You are on page 1of 74


Stress in 3D

Lecture 1: STRESS IN 3D


1.1 Introduction . . . . . . . . . . . . . . . . . . . . . 13
1.2 Mechanical Stress: Concept . . . . . . . . . . . . . . . 13
1.3 Mechanical Stress: Denition . . . . . . . . . . . . . . . 13
1.3.1 What Does Stress Measure? . . . . . . . . . . . . 13
1.3.2 Cutting a Body . . . . . . . . . . . . . . . . . 13
1.3.3 Orienting the Cut Plane . . . . . . . . . . . . . 14
1.3.4 Internal Forces on Elemental Area . . . . . . . . . . 15
1.3.5 Projecting the Internal Force Resultant . . . . . . . . 15
1.3.6 Defining Three Stress Components . . . . . . . . . . 15
1.3.7 Six More Components . . . . . . . . . . . . . . 16
1.3.8 Visualization on Stress Cube . . . . . . . . . . . . 17
1.3.9 What Happens on the Negative Faces? . . . . . . . . 17
1.4 Notational Conventions . . . . . . . . . . . . . . . . . 18
1.4.1 Sign and Subscripting Conventions . . . . . . . . . . 18
1.4.2 Matrix Representation of Stress . . . . . . . . . . . 19
1.4.3 Shear Stress Reciprocity . . . . . . . . . . . . . 19
1.5 Simplications: 2D and 1D Stress States . . . . . . . . . . . 110
1.6 Advanced Topics . . . . . . . . . . . . . . . . . . . 110
1.6.1 Changing Coordinate Axes . . . . . . . . . . . . . 110
1.6.2 A Word on Tensors . . . . . . . . . . . . . . . 111
1.7 Addendum: Action vs. Reaction Reminder . . . . . . . . . . 112


1.1. Introduction

This lecture introduces mechanical stresses in a 3D solid body. It covers definitions, notational and
sign conventions, stress visualization, and reduction to two- and one-dimensional cases. Effect of
coordinate axes transformation and the notion of tensors are briefly mentioned as advanced topics.

1.2. Mechanical Stress: Concept

Mechanical stress in a solid body is a gross abstraction of the intensity of interatomic or inter-
molecular forces. If we look at a solid under increasing magnification, say through an electron
microscope, we will see complicated features such as crystals, molecules and atoms appearing (and
fading) at different scales. The detailed description of particle-interaction forces acting at such tiny
scales is not only impractical but unnecessary for structural design and analysis. To make the idea
tractable the body is viewed as a continuum of points in the mathematical sense, and a stress state
is defined at each point by a force-over-area limit process.
Mechanical stress in a solid generalizes the simpler concept of pressure in a fluid. A fluid in
static equilibrium (the so-called hydrostatic equilibrium in the case of a liquid) can support only
a pressure state. (In a gas pressure may be tensile or compressive whereas in a liquid it must be
compressive.) A solid body in static equilibrium can support a more general state of stress, which
includes both normal and shear components. This generalization is of major interest to structural
engineers because structures, for obvious reasons, are fabricated with solid materials.

1.3. Mechanical Stress: Definition

This section goes over the process by which the stress state at an arbitrary point of a solid body is
defined. The key ideas are based on Newtons third law of motion. Once the physical scenario is
properly set up, mathematics takes over.

1.3.1. What Does Stress Measure?

Mechanical stress measures the average intensity level of internal forces in a material (solid or fluid
body) idealized as a mathematical continuum. The physical dimension of stress is

Force per unit area, e.g. N/mm2 (MPa) or lbs/sq-in (psi) (1.1)

This measure is convenient to assess the resistance of a material to permanent deformation (yield,
creep, slip) as well as rupture (fracture, cracking). Comparing working and failure stress levels
allows engineers to establish strength safety factors for structures. This topic (safety factor) is
further covered in Lecture 2.
Stresses may vary from point to point in a body. In the following we consider a arbitrary solid
body (which may be a structure) in a three dimensional (3D) setting. As previouly noted, this is
actually its idealization as a mathematical continuum. Thus the term body actually refers to that

Lecture 1: STRESS IN 3D

cut by plane

ABCD and B
discard blue
portion A

Applied Applied
loads Discard


Keep plan C

Figure 1.1. Cutting a 3D body through a plane passing by point Q.

1.3.2. Cutting a Body

The 3D solid body depicted on the left of Figure 1.1 is asumed to be in static equilibrium under the
applied loads acting on it. We want to find the state of stress at an arbitrary point Q. This point will
generally be located inside the body, but it could also lie on its surface.
Cut the body by a plane ABCD that passes through Q (How to orient the plane is discussed in the
next subsection.) The body is divided into two. Retain one portion (red in figure) and discard the
other (blue in figure), as shown on the right of Figure 1.1.
To restore equilibrium, however, we must replace the discarded portion by the internal forces it
had exerted on the kept portion. This is a consequence of Newtons third law: action and reaction.
If you are rusty on that topic please skim the Addendum at the end of this Lecture.

B Unit vector n is the exterior

normal at Q, meaning it points

A outward from the kept body.

Applied In the figure n has been

loads chosen to be parallel to x
Kept Q
body n //x
Reference frame
is a RCC system
of axes
z x Cut

Figure 1.2. Orienting the cut plane ABCD by its exterior unit normal vector n

1.3.3. Orienting the Cut Plane

The cut plane ABCD is oriented by its unit normal direction vector n  , or normal for short. See
Figure 1.2. By convention we will draw n as emerging from Q and pointing outward from the kept
body, as shown in Figure 1.2. This direction identifies the so-called exterior normal.


At this point we refer the body to a Rectangular Cartesian Coordinate (RCC) system of axes {x, y, z}.
This reference frame obeys the right-hand orientation rule. The coordinates of point Q, denoted by
{x Q , y Q , z Q }, are called its position coordinates or position components. The position vector of Q
is denoted by  
x Q = y Q (1.2)
but we will not need to use this vector here.
With respect to {x, y, z} the normal vector has components
 = ny
n (1.3)
 with respect to {x, y, z}, respectively. Since n
in which {n x , n y , n z } are the direction cosines of n 
is a unit vector, those components must verify the unit length condition
n 2x + n 2y + n 2z = 1. (1.4)
In Figure 1.2, the cut plane ABCD has been chosen with its exterior normal parallel to the +x axis.
Consequently (1.3) reduces to
= 0 .
n (1.5)
1.3.4. Internal Forces on Elemental Area
Recall that the action of the discarded (blue) portion of the body on the kept (red) portion is replaced
by a system of internal forces that restores static equilibrium. This replacement is illustrated on
the right of Figure 1.3. Those internal forces generally will form a system of distributed forces per
unit of area, which, being vectors, generally will vary in magnitude and direction as we move from
point to point of the cut plane, as pictured over there.
Next we focus our attention on point Q. Pick an elemental area A around Q that lies on the cut
plane. Call F the resultant of the internal forces that act on A. Draw that vector with origin at
Q, as pictured on the right of Figure 1.3. Do not forget to draw also the unit normal vector n
The use of the increment symbol  suggests a pass to the limit. And indeed this will be done in
equations (1.6) below, to define three stress components at Q.
1.3.5. Projecting the Internal Force Resultant
Zoom now on the elemental area about Q, omitting both the kept-body and applied loads for clarity,
as pictured on the left of Figure 1.4.
Project the internal force resultant F on the reference axes {x, y, z}. This produces three compo-
nents: Fx , Fy and Fz , as shown on the right of Figure 1.4.
Component Fx is aligned with the cut-plane normal, because n  has been taken to be parallel
to x. This is called the normal internal force component or simply normal force. On the other
hand, components Fy and Fz lie on the cut plane. These are called tangential internal force
components or simply tangential forces.

;; ;;
Lecture 1: STRESS IN 3D


;; ;;
Applied forces Applied
loads loads F

Q Q n

;; ;;
y y

z x x
Arrows are placed over F and n
to remind you that they are vectors

 over elemental area about point Q.

Figure 1.3. Internal force resultant F

Project internal force

y resultant F on
axes {x,y,z } to
x F get its normal
z and tangential
Q n //x
Q n
Figure 1.4. Projecting the internal force resultant onto normal and tangential components.

1.3.6. Defining Three Stress Components

We define the three x-stress components at point Q by taking the limits of internal-force-over-
elemental-area ratios as that area shrinks to zero:

def Fx
x x = lim normal stress component
A0 A
def Fy
x y = lim shear stress component (1.6)
A0 A
def Fz
x z = lim shear stress component
A0 A

x x is called a normal stress, whereas x y and x z are shear stresses. See Figure 1.5.
Remark 1.1. We tacitly assumed that the limits (1.6) exist and are finite. This assumption is part of the axioms
of continuum mechanics.

Remark 1.2. The use of two different letters: and for normal and shear stresses, respectively, is traditional
in American undergraduate education. It follows the influential textbooks by Timoshenko (a key contributor
to the development of Enginering Mechanics education in the US) that appeared in the 1930-40s. The main
reason for carrying along two symbols instead of one was to emphasize their distinct physical effects on
structural materials. It has the disadvantage, however, of poor fit with the tensorial formulation used in more
advanced (graduate level) courses in continuum mechanics. In those courses a more unified notation, such as
i j or i j for all stress components, is used.


y Fy
x (reproduced from
z Q n //x previous figure for
Fx convenience)
Fz A

Fx Fy Fz
xx def lim
A 0 A
xy =
A 0 A
xz =
A 0 A

Figure 1.5. Defining stress components x x , x y and x z as force-over-area mathematical


1.3.7. Six More Components

Are we done? No. It turns out we need nine stress components in 3D to fully characterize the stress
state at a point. So far we got only three. Six more are obtained by repeating the procedure: find
internal force resultant, project onto axes, divide by elemental area and take-the-limit, using two
more cut planes. The obvious choice is to pick planes normal to the other two axes: y and z.
 parallel to y and going through the motions we get three more components, one normal
Taking n
and two shear:
yy , yx , yz . (1.7)

 parallel to z we get three more compo-

These are called the y-stress components. Finally, taking n
nents, one normal and two shear:

zz , zx , zy . (1.8)

These are called the z-stress components. On grouping (1.6) , (1.7) and (1.8) we arrive at a total of
nine components, as required for full characterization of the stress at a point. Now we are done.
1.3.8. Visualization on Stress Cube
The foregoing nine stress components may be conveniently visualized on a stress cube as follows.
Cut an infinitesimal cube about Q with sides parallel to the RCC axes {x, y, z}, and dimensioned
d x, dy and dz, respectively. Draw the components on the positive cube faces (positive face is
defined below) as depicted in Figure 1.6
The three positive cube faces are those with exterior (outward) normals aligned with +x, +y and +z,
respectively. Positive (+) values for stress components on those faces are as drawn in Figure 1.6.
More on sign conventions later.
1.3.9. What Happens on the Negative Faces?
The stress cube has three positive (+) faces. The three opposite ones are negative () faces.
Outward normals at faces point along x, y and z, respectively. What do stresses on those
faces look like? To maintain static equilibrium, stress components must be reversed.

Lecture 1: STRESS IN 3D

Note that stresses are

yy forces per unit area, not
Point Q (not pictured) forces, although they look
is inside cube, at its center +y face like forces in the picture.
dy yz xy
Strictly speaking, this is
x zy +x face
a "cube" only if dx=dy=dz,
z +z face else it should be called a
zx xz
zz dz xx parallelepided; but that is
dx difficult to pronounce.
(The correct math term is
``rectangular prism'')

Figure 1.6. Visualizing stresses on faces of stress cube aligned with reference frame axes.
Stress components are positive as drawn.

For example, a positive x x points along the +x direction on the +x face, but along x on the
opposite x face. A positive x y points along +y on the +x face but along y on the x face.
To better visualize stress reversal, it is convenient to project the stress cube onto the {x, y} plane by
looking at it from the +z direction. The resulting 2D diagram, shown in Figure 1.7, clearly displays
the rule given above.

project onto {x,y} +y face yx
y yx looking along -z
x face xy
yz xy
zy xx xx
z x zx xz
xx xy +x face
zz y yx

yy y face

Figure 1.7. Projecting onto {x, y} to display component stress reversals on going from + to faces.

1.4. Notational Conventions

1.4.1. Sign and Subscripting Conventions

Stress component sign conventions are as follows.
For a normal stress: positive (negative) if it produces tension (compression) in the material.
For a shear stress: positive if, when acting on the + face identified by the first index, it points
in the + direction identified by the second index. Example: x y is + if on the +x face it points
in the +y direction.
These conventions are illustrated for x x and x y on the left of Figure 1.8.


y xy xy
+x face
stress acts on cut
stress points in
x xx plane with n along
the y direction
z +x (the +x face)

Both xx and xy are +

as drawn above
Figure 1.8. Sign and subscripting conventions for stress components.

Shear stress components have two different subscript indices. The first one identifies the cut plane
on which it acts as defined by the unit exterior normal to that plane. The second index identifies
component direction. That convention is illustrated for the shear stress component x y on the right
of Figure 1.8.

Remark 1.3. The foregoing subscripting convention plainly applies also to normal stresses, in which case the
direction of the cut plane and the direction of the component merge. Because of this coalescence some authors
(for instance, Beer, Johnston and DeWolf in their Mechanics of Materials book) drop one of the subscripts
and denote x x , say, simply by x .

Remark 1.4. The sign of a normal component is physically meaningful since some structural materials, for
example concrete, respond differently to tension and compression. On the other hand, the sign of a shear stress
has no physical meaning; it is entirely conventional.

1.4.2. Matrix Representation of Stress

The nine components of stress referred to the x,y,z axes may be arranged as a 3 x 3 matrix, which
is configured as
x x x y x z
yx yy yz (1.9)
zx zy zz
Note that normal stresses are placed in the diagonal of this square matrix.
We will call this a 3D stress matrix, although in more advanced courses this is the representation
of a second-order tensor called as may be expected the stress tensor.
1.4.3. Shear Stress Reciprocity

From moment equilibrium conditions on the infinitesimal stress cube it may be shown that

x y = yx , x z = zx , x y = zy , (1.10)

in magnitude. In other words: switching shear stress indices does not change its value. Note,
however, that index-switched shear stresses point in different directions: x y , say, points along y
whereas yx points along x.
For a proof of (1.10) see, for example, pp. 2627 of Beer-Johnston-DeWolf 5th ed.

Lecture 1: STRESS IN 3D

Property (1.10) is known as shear stress reciprocity. It follows that the stress matrix is symmetric:
x x x y x z x x x y x z
yx = x y yy yz yy yz (1.11)
zx = x z zy = yz zz symm zz
Consequently the 3D stress state depends on only six (6) independent components: three normal
stresses and three shear stresses.

1.5. Simplifications: 2D and 1D Stress States

For certain structural configurations such as thin plates, all stress components with a z subscript
may be considered negligible, and set to zero. The stress matrix of (1.9) becomes
x x x y 0
yx yy 0 (1.12)
0 0 0
This two-dimensional simplification is called a plane stress state. Since x y = yx , plane stress is
fully characterized by just three independent stress components: the two normal stresses x x and
yy , and the shear stress x y .
A further simplification occurs in structures such as bars or beams, in which all stress components
except x x may be considered negligible and set to zero, whence the stress matrix reduces to
x x 0 0
0 0 0 (1.13)
0 0 0
This is called a one-dimensional stress state. There is only one independent stress component left:
the normal stress x x .

1.6. Advanced Topics

We mention a couple of advanced topics that either fall outside the scope of the course, or will be
later covered for special cases.
1.6.1. Changing Coordinate Axes
Suppose we change axes {x, y, z} to another set {x , y , z } that also forms a RCC system. The
stress cube centered at Q is rotated to realign with {x , y , z } as illustrated in Figure 1.9. The stress
components change accordingly, as compactly shown in matrix form:
x x x y x z x x x y x z

yx yy yz becomes yx yy yz (1.14)

zx zy zz zx zy zz
Can the primed components be expressed in terms of the original ones? The answer is yes. All
primed stress components can be expressed in terms of the unprimed ones and of the direction
cosines of {x , y , z } with respect to {x, y, z}. This operation is called a stress transformation.


z x
yy z'
y 'yx 'xy
y' yz yx 'xx
xy 'yz
x' dy zy 'xz dz'
xz 'zy
z x zx 'zx
zz dz xx dy'
z' dx dx'
Note that Q is inside both cubes; they
are drawn offset for clarity 'zz

Figure 1.9. Transforming stress components in 3D when reference axes change.

For a general 3D state this operation is complicated because there are three direction cosines. In
this introductory course we will cover only transformations for the 2D plane stress state. The
transformations are simpler (and more explicit) since changing axes in 2D depends on only one
direction cosine or, equivalently, the rotation angle about the z axis.
Why bother to look at stress transformations? One important reason: material failure may depend on
the maximum normal tensile stress (for brittle materials) or the maximum absolute shear stress (for
ductile materials). To find those we generally have to look at parametric rotations of the coordinate
system, as in the skew-cut bar example studied in the first Recitation. Once such dangerous stress
maxima are found for critical points of a given structure, the engineer can determine strength safety
1.6.2. A Word on Tensors
The state of stress at a point is not a scalar or a vector. It is a more complicated mathematical object,
called a tensor (more precisely, a second-order tensor). Tensors are not covered in undergraduate
engineering courses as mathematical entities. Accordingly we will deal with stresses (and strains,
which are also tensors) using a physical approach complemented with recipes. Nonetheless for
those of you interested in the deeper mathematical aspect before reaching graduate school, here is
a short list that extrapolates tensors from two entities you already (should have) encountered in
Calculus and Physics courses.
Scalars are defined by magnitude. Examples: temperature, pressure, density, charge.
Vectors are defined by magnitude and direction. Examples: force, displacement, velocity,
acceleration, electric current.
Second-order tensors are defined by magnitude and two directions. Examples: stress, strain,
electromagnetic field strength, space curvature in GRT.
For mechanical stress, the two directions are: the orientation of the cut plane (as defined by its
outward normal) and the orientation of the internal force resultant.
All three kinds of objects (scalars, vectors, and second-order tensors) may vary from point to point.
This means that they are expressed as functions of the position coordinates. In mathematical physics
such functions are called fields.

Lecture 1: STRESS IN 3D

Remove the "bridge" (log) and

replace its effect on the elephant by

reaction forces on the legs. The elephant
stays happy: nothing happens.

;;;;;; Strictly speaking, reaction forces are
distributed over the elephant leg
contact areas. They are replaced
above by equivalent point forces,
a.k.a. resultants, for visualization convenience

Figure 1.10. How to keep a bridge-crossing elephant happy in an emergency.

1.7. Addendum: Action vs. Reaction Reminder

This is a digression about Newtons Third Law of motion, which you should have encountered in
Physics 1. See cartoon in Figure 1.10 for a reminder of how to replace the effect of a removed
physical body by that of the reaction forces acting on the kept body.

Average Stresses &
Component Design



2.1 Introduction . . . . . . . . . . . . . . . . . . . . . 23
2.2 Average Stress And Its Uses . . . . . . . . . . . . . . . 23
2.3 Safety Factor . . . . . . . . . . . . . . . . . . . . . 23
2.3.1 Design Related Definitions . . . . . . . . . . . . 23
2.3.2 Critical Safety Factor . . . . . . . . . . . . . . . 23
2.3.3 Why is Component Design Important? . . . . . . . . 25
2.3.4 Simplified Strength Design . . . . . . . . . . . . . 25
2.4 Ex.1: Axially Loaded Brittle Bar . . . . . . . . . . . . . 25
2.5 Ex.2: Axially Loaded Ductile Bar . . . . . . . . . . . . . 26
2.6 Ex. 3: Bar with Pin Connectors, Failure by Tension Break at Pin . 27
2.7 Ex. 4: Bar with Pin Connectors, Failure by Pin Shear O . . . . 28
2.8 Ex. 5: Truss Bolt Connection With Single Shear Area . . . . . 28
2.9 Ex. 6: Truss Bolt Connection With Double Shear Area . . . . . 210


2.1. Introduction

This lecture describes a highly simplified form of the 3D stress concept, in which the objective is
to get average stresses over a finite area instead of point stresses. This is applied to some simple
problems in component design. The concept of strength safety factor is discussed in some detail,
first in general terms and then in conjunction with examples.

2.2. Average Stress And Its Uses

The previous lecture emphasized that stresses vary from point to point inside a body (structure).
Thus x x = x x (x, y, z), and likewise for all other components. Mathematically, stresses form a
tensor field within the body.
Although the field viewpoint is rigorous, it may be too elaborate for practical use if its determination
requires expensive and time-consuming computations, even with the help of the computer. Often the
time-pressed engineer compromises by using a simplified stress analysis process based on average
stresses. The key idea is not to pass to the limit as done in Lecture 1 to define stresses at a point.
After appropriately cutting the body to manifest internal force(s) of interest, divide those resultants
by the finite area on which they act.
This shortcut is especially useful in preliminary design of 1D structural components when internal
force resultants can be quickly determined from statics by a Free Body Diagram (FBD). Several
illustrative examples are given in Section 2.4 and following ones.

2.3. Safety Factor

A safety factor characterizes the margin for bad things to happen to the structure. Some actual
instances of such events are shown in Figure 2.1.
2.3.1. Design Related Definitions
To quantity the foregoing statement, two definitions are introduced.
Failure mode. The structure experiences something that renders it disfunctional. Mode
identifies what. Failure may be catastrophic: a bridge or dam collapses, an airplane wing
breaks in flight, a submarine hull buckles under pressure. Or something less drastic: a building
foundation settles and cracks appear.
Design Load. A load system used for designing the structure and its components. For real
structures there may be many such systems, identified as design load cases. For example, the
design of a commercial jet aircraft may involve hundreds of load cases. Design scenarios may
include normal events while in service, as well as abnormal ones that the structure is supposed
to survive without catastrophe (e.g., total collapse) when there are lives at stake. Examples:
earthquakes, hurricanes, emergency landings, ...
2.3.2. Critical Safety Factor
Assume we have a design load case with forces P1 , P2 , . . ., and that under the given load values the
structure is fine. (Load cases may include applied moments, distributed loads, temperature changes,
etc., but for simplicity think for now of point forces.) Suppose that the loads are proportionally


Figure 2.1. Bad things can happen.

Figure 2.2. The Challenger disaster.

increased to s P1 , s P2 . . . , where s > 1 is a magnification factor, and that the first failure mode
triggered by this increase happens at s F . This s F is called the safety factor for that load case with
respect to that failure mode. If, as generally happens, there are several load cases, the process is
repeated for each one. The smallest s F encountered in this sweep is the critical safety factor, which
should meet design specifications. Obviously it should not be less than a certain target.
The foregoing description oversimplifies actual practice. For example, some loads, such as own
weight, should be kept fixed. Furthermore, statistical and cost considerations come into the picture.
Not all load cases may be equally probable (how often is an aircraft hit by a meteorite?), while the
cost of achieving adequate safety against all possible events may be prohibitive.


A detailed safety analysis that includes statistical and cost data generally requires the use of so-
phisticated computer programs. For preliminary component sizing, however, a shortcut based on
back-of-the-envelope stress analysis may be sufficient.
2.3.3. Why is Component Design Important?
The failure of a component may trigger that of the whole structure (For want of a nail the shoe was
lost ...) For instance, the well known 1986 disaster of the Challenger space shuttle, which cost $6.7
billions in 1971 dollars to build, was triggered by a malfunctioning $25 O-ring. See Figure 2.2.
2.3.4. Simplified Strength Design
A simplified strength design based on average stresses and given safety factor proceeds as follows.
A structural component such as a bar or beam is subject to known loads that come from an analysis
of the whole structure. A failure mode of the isolated component is assumed. In the ensuing
examples, that mode will represent a failure of the material when a stress level is reached.
But which stress? Here a distinction should be made between brittle and ductile materials. Brittle
materials such as fiberglass or cast iron fail by the maximum tensile normal stress reaching a f ail
level. Ductile materials such as metals or alloys fail by the maximum absolute shear stress reaching
a f ail level. Using FBD statics compute the average normal stresses avg or the average shear
stress avg over the area of an appropriate cut affected by the failure mode. The strength safety
factor is the ratio
f ail f ail
sF = or s F = (2.1)
avg avg

as appropriate to the material type and failure mode. More complicated failure criteria do exist,
but will not be covered in this course; they are studied in ASEN 4012. For strength design s F is
picked by practice, design codes, or experience, and inserted into (2.1), in which either avg or
avg is expressed in terms of the data. Design variables are solved for, and the component sized
Several examples of this procedure follow.

2.4. Ex.1: Axially Loaded Brittle Bar

P = 30 kips P P F n
x x

cut normal
Circular cross section d=? to bar axis
of diameter d & area A

Figure 2.3. Axially loaded brittle bar. Failure by tension: normal cross section breaks..

The first design example is pictured in Figure 2.3. A bar of solid circular cross section, fabricated
with brittle material, is subject to axial load P. Recall that a brittle material such as cast iron or


fiberglass fails (breaks apart) when the maximum normal stress (here the axial stress) reaches a
limit value.
Material data: cast iron, which fails by normal stress reaching f ail = 40 ksi (kilopounds/sqin).
Failure mode: cross section breaks when f ail is reached
Safety Factor: s F = 8 against failure.
Load case data: axial load P = 30 kips (kilopounds)
Find: design variable is bar diameter d in inches.
Solution. Make the normal cut shown on the right of Figure 2.3. From FBD statics, the resultant
internal force is F = P, aligned with +x. The area of the cut, a.k.a. resisting normal-stress area
or simply resisting area, is the bar cross section A = 14 d 2 . The average normal stress is

F P 4P
avg = = 1 = . (2.2)
A 4
d 2 d 2

The design condition is avg f ail /s F . Substituting for avg and solving for d yields
4P s F 4 30 kips 8
d+ = = 2.76 in (2.3)
f ail 40 kips/in2

2.5. Ex.2: Axially Loaded Ductile Bar

P = 30 kips P P o
x x 45o

skew cut at
Circular section of d=? 45o from x
diameter d & area A t

Figure 2.4. Axially loaded ductile bar: failure by yield due to crystal slip at 45 .

The next example is pictured in Figure 2.4. The configuration is identical to the previous example:
a bar of solid circular cross section under axial loads, but the material is now ductile. Failure occurs
by yield when the maximum shear stress reaches a limit value f ail .
Material data: ductile Al alloy, which fails by maximum shear at f ail = 20 ksi (kilopounds/sqin).
Failure mode: crystal slip at 45 from longitudinal bar x axis. Why 45 ? This will be the subject
of one problem in Recitation 1.
Safety Factor: s F = 5 against shear failure.
Load case data: axial load P = 30 kips = 30, 000 lbs as in last example.
Find: design variable is again diameter d in inches.


Solution. Make the skew cut shown on the right of Figure 2.4. From FBD statics,the resultant

internal force is F = P, aligned with +x. Project on cut plane: Ft = F cos 45
= F/ 2 = P/ 2.
The area of the skew cut, a.k.a. shear area, is As = A/ cos 45 = A 2 = 14 2 d 2 . The average
shear stress is
Ft P/ 2 P/ 2 P
avg = = = 1 = 1 . (2.4)
As As 4
2d 2
The design condition is avg f ail /s F . Solving for d:
2P s F 2 30 kips 5
d+ = = 2.19 in (2.5)
f ail 20 kips/in2

2.6. Ex. 3: Bar with Pin Connectors, Failure by Tension Break at Pin

zoom diameter

; ; ;; ;
P/2 P/2 is dpin

t P/2 P/2

;; h =? cross cut plane


Figure 2.5. Axially loaded ductile bar with pin connectors: failure by tension break at pin

This example is pictured in Figure 2.5. A ductile bar member (Al alloy) of rectangular cross section
h t is linked to other bar members though pin connectors. The bar transmits a tensile axial force
P. Two failure modes at the pin are possible: tension break at minimal cross section area weakened
by hole, or pin shear-off. This example considers the first possibility.
Material data: Al alloy, may fail by either maximum tensile normal stress at f ail = 268 MPa
(megaPascal) or by shear at f ail = 165 MPa.
Failure mode: tension break at minimal cross section weakened by pin hole; see Figure 2.5.
Safety Factor: s F = 6 against tension break at minimum cross section.
Load case data: axial load P = 5 kN = 5, 000 N applied to pins
Geometric data: cross section thickness t = 5 mm; pin diameter d pin = 12 mm.
Find: design variable is cross section height h in mm.
Solution. Make a transverse plane cut at the pin center as shown. The bar cross section at the
cut (obviously the minimum one) is called the corrected area Acorr = A A pin = h t d pin t.


Average normal stress over corrected area is avg = P/Acorr = P/((h d pin ) t). Design condition:
avg f ail /s F . Solving for h:

P sF 5000 N 6
h + d pin = + 12 mm = 34.4 mm (2.6)
f ail t 268 N/mm2 5 mm

2.7. Ex. 4: Bar with Pin Connectors, Failure by Pin Shear Off

;; ;; ;;
zoom ds ds =?

P/2 P
P P t P/2
cut planes pin shears off
h cross

Figure 2.6. Axially loaded ductile bar with pinhole connectors: failure by pin shearing off.

The structural configuration for this example is the same as in previous one. We now consider the
other possible failure mode, in which a pin shears off as depicted in Figure 2.6.
Material data: same as in previous example: fails by maximum tension stress at f ail = 268 MPa
or by shear at f ail = 165 MPa.
Failure mode: shear failure with a pin shearing off, see right of Figure 2.6.
Safety Factor: s F = 6 against shear failure.
Load case data: axial load P = 5 k-N = 5,000 N applied to pins, as before
Geometric data: cross section thickness t = 5 mm, pin diameter not needed.
Find: design variable is distance ds defined in the figure, expressed in mm. (Actually h and d pin
are not needed in this case.)
Solution. Make two cuts because the force P is transmitted through the two interfaces shown
in the figure. Total shear area is As = 2ds t. Resultant shear force over shear area, from FBD:
Fs = 12 P + 12 P = P. Average shear stress avg = Fs /As = P/As = P/(2ds t). Design condition:
avg f ail /s F . Solving for ds :

P sF 5,000 N 6
ds = = 18.2 mm (2.7)
2 f ail t 2 165 N/mm2 5 mm



Figure 2.7. Bolt connection with single shear area: overall sketch.

(a) (b) (c) (d)

P t
P P a
P P d bolt
P a
cut plane Fs = P avg

Figure 2.8. Bolt connection with single shear area: FBD procedural details.

2.8. Ex. 5: Truss Bolt Connection With Single Shear Area

This example deals with the design of the bolt for connecting between two truss members as shown
in Figure 2.7, using the average shear stress approach. The axial force transmitted through the bolt
is P.
Figure 2.8(a)(d) shows a set of FBDs to help understand the concept of a shear area through
which the total axial force P is transmitted (flows) from one member to the other. The cut plane
is displayed in (c). The resultant tangential force on that plane is called Fs for shear force. From
axial equilibrium, Fs = P. The shear area is colored red in (d).
Remark 2.1. It should be noted that the actual stress distribution in the vicinity of the shear area may be quite
complicated because of high stress gradients near contact zones as well as bending due to member centerline
eccentricity (more about this later). To get a realistic picture of that distribution one would need to carry out
a time-consuming 3D finite element analysis. By assuming that in the place of the bolt section a uniformly
distributed shear stress develops, however, a simple solution is readily found. Approximations involved in this
assumption are supposed to be covered by the safety factor.
Material data: metal bolt (ductile); fails by maximum shear at f ail .
Failure mode: average shear stress over shear area, colored red in (d), reaches f ail .
Safety factor: s F would typically be 4 or more to compensate for ignoring stress concentrations as
well as transfer eccentricity effects. In the derivation below s F is kept symbolic.
Load case data: axial load P to be transmitted by bolt.
Find: design variable is bolt diameter d
Solution: make the cut shown in (c) and assume that the shear stress over the shear area As =
Abolt = d 2 /4 is uniform, as discussed above. That average is given by avg = Fs /As = P/Abolt =
4P/( dbolt ). The design condition is avg f ail /s F . Solving for the bolt diameter gives

4P s F
dbolt + . (2.8)
f ail


It only remains to plug numbers.

2.9. Ex. 6: Truss Bolt Connection With Double Shear Area

(a) (b) (c) (d)

P/2 P/2
P P/2 P
P/2 P t1 P d bolt
t1 cut planes P/2 avg

Figure 2.9. Bolt connection with double shear area.

One flaw of the bolt connector studied in the foregoing example is load eccentricity caused by
member centerline offset. The resulting moment may cause additional bending stresses as well as
alignment problems. A better design that avoids that problem is shown in Figure 2.9, in which
configurational symmetry aligns inter member load transmission whence bending is eliminated.
To get the effective shear area, make two cuts as pictured in Figure 2.9(c). The effective shear
area doubles to As = 2Abolt = dbolt2
/2 and the average shear becomes avg = P/(2Abolt ) =
2P/( dbolt ). With the same design criteria of the previous example, the bolt diameter is now given


2P s F
dbolt + . (2.9)
f ail

If P, s F and f ail stay the same, this dbolt is about 30% smaller than that given by (2.8) Furthermore
s F could be cut since the lack of bending reduces stress distribution uncertainties. On the other hand,
this kind of connection is likely to be costlier to fabricate. Figure 2.10 shows a well fabricated
instance of a double shear area connector. This would be typical, for example, of a quality car
suspection device.

Figure 2.10. A well fabricated connector with double shear

area. Note that axial loads are nicely aligned.

Thin Walled
Pressure Vessels



3.1. Introduction 33
3.2. Pressure Vessels 33
3.3. Assumptions 34
3.4. Cylindrical Vessels 34
3.4.1. Stress Assumptions . . . . . . . . . . . . . . . 34
3.4.2. Free Body Diagrams . . . . . . . . . . . . . . 35
3.5. Spherical Pressure Vessel 36
3.5.1. Stress Assumptions . . . . . . . . . . . . . . . 36
3.5.2. Free Body Diagram . . . . . . . . . . . . . . . 37
3.6. Remarks on Pressure Vessel Design 37
3.7. Numerical Example 38
3.7.1. CExample: Cylindrical Tank With Bolted Lids . . . . . . 38


3.1. Introduction
This Lecture continues with the theme of the last one: using average stresses instead of point
stresses to quickly get results useful in preliminary design or in component design. We look at more
complicated structural configurations: thin wall pressure vessels, which despite their apparently
higher complexity can be treated directly by statics if both geometry and loading are sufficiently
The main difference with respect to the component configurations treated in the previous lecture is
that the state of stress in the vessel wall is two dimensional. More specifically: plane stress. As
such they will provide examples for 2D stress-displacement analysis once 2D strains and multidi-
mensional material laws are introduced in Lectures 45.

3.2. Pressure Vessels

(a) (b)

Figure 3.1. Pressure vessels used for fluid storage: (a) spherical tanks, (b) cylindrical tank.

Thin wall pressure vessels (TWPV) are widely used in industry for storage and transportation of
liquids and gases when configured as tanks. See Figure 3.1. They also appear as components of
aerospace and marine vehicles such as rocket and balloon skins and submarine hulls (although in
the latter case the vessel is externally pressurized, violating one of the assumptions listed below).
Two geometries will be examined in this lecture:
Cylindrical pressure vessels.
Spherical pressure vessels.
The walls of an ideal thin-wall pressure vessel act as a membrane (that is, they are unaffected by
bending stresses over most of their extent). A sphere is the optimal geometry for a closed pressure
vessel in the sense of being the most structurally efficient shape. A cylindrical vessel is somewhat
less efficient for two reasons: (1) the wall stresses vary with direction, (2) closure by end caps can
alter significantly the ideal membrane state, requiring additional local reinforcements. However
the cylindrical shape may be more convenient to fabricate and transport.


3.3. Assumptions

The key assumptions used here are: wall thinness and geometric symmetries. These make possible
to obtain average wall stresses analysis with simple free-body diagrams (FBD). Here is a more
detailed list of assumptions:
1. Wall Thinness. The wall is assumed to be very thin compared to the other dimensions of the
vessel. If the thickness is t and a characteristic dimension is R (for example, the radius of the
cylinder or sphere) we assume that

t/R << 1, or R/t >> 1 (3.1)

Usually R/t > 10. As a result, we may assume that the stresses are uniform across the wall.
2. Symmetries. In cylindrical vessels, the geometry and the loading are cylindrically symmetric.
Consequently the stresses may be assumed to be independent of the angular coordinate of
the cylindrically coordinate system. In spherical vessels, the geometry and the loading are
spherically symmetric. Therefore the stresses may be assumed to be independent of the two
angular coordinates of the spherical coordinate system and in fact are the same in all directions.
3. Uniform Internal Pressure. The internal pressure, denoted by p, is uniform and everywhere
positive. If the vessel is also externally pressurized, for example subject to athmospheric
pressure, p is defined by subtracting the external pressure from the internal one, a difference
called gage pressure. If the external pressure is higher, as in the case of a submarine hull,
the stress formulas should be applied with extreme caution because another failure mode:
instability due to wall buckling, may come into play. See Section 3.5.
4. Ignoring End Effects. Features that may affect the symmetry assumptions are ignored. This
includes supports and cylinder end caps. The assumption is that disturbances of the basic
stress state are confined to local regions and may be ignored in basic design decision such as
picking up the thickness away from such regions.
We study the two simplest geometries next.

3.4. Cylindrical Vessels

We consider a cylindrical vessel of radius R, thickness t loaded by internal pressure p. We use the
cylindrical coordinate system (x, r , ) depeicted in Figure 3.2(a), in which
x axial coordinate
angular coordinate, positive as shown
r radial coordinate

3.4.1. Stress Assumptions

Cut the cylinder by two normal planes at x and x + d x, and then by two planes and + d as
shown in Figure 3.2(a). The resulting material element, shown in exploded view in Figure 3.2(b)
has six surfaces. The outer surface r = R is stress free. Thus

rr = r x = r = 0 at r = R (3.2)


Wall thickness t
Free surface: rr = 0, rx = 0, r = 0
rr inside body neglected since
it varies from p to 0 over wall,
r 2R which is << & << xx
x Zero because of thin
body and xr = rx
D xx : axial stress
r x Zero because of
C axisymmetry and
x no torque
Zero because
r = r : hoop (a.k.a. Zero No tangential forces
because because r = 0 & rx =0
circumferential) =
x x

Figure 3.2. Wall material element of a pressurized cylindrical vessel referred to

cylindrical coordinates. Note that thickness is grossly exaggerated for visibility.

On the inner surface r = R t there is a compressive normal stress that balances the applied
pressure but no tangential stresses. Thus

rr = p, r x = r = 0 at r = R t (3.3)

Since the wall is thin, we can confidently assume that

r x = r = 0 for all r [R t, R] (3.4)

whereas rr varies from p to zero. Later on we will find that rr is much smaller than the other
two normal stresses, and in fact may be neglected (set to zero). Because r x = zr and r = r
on account of shear stress reciprocity, we conclude that

zr = r = 0 for all r inside wall (3.5)

The normal stresses x x and zz are called axial stress and circumferential or hoop stress, respec-
tively. The last wall stress component is x = x , which is the wall shear stress. Because of
symmetry assumptions on the geometry and loading (no torque), this stress is zero. These stress
assumptions are graphically displayed, with annotations, in Figure 3.2.
Displaying the wall stress state using the stress matrix and taking the axes in order {x, , r } for
convenience, we have
x x x xr x x 0 0
x r = 0 0 . (3.6)
r x r rr 0 0 0

Comparing this to the 2D stress state introduced in Lecture 1, we observe that the cylinder vessel
wall is in plane stress.


(a) (b) (c)

xx t xx (2R t)
t dx

r ~2R
p 2R dx
p(R 2 )
t dx Both interior and exterior
vessel radii can be taken
as R, since t << R

Figure 3.3. Free body diagrams (FBD) to get the averaged hoop and longitinal wall
stresses in a pressurized thin-wall cylindrical vessel

3.4.2. Free Body Diagrams

The nonzero normal stress components x x and in (3.6) act as sketched in Figure 3.3(a). Both
components are uniform across the wall thickness and throughout the vessel (excluding possible
end effects, which are discussed in Section 3.5). To find their value in terms of the data: p, R, and
t, we use the two FBD drawn in Figure 3.3(b,c). Details will be worked out in class. The result is

pR pR
= , x x = = 1
. (3.7)
t 2t
Neither stress depends on position. Because the hoop stress is twice the axial stress, it will be the
controlling one in a strength design. For example if R/t = 100, which is a typical vessel thickness
in aerospace applications, then = 100 p and x x = 50 p. Since rr is of the order of p as
previously discussed, it follows that neglecting it is justified.
We can summarize our findings by showing the stress matrix now expressed in terms of the data:
x x 0 0 pR 1 0 0
0 0 = 0 2 0 . (3.8)
0 0 0 2t 0 0 0

3.5. Spherical Pressure Vessel

A similar approach can be used to derived an expression for an internally pressurized thin-wall
spherical vessel. We use spherical coordinates {r, , }, as illustrated in Figure 3.4(a).
3.5.1. Stress Assumptions
Reasoning as in the preceding case, we find that
1. All shear stresses are zero: r = r = 0, r = r = 0 and = = 0.
2. The normal stress rr varies from zero on the outside free surface to the negative of the pressure
p on the inside surface. Again we will neglect this value when compared to the other normal
stresses and justify this assumption a posteriori.


(a) z (b)

(2R t)

r y ~2R

p (R 2 )

Figure 3.4. Stress analysis of a spherical pressure vessel in spherical coordinates. Once
again thickness is grossly exaggerated for visibility.

3. The normal stresses and are equal and constant over the entire vessel. For simplicity
we will use the abbreviation = = .
For convenience in writing out the stress matrix we will order the axes as {, , r }. As per the
preceding discussion, the stresses at any wall point have the configuration
r 0 0
r = 0 0 (3.9)
r r rr 0 0 0

This shows again that the vessel wall is in a plane stress state.

3.5.2. Free Body Diagram

To find we cut the sphere into two hemispheres as shown in Figure 3.4(b). The FBD gives the
equilibrium condition 2 Rt = p R 2 , whence

= (3.10)

Any section that passes through the center of the sphere yields the same result.
We can summarize our findings by showing the stress matrix expressed in terms of the original
0 0 pR 1 0 0
0 0 = 0 1 0 (3.11)
0 0 0 2t 0 0 0

Comparing to (3.9) shows that for the same p, R and t the spherical geometry is twice as efficient
in terms of wall stress. Why? This is explained in the next section.


3.6. Remarks on Pressure Vessel Design

For comparable radius, wall thickness and internal pressure the maximum normal stress in a spherical pres-
sure vessel is one half as large as that in a cylindrical one. The reason can be understood by comparing
Figures 3.5(a,b). In the cylindrical vessel the internal pressure is resisted by the hoop stress in arch action
whereas the axial stress does not contribute. In the spherical vessel the double curvature means that all stress
directions around the pressure point contribute to resisting the pressure. The cylindrical geometry, however,
can result in more efficient assignment of container space as well as stacking and better aerodynamics: a
spherical rocket does not look quite right.
One important point for designers is: what happens at the ends of a cylindrical vessel? Suppose for instance
that a cylinder is closed by hemispherical end caps, as pictured in Figure 3.6(a). If the cylinder and the end
caps were allowed to deform independently of each other under pressurization they would tend to expand as
indicated by the dashed lined in that figure. The cylinder and the ends would in general expand by different
amounts. But since physical continuity of the wall must be maintained, the necessary adjustment in the
displacement would produce local bending as well as shear stresses in the vicinity of the juncture, as pictured
in Figure 3.6(b). If thick plates are used instead of relatively flexible hemispherical ends, those juncture
stresses would increase considerably as shown in Figure Figure 3.6(b). For this reason, the ends of cylindrical
pressure vessels must be designed carefully, and flat ends are should be avoided if possible.
Most pressure vessels are fabricated from curved metal sheets that are joined by welds. Two weld types: double
dillet lap joint and double welded butt joint with V grooves are shown in Figures 3.7(a) and (b), respectively.
Of these preference should be given to the latter as it avoids across-the-weld load transmission eccentricity.
It should be emphasized that the formulas derived for TWPV in this Lecture should be used only for cases
of internal pressure. (Or, more precisely, the internal pressure exceeds the external one). If a vessel is to be
designed for external pressure, as in the case of a submarine or vacuum tank, wall buckling, whether elastic or
inelastic, may well become the critical failure mode. Should that be the case, the previous wall stress formulas
are only part of the design.

3.7. Numerical Example

3.7.1. CExample: Cylindrical Tank With Bolted Lids

This is Example 4.18 of Vables book. It is reproduced here as it combines the results of this Lecture
with the bolt-design-by-average-stress technique described in Lecture 2.
Each lid is bolted to the tank of Figure 3.8(a) along the flanges using 1-in-diameter bolts. The tank
is made from sheet metal that is 12 in thick and can sustain a maximum hoop stress of 24 ksi in
tension. The normal stress in the bolts is to be limited to 60 ksi in tension. A manufacturer can
make tanks of diameters varying from 2 ft through 8 ft in steps of 1 ft. Develop a table that the
manufacturer can use to advise custometrs of the size of the tank and the number of bolts per lid
needed to hold a desired gas pressure.
Solution. The area of each bolt is Abolt = 14 (1 in)2 = 14 sq in. From the hoop stress equation
= p R/t we get

pR 12,000
= 1
24 ksi = 24,000 psi, whence p psi (3.12)
in R


(hoop, a.k.a

p xx (axial)

(a) Spherical vessel wall (b) Cylindrical vessel wall

Figure 3.5. Why spherical vessels are more structurally efficient.

0.20 mm
0.35 mm
A 0.15 mm (b) Detail A

0.35 mm
R = 1000 mm

(a) Deformed shape (c) Deformation of the same

cylindrical pressure vessel
at a flat head

Figure 3.6. End effects in cylindrical vessels.

(a) (b)
t 45


Figure 3.7. Welds in pressure vessels: two configurations.


Nbolt = bolt A bolt = bolt /4

Nlid = p R 2

Figure 3.8. Cylindrical tank with bolted lids for Example 3.2.

The FBD of the lid is shown in Figure 3.8(b). Force equilibrium in the x direction gives

4 p R2
n Nbolt = Nlid , bolt = 60,000. (3.13)
Substituting for p gives
4 12,000 R
60,000, or n 0.8 R. (3.14)
Rewriting these inequalities in terms of the diameter D = 2R of the tank we get
p , and n 0.4D, D in inches. (3.15)
We now tabulate the maximum pressure p and the number of bolts n in terms of D as we step from
D = 2 ft = 24 in through D = 8 ft = 96 in. The values of p are rounded up to the nearest integer
multiple of 5 whereas values of n are reported by rounding up to the nearest integer.

Table 3.1. Results for Example 4.18 of Vable

Tank Diameter (ft) Max Pressure (psi) Min # of Bolts

2 1000 10
3 665 15
4 500 20
5 400 24
6 330 30
7 280 34
8 250 39


Lecture 4: STRAINS


4.1 Introduction . . . . . . . . . . . . . . . . . . . . . 43
4.2 Strain: Classication . . . . . . . . . . . . . . . . . . 43
4.3 Axial (a.k.a. Normal or Extensional) Strains . . . . . . . . . 44
4.3.1 Average Strain in 1D . . . . . . . . . . . . . . 44
4.3.2 Point Strain in 1D . . . . . . . . . . . . . . . . 45
4.3.3 Strain Units . . . . . . . . . . . . . . . . . 45
4.3.4 Point Normal Strains in 3D . . . . . . . . . . . . . 46
4.3.5 Volumetric Strain . . . . . . . . . . . . . . . 46
4.4 Shear Strains . . . . . . . . . . . . . . . . . . . . . 47
4.4.1 Average Shear Strains . . . . . . . . . . . . . . 47
4.4.2 Connecting Average Shear Strain To Displacements . . . . . 47
4.4.3 Point Shear Strains in 3D . . . . . . . . . . . . . 49
4.5 Strain-Displacement Equations . . . . . . . . . . . . . . 49
4.6 Displacement Vector Composition . . . . . . . . . . . . . 411
4.6.1 Deflection of a Truss Structure . . . . . . . . . . . . 411
4.6.2 Forces And Displacements Obey Different Composition Rules 412


4.1. Introduction

All of the material covered in the first 3 lectures pertains to statics: applied forces induce internal
forces, which induce stresses:

applied forces internal forces stresses (4.1)

We now go beyond statics into kinematics. Stresses produce deformations because real materials
are not infinitely rigid. Deformations are measured by strains. Integration of strains through space
gives displacements, which measure motions of the particles of the body (structure). As a result
the body changes size and shape:

stresses strains displacements size & shape changes (4.2)

Conversely, if the displacements are given as data (as will happen with the Finite Element Method
covered in Part IV of this course), one can pass to strains by differentiation, from strains to stresses
using material laws such as Hookes law for elastic materials, and from stresses to internal forces:

displacements strains stresses internal forces (4.3)

This lecture focuses on the definition of strains and their connection to displacements. The relation
between strains and stresses, which is given by material properties codified into the so-called
constitutive equations, is studied in the next lecture.

4.2. Strain: Classification

In general terms, strain is a macroscopic measure of deformation. Truesdell and Toupin, in their
famous Classical Field Theories review article in Handbuch der Physik, introduce the concept as
The change in length and relative direction occasioned by deformation is called, loosely, strain.
[The term strain was introduced by W. J. M. Rankine in 1851.]
The concept is indeed loose until some additional qualifiers are called upon to render the matter
more specific.
1. Average vs. Point. Average strain is that taken over a finite portion of the body; for example
using a strain gage or rosette. Point strain is obtained by a limit process in which the
dimension(s) of the gaged portion is made to approach zero.
2. Normal vs. Shear. Normal strain measures changes in length along a specific direction. It is
also called extensional strain as well as dimensional strain. Shear strain measures changes
in angles with respect to two specific directions.
3. Mechanical vs. Thermal. Mechanical strain is produced by stresses. Thermal strains are
produced by temperature changes. (The latter are described in the next lecture.)
4. Finite vs. Infinitesimal. Finite strains are obtained using exact measures of the change in di-
mensions or angles. Infinitesimal strains are obtained by linearizing the finite strain measures
with respect to displacement gradients. On account of the nature of this process, infinitesimal

Lecture 4: STRAINS

(a) Undeformed Bar


undeformed length L0
(b) Deformed Bar

deformed length L = L0 +

Figure 4.1. Undeformed and deformed bar configurations to illustrate average axial
(a.k.a. normal, extensional) strain.

strains are also called linearized strains. The looser term small strains is also found in the
5. Strain Measures. For finite strains several mathematical measures are in use, often identified
with a person name in front. For example, Lagrangian strains, Eulerian strains, Hencky strains,
Almansi strains, Murnaghan strains, Biot strains, etc. They have one common feature: as
strains get small in the sense that their magnitude is << 1, they coalesce into the infinitesimal
(linearized) version. A brief discussion of Lagrangian versus Eulerian strains is provided in
4.3.1 below.

4.3. Axial (a.k.a. Normal or Extensional) Strains

4.3.1. Average Strain in 1D
Consider an unloaded bar of length L 0 aligned with the x axis, as shown in Figure 4.1(a). In the
literature this is called the undeformed, initial, reference, original or unstretched configuration. The
strains therein are conventionally taken to be zero.
The bar is then pulled by applying an axial force, as shown in Figure 4.1(b). (The undeformed and
deformed configurations are shown offset for visualization convenience; in fact both are centered
about the x axis.) In this new configuration, called deformed, final, current or stretched, its length
becomes L = L 0 + , where = L L 0 is the bar elongation. The average axial strain over the
whole bar is defined as
bar def L L 0
av = = (4.4)
Lre f Lre f
Here L r e f is the reference length selected for the strain computation. The two classical choices are
L r e f = L 0 for Lagrangian strains, and L r e f = L for Eulerian strains. The former is that commonly
used in solid mechanics and, by extension, structures. The latter is more popular in fluid mechanics
since a liquid or gas does not retain memory from previous configurations. Whatever the choice,
the strain is a dimensionless quantity: length divided by length.
If << L 0 , which leads to the linearized strain measure (also called infinitesimal strain or small
strain) the choice of L r e f makes little difference. For example, suppose L 0 = 10 in and = 0.01
in, which would be typical of a structural material. Using superscripts L and E for Lagrangian and
Eulerian, respectively, we find

= 0.01/10 = 0.10000%, av
= 0.01/10.01 = 0.09990%. (4.5)


(a) Undeformed Bar
P Q x

uP =u uQ = uP +(uQ uP) = u+u

(b) Deformed Bar
P' Q'
Figure 4.2. Undeformed and deformed bar configurations to illustrate point axial strain.

These agree to 3 places. In what follows we will consistently adopt the Lagrangian choice, which
as noted above is the most common one in solid and structural mechanics.

4.3.2. Point Strain in 1D

As in the case of stresses covered in Lecture 1, the strain at a point is obtained by a limit process.
Consider again the bar of Figure 4.1. In the undeformed configuration mark two coaxial points:
P and Q, separated by a small but finite distance x, as shown in Figure 4.2(a). (In experimental
determination of strains, this is called the gage length.) The bar is pulled and moves to the
deformed configuration illustrated in Figure 4.2(b). (Undeformed and deformed configurations are
again shown offset for visualization convenience.)
Points P and Q move to positions P  and Q  , respectively. The axial displacements are u P = u
and u Q = u P + (u Q u P ) = u + u, respectively. The strain at P is obtained by taking the limit
of the average strain over x as this distance tends to zero:

def (u + u) u u du
 P = lim = lim = . (4.6)
x0 x x0 x dx

This relation between displacement and point-strains is called a strain-displacement equation. It

allows one to compute strains directly by differentiation should the displacement variation be known,
for example from experimental measurements.
Anticipating the generalization to 3D in 4.3.4: the {x, y, z} components of the displacement of
P in a 3D body are denoted by u, v and w, respectively. These are actually functions of position,
meaning that u = u(x, y, z), v = v(x, y, z) and w = w(x, y, z), in which {x, y, z} are the position
coordinates of P in the undeformed configuration. Then (4.6) generalizes to

x x = . (4.7)

The generalization of (4.6) to (4.7) involves three changes:

Point label P is dropped since we can make that a generic (arbitrary) point
Subscript x x is appended to identify the strain component as per rules stated later
The ordinary derivative du/d x becomes a partial derivative.

Lecture 4: STRAINS

(b) Deformed configuration

(a) Undeformed configuration


y z+w
y x+u

x P z Displacement vector PP' has components

x u,v,w along x,y,z, respectively
Figure 4.3. Undeformed and deformed cube of material in 3D. Shear strains are zero so
angles are preserved. Change in cube dimensions grossly exaggerated for visibility.

4.3.3. Strain Units

As previously noted, strain is dimensionless but in structural mechanics it is often a very small
number compared to one. To reduce the number of zeros on the left one can express that number
in per-cent, per-mill or micro units. For example

x x = 0.000153 = 0.0153 % = 153 . (4.8)

Here is the symbol for micros; by definition 1 = 106 . Sometimes this is written mm/mm or
in/in, but the unit of length is usually unnecessary.
4.3.4. Point Normal Strains in 3D
Instead of the bar of Figure 4.1, consider now a small but finite cube of material aligned with
the {x, y, z} axes, as pictured in Figure 4.3(a). The cube has side dimensions x, y and z,
respectively, in the undeformed configuration.
The cube moves to a deformed configuration pictured in Figure 4.3(b). The displacement com-
ponents are denoted by u, v, and w, respectively. The deformed cube still remains a cube (more
precisely, shear strains are assumed to be zero everywhere so angles are preserved) but side lengths
change to x + u, y + v and z + w, respectively. Here u, v and w denote appropriate
displacement increments.
The averaged normal strain components are defined as

def u + u u u def v + v v v def w + w w w

x x,av = = ,  yy,av = = , zz,av = = .
x x y y z z
Point values, at corner P of the cube, are obtained by passing to the limit:

def u u def v v def w w

x x = lim = ,  yy = lim = , zz = lim = . (4.10)
x0 x x y0 y y z0 z z
Of course this process assumes that the indicated limits exist. Components may be tagged with the
point at which strains are computed if necessary or advisable. For example: x x,P , xPx , or x x (P).


(b) Deformed configuration

(a) Undeformed configuration

(1+yy) dy

y dy (1+zz ) dz
(1+xx) dx

x dx dz
Figure 4.4. Slight modification of previous figure to illustrate the concept of volumetric strain.

4.3.5. Volumetric Strain

Figure 4.4 is a slight modification of the previous one. Here the material cube has now infinitesimal
dimensions d x, dy and dz. Upon deformation these become d x + x x d x = (1 + x x ) d x, dy +
 yy dy = (1 +  yy ) dy, and dz + zz dz = (1 + zz ) dz. If the cube volume in the undeformed and
deformed configurations are denoted by d V0 and d V , respectively, the change in cube volume is
d V d V0 = [(1 + x x )(1 +  yy )(1 +  yy ) 1] d x d y dz =
(x x +  yy + zz ) d x d y dz.
in which the last simplification assumes that strains are infinitesimal; that is, x x << 1, etc. The
relative volume change is called the volumetric strain and is denoted by v :
d V d V0 (x x +  yy + zz ) d x d y dz
v = = = x x +  yy + zz . (4.12)
dV d x d y dz
It can be shown that this quantity is a strain invariant, which means that the value does not depend
on the choice of axes.
4.4. Shear Strains
Shear strains measure changes of angles as the material distorts in response to shear stress. To
define shear strains it is necessary to look at two directions that form the plane that undergoes shear
distortion. Therefore a one-dimensional view is insufficient to describe what happens. It takes two
to shear.
4.4.1. Average Shear Strains
Figure 4.5(a) shows a cube of material undergoing a pure shear deformation in the {x, y} plane. By
looking along z we can describe the process through the two-dimensional view of Figure 4.5(b,c).
Under the action of the shear stress x y = yx , the angle formed by PQ and PR, originally /2
radians, becomes /2 radians. This change is taken as the definition of the average shear strain
associated with directions x and y:
x y,av = . (4.13)
By convention x y,av is positive if the angle {P Q, P R} decreases; the motivation being to agree
in sign with a positive shear stress.

Lecture 4: STRAINS

(a) 3D view (b) 2D view of shearing (c) 2D shear deformation

in x-y plane (grossly exaggerated
for visibility)
yx = R'

y xy = y
P Q P' Q'
x measured in radians,
x positive as shown
Figure 4.5. Average shear strain in {x, y} plane.

C' Shear-Deformed
vD = 1+ 2
vC B'
C x D A'
Undeformed vA y
90 = 2
A uA
x uD
Figure 4.6. Computing average shear strain over rectangle ABCD from corner displacement data.

4.4.2. Connecting Average Shear Strain To Displacements

A rectangle ABCD of side lengths x and y aligned with x and y respectively, undergoes the shear
deformation depicted in Figure 4.6. The original rectangle becomes a parallelogram ABCD,
whose sides are not necessarily aligned with the axes. Goal: compute the average shear strain
x y,av = in terms of the displacements of the four corners.
Observe that = 1 + 2 , where 1 and 2 are the angles indicated in Figure 4.6, with positive
senses as shown. We have
v B  v A v B A u C  u A u C A
tan 1 = = , tan 2 = = . (4.14)
u B  u A x + u B A vC  v A y + vC A
We assume that strains are infinitesimal. Consequently 1 << 1 and 2 << 1, whence tan 1 1
and tan 2 2 . Likewise, u B A << x so x + u B A x in the denominator of tan 1 and
vC A << y so y + vC A y in the denominator of tan 2 . Introducing these simplifications
into (4.14) yields
v B A u C A v u
x y,av = = 1 + 2 + = + . (4.15)
x y x y


4.4.3. Point Shear Strains in 3D

To define the shear strain x y at point P we pass to the limit in the average strain expression (4.15)
by shrinking both dimensions x and y to zero:
def v u
x y = lim x y,av = lim + . (4.16)
x y

In the limit this gives the cross partial-derivative sum

u v
x y = + = yx (4.17)
y x

This expression plainly does not change if x and y are reversed, whence x y = yx as shown above.
The foregoing limit process can be repeated by taking the angles formed by planes {y, z} and {z, x}
to define yz,av and zx,av respectively, followed by passing to the limit as in (4.17). Anticipating
the more complete analysis of 4.5, the results are

u w v w
x z = + = zx , yz = + = zy . (4.18)
z x z y

This property: x y = yx , yz = zy , and zx = x z , is called shear strain reciprocity. This is

entirely analogous to the shear stress reciprocity described in Lecture 1. It follows that the 3D
state of strain at a point can be defined by 9 components: 3 extensional and 6 shear, which can be
arranged as a 3 3 matrix
x x x y x z
yx  yy yz (4.19)
zx zy zz
On account of the shear strain reciprocity property (4.18) the above matrix is symmetric. Therefore
it can be defined by 6 independent components: three normal strains and three shear strains.

4.5. Strain-Displacement Equations

This section summarizes the connections between displacements and point strains, which have
appeared in piecewise manner so far. Consider an arbitrary body in 3D in its undeformed and
deformed configurations. A generic point P(x, y, z) moves to P  (x + u, y + v, z + w), in which
the displacement components are functions of position:

u = u(x, y, z), v = v(x, y, z), w = w(x, y, z). (4.20)

The components (4.20) define a displacement field. For visualization convenience we restrict the
foregoing picture to 2D as done in Figure 4.7. At P draw an infinitesimal rectangle PQRS of side
lengths {d x, dy} aligned with the RCC axes {x, y}. The square maps to a quadrilateral PQRS
in the deformed body as illustrated in Figure 4.7(b). To first order in {d x, dy} the mapped corner

Lecture 4: STRAINS

S'(x+u+ 6 u dx+6 u dy,y+v+ 6 v dx+6 v dy)

6x 6y 6x 6y
R'(x+u+6 u dy,y+v+6 v dy)
6y 6y

Q'(x+u+6 u dx,y+v+6 v dx)

P'(x+u,y+v) 6x 6x

R(x,y+dy) dx S(x+dx,y+dy) zoom

P(x,y) Q(x+dx,y) (b) Deformed
zoom body
(a) Undeformed Displacement P'
body vector
dx vP = v
y Position P

z x uP = u

Figure 4.7. Connecting strains to displacements: 2D view used for visualization convenience.

points are given by

P maps tp P at x + u,v
u v
Q maps to Q at x +u+ d x, y+v+ d x,
x x
u v (4.21)
R maps to R at x +u+ dy, y+v+ dy,
y y
u u v v
S maps to S at x +u+ dx + dy, y + v + dx + dy,
x y x y
as pictured in Figure 4.7(b).
To express x x in terms of displacements, take the ratio

u Q u P  x + u + ux d x (x + u) u
x x = = = . (4.22)
dx dx x
This was derived earlier in 4.3.2 in a 1D context. Notice that no passing to the limit is necessary
here because we started with infinitesimal material elements. Likewise

u R u P  y + v + v
y dy (y + v) v
 yy = = = . (4.23)
dy dy y
For the 3D case we get one more normal strain,
zz = . (4.24)


The connection of shear strains to displacement derivatives is more involved. The derivation for
x y was done in 4.4.2 and 4.4.3, which yields (4.17). The complete result for 3D is

u v v w w u
x y = + , yz = + , zx = + . (4.25)
y x z y x z

If the subscripts are switched we get

v u w v u w
yx = + , zy = + , x z = + . (4.26)
x y y z z x

Comparing these to (4.25) shows that x y = yx , etc, which proves the reciprocity property stated
in 4.4.3.

4.6. Displacement Vector Composition

In calculations of joint deflections of truss structures, it is often required to compose displacement

vectors. The composition rule is not the same as that of forces unless the vectors are mutually
orthogonal. The difference will be illustrated through a worked out example.
4.6.1. Deflection of a Truss Structure
The 2-member plane truss shown in Figure 4.8(a) was the subject of Homework Exercise 1.3, using
specific values for dimensions, loads and member properties. Here it will be initially processed in
symbolic form, keeping members lengths L AB , L BC , load q, angle , elastic modulus E (same for
both members) and bar area Ab (same for both members) as variables. Objective: find the magnitude
of the displacement B of joint B as function of the data. Infinitesimal strains are assumed (this is
to be verified after deformations and deflections are computed) and the material obeys Hooke law.
The first step is to find the internal member forces FAB and FBC from statics. The determination
of FAB is done with the FBD shown in Figure 4.8(b), by taking moments with respect to hinge
C. (This choice bypasses the need to find reactions RC x and RC y , since their contribution to the
moment equilibrium equation vanishes.) For moment calculation purposes, the uniform distributed
load q may be replaced by a resultant point force q L BC at the midpoint of member BC. We get
FAB = q L BC /(2 sin ), as pictured in Figure 4.8(b); this is tension if FAB > 0.
Next we find FBC using the FBD drawn in Figure 4.8(c), in which the previously found FAB is
used. The applied distributed load is lumped to the end joints B and C. Since q is uniform each
joint receives one half of the total force, that is, q L BC /2. Force equilibrium along either x or y
yields FBC = q L BC /(2 tan ), as shown in the figure. [Actually the quickest way to get FBC is to
consider equilibrium along the direction C B. Since the projection of q L BC /2 vanishes, FBC must
balance FAB cos whence FBC = FAB cos = q L BC cos /(2 sin ) = q L BC /(2 tan ).] Note
that both internal forces can be obtained without computing reactions.
Since Hookes law is assumed to apply and both members are prismatic, their elongations are
AB = FAB L AB /(E Ab ) and BC = FBC L BC /(E Ab ), respectively, with appropriate signs. See
Figure 4.8(d). The graphical composition of these two vectors to get B is detailed in Figure 4.8(e).
Mark AB and BC with origin at B aligned with member directions, pointing away if they are

Lecture 4: STRAINS

(a) (b)


q E, Ab q LBC
E, Ab
L BC sin


BC =

(d) E Ab
AB =
E Ab
q LBC /2 q LBC
B 2sin
B' (e)

2tan B

q LBC (LBC +LAB ) cos

2 E Ab sin sin

Figure 4.8. Computing the motion of a truss node.

positive. Then draw two perpendicular lines from the vector tips. The intersection of those lines
gives the deformed position B  of B, and the distance B B  is B . Using trigonometric relations this
value can be expressed in terms of the data as
q L BC (L BC + L AB ) cos
B = L AB . (4.27)
2E Ab sin sin

The numbers used in Exercise 1.3 were: L AB = 60 in, L BC = 66 in, = 60 and q = 80 lbs/in.
In addition to these we take Ab = 2.5 in2 and E = 30 106 psi (steel). Replacing gives

FAB = 3048 lbs, FBC = 1

FAB = 1524 lbs , both in tension,
AB = 0.002439 in, BC = 0.001341 in, B = 0.003833 in.

Since these displacements are very small compared to member lengths, the assumption of infinites-
imal strains is verified a posteriori.
4.6.2. Forces And Displacements Obey Different Composition Rules
The main point of the foregoing example is to emphasize that displacements do not compose by
the same rules as forces. The rules are graphically summarized in Figure 4.9 for two point forces
and two displacements in the plane of the figure.
Forces are combined by the well known vector-addition parallelogram rule: the tip of the resultant
is at the opposite corner of the parallelogram formed by the two vectors as sides. Displacements


(a) force (b) displacement

composition composition
Figure 4.9. Composition of (a) two forces and (b) two displacements
acting on the plane of the figure.

are combined by a cyclic quadrilateral rule as illustrated. (A cyclic quadrilateral is one that has
two opposite right angles.) If the vectors are orthogonal, the composition rules coalesce since in
that case both the parallelogram and the cyclic quadrilateral reduce to a rectangle.
If we had 3 or more 2D vectors the rules diverge. Forces can be combined by chaining by
placing them tail-to-tip. But in general 3 or more displacement vectors with common origin will
be incompatible since the perpendicular lines traced from their tips will not usually cross.
The distinction is also important in three-dimensional space. Any number of 3D force vectors can
be added by tail-to-tip chaining. Three displacement vectors with common origin can be combined
by constructing the normal plane at their tips and finding the point at which the planes intersect.
Composing more than three 3D displacement vectors is generally impossible.
An important restriction should be noted. The composition rules illustrated in Figures 4.8(e)
and 4.9(b) apply only to the case of infinitesimal deformations. This allows displacements to be
linearized by Taylor series expansion about the undeformed geometry. For finite displacements see
Section 2.4 of Vables textbook.
Remark 4.1. In advanced courses that cover tensors in arbitrary coordinate systems, it is shown that the
displacement field (or, in general, a gradient-generated vector field) transforms as a covariant field of order
one. On the other hand, a force field (or, in general, a differential-generated vector field) transforms as a
contravariant field of order one. For the pertinent math see the Wikipedia article and contravariance of tensors

Material Laws



5.1 Introduction . . . . . . . . . . . . . . . . . . . . . 53
5.2 Constitutive Equations . . . . . . . . . . . . . . . . . 53
5.2.1 Material Behavior Assumptions . . . . . . . . . . . 53
5.2.2 The Tension Test Revisited . . . . . . . . . . . . 53
5.3 Characterizing a Linearly Elastic Isotropic Material . . . . . . . 55
5.3.1 Determination Of Elastic Modulus and Poissons Ratio . . . 55
5.3.2 Determination Of Shear Modulus . . . . . . . . . . . 55
5.3.3 Determination Of Thermal Expansion Coefficient . . . . . 55
5.4 Hookes Law in 1D . . . . . . . . . . . . . . . . . . . 56
5.4.1 Elastic Modulus And Poissons Ratio In 1D Stress State . . . 56
5.4.2 Shear Modulus In 1D Stress State . . . . . . . . . . . 56
5.4.3 Thermal Strains In 1D Stress State . . . . . . . . . . 58
5.5 Generalized Hookes Law in 3D . . . . . . . . . . . . . . 59
5.5.1 Strain-To-Stress Relations . . . . . . . . . . . . . 59
5.5.2 Stress-To-Strain Relations . . . . . . . . . . . . . 510
5.5.3 Thermal Effects in 3D . . . . . . . . . . . . . . 511
5.6 Generalized Hookes Law in 2D . . . . . . . . . . . . . . 512
5.6.1 Plane Stress . . . . . . . . . . . . . . . . . 512
5.6.2 Plane Strain . . . . . . . . . . . . . . . . . . 513
5.7 Example: An Inating Balloon . . . . . . . . . . . . . . 514
5.7.1 Strains and Stresses in Balloon Wall . . . . . . . . . . 514
5.7.2 When Will the Balloon Burst? . . . . . . . . . . . 515


5.1. Introduction

Recall from the previous Lecture the following connections between various quantities that appear
in continuum structural mechanics:
internal forces stresses strains displacements size & shape changes (5.1)

displacements strains stresses internal forces (5.2)
Of these, we have studied mechanical stresses in Lecture 1 and strains in Lecture 4. How are they
linked? Through the material properties of the structural body. This is pictured by the MP symbol
above the appropriate arrow connectors. Material behavior is mathematically characterized by the
so-called constitutive equations, also called material laws.

5.2. Constitutive Equations

In this Lecture we will restrict detailed examination of constitutive behavior to elastic isotropic
materials. More complex behavior (for example: orthotropy, plasticity, viscoelasticity, and fracture)
are studied in senior and graduate level courses in Structural and Solid mechanics.
5.2.1. Material Behavior Assumptions
There is a very wide range of materials used for structures, with drastically different behavior. In
addition the same material can go through different response regimes: elastic, plastic, viscoelastic,
cracking and localization, fracture. As noted above, we will restrict our attention to a very specific
material class and response regime by making the following behavioral assumptions.
1. Macroscopic Model. The material is mathematically modeled as a continuum body. Features
at the meso, micro and nano levels: crystal grains, molecules, and atoms, are ignored.
2. Elasticity. This means the stress-strain response is reversible and consequently the material
has a preferred natural state. This state is assumed to be taken in the absence of loads at a
reference temperature. By convention we will say that the material is then unstressed and
undeformed. On applying loads, and possibly temperature changes, the material develops
nonzero stresses and strains, and moves to occupy a deformed configuration.
3. Linearity. The relationship between strains and stresses is linear. Doubling stresses doubles
strains, and viceversa.
4. Isotropy. The properties of the material are independent of direction. This is a good assumption
for materials such as metals, concrete, plastics, etc. It is not adequate for heterogenous mixtures
such as composites or reinforced concrete, which are anisotropic by nature. The substantial
complications introduced by anisotropic behavior justifies its exclusion from an introductory
5. Small Strains. Deformations are considered so small that changes of geometry are neglected
as the loads are applied. Violation of this assumption requires the introduction of nonlinear
relations between displacements and strains. This is necessary for highly deformable materials
such as rubber (more generally, polymers). Inclusion of nonlinear behavior significantly
complicates the constitutive equations and is therefore left for advanced courses.


Nominal stress
= P/A 0 Max
Strain stress
hardening Localization
Elastic Mild Steel stress
limit Tension Test
Linear elastic gage length
behavior L0
(Hooke's law is
valid over this
response region)

Undeformed state Nominal strain = L /L0

Figure 5.1. Typical tension test behavior of mild steel, which displays a well defined yield
point and extensive yield region.

Moderately from start
ductile (rubber,
Brittle (Al alloy) polymers)
(glass, ceramics,
concrete in tension)

Figure 5.2. Three material response flavors as displayed in a tension test.

Nominal stress
= P/A0
Tool steel

Note similar
elastic modulus High strength steel

Mild steel
(highly ductile)
Conspicuous yield
Nominal strain = L /L 0

Figure 5.3. Different steel grades have approximately the same elastic modulus, but very
different post-elastic behavior.


5.2.2. The Tension Test Revisited

The first acquaintance of an engineering student with lab-controlled material behavior is usually
through tension tests carried out during the first Mechanics, Statics and Structures sophomore
course. Test results are usually displayed as axial nominal strain versus axial nominal strain, as
illustrated in Figure 5.1 for a mild steel specimen taken up to failure. Several response regions
are indicated there: linearly elastic, yield, strain hardening, localization and failure. These are
discussed in the aforementioned course, and studied further in courses on Aerospace Materials.
It is sufficient to note here that we shall be mostly concerned with the linearly elastic region that
occurs before yield. In that region the one-dimensional (1D) Hookes law is assumed to hold.
Material behavior may depart significantly from that shown in Figure 5.1. Three distinct flavors:
brittle, moderately ductile and nonlinear-from-start, are shown schematically in Figure 5.2 . Brittle
materials such as glass, rock, ceramics, concrete-under-tension, etc., exhibit primarily linear be-
havior up to near failure by fracture. Metallic alloys used in aerospace, such as Aluminum and
Titanium alloys, display moderately ductile behavior, without a well defined yield point and yield
region: the stress-strain curve gradually turns down finally dropping to failure. Some materials,
such as rubbers and polymers, exhibit strong nonlinear behavior from the start. Although such
materials may be elastic there is no easily identifiable linearly elastic region.
Even for a well known material such as steel, the tension test behavior can vary significantly
depending on combination with other components. Figure 5.3 sketches the response of mild steel
with high-strength steel used in critical structural components, and with tool steel. Mild steel is
highly ductile and clearly exhibits an extensive yield region. Hi-strength steel is less ductile and
does not show a well defined yield point. Tool steel has little ductility, and its behavior displays
features associated with brittle materials. The trade off between ductility and strength is typical.
Note, however, that all three grades of steel have approximately the same elastic modulus, which is
the slope of the stress-strain line in the linear region of the tension test.

5.3. Characterizing a Linearly Elastic Isotropic Material

For an isotropic material in the linearly elastic region of its response, four numerical properties are
sufficient to establish constitutive equations. Those equations are associated with the well known
Hookes law, originally enunciated by Robert Hooke by 1660 for a spring, and later expressed
in terms of stresses and strains once those concepts appeared in the XIX Century. These four
properties: E, , G and , are tabulated in Figure 5.4.

5.3.1. Determination Of Elastic Modulus and Poissons Ratio

The experimental determination of the elastic modulus E and Poissons ratio makes use of a
uniaxial tension test specimen such as the one pictured in Figure 5.5. See slides for operational

5.3.2. Determination Of Shear Modulus

The experimental determination of the shear modulus G makes use of a torsion test specimen such
as the one pictured in Figure 5.6. See slides for operational details.


E Elastic modulus, a.k.a. Young's modulus

Physical dimension: stress=force/area (e.g. ksi)

Poisson's ratio
Physical dimension: dimensionless (just a number)

G Shear modulus, a.k.a. modulus of rigidity

Physical dimension: stress=force/area (e.g. MPa)

Coefficient of thermal expansion

Physical dimension: 1/degree (e.g., 1/ C)

E, and G are not independent. They are linked by

E = 2G (1+), G = E/(2(1+)), = E /(2G)1

Figure 5.4. Four properties that fully characterize the thermomechanical response
of an isotropic material in the linearly elastic range.

5.3.3. Determination Of Thermal Expansion Coefficient

The experimental determination of the thermal expansion coefficient can be made byt using a
uniaxial tension test specimen such as the one pictured in Figure 5.7. See slides for operational
5.4. Hookes Law in 1D
Once the values of E, , G and are experimentally determined (for widely used structural materials
they can be simply read off a manual), they can be used to construct thermoelastic constitutive
equations that link stresses and strains as described in the following subsections.
5.4.1. Elastic Modulus And Poissons Ratio In 1D Stress State
The one-dimensional Hookes law relates 1D normal stress to 1D extensional strain through two
material parameters introduced previously: the modulus of elasticity E, also called Youngs modulus
and Poissons ratio . The modulus of elasticity connects axial stress to axial strain :

= E , E= , = . (5.3)
Poissons ratio is defined as ratio of lateral strain to axial strain:
 lateral strain 
=   = lateral strain . (5.4)
axial strain  axial strain

The sign is introduced for convenience so that comes out positive. For structural materials
lies in the range 0.0 < 0.5. For most metals 0.250.35. For concrete and ceramics,
0.10. For cork 0. For rubber, 0.5 to 3 places. A material for which = 0.5 is called


Cross section of area A

(a) P P

gaged length

xx = P/A (uniform over cross section)

(b) y
P z
x Cartesian axes

xx 0 0 xx 0 0
0 0 0 Strain
state 0 yy 0
0 0 0 0 0 zz
at all points in the gaged region

Figure 5.5. Speciment for determination of elastic modulus E and Poissons ratio in the linearly elastic
response region of an isotropic material, using an uniaxial tension test.

Circular cross section

(a) T T
gaged length

For distribution of shear stresses and

strains over the cross section, cf. Lecture 7
(b) T y
x Cartesian axes

0 xy 0 0 xy 0
yx 0 0 Strain
0yx 0
0 0 0 0 0 0
at all points in the gaged region. Both the shear stress yx = xy as well
as the shear strain xy = yx vary linearly as per distance from the
cross section center (Lecture 7). They attain maximum values on the
specimen surface.

Figure 5.6. Speciment for determination of shear modulus G in the linearly elastic response region of an
isotropic material, using a torsion test.

5.4.2. Shear Modulus In 1D Stress State

The shear modulus G connects a shear strain to the corresponding shear stress :

= G , G = , = . (5.5)
Corresponding means that if = x y , say, then = x y , and similarly for the other shear
components. The shear modulus is usually obtained from a torsion test. It turns out that the


gaged length

Figure 5.7. Speciment for determination of coefficient of thermal expansion by heating a

tension test specimen and allowing it to expand freely.

3 material properties E, and G for an elastic isotropic material are not independent, but are
connected by the relations
G= , E = 2(1 + ) G, = 1. (5.6)
2(1 + ) 2G
which means that if two of them are known by measurement, the third one can be obtained from
the relations (5.6). In practice the three properties are often measured independently, and the
(approximate) verification of (5.6) gives an idea of how isotropic the material is.
5.4.3. Thermal Strains In 1D Stress State
A temperature change of T with respect to a base or reference level produces a thermal strain

T = T, (5.7)

in which is the coefficient of thermal dilatation, measured in 1/ F or 1/ C. This is typically

positive: > 0 and very small: << 1, of order 106 for most structural materials. To combine
mechanical and thermal effects in 1D, the strains are superposed:

 =  M + T = + T, (5.8)
The last expression is valid if the material is linearly elastic and obeys the 1D Hookes law.

xx = 0 and xx = 0 at

; ;;
E, , constant undeformed reference state,
over bar then bar is heated by T



Figure 5.8. Heated bar precluded from axial expansion. This bar will develop
a compressive axial stress called a thermal stress.

Example 5.1. The bar AB shown in Figure 5.8 is precluded from extending axially. It has elastic modulus
E and coefficient of dilatation > 0. The stress is zero when the bar is at the reference temperature Tr e f .
Find which axial stress develops if the temperature changes to T = Tr e f + T .
Since the bar length cannot change, the combined axial strain must be zero:

x x =  = + T = 0, (5.9)


y yy
(a) (c)
a b a b
c x
Initial shape Final shape yy
after test (2)
(b) (d)
xx zz
a b a b
c c
Final shape zz Final shape
after test (1) after test (3)
Figure 5.9. Derivation of three-dimensional Generalized Hookes Law for normal stresses and strains.
Three tension tests are assumed to be carried out along {x, y, z}, respectively, and strains superposed.

Solving for gives

= E T. (5.10)
Since E and are positive, a rise in temperature, i.e., T > 0, will produce a negative axial stress, and the bar
will be in compression. This is an example of the so-called thermally induced stress or simply thermal stress.
It is the reason behind the use of expansion joints in pavements, rails and bridges. The effect is important in
orbiting vehicles such as satellites, which undergo extreme (and cyclical) temperature changes from full sun
to Earth shade.

5.5. Generalized Hookes Law in 3D

5.5.1. Strain-To-Stress Relations

We now generalize the foregoing equations to the three-dimensional case, still assuming that the
material is elastic and isotropic. Condider a cube of material aligned with the axes {x, y, z}, as
shown in Figure 5.9. Imagine that threetension tests, labeled (1), (2) and (3) respectively, are
conducted along x, y and z, respectively. Pulling the material by applying x x along x will produce
normal strains
x x x x x x
x = , (1)
 yy = , (1)
zz = . (5.11)
Next, pull the material by yy along y to get the strains

(2) yy yy yy
 yy = , x(2)
x = , (2)
zz = . (5.12)
Finally pull the material by zz along z to get

(3) zz zz zz
zz = , x(3)
x = , (3)
 yy = . (5.13)

In the general case the cube is subjected to combined normal stresses x x , yy and zz . Since we
assumed that the material is linearly elastic, the combined strains can be obtained by superposition
of the foregoing results:
x x yy zz 1  
x x = x(1) (2) (3)
x + x x + x x = = x x yy zz .
(1) (2) (3) 1  
 yy =  yy +  yy +  yy =

= x x + yy zz . (5.14)
(1) (2) (3) x x yy zz 1  
zz = zz + zz + zz = + = x x yy + zz .
The shear strains and stresses are connected by the shear modulus as
x y yx yz zy zx x z
x y = yx = = , yz = zy = = , zx = x z = = . (5.15)
The three equations in (5.14), plus the three in (5.15), may be collectively expressed in matrix form
as 1
0 0 0
x x E 1
x x
 yy E E 0 0 0 yy

zz 1
0 0 0
= E E E
. (5.16)
x y 0 0 0 1 0 0 x y
yz 0 0 0 0 G 1 0 yz
zx 1 zx
0 0 0 0 0 G
5.5.2. Stress-To-Strain Relations
To get stresses if the strains are given, the most expedient method is to invert the matrix equation
(5.16). This gives

x x E (1 ) E E 0 0 0 x x
yy E E (1 ) E 0 0 0  yy

zz E E E (1 ) 0 0 0 zz
= . (5.17)
x y 0 0 0 G 0 0 x y

yz 0 0 0 0 G 0 yz
zx 0 0 0 0 0 G zx
Here E is an effective modulus modified by Poissons ratio:
E = (5.18)
(1 2)(1 + )
The six relations in (5.17) written out in long form are
x x = (1 ) x x +  yy + zz ,
(1 2)(1 + )
yy = x x + (1 )  yy + zz ,
(1 2)(1 + ) (5.19)
zz = x x +  yy + (1 ) zz ,
(1 2)(1 + )
x y = G x y , yz = G yz , zx = G zx .


The combination
av = 13 (x x + yy + zz ) (5.20)

is called the mean stress, or average stress. The negative of av is the pressure: p = av .
The combination v = x x +  yy + zz is called the volumetric strain, or dilatation. The negative
of v is known as the condensation. Both pressure and volumetric strain are invariants, that is,
their value does not change if axes {x, y, z} are rotated. An important relation between pressure
and volumetric strain can be obtained by adding the first three equations in (5.19), which upon
simplification and accounting for (5.20) and p = av relates pressure and volumetric strain as

p= v = K v . (5.21)
3(1 2)

This coefficient K is called the bulk modulus. If Poissons ratio approaches 12 , which happens for
near incompressible materials, K .

Remark 5.1. In the solid mechanics literature p is also defined (depending on authors preferences) as
p = av = 13 (x x + yy + zz ), which is the negative of the above one. If so, p = +K v . The definition
p = av is the most common one in fluid mechanics.

5.5.3. Thermal Effects in 3D

To incorporate the effect of a temperature change T with respect to a base or reference temperature,
add T to the three normal strains in (5.14)

x x = x x yy zz + T,
 yy = x x + yy zz + T, (5.22)
zz = x x yy + zz + T.

No change in the shear strain-stress relation is needed because if the material is linearly elastic and
isotropic, a temperature change only produces normal strains. The stress-to-strain matrix relation
(5.16) expands to
E E 0 0 0
x x 1 x x 1
E 0 0 0
 yy E E yy 1
E 1 0
zz E E 0 0 zz 1
= + T . (5.23)
x y 0 0 0 1 0 0 0
G xy
yz 1 yz 0
0 0 0 0 0
zx G zx 0
0 0 0 0 0 1

Inverting this relation provides the stress-strain relations that account for a temperature change:

x x E (1 ) E E 0 0 0 x x 1
yy E E (1 ) E 0 0 0  yy 1

zz E E E (1 ) 0 0 0 
zz 1
= ,
x y 0 0 0 G 0 0 x y 1 2 0

yz 0 0 0 0 G 0 yz 0
zx 0 0 0 0 0 G zx 0
in which E is defined in (5.18). Note that if all mechanical normal strains x x ,  yy , and zz vanish,
but T = 0, the normal stresses given by (5.24) are nonzero. Those are called initial thermal
stresses, and are important in engineering systems exposed to large temperature variations, such as
rails, turbine engines, satellites or reentry vehicles.
5.6. Generalized Hookes Law in 2D
Two specializations of the foregoing 3D equations to two dimensions are of interest in the appli-
cations: plane stress and plane strain. Plane stress is more important in Aerospace structures,
which tend to be thin, so in this course more attention is given to that case. Both specializations
are reviewed next.
5.6.1. Plane Stress
In this case all stress components with a z component are assumed to vanish. For a linearly elastic
isotropic material, the strain and stress matrices take on the form
x x x y 0 x x x y 0
yx  yy 0 , yx yy 0 (5.25)
0 0 zz 0 0 0
Note that the zz strain, often called the transverse strain or thickness strain in applications, in
general will be nonzero because of Poissons ratio effect. The strain-stress equations are easily
obtained by going to (5.14) and (5.15) and setting zz = yz = zx = 0. This gives
1   1    
x x = x x yy ,  yy = x x + yy , zz = x x + yy ,
E E E (5.26)
x y
x y = , yz = zx = 0.
The matrix form, omitting known zero components, is
E 0
x x E 
1 x x
 yy E 0
= E
yy . (5.27)
E E 0 x y
x y 1
0 0 G
Inverting the matrix composed by the first, second and fourth rows of the above relation gives the
stress-strain equations   
x x E E 0 x x
yy = E E 0  yy . (5.28)
x y 0 0 G x y


in which E = E/(1 2 ). Written in long hand,

x x = (x x +  yy ), yy = ( yy + x x ), x y = G x y . (5.29)
1 2 1 2
If temperature changes T are considered, the foregoing equations acquire extra terms on the right:
E 0
x x E  1
1 x x
 yy E 0 1
= E
yy + T , (5.30)
zz 0 1
E E x y
x y 1 0
0 0 G
x x E E 0 x x E T 1
yy = E E 0  yy 1 . (5.31)
0 0 G 1 0
xy xy

5.6.2. Plane Strain

In this case all strain components with a z component are assumed to vanish. For a linearly elastic
isotropic material, the strain and stress matrices take on the form
x x x y 0 x x x y 0
yx  yy 0 , yx yy 0 (5.32)
0 0 0 0 0 zz
Note that the zz stress, which is called the transverse stress in applications, in general will not
vanish. The strain-to-stress relations can be easily obtained by setting zz = yz = zx = 0 in
(5.19). This gives
x x = (1 ) x x +  yy ,
(1 2)(1 + )
yy = x x + (1 )  yy ,
(1 2)(1 + ) (5.33)
zz = x x +  yy ,
(1 2)(1 + )
x y = G x y , yz = 0, zx = 0.
which in matrix form, with the zero components removed, is

x x E (1 ) E 0 

yy E E (1 ) 0 xx
=  yy . (5.34)
zz E E 0
x y
x y 0 0 G
Inverting the system provided by extracting the first, second and fourth rows of (5.34) gives the
stress-to-strain relations, which are omitted for simplicity.
The effect of temperature changes can be incorporated without any difficulty.


Backup Material Only Not Covered In Lecture

Initial (reference)
diameter D0 , under inflation
inflation pressure p0 pressure
pf = p0 + p
Final (deformed)
diameter Df = D0


For simplicity assume

a spherical ballon shape Df = D0
for all calculations

Figure 5.10. Inflating balloon example problem.

5.7. Example: An Inflating Balloon

This is a generalization of Problem 3 of Recitation #2. The main change is that all data is expressed
and kept in variable form until the problem is solved. Specific numbers are plugged in at the end.
This will be the only problem in the course where some features of nonlinear mechanics appear.
These come in by writing the governing equations in both the initial and final geometries, without
5.7.1. Strains and Stresses in Balloon Wall

The problem is depicted in Figure 5.10. A spherical rubber balloon has initial diameter D0 under
inflation pressure p0 . This is called the initial configuration. The pressure is increased by p so
that the final pressure is p f = p0 + p. The balloon assumes a spherical shape with final diameter
D f = D0 , in which > 1. This will be called the final configuration. The initial wall thickness is
t0 << D0 and the final thickness is t f . Since the balloon geometry is assumed to remain spherical
for simplicity, we can apply to both configurations the stress formulas for the thin-wall spherical
vessel derived in Lecture 3.
The strains (but not stresses) are assumed to be zero in the initial configuration. The average
circumferential extensional strain assumed in the final configuration depends on whether we take
the Lagrangian or the Eulerian strain measure, which are designated by avL
and av
, respectively.

(D f D0 ) (D f D0 ) 1
 Lf = = 1,  Ef = = , (5.35)
D0 Df


Since the balloon is assumed to remain spherical and its thickness is very small compared to its
diameter, the above strains hold at all points of the balloon wall, and are the same in any direction
tangent to the sphere. If we choose the sphere normal as local z axis, the wall is in a plane stress
Next we introduce material laws. We will assume that rubber obeys the two-dimensional, plane
stress generalized Hookes law (5.31) with respect to the Eulerian strain measure, with effective
modulus of elasticity E and Poissons ratio .1 Setting x x =  yy =  Ef and x y = 0 therein and
accounting for the initial stress 0 , we obtain the inplane normal stress in the final configuration:
E E E 1
x x = yy = f = 0 + ( Ef +  Ef ) = 0 +  Ef = 0 + . (5.36)
1 2 1 1
The normal inplane wall stress is the same in all directions, so it is called simply 0 and f , for
initial and final configurations, respectively. The inplane shear stress vanishes in all directions.
Assume D0 , t0 , E and are given as data. An interesting question: what is the relation between p
(the excess or gage pressure) and the diameter D f = D0 ? And, is there a maximum pressure that
will cause the balloon to burst?
5.7.2. When Will the Balloon Burst?
To relate p and it is necessary to express the wall stresses 0 and f in terms of geometry and
internal pressure. This is provided by equation (3.10) in Lecture 3, derived for a thin-wall spherical
vessel. In that equation replace p, R and t by quantities in the initial and final configurations:
p0 R 0 p0 D 0 pf Rf ( p0 + p) D f ( p0 + p) D0
0 = = , f = = = . (5.37)
2t0 4t0 2tf 4tf 4tf
All quantities in the above expressions are known in terms of the data, except t f . A kinematic
analysis beyond the scope of this course shows that
t f = 1 + 2 1 t0 . (5.38)
We can check (5.38) by inserting two limit values of Poissons ratio:
= 0: t f = t0 . This is correct since the thickness does not change.
= 1/2: t f = t0 / 2 . Is this correct? If = 1/2 the material is incompressible and does not
change volume. The initial and final volume of the thin-wall spherical balloon are
V0 = D02 t0 and V f = D 2f t f = 2 D02 t f , respectively. On setting V0 = V f and
solving for t f we get t f = t0 / 2 .
To obtain p in terms of , replace (5.38) into (5.37), equate this to (5.36) and solve for p. The
result provided by Mathematica is
4Et0 (1 )(2 + 2 (1 2)) + D0 p0 (1 )(4 + 2 (2 4))
p= (5.39)
D0 4 (1 )
1 This is a very rough approximation since constitutive equations for rubber (and polymers in general) are highly nonlinear.
But getting closer to reality would take us into the realm of nonlinear elasticity, which is a graduate-level topic.


p (MPa)
7 =0
2 =1/2
0 = D f /D0
1 1.5 2 2.5 3 3.5 4

Figure 5.11. Inflating pressure (in MPa) versus diameter expansion ratio
= D f /D0 for a balloon with E = 1900 MPa, D0 = 50 mm, p0 = 0 MPa,
t0 = 0.18 mm, 1 4 and Poissons ratios = 0 and = 12 .

This expression simplifies considerably in the two Poissons ratio limits:

4E t0 ( 1) + D0 p0 (2 )
p|=0 = (5.40)
D0 2

8E t0 ( 1) + D0 p0 (2 3 )
p|=1/2 = (5.41)
D0 4
Pressure versus diameter ratio curves given by (5.40) and (5.41) are plotted in Figure 5.11 for
the numerical values indicated there. Those values correspond to the data used in Problem 3 of
Recitation 2, in which = 1/2 was specified from the start.
Rubber (and, in general, polymer materials) are nearly incompressible; for example 0.4995 for
rubber. Consequently, the response depicted in Figure 5.11 for = 1/2 is more physically relevant
than the other one.
Do the response plots in Figure 5.11 tell you when an inflating balloon is about to collapse? Yes.
This is the matter of a (optional) Homework Exercise.

Plane Stress



6.1 Introduction . . . . . . . . . . . . . . . . . . . . . 63
6.2 Thin Plate in Plate Stress . . . . . . . . . . . . . . . . 63
6.3 2D Stress Transformations . . . . . . . . . . . . . . . . 65
6.3.1 Why Are Stress Transformations Important? . . . . . . 65
6.3.2 Method of Equations . . . . . . . . . . . . . . . 65
6.3.3 Double Angle Version . . . . . . . . . . . . . . 66
6.3.4 Principal Stresses, Planes, Directions, Angles . . . . . . . 66
6.3.5 Maximum Shear Stresses . . . . . . . . . . . . . 67
6.3.6 Principal Stress Element . . . . . . . . . . . . . . 68
6.3.7 Mohrs Circle . . . . . . . . . . . . . . . . . 68
6.4 What Happens in 3D? . . . . . . . . . . . . . . . . . . 610
6.4.1 Including the Plane Stress Thickness Dimension . . . . . 610
6.4.2 3D Mohr Circles . . . . . . . . . . . . . . . . 610
6.4.3 Overall Maximum Shear . . . . . . . . . . . . . 611
6.4.4 Plane Stress Revisited . . . . . . . . . . . . . . 612
6.4.5 The Sphere Paradox . . . . . . . . . . . . . . . 613


6.1. Introduction

This Lecture deals with the plate stress problem. This is a two-dimensional stress state, briefly
introduced in ? of Lecture 1. It occurs frequently in two kinds of aerospace structural components:
1. Thin wall plates and shells; e.g., aircraft and rocket skins, and the pressure vessels of Lecture 3.
2. Shaft members that transmit torque. These will be studied in Lectures 79.
The material below focuses on thin flat plates, and works out the associated problem of plane stress

6.2. Thin Plate in Plate Stress

In structural mechanics, a flat thin sheet of material is called a plate. The distance between the plate
faces is the thickness, which is denoted by h. The midplane lies halfway between the two faces.
The direction normal to the midplane is the transverse z
direction. Directions parallel to the midplane are called Top surface
in-plane directions. The global axis z is oriented along
the transverse direction. Axes x and y are placed in the
midplane, forming a right-handed Rectangular Carte-
sian Coordinate (RCC) system. Thus the equation of y
the midplane is z = 0. The +z axis conventionally
defines the top surface of the plate as the one that it x
intersects, whereas the opposite surface is called the Figure 6.1. A plate structure in plane stress.
bottom surface. See Figure 6.1.
A plate loaded in its midplane is said to be in a state of plane stress, or a membrane state, if the
following assumptions hold:
1. All loads applied to the plate act in the midplane direction, as pictured in Figure 6.1, and are
symmetric with respect to the midplane.
2. All support conditions are symmetric about the midplane.
3. In-plane displacements, strains and stresses are taken to be uniform through the thickness.
4. The normal and shear stress components in the z direction are zero or negligible.
The last two assumptions are not necessarily consequences of the first two. For those to hold, the
thickness h should be small, typically 10% or less, than the shortest in-plane dimension. If the plate
thickness varies it should do so gradually. Finally, the plate fabrication must exhibit symmetry with
respect to the midplane.
To these four assumptions we add an adscititious restriction:
5. The plate is fabricated of the same material through the thickness. Such plates are called
transversely homogeneous or (in aerospace) monocoque plates.
The last assumption excludes wall constructions of importance in aerospace, in particular composite
and honeycomb sandwich plates. The development of mathematical models for such configurations
requires a more complicated integration over the thickness as well as the ability to handle coupled
bending and stretching effects. Those topics fall outside the scope of the course.





Figure 6.2. Mathematical model of plate in plane stress. (Symbols  and , used to
denote the plate interior and the boundary, respectively, are used in advanced courses.)

In-plane internal forces y + sign conventions

for internal forces,
dx dy stresses and strains
h pyy
Thin plate in plane stress y x
pxx pxy dy
z x dx

dx dy In-plane stresses In-plane body forces

y dx dy
dx dy
yy y h
x y
x xx xy = yx by

In-plane strains In-plane displacements

dx dy dx dy
h h
yy y uy y
x xx xy = yx x u x

Figure 6.3. Notational conventions for in-plane stresses, strains, displacements and
internal forces of a thin plate in plane stress.

Remark 6.1. Selective relaxation from assumption 4 leads to the so-called generalized plane stress state, in
which nonzero zz stresses are accepted; but these stresses do not vary with z. The plane strain state described
in ? of Lecture 5 is obtained if strains in the z direction are precluded: zz = x z = yz = 0.

Remark 6.2. Transverse loading on a plate produces plate bending, which is associated with a more complex
configuration of internal forces and deformations. This topic is studied in graduate-level courses.

Remark 6.3. The requirement of flatness can be relaxed to allow for a curved configuration, as long as
the structure, or structure component, resists primarily in-plane loads. In that case the midplane becomes a
midsurface. Examples are rocket and aircraft skins, ship and submarine hulls, open parachutes, boat sails
and balloon walls. Such configurations are said to be in a membrane state. Another example are thin-wall
members under torsion, which are covered in Lectures 89.
The plate in plane stress idealized as a two-dimensional problem is illustrated in Figure 6.2.
In this idealization the third dimension is represented as functions of x and y that are integrated


yy tt
(a) (b) tn
yx nt nn
P xx P

y t y
Global axes Local axes
x,y stay fixed x x n,t rotate by
z z
Figure 6.4. Plane stress system referred to global axes x, y (data) and to local rotated axes n, t.
(Locations of point P in (a,b) coincide; they are drawn offset for visualization convenience.)

through the plate thickness. Engineers often work with internal plate forces, which result from
integrating the in-plane stresses through the thickness. See Figure 6.3.
In this Lecture we focus on the in-plane stresses x x , yy and x y and their expressions with respect
to an arbitrary system of axes

6.3. 2D Stress Transformations

The stress transformation problem studied in this Lecture is illustrated in Figure 6.4. Stress com-
ponents x x , yy and x y at a midplane point P are given with respect to the global axes x and y,
as shown in Figure 6.4(a). The material element about P is rotated by an angle that aligns it with
axes n, t, as shown in Figure 6.4(b). (Note that location of point P in (a,b) coincide they are
drawn offset for visualization convenience.)
The transformation problem consists of expressing nn , tt , and nt = tn in terms of the stress
data x x , yy and x y , and of the angle . Two methods, one analytical and one graphical, will
be described here. Before that is done, it is useful to motivate what are the main uses of these
6.3.1. Why Are Stress Transformations Important?

The transformation problem has two major uses in structural analysis and design.
Find stresses along a given skew direction. Here is given as data. This has several
applications. Two examples:
1. Analysis of fiber reinforced composites if the direction of the fibers is not aligned with the
{x, y} axes. A tensile normal stress perpendicular to the fibers may cause delamination,
and a compressive one may trigger local buckling.
2. Oblique joints that may fail by shear parallel to the joint. For example, welded joints.
Find max/min normal stresses, max in-plane shear and overall max shear. This may be
important for strength and safety assessment. Here finding the angle is part of the problem.
Both cases are covered in the following subsections.


6.3.2. Method of Equations

The derivation given on pages 524-525 of Vable or in 7.2 of Beer-Johnston-DeWolf is based on
the wedge method. This will be omitted here for brevity. The final result is

nn = x x cos2 + yy sin2 + 2 x y sin cos ,

tt = x x sin2 + yy cos2 2 x y sin cos , (6.1)
nt = (x x yy ) sin cos + x y (cos sin ).
2 2

Couple of checks are useful to verify these equations. If = 0 ,

nn = x x , tt = yy , nt = x y , (6.2)

as expected. If = 90 ,
nn = yy , tt = x x , nt = x y . (6.3)
which is also OK (can you guess why nt at = 90 is x y ?). Note that

x x + yy = nn + tt . (6.4)

This sum is independent of , and is called a stress invariant. (Mathematically, it is the trace of the
stress tensor.) Consequently, if nn is computed, the fastest way to get tt is as x x + yy nn ,
which does not require trig functions.
6.3.3. Double Angle Version
For many developments it is convenient to express the transformation equations (6.1) in terms of
the double angle 2 by using the well known trigonometric relations cos 2 = cos2 sin2
and sin 2 = 2 sin cos , in addition to sin2 + cos2 = 1. The result is

x x + yy x x yy
nn = + cos 2 + x y sin 2,
2 2 (6.5)
x x yy
nt = sin 2 + x y cos 2.

Here tt is omitted since, as previously noted, it can be quickly computed as tt = x x + yy nn .

6.3.4. Principal Stresses, Planes, Directions, Angles
The maximum and minimum values attained by the normal stress nn , considered as a function of
, are called principal stresses. (A more precise name is in-plane principal normal stresses, but
the qualifiers in-plane and normal are often dropped for brevity.) Those values occur if the
derivative dnn /d vanishes. Differentiating the first of (6.1) and passing to double angles gives

= 2( yy x x ) sin cos + 2x y (cos2 sin2 )
d (6.6)
= ( yy x x ) sin 2 + 2x y cos 2 = 0.


This is satisfied for = p if

2 x y
tan 2 p = (6.7)
x x yy

There are two double-angle solutions 21 and 22 given by (6.7) in the range [0 360 ] or
[180 180 ] that are 180 apart. (Which range is used depends on the textbook; here
we use the first one.) Upon dividing those values by 2, the principal angles 1 and 2 define the
principal planes, which are 90 apart. The normals to the principal planes define the principal
stress directions. Since they differ by a 90 angle of rotation about z, it follows that the principal
stress directions are orthogonal.
As previously noted, the normal stresses that act on the principal planes are called the in-plane
principal normal stresses, or simply principal stresses. They are denoted by 1 and 2 , respectively.
Using (6.7) and trigonometric relations it can be shown that their values are given by

x x + yy x x yy
1,2 = + x2y . (6.8)
2 2

To evaluate (6.8) it is convenient to go through the following staged sequence:

1. Compute 
x x + yy x x yy
av = and R=+ + x2y . (6.9)
2 2
Meaning of these values: av is the average normal stress at P (recall that x x + yy is an
invariant and so is av ), whereas R is the radius of the Mohrs circle, as described in 6.3.7
below, thus the symbol. Furthermore, R represents the maximum in-plane shear stress value,
as discussed in 6.3.5 below.
2. The principal stress values are

1 = av + R, 2 = av R. (6.10)

3. Note that the a priori computation of the principal angles is not needed to get the principal
stresses if one follows the foregoing steps. If finding those angles is of interest, use (6.7).
Comparing (6.6) with the second of (6.5) shows that dnn /d = 2nt . Since dnn /d vanishes for
a principal angle, so does nt . Hence the principal planes are shear stress free.
6.3.5. Maximum Shear Stresses
Planes on which the maximum shear stresses act can be found by setting dnt /d = 0. A study
of this equation shows that the maximum shear planes are located at 45 from the principal
planes, and that the maximum and minimum values of nt are R. See for example, 7.3 of the
Beer-Johnston-DeWolf textbook.
This result can be obtained graphically on inspection of Mohrs circle, covered later.


2 =10 psi |max |= R=50 psi
yy = 20 psi 2= 108.44
(a) (b)
xy = yx =30 psi 1 =110 psi
(c) 18.44 +45
= 63.44
P xx =100 psi P 1=18.44 P
x x
principal principal
planes planes
y t y
n planes
x (d) 45 principal
x element
plane of max 45
inplane shear P

Figure 6.5. Plane stress example: (a) given data: stress components x x , yy and x y ; (b) principal stresses and
angles; (c) maximum shear planes; (d) a principal stress element (actually four PSE can be drawn, that shown is one
of them). Note: locations of point P in (a) through (d) coincide; they are drawn offset for visualization convenience.

Example 6.1. This example is pictured in Figure 6.5. Given: x x = 100 psi, yy = 20 psi and x y = 30 psi, as
shown in Figure 6.5(a), find the principal stresses and their directions. Following the recommended sequence
(6.9)(6.10), we compute first
100 + 20 100 20
av = = 60 psi, R=+ + 302 = 50 psi, (6.11)
2 2
from which the principal stresses are obtained as
1 = 60 + 50 = 110 psi , 2 = 60 50 = 10 psi. (6.12)
To find the angles formed by the principal directions, use (6.7):
2 30 3
tan 2 p = = = 0.75, with solutions 21 = 36.87 , 22 = 21 + 180 = 216.87 ,
100 20 4 (6.13)
whence 1 = 18.44 , 2 = 1 + 90 = 108.44 .

These values are shown in Figure 6.5(b). As regards maximum shear stresses, we have |max | = R = 50 psi.
The planes on which these act are located at 45 from the principal planes, as illustrated in Figure 6.5(c).
6.3.6. Principal Stress Element
Some authors, such as Vable, introduce here the so-called principal stress element or PSE. This is
a wedge formed by the two principal planes and the plane of maximum in-plane shear stress. Its
projection on the {x, y} plane is an isosceles triangle with one right angle and two 45 angles. For
the foregoing example, Figure 6.5(d) shows a PSE.
There are actually 4 ways to draw a PSE, since one can join point P to the opposite corners of
the square aligned with the principal planes in two ways along diagonals, and each diagonal splits
the square into two triangles. The 4 images may be sequentially produced by applying sucessive
rotations of 90 . Figure 6.5(d) shows one of the 4 possible PSEs for the example displayed in that
Figure. The PSE is primarily used for the visualization of material failure surfaces in fracture and
yield, a topic only covered superficially here.


22 = 36.88 +180 = 216.88

= shear
(a) Point in plane stress stress max = 50
yy = 20 psi 40
(a) H
xy = yx =30 psi 30
Radius R = 50
= normal
P xx =100 psi 10
0 20 40 60 80 100 stress
2 = 10 10
20 1 = 110
40 V
21 = 36.88

x min = 50
(b) Mohr's circle
coordinates of blue points are
H: (20,30), V:(100,-30), C:(60,0)

Figure 6.6. Mohrs circle for plane stress example of Figure 6.5.

6.3.7. Mohrs Circle

Mohrs circle is a graphical representation of the plane stress state at a point. Instead of using the
methods of equations, a circle is drawn on the {, } plane. The normal stress () and the share
stress ( ) are plotted along the horizontal and vertical axes, respectively, with as a parameter.
All stress states obtained as the angle is varied fall on a circle called Mohrs circle.1
This representation was more important for engineers before computers and calculators appeared.
But it still retains some appealing features, notably the clear visualization of principal stresses and
maximum shear. It also remains important in theories of damage, fracture and plasticity that have
a failure surface.
To explain the method we will construct the circle corresponding to Example 6.1, which is re-
produced in Figure 6.6(a) for convenience. Draw horizontal axis = nn () to record normal
stresses, and vertical axis = nt () to record shear stresses. Mark two points: V (for vertical
cut, meaning a plane with exterior normal parallel to x) at (x x , x y ) = {100, 30} and H (for
horizontal cut, meaning a plane with exterior normal parallel to y) at ( yy , x y ) = (20, 30).
The midpoint between H and V is C, the circle center, which is located at ( 12 (x x + yy ), 0) =
(av , 0) = (60, 0). Now draw the circle. It may be verified that its radius is R as given by (6.9);
which for Example 6.1 is R = 50. See Figure 6.6(b).
The circle intersects the axis at two points with normal stresses avg + R = 110 = 1 and
av R = 10 = 2 . Those are the principal stresses. Why? At those two points the shear stress nt
vanishes, which as we have seen characterizes the principal planes. The maximum in-plane shear
occurs when = nt is maximum or minimum, which happens at the highest and lowest points of
the circle. Obviously max = R = 50 and min = R = 50. What is the normal stress when the
shear is maximum or minimum? By inspection it is av = 60 because those points lie on a vertical
line that passes through the circle center C.
1 Introduced by Christian Otto Mohr (a civil engineer and professor at Dresden) in 1882. Other important personal
contributions were the concept of statically indeterminate structures and theories of material failure by shear.


Other features, such as the correlation between the angle on the physical plane and the rotation
angle 2 traversed around the circle will be explained in class if there is time. If not, one can find
those details in Chapter 7 of Beer-Johnston-DeWolf textbook, which covers Mohrs circle well.

6.4. What Happens in 3D?

Despite the common use of simplified 1D and 2D structural models, the world is three-dimensional.
Stresses and strains actually live in 3D. When the extra(s) space dimension(s) are accounted for,
some paradoxes are resolved. In this final section we take a quick look at principal stresses in 3D,
stating the major properties as recipes.
6.4.1. Including the Plane Stress Thickness Dimension
To fix the ideas, 3D stress results will be linked to the plane stress case studied in Example 6.1. and
pictured in Figure 6.5. Its 3D state of stress in {x, y, z} coordinates is defined by the 3 3 stress
100 30 0
S = 30 20 0 (6.14)
0 0 0
It may be verified that the eigenvalues of this matrix, arranged in descending order, are

1 = 110, 2 = 10, 3 = 0. (6.15)

Now in 3D there are three principal stresses, which act on three mutually orthogonal principal
planes that are shear stress free. For the stress matrix (6.14) the principal stress values are 110, 10
and 0. But those are precisely the eigenvalues in (6.15). This result is general:

The principal stresses in 3D are the eigenvalues of the 3D stress matrix

The stress matrix is symmetric. A linear algebra theorem says that a real symmetric matrix has a
full set of eigenvalues and eigenvectors, and that both eigenvalues and eigenvectors are guaranteed
to be real. The normals to the principal planes (the so-called principal directions) are defined by
the three orthonormalized eigenvectors, but this topic will not be pursued further.
In plane stress, one of the eigenvalues is always zero because the last row and column of S are null;
thus one principal stress is zero. The associated principal plane is normal to the transverse axis z,
as can be physically expected. Consequently zz = 0 is a principal stress. The other two principal
stresses are the in-plane principal stresses, which were those studied in 6.3.4. It may be verified
that their values are given by (6.8), and that their principal directions lie in the {x, y} plane.
Some terminology is needed. Suppose that the three principal stresses are ordered, as usually done,
by decreasing algebraic value:
1 2 3 (6.16)
Then 1 is called the maximum principal stress, 3 the minimum principal stress, and 2 the inter-
mediate principal stress. (Note that this ordering is by algebraic, rather than by absolute, value.)
In the plane stress example (6.14) the maximum, intermediate and minimum normal stresses are
110, 10 and 0, respectively. The first two are the in-plane principal stresses.


Overall All possible

= shear max shear stress states lie
in the shaded = shear
stress area between stress max
= 55
the outer and max
= 50
Inner Mohr's inner circles 50
circles 40
= normal 3 = 0 10 = normal
stress 0 20 40 60 80 100 stress
3 2 10
1 2 = 10
20 1 = 110
Outer Mohr's
circle 40

Yellow-filled circle is the inplane one

Figure 6.7. Mohrs circles for a 3D stress state: (a) general case; (b) plane stress example of Figure 6.5. In (b)
the Mohr circle of Figure 6.6 is the rightmost inner circle. Actual stress states lie on the grey shaded areas.

6.4.2. 3D Mohr Circles

More surprises (pun intended): in 3D there are actually three Mohr circles, not just one. To draw
the circles, start by getting the eigenvalues 1 , 2 and 3 of the stress matrix, for example through
the eig function of Matlab. Suppose they are ordered as per the convention (6.16). Mark their
values on the axis of the vs. plane. Draw the 3 circles with diameters {1 , 2 }, {2 , 3 } and
{1 , 3 }, as sketched in Figure 6.7(a).2 For the principal values (6.15) the three Mohr circles are
drawn in Figure 6.7(b), using a scale similar to that of Figure 6.6.
Of the three circles the wider one goes from 1 (maximum normal stress) to 3 (minimum normal
stress). This is known as the outer or big circle, while the two others are called the inner circles.
It can be shown (this is proven in advanced courses in continuum mechanics) that all actual stress
states at the material point lie between the outer circle and the two inner ones. Those are marked
as grey shaded areas in Figure 6.7.
6.4.3. Overall Maximum Shear
For ductile materials such as metal alloys, which yield under shear, the 3D Mohr-circles diagram
is quite useful for visualizing the overall maximum shear stress at a point, and hence establish the
factor of safety against that failure condition. Looking at the diagram of Figure 6.7(a), clearly the
overall maximum shear, called max overall
, is given by the highest and lowest point of the outer Mohrs
circle, marked as a blue dot in that figure. (Note that only its absolute value is of importance for
safety checks; the sign has no importance.) But that is also equal to the radius Router of that circle.
If the three principal stresses are algebraically ordered as in (6.16), then
3 1
= Router = (6.17)
Note that the intermediate principal stress 2 does not appear in this formula.
2 In the general 3D case there is no simple geometric construction of the circles starting from the six x, y, z independent
stress components, as done in 6.3.7 for plane stress.


If the principal stresses are not ordered, it is necessary to use the max function in a more complicated
formula that selects the largest of the 3 radii:
 1 2   2 3   3 1 
max = max 
overall , ,  (6.18)
2   2   2 

Taking absolute values in (6.18) is important because the max function picks up the largest algebraic
value. For example, writing max
= max(30, 50, 20) = 30 picks up the wrong value. On the
other hand max max(|30|, | 50|, |20|) = 50 is correct.

6.4.4. Plane Stress Revisited

Going back to plane stress, how do overall maximum shear and in-plane maximum shear compare?
Recall that one of the principal stresses is zero. The ordering (6.16) of the principal stresses is
assumed, and one of those three is zero. The following cases may be considered:
(A) The zero stress is the intermediate one: 2 = 0. If so, the in-plane Mohr circle is the outer
one and the two shear maxima coincide:
1 3
= max
= (6.19)

(B) The zero stress is either the largest one or the smallest one. Two subcases:
(B1) If 1 2 0 and 3 = 0 : max
= ,
2 (6.20)
(B2) If 3 2 0 and 1 = 0 : max
= .

In the plane stress example of Figure 6.5, the principal stresses are given by (6.12). Since both
in-plane principal stresses (110 psi and 10 psi) are positive, the zero principal stress is the smallest
one. We are in case (B), subcase (B1). Consequently max overall
= 12 1 = 55 psi, whereas max =
( 2 ) = 50 psi.
2 1


= shear = shear
stress stress
(a) (b) max
= 40
50 50
40 40
30 inplane
max = 0
30 inplane
max = 0
20 3 = 0 20
10 = normal 10 = normal
0 20 40 60 80 100 stress 0 20 40 60 80 100 stress
0 0
10 10
20 1 = 2=80 20 1 = 2=80
30 30
40 40
50 50

Figure 6.8. The sphere paradox: (a) Mohrs in-plane circle reduced to a point, whence max = 0;
overall = 40.
(b) drawing the 3D circles shows that max

6.4.5. The Sphere Paradox

Taking the third dimension into account clarifies some puzzles such as the sphere paradox.
Consider a thin spherical pressure vessel with p = 160 psi, R = 10 in and t = 0.1 in. In Lecture 3,
the wall stress in spherical coordinates was found to be = p R/(2 t) = 80 ksi, same in all
directions. The stress matrix at any point in the wall, taking z as the normal to the sphere at that
point, is  
80 0 0
0 80 0 (6.21)
0 0 0
The Mohr circle reduces to a point, as illustrated in Figure 6.8(a), and the maximum in-plane shear
stress is zero. And remains zero for any p. If the sphere is fabricated of a ductile material, it should
never break regardless of pressure. Plainly we have a paradox.
The paradox is resolved by considering the other two Mohr circles. These additional circles are
pictured in Figure 6.8(b). In this case the outer circle and the other inner circle coalesce. The
overall maximum shear stress is 40 ksi, which is nonzero. Therefore, increasing the pressure will
eventually produce yield, and the paradox is resolved.