges. 200
Distance of edge thickening: Span = 25 m
40° 40°
0.38 rt to 0.76 rt
i.e. 0.29 m to 0.58 m
Let the edge be thickened to a distance of 600 mm.
Figure 14.3 shows the section selected.
Fig. 14.3
14.4 ANALYSIS
Various methods of analysis found in literature may be listed as:
(i) Beam theory
(ii) Membrane theory
(iii) Bending theory
(iv) Finite element method
Beam theory is explained in this chapter. Membrane theory and Bending theory are taken in the next
two chapters. Finite element method is treated as out of the scope of this book. For this method, readers
may refer the books on the finite element analysis.
14.5 BEAM THEORY

This theory can be used for the analysis of cylindrical shells with span more than three times the chord
width. This theory is based on the assumption that plane cross-section remains plane even after defor-
mation. In other words, it assumes longitudinal stress varies linearly across the depth of section. This
assumption holds good fairly well for long shells. Hence, one can use it for the analysis of such shells.
This theory is relevant even today to study because it gives feel of the structural behaviour and the
designer can avoid mixing up of the sign conventions in the sophisticated analysis and committing the
blunder. This method has the following advantages:
1. It brings shell analysis within the reach of those who are not familiar with the technique of
advanced mathematics.
2. Shells with non-circular directices can be dealt with easily.
3. Shells with non-uniform thickness also can be handled without much difficulty.
4. Structural action of the shell is easily visualised.
This theory divides the shell action into
(i) beam action
(ii) arch action
14.5.1 Beam Action
A long cylindrical shell is analysed as a beam with curved cross-section, supported on end frames. To
start with entire concrete section is assumed effective to determine the longitudinal stresses and shear
stresses. In the design wherever tension is found, concrete is treated as ineffective and steel is provided
to take the tension. Consider the typical cross section of the shell shown in Fig. 14.4.
y
h
N A
2a
2b fk
fn
f R
df
R - radius of shell
fk - semi central angle
2a - depth of edge beam
2b - width of edge beam
h - thickness of shell.
Fig. 14.4
Let the neutral axis be at a distance y from the horizontal axis through crown. Then,
Moment of area about the horizontal axis at crown
y=
Total area of shell
È fk ˘
{ }
Í 2 a ¥ 2 b R (1 - cos f k ) + a + Ú Rd fh ( R - R cos f ) ˙ ¥ 2
Í ˙˚
=Î
0
ÈÎ 2 a ¥ 2 b + hRf k ˘˚ ¥ 2
=
{ }
4 ab R (1 - cos f k ) + a + R 2 h {f k - sin f k }
...eqn. 14.1
4 ab + hRf k
If φn is the semicentral angle made by the neutral point on shell as shown in the figure, then it can
be determined from the relation:
R (1 - cos f n ) = y . ...eqn. 14.2
Then the moment of inertia of the section about neutral axis ‘I’ is given by
I = Moment of inertia of the edge beams + Moment of inertia of the shell.
È1 fk ˘
{
= Í ¥ 2 b (2 a ) + 2 a ¥ 2 b R (1 - cos f k ) - y + a } Ú R d fh R 2 (cos f - cos f n ) ˙ ¥ 2
3 2 2
+
ÍÎ 12 0 ˙˚
M
Using the expression , s = y, stresses at any required depth can be found, in which M is the
I
moment. It is maximum at centre of span as it is a simply supported beam of span L subjected to
wL2
uniformly distributed load w i.e. maximum moment is = .
8
Obviously bending stress is zero at neutral axis and is having maximum compressive at crown point
in midspan. This stress should be checked for bending compression. To take care of tensile stresses
provide longitudinal reinforcement to take complete tension. It may be observed that only a small
portion of shell (that is below N-A) and edge beams are in tension. Tensile force in the edge beam may
be considered equal to the tensile stress at mid depth of the edge beam multiplied by the area of the edge
beam. Reinforcement found is provided at closer interval in the lower portion and spacing is uniformly
increased towards the top.
The beam with this curved shape is to be designed to withstand shear stresses also. Maximum shear
force is equal to 1/2 the total load and occurs near end frames.
wL
V= .
2
( ay ) , where
V
Maximum shear stress =
bI
b = width of shell at neutral axis = 2h
ay = moment of area above neutral axis about the neutral axis
fn
= 2 Ú Rd fh ( R cos f - R cos f n )
0
f
= 2 R 2 h ÈÎsin f - f cos f n ˘˚ 0 n
= 2 R 2 h ÈÎ sin f n - f n cos f n ˘˚
Shear reinforcement is provided diagonally at corners to a maximum length of L/4.
14.5.2 Arch Action
Consider the freebody diagram of a shell of unit length as shown in Fig. 14.5.
Unit length
fk fk
Fig. 14.5 Arch action
The equilibrium of the unit length of shell is maintained by two sets of forces, namely the load w per
unit length of shell and the specific shear d = ( ay ) where the specific shear is defined as difference
w
bI
in shear forces between the two edges of the unit length of shell.
The specific shear acting at any point in the direction of the tangent to the shell arch may be resolved
into its horizontal and vertical components. It is clear that the sum of the vertical components of specific
shear balances the vertical load on the arch and the sum of the horizontal components will be zero.
If it is a single barrel shell, it is obvious that the elemental shell arch do not develop any restraining
forces or moments at its ends. Hence, we have a statically determinate free arch for the analysis.
If it is an intermediate shell of a multiple shell, the ends will behave like fixed ends. Hence, fixed
arch analysis is necessary. The column analogy or the elastic centre method is suitable for the analysis
of fixed archies.
Example 14.2. Design a reinforced concrete circular shell with the following particulars.
Radius R=3m
Span L = 15 m
Semi central angle φk = 60°
Thickness of shell h = 75 mm
75 mm
N A
fn
R=
3m fk
Fig. 14.6
Loads
Self weight = 0.075 × 1 × 1 × 25 = 1.875 kN/m2
Water proofing cover and occasional live load = 1 kN/m2
∴ Total load = 2.875 kN/m2
Weight per metre run of shell
p
w = 2.875 ¥ R ¥ 2 f k = 2.875 ¥ 3 ¥ 2 ¥ = 18.06 kN m 2
3
Maximum bending moment
wL2 152
= 18.06 ¥ = 508 kN-m
8 8
Since, edge beams are not provided, a = b = 0
R 2 h (f k - sin f k ) R (f k - sin f k )
∴ y= =
hRf k fk
Èp ˘
3 Í - sin 60 ˙
= Î 3 ˚ = 0.519 m
p
3
∴ φn is obtained by the relations
y = R - R cos f n
0.519 = 3(1 − cos φn)
∴ φn = 34.2°
fk
I = 2 Ú Rd f h R 2 (cos f - cos f n ) d f
2
0
fk
È 1 + cos 2 f ˘
= 2 R 3h Ú ÍÎ 2
- 2 cos f cos f n + cos 2 f n ˙ d f
˚
0
È 1 ˘
Í f k + 2 sin 2 f k ˙
= 2R h Í
3
- 2 cos f n sin f k + f k cos f n ˙
2
Î 2 ˚
Èp 1 ˘
Í 3 + 2 sin (2 ¥ 34.2) p ˙
= 2 ¥ 3 ¥ 0.075 Í
2
- 2 cos34.2 ¥ sin 60 + cos 34.2 ˙
2
Î 2 3 ˚
= 0.0968m 4 = 0.0968 ¥ 1012 mm 4

∴ Compressive stress at crown
M 508 ¥ 10 6
= y= ¥ 519 = 2.723 N mm 2 < 8.5 N mm 2
I 0.0968 ¥ 1012
Rise of shell = R - R cos60 = 3 (1 - cos60) = 1.5 m
∴ The distance of lowest point from NA = 1.5 − 0.519 = 0.981 m = 981 mm
∴ Maximum Tensile Stress in shell
508 ¥ 10 6
= ¥ 981 = 5.148 N mm 2
0.0968 ¥ 1012
∴ Tensile force per metre run of shell = 5.148 × 75 × 1000 = 386116 N
386116
∴ Ast = = 2574 mm 2
150
Provide 16 mm bars at 75 mm c/c. This may be changed to 12 mm bars at 200 mm c/c near neutral
axis and is maintained in the compression zone.
Design for Shear
15
Maximum shear force = 18.064 ¥ = 135.48 kN
2
At neutral axis
ay = 2 R 2 h ÈÎ sin f n - f n cos f n ˘˚
È 34.2 ¥ p ˘
= 2 ¥ 32 ¥ 0.075 Ísin 34.2 - cos2 34.2 ˙
Î 180 ˚
= 0.09234 m3 = 0.09234 × 109 mm3
∴ Shear strees at neutral axis
(ay )
V
q=
bI
135.48 ¥ 1000
= ¥ 0.09234 ¥ 10 9
2 ¥ 75 ¥ 0.0968 ¥ 1012
= 0.86 N/mm2
τv = 1.5 × 0.86 = 1.29 N/mm2 < τc max

Hence, shear reinforcement can be provided to take care of shear.
p
¥ 12 2
Ast provided at N-A = 4 ¥ 1000 = 565.48 mm 2 per meter width
200
Ast 565.48
Percentage of reinforcement = ¥ 100 = ¥ 100 = 0.754
h ¥ 1000 75 ¥ 1000
τc = 0.57 N/mm2
Thus τc < τv < τc max.
Hence, shear reinforcement is to be provided.
Shear force per metre width of shell
Vu = τv × h × 1000 = 1.29 × 75 × 1000 = 96750 N.
∴ Vus = 96750 − τc h × 1000
= 96750 − 0.57 × 75 × 1000 = 54000 N
using 8 mm bars [note it is single legged is shells] at 45°
p
¥ 8 2 ¥ 0.87 ¥ 415 ¥ 1000 [sin 45∞ + cos 45∞]
s= 4 = 475 mm
54000
Hence, provide nominal shear reinforcement of 8 mm at 200 mm c/c.
Arch Action
Load per unit length of shell, w = 18.06 kN
18.06 ¥ 1000
∴ Specific shear qs =
(2 h ) I
( ay )
2 R 2 h [sin f - f cos60]
18060
=
2 ¥ 75 ¥ 0.1613 ¥ 10 12
18060 ¥ 2 ¥ (3000 ) ¥ 72
2
= [sin f - 0.5f]
2 ¥ 75 ¥ 0.1613 ¥ 1012
= 1007.68 [sin φ − 0.5 φ]
The arch is divided into a suitable number of parts for further calculation. In this case, let us divide
it into 12 parts at 10° intervals as shown in Fig. 14.7. The specific shear at centre of each part is
calculated using the above expression and is shown in Table 14.2. Specific shear force
10
Ts = q s A = q s ¥
p ¥ 75
180
∴ Net vertical force, V = ws − Ts sin φ, where ws is the vertical load on each part.
18.06
ws = = 1.505 kN = 1505 N
12
and net horizontal force H = Ts cos φ
fn
fk
Fig. 14.7
Moment at crown = ∑(V R sin φ + H R (1 − cos φ))
Moment at φ = Moment at crown − ∑(V R sin φ + H R (1 − cos φ))
In the expression for moment at φ summation is to be considered for the points above the one under
consideration. Table 14.1 shows these calculations.
Table 14.1 Calculation Table
Section φ qs in N/mm2 Ts in N V in N H in N VR sin φ + M in

HR(1 − cos φ) N-m
Deg. Radians in N-m
1 5 0.0873 43.8 574.1 1455 572.0 386.96 340.5

2 15 0.2618 128.9 1687.3 1068.3 1630.0 996.1 −655.5
3 25 0.4363 206.0 2696.8 365.3 2444.0 1150.1 −1805.6
4 35 0.6109 270.2 3537.0 −523.7 2897.0 670.6 −2476.3
5 45 0.7854 316.8 4147.2 −1427.5 2933.0 −451.0 −2025.3
6 55 0.9599 341.8 4474.0 −2159.9 2566.0 −2025.5 0
∑ 727.5
∴ Moment at crown 727.5
Moment for design = 2476.3 N-m
Mu = 1.5 × 2476.3 = 3714.45 N-m
Taking effective depth = 75 − 25 = 50 mm
Ê Ast 415 ˆ
3714.45 ¥ 10 3 = 0.87 ¥ 415 ¥ Ast ¥ 50 Á 1 - ¥
Ë 1000 ¥ 50 25 ˜¯
Ê A ˆ
205.8 = Ast Á 1 - st ˜
Ë 3012 ¯
or Ast2 - 3012 Ast + 205.8 ¥ 3012 = 0
Ast = 222 mm2
p
¥ 82
Provide 8 mm bars at 4 ¥ 1000 ª 220 mm c c
222
The bars are to be provided on appropriate side as shown in Fig. 14.8.
12 f @ 200
8 f @ 220
12 f @ 70
8 f @ 200
8 f @ 220
12 f @ 70
12 f @ 200
Fig. 14.8
Example 14.3. An intermediate shell of a multiple cylindrical shell roof of span 25 m is to be built.
Each shell unit has a chord width of 8 m. Fix up overall size of the shell and design longitudinal and
shear reinforcement. Explain how analysis for arch action will be carried out and then give typical
reinforcement details.
Solution.
L = 25 m Chord width = 8 m
Let θk = 40°.
4
Then R= = 6.22 m
sin 40∞
Rise = R − R cos 40° = 6.22(1 − cos 40°)
= 1.46 m.
1 1
The overall depth of shell should be between th to th span i.e. in this case it is to be between
12 6
25 25
to m. Let us select overall depth = 3.0 m.
12 6
∴ Edge beam depth = 3.0 − 1.46 = 1.54 m. Let the thickness of shell be = 75 mm.
∴ Width of edge beam = 2 to 3 times 75 mm. Let us select width of edge beam = 200 mm.
Figure 14.9 shows the typical section selected.
Unit length
fk fn
Fig. 14.9
Half of the edge beam is assumed to take care of shell on either side
∴ 2b = 0.1 or b = 0.05 m
2a = 1.54 m or a = 0.77 m
∴ Distance of N-A from the crown
4 ab ÈÎ R (1 - cos f k ) + a ˘˚ + R 2 h ( f k - sin f k )
y=
4 ab + Rhf k
È p ˘
4 ¥ 0.77 ¥ 0.05 ÈÎ6.22 (1 - cos 40∞) + 0.77 ˘˚ + 6.22 2 ¥ 0.075 Í 40 ¥ - sin 40∞˙
= Î 180 ˚
p
4 ¥ 0.77 ¥ 0.05 + 6.22 ¥ 0.075 ¥ 40 ¥
180
= 1.05 m
Let the intersection of neutral axis with shell make angle φn with vertical through crown [Ref.
Fig. 14.10]. Then
R(1 − cosφn) = 1.05
N A
fn
R= fk
3m
Fig. 14.10
6.22(1 − cosφn) = 1.05

∴ φn = 33.76°
¥ 0.1 ¥ 1.54 3 + 1.6 ¥ 0.1{6.22 (1 - cos 40∞) - 1.05 + 0.77} ¥ 2

1 2
I=
12
È 40 ¥ p 1 ˘
Í 180 + 2 sin 80∞ 40 ¥ p ˙
+ 2 ¥ 6.22 ¥ 0.075 Í
3
- 2 cos33.765∞ ¥ sin 40∞ + cos 33.765∞˙
2
Î 2 180 ˚
= 0.828276 m4 = 0.828276 × 1012 mm4
Loads
Self weight of shell = 0.075 × 25 = 1.875 kN/m2
Occasional live load plus finishing load = 1 kN/m2
Total load on shell = 2.875 kN/m2
p
∴ Load per metre run of shell = 2.875 ¥ 80 ¥ ¥ 6.22
180
= 24.97 kN/m
Weight of edge beams = 0.1 × 1.54 × 25 × 2 = 7.7 kN/m
Valley finishing and weight of edge finishing = 1.5 kN/m
∴ Total load W = 34.17 kN/m
Design for Beam Action
wL2 25 2
Maximum moment M = = 34.17 ¥ = 2669.531 kN-m
8 8
wL 25
Maximum shear V = = 34.17 ¥ = 427.125 kN
2 2
∴ Compressive stress at crown
M 2669.531 ¥ 10 6
sc = y= ¥ 1.05 ¥ 103
I 0.828276 ¥ 1012
= 3.384 N mm 2 < 7.0 N mm 2

where 7.0 N/mm2 is permissible stress in bending compression in M:20 concrete.
Hence, the section selected is safe in bending compression.
Tensile Stress at Centre of Edge Beam:
Distance of centre of edge beam from neutral axis:
y = R (1 − cos φk) + a − R (1 − cos φn)
= 6.22 (1 − cos 40°) + 0.77 − 6.22 (1 − cos 33.765°)
= 1.176 m
Tensile stress in edge beam at its centre
M 2669.531 ¥ 106
s av = y= ¥ 1.176 ¥ 103 = 3.79 N mm 2
I 0.828276 ¥ 1012
Total tensile force in edge beam

= 3.79 × 1.54 × 1000 × 0.1 × 1000
= 583698 N
583698
∴ Ast reqd = = 3891 mm 2
150
Using 28 mm bars, number of bars required
3891
= =7
p
¥ 28 2
4
Provide 7 bars of 28 mm diameter Fe-415 steel on each face of edge beam, since, another half of the
beam has to take care of loads from the adjoining shell.
Lower portion of shell is also in tension. Height of this portion
= R (1 - cos f k ) - y
= 6.22 (1 − cos 40°) − 1.05
= 0.405 m
Maximum tension in this tensile portion of shell is at the junction with edge beam.
M
This stress = ¥y
I
2669.531 ¥ 106
= ¥ 0.405 ¥ 1000
0.828276 ¥ 1012
= 1.305 N/mm2
Thickness of shell at junction is 30% more than shell thickness.
i.e. Thickness = 1.3 × 75 = 100 mm (say)
1.305 ¥ 100 ¥ 1000
∴ Tensile steel required = = 870 mm 2 per metre length.
150
∴ Provide 10 mm bars at 90 mm centre to centre in the lower portion of shell. After providing 4 such
bars, in the remaining portion nominal reinforcement of 10 mm @ 200 mm c/c may be provided
[Ref. Fig. 14.11].
10 mm @ 200
(a) ( b) Enlarged view of end beam
8m End frame
10 mm @ 90
10 mm @ 220
6.25
10 mm @ 200 c/c
12.5
To be designed for
arch action
(c ) Plan
Fig. 14.11 Typical Reinforcement in cylindrical shell
Design for Shear

Shear stress is maximum at the neutral axis in the end section. Here
ay = 2 R 2 h ÈÎ sin f n - f n cos f n ˘˚
È 33.765 ¥ p ˘
= 2 ¥ 6.22 2 ¥ 0.075 Ísin 33.765∞ - cos33.765˙
Î 180 ˚
= 0.382318 m3
= 0.382318 × 109 mm3
∴ Shear stress at N-A
q=
V
(ay )
bI
427.125 ¥ 1000
= ¥ 0.382318 ¥ 10 9
(2 ¥ 75) ¥ 0.828276 ¥ 1012
= 1.314 N/mm2
p 4 ¥ 10 2
Ast provided = ¥ 1000 = 392.7 mm 2
200
392.7
∴ % steel = ¥ 100 = 0.523
75 ¥ 1000
∴ τc = 0.48 N/mm2. [for M: 200 mm].
Hence, shear reinforcement is to be designed.
Factored shear force per metre length of shell
Vu = 1.5 × 1.314 × 1000 × 75 = 147825 N
∴ Vus = Vu − τcbd
Vus = 147825 − 0.48 × 75 × 1000 = 111825 N
Using 8 mm stirrups at 45°,
0.87 f y Asv d
Sv = (sin a + cos a )
Vus
p
0.87 ¥ 415 ¥ ¥ 82 ¥ 1000 (sin 45∞ + cos 45∞)
= 4
111825
= 229 mm.
Provide 8 mm bars at 220 mm c/c.
Design for Arch Action
The intermediate shell acts as a fixed arch. The specific shear may be found at centres of several equal
parts. Horizontal and vertical loads due to specific shear may be determined for each part. The vertical
downward load due to self weight of each segment may be calculated. Then the arch may be analysed
by elastic centre method to get transverse moments at the centre of each part. Then the reinforcement
required to resist the transverse moment may be determined. The transverse moment is −ve near edges
and positive at crown. Hence, the reinforcement is on upper side at edges while it is on lower side at
crown. However, since the thickness is small sometimes one layer on upper side and another layer on
lower side are also provided.
Figure 14.11 shows the details of reinforcement.
QUESTIONS
1. Draw a neat sketch of a single barrel shell and indicate its various parts.
2. List the Indian standard recommendations for the design of cylindrical shell and fix up the overall dimensions
for a shell to cover an area 24 m × 8 m.
3. What are the advantages of beamy theory for the analysis of cylindrical shells? What are its limitations?
4. Briefly explain Leudgreen’s beam theory for the analysis of a shell with edge beam giving necessary equations
for beam action. Arch action may be explained qualitatively.
5. Design longitudinal and shear reinforcement in a circular cylindrical shell of span 24 m and chord width 8 m.
Use beam theory.
6. Draw neatly typical reinforcements in a cylindrical shell roof with edge beam.
Chapter
15
Membrane Analysis of
Cylindrical Shell Roofs
In the membrane theory, the shell is idealised as a membrane incapable of resisting bending stresses.
In other word, in this theory, bending of element of a shell is ignored and it is treated as though it is
under the forces through the skin of the shell structure.
In case of cylindrical shells bending of the element is not negligible. To get reasonably good behav-
iour of shell, the bending should be considered. Hence, one should design cylindrical shell roofs only
after analyzing by bending theory. However, we study the membrane theory first because of the follow-
ing reasons.
1. It is useful in many practical cases in gaining some insight into the structural behaviour of a shell.
2. We see later that the membrane theory can be used as a particular integral in the bending theory.

Figure 15.1(a) shows the coordinate system selected and 15.1(b) shows a typical element subject to the
membrane forces.
Nx
Nxf
y=f Nfx
x x z Nf
dx
y
+ x z
Nf
+
R Nfx
+
df f +
Nx
Nxf
(a) (b)
Fig. 15.1
MEMBRANE ANALYSIS OF CYLINDRICAL SHELL ROOFS 183
Note:
1. The origin of the coordinate system is at the apex of the shell directrix at mid span.
2. The coordinate x is measured along the crown generator.
3. The coordinate φ or y is along the tangent to directrix.
4. The coordinate z is in the radial inward normal.
The membrane forces acting on the element are
1. longitudinal tension Nx per unit width.
2. tension is transverse (φ or y) direction Nφ per unit width.
3. shear forces Nxφ and Nφx per unit width.
Measuring φ from the crown and taking an element of size dx × Rdφ we may note,
∂N x
N x+ = N x + dx
∂x
∂N f
N f+ = N f + df
∂f
∂N xf
N x+f = N xf + dx
∂x
∂N fx
N f+x = N fx + df
∂f
Let X, Y and Z be components of load intensity on the element.
Consider the equilibrium of forces in x-direction.
∑ Fx = 0 →
Nx+Rdφ − NxRdφ + N+φx dx − Nφx dx + X dx Rdφ = 0
Substituting for Nx+ and N+φx we get,
Ê ∂N x ˆ Ê ∂N fx ˆ
ÁË N x + dx ˜ Rd f - N x Rd f + Á N fx +
¯
d f˜ dx - N fx dx + XRdxd f = 0
∂x Ë ∂f ¯
Simplifying and dividing throughout by R dx dφ, we get
∂N x 1 ∂N fx
+ +X =0 …eqn. 15.1
∂x R ∂f
Similarly ∑Fy = 0, gives
Ê ∂N f ˆ Ê ∂N x f ˆ
ÁË N f + ∂f d f˜¯ dx - N f dx + ÁË N xf + ∂x dx ˜¯ Rd f - N xf Rdf + YRdx d f
Simplifying and dividing throughout by Rdxdφ, we get,

∂N x f 1 ∂N f
+ +Y = 0 …eqn. 15.2
∂x R ∂f
Consider the equilibrium of forces in z-direction.
i.e. ∑Fz = 0 →
Ê dx ˆ df
N f+ dx sin Á ˜ + N f dx sin + ZdxRd f = 0.
Ë 2¯ 2
Ê df ˆ df
Since, dφ is small angle, sin Á ˜ =
Ë 2 ¯ 2
Ê ∂N f ˆ d f df
ÁË N f + ∂f d f˜¯ 2 + N f 2 + Z Rdx d f = 0
Neglecting small quantity of higher order, we get,

Nφ + ZR = 0 …eqn. 15.3
Taking moment about z-axis through the centre of element, we get
dx dx Rd f Rd f
N xf Rd f + N x+f Rd f - N fx dx - N f+x dx =0
2 2 2 2
Substituting for N+xφ and N+φx and then neglecting small quantities of higher order, we get,
Nxφ = Nφx.
Hence, the equations of equilibrium are
∂N x ∂N x f ¸
+ +X =0 Ô
∂x Rd f ÔÔ
∂N x f ∂N f ˝ …eqn. 15.4
+ +Y = 0 Ô
∂x Rd f
and Ô
N f + ZR = 0 Ô˛
For any given loading the above three equations of equilibrium can be solved to get the membrane
forces. Nx, Nφ and Nxφ.
Example 15.1. Find the expressions for membrane forces in a circular cylindrical shell roof subjected
to self weight only.
Solution.
Self weight be g/unit surface area.
Then its X, Y, Z components are given by (Ref. Fig. 15.2)
X=0
Y = g sinφ f
Z = g cosφ y
z
From third equation of equilibrium, we get, g
Nφ = −ZR = −gR cosφ
From second equation of equilibrium,
∂N xf ∂N f f
+ +Y = 0
∂x R∂f
∂N xf
( - gR )( - sin f) + g sin f = 0
1
+ Fig. 15.2
∂x R
∂N xf
i.e. = -2 g sin f
∂x
Integrating w.r.t. x on both side we get,
N xf = -2 gx sin f + f1 ( f )
where f1(φ) is constant of integration.
The boundary condition is,
at x = 0, Nxφ = 0 due to symmetry.
From this condition, we get,
0 = 0 + f1(φ).
∴ Nxφ = −2gx sinφ
From equation of equilibrium (1), we have
∂N x 1 ∂
+ ( -2 gx sin f) = 0
∂x R ∂f
∂N x 1
i.e. = 2 gx cos f
∂x R
1 Ê x2 ˆ
∴ Nx = 2 g Á ˜ cos f + f2 ( f)
R Ë 2¯
gx 2
= cos f + f2 ( f )
R
where f2(φ) is the constant of integration.
Boundary condition: Since the end frame is assumed to be simple support, at x = L/2, Nx = 0
g L2
0= cos f + f 2 ( f )
R 4
g L2
or f2 ( f ) = - cos f
R 4
gx 2 g L2
Hence, Nx = cos f - cos f
R R 4
gL2 Ê 4 x2 ˆ
=- ÁË 1 - ˜ cos f
4R L2 ¯
Thus, the expressions for the membrane forces are
Nφ = −gR cos φ ¸
Ô
Nxφφ = −2gx sin φ ÔÔ …eqn. 15.5
˝
gL2 Ê 4x 2 ˆ Ô
and Nx = - ÁË 1 - 2 ˜¯ cos f Ô
4R L
Ô˛
Example 15.2. Determine the expressions for membrane forces in a circular cylindrical shell subject to
a snow load of p0/unit horizontal area.
Solution. Load is p0/unit horizontal area.
It is equal to p0 cosφ per unit surface (which can be easily seen from Fig. 15.3.)
1
P0
1
cosf
f
Fig. 15.3
∴ X=0
Y = p0 cosφ. sinφ
Z = p0 cosφ. cosφ = p0 cos2φ
From equation of equilibrium (3),
Nφ = −ZR = −p0 cos2φR
= −p0R cos2φ
From equation of equilibrium in φ-direction,
∂N x f 1 ∂N f
+ +Y = 0
∂x R ∂f
∂N xf
( - p0 R ) 2 cos f ( - sin f) + p0 cos f sin f = 0
1
i.e. +
∂x R
∂N xf
i.e. = -3 p0 cos f sin f
∂x
∴ N xf = -3 p0 x cos f sin f + F1 ( f )
Due to symmetry, at x = 0, Nxφ = 0
∴ 0 = 0 + f1(φ).
Hence, Nxφ = −3p0x cosφ sinφ
3
=- p x sin 2 f
2 0
From equation of equilibrium in x-direction,

∂N x 1 ∂N xf
+ + X = 0.
∂x R ∂f
∂N x 1 Ê 3 ˆ
i.e. + Á - p0 ˜ x 2 cos 2 f = 0
∂x RË 2 ¯
∂N x 3 p0 x cos2 f
or =
∂x R
3 p0 x 2
∴ Nx = cos 2 f + f 2 ( f )
R 2
As end frames are treated as simple supports, at x = L/2, Nx = 0.
3 p0 L2
∴ 0= cos 2 f + f 2 ( f )
R 8
3 p0 L2
or f2 ( f ) = - cos 2 f
R 8
3 p0 x 2 3 p L2
Hence, Nx = cos 2 f - 0 cos 2 f
R 2 R 8
3 p0 L2 È 4 x 2 ˘
=- Í1 - 2 ˙ cos 2 f
R 8 Î L ˚
Thus in this case
Nφ = − p0R cos2 φ
3 …eqn. 15.6
N kf = - p 0 x sin 2 f
2
p 0 L2 Ê 4x 2 ˆ
and Nx = - 3 Á 1 - 2 ˜ cos 2 f
8R Ë L ¯
Example 15.3. Find the membrane forces in a circular cylindrical shell subject to a sinusoidal loading
4g px
of intensity cos per unit surface area acting vertically downward.
p L
Solution. Now vertical downward load intensity
4g px
= cos per unit surface area
p L
[Note: It is first term of equivalent Fourier series for self weight].
Load components are,
X=0
4g px
Y= cos ◊ sin f
p L
4g px
Z= cos cos f
p L
From third equation of equilibrium,
4 gR px
N f = - ZR = -
cos cos f
p L
From the equation of equilibrium in φ-direction,
∂N x f 1 ∂N f
= + + Y = 0, we get,
∂x R ∂f
∂N x f 1 Ê 4 gR ˆ px px
˜ cos ( - sin f ) +
4g
+ Á- cos sin f = 0.
∂x RË p ¯ L p L
∂N xf 8g px
=- cos sin f
∂x p L
px
◊ sin f + f1 ( f )
8g L
∴ N xf = - ◊ sin
p p L
Due to symmetry at x = 0, Nxφ = 0
∴ 0 = 0 + f1(φ) i.e. f1(φ) = 0
8g px
∴ N xf = - ◊ sin f
L sin
p 2
L
From equation of equilibrium in x-direction,
∂N x 1 ∂N xf
+ +X =0
∂x R ∂f
∂N x 1 Ê 8 gL ˆ px
i.e. + Á - 2 ˜ sin cos f = 0
∂x RË p ¯ L
∂N x 8 gL px
∴ = 2 sin ◊ cos f
∂x p R L
8 gL LÊ px ˆ
Nx = ¥ Á - cos ˜¯ cos f + f2 f
p R 2 pË L
8 gL2 px
=- cos ◊ cos f + f2 ( f)
p R 3
L
L
At x= , N x = 0.
2
∴ 0 = 0 + f2 ( f)
8 gL2 px
∴ Nx = - cos ◊ cos f
p R
3 L
Thus in this case,
4gR px
Nf = - cos cosf
p L
8gL px
N xf = - sin ◊ sinf ...eqn. 15.7
p 2 L
8gL2 px
and Nx = - cos ◊ cosf
p R
3 L
15.2 CYLINDRICAL SHELLS WITH PARABOLIC, CATENARY AND CYCLOID

DIRECTRICES
Intrinsic equation is that equation which relates radius of curvature and the angle. It is interesting to
note that the intrinsic equations for parabolic, catenary, circular and cycloid are having general form.
R = R0 cosnφ
where Ro—radius of curvature at crown (φ = 0)
If n = −3, it is parabolic
= −2, it is catenary
= 0, it is circular, and
= 1, it is cycloid.
The shapes of these curves look as shown in Fig. 15.4.
Cycloid
Circular
Catenary
Parabolic
R0
fk
fk
Fig. 15.4
Hence, it is possible to get membrane solution for all the above 4 cases in the general form. The
examples 15.4 and 15.5 illustrate this point.
Example 15.4 Give membrane analysis of cylindrical shells with parabolic, catenary, circular and
cycloid subject to self weight g/unit surface area.
Solution. For self weight, X=0
Y = g sin φ
Z = g cos φ
From equation of equilibrium of forces in z-direction,
Nφ = −ZR
= − g cos φ.R0 cosnφ
= − gR0 cosn+1φ
From equation of equilibrium of forces in φ-direction, we have
∂N x f 1 ∂N f
+ +Y = 0
∂x R ∂f
∂N x f
(- gR0 ) (n + 1) cos n f ( - sin f) + g sin f = 0
1
+
∂x R
∂N x f (n + 1) gR0
+ cos n f sin f + g sin f = 0
∂x R0 cos fn
∂N xf
i.e. + ( n + 2 ) sin f = 0
∂x
∂N x f
i.e. = - g ( n + 2 ) sin f
∂x
Integrating both sides w.r.t. x, we get,
Nxφ = −gx(n + 2) sin φ + f1(φ)
where f1(φ) is constant of integration.
The boundary condition is, due to symmetry, at x = 0, Nxφ = 0.
i.e. 0 = 0 + f1(φ) i.e. f1(φ) = 0.
Hence, Nxφ = −gx (n + 2) sin φ
From the equation of equilibrium of forces in x-direction, we have,
∂N x 1 ∂N xf
+ + X = 0.
∂x R ∂f
∂N x
◊ ( - gx ) ( n + 2 ) cos f = 0
1
i.e. +
∂x R0 cos n f
∂N x gx ( n + 2 )
i.e. =
∂x R0 cos n -1 f
gx 2 ( n + 2 )
∴ Nx = + f2 ( f)
2 R0 cos n -1 f
As end frames are assumed as simple supports,

L
at x= , N x = 0.
2
gL2 ( n + 2 )
i.e. 0= + f2 ( f )
8 R0 cos n -1 f
gL2 ( n + 2 )
or f2 ( f) = -
8 R0 cos n -1 f
gx 2 ( n + 2) gL2 ( n + 2 )
∴ Nx = -
2 R0 cos n -1 f 8 R0 cos n -1 f
gL2 ( n + 2 ) Ê 4 x2 ˆ
=- Á 1 - ˜
8 R0 cos n -1 f Ë L2 ¯
Thus,
N f = - gR 0 cos n+1 f
N xf = - gx (n + 2 ) sin f …eqn. 15.8
gL2 (n + 2 ) Ê 4x 2 ˆ
Nx = - -
8R 0 cos n -1 f ÁË ˜
1
L2 ¯
n = −3 is for parabolic, n = −2 is for catenary, n = 0 is for circular and n = 1 is for cycloid.
Example 15.5. Find the membrane forces in the cylindrical shells with parabolic, catenary, circular and
cycloid directrices due to snow load of intensity p0 per unit horizontal area.
Solution. Let the intrinsic equation be R = R0 cosnφ
Snow load of p0/per unit horizontal area
= p0 cosφ/per unit surface area
∴ Load components are
X=0
Y = p0 cosφ sinφ
Z = p0 cos2φ
From the equation of equilibrium of forces in z-direction, we get,
N f = - ZR
= - p0 cos 2 f ◊ R0 cos n f
= - p0 R0 cos n + 2 f
From the equation of equilibrium of forces in φ-direction, we get,
∂N x f 1 ∂N f
+ +Y = 0
∂x R ∂f
∂N x f
◊ ( - p0 R0 ) ( n + 2 ) cos n +1 f ( - sin f ) + p0 cos f sin f = 0
1
+
∂x R0 cos f n
∂N x f
i.e. + p0 ( n + 2 ) sin f cos f + p0 cos f sin f = 0
∂x
∂N xf
i.e. = - p0 ( n + 3) sin f cos f
∂x
( n + 3)
= - p0 sin 2 f
2
p0 x ( n + 3)
∴ N xf = - ¥ sin 2 f + f1 ( f) .
2
From boundary condition, Nxφ = 0 at x = 0, we get,
0 = 0 + f1 ( f) or f1 ( f) = 0
p0 x ( n + 3)
∴ N xf = - sin 2 f
2
From the equation of equilibrium of forces in x-direction, we have
∂N x 1 ∂N f
+ +X =0
∂x R ∂f
∂N x n+3
( - p0 x )
1
i.e. + 2 cos 2 f = 0
∂x R0 cos f
n 2
∂N x cos 2 f
= p0 x ( n + 3)
∂x R0 cos n f
∴ N x = p0
x2
(n + 3) cos 2 nf + f2 ( f )
2 R0 cos f
From the boundary condition, Nx = 0 at x = L/2, we get,
0 = p0
L2
(n + 3) cos2 fn + f2 ( f) .
8 R0 cos f
f2 ( f) = - p0
L2
(n + 3) cos2 fn
8 R0 cos f
(n + 3) cos 2 fn - 0 (n + 3) cos 2 fn
p0 x 2 P L2
Hence, Nx =
2 R0 cos f 8 R0 cos f
Ê 2ˆ
(n + 3) cos 2 fn Á1 - 4 x2 ˜
p0 L2
=-
8 R0 cos f Ë L ¯
Thus, the solution is
N f = - p 0 R 0 cos n+2 f
- p 0 x (n + 3 ) … eqn. 15.9
N xf = sin 2 f
2
Ê 2ˆ
(n + 3 ) cos 2 fn Á 1 - 4x2 ˜
p 0 L2
Nx = -
8 R 0 cos f Ë L ¯
n = −3 is for parabolic
n = −2 is for catenary
n = 0 is for circular and
n = 1 is for cycloid
The membrane forces for cylindrical shells with different directrices due to self weight and snow
load are shown in Table 15.1.
Table 15.1 Membrane forces
Directrix Nφ Nxφ Nx
DL SL DL SL DL SL
1 1
Parabolic - - sin φ 0 cos4φ 0
cos2 f cos f
n = −3
1 sin 2 f
Catenary - −1 0 - 0 −cos2φ cos2φ
cos f 2
n = −2
3
Circular −cosφ −cos2φ −2sin φ - sin 2 f −2cosφ −3cos2φ
2
n=0
4cos2 f
Cycloid −cos2φ −cos3φ −3sin φ −2sin2φ −3 -
cos f
n=1
gL2 Ê 4x2 ˆ p0 L2 Ê 4 x2 ˆ
Common gR0 p0R0 gx p0x
8 R0 ÁË1 - 2 ˜¯ 8 R0 ÁË1 - 2 ˜¯
L L
multiplying
factor
15.3 CYLINDRICAL SHELL WITH SEMI-ELLIPTIC DIRECTRIX

Figure 15.5 shows a semiellipse with major axis = 2a and minor axis = 2b. Its radius of curvature at any
central angle φ is given by
a2b2
R=
(a 2 sin 2 f + b2 cos2 f)3 2
a a
f
R
Fig. 15.5
(1) For self weight X = 0, Y = g sin φ, Z = g cos φ

From equation of equilibrium for forces in z-direction, we have
Nφ = −ZR
a 2 b 2 cos f
= -g
(a 2 sin 2 f + b2 cos2 f)3 2
From equation of equilibrium in φ-direction,
∂N x f 1 ∂N f
+ +Y = 0
∂x R ∂f
∂N x f
=-
( a 2 sin 2 f + b 2 cos 2 f )
32
ga 2 b 2
È - sin f
∴ Í 2 2
∂x 2 2
a b (
ÍÎ a sin f + b 2 cos 2 f )3 2
+ cos f
2
(
3 2 2
a sin f + b 2 cos2 f )-5 2 (2 a 2 sin f cos f + 2b2 cos f ( - sin f))˘˙˚
= - g sin f Í1 +
(
È 3 a 2 cos 2 f - b 2 cos 2 f
+ 1˙
˘ )
ÍÎ (
a 2 sin 2 f + b 2 cos 2 f ˙˚ )
È 3 (a 2 - b 2 ) cos 2 f ˘
∴ N xf = - gx sin f Í 2 + 2 2 ˙ + f1 ( f )
ÎÍ a sin f + b 2 cos 2 f ˚˙
( )
-3 2
[Note: In differentiating φ, u = cos φ and v = a 2 sin 2 f + b 2 cos2 f . Hence, d (uv ) ]
dx
At x = 0, Nxφ = 0 (due to symmetry)
∴ f1(φ) = 0
È 3 (a 2 - b 2 ) cos2 f ˘
∴ N xf = - gx sin f Í 2 + ˙
ÍÎ a 2 sin 2 f + b 2 cos 2 f ˙˚
From equation of equilibrium of forces in x-direction, we have
∂N x 1 ∂N xf
+ + X = 0.
∂x R ∂f
∂N x 1 ∂N x f
=- since X = 0 .
∂x R ∂f
=+
( a 2 sin 2 f + b 2 cos2 f)
32
È
gx Í 2 cos f + 3 (a 2 - b 2 )
ÏÔ cos3 f - 2 cos f sin 2 f ¸Ô
Ì 2 2 ˝
a 2 b2 ÓÔ a sin f + b cos f ˛Ô
2 2
ÎÍ
cos2 f sin f 2 (a 2 - b 2 ) cos f sin f ˘

- ˙
a 2 sin 2 f + b 2 cos 2 f ˙˚
È 2 ab 3 ( a 2 - b 2 ) Ï 2 2a 2 ¸˘
= gx cos f Í 3 + Ìcos f - 2 sin f ˝˙
2
ÍÎ a aba Ó b ˛˙˚
ab
a=
(a 2 sin 2 f + b2 cos2 f)3 2
where
gx 2 È 2 ab 3 (a 2 - b 2 ) Ï 2 2a 2 ¸˘
∴ Nx = cos f Í + Ì cos f - sin 2 f˝ ˙ + f 2 ( f)
2 ÍÎ a 3
aba Ó b 2
˛ ˙˚
From B-C, Nx = 0 at x = L/2, f2(φ) may be found and on simplification we get,
- gL2 Ê 4 x2 ˆ È 2 ab 3 (a 2 - b 2 ) ˘ È 2 2a 2 ˘
Nx = ÁË 1 - 2 ˜
cos f Í + ˙ Í cos f - sin 2 f ˙
8 L ¯ Îa 3
aba ˚ Î b 2
˚
(2) For snow load:
Let snow load be p0 per unit horizontal load. Then
X = 0, Y = p0 cos φ sin φ Z = p0 cos2φ
Proceeding on the same line as earlier, it can be shown that
a 2 b 2 cos 2 f
N f = - p0
a 2 sin 2 f + b 2 cos 2 f
a 2 cos f sin f
N xf = -3 p0 x
a 2 sin 2 f + b 2 cos 2 f
3 p0 L2 Ê 4 x 2 ˆ b 2 cos2 f - a 2 sin 2 f
Nx = - ÁË 1 - ˜
8 ( )
L2 ¯ a 2 sin 2 f + b 2 cos2 f 1 2
15.4 COMMENTS ON MEMBRANE THEORY

1. A thin shell acts partially as an arch and partially as a beam. Due to arch action (in the form of
Nφ) the load is transferred to edge beam and due to beam action (in the form of Nxφ) the load is
transferred to the end frame.
2. If the directrix choosen is the funicular curve of the applied loading (catenary for self weight and
parabola for snow load), the shell degenerates itself into a series of independent arches and beam
action completely disappears.
3. The distribution of longitudinal force Nx across the cross section is not linear as in an ordinary
beam, except in the case of circular directrix under self weight.
4. The variation of Nφ is independent of x.
5. For a shell with free edges, Nφ and Nxφ should be zero at edges. But membrane theory gives
definite values. The membrane theory fails to give good results at edges.
In actual shells, edges bend in transverse direction (to avoid these unbalanced forces) introducing
transverse moment Mφ and radial shear. Earlier research was to assess these edge perturbations and
superpose them with membrane values.
QUESTIONS
1. Derive the differential equations of equilibrium using membrane forces only. Find the stresses Nx, Ny and Nxy
due to uniformly distributed load acting on the shell surface.
2. Find the membrane stresses in a circular cylindrical shell subject to snow load.
3. Find the membrane forces in a circular cylindrical shell subject to a sinusoidal load.
4. What is intrinsic equation? Give it for
(a) Circular
(b) Conoid
(c) Parabolic and
(d) Catenary directrices.
Derive the general membrane solution for above type of cylindrical shells subjected to self weight only.
5. Comment on membrane theory for cylindrical shells.
BENDING THEORY OF CYLINDRICAL SHELL ROOFS 197
Chapter
16
Bending Theory of Cylindrical
Shell Roofs
In the previous chapter, we have seen that membrane theory do not satisfy the edge conditions. This was
realized as early as in 1930. Though considerable load transfer in shells is by membrane action, bending
is not completely avoided. The moments and transverse shears act on the shell element. Hence, in
deriving the equilibrium equations for elements, moments and transverse shears should be considered.
Finsterwalder and Dishinger of Germany gave a bending theory in 1930. Schorer of America gave a
simplified solution in 1936. The contributions of Donnel (1933–34), Karman (1941) and Jenkins (1947)
lead to the bending theory known as DKJ theory which can be applied to circular cylindrical shells of
all dimensions. In this chapter, shell element and forces acting on it are explained, making sign conven-
tions clear. Then equations of equilibrium are derived and relations between stress resultants and radial
displacement ‘w’ are derived. As it is highly impossible to go ahead with exact relations to find the
solution, the assumptions made by various researchers are presented and the solution by Schorer and
DKJ are presented. Statical checks to be applied are also presented.
16.1 A TYPICAL SHELL ELEMENT

A typical shell element and various forces acting on it are shown in Fig. 16.1.
The coordinate system x, y, z is same as the one selected in membrane theory.
In the figure, all forces and moments are shown in their positive senses. It may be noted that the sign
convention followed is that on positive face if the force is acting in positive direction, it is positive. At
the same time, if the force is acting in negative direction on negative face, then also it is positive force.
The moments are positive if they are produced by positive forces acting in positive direction of z. This
convention results into tensile forces and sagging moments as positive. For shears and twisting moment
one should carefully note the sign conventions used. It is obvious that in the element shown:
∂N x ∂M x
N x+ = N x + dx M x+ = M x + dx
∂x ∂x
∂N f ∂M f
N f+ = N f + df M f+ = M f + df
∂f ∂f
∂N xf ∂M x f
N x+f = N xf + dx M x+f = M xf + dx
∂x ∂x
Nx Q x
Nxf
O
Qf
Nfx
z Nf
x
y
dx
+
+
Nfx
Nf
x
+
Qf
+ Nxf
f
+ +
df Q x Nx
(a) Shell element (b) Forces on shell element
Mxf
Mx
Mx +
Mxf Mf
Mfx Mfx
Mf
+
+
Mfx +
Mf Mfx
+
+ Mx
Mxf
+ Mf
Mx
+
Mxf
(c ) Moments on shell elements (d ) Another method of representing moments

(4 closed right hand fingers indicate direction
of moment and stretched thumb indicates axis
about which moment act)
Fig. 16.1
∂N fx ∂M fx
N f+x = N fx + df M f+x = M fx + df
∂f ∂f
∂Q x
Qx+ = Qx + dx X , Y , Z are the load components in x, y, z
∂x directions respectively.
∂Qf
Qf+ = Qf + df
∂f

Three equations of equilibrium can be found by considering force equilibrium in x, y and z directions
and another three by considering moment equilibrium. It may be noted that all forces and moments
shown on the element are in per unit length.
∑ Forces in x-direction = 0 →
∂N x ∂N x f
+ +X =0 …eqn. 16.1
∂x R∂f
∑ Forces in y-direction = 0 →
Ê ∂N f ˆ Ê ∂N x f ˆ
ÁË N f + ∂f d f˜¯ dx - N f dx + ÁË N xf + ∂x dx ˜¯ Rd f
È df Ê ∂Qf ˆ d f ˘
- N xf Rd f - ÍQf dx + Á Qf + d f˜ dx ˙ + YdxRd f = 0.
ÎÍ 2 Ë ∂f ¯ 2 ˚˙
Neglecting small quantities of higher order and dividing throughout by Rdxdφ, we get,
∂N x f ∂N f Qf
+ - +Y = 0 …eqn. 16..2
∂x R∂f R
∑ Forces in z-direction = 0 →
Ê ∂Q x ˆ Ê ∂ Qf ˆ
ÁË Q x + dx ˜ R d f - Q x Rd f + Á Qf +
¯
d f˜ dx
∂x Ë ∂f ¯
Ê ∂N f ˆ d f df
-Q x dx + Á N f + d f dx + N f dx + Zdx Rd f = 0
Ë R∂f ˜¯ 2 2
Neglecting small quantities of higher order and dividing throughout by Rdxdφ, we get,
∂Qx ∂Qf N f
+ + +Z =0 …eqn. 16.3
∂x R∂f R
Now consider the moment equilibrium conditions.
∑ Moments in x-direction = 0 →
Ê ∂M x ˆ Ê ∂M fx ˆ
ÁË M x + dx ˜ R d f - M x R d f + Á M fx +
¯
d f˜ dx - M fx dx - Qx Rd f dx = 0
∂x Ë ∂f ¯
Simplifying we get,
∂M x 1 ∂M x f
+ - Qx = 0 …eqn. 16.4
∂x R ∂f
∑ Moments in y-direction = 0 →
∂M x f 1 ∂M f
we get, + - Qf = 0 …eqn. 16.5
∂x R ∂f
∑ Moments about z-axis = 0, gives
Ê ∂M fx ˆ d f df
ÁË M fx + ∂f d f˜¯ dx 2 + M fx dx 2 + N xf Rd f dx - N fx dx Rd f = 0
Neglecting small quantities of higher order and dividing throughout by Rdxdφ, we get
M fx
+ N xf - N fx = 0 …eqn. 16.6
R
Thus we have got six equations of equilibrium and 10 unknowns i.e. Nx, Nφ, Nxφ, Nφx, Qx, Qφ, Mx, Mφ,
Mxφ and Mφx. Hence, the problem is not statically determinate. After studying deformations all stress
resultants are expressed in terms of single displacement i.e. ‘w’ in z-direction. Then a shell equation is
derived in terms of ‘w’ only, which will be 8th order differential equation. Solution of that equation
satisfying the boundary conditions gives expression for ‘w’, using which any stress resultant can be
found.
16.3 DEFORMATION IN SHELL

Let the origin of the coordinate system be at the crown of mid-span section. Let
x – be in the longitudinal direction
y or φ – be in the tangential direction and
z – be in the radial inward direction.
Consider the point A with coordinates x, φ and z, all having positive values. Let uA—displacement in
longitudinal direction, positive in the direction of increasing x.
vA— displacement along a circle of radius (R − z), positive in the direction of increasing φ and
wA— radial displacement, positive in the inward direction.
Let u, v and w be the displacement of the middle surface of the point which has also the coordinates
x and φ, but z = 0. The relations among these displacements will be derived first, considering the shell
is thin. The following additional assumptions are also made:
1. All points lying on a normal to the middle surface before deformation, do the same after defor-
mation also. In other words, shear deformations are considered negligible.
2. The stresses in radial direction (σz) is considered negligible and hence, the deformation in
z-direction is negligible. In other word point A remains at distance z even after deformation.
3. All displacements are small i.e. they are negligible compared to the radius of curvature of the
middle surface and that their first derivatives i.e. slopes are negligible compared with unity.
In short, we are considering small deflection theory of thin shells.

(i) Relation between uA and u
A0 and A are the original positions of two points (Ref. Fig. 16.2). A0 is on the middle surface
and A is at a distance z from A0 on the normal to the surface. After deformation let the points
move to the positions A0′ and A′ respectively. According to the assumptions even now A′ is on
normal to the middle surface and is at a distance z. From the figure, it is easily seen that,
∂w …eqn. 16.7
uA = u - z
∂x
A0 x axis
uA
w
z
dw
A0¢ dx
Bent middle
surface
z dw
z axis dx
A¢
Fig. 16.2
(ii) Relation between vA and v

Figure 16.3 shows a transverse section through the shell. The point A0 on the middle surface
is displaced by v along the middle surface to point A0′. The point A which was at distance z
from A0 is displaced to A′ and it remains at right angles to middle surface at A0′.
n
A0
A0¢
B
A
A¢
A² B0
w
nA
dw
z dw
Rdf
Fig. 16.3
Let A″ be at right angles to middle surface at A0′ at a distance z. Then the displacement of A may be
looked as consists of two parts:
R-z
(i) A moves to A″ due to shortening of radius = v
R
∂w ∂w
(ii) Then A″ moves to A′ due to rotation by angle i.e. by a distance z
R∂f Rd f
R-z 1 ∂w
∴ vA = v-Z …eqn. 16.8
R R ∂f
(iii) Relation between wA and w
Since, the shell is assumed thin and displacements are considered small, we get,
wA = w. …eqn. 16.9
16.4 STRAIN DISPLACEMENT RELATIONS

As a next step in assembling shell equation, we establish strain-displacement relations.
dx
A0 B0
B0′
A0′
u + ∂u dx
∂x
Fig. 16.4
Consider an element AB of length dx at the middle surface. Referring to Fig. 16.4, after deformation
let the element A0B0 move to the position A0′B0′. Let A0 move by distance u in x-direction. Then B0
∂u
moves by some other distance. If the rate of change of this displacement is , then BB′ is equal to
∂x
∂u
u+ dx.
∂x
∴ Strain εxc in x-direction is given by
∂u
u+ dx - u
e xc =
Extension
= ∂x
Original Length dx
∂u
=
∂x
∴ At A,
∂u A …eqn. 16.10
ex =
∂x
Referring to Fig. 16.5, the strain at middle surface in φ-direction is given by,
∂v v
v+ df - v A0
∂f w
e fc = - A¢
Rd f R B0 v + ∂v df
∂f
1 Ê ∂v ˆ
= Á - w˜
R Ë ∂f ¯
w ¢ w
[Note: is shortening due to radial
R
inward displacement w].
∴ At A, Fig. 16.5
1 Ê ∂v A ˆ
ef = Á - wA ˜ …eqn. 16.11
R - z Ë ∂f ¯
Referring to Fig. 16.6, in which shear deformations are shown in their positive senses, we get,
∂v 1 ∂v
g xfc = + ¶v dx
∂x R ∂f ¶x
∴ At A,
∂v A 1 ∂u A
g xf = + …eqn. 16.12 dx
∂x R - z ∂f
∂u A ∂v ∂2 w
ex = = -Z 2 …eqn. 16.13
∂x ∂x ∂x
Rd
f
From eqns. 16.8 and 16.11, we get

¶v
df
¶x
1 ∂v Z ∂ w2
w
ef = - -
R ∂f R ( R - z ) ∂f 2 R - Z Fig. 16.6
1 ∂v Z ∂2w w
= - -
R ∂f Ê Z ˆ ∂f 2
Ê Zˆ …eqn. 16.14
R 2 Á1 - ˜ R Á1 - ˜
Ë R¯ Ë R¯
Similarly from eqns. 16.9 and 16.12, we get
1 ∂u R - Z ∂v Ê Z Z ˆ ∂2w
g xf = + -Á + ˜
R - Z ∂f R ∂x Ë R R - Z ¯ ∂x∂f
1 ∂u Ê Z ˆ ∂v 1 Ê 1 ˆ ∂2w
= + Á1 - ˜ - 1+ …eqn. 16.15
Ê Z ˆ ∂f Ë R ¯ ∂f R Á Z ˜ ∂f 2
R Á1 - ˜ ÁË 1 - ˜¯
Ë R¯ R
16.5 RELATIONS BETWEEN STRESS RESULTANTS AND DISPLACEMENTS

For stress analysis linear elastic theory is used. Hence, from Hooke’s law,
(e x + me f )
E
sx = …eqn. 16.16
1 - m2
(me x + e f )
E
sf = …eqn. 16.17
1 - m2
E
t xy = g
2 (1 + m ) xf
and …eqn. 16.18
In the equations 16.16 to 16.18, by substituting the relations of strains with displacements
(eqns. 16.13 to 16.15) we get stress-displacement relations.
Now consider an element of unit dimension at middle section as shown in Fig. 16.7. At any distance
z from the middle surface the length of element in x-direction is unity but in φ-direction, it is different.
Referring to the Fig. 16.6,
Rdf = 1
1
or df =
R
Middle surface
Surface at distance - z
R
-Z
R
df
Fig. 16.7
Elemental length in φ-direction at distance z from middle surface

= (R - Z ) df
1
= (R - Z )
R
Ê Zˆ
= Á1 - ˜
Ë R¯
Hence, the stress resultants per unit length are as shown below:
h2 h2
Ê Zˆ
Nx = Ú s x ÁË1 - R ˜¯ dz Nf = Ú s f ¥ 1 ¥ dz
-h 2 -h 2
h2 h2
Ê Zˆ
N xf = Ú t xf Á1 - ˜ dz N fx = Ú t fx ¥ 1 ¥ dz
Ë R¯
-h 2 -h 2
h2 h2
Ê Zˆ
Mx = Ú s x ÁË1 - R ˜¯ z dz Mf = Ú s f ¥ 1 ¥ z ¥ dz
-h 2 -h 2
h2 h2
Ê Zˆ
M xf = Ú t xf Á1 - ˜ z dz M fx = Ú t fx ¥ 1 ¥ z ¥ dz
Ë R¯
-h 2 -h 2
In the above expressions by replacing stresses in terms of strains and then in terms of displacements,
stress resultants are obtained in terms of displacement. For example,
h2
Ê Zˆ
Nx = Ú s x Á1 - ˜ dz
Ë R¯
-h 2
h2
Ê Zˆ
Ú (e x + me f ) ÁË1 - R ˜¯ dz
E
=
1- m 2
-h 2
E Ê ∂u
h2
∂ 2 w m ∂v m Z ∂ 2 w w ˆÊ Zˆ
2 Ú Á ∂x
= - Z + - -m ˜ ÁË1 - ˜¯ dz
1 - m -h 2 Ë ∂x 2 R ∂f R R - Z ∂f 2 R-Z¯ R
E È ∂u
h2
∂ 2 w m ∂v mZ ∂2w mw ˘Ê Zˆ
2 Ú
= Í - Z + - - ˙ ÁË1 - ˜¯ dz
1 - m - h 2 Í ∂x ∂x 2 R ∂f R (1 - Z R ) ∂f
2 2
Ê Zˆ
R Á1 - ˜ ˙
R
ÍÎ Ë ¯
R ˚˙
E
h2 È Ê ∂u ∂ 2 w m ∂v Ê Z ˆ m Z ∂ 2 w mw ˆ ˘
= Ú Á Í
1 - m 2 -h 2 ÎÍ Ë ∂x
- z +
∂x 2 R ∂f Ë
Á 1 - ˜-
R ¯ R 2 ∂f 2
- ˙ dz
R ¯˜ ˚˙
Now rearranging by collecting terms without z, with z and with z2, we can write
E È Ê ∂u m ∂v mw ˆ Z Ê ∂u m ∂v m ∂ 2 w ˆ Z 2 ∂ 2 w ˘
h2
Nx = Ú Í + - - Á + - ˜+
1 - m 2 - h 2 ÎÍ ÁË ∂x R ∂f R ˜¯ R Ë ∂x R ∂f R ∂f 2 ¯ R ∂x 2 ˙˚
˙ dz
h h
Noting that integration in the limits - to
2 2
(a) of 1 is h
(b) of z is 0
h3
(c) of z2 is , we find
12
E È Ê ∂u m ∂v mw ˆ h3 ∂2 w ˘
Nx = Í + - h + ˙
1 - m 2 ÎÍ ÁË ∂x R ∂f R ˜¯ 12 R ∂x 2 ˚˙
Eh Ê ∂u m ∂v mw ˆ D ∂ 2 w
= + - + …eqn. 16.19
1 - m 2 ÁË ∂x R ∂f R ˜¯ R ∂x 2
Eh 3
D=
( )
where is flexural rigidity
12 1 - m 2
To find Nφ,
h2
Nf = Ú s f dz
-h 2
h2
(e f + me x ) dz
E
= Ú
-h 2 1 - m
2
E
h2È Ê 1 ∂v Z ∂2 w w ˆ Ê ∂u ∂2 w ˆ ˘
= Ú Á Í - - ˜
1 - m 2 - h 2 ÍÎ Ë R ∂f R ( R - Z ) ∂f 2 R - Z ¯
+ m ÁË ∂x - z ˜ ˙ dz
∂x 2 ¯ ˙˚
E
h2
È Ê 1 ∂v ∂u ˆ 1 Ê R ˆ ∂2 w w ∂2w ˘
= Ú Í
1 - m 2 - h 2 ÎÍ ËÁ R ∂f
+ m + Á
∂x ¯˜ R Ë
1 - ˜ -
R - Z ¯ ∂f 2 R - Z
- mZ ˙ dz
∂x 2 ˚˙
Note that
h2
1 Ê hˆ Ê hˆ
= ÈÎ log ( R - Z )˘˚ - h 2 = log Á R +
h2
I1 = Ú R-Z Ë 2
˜¯ - log ÁË R - ˜¯
2
-h 2
Ê hˆ Ê h ˆ
R+ 1+
Á ˜
2 = log Á 2R ˜
= log Á ˜ Á
h h ˜
ÁR- ˜ Á1- ˜
Ë 2¯ Ë 2R ¯
Ê h ˆ Ê h ˆ
= log Á 1 + ˜¯ - log ÁË1 - ˜
Ë 2R 2R ¯
x2 x3 x 4
Noting that log (1 + x ) = x - + - + ...
2 3 4
È x2 x3 x 4 ˘
and log (1 - x ) = - Í x + + + + ...˙
Î 2 3 4 ˚
h h2 h3 h4 Ê h h2 h3 h4 ˆ
We find I1 = - + - + Á + + + ˜
2R 8R 2
24 R 3
4 ¥ 64 R 4 Ë 2R 8R 2
24 R 3
4 ¥ 64 R ¯
4
h h3
= + .
R 12 R 3
Integration of 1 results into h and integration of Z results into zero.
E È Ê 1 ∂v ∂u ˆ 1 ∂2 w Ê h3 ˆ Êh h3 ˆ ˘
∴ Nf = Í + m + Á - - ˜ - + ˙
ËÁ R 12 R 3 ¯˜ ˚˙
h h h w
1 - m 2 ÎÍ ËÁ R ∂f ∂x ¯˜ R ∂f 2 Ë 12 R 2 ¯
E È Ê 1 ∂v ∂u w ˆ Ê ∂2w ˆ h3 ˘
= Í + m - h - Á 2 + w ˜ ˙
1 - m 2 ÎÍ ÁË R ∂f ∂x R ˜¯ Ë ∂f ¯ 12 R 3 ˚˙
Eh È 1 ∂v ∂u w ˘ D Ê ∂2w ˆ
= + m - - +
∂x R ˚˙ R 3 ÁË ˜
Í w …eqn. 16.20
1 - m 2 Î R ∂f ∂f 2 ¯
Proceeding on the same line all stress resultants can be obtained as shown below:
Eh È ∂u m ∂v w ˘ D ∂2w
Nx = Í + - m +
1 - m 2 Î ∂x R ∂f R ˙˚ R ∂x 2
Eh È 1 ∂v w ∂u ˘ D Ê ∂2w ˆ
Nf = Í - + m ˙ - Á w + ˜
1 - m 2 Î R ∂f R ∂x ˚ R 3 Ë ∂f 2 ¯
Eh Ê 1 - m ˆ È 1 ∂u ∂v ˘ D 1 - m Ê ∂u ∂2w ˆ
N fx = Á ˜Í + ˙ - Á + ˜
1 - m 2 Ë 2 ¯ Î R ∂f ∂x ˚ R 2 2 Ë R∂f ∂x∂f ¯
Eh 1 - m È 1 ∂u ∂v ˘ D 1 - m Ê ∂v 1 ∂ 2 w ˆ
N xf = Í + ˙- Á - ˜
1 - m 2 2 Î R ∂f ∂x ˚ R 2 2 Ë ∂x R ∂x∂f ¯
È ∂ 2 w m ∂ 2 w 1 ∂v m ∂ v ˘
Mx = -D Í 2 + 2 + + ˙
ÎÍ ∂x R ∂f 2 R ∂x R 2 ∂f ˙˚
D Ê ∂2 w 2 ˆ
2 ∂ w
Mf = - Á w + + mR ˜
R2 Ë ∂f 2 ∂x 2 ¯
È 1 ∂2w 1 ∂u 1 ∂v ˘
M fx = - D (1 - m ) Í - + ˙
Î R ∂x∂f 2 R ∂f 2 R ∂x ˚
2
È 1 ∂ 2 w 1 ∂v ˘
M xf = - D (1 - m ) Í + ˙
Î R ∂x∂f R ∂x ˚
The above relations are known as ‘Stress Resultants to Flugge’s Accuracy’.
16.6 SIMPLIFIED STRESS RESULTANTS AND DISPLACEMENT RELATIONS

As it is very difficult to make use of the stress resultants to Flugge’s accuracy, researchers made
assumptions to form shell equations. Usual assumptions are (i) to consider the strains across a section
uniform (ii) take the value of Poisson’s ratio µ = 0. Finsterwalder and Schorer went a step forward and
assumed Mx = 0 and Mxφ = 0. Table 16.1 shows the simplified stress resultant – displacement relations
assumed by different researchers. It may be noted that except Holland, other three have omitted the
terms which do not figure in ‘Disk action, Plate action and Membrane action’.
16.7 DKJ THEORY

The assumptions Mx = Mxφ = 0, make Finsterwalder and Schorer theories applicable only to long shells.
Holland and DKJ theories can be applied to all classes of shells. However, DKJ theory is commonly
used shell theory.
In 1933-34, Donnel developed the theory for his studies on stability of thin walled circular cylinders.
Kármán and Tsien employed the same theory in 1941 for their investigations on the buckling of cylin-
drical shells. In 1947, Jenkins published a book in which he presented the theory developed by Donnel
and Kármán in the form suitable for the analysis of cylindrical shell roof. Hence, the theory in now
known as Donnel-Kármán-Jenkins in short DKJ theory.
Table 16.1 Simplified stress resultants—Displacement relations
Stress resultant Finsterwalder Schorer Holland DKJ
∂u ∂v ∂v ∂v
Nx Eh Eh Eh Eh
∂x ∂x ∂x ∂x
Ê ∂v wˆ Ê ∂v wˆ Ê ∂v wˆ Ê ∂v wˆ
Eh Á - Eh Á - Eh Á - Eh Á -
Nφ Ë R∂f R ˜¯ Ë R∂f R ˜¯ Ë R∂f R ˜¯ Ë R∂f R ˜¯
Eh Ê 1 ∂u ∂v ˆ Eh Ê 1 ∂u ∂v ˆ Eh Ê 1 ∂u ∂v ˆ Eh Ê ∂u ∂v ˆ
+ + + +
Nxφ = Nφx 2 ÁË R ∂f ∂x ˜¯ 2 ÁË R ∂f ∂x ˜¯ 2 ÁË R ∂f ∂x ˜¯ 2 ÁË R∂f ∂x ˜¯
∂2 w ∂2 w
Mx 0 0 -D -D
∂x 2
∂x 2
D ∂2 w D ∂2 w Ê 1 ∂2 w ˆ D ∂2 w
Mφ - - -D Á 2 + w˜ -
R 2 ∂f 2 R 2 ∂f 2 Ë R ∂f 2
¯ R 2 ∂f 2
Ê 1 ∂ 2 w 1 ∂v ˆ D ∂2 w
Mxφ = Mφx 0 0 -D Á + ˜ -
Ë R ∂x∂f R ∂x ¯ R ∂x∂f
In this theory (and also in Shorer’s theory) the term Qφ appearing in equation of equilibrium 16.2 is
dropped, since such term do not appear in the corresponding equation of equilibrium of disc, plate or
the membrane shell. Hence, the equation is rewritten as,

∂N x f 1 ∂N f
+ +Y = 0.
∂x R ∂f
Since, in this theory, Nxφ is taken as equal to Nφx, equation of equilibrium 16.6 shows Mxφ = 0. But in
D ∂2 w
this theory Mxφ is taken as - . Thus, these is clear violation of fundamental principle of mechan-
R ∂x∂f
ics. However, still this theory is used, since, Mxφ is generally small quantity.
Since, Nxφ = Nφx and Mxφ = Mφx, the theory is developed below using Nxφ for shearing forces and Mxφ
for twisting moment.
We have, now, the following equilibrium equations.
∂N x 1 ∂N xf
+ +X =0 ...(1)
∂x R ∂f
∂N x f 1 ∂N f
+ +Y = 0 ...(2)
∂x R ∂f
∂Qx 1 ∂Qf N f
+ + +Z =0 ...(3)
∂x R ∂f R
∂M x 1 ∂M x f
+ - Qx = 0 ...(4)
∂x R ∂f
∂M x f 1 ∂M f
+ - Qf = 0 ...(5)
∂x R ∂f
The following are the stress-resultant and displacement relations:
∂u
N x = Eh ...(6)
∂x
Ê ∂u wˆ
N f = Eh Á - ˜ ...(7)
Ë R∂f R ¯
Eh Ê 1 ∂u ∂v ˆ
N x f = N fx = +
2 ÁË R ∂f ∂x ˜¯ ...(8)
∂2 w
Mx = -D ...(9)
∂x 2
1 ∂2w
Mf = -D ...(10)
R 2 ∂f 2
D ∂2 w
and M xf = M fx = - ...(11)
R ∂x∂f
Thus, we have 11 equations and number of unknowns are—

Nx, Nφ, Nxφ = 3
Qx, Qφ = 2
Mx, Mφ, Mxφ = 3
u, v and w = 3
Total 11
Hence, the problem can be solved.
From equation (5),
∂M x f 1 ∂M f
Qf = +
∂x R ∂f
D ∂3 w D ∂3 w
=- -
R ∂x 2 ∂f R 3 ∂f 3
D ∂ Ê ∂2w 1 ∂2w ˆ
=- +
R ∂f Ë ∂x 2 R 2 ∂f 2 ¯˜
Á
D ∂ ( 2
=- — w) ...(12)
R ∂f
From eqn. (4),
∂M x 1 ∂M x f
Qx = +
∂x R ∂f
∂3 w 1 Ê D ˆ ∂3 w
= -D + Á- ˜
∂x 3 R Ë R ¯ ∂x∂f 2
È ∂3 x 1 ∂3 w ˘
= -D Í 3 + 2 ˙
ÎÍ ∂x R ∂x∂f 2 ˙˚
∂ Ê ∂2w 1 ∂2 w ˆ
= -D +
∂x ËÁ ∂x 2 R 2 ∂f 2 ¯˜
∂ ( 2
= -D — w) ...(13)
∂x
From eqn. (3),
∂Qx ∂Qf
Nf = -R - - ZR
∂x ∂f
∂ 2 (— 2 w ) D ∂ 2 (— 2 w )
= +R ◊ D + - ZR
∂x 2 R ∂f 2
È ∂ 2 (— 2 w ) 1 ∂ 2 ˘
= DR Í + ( — 2 )
w ˙ - ZR
ÍÎ ∂x R 2 ∂f 2 ˙˚
2
Ê ∂2 1 ∂2 ˆ
= DR Á 2 + 2 2 ˜ (— 2 w ) - ZR
Ë ∂x R ∂f ¯
= DR — 2 (— 2 w ) - ZR
= DR (— 4 w ) - ZR ...(14)
From eqn. (2),
∂N x f ∂N f
=- -Y
∂x R∂f
∂ ( 4 ∂Z
= -D — w) + -Y ...(15)
∂f ∂f
From eqn. (1),
∂N x ∂N x f
=- -X
∂x R∂f
∂2 N x ∂ 2 N xf ∂X
∴ =- -
∂x 2
R∂x∂f ∂x
D ∂ 2 (— 4 w ) ∂ 2 Z 1 ∂Y ∂X
= - + - ...(16)
R ∂f 2
R∂f 2 R ∂f ∂x
Thus, now we have expressed all stress resultants in only one unknown ‘w’. To form compatibility
equations consider the force displacement equations 6, 7 and 8.
∂u
N x = Eh ...(6)
∂x
Ê 1 ∂u w ˆ
N f = Eh Á -
Ë R ∂f R ˜¯
...(7)
Eh Ê 1 ∂u ∂v ˆ
N xf = +
and 2 ÁË R ∂f ∂x ˜¯ ...(8)
To eliminate v in the above equations:

∂N f Ê 1 ∂2 v 1 ∂w ˆ
= Eh Á -
Ë R ∂x∂f R ∂x ˜¯
...(a)
∂x
1 ∂N xf Ê 1 ∂ 2u 1 ∂2 v ˆ
= Eh Á 2 2 +
R ∂x∂f ˜¯
2 ...(b)
R ∂f Ë R ∂f
Subtracting eqn. (b) from eqn. (a), we get,

∂N f 1 ∂N x f 1 ∂w 1 ∂2u
-2 = - Eh - Eh 2
∂x R ∂f R ∂x R ∂f 2
È 1 ∂w 1 ∂2u ˘
= - Eh Í + 2 2˙ ...(17)
ÎÍ R ∂x R ∂f ˚˙
To eliminate u from eqn. (6) and (17):
1 ∂2 N x 1 ∂ 3u
From eqn. 6, = ◊ Eh ...(c)
R 2 ∂f 2 R2 ∂x∂f 2
∂2 N f 2 ∂ N xf
2
È ∂2w 1 ∂ 3u ˘
From eqn. 17, - ◊ = - Eh Í + ˙ ...(d)
∂x 2 R ∂x∂f ÎÍ R∂x
2
R 2 ∂x∂f 2 ˚˙
Adding eqn. (c) and eqn. (d) we get,
∂2 N f 1 ∂2 N x 2 ∂ N xf
2
∂2w
+ - = - Eh
∂x 2 R 2 ∂f 2 R ∂x∂f R∂ x 2
Differentiating both sides twice w.r.t. x we get,
∂4 N f 1 2 ∂ N xf
4
∂4 N x Eh ∂ 4 w
+ 2 2 2 - = -
∂x 4 R ∂x ∂f R ∂x 3 ∂f R ∂x 4
∂N x f ∂2 N x
Substituting the values of N f , and from eqn. 14, 15 and 16, we get equations.
∂x ∂x 2
∂ 4 (— 4 w ) ∂4Z 1 È D ∂ 4 (— 4 w ) ∂ 4 Z 1 ∂ 3Y ∂3 X ˘
DR -R + Í - + - ˙
∂x 4 ∂x 4 R 2 ÎÍ R ∂f 4 R∂f 4 R ∂f 3 ∂x∂f 2 ˚˙
2È ∂ 4 (— 4 w ) ∂4Z ∂ 3Y ˘ Eh ∂ 4 w
- Í-D + - ˙ = -
R ÎÍ ∂ x 2 ∂y 2 ∂x 2 ∂f 2 ∂x 2 ∂f ˚˙ R ∂x 4
Dividing throughout by R and bringing load terms to right hand side, we get
È ∂ 4 (— 4 w ) 2 ∂ 4 (— 4 w ) ∂ 4 (— 4 w ) ˘ Eh ∂ 4 w
DÍ + 2 + ˙+
ÎÍ ∂x R ∂x 2 ∂f 2 R 4 ∂f 4 ˚˙ R 2 ∂x 4
4
È ∂4 Z 1 ∂4Z 2 ∂ 4 Z ˘ 1 ∂ 3Y 2 ∂ 3Y 1 ∂3 X
=Í 4 + 4 + ˙ - - +
ÎÍ ∂x R ∂f 4 R 2 ∂x 2 ∂f 2 ˚˙ R 4 ∂f 3 R 2 ∂x 2 ∂f R 2 ∂x∂f 2
∂4w 1 ∂ 3Y ∂ 3Y ∂3 X
D— 4 (— 4 w ) + Eh
2 1
i.e. = — 4Z - - +
∂x 4 R 4 ∂f3 R 2 ∂x 2 ∂f R 2 ∂x∂f 2
Eh ∂ 4 w 1È 4 1 ∂ 3Y 2 ∂ 3Y 1 ∂3 X ˘
i.e. —8 w + = Í— Z = 4 - + ˙
R 2 D ∂x 2 D ÍÎ R ∂f3 R 2 ∂x 2 ∂f R 2 ∂x∂f 2 ˙˚
where ∇w8 = ∇4(∇w4)
Ê ∂4 2∂ 4 ∂4 ˆ Ê ∂4w 2 ∂4w ∂4w ˆ
=Á 4 + 2 2 2 + 4 4 ˜Á 4 + 2 2 2 + 4 4˜
Ë ∂x R ∂x ∂f R ∂ f ¯ Ë ∂x R ∂x ∂f R ∂f ¯
∂8 w 4∂8 w 6 ∂8 w 4∂ 8 w ∂8 w
= + + + +
∂x 8 R 2 ∂x 6 ∂f 2 R 4 ∂x 4 ∂f 4 R 6 ∂x 2 ∂f 6 R 8 ∂f8
Eh
Since m = 0, D = = Eh.
1 - m2
∴ The eqn. of shell is,
h ∂4 w 1 È 4 1 Ê 1 ∂3 Y 2 ∂3 Y ˆ 1 ∂3X ˘
—8 w + = Í—Z - Á 3 + + ˙ ...eqn. 16.21
R 2 I ∂x 4 D ÍÎ R Ë R ∂f 3 R ∂x 2 ∂f ˜¯ R 2 ∂x∂f 2 ˙˚
Equation 16.21 may be called as shell equation as per DKJ theory.

The solution of the above equation for given loading gives expression for w at any point and knowing
‘w’, any stress resultant may be found.
16.8 DETERMINATION OF PARTICULAR INTEGRAL FOR DEAD LOAD

Consider the case of dead load g/unit surface area. Expanding it in the form of Fourier series we get,
•
4 npx
g=g Â m p
cos
L
m =1,3,...
Taking only first term of loading, we get,

4g px
gª cos
mp L
4g p
= g ¢ cos kx where g ¢ = and k =
mp L
Dead load components are,
X = 0, Y = g ¢ sin f cos kx and z = g ¢ cos f cos kx .
Let w be in the form,
w = Cg ¢ cos f cos kx
Then from plate equation, we get,
h ∂4 w 1 ÈÊ 4 2k 2 1 ˆ 1Ê 1 2 ˆ˘
—8 w + = Í Á k + 2 + 4 ˜ - Á - 3 - k 2 ˜ ˙ g ¢ cos f cos kx
R 2 I ∂x 4 D ÎÍ Ë R R ¯ RË R R ¯ ˚˙
h ∂ 4 w = 1 È k 4 + 2 k + 1 + 1 + 2 k ˘ g ¢ cos kx cos f
4
Ê ∂2 1 ∂2 ˆ
2 2
i.e. Á 2 + 2 2˜ w+ 2 Í ˙
Ë ∂x R ∂f ¯ R I ∂x 4 DÎ R2 R4 R4 R2 ˚
ÈÊ 1 ˆ
4
h ˘ 1È 4k 2 2 ˘
i.e. ÍÁ k 2 + 2 ˜ + 2 k 4 ˙ C = Ík 4 + 2 + 4 ˙
ÎÍ Ë R ¯ R I ˙˚ DÎ R R ˚
∴ Multiplying both sides by R8, we get
È 2 2 hR 6 4 ˘ R4 4 4
Í ( k R + 1) 4
+ k ˙ C = [k R + 4 k 2 R 2 + 2 ]
Î I ˚ D
R4 k 4 R 4 + 4k 2 R2 + 2
∴ C= ...eqn. 16.22
D (1 + k 2 R 2 )4 + k 4 R 4 h R 2
I
In an analysis problem, all terms in RHS of eqn. 16.22 are known. Hence, C is found and hence, w
is found. Substituting the expression of w in eqns. 6 to 16, expressions for all stress resultants and
displacements can be determined. These expressions are given below:
EI
M f = 2 Cg ¢ cos f cos kx
R
M x = EIk 2Cg ¢ cos f cos kx
EIk
M x f = M fx = - Cg ¢ sin f sin kx
R
Ê k2 ˆ
Q x = - EI Á k 3 + ˜ Cg ¢ cos f ◊ sin kx
Ë R¯
Ê 1 k2 ˆ
Qf = - EI Á 3 + ˜ Cg ¢ sin f cos kx
ËR R¯
N f = R ( H1 - 1) g ¢ cos f cos kx
Ê 2k 2 1 ˆ
where H1 = EI Á k 4 + 2 + 4 ˜ C + 2
Ë R R ¯
H1
N xf = N fx = g ¢ sin f sin kx
k
H1
Nx = g ¢ cos f cos kx
k2R
H1
u= g ¢ cos f cos kx
Ehk 3 R
H1 H 2
v=- g ¢ sin f sin kx
k
1 È 1 ˘
where H2 = Í 2+ 2 2˙
Ehk Î k R ˚
and w = Cg ¢ cos f cos kx ...eqn. 16.23

If for any particular shell if the above values are calculated, it will be found that Mx, Mφ, Mxφ, Qx and
Qφ values are very small (hardly 0.3 percent of final values) and Nx, Nφ, Nxφ values are almost equal to
those obtained from membrane theory. Hence, we replace particular solution by membrane solution.
16.9 HOMOGENEOUS SOLUTION

For this part, the equation is,
h ∂4 w
—8 w + =0
R 2 I ∂x 4
The solution consists of two parts as shown below:
npx nf
w = f ( f ) cos + f ( x )cos .
L fk
The first part gives the disturbances from the edge and the second part gives the disturbances from
the traverses. As the equation is of the eighth order, there will be eight constants in the first term and
eight in the second term. If the boundary conditions at traverses are assumed as simply supported only
first term may be considered. In that case
n px
w = f ( f ) cos
L
npx
= He mf cos
L
considering only first term of ‘w’,
p
w = He mf cos kx where k =
L
The equation reduces to the form,
4
Ê 2 1 2ˆ h 4
ÁË - k + 2 m ˜¯ + 2 k = 0
R R I
6
or (m 2 - k 2 R 2 )4 + hR k 4 = 0.
I
hR 6 4
Substituting r8 = k
I
hR 6 12 R 6 k 4
= k4 = ,
1
¥ 1 ¥ h3 h2
12
We get, (m 2 - k 2 R 2 )4 + r8 = 0
or (m 2 - k 2 R 2 )4 = -r8
4
Ê m2 - k 2 R2 ˆ
i.e. Á ˜ = -1 = cos (2 np + p ) + i sin (2 np + p )
Ë r2 ¯
m2 - k 2 R2 2 np + p ( 2 np + p )
∴ = cos + i sin where n = 0, 1, 2, 3, ...
r 2 4 4
1
=± (1 ± i )
2
È k 2 R2 1 ˘
∴ m 2 = r2 Í 2 ± (1 ± i )˙
ÎÍ r 2 ˚˙
r2 È k 2 R 2 ˘
= Í 2 ± 1 ± i˙
2 ÍÎ r2 ˙˚
r2
= [g ± 1 ± i]
2
k 2 R2
where g= 2
r2
r
m = ± 4 [g ± 1 ± i]
12
∴
2
The eight roots m1, m2, ..., m8 may now be written as,
m1 = α1 + iβ1 m5 = −m1
m2 = α1 − iβ1 m6 = −m2
...(a)
m3 = α2 + iβ2 m7 = −m3
m4 = α2 − iβ2 m8 = −m4
where
12
È
r Í (1 + g )2 + 1 + 1 + g ˘˙
a1 = 4
2 ÍÎ 2 ˙˚
12
È
r Í (1 - g )2 + 1 - (1 - g ) ˘˙
a2 = 4
2 ÍÎ 2 ˙˚
12
È
r (1 + g )2 + 1 - (1 + g ) ˘˙
b1 = 4 Í
2 ÎÍ 2 ˚˙
12
È
r Í (1 - g )2 + 1 + (1 - g ) ˘˙
and b2 = 4 ...(b)
2 ÍÎ 2 ˙˚
Thus, w = ÈÎ H1e m1f + H 2 e m2 f + ... + H8 e m8 f ˘˚ cos kx ...(c)

where H1, H2, ..., H8 are arbitrary constants and are to be determined from boundary conditions at edges.
Substituting for m1, m2, ..., m8 in terms of α and β (from eqn. (a)), we get,
w = ÈÎ H1e ( 1 1 ) + H 2 e( 1 1 ) + H 3 e( 2 2 ) + H 4 e( 2 2 )
a + ib f a - ib f a + ib f a - ib f
+ H 5 e ( 3 3 ) + H 6 e ( 1 1 ) + H 7 e ( 2 2 ) + H 8 e ( 2 2 ) ˘˚ cos kx
- a + ib f - a - ib f - a + ib f - a - ib f
= ÈÎ H1e a1f (cos b1f + i sin b1f ) + H 2 e a1f (cos b1f - i sin b1f )
+ H 3 e a 2 f (cos b 2 f + i sin b 2 f ) + H 4 e a 2 f (cos b 2 f - i sin b 2 f )
+ H 5 e -a1f (cos b1f + i sin b1f ) + H 6 e -a 2 f (cos b1f - i sin b1f )
+ H 7 e -a 2 f (cos b 2 f + i sin b 2 f ) + H 8 e -a 2 f (cos b 2 f - i sin b 2 f ) ˘˚ cos kx
= ÈÎ e a1f ( H1 + H 2 ) cos b1f + i ( H1 - H 2 ) e a1f sin b1f
+ ea 2 f ( H 3 + H 4 ) cos b 2 f + i ( H3 - H 4 ) ea 2 f sin b2 f
+ e -a1f ( H 5 + H 6 ) cos b1f + i ( H 5 - H 6 ) e -a1f sin b1f
+ e -a 2 f ( H 7 + H 8 ) cos b 2 f + i ( H 7 - H 8 ) e -a 2 f sin b 2 f ˘˚ cos kx

It is to be noted that the arbitrary constants H1, H2, ..., H8 are complex numbers. Since, w is real, it
follows that (H1 + H2), i (H1 − H2), ..., i (H7 − H8) should be real. It means that H1 and H2, H3 and H4,
H5 and H6, H7 and H8 should be conjugate pairs. Introducing real constants as,
A1 = H1 + H2, A2 = i (H1 − H2), A3 = H3 + H4, A4 = i (H3 − H4)
A5 = H5 + H6, A6 = i (H5 − H6), A7 = H7 + H1 and A8 = i (H7 − H6),
we get,
w = ÈÎ A1e a1f cos b1f + A2 e a1f sin b 1f + A3 e a 2 f cos b 2 f + A4 e a 2 f sin b2 f
+ A5 e -a1f cos b1f + A6 e -a1f sin b1f + A7 e -a 2 f cos b 2 f + A8 e -a2 f sin b2 f ˘˚ cos kx
A1 + A5 A - A5
Let = C1 and 1 = C2
2 2
then, A1 = C1 + C2, A5 = C1 − C2
∴ A1e a1f cos b1f + A2 e -a1f cos b1f
= ÈÎ (C1 + C2 ) e a1f + (C1 - C 2 ) e -a1f ˘˚ cos b1f
= C1 (e a1f + e -a1f ) cos b1f + C 2 (e a1f - e -a1f ) cos b1f

= 2C1 cosh α1φ cos β1φ + 2C2 sinh α1φ cos β1φ
Similarly, substituting A3 = C3 + C4, A4 = C3 − C4, A5 = C5 + C6, A6 = C5 − C6, A7 = C7 + C8 and
A8 = C7 − C8,
we get,
w = 2[C1 cosh α1φ cos β1φ + C2 sinh β1φ cos β1φ
+ C3 cosh α1φ sin β1φ + C4 sinh α1φ ⋅ sin β1φ
+ C5 cosh α2φ cos β2φ + C6 sinh β2φ cos β2φ
+ C7 cosh α2φ sin β2φ + C8 sinh α2φ sin β2φ] cos kx ...(16.24)
For symmetric loading, anti-symmetric terms should vanish. Hence, displacement for symmetric
case is,
w = 2[C1 cosh α1φ cos β1φ + C4 sinh α1φ sin β1φ
+ C5 cosh α2φ cos β2φ + C8 sinh α2φ sin β2φ] cos kx
= 2 (a cos β1φ cosh α1φ − b sin β1φ sinh α1φ
+ c cos β2φ cosh α2φ − d sin β2φ sinh β2φ) cos kx ...(16.25)
where C1 = a, C2 = −b, C3 = c and C4 = −d.
Substituting the value of w in the stress resultant expressions 16.23, all stress resultants can be
expressed in ‘w’. In Table 16.2 and 16.3, they are arranged in the convenient form in which,
F – represent stress resultant/displacement a, b, c and d are arbitrary constants
∂M x ¸
Qx¢ = Q x +
∂x ÔÔ
∂M f ˝ Vertical reactions
Qf¢ = Qf + Ô
R∂r Ô˛
a1 4 2 4
2
m1 = m2 = a 2 ◊
r r
b1 4 2 b2 4 2
n1 = n2 =
r r
Table 16.2 Stress resultants and displacements due to complementary function (Even)
F=R [(aB1 − bB2) cosβ1φ cosh α1φ − (aB2 + bB1) sinβ1φ sinh α1φ
+(cB3 − dB4) cosβ2φ cosh α2φ − (cB4 + dB3) sinβ2φ sinh α2φ]
F R B1 B2 B3 B4
2 EI r2 r2 r2 r2
Mφ 2
cos kx (1 + g ) ( g - 1)
R 2 2 2 2
Mx -2 EIk cos kx 2 1 0 1 0
3
2 EIk
Qx - sin kx 1 1 −1 1
g
2k 4
Nφ -2 EI cos kx 0 1 0 −1
g2
2k 4
Nx 2 EIR cos kx −1 1+γ 1 1−γ
g3
2 EIk 3
Qx′ - sin kx γ+2 2 γ−2 2
g
2k 3
u 2 EI sin kx −1 1+γ 1 1−γ
hg 3
w 2 cos kx 1 0 1 0
Table 16.3 Stress resultants and displacements due to complementary function (Odd)
F=R [(aB1 − bB2) cosβ1φ sinh α1φ − (aB2 + bB1) sinβ1φ cosh α1φ
+(cB3 − dB4) cosβ2φ sinh α2φ − (cB4 + dB3) sinβ2φ cosh α2φ]
F R B1 B2 B3 B4
2 EIk 3
Qφ cos kx m1 – n1 m1 + n1 – m 2 – n2 m2 – n2
( g )3 2
2 EIk 3
Qφ′ cos kx m1 (1–γ) – n1 m1 + n1 (1–γ) – m2 (1+γ) – n2 m2 – n2 (1 + γ)
g3 2
2k 4
Nxφ 2 EIR sin kx −n1 m1 n2 −m2
g3 2
Mxφ -2 EIK α1 β1 α2 β2
sin kx
R
2k3
v -2 IR cos kx m1 + n1(1−γ) n1 − m1(1−γ) − m2 + n2(1+γ) −n2 − m2(1+γ)
hg 7 2
2 a1 ( RB1 ) v 2b1 ( RB2 ) v 2 a 2 ( RB3 ) v 2b 2 ( RB4 ) v
θ cos kx - + - + - + - +
R R R R R R R R
Note: In the Table 16.3, θ is the rotation of the tangent.

v
A v R
B
w
A¢
¶w
R¶f
B¢
w + ¶w
v
R
Fig. 16.8

∂w
Due to differential radial displacement, tangent at A will rotate by R∂f
Because of circumferential displacement v, at point A, tangent rotates by additional angle v/R,

∂w v
∴ q= + ...eqn. 16.26
R∂f R
16.10 EDGE CONDITIONS

To determine the arbitrary constants in the expression for displacement w, we have to make use of the
edge conditions. The edge conditions in a shell naturally depend upon the method of supporting the
edges. At each edge φ = φk, one has to look for any four of the moments, forces and displacements at
edges, namely any four of Mφ, Nφ, Qφ′, Nφx, u, v, w and θ. Such conditions for some of the common
support conditions are presented below:
1. Edge Unsupported
In this case, since the shell edge is free, the forces and moments at the edges φ = φk, will be zero. Thus,
Mφ = 0, Nφ = 0, Qφ′ = 0 and Nφx = 0. ...eqn. 16.27
2. Edge Supported on Unyielding Wall

In such case, the wall restricts vertical and horizontal displacement at the shell but allows rotation.
Horizontal and vertical components are found by taking components of v and w. They are
w sin Qk − v cos φk and w cos φk + v sin φk.
Hence, the boundary conditions are,

w sin Qk − v cos φk = 0, w cos φk + v sin φk = 0.
Since, edge rotation is permitted, the other two edge conditions are,
Mφ = 0 and Nxφ = 0. ...eqn. 16.28
3. Edge Rigidly Held (Fixed Edge)
This is not so common case. Since, it is fixed edge conditions, all displacements are zero at the edge.
Thus,
uφk = 0, vφk = 0, wφk = 0, θφk = 0. ...eqn. 16.29
4. Edge Provided with Edge Beam
This is a common case. For outer edge of a multiple shell or the edges of single barrel shell, the
following boundary conditions may be assumed:
(a) θ = 0, i.e. no rotation of shell edge owing to relatively great stiffness of edge beam.
(b) RH = 0, i.e. edge beam is incapable of withstanding lateral thrust. Since, Nφ and Qφ′ are the forces
giving horizontal component of forces,
RH = Nφ cosφk − Qφ′ ⋅ sinφk = 0
(c) The longitudinal displacements of shell and edge beam should be the same.
i.e. vS = vB.
(d) The vertical displacement of shell and edge beam should be the same.
i.e. w cosφk − v sinφk = wB ...eqn. 16.30
For inner edge of a multiple shell, the following boundary conditions may be assumed:
(a) v = 0
(b) Displacement in horizontal direction = 0
i.e. w sinφk + v cosφk = 0.
(c) uS = uB, and
(d) ws = w cos φk − v sin φk = wB. ...16.31
16.11 EDGE BEAM THEORY

The relationship between edge beam forces and displacements with respect to the shell edges forces
should be found so as to apply necessary boundary conditions. Edge beam theory deal with this aspect
of shell analysis.
Shell exerts following forces on the edge beam: Nφ, Nφx, Qφ, Mφ and Mxφ.
Equal and opposite stress resultants develop on edge beam. Let the connection be at the corner of the
edge beam. As a result of these forces from the shell edge beam is subjected to the following forces:
(a) F – longitudinal force,
(b) S – vertical shear and
(c) Mx – bending moment.
Consider the equilibrium of unit length of edge beam as shown in Fig. 16.9.
∑ Forces in x-direction = 0, gives,
∂F
-F + F + ¥ 1 - N fx = 0
∂x
Mx
Qf
F
Mf
Mfx Nf
Q
Nfx
Mf Mfx +
Mx
1
=
Nf dx
+
Nfx Qf F
(a) Shell edge forces + (b) Forces on edge beam

S
Fig. 16.9
∂F
i.e. = N fx
∂x
∴ F = ∫Nφx dx
= ∫[Nφx] sin kx dx
1
= - ÈÎ N fx ˘˚ cos kx ...16.32(a)
k
where [Nφx] sin kx = Nφx.
Similarly for all stress resultants, the magnitude excluding sin kx or cos kx associated with will be
marked as a bracked quantity and edge beam theory derived.
∑ Forces in vertical direction = 0, gives,
∂S
- Qf cos f k - N f sin f k + W ¢ = 0
∂x
where W ′ is weight of edge beam per unit length.
∴ (
S = Ú Qf cos f k + N f sin f k - W ¢ dx)
=
1
{ }
ÈQf ˘ cos f k + ÈÎ N f ˘˚ sin f k - [W ¢] sin kx
k Î ˚
∑ Moments in x-direction = 0, gives,

∂M
+ N xf a1 - M xf cos f k - S ¥ 1 = 0
∂x
{ }
M = - Ú ÈÎ N xf ˘˚ a1 - ÈÎ M xf ˘˚ cos f k - [ S ] sin kx
=
1
{ }
È N xf ˘˚ a1 - ÈÎ M xf ˘˚ cos f k - [ S ] cos kx
k Î
1Ï cos f k sin f k [W ¢] ¸
= ÌÈÎ N xf ˘˚ a1 - ÈÎ M xf ˘˚ cos f k - ÈÎQf ˘˚ - ÈÎ N f ˘˚ + ˝ cos kx
kÓ k k k ˛
Stress at junction,
∂u B F M
E = - a1
∂x A IB
1 ÊF M ˆ
E Ú ÁË A I B ˜¯
∴ uB = - a1 dx
1 È 1 a1 Ï Qf ÈÎ N f ˘˚ [W ¢] ¸˘˙
E Ú ÎÍ Ak Î
= Í - È N ˘ - Ì È N ˘ a - È M ˘ cos f - cos f - sin f + ˝ cos kx dx
x f ˚ I k ÔÓÎ x f ˚ 1 Î x f ˚ k
k k
k k
k Ô˛˚˙
B
ÈÈ ˘ Ï È Qf ˘ ÈNf ˘ [W ¢] ¸˘˙
Ì È N xf ˘ a1 - È M xf ˘ cos f k - Î ˚ cos f k - Î ˚ sin f k +
1 a
= ÍÍ - ÈN ˘ - 1 2
2 Î xf ˚ ˙
˝ sin kx
ÍÎ Î AEk ˚ IBk ÔÓ Î ˚ Î ˚ k k k Ô˛ ˙˚
Vertical deflection of edge beam is given by,
∂ 2 wB
EI B =M
∂x 2
1Ï È Qf ˘ ÈNf ˘ [W ¢] ¸
= ÌÈ N xf ˘ a1 + È M xf ˘ cos f k - Î ˚ cos f k - Î ˚ sin f k + ˝ cos kx.
k ÓÔÎ ˚ Î ˚ k k k ˛Ô
1 Ï Q ÈNf ˘ [W ¢] ¸
∴ wB = - ÌÈ N ˘ a + È M ˘ cos f k - f cos f k - Î ˚ sin f k + ˝ cos kx
3 ÔÎ xf ˚ 1 Î xf ˚
I Bk Ó k k k ˛Ô
16.12 SUMMARY OF CALCULATIONS

1. Select the overall dimensions of the shell using I.S. Recommendations. Calculate k = π/L.
2. Find membrane or particular solution.
3. Calculate:
18 18
Ê hR 6 k 4 ˆ Ê 12 R 6 k 4 ˆ
(a) r = Á =Á
Ë I ˜¯ Ë h 2 ¯˜
k 2 R2
(b) g = 2
r2
(c) Find
12
È
r (1 + g )2 + 1 + 1 + g ˘˙
a1 = 4 Í
2 ÎÍ 2 ˚˙
12
È
r Í (1 - g )2 + 1 - (1 - g ) ˘˙
a2 = 4
2 ÍÎ 2 ˙˚
12
È
r Í (1 + g )2 + 1 - (1 + g ) ˙˘
b1 = 4
2 ÍÎ 2 ˙˚
12
È
r (1 - g )2 + 1 - (1 - g ) ˘˙
and b2 = 4 Í
2 ÎÍ 2 ˚˙
4
2
(d) Find m1 = a1
r
4
2
m2 = a2
r
4
2
n1 = b1
r
4
2
n2 = b2
r
(e) Select the boundary conditions at the edge φ = φk. Each equation can be formed by collecting
coefficients of a, b, c and d and writing the right hand side of equation.
(f) Solve simultaneous equations to get arbitrary constants a, b, c and d.
(g) Using Table 16.2 and 16.3 find the stress resultants.
Since, the calculations are lengthy bending analysis is not meant for hand calculation. It is suggested
that going through above steps one should develop analysis package and use it.
16.13 STATICAL CHECKS

For confirming the validity of the results obtained, the following statical checks should be applied.
1. Sum of the horizontal forces in x-direction should be zero.
To apply this check, Nx should be found at mid point of various equal segments (say 5° or 10°).
4 5
6
3
7
2
y6
8
1 y2
df
Fig. 16.10 Shell arch divided into 8 equal segments
Let these values be N x1 , N x2 ,..., N x L as shown in Fig. 16.10. Apply Simpson’s rule to sum up
there forces. Thus
Rd f
Â Nx = È N x + 4 N x2 + 2 N x3 + 4 N x4 + ... N x L ˘˚
3 Î 1
where δφ is the angular length of the segment and NxL is the last ordinate. Then the statical check
Â Nx - F
is, if F is the sum of longitudinal forces in edge beams, then error = ¥ 100 .
ÂNx
This error should be within the limit.
2. Sum of vertical components of Nxφ along the rim of end frame (i.e. at x = L/2) should be equal
to half the total vertical load on the shell.
Now, vertical component of Nxφ
= Nxφ sin φk
For finding the total vertical component due to Nxφ at x = L/2, find the values at mid point of
various equal segments of length Rδφ. Then from Simpson’s rule
L
Â N xf sin f = R ¥ 2 f x ¥
¥ g.
2
One can make use of symmetry and consider only half the shell arch.
3. Check for Longitudinal Moments
About any horizontal axis y − y, in the cross section, sum of all moments must be equal to statical
moment. This is applied at mid span (x = 0).
Referring to Fig. 16.10, various internal moments are
(a) Moment Mx in the shell.
1
Â Mx = Rdf ÈÎ M x1 + 4 M x 2 + 2 M x3 + 4 M x4 + ... + M x L ˘˚
3
(b) Moment due to Nx forces
Rd f
= È y1 N x1 + 4 y2 N x2 + 2 y3 N x3 + 4 y4 N x4 + ... + y L N xL ˘˚
3 Î
where y1, y2 ... are the lever arms of Nx forces in the middle of segments.
= R (cos f - cos f k ) + h 2
(c) Edge Beam Moment = 2M1
(d) Statical moment is due to sinusoidal loading on the shell and that on the edge beam:
At mid span.
È L2 L2 ˘
= Í 2 ( R ¥ 2fk g ¢) + w ¢ 2 ˙
Îp p ˚
Sum of moments ( a ) , (b ) , (c ) - moment ( d )

The error is, . It should be within the acceptable limit.
moment ( d )
16.14 LONG AND SHORT SHELLS

In the previous discussion it has been said that beam theory and Schorer theory are applicable only for
long shells. D.K.J. theory is applicable for long as well as short cylindrical shells. However, the terms
long and short shells have no precise meaning. It is difficult to exactly demarcate between them. Several
criteria have been proposed from time to time by different authors. Some of them are presented in this
article.
1. Basis of Applicability of Beam Theory
Beam theory is the simple method for the analysis of cylindrical shell. It is hoped on the assumption that
along the depth of beam stress variation is linear. After studying the stress distribution across the depth
by more precise theories it has been suggested that beam theory is applicable if,
(i) L/R ≥ 5 for shells without edge beams
(ii) L/R ≥ 3 for shells with edge beams
Hence a shell is called long shell if L/R ≥ 5, if it is without edge beam and if L/R ≥ 3, if it is with edge
beam.
2. Basis of Ignoring Mx, Qx, Mxy
There are theories like Schorer theory in which the terms Mx, Qx and Mxy in a shell element are taken
as zero. Such theories are applicable only if, L/R ≥ π. Hence, on the basis of ignoring Mx, Mxy, Qx shells
may be called as long shell if L/R ≥ π.
3. Extent of Travel of Edge Disturbances
Disturbance emanating from the edges of shell may or may not penetrate beyond the crown. In long
shells they do not penetrate crown i.e. actual forces at crown are the membrane forces only. On this
basis ASCE classifies shell as long if L/R > 16.
4. Aas Jacobsen’s Classification
Aas Jacobsen was perhaps the first to classify shells into long, intermediate and short. According to
him,
the shell is long, if ρ = 4 to 7
intermediate, if ρ = 7 to 10
short, if ρ = 10 to 20.
QUESTIONS
1. Draw a typical cylindrical shell element and indicate various membrane forces, transverse shears and moments.
Give the relations among forces and moments on positive faces and negative faces.
2. Derive the six equations of equilibrium of a shell element subject to bending.
3. Enumerate the basic assumptions made in the analysis of cylindrical shells in the following theories:
(i) D. K. J. theory
(ii) Schorer’s theory.
Comment on the above two theories.
4. Discuss edge conditions to be used in the following cases of cylindrical shell analysis:
(i) Edge unsupported
(ii) Edge supported on unyielding wall
(iii) Fixed edge
(iv) Inner edge of multiple shell.
Chapter
17 Analysis of Symmetrically
Loaded Cylindrical
Water Tanks and Pipes
Circular pipes and water tanks are usually subjected to symmetrical load and they are commonly used
structure. In this chapter, the analysis of these structures is presented.
17.1 FORCES ON AN ELEMENT

Figure 17.1 shows a typical cylindrical pipe/water tank and Figure 17.2 shows an element in which
x – longitudinal direction
y or φ – tangential direction and
z – radial inward direction.
Fig. 17.1 A Cylindrical Pipe/Water tank

ANALYSIS OF SYMMETRICALLY LOADED CYLINDRICAL WATER TANKS AND PIPES 229
Qx
Nx
Qf
Nxf
Nf
Nfx
y x
z
+
Nf
N +
fx
+
+ + Nx
Qf Nxf
Qx+
df
R
Fig. 17.2 Typical element
Due to symmetry,
(i) Shearing forces and twisting moments are zero i.e.
Nxφ = Nφx = 0
Qφ = 0
Mxφ = Mφx = 0.
(ii) There is no variation of forces and moments with respect to φ, i.e.
Nφ – Constant
Mφ – Constant.
Thus, there is variation of forces with respect to x only i.e.
∂N x
N x+ = N x + dx
∂x
∂Q x
Qx+ = Qx + dx and
∂x
∂M x
M x+ = M x + dx.
∂x

Let X and Z be component of load in x and z direction respectively.
(i) ∑ Forces in x-direction = 0, gives
Ê ∂N x ˆ
ÁË N x + dx ˜ R d f - N x Rd f + XdxRd f = 0
¯
∂x
∂N x
i.e. + X = 0. ...eqn. 17.1
∂x
(ii) ∑ Forces in z-direction = 0, gives
Ê ∂Q x ˆ df df
ÁË Q x + dx ˜ Rd f - Q x Rd f + N f dx sin
¯
+ N f dx sin + Zdx Rd f = 0.
∂x 2 2
df df
Since, φ is a small angle sin = and hence, the equilibrium equation reduces to
2 2
∂Qx
dx Rd f + N f dx d f + Z R dxd f = 0
∂x
∂Q x N f
i.e. + +Z =0 ...eqn. 17.2
∂x R
(iii) ∑ Moments in x-direction = 0, gives
Ê ∂M x ˆ dx Ê ∂Q x ˆ dx
ÁË M x + dx ˜ Rd f - M x Rd f - Q x Rd f
¯
- Á Qx +
Ë
dx ˜ Rd f ◊
¯
=0
∂x 2 ∂x 2
After neglecting small quantity of higher order,
∂M x
Rdxd f - Qx Rd fdx = 0
∂x
∂M x
i.e. - Qx = 0 ...eqn. 17.3
∂x
17.3 STRESS RESULTANTS

First equation gives Nx value directly and Nx term is not appearing in any of the subsequent equations.
Hence, Nx can be determined independently. Due to other load components Nx is not at all affected.
Now consider the equations 2 and 3. There are three unknowns namely, Qx, Nφ and Mx. As the
number of unknowns are more than number of equations, compatibility of displacements should be
considered.
Let u, v and w be the displacement components in x, y and z-directions respectively. Due to symmetry
v is zero. Thus, only u and w exist.
Figure 17.3 shows the deformed position of the middle surface.
∂u
Now, ex = ...eqn. 17.4
∂x
Middle surface
Fig. 17.3 Deformed position of middle surface
Final length in f direction - Original length

ef =
Original length
( R - w ) d f - Rd f w
= =- ...eqn. 17.5
R df R
s x ms f
ex = - , and
E E
ms x s f
ef = - +
E E
s x m 2s x
∴ e x + me f = -
E E
(e x + me f )
E
∴ sx =
1 - m2
(e x + me f ) , where h is the thickness.

Eh
∴ N x = hs x =
1 - m2
(e f + me x )
Eh
Similarly, Nf =
1 - m2
(e x + me f )
Eh
Thus, Nx = ...eqn. 17.6(a)
1 - m2
(e f + me x )
Eh
and Nf = ...eqn. 17.6(b)
1 - m2
For the loading Y and Z only, Nx = 0.
∴ e x = -me f ...eqn. 17.7
Substituting eqn. 7 in eqn. 17.6(b), we get,
Nf =
Eh
1 - m2
(e f - m2ef )
= Ehe f
Ehw
∴ Nf = - ...eqn. 17.8
R
Due to symmetry, there is no change in curvature in the circumferential direction. The curvature in
∂2 w
x-direction is equal to - .
∂x 2
∂2 w
∴ Mx = -D and ...eqn. 17.9(a)
∂x 2
∂2 w
M f = - Dm
∂x 2
= mM x ...eqn. 17.9(b)
Eh 3
D=
where
(
12 1 - m 2 )
From equation of equilibrium 17.3,
∂M x ∂2 w
Qx = = -D 3
∂x ∂x
From equation 17.2,
∂4w Ehw 1
-D - ¥ + Z = 0.
∂x 4
R R
∂4w Eh Z
∴ - - w+ =0
∂x 4
R D 2
D
∂4 w Eh Z
i.e. + w=
∂x 4
R D 2
D
∂4 w Z
i.e. + 4b 4 w = ...eqn. 17.10
∂x 4
D
Eh
where 4b 4 =
R2 D
b4 =
Eh
=
Eh
=
(
3 1 - m2 )
2 3 2 2
4R D Eh R h
i.e. 4 R2
(
12 1 - m 2
)
Equation 17.10 is to be solved to get particular integral and complementary solution. Particular
integral depends upon the nature of loading and complementary solution depends upon the boundary
conditions. Thus, the total solution
w = W1 + W2
where w1 — particular integral
and w2 — complementary solution
Complementary solution is given by
∂ 4 w2
+ 4 b 4 w2 = 0
∂x 4
(∂4 + 4b 4 ) w2 = 0
∴ d 4 = -4b 4 = 4b 4 ÈÎ cos (2 np + p ) + i sin (2 np + p ) ˘˚
È 2 np + p 2 np + p ˘
∴ d = 2 b Í cos + i sin
Î 4 4 ˙˚
where n = 0, 1, 2 and 3.
È 1 1 ˘
i.e. d1 = 2 b Í +i ˙ = b [1 + i ]
Î 2 2˚
È 1 1 ˘
d2 = 2 b Í- +i ˙ = b [ -1 + i ]
Î 2 2˚
È 1 1 ˘
d3 = 2 b Í - -i ˙ = b [ -1 - i ]
Î 2 2˚
È 1 1 ˘
d4 = 2 b Í -i ˙ = b [1 - i ]
Î 2 2˚
( ) bx ( ) bx ( ) bx ( ) bx
∴ w2 = C1¢e 1+ i + C 2¢ e 1- i + C3 e -1-i + C 4 e -1+ i
= C1¢ebx [cos bx + i sin bx ] + C 2¢ e bx [cos bx - i sin bx ]
+ C3¢ e -bx [cos bx - i sin bx ] + C 4¢ e -bx [cos bx + i sin bx ]

= ebx (C1¢ + C2 ) cos bx + ebx (C1¢ - C2¢ ) i sin bx
+ e -bx (C3¢ + C 4¢ ) cos bx + e -bx (C 4 - C3 ) i sin bx

It is to be noted that C1′, C2′, C3′ and C4′ are complex numbers. Since, w is real, it follows that
C1′ + C2′, i(C1′ − C2′), C3′ + C4′ and i(C3′ − C4′) should be real. It means C1 and C2 are conjugate terms and
C3 and C4 are also conjugate. Hence, let
C1 = C1′ + C2′, C2 = i(C1′ − C2′), C3 = (C3′ + C4′)
and C4 = i(C3′ − C4′), where C1, C2, C3 and C4 are real terms.
∴ w2 = C1ebx cos bx + C 2 ebx sin bx + C3 e -bx cos bx + C 4 e -bx sin bx
= ebx (C1 cos bx + C2 sin bx ) + e -bx (C3 cos bx + C 4 sin bx ) ...eqn. 17.11
Example 17.1. Analyse a water tank of radius R and depth ‘d’, if its one edge is fixed at base slab and
top is free. Assume the thickness of wall uniform.
Solution. Figure 17.4 shows a typical water tank. Let the coordinates be selected as shown in the figure.
2R
Fig. 17.4 A Typical water tank
In this case at x, the load component is

Z = -g ( d - x )
where γ—unit weight of water.
d 4w g (d - x )
∴ 4
+ 4b 4 w = -
dx D
Particular Solution:
(d 4 + 4b 4 ) w1 = -g (dD- x )
Ê d4 ˆ g (d - x )
4b 4 Á1 + 4 ˜ w1 = -
Ë 4b ¯ D
-1
-g ( d - x ) Ê d4 ˆ
w1 = Á 1 + ˜
4b 4 D Ë 4b 4 ¯
g (d - x ) È 1 d 4 ˘
=- Í1 + - ... ˙
4b 4 D ÎÍ 2 4b 4 ˙˚
g (d - x )
=-
4b 4 D
g (d - x )
=-
Eh
D
R2 D
R2
∴ w1 = -g ( d - x ) ...eqn. 17.12
Eh
The same result is obtained if we consider membrane solution, as shown below:
Z = −γ(d − x)
Referring to Fig. 17.5
A¢ B¢
Deformed
surface
B
A
Original
surface
Fig. 17.5 Cross section of a water tank
R
Hoop stress = g ( d - x )
h
g (d - x ) R
∴ Circumferential strain =
Eh
A ¢B ¢ - AB (d - x ) R
i.e. =g
AB Eh
( R - w ) d f - Rd f (d - x ) R
i.e. =g
Rd f Eh
w g (d - x ) R
i.e. - =
R Eh
-g ( d - x ) R 2
or w=
Eh
Thus, the particular solution and the membrane solutions are the same. Thus, here also particular
solution can replace membrane solution. The complementary solution which depends on boundary
conditions may be looked as edge perturbations from boundaries.
Total solution,
w = w1 + w2
g (d - x ) R 2
= eβx(C1 cos βx + C2 sin βx) + e−βx(C3 cos βx + C4 sin βx) - ...eqn. 17.13
Eh
In any stable structure disturbances at one end must go on reducing at the other end. w can go on
reducing with increase in x if and only if C1 and C2 are zero.
Hence,
g (d - x ) R 2
w = e -bx (C3 cos bx + C 4 sin bx ) -
Eh
C3 and C4 are to be determined from the boundary conditions. In this case, the boundary conditions
available are,
wx =0 = 0 ...(1)
∂w
=0 ...(2)
∂x x =0
From B.C. (i), we get,
gdR 2
0 = C3 -
Eh
gdR 2
or C3 = .
Eh
From B.C. (ii), we get,
gR 2
-be -bx (C3 cos bx + C 4 sin bx ) + e -bx ( -C3b sin bx + C 4b cos bx ) + = 0.
Eh x =0
gh 2
i.e. -bC3 + C 4b + =0
Eh
gR 2 gdR 2 gR 2
or, C 4 = C3 - = -
Ehb Eh Ehb
gR 2 Ê 1ˆ
= ÁË d - b ˜¯
Eh
gR 2 È -bx Ï Ê 1ˆ ¸˘
∴ w=- Í d - x - e Ìd cos bx + Á d - ˜ sin bx ˝˙ ...eqn. 17.14
Eh ÎÍ ÔÓ Ë b¯ Ô˛˚˙
Ehw È Ï Ê 1ˆ ¸˘
∴ Nf = - = gR Í d - x - e -bx Ì d cos bx + Á d - ˜ sin bx ˝ ˙ ...eqn. 17.15
R ÎÍ ÔÓ Ë b¯ Ô˛ ˚˙
∂w gR 2
Now, = be -bx ( -C3 cos bx - C 4 sin bx - C3 sin bx + C 4 cos bx ) +
∂x Eh
∂2 w
∴ = b2 e -bx ÈÎ(C3 cos bx + C4 sin bx + C3 sin bx - C4 cos bx ) + C3 sin bx
∂x 2
-C 4 cos bx - C3 cos bx - C 4 sin bx ˘˚
= 2b2 e -bx (C3 sin bx - C4 cos bx )
gR 2 È Ê 1 ˆ˘
= 2b 2 e -bx Í d sin bx - Á d - cos bx ˜ ˙
Eh Î Ë b ¯˚
From eqn. 17.9(a),
∂2 w
Mx = -D
∂x 2
gR 2 È Ê 1ˆ ˘
∴ M x = -2b 2 De -bx Í d sin bx - Á d - ˜ cos bx ˙ ...eqn. 17.16
Eh Î Ë b¯ ˚
Convergence Study
Table 17.1 shows the values of βx, e−βx sinβx and e−βx cosβx. Figure 17.6 shows the variation of the
functions e−βx sinβx and e−βxcosβx with respect to e−βx. It may be observed that the two functions are
converging fast. Hence, deflection and all stress resultants are converging fast. For example, for a water
tank with R = 5 m, d = 4.5 m, h = 150 mm, if µ is taken zero,
b = 4 (
3 1 - m2 )= 3
= 5.333
5 (0.15)
2 2 2 2
R h
∴ β = 1.52
Table 17.1 Convergence study of the functions
βx e−βx sinβx e−βx cosβx
0 0 1.0
1 0.3096 0.1988
2 0.1231 −0.0563
3 0.0071 −0.0493
4 −0.0139 −0.0120
5 −0.0065 0.0019
6 −0.0007 0.0024
7 0.0006 0.0007
sin bx
0.30%
-bx
0 1 2 3 4 5 6 7
e
bx
-bx
(a) Convergence of the function e sin bx
1.0
cos bx
-bx
0 1 2 3 4 5 6 7
e
bx
-bx
(b) Convergence of the function e cos bx
Fig. 17.6 Convergence of the functions
At x = 0, at edge fixed with slab,

e−βx = 1, sinβx = 0, cosβx = 1
-2b 2 gR 2 È Ê 1ˆ˘
∴ M0 = D Í- Á d - ˜ ˙
Eh Î Ë b¯˚
Eh 3
( d - b)
1
= 2b 2 gR 2 ¥
(
12 1 - m 2
) Eh
h2
= 2 b 2 gR 2 (d - b)
12
Substituting β = 1.52, γ = 9.8 kN/m3 and the other values we get,

(0.15)2 Ê 1 ˆ
M 0 = 2 ¥ 1.52 2 ¥ 9.8 ¥ 5.0 2 ÁË 4.5 - ˜
12 1.52 ¯
= 8.156 kN-m
At the free edge x = 4.5 m, since
D Eh 3 1 h2
= ¥ =
Eh 12 (1 - 0 ) Eh 12
(0.15)2
∴ M 0 = 2 ¥ 1.52 2 ¥ 9.8 ¥ 52 ¥ ¥ e -1.52 ¥ 4.5
12
È Ê 1 ˆ ˘
Í 4.5sin 4.5 ¥ 1.52 - ÁË 4.5 - ˜¯ cos 4.5 ¥ 1.52 ˙
Î 1.52 ˚
= 0.002 kN-m ≈ 0.
∴ The boundary condition at top edge do not influence moment at fixed edge.
d
At x= = 1.5 m, \ bx = 1.5 ¥ 1.52 = 2.28
3
0.152 -2.28 È Ê 1 ˆ ˘
M x = 2 ¥ 1.52 2 ¥ 9.8 ¥ 52 ¥ e Í 4.5sin 2.28 - ÁË 4.5 - ˜¯ cos 2.28 ˙
12 Î 1.52 ˚
= 1.284 kN-m
It may be noted that in R.C.C. water tank design, it is assumed that in top 2/3 depth vertical rein-
forcement may be only nominal reinforcement. The assumption is justified. In the design it is also
assumed that the moment at base is due to the pressure shown in Fig. 17.7.
d/3
gd
Fig. 17.7 Approximate estimation of moment Mx
1 d Ê 1 d ˆ gd 3
i.e. M0 = ¥ gd ¥ ¥ Á ¥ ˜ =
2 3 Ë3 3¯ 54
Hence, in this case approximated moment is

9.8 ¥ 4.53
M0 = = 16.538 kN-m.
54
This is almost double of actual value of 8.156 kN-m. Thus, in the approximate method moment is
overestimated.
17.4 ANALYSIS OF LONG CIRCULAR PIPES

In the analysis of shells, particular solution can be replaced by membrane solution and complementary
solution can be considered as edge perturbations. Hence, the two solutions can be found separately and
superposed to get total solution. The perturbation at the edges is due to edge moment M0 and edge
transverse shear Q0. Due to antisymmetry M0 and Q0 are uniformly distributed along the edges. The
deflection function w due to M0 and Q0 may be found as explained below (Ref. Fig. 17.8).
Q0
M0
M0
Q0
Fig. 17.8 Pipe subjected to uniform M0 and Q0
Since, the forces applied at the end x = 0 produce a local bending which dies out rapidly with x from
the loaded end, we conclude that C1 and C2 in the general solution must vanish.
i.e. C1 = C2 = 0
∴ w = e−βx(C3 cos βx + C4 sinβx)
The boundary conditions to be satisfied are:
Ê ∂2 w ˆ
Mx = -D Á 2 ˜ = M0 ...(i)
x =0 Ë ∂x ¯ x = 0
∂M x Ê ∂3 w ˆ
and Qx = = -D Á 3 ˜ = Q0 ...(ii)
x =0 ∂x x =0 Ë ∂x ¯ x = 0
∂w
Now, = be -bx ( -C3 cos bx - C 4 sin bx - C3 sin bx + C 4 cos bx )
∂x
∂2 w
= b 2 e -bx (C3 cos bx + C 4 sin bx + C3 sin bx - C 4 cos bx
∂x 2
+ C3 sin bx - C 4 cos bx - C3 cos bx - C 4 sin bx )
= 2b 2 e -bx (C3 sin bx - C 4 cos bx )
∂3 w
∴ = 2b3 e -bx ( -C3 sin bx + C 4 cos bx + C3 cos bx + C 4 sin bx )
∂x 3
From B.C. (i), we get,
M0
- D 2b 2 ( -C 4 ) = M 0 or C4 =
2b2 D
From B.C. (ii), we get,
D 2b3 (C 4 + C3 ) = Q0
Q0
i.e. C3 + C 4 = -
2b 3 D
Q0 M0
i.e. C3 = - -
2b D3
2b 2 D
(Q0 + bM 0 )
1
or C3 = -
2b 3 D
e -bx
∴ w= ÈÎbM 0 (sin bx - cos bx ) - Q0 cos bx ˘˚ ...eqn. 17.17
2b 3 D
The maximum deflection occurs at x = 0.
1
wmax = w0 = - ÈÎbM 0 + Q0 ˘˚ ...eqn. 17.18
2b 3 D
The negative sign for this deflection is due to the fact that w is taken as positive towards the axis of
the cylinder. The slope at x = 0 is given by
∂w
= b ( -C 3 + C 4 )
∂x x =0
È M ˘
= b Í (Q0 + bM 0 ) 3 + 20 ˙
1
Î 2b D 2 b D ˚
1
= ÈÎQ0 + 2bM 0 ˘˚ ...eqn. 17.19
2b 2 D
17.5 ANALYSIS OF WATER TANKS AS COMBINATION OF MEMBRANE

SOLUTION AND EDGE PERTURBATIONS M0, Q0
The water tank may be analysed as the combination of the following two cases:
(i) Membrane analysis (Free edge at bottom also) and
(ii) Free edge subject to M0 and Q0 at bottom with the conditions that
w = 0 at x = 0 ¸
Ô
∂w
= 0 at x = 0 ˝Ô
for fixed base
and
∂x ˛
w = 0 at x = 0 ¸
and M x = 0 at x = 0 ˝˛
for hinged base.
Analysis of Water Tank with Fixed Base:

Membrane displacement is given by
-g ( d - x )
w1 =
4b 4 D
ÈÎbM 0 (sin bx - cos bx ) - Q0 cos bx ˘˚

1
and w2 = -
2b 3 D
(bM 0 + Q0 )
1
∴ w2 x =0
=-
2b 3 D
∂w2
(2bM 0 + Q0 )
1
∴ =
∂x x =0 2b 2 D
Now, boundary conditions are at x = 0,
∂w
w=0 ...(1) =0
and ...(2)
∂x
Noting that w = w1 + w2, from boundary condition (i), we get
gd
(bM 0 + Q0 ) = 0.
1
- -
4b D4
2b 3 D
-gd
or (bM 0 + Q0 ) = 2b
...(iii)
From boundary condition (ii), we get,

g
(2bM 0 + Q0 ) = 0
1
+
4b D 4
2b 2 D
g
i.e. 2bM 0 + Q0 = - ...(iv)
2b 2
From equation (iii) and (iv), we get,

gd g
bM 0 = - 2
2b 2b
g Ê 1ˆ
or M0 = 2 Á
d- ˜ ...(v)
2b Ë b¯
From equation (iii),
gd
Q0 = - - bM 0
2b
gd g Ê 1ˆ
=- - Á d- ˜
2b 2b Ë b¯
gÈ 1 ˘
= Í -d + ˙
bÎ 2b ˚
g (d - x ) e -bx È Ê gd g ˆ gÊ 1ˆ ˘
∴ w=- + 3 Í Á 2b
- 2 ˜ (sin bx - cos bx ) - Á - d + ˜ cos bx ˙
4b 4 D 2b D ÎÍ Ë 2b ¯ bË 2b ¯ ˚˙
g (d - x ) e -bx È gd gd g ˘
=- + 3 Í
sin bx + cos bx - 2 sin bx ˙
4b 4 D 2b D Î 2b 2b 2b ˚
g (d - x ) e -bx È g ˘
=- + Í gd sin bx + gd cos bx - b sin bx ˙
4b 4 D 2b D Î
4
˚
gÈ -bx ÏÊ 1ˆ ¸˘
=- Í d - x - e ÌÁ d - ˜ sin bx + d cos bx ˝˙ ...eqn. 17.20
4b D ÍÎ
4
ÔÓË b¯ Ô˛˙˚
This is same as eqn. 17.14 obtained from first principle.
(b) Tank with Hinged Base:
The total solution is,
g (d - x ) e -bx
w=- + ÈÎbM 0 (sin bx - cos bx ) - Q0 cos bx ˘˚
4b 4 D 2b 3 D
The boundary conditions are at x = 0,
w = 0 and M0 = 0.
gd 1
∴ we get 0=- + ÎÈ -Q0 ˚˘ = 0 .
4b D
4
2b 3 D
gd
∴ Q0 = -
2b
g (d - x ) e -bx
∴ w=- + ( -Q0 cos bx )
4b 4 D 2b 3 D
g (d - x ) e -bx Ê gd ˆ
=- - - cos bx
4b 4 D 2b 3 D ÁË 2b ˜¯
g (d - x ) gd
=- + e -bx cos bx
4b D 4
4b D
4
g ÈÎ d - x - de -bx cos bx ˘˚
=-
4b 4 D
gR 2 È
Î d - x - de cos bx ˘˚
-bx
=- ...eqn. 17.21
Eh
The same result may be obtained from the first principle also.
QUESTIONS
1. Derive the equations of equilibrium for a symmetrically loaded cylindrical pipes.

2. The equation of equilibrium for a symmetrically loaded pipe is
∂4 w Z
+ 4b 4 w =
∂x 4
D
Eh
where 4b 4 = .
R2 D
Determine the complementary function for it.
3. Show that the particular solution in the equilibrium equation of symmetrically loaded pipe may be replaced by
membrane solution.
4. A long cylindrical pipe is subjected to edge moment M0 and edge shear Q0 per unit circumferential length.
Determine the expression for displacement.
MEMBRANE THEORY FOR SHELLS OF REVOLUTIONS 245
Chapter
18
Membrane Theory for
Shells of Revolutions
Middle surface of a shell of revolution is obtained by rotating a plane curve around an axis, which is
denoted as shell axis. In this chapter, analysis of such shells by membrane theory is presented.
18.1 GEOMETRY OF SHELL OF REVOLUTION

Figure 18.1 shows a typical shell of revolution. By bisecting the middle surface of the shell with two
series of plane, one containing the axis and another perpendicular to the axis, we create a net of lines.
These network of curves may be called as meridians and parallels.
Shell axis
n
ia
id
er
M
q ABCD - shell element

dq
r0 90°
A
B D
r2
r1
f
C df
Parallel circle
Fig. 18.1
Let,
r1 — radius of curvature of the meridian.
r2 — length of the shell normal to meridian at point A upto the shell axis.
r0 — radius of curvature of parallel circle.
φ — the angle between the shell axis and shell normal.

θ — the angle in the plane of parallel circle between a reference direction and the radius joining.
It may be noted that the angle between the shell axis and the radius of the parallel circle r0 is 90°.
∴ r0 = r2 sin φ ...eqn. 18.1

Consider an element formed by two meridians and two parallels as shown in Fig. 18.2.
Nq Nqf Nf dq
Nfq
r0
90°
x
y
z
+ +
+
Nqf N q r2 f
Nfq
+
Nf
r1
f
df
Fig. 18.2
Let Nθ, Nφ, Nθφ and Nφθ be the forces per unit length. Then obviously,
∂N q
N q+ = N q + dq
∂q
∂N f
N f+ = N f + df ...eqn. 18.2
∂f
+
∂N qf
N qf = N qf + dq and
∂q
+
∂N fq
N fq = N fq + df
∂f
Let,
X — Component of load in θ-direction per unit area.
Y — Component of load in φ-direction per unit area.
Z — Component of load in direction normal to the element in the inward direction, per unit area.
Noting that in membrane theory bending moment and transverse shear are neglected, the element is
in equilibrium under the action of Nθ, Nφ, Nθφ and Nφθ.
Three equations of equilibrium may be derived for the element by considering the forces in three
mutually perpendicular directions θ, φ and Z.
1. Equilibrium of forces in θ direction
Component of various forces in θ-direction are:
(a) Nθ forces:
Ê ∂N q ˆ
- N q r1d f + Á N q + d q˜ r1d f
Ë ∂q ¯
∂N q
= r1 ◊ dq .
∂q
(b) Nφ forces – No component.
(c) Nφθ forces:
Ê ∂N fq ˆ Ê ∂r ˆ
- N fq r0 dq + Á N fq + d f˜ Á r0 + 0 d f˜ dq
Ë ∂f ¯Ë ∂f ¯
∂r0 ∂N fq
= N fq d fdq + r0 d fdq
∂f ∂f
∂
=
∂f
( )
r0 N fq d f dq
[Small quantity of higher order neglected].

(d) Nθφ forces:
f 2
Nqf r1 df cos dq/
2
f
/
r 1d
dq
N qf
sf
co
df
f Nqf r1 df cosf
1
qf r
N +
Resultant
(a) Vertical section (b) Plan view
Fig. 18.3

On face AB of the element, upward force
= Nθφ r1 dφ [Ref. Fig. 18.3(a)]
Its radial inward component
= Nθφ r1 dφ cosφ [Ref. Fig. 18.3(b)]
Similarly on face CD, downward force
Ê ∂N qf ˆ
= Á N qf + dq˜ r1d f
Ë ∂q ¯
Its radial inward component,
Ê ∂N qf ˆ
= Á N qf + dq˜ r1d f cos f [Ref. Fig. 18.3(b)]
Ë ∂q ¯
Hence, the component of these Nθφ forces in ‘θ’ direction
Ê ∂N qf ˆ dq dq
= Á N qf + dq˜ r1d f cos f ◊ + N qf r1d f cos f ◊
Ë ∂q ¯ 2 2
= N qf r1 cos f d f dq
[After neglecting small quantities of higher order].
(e) Load Component.
= X r1dφ r0 dθ
= X r0r1 dθdφ
∴ The equation of equilibrium is,
∂N q ∂
r1
∂q
d qd f +
∂f
( )
r0 N qf d qd f + N qfr1 cos fd qd f + Xr0 r1d qd f = 0
∂N q ∂
i.e.
∂q
r1 + N qf r1 cos f +
∂f
( )
N qf r0 + Xr1r0 = 0 ...eqn. 18.3
2. Equation of Equilibrium of forces in φ direction.

(a) Component of Nφ forces in φ direction:
Ê ∂N f ˆ Ê ∂r ˆ
- N f r0 dq + Á N f + d f˜ Á r0 + 0 d f˜ dq
Ë ∂f ¯Ë ∂f ¯
∂r0
= Nf d f dq + r0 N f d fd q
∂f
∂
=
∂f
( )
r0 N f d f dq
[Small quantity of higher order neglected].

(b) Component of Nθ forces:

Looking at Fig. 18.4(a), tangential outward forces in parallel plane are Nθr1dφ and Nθ+r1 dφ.
sf
co
df
dq
r1
q
N
Nq r1 dq df
dq/2
Nq r1 df
f dq/2
qd
q
r 1d
N
+
Nq r1 df
(a) Horizontal plane ( b) Vertical plane
Fig. 18.4
Their radial inward component

dq dq
+ N q+ r1d f
= N q r1d f
2 2
= Nθr1dφ dθ
[After neglecting small quantity of higher order]
∴ Its component in φ-direction = − Nθr1 cos φ dφ dθ (Ref. Fig. 18.4(b))
(c) Nφθ forces − no component.
(d) Component of Nθφ forces:
Ê ∂N qf ˆ
= - N qf r1d f + Á N qf + dq˜ r1d f
Ë ∂q ¯
∂N qf
= ◊ r1dq d f
∂q
(e) Component of load:
= Y r0 r1 d q d f
∴ Equation of equilibrium is
∂ ∂N qf
∂f
( )
N fr0 - N q r1 cos f +
∂q 1
r + Yr0 r1 = 0 ...eqn. 18.4
3. Equation of Equilibrium of forces in z-direction

(a) Component of Nθ forces:
From Fig. 18.5(a), it is clear that, there is a component Nθr1dφ dθ in the horizontal plane. From
Fig. 18.5(b), it is clear that, the component in z-direction
= Nθr1dφ dθ sinφ.
dq/2
Nq r1df f Nq r 1 df dq
dq/2
q
fd
r 1d
Nq
Nq r
1 df d
q si
nf
f
+
Nq r1 df
(a) (b)
Fig. 18.5
(b) Component of Nφ forces in z-direction.

Referring to Fig. 18.6, we find component of Nφ forces in z-direction
df Ê ∂N f ˆ Ê ∂r ˆ dq d f
= N f r0 dq ◊ + Á Nf + d f˜ Á r0 + 0 d f˜
2 Ë ∂f ¯Ë ∂f ¯ 2
Nf r0 dq
N
f r0 d
fd
q
f
+
Nf r0 dq
df/2
df/2
Fig. 18.6
= Nφr0dφ dθ
[Small quantities of higher order neglected]
(c) Nθφ and Nφθ have no components.
(d) Component of external load:
= Z r0r1dθdφ
∴ The equation of equilibrium is
Nθr1 sinφ + Nφr0 + Z r0r1 = 0
N q sin f N f
or + +Z =0
r0 r1
Since, r0 = r2 sinφ, we get,
Nf Nq
+ +Z =0 ...eqn. 18.5
r1 r2
Thus, the equations of equilibrium are:
∂N q ∂
r1
∂q
+
∂f
( )
r0 N fq + r1 N qf cos f + r0 r1 X = 0
∂ ∂N qf
∂f
( )
r0 N f - r1 N q cos f + r1
∂q
+ r0 r1 Y = 0 ...eqn. 18.6
Nf Nq
+ + Z = 0.
r1 r2
18.3 EQUATIONS OF EQUILIBRIUM FOR AXI-SYMMETRICALLY LOADED SHELLS

When in addition to the symmetry of forms, shells of revolution are subjected to axi-symmetrical loads,
then,
(i) The forces are independent of θ i.e. all terms involving differentiation w.r.t. θ will vanish. Thus,
Nθ+ = Nθ
Nθφ+ = Nθφ
(ii) Shearing forces cannot exist, because if they exist there will be variation of forces in φ direction
as well as in θ-direction. But due to symmetry Nθ cannot vary. Hence,
Nθφ = Nφθ = 0.
(iii) Due to symmetry load component X = 0. Hence, the forces on the element are as shown in
Fig. 18.7.
There are only two forces Nφ and Nθ. They can be found by writing equations of equilibrium in
φ-direction and Z-direction.
1. ∑ Forces in φ-direction = 0:
(a) Component of Nφ forces:
Ê ∂N f ˆ Ê ∂r ˆ
- N f r0 dq + Á N f + d f˜ dq Á r0 + 0 d f˜
Ë ∂f ¯ Ë ∂f ¯
Nq A
Nf
dq
r0
B
D
+
+ Nq r2
Nf C
dq
Fig. 18.7
∂N f ∂r0
= r0 d fdq + N f d f dq
∂f ∂f
∂
=
∂f
(
r0 N f d f dq )
(b) Component of Nθ forces: Referring to Fig. 18.4, component of Nθ forces
= - N q r1 cos f d f dq .
(c) Load component
= Yr0 r1d q d f .
∂
∂f
( )
r0 N f - r1 N q cos f + r0 r1Y = 0 ...eqn. 18.7
2. ∑ Forces in Z-direction = 0:
As derived in Art 18.2.3, this equation is
Nf Nq
+ +Z =0 ...eqn. 18.8
r1 r2
18.4 SOLUTION OF EQUATIONS OF EQUILIBRIUM

By solving equations 18.7 and 18.8, the membrane forces Nθ and Nφ can be found.
From eqn. 18.8,

r2
N q = - Zr2 - N f
r1
Substituting it is eqn. 18.7, we get,
∂
∂f
( )
r0 N f + Zr1r2 cos f + N f r2 cos f + r0 r1Y = 0
Multiplying each term by sinφ, we get,

∂
∂f
( )
r0 N f ¥ sin f + Zr1r2 cos f sin f + N f r2 cos f sin f + r0 r1Y sin f = 0
Noting that r0 = r2 sinφ, we get,

∂
∂f
( )
r0 N f ¥ sin f + N f r0 cos f = - r1r2 ( Z cos f + Y sin f) sin f
∂
i.e. ∂f
( )
r0 N f sin f = - r1r2 ( Z cos f + Y sin f) sin f
r0 N f sin f = - Ú r1r2 ( Z cos f + Y sin f) sin fd f + C
i.e. r2 N f sin 2 f = - Ú r1r2 ( Z cos f + Y sin f) sin fd f + C
1 È r1r2 ( Z cos f + Y sin f ) sin fd f + C ˘

i.e. Nf = -
r2 sin 2 f Î
Ú ˚
To get physical meaning for it, let us multiply numerator and denominator by 2π.
1 È 2 pr1r2 (Y sin f + Z cos f ) sin fd f + C ˘
Nf = -
2 pr2 sin 2 f Î
Ú ˚ ...eqn. 18.9
where C is new arbitrary constant.
Elemental strip at f
r0
f
90-f
Nf Nf
r2
Fig. 18.8
A simple physical meaning of the above equation is possible. Referring to Fig. 18.8, the term
2πr1r2 sinφ dφ stands for the surface area of an elemental strip of the shell. The term (Y sinφ + Z cosφ)
stands for the vertical component of the forces per unit area acting on this elemental strip. Hence, the
term 2πr1 r2 sinφ(Y sinφ + Z cosφ) stands for the vertical load acting on the strip. The integral ∫2πr1r2
sinφ (Y sinφ + Z cosφ)dφ represents the vertical load on the shell upto the level where the meridian angle
is φ.
2πr2 sin2φ Nφ represents the vertical component of Nφ acting around the circle of latitude φ. Thus, the
equation 18.9 is nothing but a mathematical statement of the vertical equilibrium of portion of shell
above the parallel circle at φ. Hence, it can be stated as
W
Nf = - ...eqn. 18.10
2 pr2 sin 2 f
where W is the total vertical load acting on the dome above the level denoted by φ.
The constant of integration C can be made use of to account for concentrated load, if any, applied
at the crown or as a ring load above this level. If no such concentrated load exists, C = 0.
Example 18.1. Determine the membrane forces in a hemispherical shell subjected to self weight only.
Solution. Let the self weight per unit area be ‘g’ and radius of shell be ‘a’. [Ref. Fig. 18.9(a)]
Area of shell above φ
f
= Ú a db 2 pr0
0
f
= Ú a ◊ 2 p ◊ a sin b ◊ db
0
-ga/2 -ga/2
r0
g
a
b
f
db
a -ga ga
(b) Variation of N f (c ) Variation of N q
(a)
Fig. 18.9
= 2 pa 2 [ - cos b]0
f
= 2 pa 2 (1 - cos f )
Total vertical downward load above the level of φ is

w = 2 pa 2 g [1 - cos f]
2 pa 2 g (1 - cos f)
Nf = -
2 pr0 sin f
2 pa 2 g (1 - cos f)
=- , since r0 = a sinφ
2 pa sin 2 f
ga (1 - cos f )
=-
1 - cos 2 f
ga
=- ...(Ans.)
1 + cos f
To find Nθ,
Z = g cosφ
∴ From the equation of equilibrium,
Nf Nq
+ + g cos f = 0, we get,
r1 r2
ga N
- + q + g cos f = 0 [Since r1 = r2 = a for a spherical shell]
a (1 + cos f) a
Ê 1 ˆ
∴ Nq = a ◊ g ◊ Á - cos f˜ ...(Ans.)
Ë 1 + cos f ¯
The variations of Nφ and Nθ forces for a hemispherical shell are as shown in Fig. 18.9(b) and 18.9(c).
Example 18.2. Determine membrane forces in a hemispherical shell due to concentrated load at crown
only.
Solution. Let P be the concentrated load at crown and ‘a’ be the radius of hemispherical shell.
∴ The vertical equilibrium of the shell above the parallel circle at φ gives,
P
Nf = - ...(Ans.)
2 pa sin 2 f
Then, to find Nθ, we know Z = 0 for this loading.
Nf Nq
∴ + + 0 = 0.
a a
P
i.e. Nq = - N f = ...(Ans.)
2 pa sin 2 f
Example 18.3. Determine the membrane forces in a hemispherical shell due to snow load only.
Solution. Let snow load be p0 per unit horizontal area of shell surface. Hence, the total load upto the
circle of latitude φ is,
W = p0 πr02
= p0 π a2 sin2φ
W W
∴ Nf = - =-
2 pr0 sin f 2 pa sin 2 f
p0 pa 2 sin 2 f
=-
2 pa sin 2 f
pa a
=- ...(Ans.)
2
Now, intensity of load on surface at φ
p0 ¥ 1 1
= = p0 cos f
1
cos f
co 1
sf
∴ Z = p0 cos f ◊ cos f
= p0 cos 2 f
∴ From the equation of equilibrium,
f
Nf N
+ q + Z = 0, we get,
r1 r2
Fig. 18.10
p a N
- 0 + q + p0 cos2 f = 0
2 a
a a
- P0 - P0
2 2
f
a a
- P0 + P0
2 2
a
(a) (b) Variation of N f (c ) Variation of N q
Fig. 18.11
Ê p a ˆ
N q = - Á 0 - p0 cos 2 f˜ ¥ a
Ë2¥a ¯
=
ap0
2
(
1 - 2 cos2 f )
p0 a
=+cos2 f ...(Ans.)
2
The variation of membrane forces are shown in Fig. 18.11.
18.5 MEMBRANE ANALYSIS OF CONICAL SHELLS

Referring to Fig. 18.12, it is clear that φ cannot be taken as a coordinate. Let the base angle be α. Taking
‘s’ which is the distance from the apex to the point under consideration, as a coordinate, we note,
φ=α
r0 = s cos α
a a
S
r0
a
r2
Fig. 18.12 Conical Shell
r1 = α, and
r2 = s cot α
W
∴ NS = -
2 pr0 sin f
W
=- ...eqn. 18.11
2 ps cos a sin a
NS Nq
and + + Z = 0.
a s cot a
∴ N q = - Zs cot a ...eqn. 18.12
Example 18.4. Find the membrane forces in the umbrella type conical shell shown in Fig. 18.13.
r0
a g cos f = g cos a
f
r2
Fig. 18.13
Solution. Load below a horizontal section at a distance ‘s’ from apex

L
W = Ú 2 ps ◊ cos a ds g
S
L
È s2 ˘
= 2 pg Í ˙ cos a
Î 2 ˚S
= pg ( L2 - S 2 ) cos a
W
∴ NS = +
2 ps cos a sin a
pg ( L2 - S 2 ) cos a
=+
2 pS cos a sin a
g ( L2 - S 2 )
=+ ...(Ans.)
2 S sin a
Now Z = g cos α. (Ref. Fig. 18.13)
∴ Nθ = −Zs cot α
= −g cos α ⋅ s ⋅ cot α
= −gs cos α ⋅ cot α ...(Ans.)
18.6 ROTATIONAL HYPERBOLOID OF ONE SHEET

This type of shells are commonly used as cooling towers in thermal stations. If a hyperbolic curve is
rotated about a vertical axis, passing through the pole, such shells are generated.
Figure 18.14 shows a typical rotational hyperboloid of one sheet.
f0
a r0 axis
O
f
r1
r0
f r2
z-axis
Fig. 18.14
with r0 and z-axis selected as shown in Fig. 18.14, the equation of hyperbola is
r02 z2
- =1
a2 b2
z2
i.e. r0 = a 1 + ...eqn. 18.13
b2
The principal radii of curvatures r1 and r2 are given by,
32
È1 + ( f ¢ ( z )) 2 ˘
r1 = Î ˚ ...eqn. 18.14
f ¢ (z)
and r2 = f(z)[1 + (f′(z))2]1/2 ...eqn. 18.15
where f (z) = r0
z2
Now, f ( z ) = r0 = a 1 +
b2
z
2
∂r 1 b2 a z
∴ f ¢ (z ) = 0 = - a =-
∂z 2 z 2 b 2
z2
1+ 1+
b2 b2
a2 z
=-
b 2 r0
È -1 2 z ˘
a Í 1 2 b2 ˙
f ¢¢ ( z ) = - 2 Í +z 32
˙
b Í z2 Ê z 2ˆ ˙
Í 1+ 2 Á 1+ 2 ˜ ˙
Î b Ë b ¯ ˚
È z2 z2 ˘
Í 1+ 2 - 2 ˙
a
=- 2 Í b b ˙
b ÍÊ 2ˆ
32 ˙
Í Á1 + z ˜ ˙
ÎÍ Ë b2 ¯ ˚˙
a 1
=- 2 32
b Ê r02 ˆ
ÁË 2 ˜¯
a
a4
=-
b 2 r03
32
È1 + ( f ¢ )2 ˘
∴ r1 = Î 2 ˚
f (z)
¢¢
32
È a4z4 ˘
Í1 + 4 3 ˙
= ÎÍ b r0 ˚˙
(
- a 4 b 2 r03 )
32
b2 È a4z4 ˘
=- r3
4 0 Í1 + 4 3 ˙
a ÎÍ b r0 ˚˙
32
È r 2 z4 ˘
b2
= - 4 ¥ a Í 04 + 4 ˙ 6
a Îa b ˚
32
È r2 z4 ˘
= - a b Í 04 + 4 ˙
2 2
...eqn. 18.16
Îa b ˚
12
rz = f ( z ) ÈÎ1 + ( f ¢ ( z )) ˘˚
2
12
È a4z2 ˘
= r0 Í1 + 4 2 ˙
ÍÎ b r0 ˙˚
12
È r 2 z2 ˘
= a Í 04 + 4 ˙
2
...eqn. 18.17
Îa b ˚
b2
r1 = -
r23 ...eqn. 18.18
a4
The equation of hyperbola (Eqn. 18.13) may be rewritten as
b 2
z=± r - a2
a 0
∂z
( )
b1 2 -1 2
∴ = tan f = r - a2 ¥ 2 r0
∂r0 a2 0
b r0
=
a (r 2 - a 2 )1 2
∴ cot f = ±
a r0 - a
2
(
2
)1 2
b r0
cot 2 f =
(
a 2 r0 - a
2 2
) = a 2 Ê1 - a 2 ˆ
∴ Á ˜
b2 r02 b2 Ë r02 ¯
b2 a2
cot 2 f = 1 -
a2 r02
a2 b2 a 2 sin 2 f - b 2 cos 2 f
∴ =1- cot 2 f =
r02 a2 a 2 sin 2 f
a 4 sin 2 f
r02 =
a 2 sin 2 f - b 2 cos2 f
a 2 sin f
r0 =
(a 2 sin 2 f - b2 cos2 f)1 2
or ...eqn. 18.19
But r0 = r2 sin f
r0 a2
r2 = =
( )1 2
Hence, sin f ...eqn. 18.20
a 2 sin 2 f - b 2 cos 2 f
b2 a2b2
and r1 = - r23 = - ...eqn. 18.21
a4 (a 2 sin 2 f - b2 cos2 f)3 2
After noting the above geometric relations of hyperbola, it is possible to analyse the rotational
hyperboloid of one sheet for any type of loading.
Example 18.5. Analyse a typical rotational hyperboloid of one sheet subjected to self weight g/unit
surface area and find the membrane forces.
Solution. Let W be the total load of the shell above level φ. Then,
f0
W = g Ú 2 pr0 r1d f
f
f0
a 2 sin f (- b2 a 2 )
= 2 pg Ú df
f (a 2 sin 2 f - b2 cos2 f)1 2 (a 2 sin 2 f - b2 cos2 f)3 2
f0
sin f
= -2 pga 4 b 2 Ú df
f (a 2
sin f - b cos f
2 2 2
) 2
a
Let us substitute, cos f = x.
a + b2
2
a
then - sin f d f = d x.
a2 + b2
f0
a dx
∴ W = 2 pga b ÚÈ
4 2
2
a +b
2 2
f Ê a2 ˆ a2 ˘
Í a Á1 - 2 x - x ˙
2 2 2 2
˜ b
Î Ë a + b2 ¯ a2 + b2 ˚
f0
a 5b 2 1 dx
= 2 pg ¥ 4 ÚÊ 2
a +b a + b2 2 ˆ
2 2 a 2
f
ÁË 1 - x ˜
a2 + b2 ¯
f0
2 pg ab 2 dx
= Ú
a +b 2 2
f (1 - x 2 )2
x
2 pg ab 21 È 2x 1+ x˘
= ¥ Í + log
a +b
2 2 4 Î1 - x 2 1 - x ˙˚
x0
pg ab 2
= ÈÎ f ( x ) - f (x 0 ) ˘˚
2 ¥ a2 + b2
2x 1+ x
where f (x ) = + log
1- x 2 1- x
W
∴ Nf = -
2 pr0 sin f
ab 2
ÈÎ f ( x ) - f (x 0 )˘˚
g 1
=-
a 2 + b 2 r0 sin f
4
Now, r0 sinφ = r2 sin2 φ
a 2 sin 2 f
=
(a 2 sin 2 f - b2 cos2 f)1 2
È a2 ˘
a 2 Í1 - 2 x2 ˙
= Î a +b 2
˚
12
È 2Ê a2 ˆ a2 ˘
Í a Á1 - 2 x 2
˜ - b 2
x2 ˙
Î Ë a +b ¯ a +b
2 2 2
˚
a 2 ÈÎ a 2 + b 2 - a 2 x 2 ˘˚ (a 2 + b 2 )
= 12
È a 2 + b2 2 ˘
a Í1 - 2 x ˙
Î a + b2 ˚
=
(
a a2 + b2 - a2x2 )
(a 2 + b 2 ) (1 - x 2 ) 12
ab 2 (a 2 + b2 ) (1 - x 2 )1 2
È f (x ) - f (x 0 )˘˚
g
∴ Nf = -
4 a 2 + b2 a a + b - a x
2 2
(
2 2
)Î
1- x 2
ÈÎ f ( x ) - f (x 0 ) ˘˚
g
= - b 2 a 2 + b2
4 a + b2 - a2x2
2
( ) ...eqn. 18.22
To find Nθ;
Nf Nq
+ = - g cos f
r1 r2
r2
N q = - gr2 cos f - Nf
r1
a2
r2 =
(a 2 sin 2 f - b2 cos2 f)1 2
a2 a2
= =
(a (1 - cos f) - b ) È a 2 - ( a 2 + b 2 ) cos 2 f ˘
12 12
2 2 2
cos 2 f Î ˚
but (a 2 + b 2 ) cos2 f = a 2 x 2
a2 a
∴ r2 = =
(a 2
-a x 2 2
) 12
(1 - x2 )1 2
b2
r1 = - r23
a4
b2 a3
=- ¥
a4 (1 - x 2 )3 2
-b2
=
(
a 1 - x2 )3 2
r2
∴ N q = - gr2 cos f - Nf
r1
a a a a
(1 - x2 )
32
= -g ◊ x+ ¥ Nf
(1 - x ) 2
12
a2 + b2 (1 - x )
2
12
b 2
x
=-
ga 2
+
a2
(1 - x2 ) N f
a +b2 2
(1 - x )
2
12
b2 ...eqn. 18.23
Equations 18.22 and 18.23 give the membrane forces.

dx 1 È 2x 1+ x˘
To prove, I = Ú = + log :
(1 - x )2
2
4 ÍÎ 1 - x 2 1 - x ˙˚
Let, ξ = sinθ
dξ = cosθ dθ
cos q dq 1
I =Ú =Ú dq = Ú sec 3 q dq
cos q
4
cos3 q
= Ú sec 2 q ◊ sec q ◊ dq
d
=Ú (tan q) ◊ sec q ◊ dq
dq
= secθ tanθ − Ú tanθ ⋅ secθ tanθ dθ
= secθ tanθ − Ú secθ (sec2θ − 1) dθ
= secθ tanθ − Ú sec3θdθ + Ú secθ dθ
i.e. 2I = secθ tanθ + Ú secθ dθ
1
∴ I= ÎÈ sec q tan q + log (sec q + tan q ) ˚˘
2
1 È sin q Ê 1 sin q ˆ ˘
= Í 2 + log ÁË + ˜˙
2 Î cos q cos q cos q ¯ ˚
1È x 1+x ˘
= Í + log ˙
2 Í1 - x 2
1 - x 2 ˚˙
Î
1È x 1+ x ˘
= Í + log ˙
2 Î1 - x 2
1- x ¥ 1+ x ˚
1È x 1+ x ˘
= Í + log ˙
2 ÎÍ 1 - x 2
1 - x ˙˚
1È x 1 1+ x˘
= Í + log
2 Î1 - x 2
2 1 - x ˚˙
1 È 2x 1+ x˘
= Í + log .
4 Î1 - x 2 1 - x ˚˙
QUESTIONS
1. Derive the equations of equilibrium for a shell of revolution, neglecting bending of the element.
2. A shell of revolution is subjected to symmetrical load. Considering membrane theory, determine the equations
of equilibrium. Bring out physical meaning for the expression Nφ.
3. Determine the membrane forces in a hemispherical dome subject to
(i) Self weight ‘g’ per unit area
(ii) Snow load ‘p’ per unit projected area.
4. Determine membrane forces in an inverted umbrella type conical shell supported centrally and subjected to self
weight only. Assume uniform thickness.
Chapter
19
Membrane Theory for
Shells of Translation
Elliptic paraboloid, hyperbolic paraboloid and conoids are doubly curved shells generated by transla-
tion of one curve over the other. In this chapter, membrane theory of such shells is dealt.
19.1 ASSUMPTIONS
In developing the membrane theory, the following assumptions are made:
1. The thickness of the shell is small compared to other two dimension. In other words, the shell is
treated as thin i.e. the stresses normal to shell surface are ignored.
2. The deformation is small compared to the thickness of the shell i.e. stresses in the middle surface
are assumed zero.
3. Points on lines normal to the middle surface before deformation remain on the same normal even
after deformation. In other words, shear deformations are ignored.
4. The material of the shell is homogeneous, isotropic and linearly elastic.
5. The thickness of the shell is uniform.
19.2 COORDINATE AXES

The curvilinear coordinates may be convenient to generate the geometry of the shell. But use of the
consequent equations to satisfy the boundary conditions at the rectangular boundaries encountered in
practice is very difficult. Hence, the cartesian coordinates are used. Right hand system of cartesian
coordinates are used. Figure 19.1 shows a typical element in space and its projected element in x-y
plane.
19.3 MONGE’S NOTATIONS

The surface is represented by z = f (x, y). The following notations are used to write the equations
conveniently. These notations are known as Monge’s notations:
∂z ∂z
= p, =q
∂x ∂y
∂2 z ∂2 z ∂2 z
= r, = s, 2 = t ...eqn. 19.1
∂x 2 ∂x ∂y ∂y
A tx
f
q x
y B
ty
y D
C
x
A¢ dx B¢
dy
D¢ C¢
y
z
Fig. 19.1 Typical Element and Coordinate direction
19.4 PROPERTIES OF THE ELEMENT

Let at A,
tx – Tangent to the surface in x-direction
ty – Tangent to the surface in y-direction
φ – Angle between tx and x-direction
ψ – Angle between ty and y-direction
θ – Interior surface angle
dsx – Surface length of element in x-direction
dsy – Surface length of element in y-direction
dx – Projection of dsx on x-y plane
dy – Projection of dsy on x-y plane.
Referring to Figure 19.2,
ds x
dz
dx
Fig. 19.2 Side AB of element in x-z plane
∴ ds 2x = dx 2 + dz 2
ds x = dx 2 + dz 2
2
Ê dz ˆ
= dx 1 + Á ˜
Ë dx ¯
= dx 1 + p 2 ...eqn. 19.2
MEMBRANE THEORY FOR SHELLS OF TRANSLATION 269
Similarly,
ds y = dy 2 + dz 2
= dy 1 + q 2 ...eqn. 19.3
dx 1
cos f = = ...eqn. 19.4
ds x 1 + p2
dy 1
cos y = = ...eqn. 19.5
ds y 1 + q2
cos q = sin f ◊ sin y
= 1 - cos2 f 1 - cos 2 y
1 1
= 1- 1-
1+ p 2
1 + q2
p2 q2
=
1 + p2 1 + q2
pq
= ...eqn. 19.6
1+ p 2
1 + q2
Surface area of the element
dA = dsx dsy sinθ
= dx 1 + p 2 dy 1 + q 2 1 - cos 2 q
p2q2
= dx dy 1 + p 2 1 + q 2 1 -
(1 + p2 )(1 + q 2 )
= dx dy 1 + p 2 1 + q 2
(1 + p2 )(1 + q 2 ) - p2 q 2
( 1 + p2 )( 1 + q2 )
= dx dy 1 + p 2 + q 2 ...eqn. 19.7
19.5 MEMBRANE ANALYSIS

Let the membrane forces in the element be Nx, Ny and Nxy.
In the analysis, it is convenient to use the components of membrane forces parallel to the axes x and
y, imagined to act on the projected surface in x-y plane. These forces are called pseudo forces and they
are denoted by ‘n’ with suitable suffixes. Thus, the pseudo stress resultant nx is such that it exert the
same force in the x-direction on the projected side A′D′ as the membrane force does on side AD. Figure
19.3 shows the membrane forces and pseudo forces.
Ny
Nyx
Nxy A
Nx
B
+
Nx
D
+
Nxy
+ C
Nyx
+
Ny
0
x
ny
nyx
nxy A¢ B¢
+
nx nx
+
D¢ + C¢
nxy
nyx
+
ny
Fig. 19.3 Membrane and Pseudo forces
Relationship between Pseudo and Membrane Forces:

1. Nx and nx forces:
Normal force on AD = Nx dsy
Its component in x-direction = Nx dsy cosφ
From the definition of pseudo force,
n x dy = N x ds y cos f
ds y
or nx = N x cos f
dy
1
= Nx ◊ cos f
cos y
1
= N x 1 + q2 ¥
1 + p2
1 + q2
i.e. nx = Nx ...eqn. 19.8
1 + p2
1 + p2
Similarly, ny = Ny ...eqn. 19.9
1 + q2
2. nxy and Nxy forces:
Shear force on AD = Nxy dsy
Its component in y-direction = Nxy dsy cos ψ. According to the definition of the pseudo forces, this
must be equal to nxy dy.
∴ nxy dy = Nxy dsy cosψ
ds y
i.e. n xy = N xy cos y
dy
1
= N xy ◊ cos y
cos y
i.e. nxy = Nxy ...eqn. 19.10
Similarly, nyx = Nyx ...eqn. 19.11
But nxy = nyx
∴ Nxy = Nyx ...eqn. 19.12
Pseudo Loads:
The pseudo loads X, Y, Z in the directions x, y and z are so defined that
Real Load × Surface area of the element
= Pseudo load × Projected area of the element.
Hence, if Fx, Fy and Fz are intensity of load components on the element in x, y and z directions,
then, according to the definition of pseudo load,
Fx 1 + p 2 + q 2 dx dy = X dx dy
Thus, X = Fx 1 + p 2 + q 2
Similarly, Y = Fy 1 + p 2 + q 2 ...eqn. 19.13
Z = Fz 1 + p 2 + q 2
Equations of Equilibrium
Referring to the projected element A′B′C′D′,
∑ Forces in x-direction = 0, gives
Ê ∂n x ˆ Ê ∂N yx ˆ
ÁË - n x + n x dx ˜ dy + Á - n yx + n yx +
¯
dy ˜ dx + X dx dy = 0
∂x Ë ∂y ¯
∂n x ∂n xy
i.e. + +X =0 ...eqn. 19.14
∂x ∂y
∑ Forces in y-direction = 0, gives
∂n y ∂N xy
+ + y = 0. ...eqn. 19.15
∂y ∂x
∑ Forces in z-direction = 0:
To assemble this equation, the components of various forces in z-direction are to be found.
Vertical component of N
= Nx dsy sin φ, But from definition of pseudo forces Nx dsy cos φ = nx dy
dy
= nx sin f
cos f
= nx tanφ dy
Similarly, vertical component of Ny = ny tanψ dx
vertical component of Nxy = nxy tanψ dy
and vertical component of Nyx = nyx tanφ dx
Net vertical force of Nx - forces
= - n x tan f dy + ( n x tan f ) dy
+
∂
= - n x tan f dy + n x tan f dy + (n tan f) dx dy
∂x x
∂
= (n tan f) dxdy
∂x x
Similarly,
∂
net vertical force due to Ny forces =
∂y
( )
n y tan y dx dy
∂
net vertical force due to Nxy forces =
∂x
( )
n xy tan y dx dy
∂
net vertical force due to Nyx forces =
∂y
( )
n yx tan f dx dy

∂ ∂ ∂
∂x
( n x tan f) +
∂y
(
n y tan y +
∂x
) (
n xy tan y )
∂
+
∂y
( )
n yx tan f + Z = 0
∂z ∂z
But tan f = and tan y = .
∂x ∂y
∴ Equation of equilibrium is,

∂ Ê ∂z ˆ ∂ Ê ∂z ˆ ∂ Ê ∂z ˆ ∂ Ê ∂z ˆ
ÁË n x ˜¯ + Á n y ˜ + Á n xy ˜ + ÁË n yx ˜¯ + Z = 0
∂x ∂x ∂y Ë ∂y ¯ ∂x Ë ∂y ¯ ∂y ∂x
∂n x ∂ z ∂ 2 z ∂n y ∂z ∂ 2 z ∂n xy ∂z
i.e. ◊ + nx 2 + + ny 2 + ◊
∂x ∂x ∂x ∂y ∂y ∂y ∂ x ∂y
∂ 2 z ∂n yx ∂z ∂2 z
+ n xy + ◊ + n yx +Z =0
∂x∂y ∂y ∂x ∂ x ∂y
Since, nxy = nyx, we get,
∂2 z ∂2 z ∂2 z
nx + 2 n xy + ny 2
∂x 2 ∂ x ∂y ∂y
∂z Ê ∂n x ∂n xy ˆ ∂z Ê ∂n y ∂n xy ˆ
+ + + + +Z =0
∂x ÁË ∂x ∂y ˜¯ ∂y ÁË ∂y ∂x ˜¯
But from equation 1,
∂n x ∂n xy
+ = -X
∂x ∂y
and from equation 2,
∂n y ∂n xy
+ = -Y
∂y ∂x
∴ The equation of equilibrium is
∴ rn x + 2sn xy + tn y = pX + qY - Z ...eqn. 19.16
19.6 PUCHERS STRESS FUNCTION

In 1934, Pucher introduced a stress function φ so as to reduce the three equations of equilibrium in three
unknowns, namely nx, ny and nxy, to one equation of equilibrium in only one unknown φ. The stress
function φ is so defined that,
∂2f
n xy = -
∂x∂y
Then from Eqn. 19.14,
∂n x ∂3 f
= -X
∂x ∂x ∂ y 2
∂2 f
or nx = - Ú X dx
∂y 2
From equation 19.15,

∂n y ∂3f
= -Y
∂y ∂ x 2 ∂y
∂2 f
∴ ny = - Ú Y dy
∂x 2
Substituting these valuation in Eqn. 19.16, we get,
Ê ∂2 f ˆ Ê ∂2f ˆ Ê ∂2f ˆ
r Á 2 - Ú X dx ˜ + 2 S Á - ˜ + t Á - Ú Ydy˜
Ë ∂y ¯ Ë ∂x ∂y ¯ Ë ∂x 2 ¯
= pX + qY - Z
∂2 f ∂2f ∂2f
or r - 2S + t 2 = pX + qY - Z + r Ú Xdx + t Ú Ydy ...eqn. 19.17
∂y 2 ∂x∂y ∂x
19.7 SYNCLASTIC, DEVELOPABLE AND ANTICLASTIC SHELLS

1 1
If r, s and t are curvatures, the principal curvatures and are given by
R1 R2
2
1 r +t Êr -tˆ
= + Á + s2
R1 2 Ë 2 ˜¯
2
1 r +t Êr -tˆ
= - Á + s2
R2 2 Ë 2 ˜¯
Ï ¸Ï ¸Ô
1 Ôr + t ÔÔ r + t
2 2
1 Êr -tˆ Êr -tˆ
∴ ¥ =Ì + Á + 2
˝Ì - + 2
˝
Ë 2 ˜¯ ÁË ˜
s s
R1 R2 ÔÓ 2 ÔÔ 2
˛Ó 2 ¯ ˛Ô
Êr +tˆ
2
ÔÏÊ r - t ˆ 2 Ô¸
=Á ˜ - ÌÁ ˜ + s2 ˝
Ë 2 ¯ ÔÓË 2 ¯ Ô˛
= rt - s 2
A shell is synclastic, developable or anticlastic according as
= rt - s 2 £
1
R1 R2 > 0. ...eqn. 19.18
19.8 MEMBRANE THEORY OF SYNCLASTIC SHELLS

Among synclastic shells, rotational paraboloid and the elliptic paraboloid are the two surfaces most
frequently favoured roofs to cover very large column free rectangular or square areas.
When a convex parabola moves over another convex parabola or when a concave parabola moves
over another concave parabola elliptic paraboloid surface is generated. If both parabolas are identical
the surface generated is known as rotational paraboloid. The vertical sections of an elliptic paraboloid
are parabolas while the horizontal sections are ellipses. Hence, the surface has the name elliptic parabo-
loid. The vertical sections of rotational paraboloid are parabolas while the horizontal sections are
circles.
A shell roof in the form of an elliptic paraboloid or rotational paraboloid over a rectangular or square
ground plan area is usually supported by shear diaphragm on all the four edges (Refer Figure 19.4). The
diaphragm are assumed to be stiff enough in their own planes to receive the shell but they cannot carry
any load applied normal to their plane. In other words, the diaphragm provide simple support to the
shell.
0 x
y 2b
z
fy
fx
a a
Fig. 19.4 Elliptic Paraboloid
The surface of the shell is mathematically represented by

Z = f1(x) + f2(y)
where f1(x) is the equation of the parabola in x-direction and f2(y) is the equation of the parabola in
y-direction.
If the origin is selected at the crown of the shell (Refer Figure 19.4),
fx fy
f1 ( x ) = 2
x 2 and f2 ( x ) = 2
y2 .
a b
∴ The equation of surface of shell is
fx fy
z= 2
x2 + y2 ...eqn. 19.19
a b2
∂z 2 f x ¸
∴ p= = x ...(a) Ô
∂x a 2
Ô
∂z 2 f y Ô
q= = y ...(b) Ô
∂y b 2 Ô
Ô
∂2 z Ô
...(c) Ô˝
2 fx
r= =
∂x 2
a2 Ô ...eqn. 19.20
Ô
∂2 z Ô
s= =0 ...(d) Ô
∂x∂y
Ô
Ô
∂2 z 2 fy Ô
and t= = ...(e) Ô˛
∂y 2
b2
Membrane Analysis for Snow Load

Puchers equation for the shell is
∂2 f ∂2f ∂2 f
r - 2s + t 2 = pX + qY - Z + Ú Xdx + Ú Ydy
∂y 2 ∂x∂y ∂x
For snow load X = Y = 0 and Z = p0
Hence, Pucher’s equation reduces to
2 fx ∂2f 2 f y ∂2 f
+ = - p0
a 2 ∂y 2 b 2 ∂x 2
Particular integral may be assumed as,
a2
f1 = - p y2 ...eqn. 19.21
4 fx 0
Homogeneous solution may be written as
2 fx ∂2f 2 f y ∂2 f
+ =0
a 2
∂y 2
b 2 ∂x 2
Seeking the solution in the form
φ2 = X Y, where X is the function of x only and Y is the function of y only,
2 fx 2 fy
we get, Y ¢¢X + YX ¢¢ = 0.
a2 b2
f y X ¢¢ f Y ¢¢
i.e. 2 X
= - x2
b a Y
f y a 2 X ¢¢ Y ¢¢
i.e. 2
=-
fx b X Y
The above relation holds good if and only if the right hand side and left hand side expressions are
equal to constant. Let the constant be λ2. Thus,
f y a 2 X ¢¢ Y ¢¢
2
=- = l2 .
fx b X Y
It leads to two ordinary differential equations,
Y″ + λ2 Y = 0 ...(a)
f x b2 2
and X ¢¢ - l =0 ...(b)
f y a2
From equation (a),
Y = cos λy
and from equation (b),
Ê fx b ˆ
X = cosh Á lx ˜
Ë fy a ¯
= cosh βx
fx b
where b= ◊ ◊l
fy a
∂2 f
Keeping in mind that N y = is to be zero at y = ±b, we try a solution in the form
∂x 2
•
f2 = - Â Am cosh b n x ◊ cos l n y. ...eqn. 19.22
n =1,3,...
np f x b np f x np
where ln = and b n = ◊ = .
2b f y a 2b f y 2a
∴ The total solution is
a2
f = - Â Am cosh b n x ◊ cos l n y - p y2 ...eqn. 19.23
4 fx 0
Hence, the expressions for the pseudo stress resultants are:
∂2 f •
a2 ¸
nx =
∂y 2
= Â An l 2n cosh b n x cos l n y -
2 fx
p0 Ô
n =1,3,... Ô
Ô
Ô
∂ f
2
˝
n y = 2 = - Â Anb 2n cosh b n x cos l n y
∂x Ô ...eqn 19.24
Ô
Ô
∂ f
2
Ô
n xy = - = - Â Anb n l n sinh b n x sin l n y ˛
∂x ∂y
Because the traverses cannot receive any load normal to their planes, the boundary conditions to be
satisfied are
nx = 0 at x = ±a ...(1)
and ny = 0 at y = ±b. ...(2)
Boundary condition (2) is automatically satisfied, since, at y = ±b, sin λny = 0. Boundary condition
(1) helps in finding An. To apply this boundary condition, it is necessary to expand the uniform load p0
in Fourier series form in the y-direction. Thus, we take,
• n -1
4 p0
p0 = Â ( -1) 2 cos l n y
n =1,3,... np
Then from boundary condition (1), we get

n -1
a2 4p
0 = Â An l 2n cosh b n a cos l n y - Â 0 ( -1) 2 cos l n y
2 fx np
For all values of n,
n -1
a 2 4 p0
An l 2n cosh b n a cos l n y = ( -1) 2 cos l n y
2 f x np
n -1
2 p0 a 2 ( -1) 2
∴ An = ...eqn. 19.25
l 2m f x np cosh b n a
Hence,
¸
ÏÔ 2
p0 a 2 n -1
cosh b n x cos l n y 1 ¸Ô Ô
nx = Ì Â ( -1) 2 - ˝
fx ÓÔ p n cosh b n a 2 ˛Ô Ô
Ô
Ô
p0 b 2 ÏÔ 2 n -1
cosh b n x ◊ cos l n y ¸Ô Ô
ny = - Ì Â ( -1) 2 ˝ ˝ ...eqn. 19.26
f y ÓÔ p n cosh b n a ˛Ô Ô
Ô
p ab ÔÏ 2 n -1
sinh b n x ◊ sin l n y Ô¸ ÔÔ
n xy = - 0 Ì Â ( -1) 2 ˝
f x f y ÓÔ p n cosh b n a ˛Ô Ô
˛
The corresponding membrane stresses may be found by using the following relations:
1 + p2 ¸
Nx = x nx Ô
1+ q 2
Ô
Ô
Ô ...eqn. 19.27
1 + q2 ˝
Ny = ny Ô
1 + p2 Ô
Ô
and N xy = n xy . Ô
˛
Convergence Study
Parme A.L. reported the following convergence study:
nx and ny converge rapidly and hence, three to four terms of the series give satisfactory accuracy.
However, the expression for shear converge rather slowly at x = ±a. To force nxy to converge rapidly, the
following recommendations has been made.
p0 ab ÔÏ 2 n -1
sinh b n a sin l n y Ô¸
n xy =- Ì Â ( -1) 2 ˝
x=a f x f y ÔÓ p n cosh b n a Ô˛
ÈÏ n -1 ¸ ˘
pa ab Í Ô 2 ( -1) 2 È sinh b n a ˘ Ô ˙
=- Ì Â Í cosh b a - 1˙ + 1˝ sin l n y ˙
f x f y ÍÎ ÔÓ p n Î n ˚ ˛ Ô ˚
n -1
• ( -1) 1 Ê py py ˆ
2
Â
2
However, sin l n y = log Á sec + tan ˜
n =1, 3,... n 4 Ë 2b 2b ¯
ab È 1 Ê py py ˆ
∴ n xy = - p0 Í log ÁË sec + tan ˜
f x fy Î 2p 2b 2b ¯
n -1 ˘
( -1)
- Â (1 - tanh b n a ) sin l n y ˙˙
2 2
p n ˚
fx
For values > 1, tanh b n a = 1 and hence, the second term in the expression may be ignored except
fy
perhaps for n = 1.
py py
At y = ±b, sec and tan are infinite. Hence, it indicates nxy at the corner is infinite. This would
2b 2b
be true, if the corners were completely free of normal forces and if the shell had no bending resistance.
However, because of the integral action of the supporting ribs and shell, normal forces do exist at the
corner. These normal forces alter the resistance to the extent that, the shear does not need to be infinite
to satisfy statics. Moreover at the corner some of the load can be and is resisted by flexural resistance.
From the studies made on cylindrical shells, it has been found that this flexural action is confined to a
distance of approximately 0.4 rt from the rib, in which r is the radius of curvature of the shell and t
is the thickness. Therefore, it is felt that the expression for shear do not apply within the distance
0.4 rt from the corner. Shear can be considered maximum at the point y = b - 0.4 rt .
19.9 MEMBRANE THEORY OF ANTICLASTIC SHELLS

Hyperbolic paraboloid and conoid are the commonly used anticlastic shells.
19.9.1 Hyperbolic Paraboloid
Geometry: If a convex parabola moves over a concave parabola or vice versa, a hyperbolic paraboloid
is generated. Vertical sections of the surfaces are paraboloid and the horizontal sections are hyperbolas.
Figure 19.5 shows a typical H.P. shell.
x
fy
y z
fx
Fig. 19.5 Hyperbolic Paraboloid
Let the parabolas used be

fx
z1 = 2
x2
a
fy
z2 = y2 .
b2
The geometry of the shell is defined by
fx fy
z = - z1 + z2 = - 2
x2 + y2 .
a b2
∂2 z
= 0 i.e. x and y are principal directions
∂x∂y
1 ∂2 z 2 f
∴ = 2 = 2x
R1 ∂x a
1 ∂2 z 2 fy
= 2 = 2
R2 ∂y b
where R1 and R2 are the principal curvatures. Thus, in terms of principal curvatures, the surface may
be defined as
x2 y2
z=- + ...(a)
2 R1 2 R2
setting z = 0, we get
Ê x y ˆÊ x y ˆ
+ - + =0 ...(b)
Á 2R ˜ Á 2 R2 ˜¯
Ë 1 2 R2 ¯ Ë 2 R1
Equation (b) represents a pair of straight lines existing on the surface. Their inclination to x-axis is
given by (Ref. Figure 19.6).
R2
tan g = .
R1
g
g
Fig. 19.6
If R2 = R1, tan γ = 1 or γ = 45°. Hence, the angle between two pairs of straight lines is 90° i.e. they
are orthogonal. Such surfaces are known as rectangular hyperbolic paraboloid.
If the asymptotes are chosen as coordinate axes (Ref. Figure 19.7), let x′, y′ be the coordinate system.
O g C
x axis
g
B g y¢
g D
E
y¢ axis
x¢
x¢axis
y axis
Fig. 19.7
Then looking at point A,

x = OC = BD − BE
= x′ cos γ − y′ cos γ
= (x′ − y′) cos γ
y = AC = OE + AD
= y′ sin γ + x′ sin γ
= (x′ + y′) sin γ
∴ Equation of hyperbolic paraboloid with respect to asymptotic axes is,
z=-
x2
+
y2
=-
( x ¢ - y ¢ ) cos2 g ( x ¢ + y ¢ ) sin 2 g
+
2 2
2 R1 2 R2 2 R1 2 R2
È 2 ˘
Í - ( x ¢ - y ¢ ) cos g R + ( x ¢ + y ¢ ) sin g ˙
1 2 2 R2
=
2 R2 Î 1 ˚
- ( x ¢ - y ¢ ) cos 2 g tan 2 g + ( x ¢ + y ¢ ) sin 2 g ˘˚

1 È 2 2
= Î
2 R2
- ( x ¢ - y ¢ ) + ( x ¢ + y ¢ ) ˘˚ sin 2 g
1 È 2 2
= Î
2 R2
2
=+ x ¢y ¢ sin 2 g
R2
For rectangular hyperbolic paraboloid γ = 45°.
1
∴ z=- x ¢y ¢.
R
Looking at Fig. 19.8, it may be visualised as surface made up of straight line generaters. Then
xf f
a
Fig. 19.8
x y
Z= f◊
a b
f xy
= xy =
ab c
ab
where c = is radius of curvature.
f
Analysis of Hyper Shells (Rectangular Hyperbolic Shells)

The shell surface is given by
xy ab
z= where c = .
c f
∂z y ∂z x
p= = , q= =
∂x c ∂y c
∂2 z 1 ∂2 z
r = 0, s= = t= = 0.
∂x ∂y c ∂y 2
Hence, the equation of equilibrium reduces to
2s nxy = −Z + pX + qY
For self wt:
X = 0, Y = 0, Z = g 1 + p2 + q2
1 x 2 y2
∴ 2 n xy = - g 1 + 2 + 2
c c c
gc x2 y2
n xy = - 1+ 2 + 2
2 c c
g 2
=- c + x 2 + y2
2
From equation of equilibrium 1,
∂n x ∂n xy
∂x
=-
∂y
g 1
(
= + ◊ 2 y ¥ c2 + x 2 + y2
2 2
)-1 2
g y
=
2 c + x2 + y2
2
nx =
g Ú y dx + C.
∴ 2 c2 + x 2 + y2
log ÈÎ x + c 2 + x 2 + y 2 ˘˚ + f1 ( y )
gy
=
2
From equation of equilibrium 2,
∂n y ∂n xy g ( -2 x )
=- =-
∂x ∂x 2 2 c + x 2 + y2
2
log ÈÎ y + c 2 + x 2 + y 2 ˘˚ + f2 ( x )
gx
=
2
f1(y) and f2(x) are to be evaluated from the boundary conditions.
Umbrella Roof: Figure 19.9 shows a typical umbrella type H.P. Shell.
x
y
z
Fx
a a
Fig. 19.9 A typical umbrella type H.P. Shell
End beams are thin and deep.

∴ nx = 0 at x = a
and ny = 0 at y = b.
f1 ( y ) = - log ÈÎ a + c 2 + a 2 + y 2 ˘˚
gy
∴
2
log ÈÎ b + c 2 + x 2 + b 2 ˘˚
gx
f2 ( x ) = -
2
gy x + c2 + x 2 + y2
∴ nx = log
2 a + c2 + a2 + y2
gx y + c2 + x2 + y2
ny = log
2 b + c2 + x 2 + b2
For a shallow shell:

p2 and q2 are negligible compared to unity. Hence,
Z = g 1 + p2 + q2 = g
∴ Equation of equilibrium is
1
2 ¥ n xy = - g
c
gc
∴ n xy = -
2
nx = 0 and ny = 0.
Thus, a shallow shell subjected to dead load only is in a state of pure shear.
Shallow and Deep shell
For a deep shell,
g 2
n xy = - c + x2 + y2
2
gc Ê x 2 + y2 ˆ
=- 1+ Á
2 Ë c 2 ˜¯
gc È a 2 + b 2 ˘
∴ n xy =- Í1 + ˙
max 2 Î c2 ˚
gc È 1 a 2 + b 2 ˘
=- Í1 + + ...˙
2 Î 2 c 2
˚
gc È 1 a 2 + b 2 ˘
=- Í1 + ˙
2 Î 2 c2 ˚
a
Let a ≤ b, so that £1
b
a 2 + b2 a 2 + b2
Hence, =
2c 2 a 2 b2
2
f2
1Ê a2 ˆ f 2 1 f2 a2
= + £ <1
2 ÁË ˜
1 since
b2 ¯ a 2 2 a2 b2
f 1 a 2 + b2 1
Thus, if = , 2
£
a 10 2c 100
or 1 percent of first term.
f 1 a 2 + b2 1
Again, if = , 2
£ of first term
a 5 2c 25
i.e. less than 4% of first term.
f
Hence, for all practical purpose, a H.P. shell may be considered shallow, if is less than or equal
a
1
to .
5
Example. Design a 10 m × 12 m invested umbrella type H.P. shell.
A E B
H F
D
C
G
Fig. 19.10
Figure 19.10 shows a typical invested umbrella type of H.P. Shell. In this shell,
a = 5 m and b = 6 m.
1 1
To make it shallow amount of pulling down of a corner is kept th to th of least dimension i.e.
5 7
‘a’. Hence, in this case,
5
f = 1m to m.
7
Let it be 1m.
ab 5 ¥ 6
∴ Radius of curvature c = = = 30 m.
f 1
1 1
Thickness of the roof is kept between to th of radius of curvature. Thus,
400 500
1 1
t= ¥ 30 ¥ 1000 to ¥ 30 ¥ 1000 = 75 mm to 60 mm.
400 500
Let t = 60 mm.
Loads:
Dead load = 0.060 × 1 × 1 × 25 = 1.5 kN/m2
1
Live load: In this case, slope of line joining springing point and the crown is a = tan -1 < 10∞.
52 + 6 2
∴ L.L = 0.75 kN/m2
Let total load on shell (including finishing load) be 2.4 kN/m2.
Since, the shell is shallow, it is subjected to pure shear,
gc 2.4 ¥ 30
n xy = -
=- = 36 kN m.
2 2
Pure shear produces tensile/compressive stresses in diagonal directions, the magnitude being the
same (36 kN/m).
36 ¥ 1000
∴ Ast = = 240 mm 2 ,
150
Using 8 mm diameter bars,
p
¥ 82
spacing = 4 ¥ 1000 = 209 mm.
240
Provide 8 mm bars at 200 mm c/c. Actually these bars are required in the direction of tensile stresses
(45° to x axis) and distribution bars are required at 90° to that direction. In these directions, it is difficult
to bend the bars. Hence, usually reinforcement is provided in x and y directions in which bars are to be
straight. Provide 8 mm bars at 200 mm c/c in x and y directions so that the component of steel at 45°
is sufficient to take diagonal tension.
Design of Peripheral Edge Beam
Along the edges of the shell, thin but sufficiently deep edge beams are to be provided to take up shear
from the shell and transfer the load to columns. Figure 19.11(a) shows peripheral edge beams with the
load transferred by shell and Figure 19.11(b) shows the transfer of load by beams connecting peripheral
edge beam and columns. It may be noted that peripheral edge beams are subjected to tensile forces and
the beams connected to column are subjected to compressive stresses, the magnitude being same as
shear in the shell (36 kN/m).
Maximum tension in peripheral beam is at section B in 6 m beam.
T = 36 × 6 = 216 kN.
216 ¥ 1000
Ast = = 1440 mm 2 .
150
Provide 8 bars of 16 mm diameter.
Provide nominal shear reinforcement. Use same section for 5 m beams also.
A E
B
(a) Force on peripheral beam
F
H
(b) Force on beam connecting peripheral beam and column
Fig. 19.11
Edge Beam Dimension

To control crack width, maximum permissible direct tension in M: 25 concrete = 1.3 N/mm2.
280
for M: 25 concrete, m = ª 11.
3 ¥ 8.5
Hence, area of concrete required is
216 ¥ 1000
1.3 =
A + (11 - 1) ¥ 1440
A = 151753 mm2.
Provide, 225 × 675 mm beams.
These edge beams should be connected to shell symmetrically so as to avoid bending.
Edge beam
Shell
Design of Beams Connecting Edge Beams and Column
Length of beam = 6 + 1 = 6.00276 m.

2 2
∴ Maximum compressive force = 36 × 6.00276

= 218.98 kN.
From architectural point of view, the size of beam is kept same as peripheral beam.
= 225 × 675 mm.
Compressive stress in concrete

218.98 ¥ 1000
= = 2.16 N mm 2 .
150 ¥ 675
Hence, concrete alone can resist it. Provide minimum reinforcement.
0.8
Asc =
¥ 225 ¥ 675 = 1215 mm 2 .
100
Provide 6 bars of 16 mm diameter. Use same section for 5 m beams also.
225
225
Nominal Nominal
675 mm 675 mm
Shell
8 of 16 f 6 of 16 f
(a) Peripheral edge beam (b)
Fig. 19.12
The sections are shown in Figure 19.12.
19.9.2 Membrane Analysis of Conoid
Figure 19.13 shows a typical conoid. It may be generated by moving a straight line with one of its end
on a straight line and the other end on a plane curve, keeping the line parallel to itself on a projected
plane. The projected plane is known as director plane. The plane curve and the straight line are known
as directrices. The conoid is inefficient in transferring the load by membrane action near straight line
directrix. Hence, many times truncated conoids (Fig. 19.14) are used.
Geometry of Conoid: The plane curve used as a directrix may be circular, parabolic or catenary.
Accordingly the conoids are known as circular, parabolic and catenary conoids. Of these the commonly
used conoid is the parabolic.
The glazing provided at plane curve end may be vertical or inclined as shown in Fig. 19.15. In this
article, analysis is carried out for type–I parabolic conoid.
ne
pla
to r
ir ec
D
St. line director
l
f
Curved director
B
Fig. 19.13 A conoid Fig. 19.14 Truncated conoid
Type I conoid Type II conoid
Fig. 19.15 Types of Conoids
The equation of type–I parabolic conoid, with x, y, z, axes selected as shown in Figure 19.16.
4 f Ê B2 ˆx
z=- 2 Á
- y2 ˜
B Ë 4 ¯ l
0
∂z 4 f Ê B2 ˆ a Ê B2 ˆ
∴ p= =- 2Á - y2 ˜ = - Á - y2 ˜
∂x lB Ë 4 ¯ Ë
2 4 ¯ y
z
∂z 8 f xy
q= = = axy
∂y lB 2
l
∂2 z
r= =0 f
∂x 2
∂ 2 z 8 fy
s= = = ay x
∂x∂y lB 2 B
∂2 z 8f 8f
t= = x = ax where a = Fig. 19.16 Geometry of Parabolic Conoid of type–I
∂y 2 lB 2 lB 2
Analysis for Snow Load

All cross sections of shell are parabolas. For snow load parabola is a funicular shape. Hence, for snow
load the shell degenerates into a series of independent arches. In other words, nx = nxy = 0. Hence, the
equilibrium equation assumes the form
tny = −p0
p0
or ny = -
t
lB 2 p
i.e. n y = - p0 =- 0
8 fx ax
∴ An appropriate stress function is
p0
f0 = - x ( log x - 1)
a
Analysis for Dead Load
If dead load is g/unit surface area,
X = 0, Y = 0, Z = g 1 + p 2 + q 2
È 1
Î 2
( ˘
ª g Í1 + p 2 + q 2 ˙
˚
)
È 2 ˘
Í 1 a 2 Ê B2 ˆ 1 2 2 2˙
i.e. Z = g 1+ ¥ Á -y ˜ + a x y
2
ÍÎ 2 4 Ë 4 ¯ 2 ˙˚
È Ï B4 2 B 2 y 2 y 4 1 2 2 2 ¸˘
= g Í1 + a 2 Ì - + + a x y ˝˙
Î Ó 8 ¥ 16 8 ¥ 4 8 2 ˛˚
È a2 B4 ˘ È 4 1 2 2 x 2 y2 ˘
2 y
= g Í1 + ˙ + ga Í - B y + ˙
Î 8 ¥ 16 ˚ Î 8 16 2 ˚
È 64 f 2 B 4 ˘ È 4
2 y 1 x2 y2 ˘
= g Í1 + 2 4 ˙ + ga Í - B2 y2 + ˙
Î l B 8 ¥ 16 ˚ Î 8 16 2 ˚
È 1 f2˘ È 4 1 2 2 x 2 y2 ˘
2 y
= g Í1 + ˙ + ga Í - B y + ˙
Î 2 l2 ˚ Î 8 16 2 ˚
Hence, from equation of equilibrium, we get
È 1 f2˘ È 4 1 2 2 x2 y2 ˘
2 y
2 an xy + axn y = - g Í1 + ˙ - ga Í - B y + ˙
Î 2 l2 ˚ Î 8 16 2 ˚
The above equation in terms of the stress function φ is,
∂2f ∂2 f È 1 f2˘ È 4 B2 y2 x 2 y2 ˘
2 y
- 2 ay + ax 2 = - g Í1 + ˙ - ga Í - + ˙
∂x∂y ∂x Î 2 l2 ˚ Î8 16 2 ˚
First part of the load term is constant and may be regarded similar to snow load. Hence, the solution
for the first part of load is
gÈ f2 ˘
f1 = -Í1 + ˙ x ( log x - 1)
a Î 2l 2 ˚
For the second part of the load let the stress function be φ2.
∂ 2 f2 ∂ 2 f2 È 4 B2 y2 x 2 y 2 ˘
2 y
Hence, -2 ay + ax = - ga Í - + ˙
∂x ∂ y ∂x 2 Î8 16 2 ˚
∂2 f2 ∂2 f2 È y4 B2 y2 x 2 y2 ˘
i.e. -2 y +x = - ga Í - + ˙
∂x∂y ∂x 2 Î8 16 2 ˚
The above equation is satisfied by a stress function of the form
f 2 = Â Amn x m y n
or f 2 = Â Amn ( x m - l m ) y n
The latter form is preferred as it can be made to satisfy the boundary condition at x = l easily. Using
chosen stress function, we get,
È y 4 1 2 2 x 2 y2 ˘
Â -2 y Amn mn x m -1 n -1
y + x Amn m ( m - 1) x m -2 n
y = - ga Í - B y +
Î 8 16 2 ˚
˙
È y 4 B2 y2 x 2 y2 ˘
i.e. Amn x m -1 y n m ( -2 n + m - 1) = - ga Í - + ˙
Î 8 16 2 ˚
The values of Amn for each of the load term may be found seperately and total solution obtained.
For first term:
y4
Amn x m -1 y n m ( -2 n + m - 1) = - ga
8
Comparing the powers of x and y, we get
m − 1 = 0 i.e. m = 1
and n = 4.
ga
∴ Amn ¥ 1 ( -2 ¥ 4 + 1 - 1) = -
8
ga
or Amn =
64
∴ Solution for first term of load is
ga
f2 = ( x - l ) y 4 = - ga (l - x ) y 4
64 64
Similarly for second term of load

m = 1 and n = 2
ga 2
∴ Amn ¥ 1 ( -2 ¥ 2 + 1 - 1) = B
16
ga
or Amn = - ( x - l ) B2
64
ga
Hence, f2 = - ( x - l ) B 2 x y 2 = ga (l - x ) B 2 xy 2
64 64
For third term of load,
m = 3 and n = 2.
1
∴ Amn ¥ 3 ( -2 ¥ 2 + 3 - 1) = - ga
2
1
or Amn = ¥ ga
12
f2 = (x - l ) y
ga 3 3 2
12
ga ( 3
=- l - x3 ) y2 .
12
∴ The solution is,
2
f2 = -
ga
(l - x ) y 4 + gaB (l - x ) y 2 - ga (l 3 - x 3 ) y 2
64 64 12
The total solution for self weight is
f = f1 + f 2
gÈ f2 ˘ gaB 2
=- Í1 + ˙ x ( log x - 1) -
ga
(l - x ) y 4
+ (l - x ) y 2 - ga (l 3 - x 3 ) y 2
a Î 2l ˚2 64 64 12
∂2 f È3 B2 1( 3 3)
˘
nx = = - ga Í (l - x ) y 2
- ( l - x ) + l - x ˙
∂y 2 Î 16 32 6 ˚
∂2 f ga 2 g È f2 ˘
ny = = xy - Í1 + 2 ˙
∂x 2 2 ax Î 2l ˚
∂2f È y 3 yB 2 x 2 y ˘
n xy = - = ga Í - + ˙
∂x∂y Î 16 32 2 ˚
It may be easily verified that the boundary conditions n x x =l = 0 is satisfied. The boundary condition
nxy = 0 along y = 0 (due to symmetry) is also satisfied. Since, conoid is of hyperbolic type, no boundary
conditions may be prescribed on the edges x = ±B/2 which are (open boundaries).
QUESTIONS
1. Explain the term ‘Pseudo forces’ and derive the relationship between pseudo forces and membrane forces. Use
Monge’s notations.
2. Derive the equations of equilibrium for the analysis of doubly curved shells. Use Pseudo forces and Puchers
stress function.
3. State the Pucher’s equation of equilibrium for the analysis of doubly curved shell. Derive membrane solution
for a shell subject to snow load only.
4. Find the membrane solution for a rectangular hyperbolic parabola subject to self weight only. Show that if the
1
ratio of the amount of pulling down of a corner to shorter side is less than or equal to , it may be treated as
5
shallow shell.
5. Design an inverted umbrella type H.P. shell to cover an area of 12 m × 15 m.
6. Differentiate between full conoid and truncated conoid and explain the relative merits and demerits. Discuss
the need for a bending theory of conoids.
References and Bibliography
1. Aas-Jabobsen A., Zyhndershlen mit Underlichen Krummugshalbmersser and Verdnderlicher
Schdlenstdrke, Barigenieur, Vol-18, Berlin, 1937.
2. Arya A.S., Elastic Theories of Cylindrical Shells, Proceedings of the Seminar on Shell Structures,
Bengal Engineering College, Shibpur, Jan 1962 – Indian Industries, Oxford Book and Stationary
Co., Calcutta, Vol-5, No.-3, Paper No:3
3. Baikov V.N., Reinforced Concrete Structures, (translated by Savin I.V.) M.R. Publishers, Mascow,
1978.
4. Bairagi N.K., A Text Book of Plates Analysis, Khanna Publications, New Delhi, 1986.
5. Bairagi N.K., Shell Analysis, Khanna Publications, New Delhi, 1990.
6. Baker E.H., Kovalevsky and Rich RFL, Structural Analysis of Shells, Mc Graw Hill, 1972.
7. Banerjee S.P., Numerical Analysis of Conoidal Shells, International Association for Bridge and
Structural Engineering, Vol-24, 1964.
8. Bhavikatti S.S. Study of Circular Cylindrical Shells with Various End Conditions, M.Tech. thesis
UOR Roorkee, 1967.
9. Bhise V.M., Apte V.P. and Phadke B.S., Membrane Analysis of Cotenery and Circular Conoidol
Shells, Indian Concrete Journal, Vol-36, No-5 May 1962.
10. Billington D.P., Thin Shell Concrete Structures”, McGraw Hill Book Company, New York, 1965.
11. Candela F., General Formulas for Membrane Stresses in Hyperbolic Paraboloidical Shells. Journal
of the American Concrete Institute, Proceedings, Vol-57, October 1960.
12. Candela F., Structural Application of Hyperbolic Paraboloidical Shells, Journal of American Concrete
Institute, Proceedings Vol-51, January 1955.
13. Chandrashekhar K., Theory of Plates, University Press Ltd., 2001.
14. Chatterjee B.K., Theory and Design of Concrete Shells, Oxford and IBH Publishing Co., Calcutta,
1971.
15. Chavan V.K., Practical Design of Cylindrical Shell Roofs, Popular Prakashan, Bombay, 1967.
16. Directives for the Design, Calculation and Execution of Cylindrical Shells in Reinforced Concrete,
Report No:12, Report of the Commission Schaaldaken, Commission for conducting Research,
Gravenhage, Holland, December, 1956.
17. Dischinger F., Die Strange Theorie der Kreiszyhinderschale in ihrer Anwendudng anf die Zeiss-
Dywidagschalen, betan un Eisen, Vol-34, 1935.
18. Donnell l.H., The Stability of Thin Walled Tubes under Torsion, National Adviosory Committee for
Aeronautics, report 479, Washington D.C., 1934.
19. Finsterwalder U, Die Quervesteinflen Zylinderschen Schalengewolbe mit Kreissegmentformigen

Queschnitt, Ing-Arch, Vol-4, 1933.
20. Finsterwalder U., Di Theorie der Kreiszylindrischen Schalengewoble System Zeiss-Dywidag Und
ihre Answerdung and die Grossmarkthalle in Budapest, International Association for Bridges and
Structural Engineering , Vol-I, 1932.
21. Fischer L., Design of Cylindrical Shells with Edge Beams, Journal of ACI Vol-52, p-481, December
1955.
22. Fischer L., Model Investigation for a Conoidal shell 30m x 60m for the Jersulam swimming pool,
Proceedings of International Association for Shell Structures, Symposium on Shell Research, North
Holland Publishing Company, Delft., 1961.
23. Fisher L., Determination of membrane Stresses in Elliptic Paraboloids using Polymials, Journal of
ACI, Vol-32, No-4, October 1960.
24. Fisher L., Theory and Practice of Shell Structures, Wilhelm Ernst and John, Berlin – Munich, 1968.
25. Flugge, W., Stresses in Shells, Springer-Verlog OHG, Berlin, 1960.
26. Gibson J.E., Computer Analysis of Cylindrical Shells, Spon Limited, London, 1961.
27. Gibson J.E., Linear Elastic Theory of Thin Shells Pergraman Press, 1965.
28. Goldberg J.E. and Leve H.L., Theory of Prismatic Folded Plate Structures, Publications of
International Association for Bridge and Structural Engineering Vol-17, 1957, pp:59-86.
29. Haas, A.M., Design of Thin Concrete Shells – Vol-2, Negative Curvature Index, John Wiley and
Sons, Inc., New York, 1967.
30. Holland I, Analysis of Folded Plate Structures by Transfer Matrices, Proceedings of the Colloquim
on Simplified Calculation Methods, Brussels September 4-6, 1961, North Holland Publishing
Company, Amsterdam.
31. Holland I., Design of Circular Cylindrical Shells, Norges Tekniske, Vitenskapsakademi, 1957.
32. Indian Standard Criteria for Design of Reinforced Concrete Shell Structures and Folded Plates, IS
2210-1988, Bureau of Indian Standards, New Delhi.
33. J.N. Bandopadhyay, Thin Shell Structures - Classical and Modern Analysis, New Age International
Publishers, New Delhi, 2008.
34. Jenkins R.S., Theory and Design of Cylindrical Shell Structures, O.N. Arun Group of Consulting
Engineers, Bulletin 1, London 1947.
35. Jenssen O., Methods for Analysis of Concrete Shell Roofs, Proceedings of the second Symposium
on Concrete Shell Roof Construction, October 1957, P-111
36. Karman T.V. and Tsein H.S., The Buckling of Thin Cylindrical Shells under Axial Compression,
Journal of Aeronautical Science, Vol-8, No- 8, June 1941.
37. Kazuhiko, Mashita, Yasav Tamaba and Midiya Hara, Ultimate Strength Analysis of Reinforced
Circular Cylindrical Shells Stiffened with Edge Beams Proc. 3rd Summer Collogium on Shell and
Spatial Structures, Taegue, South Korea, August 1-10, 1990, pp:223-235
38. Krass H., Thin Elestic Shells, John Wiley and Sons, 1967.
39. Lakshmy T.K. and Bhavikatti S.S., Optimum Design of Multiple Barrel Shells, Bulletin of
International Association of Shells and Spatial Structures, Vol-I, 36, pp:53-63, April 1995.
40. Lakshmy T.K. and Bhavikatti S.S., Optimum Design of Single Barsel Shells, Proc. National Seminar
on Structural Mechanics, IISc Bangalore, pp:27-28, 1994.
41. Lakshmy T.K. and Bhavikatti S.S., Optimum Design of Trough –Type RC Folded Plate Roofs,
Computers and Structures, Vol-57, No-1, pp:125-130, 1995.
REFERENCES AND BIBLIOGRAPHY 297
42. Lakshmy T.K. and Bhavikatti S.S., Optimum Design of V-Type RCFolded Plate Roofs, National
Seminar on Recent Trends in Concrete and Steel Structures, IIT Madras (Chennai), April 26-27,
1997.
43. Lakshmy T.K. Optimum Design of large –span R.C. Structures, Ph.D. thesis, Mangalore University
1993.
44. Lundgren H., Cylindrical Shells, The Danish technical Press, The Institute of Danish Civil Engineers,
Copenhagen, 1949.
45. Manual No:31, ASCE, 1952., Design of Cylindrical Concrete Shell Roofs,
46. Parme A.L., And Corner H.W., Design Constants for Interior Cylindrical Concrete Shells, ACI
Journal, Proceedings, V-58, No-1, July 1961.
47. Parme A.L., Hyperbolic Paraboloid and other Shells of Double Curvature, Journal of Structural
Division, Proceedings of the American Society of Civil Engineers, Vol-82, No-ST5, September
1956.
48. Phase-I report on Folded Plate Construction, task Committee on Folded Plate Construction of the
Committee on Masonry and reinforced Concrete, Journal of the Structural Division, Proceedings
ASCE, Vol-89, No-ST 6, 1963 pp:365-406
49. Pucher A, Uber Den Spannugzustand in Doppett Gerkriimmten Flachen, Vol-33, Beton Und Eisen,
1934.
50. Ramaswamy G.S. and Keshava Rao M.N., The Membrane Theory Applied to Hyperbolic Paraboloid
Shells, The Indian Concrete Journal, Vol-35, September 1961.
51. Ramaswamy G.S., Design and Construction of Concrete Shell Roofs, CBS Publisher and Distributors,
New Delhi, 1986.
52. Rao G.S.,Membrane Analysis of a Conoidal Shell with a Parabolic Direction, Indian Concrete
Journal, Vol-35, September, 1961.
53. Reiss M and Yitzak Michael, Analysis of Short Folded Plates, Journal of structural Division
Proceedings ASCE Vol-91, No.ST5, 1965, pp:233-254.
54. Schorer H, Live load Action on Thin Cylindrical Shells, Proceedings ACSE Vol-61, 1935
55. Simpson H., Design of Folded Plate Roofs, Journal of the Structural Division, Proc. ASCE Vol-84,
No-ST3, 1958, pp:1508-1-1508-21
56. Soare M., Membrane Theory of Conoidal Shells, Cement and Concrete Association, 52.Grossveror
gardens, London 1957.
57. Szilard R. Theory and Analysis of Plates—Classical and Numerical Methods, Prentice Hall, 1994.
58. Tamhankar M.G. and Jain R.D., Computer Analysis of Folded Plates, ICJ October 65.
59. Timoshenko S.P. and Wrinoskt-Kreiger, Theory of Plates and Shells, Mc Graw Hill Book Co.,
NewYork, 1959.
60. Whitney C S and Andersson B.G and Birnabanum, Reinforced Concrete Folded Plate Construction,
Journal of the Structural Divn. Proc. ASCE Vol-85, No-ST8, 1959, pp:15-43.
61. Winter G and Pei M, Hipped Plate Construction, Journal of the ACI, Proceedings Vol-43, 1947,
pp:505-531.
62. Yitzaki David, The Design of Prismatic and Cylindrical Shell Roofs, Haifa Science Publishers
Haifa Israel, 1958.
63. Yitzhaki D., The Design of Prismatic and Cylindrical Shell Roofs, North Holland, Publishing
Company, Amsterdam, 1959.

Theory of Plates and Shells

Uploaded by

Copyright:

Available Formats

Theory of Plates and Shells

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Theory of Plates and Shells

Uploaded by

Copyright:

Available Formats

This page

NEW AGE INTERNATIONAL (P) LIMITED, PUBLISHERS

All rights reserved.

ISBN (13) : 978-81-224-3492-7

PUBLISHING FOR ONE WORLD

1. Introduction to Plates 1–4

2. Pure Bending of Plates 5–22

3. Small Deflections of Laterally Loaded Plates 23–30

4. Fourier Series of Loadings 31–42

5. Navier’s Solution for Rectangular Plates 43–46

6. Levy’s Solution for Rectangular Plate Analysis 47–54

7. Rectangular Plates with Various Edge Conditions 55–72

8. Circular Plates Bent Axi-symmetrically 73–91

9. Plates of Other Shapes 92–105

10. Energy Method 106–119

11. Finite Difference Method 120–141

12. A Folded Plate Roofs 142–157

13. Introduction to Shells 158–163

14. Introduction to Cylindrical Shell Roof 164–181

15. Membrane Analysis of Cylindrical Shell Roofs 182–196

16. Bending Theory of Cylindrical Shell Roofs 197–227

17. Analysis of Symmetrically Loaded Cylindrical

18. Membrane Theory for Shells of Revolutions 245–266

19. Membrane Theory for Shells of Translation 267–294

References and Bibliography 295–297

Fig. 1.1 Shapes of plates

Fig. 1.2 Edge conditions

1.1 COORDINATE SYSTEMS

Fig. 1.3 Different orientation of coordinates with right hand rule

Fig. 1.4 Polar coordinates for circular plate

1.2 STRESSES ON AN ELEMENT

Fig. 1.5 Stresses on an element

1.3 TYPES OF THEORIES OF PLATES

2.1 SLOPE IN SLIGHTLY BENT PLATE

Fig. 2.1 (a) Plan view of element

Fig. 2.1 (b) Sectional view of element

Expression for Maximum Slope:

2.2 CURVATURE OF SLIGHTLY BENT PLATE

1 ∂ Ê ∂w ˆ ∂2w ...eqn. 2.8

Similarly, curvature in y-direction

Fig. 2.2 Positive sense of curvature

Curvature in any direction

1 Ê 1 1 ˆ Ê 1 1 ˆ cos 2 a 1 ...eqn. 2.10

Adding eqns. 2.10 and 2.11, we get

2.3 PRINCIPAL CURVATURE

2 rxy Fig. 2.4

Fig. 2.5 Mohr’s circle for curvatures

2.4 DISPLACEMENT—STRAIN RELATIONS

Fig. 2.6 Deflected shape of element

2.5 STRAINS IN TERMS OF w

u rotated middle surface

Fig. 2.7 Displacement u in terms of w

From the figure, it is clear that

2.6 STRAIN-STRESS RELATIONS

Fig. 2.9 Stresses on an element

From theory of elasticity, we know

2.7 EXPRESSIONS FOR MOMENTS

Fig. 2.10 Sign convention for moments

2.8 MOMENT IN ANY DIRECTION

Fig. 2.11 Element considered