BASIC CONCEPTS OF STRUCTURAL MECHANICS

1. Introduction
The development of the Finite Element Method (FEM) [1] is a significant event in the area of
engineering analysis and design. The initiation of the method and its basic developments are
due to the structural engineers who have contributed a lot to establish the method.
Subsequently it has drawn attentions of engineers and scientist of different fields including
mathematicians who have enriched the method and made it suitable for application to
problems of other fields. As a result of that, the method got a general shape which is standing
on a strong mathematical foundation. The mathematical approach for introducing the FEM
has the advantage that it gives a general understanding of the method but it may not be
appealing to an engineer as the number of abstractions in various mathematical steps is much
more in this approach. On the other hand, the engineering approach lacks from generality as
the method is explained in the context of a particular field of engineering but this approach is
much more convenient for an engineer in that field for understanding the subject. As the
objective here is to solve structural mechanics problems, the engineering approach is
followed and the method is explained in the light of structural mechanics. Thus the basic
concepts of solid or structural mechanics are essential which are explained in the following
sections [2].

2. Axis System
The cartesian or rectangular coordinate system of mutually perpendicular axes Ox, Oy and Oz
as shown in Fig. 1 will be used as the reference in most of the cases. The right handed system
as shown in Fig. 1 will be followed in most of the cases for the axis system and other relevant
quantities or parameters.

Fig. 1

3. Force
In this study, force is a general term which will indicate both force and moment since both
the quantities have similar feature. They are vector quantities having magnitude as well as
direction. The direction of a moment is indicated by the normal to the plane on which it acts.
A typical case is shown in Fig. 2 where a right handed system is followed as mentioned
earlier. The double arrowhead is used to indicate this as moment where a single arrow is used
for force. In a general case of three dimensional system, a force P has three components Px,
Py and Pz along x, y and z directions respectively where Px = P cosα, Py = P cosβ and Pz = P
cosγ (see Fig. 3). In the other way, it may be mentioned that a force P can be completely
defined by the three components Px, Py and Pz. The boldface is used to indicate that the force
P is a vector quantity while its magnitude is indicated P. In a compact form cosα, cosβ and
cosγ are written as l, m and n, and these are defined as direction cosines. These are also
applicable for a moment in the similar manner.

Fig. 2

Fig. 3
4. Displacement
The particles of a body or a structure undergo movement due to its deformation by some
means which is represented by displacement (linear movement) and rotation (rotational
movement). Similar to the case of force, the displacement is a general term which indicates
both displacement and rotation since both the quantities have similar feature. In a most
general case, the possible components of displacement and rotation at a point are three and
these are along the direction of x, y and z. In this study, the components of displacement are
indicated by u, v and w and that of rotation are θx, θy and θz.

5. Stress
As a general case, consider a solid objected, as shown in Fig. 4, subjected to a system of
external forces acting at different points on it which is in equilibrium. It may be imagined that
the solid object is made with infinite number of particles. Under the action of external forces,
the body will be in equilibrium and it will try to deform which will be resisted by the particles
within the body. In this process particles within the solid will be subjected to a system of
2
forces developed internally due to the action and reaction between the adjacent particles. The
intensity of these internally developed forces, which resist the deformation of the solid, is
defined as stress. For a complete definition of stress at a point within the solid the orientation
is important as the value of stress changes with the change of orientation. In this context an
imaginary plane may be considered to pass through the point as shown in Fig. 4 where the
orientation is specified by the direction of the normal of the imaginary plane. Considering
one half of the solid, a small area ∆A surrounding the point on the imaginary plane as shown
in Fig. 4 may be taken. If ∆R is the resultant force acting on this small area, the resultant
stress σr will be ∆R/∆A, if ∆A tends to zero. The resultant stress may be resolved into two
components where one of the components, known as normal stress σ, is normal to the plane
and the other component is tangential to the plane which is known as shear stress τ (see Fig.
4). Again the shear stress may be resolved into two suitable components in that plane. For a
complete definition of stress at a point a cubic element as shown in Fig. 5 may be taken
where the normal to the faces are aligned according to axis system. On each face there are
three components of stresses where one of the components is the normal stress having one
subscript and the other two are shear stresses having two subscripts. The subscript of the
normal stress is identical to the first subscript of shear stress which is corresponding to the
normal to that plane while the second subscript of the shear stress is according to its
direction. As the body and any part of it is in equilibrium, each stress component has a pair
which is found to act in the opposite face having equal magnitude but opposite direction. Any
stress component is considered as positive if its direction and the direction of the outward
normal of the plane where it acts are both positive or both negative (according to the axis
system). At this level there are three normal stress components σx, σy and σz and six shear
stress components τxy, τyx, τyz, τzy, τzx and τxz. The six shear stress components can be reduced
to three by considering equilibrium of the element for moment about x, y and z which gives
τxy = τyx, τyz = τzy and τzx = τxz.

Fig. 4

3
 Fig. 5

6. Strain
The deformation of the body is indicated by strain which is measured by the relative change
of position between the adjacent particles. Similar to the case of stress, there are three
components of the normal strain (εx, εy and εz) and three components of the shear strain (γxy,
γyz and γzx) at a point. The normal strain at a point along a particular direction is measured by
the change in length in that direction of an infinitesimal line element taken at that point in
that direction divided by its original length. The shear strain at a point corresponding to a
particular direction and perpendicular to that is measured by the change in angle between two
line elements taken at that point which are aligned in these directions. For normal strain,
elongation is considered as positive while reduction of angle between the two line elements at
right angle is considered as positive for shear strain. Fig. 6 shows two line elements PA and
PB at a point P which are aligned in the direction of x and y respectively before deformation.
They have been shifted to P'A' and P'B' after deformation as shown. The change in length of
PA measured in the direction of x is (u + ∂u/∂x dx - u) ∂u/∂x dx (see Fig. 6). Thus the normal
strain in the direction of x is εx = (∂u/∂x dx) / dx = ∂u/∂x. The change in orientation of the line
element PA is (v + ∂v/∂x dx - v) / dx = ∂v/∂x while it is ∂u/∂y for the line element PB (see Fig.
6). Thus the shear strain γxy = ∂u/∂y + ∂v/∂x, which is the total change in angle between the
two line elements. The other strain components can be easily obtained following the similar
approach. These are basically strain-displacement relationship which is written for all the
strain components as follows
εx = ∂u/∂x, εy = ∂v/∂y and εz = ∂w/∂z (1)
γxy = ∂u/∂y + ∂v/∂x, γyz = ∂v/∂z + ∂w/∂y and γzx = ∂w/∂x + ∂u/∂z. (2)

4
 Fig. 6

7. Stress-Strain Relationship
According to Hooke's law, stress is proportional to strain up to a certain limit where the
constant of proportionality is the material constant. For normal stress and normal strain, the
material constant is the modulus of elasticity E while it is modulus of rigidity G for shear. So
it may be written as
σ = Eε (3)
τ = Gγ . (4)
Again a component of normal stress has some contribution on normal strain in a direction
perpendicular to the direction of that stress. The ratio of this strain in the transverse direction
to the strain in the direction of the stress is called Poisson's ratio (ν). The strain in the
transverse direction is negative if the stress and positive. With the help of Poisson's ratio ν,
the modulus of rigidity G may be expressed in terms of modulus of elasticity E as [1]
G = E/2(1+ν). (5)
Based on the above discussion, the components of strain in the different directions due to σx
may be written as
εx = σx /E , εy = -ν σx /E and εz = -ν σx /E.
The contribution of σy and σz may be obtained in the similar manner. Thus εx, εy and εz
produced by σx, σy and σz may be obtained by superposing the contribution of the individual
stress components as
1 1 1
εx = [σ x − ν (σ y + σ z )] , ε y = [σ y − ν (σ z + σ x )] and ε z = [σ z − ν (σ x + σ y )] . (6)
E E E
The above equation shows that the components of normal stress and normal strain are having
coupling between them but it is not found in the case of shear where γxy, γxy and γxy may be
simply expressed as
γ xy = τ xy / G , γ yz = τ yz / G and γ zx = τ zx / G . (7)

5
In the above equations (6-7), the components of strain are expressed in terms of stress. The
reverse relationship may be easily obtained [2] and it may be expressed as

E (1 − ν )  ν
σx = εx + (ε y + ε z ) , σ y = E (1 − ν ) ε y + ν (ε z + ε x ) and
(1 + ν )(1 − 2ν )  1 −ν  (1 + ν )(1 − 2ν )  1 −ν 
E (1 − ν )  ν
σz = εz + (ε x + ε y ) ; (8)
(1 + ν )(1 − 2ν )  1 −ν 
τ xy = Gγ xy , τ yz = Gγ yz and τ zx = Gγ zx . (9)

With the help of equation (5), the above equations (8-9) may be expressed in matrix form as
{σ } = [D]{ε } (10),

where {σ } is the stress vector, {ε } is the strain vector and [D ] is the elasticity or rigidity
matrix which are as follows

{σ } = [σ x σ y σ z τ xy τ yz τ zx ]T , {ε } = [ε x ε y ε z γ xy γ yz γ zx ]T and

 ν ν 
 1 1 −ν 1 −ν
0 0 0 
 ν ν 
 1 0 0 0 
1 − ν 1 −ν 
 ν ν
1 0 0 0 

[D] = E (1 − ν ) 1 − ν 1 −ν
(1 − 2ν )  .
(1 + ν )(1 − 2ν )  0 0 0 0 0 
 2(1 − ν ) 
 (1 − 2ν ) 
 0 0 0 0
2(1 − ν )
0 
 
 0 (1 − 2ν ) 
0 0 0 0
 2(1 − ν ) 

8. Type of Structural Problems and their modelling

Any structure or structural member may be considered as a three-dimensional object having
all the three dimensions but there are many cases where all the dimensions are not
predominant as they are not of the same order. The three dimensional modelling of such
structures is unnecessary as the involvement is too much in a three dimensional analysis. In
some cases, the structural problem may be reduced to a two dimensional problem where one
of the dimensions is very small compared to the others or it may be reduced to a one
dimensional problem in some cases where only one of the dimensions is predominant. There
is another category of structures where one of the dimensions may be very large compared to
the others and there is no variation of geometry, loading and boundary condition along that
direction. In these cases, the behaviour of the structure does not vary in that direction which

6
helps to eliminate the effect of that direction in the structural modelling and it helps to reduce
the size of the problem. In the following sections, the different types of structural problems of
basic category are described.

8.1. Three dimensional model

The three dimensional model is the most general one for analysing a structure or a structural
member. It has all six components of stress (σx, σy, σz, τxy, τyz and τzx) and strain (εx, εy, εz, γxy,
γyz and γzx) and the relationship between them is given in equations (6-9). In this case, the
structural deformation is defined by u, v and w which can be used to get the six strain
components using equations (1-2). Unless it becomes absolutely necessary, the three
dimensional analysis is usually avoided as the computational involvement is huge. Usually a
three dimensional model is not required to solve a common structural engineering problems
but it has uses in some specific problems such as analysis of stress distribution in rock or soil
mass, complex structural joints, machine components and some similar object.

8.2. Two dimensional model

The two dimensional model may be applied to the analysis of a large number of structural
engineering problems without any significant loss of accuracy in the solution of the problems.
In this case, the effect of one of the dimensions can be eliminated which helps to reduce the
problem size considerably. The different types of two dimensional problems are discussed in
the following sections.

8.2.1. Plane stress problem

The condition of a structure having one of its dimensions in a particular direction, say z, is
very small compared to its other dimensions and subjected to loading parallel to x-y plane and
uniformly distributed in the z direction as shown in Fig. 7 is defined as the plane stress
condition. Thus the structure is a flat plate having a very small thickness t in the direction of z
and it is subjected to loading in its plane uniformly distributed over the thickness. In this
situation the stress components σz, τyz and τzx are zero on both the faces (parallel to x-y plane)
of the plate and it is assumed that they are also zero within the plate. Thus the stress
components remains are σx, σy and τxy and they are assumed to be independent of z i.e., they
do not vary along the plate thickness. The problem is defined as plane stress problem as all
the effective stress components are dependent on x and y i.e., the parameters of a particular
plane. Now substituting σz, τyz and τzx as zero in equations (6-7), the stress components may
be expressed in terms of strain components as
E E
σx = [ε x + νε y ] , σ y = [ε y + νε x ] and τ xy = Gγ xy . (11)
1 −ν 2
1 −ν 2
With the help of equation (5) the above equation may be written in the matrix form as

7
σ x  1 ν 0  εx 
  E ν 1   ε  or {σ } = [D ]{ε } .
σ y  =  0  y  (12)
τ  1 − ν
2

 xy   0 0 (1 − ν ) / 2 γ xy 

In analysing structures like shear wall in tall buildings, bulkhead in aircraft and ships and
some similar objects, the structure can be satisfactorily modelled by considering it as a plane
stress problem.

Fig. 7

8.2.2. Plane strain problem

A similar simplification is possible for the other extreme where the dimension of the structure
in the z direction is very large as shown in Fig. 8. If the geometry, loading and boundary
condition do not vary in the longitudinal direction i.e., z direction and the loading acts
perpendicular to the direction of z, it may be assumed that the condition at any section
perpendicular to z is identical to that at any other section perpendicular to z. In this situation,
the investigation of a slice between two adjacent sections is sufficient to get the behaviour of
the whole structure. In that case, the boundary conditions at the end sections should be such
that their movement is absolutely free in x-y plane but fully restrained in the direction of z. It
makes the displacement component w along z as zero at any section. Moreover, the variations
of other displacement components u and v are independent of z and they are simply function
of x and y only. Substituting the above conditions in equations (1-2), the strain components εz
(= ∂w/∂z), γyz (= ∂v/∂z + ∂w/∂y) and γzx (= ∂w/∂x + ∂u/∂z) become zero. Again the nonzero
strain components εx (= ∂u/∂x), εy (= ∂v/∂y) and γxy (= ∂u/∂y + ∂v/∂x) are dependent on the
parameters of x-y plane which justify the name of the problem i.e., plane strain problem. With
the help of third component of equation (6) and taking εz as zero, the normal stress
component in the longitudinal direction σz may be expressed in terms of σy and σz as
σ z = ν (σ x + σ y ) .
Substituting the above equation in first two components of equation (6), the normal stress
components may be expressed in terms of normal strain components as
E (1 − ν )  ν  E (1 − ν )  ν 
σx = εx + ε y  and σ y = εy + εx . (13)
(1 + ν )(1 − 2ν )  1 −ν  (1 + ν )(1 − 2ν )  1 −ν 

8
The effective shear stress component τxy may be expressed in terms of corresponding shear
strain component with the help of equations (5) and (7) as
E
τ xy = γ xy . (14)
2(1 + ν )
Now the above equations (13-14) may be expressed in terms of matrix form as

 ν 
 1 1 −ν
0 
σ x   ε x 
  E (1 − ν )  ν  
σ y  = 1 0   ε y  or {σ } = [D ]{ε } . (15)
τ  (1 + ν )(1 − 2ν ) 1 − ν  
 xy   1 − 2ν  τ xy 
 0 0
2(1 − ν ) 

The above equation may also be obtained simply by substituting εz, γyz and γzx as zero in
equation (10). The plane strain model can be satisfactorily applied to the analysis of long and
straight pipe, tunnel, dam, retaining wall, and some similar structures.

Fig. 8

8.2.3. Axisymmetric problem

A structure or a solid is defined as solid of revolution or axisymmetric solid if its geometry
can be generated by rotating a plane surface about a nonintersecting axis or line in its plane
through an angle of 2π as shown in Fig. 9. The most convenient way of representing such
structural geometry is to use cylindrical coordinate system ( x, y,θ ), where x may be in the
radial direction, y is along the axis of rotation and θ is the rotation about y. In such a
structure, the geometry does not vary in the direction of θ and the structure is always
symmetrical with respect to any plane passing through y. If the loading and boundary
conditions imposed on such structure do not vary in the direction of θ i.e., symmetrical with
respect to any plane passing through y, the structural problem is defined as asisymmetrical
problem. The situation is more or less identical to that of plane strain problem where the
direction z in plane strain corresponds to θ in the present case. Thus the behaviour of the

9
whole structure can be obtained by analysing a single section. Now the strain displacement
relationship defined in terms of cartesian coordinate system (1-2) may be expressed in terms
of cylindrical coordinate system as follows [2]
εx = ∂u/∂x, εy = ∂v/∂y and εθ = u/x + ∂w/(x∂θ)
γxy = ∂u/∂y + ∂v/∂x, γyθ = ∂v/(x∂θ) + ∂w/∂y and γθx = ∂w/∂x - w/x+ ∂u/(x∂θ).
In the present case, w is zero at any section and u,v are function of x,y which is similar to the
case of plane strain problem and it may be substituted in the above equation and it leads to
εx = ∂u/∂x, εy = ∂v/∂y and εz = u/x (16)
γxy = ∂u/∂y + ∂v/∂x, γyz = 0 and γzx = 0. (17)
The stress strain relationship can be obtained by substituting the above conditions in equation
(10) with appropriate modification of the stress and strain components as

σ x   1 ν /(1 − ν ) ν /(1 − ν ) 0  ε x 
σ  ν /(1 − ν )  ε 
 y E (1 − ν )  1 ν /(1 − ν ) 0   y 
 =
σ θ  (1 + ν )(1 − 2ν ) ν /(1 − ν ) ν /(1 − ν ) 1 0   εθ 
τ xy   
 0 0 0 (1 − 2ν ) / 2(1 − ν ) γ xy 

or {σ } = [D ]{ε } . (18)
The axisymmetric model can be applied to the analysis of circular footing resting on soil,
cylindrical structures and so on.

y (v)

x (u)

Fig. 9

8.2.4. Plate bending problem

If a thin flat plate is subjected to loading perpendicular to the plane of the plate and it is
properly supported, the plate undergoes bending [3]. This structural problem is known as
plate bending problem. There are two types of plate bending theory; one is the thin plate

10
theory or classical plate theory and the other one is the thick plate theory. If the span to
thickness ratio of a plate is 20 or higher, it is considered as a thin plate while a plate having
this ratio lower than 20 is considered as thick plate. The basic difference between these two
theories lies with the effect of transverse shear deformation which is considered in thick plate
theory but not in thin plate theory. Both the plate theories are explained below.

8.2.4.1. Thin plate theory

For the derivation of the plate theory, the middle plane of the plate is considered as the
reference plane, which may be taken as the x-y plane. The basic assumptions made in this
theory are: 1) The middle plane remains the neutral plane having no strain during bending, 2)
The normal stress component σz is neglected as it is very small compared to other stress
components, 3) the transverse displacement w i.e., displacement in the z direction has no
variation in the thickness or z direction and 4) A linear element normal to the middle plane
before bending remains linear and normal to the middle plane after bending as shown in Fig.
10. With the help of last assumption, the displacement components u and v of a point, say Q,
located at distance of z from the reference plane may be expressed as
u = − z ∂w / ∂x and v = − z ∂w / ∂y (19)

where w may be considered as the displacement of reference plane at P in the direction of z.

The above expressions conforms that the transverse shear strains γyz and γzx are zero and it can
be verified by substituting them in equation (2). The other strain components, which are
effective, may be obtained by taking appropriate derivatives (1-2) of u and v (19) as follows

ε x = ∂u / ∂x = − z ∂ 2 w / ∂x 2 , ε y = ∂v / ∂y = − z ∂ 2 w / ∂y 2
and γ xy = ∂u / ∂y + ∂v / ∂x = −2 z ∂ 2 w / ∂xy . (20)

In the second assumption, σz is taken as zero to consider a plane stress condition for the
different layers parallel to the reference plane. Using equation (11) and the above equation
(20), the stress components may be expressed as
Ez Ez
σx = − [∂ 2 w / ∂x 2 + ν ∂ 2 w / ∂y 2 ] , σ y = − [∂ 2 w / ∂y 2 + ν ∂ 2 w / ∂x 2 ]
1 −ν 2
1 −ν 2

and τ xy = −2 zG ∂ 2 w / ∂x∂y (21).

The above equations (21) indicate that the variation of stress components in the direction of
plate thickness is linear. Utilising this well defined variation, the stress resultants per unit
length Mx, My and Mxy (see Fig. 11) produced by σx, σy and τxy respectively may be expressed
as
t/2
Et 3  ∂ 2 w ∂2w 
Mx = ∫ σ x z dz = −  + ν ,
−t / 2 12(1 − ν 2 )  ∂x 2 ∂y 2 

11
 t/2
Et 3  ∂ 2 w ∂2w 
M y = ∫ σ y z dz = −  + ν 2 
−t / 2 12(1 − ν 2 )  ∂y 2 ∂x 
Gt 3 ∂ 2 w
t/2
and M xy = − ∫ τ xy z dz = (22)
−t / 2
6 ∂x∂y

where M xy is the twisting moment and M x and M y are the bending moments. Using
equation (5), the above equation may be expressed in the matrix form as

 M x   D νD 0   − ∂ w / ∂x 
2 2

     − ∂ 2 w / ∂y 2  or {σ } = [D ]{ε }
 M y  = νD D 0   (23)
M   0 0 D (1 − ν ) / 2 2 ∂ w / ∂x∂y 
 2
 xy  

where D = Et 3 / 12(1 − ν 2 ) is called the flextural rigidity of the plate. Equation (23) gives the
moment curvature relationship, which may be considered as the stress strain relationship in
plate bending problem. It is now clear that the present problem became a two dimensional
problem as the curvature or generalised strain components are function of w, x and y which
are the parameters of the reference plane. The plate problem has a large number of
applications in various engineering constructions.

Fig. 10

12
 Qy
Qx
M yx
M xy Mx
y x
z P My

dy
dx ∂Mxy
∂My Mxy + dx
My + dy ∂x
∂y
∂ Qx
Qx + dx
∂ M yx ∂x
M yx + dy ∂Mx
∂y Mx + dx
∂ Qy ∂x
Qy + dy
∂y

Fig. 11

8.2.4.2. Thick plate theory

In thick plates, the effect of shear deformation is taken in to account in addition to the
bending deformation taken in thin plate theory. As the transverse shear stresses vary
parabolically in the direction of plate thickness, the shear strain is also produced in the similar
manner. It becomes maximum at the neutral level and reduced to zero at the free surfaces.
The deformation due to shear strain may be superposed with the bending deformation and it
is shown in Fig. 12. Now the line element normal to the middle plane before bending became
distorted after bending and it has made the problem complex. Mindlin has simplified the
problem with the recommended that the shear strain is uniform over the plate thickness which
has helped to make line element straight but not necessarily normal to the deflected middle
surface as shown in Fig. 12. Now the bending rotations θx and θy are not simply the slope of
the deflected middle plane and they may be represented as

13
 Fig. 12

θ x = ∂w / ∂x − φ x and θ y = ∂w / ∂y − φ y (24)

where φx and φy are the average shear rotations.

In the present case the displacement component u and v may be expressed as
u = − zθ x and v = − z θ y . (25)

With the procedure followed in the previous section, the stress resultants Mx, My and Mxy may
be expressed as
M x = − D (∂θ x / ∂x + ν∂θ y / ∂y ) , M y = − D (∂θ y / ∂y + ν∂θ x / ∂x )

D (1 − ν )
and M xy = (∂θ x / ∂y + ∂θ y / ∂x ) . (26)
2
With the help of equations (7) and (24), the average value of the transverse shear stress may
expressed as
τ zx = G (∂w / ∂x − θ x ) and τ yz = G (∂w / ∂y − θ y ) . (27)

Now the shear force Qx and Qy may be written as

Q x = Gtβ (∂w / ∂x − θ x ) and Q y = Gtβ (∂w / ∂y − θ y ) . (28)

where β is the shear correction factor and it is obtained by equating the strain energy due to
shear strain corresponding to average value and actual variation. For a plate problem, it has a
value of 5/6.

8.2.5. Shell theory

Shell structures are similar to plates in the sense that the thickness is very small compared to
its other dimensions but the shell structure has a curved surface whereas a plate has an
absolutely flat surface. A shell may be subjected to any form of loading and it need not be

14
normal to the shell surface like that of plates. Thus the situation in this structure is much more
general as well as complex. Due to its general nature, a shell structure experiences bending
action as well as membrane action. Thus a plate bending problem or a plane stress problem
may be considered as a special case of this problem. As the structure is quite complex, the
theory and finite element application of it is presented in a separate chapter.

8.3. One dimensional problems

The problem associated with a structure or a structural member having one of its dimensions,
say length, is predominant while other two dimensions are very small compared to its length
comes under the category of one dimensional problem. In most of the cases, a large number
of slender members i.e., one dimensional structural members are found to be combined or
connected in a desired sequence and it gives a large structure. These structures may be
defined as skeleton structures having a network of slender members. The skeleton structures
may be broad divided in to two groups. They are 1) truss and 2) frame structures which
described in the following sections.

8.3.1. Truss
A truss is a network of a number of slender members connected only at their ends by
frictionless hinge joints and sustains loads only at the joints as shown in Fig. 13. The force
produced in any member is due to the relative movements between two joints at its ends as
the member does not sustain any load. Furthermore the joints are frictionless hinges which
ensure that the member forces are simply axial tension or axial compression. So a truss
member is free from bending. If all the members in a truss lie in a particular plane and the
loads act in that plane, it is defined as plane or two dimensional (2D) truss otherwise it is
space or three dimensional (3D) truss. In such a structure, its behaviour of the individual
members can be completely defined if the movements or displacements of the joints are can
be determined. As a joint connects a number of members oriented in different directions, its
resultant displacement may have any direction which may be conveniently represented in
terms of its components corresponding to a suitable axis system. Thus the number of
independent displacement components at a joint may be taken as 2 (u and v) in 2D truss and 3
(u, v and w) in 3D truss.

15
 Fig. 13

8.3.2. Frame
Frame is also a network of a number of slender members connected at their ends but the
joints are rigidly connected and the loads may act at anywhere within the structure (see Fig.
14). A joint is treated as a rigid one where the angle between the member ends meeting the
joint do not change with the structural deformation due to the external loading. Thus the
rotation of the member ends meeting at a joint is identical which is considered as the rotation
of the joint. Similar to the case of truss, a frame is defined as plane frame (2D) if all the
members lie in a particular plane and the loads act in that plane. A member of a plane frame
free from any load on it is somewhat similar to the truss member except the type of joints at
its end. These joints will produce bending in the member as they resist the member end to
rotate freely (see Fig 15) to get a straight configuration. In that case, the member ends are
neither fully free to rotate nor fully restrained against rotation. The bending is further
produced in a member when it sustains load on it. Thus the member forces are shear force
and bending moment produce by bending of the member in addition to the axial force. In this
case, the independent displacement components at the joints are u, v and θ if the frame is
defined in x-y plane. Similar to the case of plane frame, there is another category of frame
structure known as grillage as shown in Fig. 16 where all the members lie in a particular
plane, say x-y, but the loads act in the direction of z i.e., perpendicular to the plane of the
structure. In a member of such structure, bending is produced by the loading acts on it and
also by the influence of the members connected at its end due to their continuity. Furthermore
the members encounter twisting due to the bending of the members connected at its end but
having different orientation. Thus the member forces in a grillage member are shear force,
16
bending moment and twisting moment or torsion and the independent displacement
components at the joints are w, θx and θy. A combination of all the aspects of plane frame and
grillage is found in space frame. In this case the member forces are axial force, twisting
moment and a pair of shear force and bending moment produced by biaxial bending and the
independent displacement components at the joints are u, v, w, θx, θy and θz. Thus a plane
frame or a grillage may be considered as a special case of space frame. Similarly, a beam
having any number of span as shown in Fig. 17 may be considered as a special case of plane
frame, grillage or space frame where all the members are aligned in specific direction, say x,
and the loads act perpendicular to that i.e., y. In that case the member forces are shear force
and bending moment and the independent displacement components at the joints are v and θ .

Fig. 14

Fig. 15

17
 Fig. 16

Fig. 17

9. Degrees of Freedom
The degrees of freedom (DOF) of a system are the number of independent parameters
required to define the system completely. In the present case, a structure or a solid is the
system and the parameters are usually the displacement components. As an example, a
structural system consists of three rigid members as shown in Fig. 18 may be considered. The
degrees of freedom of the system are two and they are m and n. With these two parameters,
the displacement at any point within the structure can be obtained easily. In any structural
mechanic problem, the basic unknown is the displacement. Once it is obtained, the strain can
be determined by taking its derivative and the strain is multiplied with the material constant
to get the stress. Thus the degrees of freedom are nothing but the displacement components in
the context of the present problem.

10. Discrete and Continuous System

The system shown in Fig. 18 is a discrete system as it can be entirely defined by the two
discrete parameters, m and n. On the other hand, an elastic continuum may be considered as a
continuous system. A continuum is defined as a domain in which matter exits everywhere
i.e., it consists of an infinite number of particles. In an elastic continuum the displacement
components of the individual particles are independent. Thus it has infinite degrees of
freedom. In that case, the problem can be defined using the mathematical fiction of an
infinitesimal, which leads to differential equations. The classical approach may be applied to
solve these equations and it leads to an analytical solution of the problem which may have
feasible solution only for some simple problems. In finite element analysis, the continuum is
divided in to a number of units known as elements and they are connected at some discrete
points known as nodes. Now the degrees of freedom are taken only at these nodes and the
displacement and other quantities within an element are approximated in terms of nodal
displacements of that element. It helps to represent the continuous system approximately in to
a discrete system having finite degrees of freedom which can be solved numerically using
computers even with a large number of DOF. In this context, two dimensional and three
dimensional structures come under the category of continuous system while skeleton
structures come under discrete system.

18
 Fig. 18

11. Strain Energy

The concept of strain energy is very important in structural mechanics. As a simple example,
a prismatic bar subjected to an axial tension as shown in Fig. 19 is considered. As the
deflection or elongation of the bar may be defined as δ ′ = P ′l ( AE ) , the load deflection
diagram will be linear where δ ′ is the deflection for a particular value of P ′ (see Fig. 19). If
the load is incremented by dP ′ , the deflection will be incremented by dδ ′ (see Fig. 19). In
this process, the load P ′ will do a positive work of P ′dδ ′ which can be represented by the
area of shaped strip in Fig. 19. This work will be stored in the bar in the form of energy
known as strain energy. If the load increment dP ′ is now withdrawn, the tip of the bar will
lift upward by an amount dδ ′ by releasing that energy P ′dδ ′ . The total energy stored in the
bar under a tensile load P is equal to the sum of area of all such elemental strip ( P ′dδ ′ )
between O and B and it may be represented by the area of the triangle OAB (see Fig. 19).
Thus the strain energy U may be written as U = (1 2)Pδ which may be redefined as

Fig. 19
1 1  P δ  1
U= Pδ =     Al = σε Al = U 0 Al ,
2 2  A  l  2

where σ = P / A is the stress, ε = δ / l is the strain, Al is the volume and U 0 is strain energy
per unit volume or strain energy density. Though the strain is constant over the entire bar in

19
this problem but the concept can be applied in any problem at the level of an infinitesimal
element as shown in Fig. 20. In that case strain energy due to σ x may be written as

dU = (1 2 )(σ x dydz )(ε x dx ) = 1 2 σ x ε x dxdydz

Considering all the components of stress and strain, the strain energy density of a solid may
be written as
U 0 = (1 2 )(σ x ε x + σ y ε y + σ z ε z + τ xy γ xy + τ yz γ yz + τ zx γ zx ) . (29)

Fig. 20

12. Principal of Virtual Work

The principle is first explained with a particle which is in equilibrium under the action of a
system of forces. If a virtual displacement is given to the particle, the principle states that the
work done by the system of forces due to the virtual displacement will be zero and it can be
verified easily as follows.
If δu , δv and δw are the components of the virtual displacement and ∑ X , ∑ Y and
∑ Z are the summation of corresponding components of the forces, the virtual work may be
expressed as δu ∑ X + δv ∑ Y + δw ∑ Z and it will be zero as ∑ X = 0 , ∑ Y = 0 and ∑ Z = 0 .
In the case of the particle, the virtual displacement may have any value provided the
magnitude and direction of the forces remain unchanged.
Now consider an elastic body which is in equilibrium under the action of externally
applied forces. The elastic solid may be considered as a system of particles where all the
particles are subjected to a system of forces and they are all in equilibrium. Thus the total
virtual work due to external forces and internal forces is zero. Now following the concept
used in the previous section for the strain energy, the virtual work due to internal forces may
be written as

δWi = ∫ (σ xδε x + σ y δε y + σ z δε z + τ xyδγ xy + τ yz δγ yz + τ zxδγ zx )dxdydz (30).

V

20
The virtual work due to external forces which may be body force having components X, Y
and Z acting over a domain Ω, surface force having components X , Y and Z acting over a
surface area S and concentrated forces Pi having components Pix , Piy and Piz may be written
as

δWe = ∫ ( Xδu + Yδv + Zδw) dΩ + ∫ ( Xδu + Y δv + Z δw) dS

Ω S
n
+ ∑ ( Pix δu i + Piy δvi + Piz δwi ) (31),
i =1

where n is the number of concentrated forces.

The work done by internal forces is taken as negative as the work is against the mutual forces
between the particles. Thus the final expression for an elastic solid according to the principle
of virtual work may be written as
δWi − δWe = 0 or δWi = δWe (32).
In case elastic solids, the virtual displacement should be sufficiently small so as to satisfy the
material continuity, boundary conditions and other aspects.

13. Maxwell-Betti Reciprocal Theorem

The theorem states that the displacement at a point A in a direction r produced by a force P
applied at some another point B in a direction s and the displacement at B in the direction s
produced by P applied at A in the direction r are equal. It can be easily verified using the
concept of strain energy and it is not presented here as it is available in any standard text on
structural mechanics [2].

References
1. Zienkiewicz, O. C. (1977) The Finite Element Method, 3 rd Ed, McGraw-Hill Book Co.,
U.K.
2. Timoshenko, S.P. and Gooddier, J.N. (1970) Theory of Elasticity, 3rd Ed. McGraw-Hill
Book Co., Singapore.
3. Timoshenko, S.P. and Woinowsky-Kriger, S. (1970) Theory of Plates and Shells, 2nd Ed.
McGraw-Hill Book Co., Singapore.

21