1

Theory of Failure
1.1 INTRODUCTION
A structural or machine element subjected to increasing load (static or dynamic) fails and becomes
unserviceable at a certain value of the load. The failure may be (i) brittle failure (fracture) caused by
separation of material without measurable deformation due to loss of cohesive bond between the
adjacent molecules or (ii) ductile failure (yielding) manifested by first measurable inelastic deformation
as a result of slipping along critically oriented planes due to sudden breakdown of the molecular
structure of the material. However, it is important to note that brittle failure does not occur in brittle
material alone. It can take place in ductile material also depending on state of stress, speed and nature
of loading, shape and size of the element, thermal state of the element with respect to its critical
temperature (temperature above which resistance to slipping in less than resistance to separation of
material) and some other parameters. Again the above mentioned load may create uniaxial or multiaxial
principal stresses of same or opposite sign. In case of deformation it may cause change of volume,
change of shape or both. Depending on above factors the quantities such as (a) principal stresses (b)
principal strains (c) maximum shear stress (d) total elastic strain energy density (e) distortion energy
density (f) octahedral shear stress etc. will attain certain values at different points of the loaded body.
Out of the said points, at some critical ones, the values of the above parameters will reach their
maximum and eventually failure process of the structure will start from there. Now the question is
which of the aforesaid characteristic quantities plays the most important role in causing the said failure
of the element. To get the answer, the researchers conducted experiments like simple tension test of
ductile and brittle material (uniaxial principal stress), torsion test of solid circular shaft (pure shear), test
under uniform hydrostatic pressure (pure dilatation), test on tubes with both ends closed and subjected
to internal pressure and variable axial load (multiaxial principal stress) and many others. Out of them
the results obtained from the stress-strain graph of mainly first three tests only were found to be reliable
enough to be used to predict the failure criterion of different multiaxial stress system caused by different
types of loading. Other test results were not reliable for the above purpose because their test conditions
could not be made ideal for some practical difficulties due to which many undesirable forces like
friction etc. appeared along with the desirable ones during experiments. Depending mainly on the
results obtained from first three tests stated above, different researchers, on the basis of their own
1

Advanced Strength of Materials

perception and analysis, put forward their theories stating the critical condition to be satisfied for the
onset of failure process. These theories which deal with the relation between a strength characteristic
under different forms of combined loading and analogous strength characteristic under simple loading
are known as theories of failure. All the theories as will be discussed and seen later predict the same
elastic failure load for uniaxial loading. So it is impossible to determine which of theories and
characteristic quantities is responsible for failure in this case. On the other hand, as no test data for
different multiaxial loading are used in the formulations of the theories, different theories will predict
different elastic failure load for multiaxial stress system. Naturally the theory which predicts a failure
load nearest to that obtained from actual test data under similar loading of the element shall be the most
acceptable one for the given situation. The statement of the theories, explanation about their applicability
and limitations with respect to different practical cases, to be taken up in the next part of the chapter,
will throw more light into the matter.

1.2 IMPORTANT EXPERIMENTAL RESULT

(a) Brittle material fails by separation of material due to breakdown of the cohesive bond.
(b) In case of ductile material there is a temperature known as critical temperature. Below this
temperature failure takes place by separation of material (brittle fracture) and above this temperature
failure takes place by slipping or sliding (yielding). This critical temperature changes with speed
of loading, state of stress, shape and size of the specimen etc.
(c) Yield strength in shear (e) obtained from pure torsion test  0.57 yield strength in tension (e)
for ductile material obtained from uniaxial tensile test.
(d) Hydrostatic pressure up to a very high value produces only small change in volume and practically
no inelastic deformation. As such the part of total elastic potential energy absorbed due to change
of volume caused by hydrostatic stress (1 = 2 = 3) plays no role in the onset failure process.
(e) The stress at which brittle failure takes place in a ductile material is almost independent of
temperature but the stress at which ductile failure takes place decreases nonlinearly with temperature.
The above two failure stress versus temperature graphs intersect at the point of critical temperature.
(f) Taking e as the elastic strength in tension (elastic limit or yield stress) and e, the elastic strength
in compression (i) for ductile material e  e and (ii) for brittle material e < e.
(g) In 2-D case also if equal tension or compression is applied in two perpendicular direction, it will
have almost no influence in creating inelastic action or failure.

1.3 STATEMENT OF THE THEORIES OF FAILURE

There are mainly six theories of failure. Apart from these, there are a few others which are either
modified version of some of the above theories or independent ones. The main theories of failure are:
(a) Maximum Principal Stress Theory
This theory was proposed by an English engineer W.T.M. Rankine. It states that for any state of stress
in a body failure will start when at any point of the body at least one of the principal stresses reaches
the value at which failure occurs in a simple uniaxial tension (or compression) test on the same material.
Here failure means yielding or fracture whichever occurs first.
(i) In a 3-D case if 1, 2 and 3 are three principal stresses then the governing equations for starting
the failure process are.
1 = e, 2 = e, 3 = e, 1 = e, 2 = e and 3 = e
(1.3.1)

Theory of Failure

It should be noted that out of the above six equations at least one equation has to be satisfied to start
the failure process.
Considering the variables 1, 2 and 3 as coordinate axes in 3-D system, the above equations
represent six planes. If e = e, the planes will enclose a cubical space. At different points of a loaded
element the principal stresses will be such that their representative position in the above stress coordinate system are either inside the above cube (safe against failure) or on the surface of the cube
(failure has just started) or outside the cube (have already failed).
(ii) In 2-D case [Fig. 1.3(a)], 3 = 0 and s1 , s 2 =

sx + sy
2

sx - sy
2

+ t

Here governing equations of failure are

1 = e, 2 = e, 1 = e and 2 = e
(1.3.2)
Out of these four equations at least one equation has to be satisfied to start the failure process. The
above equations will represent four straight lines enclosing a square area of safe zone against failure
[Fig. 1.3 (b), (c)].
s2
se

sy

se
se

sy
t

se

sx

s1

sx

se

sx

se

se

sy
(a)

For brittle material

(b)

se

For ductile material

(c)

Fig. 1.3

Special Case: When a solid circular shaft of diameter d is subjected to uniform bending moment
M and torque T.
sx =

M d 32 M
T d 16T
, s y = 0 and t = = 3
=
I 2 pd 3
J 2 pd
2

s
s
16
s1 , s 2 = x x + t 2 = 3 M M 2 + T 2

2
2
pd

(1.3.3)

These principal stresses must satisfy the failure conditions stated above to start the failure process in
the shaft.
Although this theory has been proposed to predict all types of failure, experiments suggest that its
suitability is limited to mainly brittle failure.
(b) Maximum Principal Strain Theory
This theory was put forward by B. de Saint Venant, the great French researcher in the field of elasticity.
It states that in a body subjected to any state of stress failure will start when magnitude of any of the

Advanced Strength of Materials

principal strains at any point of the body reaches the value equal to that required to start failure process
in a uniaxial tensile (or compressive) test on the same material.
(i) In 3-D case, the governing equations for starting failure process are
s
1
s
1
e1 =
s1 - n ( s 2 + s 3 ) = - e
s1 - n (s 2 + s 3 ) = e
E
E
E
E
s e
s e
1
1
e2 =
s 2 - n (s 3 + s1 ) =
s 2 - n (s 3 + s1 ) =
and e 2 =
E
E
E
E
s
1
s
1
e3 =
s 3 - n ( s1 + s 2 ) = e
e3 =
s 3 - n (s1 + s 2 ) = - e

E
E
E
E
which lead to the equations
e1 =

{
{
{

{
{
{

}
}
}

s1 - n (s 2 + s 3 ) = s e
s 2 - n ( s 3 + s1 ) = s e
s 3 - n ( s1 + s 2 ) = s e

}
}
}

s1 - n (s 2 + s 3 ) = -s e

and s 2 - n (s 3 + s1 ) = -s e

s 3 - n (s1 + s 2 ) = -s e

(1.3.4)

This theory suggests that out of the above six equations at least one has to be satisfied to start the
failure process. The above six equations represent six different planes in 123 coordinate axes
system. These planes will enclose a volume of space. The significance of any point within this volume,
on the surface of the planes and outside the said volume is exactly same as that explained in Art.
1.3(a)(i).
(ii) In 2-D case where 3 = 0, the governing equations for the onset of the failure process are
s
1
s1 - ns 2 } = e
{
E
E
se
1
e 2 = {s 2 - ns1 } =
E
E
e1 =

1
e1 = {s1 - ns 2 } = - e

E
E
and

s
1

e 2 = {s 2 - ns1 } = - e

E
E

s2
B

B2

B1
A2
A

C1
A1

D1

C
C2

D2

Fig. 1.3(d)

s1

A1[-se, 0], B[0, se]

n
B1[0, se], A[- se , 0]
n
C1[se, 0], D[0, - se]
n
D1[0, -se], C[ se , 0]
n

Theory of Failure

which lead to the equations

s1 - ns 2 = s e
s - ns 2 = -s e
(1.3.5)
and 1

s 2 - ns1 = s e
s 2 - ns1 = -s e
Here also out of the above four equations at least one equation has to be satisfied to start the failure
process. The above equations will also represent the equations of four straight lines enclosing an area
which represent safe zone against failure [Fig. 1.3(d)]. In drawing the graph it has been assumed that
e = e.
Although this theory removes certain weaknesses of Principal stress theory, still it fails to answer
many pertinent questions. As such it serves as a better option to predict brittle failure but not all types
of failure.
(c) Maximum Shear Stress Theory
This theory is known as Trescas theory of failure after the name of the great researcher H. Tresca. It
is sometimes called Coulombs or Guests theory due to their contribution in this regard. This
theory states that for any state of stress, failure will start when maximum shear stress at any point is
equal to the maximum shear stress at yield point in a simple tension (or compression) test on the same
material.
s
- s min
. The planes on which
We know from the concept of Mohrs circle of stress that t max = max
2
max occurs are known as slip planes and they are inclined at 45 to principal planes. So the maximum
s
shear stress in uniaxial tension test t max = max . This theory indirectly suggests that yield stress in
2
s
shear = Half of the yield stress in simple tension (or compression) i.e. t e = e . It means that failure
2
se
will start when t max =
.
2
(i) In 3-D case the governing equations are derived as follows:
s
s1 - s 2
= t e = e or s1 - s 2 = s e .
2
2
Proceeding in this way we get six following equations at least one of which must be satisfied to
commence the yielding or failure process.
s1 - s 2 = s e
s1 - s 2 = - s e

s 2 - s 3 = s e and s 2 - s 3 = - s e
s 3 - s1 = s e
s 3 - s1 = - s e

(1.3.6)

As this theory results from the test data on ductile material, elastic limit (or yield stress) in tension
and compression have been taken equal. The above six equations represent six planes which enclose a
volume of space in 1 2 3 coordinate system. The significance of points within the volume, on the
surface of the planes and outside the volume is exactly same as that explained in Art. 1.3(a)(i) and
1.3(b)(i).
sx - sy
2

+ t

2
2
Here also we shall get six following equations at least one of which must be satisfied for the
commencement of the failure process.
(ii) In 2-D case where s 3 = 0, s1 , s 2 =

sx + sy

Advanced Strength of Materials

1 2 = e
2 3 = e or 2 0 = e
3 1 = e or 0 1 = e

and
or
or
or
or

1 2 =
2 3 =
2 0 =
2 =
3 1 =
0 1 =
1 =

e
e
e
e
e
e
e

Rewriting the equations, we get

1 2 = e
1 2 = e

2 = e
and
2 = e
(1.3.7)

1 = e
1 = e
The above six equations represent six straight lines in 1 2 coordinate system as shown in the
Fig. 1.3(e).
The adjacent diagram has same significance as similar diagrams in other theories.
s max - s min
s2
, in 2-D cases where 3 = 0
Again as t max =
(0, se)
2
(se, se)

t max =

s1
s
or 2 where 1 and 2 are of same sign
2
2

s1 - ( -s 2 )
s + s2
= 1
when they are of
2
2
opposite sign. This, in essence, provide us with same six
equations as mentioned earlier.
but t max =

Although this theory has certain inadequacies in the

matter of prediction of brittle failure, it has proved to be
a powerful tool in predicting the phenomenon of ductile
failure particularly in those cases where shear force
developed is of high magnitude.

(- se, 0)

s1
(se, 0)

(- se, - se)

(0, - se)

Fig. 1.3(e)

(d) Maximum Strain Energy Theory

This theory was put forward by Beltrami and Haigh. It states that the failure process will commence in
a body subjected to any state of stress when density of elastic potential energy absorbed by the body
at any point reaches the value required to be absorbed by a specimen of same material under simple
tension (or compression) test at a stress equal to its elastic limit or yield point.
(i) In a 3-D case, total elastic energy density or total elastic energy/unit volume
(Vd) =

1
( + 22 + 33)
2 1 1

1
n
(12 + 22 + 32)
( + + 31)
2E
E 1 2 2 3
As per this theory for failure to occur

Theory of Failure

7
2
1
n
s 2 (s e )
(21 + 22 + 23)
(12 +23 + 31) = e or
2E
E
2 E
2 E

If
e = e,
we get
(12 + 22 + 32) 2(12 +23 + 31) = e2
(1.3.8)
The above equation represents a curved surface enclosing a volume of space the meaning of which
is exactly same as the similar enclosed surface in other theories of failure.
(ii) In 2-D case where 3 = 0, the above governing equation reduces to
21 + 22 212 = e2
(1.3.9)
The above equation represents the equation of an ellipse as shown in the figure.
se

(0, se)

se

//////////

2(1 - n)

////
/////
/

/
///
////

se

////////////

,-

// //

se
2(1 - n)

45

//////////////

//
/
/
////
//

(se, 0)

// /

(- se, 0)

2(1 - n)

/
///
/
/
/
////
///
/////////// (0, - se)

2(1 - n)

Fig. 1.3(f)

Like other theories of failure the enclosed area by the above graph indicates the safe zone against
failure. Unfortunately this theory failed to predict failure load in most of the multiaxial stress system
satisfactorily. This is mainly because it did not take into account the fact that out of the total energy
absorbed in a system due to dilatation and distortion, the part connected with dilatation does not play
any role in causing failure.
(e) Maximum Distortion Energy Theory
This theory was established and developed through the analytical work of Hencky, von Mises and
Huber and experimental work of Bridgman. It states that a body under any combined stress system will
fail if at any point distortion energy per unit volume is equal to the distortion energy per unit volume
of a test specimen of same material under uniaxial tension (or compression) at a stress equal to its
elastic limit or yield stress.
(i) Let us consider a 3-D principal stress system [Fig. 1.3(g)]
We can write

s1 =

s1 + s 2 + s 3 s1 - s 2 s 3 - s1
s - s 2 s 3 - s1
+
=s+ 1
3
3
3
3
3

s2 =

s1 + s 2 + s 3 s 2 - s 3 s1 - s 2
s - s 3 s1 - s 2
+
=s+ 2
3
3
3
3
3

Advanced Strength of Materials

8
s3
s

s1

s2
s2 - s3

s1 - s2
3
(Pure
dilatation)

s1 - s2
3

(Pure shear)

s 3 - s1
3
s3 - s1

+
(Pure shear)

s2 - s3

(Pure shear)

Fig. 1.3(g)

s1 + s 2 + s 3 s 3 - s1 s 2 - s 3
s - s1 s 2 - s 3
+
=s+ 3
3
3
3
3
3
The above resolution of 1, 2 and 3 into components are shown in Fig. 1.3(g). Again we know that
(1) two equal and opposite stresses on two mutually perpendicular planes creates only pure shear effect
and causes distortion or shear deformation only, (2) in small deformation analysis where F and
, the deformation caused by the resultant force is equal to algebraic sum of the deformation caused
by the components and (3) the mean stress s will cause only change in volume and no distortion.
Distortion or shear energy/unit volume
= Total strain energy/unit volume energy/unit volume due to change of volume only.
s3 =

1
1
( s1 e1 + s 2 e 2 + s 3 s 1 ) - 2 s e v
2

1 2
1 s1 + s 2 + s 3
s1 + s 22 + s 32 - 2 n (s1s 2 + s 2 s 3 + s 3 s1 )

2E
2K
3

E
and simplifying
3 (1 - 2 n )
we get distortion energy/unit volume
Putting K =

1 2
s1 + s 22 + s 32 - (s1s 2 + s 2 s 3 + s 3s1 )
6G

1
E
(s1 - s 2 )2 + (s 2 - s 3 )2 + (s 3 - s1 )2 where G = 2 (1 + n )
12G

According to distortion energy theory failure process will begin when

Theory of Failure

s 2e
1
2
2
2
s
s
+
s
s
+
s
s
=
( 1 2 ) ( 2 3 ) ( 3 1 ) 6G Putting s1 = s e , s 2 = s 3 = 0
12G

12 + 22 + 32 12 23 31 = e2
(1.3.10)
Like previous theories this equation also represents a curved surface enclosing a volume of space
having similar meaning as before.
(ii) In 2-D case, where 3 = 0, the governing equation reduces to
12 + 22 12 = e2
(1.3.11)

or

Note: The expression for distortion energy/unit volume can be obtained by just putting = 1/2 in the expression of total
elastic energy/unit volume after writing E = 2G(1 + ). This is because when = 1/2 change of volume is zero. As
a result dilatation energy/unit volume is zero. So expression for total energy/unit volume and that for distortion
energy/unit volume becomes same.

The above equation (1.3.11) represents an ellipse in 1 2 co-ordinate system as shown in the
Fig. 1.3(h) having significance same as similar diagrams in previous theories.
s2
(0, se)

//////////

/
///
////

//////////////

s1

// /

(se, 0)

//
////
////
/
/
/
/
////////
(- s , - s )
e

////
/////
/

////////////
45

// //

//
/
/
////
//

(- se, 0)

(se, se)

///

(0, - se)

Fig. 1.3(h)

This theory has been found to be wonderfully suited to predict the failure load in case of ductile
failure. In many cases, it is better than the theory of maximum shear stress.
(f) Theory of Maximum Octahedral Shear Stress
This is nothing but the theory of maximum distortion energy stated in a different way. It states that
failure process will commence in a body subjected to any stress system if the shearing stress on an
octahedral plane anywhere in the body is equal to the octahedral shear stress of a test specimen of same
material under uniaxial tension (or compression) at its elastic limit or yield point.
We know shearing stress on an octahedral plane of an element subjected to all three principal
stresses.
t *0 =

1
3

(s 1 - s 2 ) 2 + ( s 2 - s 3 ) 2 + ( s 3 - s 1 ) 2

where the octahedral plane is a plane inclined equally

to all axes a = b = g = cos -1

3
* Refer to Art. 24, Chapter 3 of Advanced Mechanics of Materials by Seely and Smith.

Advanced Strength of Materials

10

According to this theory

1
2
se
[(1 2)2 + (2 3)2 + (3 1)2]1/2 =
[Putting 1 = e, 2 = 3 = 0]
3
3
or
(1 2)2 + (2 3)2 + (3 1)2 = 2e2
or
12 + 22 + 32 12 23 31 = e2
(1.3.11)
This is exactly the same equation as in case of the maximum distortion energy theory.

(g) Other Theories of Failure

Although maximum shear stress theory is reasonably fine in predicting the failure load in case of ductile
material it is not accurate in case of brittle materials where compressive strength (e) is quite high in
comparison to its tensile strength (e). C. Duguet presented the above theory in a modified form to take
care of the above fact the graphical plot of which is shown in Fig. 1.3(i).
This theory was further refined by Griffith and Irwin.
s2
According to them apart from different parameters
influencing failure process, surface energy of microscopic tension cracks, if its release rate is more than
(se, se)
absorption rate, will speed up the propagation of cracks
and hasten the failure process. The effect of the surface
s1
energy of cracks is shown by dotted line in Fig. 1.3(h).
None of the above theories was suitable for assessing
the strength (mainly shear strength) of soil. The most
commonly used theory of incipient shear failure in soil
was first propounded by Coulomb and later generalized
by Otto Mohr. Mathematically Coulombs theory in (- s , - s )
e
e
this connection is a functional relation (assumed to be
linear) between normal stress on a plane and available
Fig. 1.3(i)
shear strength available on that plane. It is written as
e = C + tan where C and are empirical parameters known as apparent cohesion and angle
of shearing resistance (or angle of internal friction) respectively. It is known as Coulombs Failure,
Envelope.
t

te = C + s tan f
(Failure envelope)
f

C = 0 for sand, f = 0 for clay

s (Compression)
Failure Envelope of Coulombs Theory

Fig. 1.3(j)

While any Coulombs equation is valid for a given small range of with known boundary values, the
Mohrs curved envelope having equation e = f() is valid for all values of .