520 C H A P T E R 10 S T R A I N T R A N S F O R M AT I O N

*10.7 Theories of Failure

When an engineer is faced with the problem of design using a specific
material, it becomes important to place an upper limit on the state of
stress that defines the material’s failure. If the material is ductile, failure is
usually specified by the initiation of yielding, whereas if the material is
brittle, it is specified by fracture. These modes of failure are readily
defined if the member is subjected to a uniaxial state of stress, as in the
case of simple tension; however, if the member is subjected to biaxial or
triaxial stress, the criterion for failure becomes more difficult to establish.
In this section we will discuss four theories that are often used in
engineering practice to predict the failure of a material subjected to a
multiaxial state of stress. No single theory of failure, however, can be
applied to a specific material at all times, because a material may behave
in either a ductile or brittle manner depending on the temperature,
rate of loading, chemical environment, or the way the material is shaped
or formed. When using a particular theory of failure, it is first necessary
to calculate the normal and shear stress at points where they are the
largest in the member. Once this state of stress is established, the
principal stresses at these critical points are then determined, since each
of the following theories is based on knowing the principal stress.

Ductile Materials
Maximum-Shear-Stress Theory. The most common type of yielding
of a ductile material such as steel is caused by slipping, which occurs along
the contact planes of randomly ordered crystals that make up the
material. If we make a specimen into a highly polished thin strip and
45
subject it to a simple tension test, we can actually see how this slipping
causes the material to yield, Fig. 10–26. The edges of the planes of
slipping as they appear on the surface of the strip are referred to as
Lüder’s lines on
mild steel strip
Lüder’s lines. These lines clearly indicate the slip planes in the strip,
which occur at approximately 45° with the axis of the strip.
The slipping that occurs is caused by shear stress. To show this,
consider an element of the material taken from a tension specimen,
when it is subjected to the yield stress sY, Fig. 10–27a. The maximum
shear stress can be determined by drawing Mohr’s circle for the element,
Fig. 10–27b. The results indicate that

10 sY
tmax = (10–26)
Fig. 10–26 2
 10.7 THEORIES OF FAILURE 521

Furthermore, this shear stress acts on planes that are 45° from the planes T
sY
of principal stress, Fig. 10–27c, and these planes coincide with the
direction of the Lüder lines shown on the specimen, indicating that
indeed failure occurs by shear.
Using this idea, that ductile materials fail by shear, in 1868 Henri
Tresca proposed the maximum-shear-stress theory or Tresca yield
criterion. This theory can be used to predict the failure stress of a ductile
material subjected to any type of loading. The theory states that yielding
of the material begins when the absolute maximum shear stress in the
Axial tension
material reaches the shear stress that causes the same material to yield
when it is subjected only to axial tension. Therefore, to avoid failure, it is T (a)
required that tabs
max
in the material must be less than or equal to sY> 2,
where sY is determined from a simple tension test.
For application we will express the absolute maximum shear stress in
terms of the principal stresses. The procedure for doing this was
discussed in Sec. 9.5 with reference to a condition of plane stress, that is, s2  0 s1  sY
s
where the out-of-plane principal stress is zero. If the two in-plane A(0, 0)
90
principal stresses have the same sign, i.e., they are both tensile or both
compressive, then failure will occur out of the plane, and from Eq. 9–13,
sY
max 
2
s1 sY
tabs = savg 
2
max
2  (b)

y¿ sY x¿
If instead the in-plane principal stresses are of opposite signs, then tmax  sY
2 savg 
failure occurs in the plane, and from Eq. 9–14, 2
45
x
s1 - s2
tabs =
max
2

Using these equations and Eq. 10–26, the maximum-shear-stress theory (c)
for plane stress can be expressed for any two in-plane principal stresses s1 Fig. 10–27
and s2 by the following criteria:
s2

ƒ s1 ƒ = sY sY
r s1 , s2 have same signs
ƒ s2 ƒ = sY (10–27)

ƒ s1 - s2 ƒ = sY6 s1 , s2 have opposite signs s1

sY sY 10

A graph of these equations is given in Fig. 10–28. Clearly, if any point sY
of the material is subjected to plane stress, and its in-plane principal
stresses are represented by a coordinate (s1, s2) plotted on the boundary Maximum-shear-stress theory
or outside the shaded hexagonal area shown in this figure, the material
will yield at the point and failure is said to occur. Fig. 10–28
 522 C H A P T E R 10 S T R A I N T R A N S F O R M AT I O N

s3 Maximum-Distortion-Energy Theory. It was stated in Sec. 3.5 that

an external loading will deform a material, causing it to store energy
internally throughout its volume. The energy per unit volume of material
is called the strain-energy density, and if the material is subjected to a
uniaxial stress the strain-energy density, defined by Eq. 3–6, becomes

1
s1 u = sP (10–28)
s2 2

If the material is subjected to triaxial stress, Fig. 10–29a, then each

(a) principal stress contributes a portion of the total strain-energy density,
so that

1 1 1
u = s1P1 + s2P2 + s3P3
ⴝ

2 2 2

Furthermore, if the material behaves in a linear-elastic manner, then

savg Hooke’s law applies. Therefore, substituting Eq. 10–18 into the above
equation and simplifying, we get

C s 2 + s22 + s32 - 2n1s1s2 + s1s3 + s3s22 D

1
u = (10–29)
2E 1

This strain-energy density can be considered as the sum of two parts,

savg one part representing the energy needed to cause a volume change of
savg
the element with no change in shape, and the other part representing the
energy needed to distort the element. Specifically, the energy stored in the
element as a result of its volume being changed is caused by application
(b) of the average principal stress, savg = 1s1 + s2 + s32>3, since this stress
causes equal principal strains in the material, Fig. 10–29b. The remaining
portion of the stress, 1s1 - savg2, 1s2 - savg2, 1s3 - savg2, causes the
energy of distortion, Fig. 10–29c.
ⴙ Experimental evidence has shown that materials do not yield when
subjected to a uniform (hydrostatic) stress, such as savg discussed above.
As a result, in 1904, M. Huber proposed that yielding in a ductile
(s3  savg )
material occurs when the distortion energy per unit volume of the
material equals or exceeds the distortion energy per unit volume of the
same material when it is subjected to yielding in a simple tension test.
This theory is called the maximum-distortion-energy theory, and since it
10 was later redefined independently by R. von Mises and H. Hencky, it
sometimes also bears their names.
To obtain the distortion energy per unit volume, we will substitute the
(s1  savg ) stresses 1s1 - savg2, 1s2 - savg2, and 1s3 - savg2 for s1 , s2 , and s3 ,
(s2  savg ) respectively, into Eq. 10–29, realizing that savg = 1s1 + s2 + s32>3.
Expanding and simplifying, we obtain

C 1s1 - s222 + 1s2 - s322 + 1s3 - s122 D

(c)
1 + n
Fig. 10–29 ud =
6E
 10.7 THEORIES OF FAILURE 523

In the case of plane stress, s3 = 0, and this equation reduces to s2

A s12 - s1s2 + s22 B

1 + n
ud = sY
3E
For a uniaxial tension test, s1 = sY , s2 = s3 = 0, and so
sY
1 + n 2 s1
1ud2Y = s sY
3E Y
Since the maximum-distortion-energy theory requires ud = 1ud2Y , then
sY
for the case of plane or biaxial stress, we have

s12 - s1s2 + s22 = sY2 (10–30) Maximum-distortion-energy theory

This is the equation of an ellipse, Fig. 10–30. Thus, if a point in the Fig. 10–30
material is stressed such that (s1, s2) is plotted on the boundary or
outside the shaded area, the material is said to fail.
A comparison of the above two failure criteria is shown in Fig. 10–31.Note
that both theories give the same results when the principal stresses are
equal, i.e., s1 = s2 = sY, or when one of the principal stresses is zero and
the other has a magnitude of sY. If the material is subjected to pure shear, t,
then the theories have the largest discrepancy in predicting failure. The s2
stress coordinates of these points on the curves can be determined by Pure shear
considering the element shown in Fig. 10–32a. From the associated Mohr’s sY (sY, sY)
circle for this state of stress, Fig. 10–32b, we obtain principal stresses s1 = t
and s2 = - t. Thus, with s1 = - s2, then from Eq. 10–27, the maximum-
shear-stress theory gives 1sY >2, -sY >22, and from Eq. 10–30, the
sY
s1
sY
maximum-distortion-energy theory gives 1sY > 23, -sY > 232, Fig.10–31.
sY , sY
Actual torsion tests, used to develop a condition of pure shear in a 3

3
ductile specimen, have shown that the maximum-distortion-energy sY
(sY,sY) sY , sY
theory gives more accurate results for pure-shear failure than the 2  2
maximum-shear-stress theory. In fact, since 1sY> 132>1sY>22 = 1.15,
the shear stress for yielding of the material, as given by the maximum- Fig. 10–31
distortion-energy theory, is 15% more accurate than that given by the
maximum-shear-stress theory.

s2  t s1  t
s
t 10
90

A (t, 0)

t
(a) (b)

Fig. 10–32
 524 C H A P T E R 10 S T R A I N T R A N S F O R M AT I O N

Brittle Materials
Maximum-Normal-Stress Theory. It was previously stated that
brittle materials, such as gray cast iron, tend to fail suddenly by fracture
with no apparent yielding. In a tension test, the fracture occurs when
the normal stress reaches the ultimate stress sult , Fig. 10–33a. Also,
brittle fracture occurs in a torsion test due to tension since the plane of
fracture for an element is at 45° to the shear direction, Fig. 10–33b. The
fracture surface is therefore helical as shown.* Experiments have further
shown that during torsion the material’s strength is somewhat unaffected
by the presence of the associated principal compressive stress being at
right angles to the principal tensile stress. Consequently, the tensile stress
needed to fracture a specimen during a torsion test is approximately the
Failure of a brittle material
same as that needed to fracture a specimen in simple tension. Because of
in tension this, the maximum-normal-stress theory states that a brittle material will
fail when the maximum tensile stress, s1, in the material reaches a value
(a)
that is equal to the ultimate normal stress the material can sustain when
it is subjected to simple tension.
If the material is subjected to plane stress, we require that

45 ƒ s1 ƒ = sult
(10–31)
ƒ s2 ƒ = sult

45 These equations are shown graphically in Fig. 10–34. Therefore, if the
stress coordinates 1s1 , s22 at a point in the material fall on the boundary
or outside the shaded area, the material is said to fracture. This theory is
generally credited to W. Rankine, who proposed it in the mid-1800s.
Failure of a brittle material Experimentally it has been found to be in close agreement with the
in torsion behavior of brittle materials that have stress–strain diagrams that are
(b) similar in both tension and compression.

Fig. 10–33 Mohr’s Failure Criterion. In some brittle materials tension and
compression properties are different. When this occurs a criterion based
on the use of Mohr’s circle may be used to predict failure. This method
s2 was developed by Otto Mohr and is sometimes referred to as Mohr’s
failure criterion. To apply it, one first performs three tests on the material.
A uniaxial tensile test and uniaxial compressive test are used to
sult determine the ultimate tensile and compressive stresses 1sult2t and
10 1sult2c , respectively. Also a torsion test is performed to determine the
material’s ultimate shear stress tult. Mohr’s circle for each of these stress
sult sult s1

sult

Maximum-normal-stress theory

Fig. 10–34 *A stick of blackboard chalk fails in this way when its ends are twisted with the fingers.
 10.7 THEORIES OF FAILURE 525

conditions is then plotted as shown in Fig. 10–35. These three circles are
contained in a “failure envelope” indicated by the extrapolated colored
curve that is drawn tangent to all three circles. If a plane-stress condition Failure envelope
at a point is represented by a circle that has a point of tangency with the
envelope, or if it extends beyond the envelope’s boundary, then failure is s
said to occur. (sult)c (sult)t
We may also represent this criterion on a graph of principal stresses s1 tult
and s2. This is shown in Fig. 10–36. Here failure occurs when the absolute
value of either one of the principal stresses reaches a value equal to or
greater than 1sult2t or 1sult2c or in general, if the state of stress at a point t
defined by the stress coordinates 1s1 , s22 is plotted on the boundary or
Fig. 10–35
outside the shaded area.
Either the maximum-normal-stress theory or Mohr’s failure criterion
can be used in practice to predict the failure of a brittle material. s2
However, it should be realized that their usefulness is quite limited. A
tensile fracture occurs very suddenly, and its initiation generally depends
on stress concentrations developed at microscopic imperfections of (sult)t
the material such as inclusions or voids, surface indentations, and
small cracks. Since each of these irregularities varies from specimen
to specimen, it becomes difficult to specify fracture on the basis of a s1
(sult)c (sult)t
single test.

(sult)c

Mohr’s failure criterion

Fig. 10–36

Important Points

• If a material is ductile, failure is specified by the initiation of yielding, whereas if it is brittle, it is specified
by fracture.
• Ductile failure can be defined when slipping occurs between the crystals that compose the material. This
slipping is due to shear stress and the maximum-shear-stress theory is based on this idea.
• Strain energy is stored in a material when it is subjected to normal stress. The maximum-distortion-energy
theory depends on the strain energy that distorts the material, and not the part that increases its volume.
• The fracture of a brittle material is caused only by the maximum tensile stress in the material, and not the 10
compressive stress. This is the basis of the maximum-normal-stress theory, and it is applicable if the
stress–strain diagram is similar in tension and compression.
• If a brittle material has a stress–strain diagram that is different in tension and compression, then Mohr’s
failure criterion may be used to predict failure.
• Due to material imperfections, tensile fracture of a brittle material is difficult to predict, and so theories of
failure for brittle materials should be used with caution.
 526 C H A P T E R 10 S T R A I N T R A N S F O R M AT I O N

EXAMPLE 10.12
The solid cast-iron shaft shown in Fig. 10–37a is subjected to a torque
of T = 400 lb # ft. Determine its smallest radius so that it does not fail
according to the maximum-normal-stress theory. A specimen of cast
iron, tested in tension, has an ultimate stress of 1sult2t = 20 ksi.

tmax

s
s2 s1
T  400 lbft

T  400 lbft
tmax
r t

(a) (b)

Fig. 10–37

SOLUTION
The maximum or critical stress occurs at a point located on the surface
of the shaft. Assuming the shaft to have a radius r, the shear stress is

Tc 1400 lb # ft2112 in.>ft2r 3055.8 lb # in.

tmax = = =
J 1p>22r 4
r3

Mohr’s circle for this state of stress (pure shear) is shown in Fig. 10–37b.
Since R = tmax , then
3055.8 lb # in.
s1 = - s2 = tmax =
r3
The maximum-normal-stress theory, Eq. 10–31, requires
ƒ s1 ƒ … sult
3055.8 lb # in.
10
… 20 000 lb>in2
r3
Thus, the smallest radius of the shaft is determined from
3055.8 lb # in.
= 20 000 lb>in2
r3
r = 0.535 in. Ans.
 10.7 THEORIES OF FAILURE 527

EXAMPLE 10.13
The solid shaft shown in Fig. 10–38a has a radius of 0.5 in. and is made
of steel having a yield stress of sY = 36 ksi. Determine if the loadings
cause the shaft to fail according to the maximum-shear-stress theory
and the maximum-distortion-energy theory.

SOLUTION
The state of stress in the shaft is caused by both the axial force and the
torque. Since maximum shear stress caused by the torque occurs in
the material at the outer surface, we have A
P -15 kip
sx = = = - 19.10 ksi 15 kip
A p10.5 in.22
3.25 kip # in. 10.5 in.2 0.5 in. 3.25 kipin.
Tc
txy = = = 16.55 ksi
2 10.5 in.2
p 4 (a)
J
The stress components are shown acting on an element of material
at point A in Fig. 10–38b. Rather than using Mohr’s circle, the principal 16.55 ksi
stresses can also be obtained using the stress-transformation Eq. 9–5.
sx + sy sx - sy 2 19.10 ksi
s1,2 = ; a b + txy2
2 B 2
-19.10 + 0 -19.10 - 0 2
= ; a b + 116.5522 (b)
2 B 2
= - 9.55 ; 19.11 Fig. 10–38
s1 = 9.56 ksi
s2 = - 28.66 ksi

Maximum-Shear-Stress Theory. Since the principal stresses have

opposite signs, then from Sec. 9.5, the absolute maximum shear stress
will occur in the plane, and therefore, applying the second of
Eqs. 10–27, we have
ƒ s1 - s2 ƒ … sY

ƒ 9.56 - 1 -28.662 ƒ … 36
?

38.2 7 36
Thus, shear failure of the material will occur according to this theory. 10
Maximum-Distortion-Energy Theory. Applying Eq. 10–30, we have
A s12 - s1s2 + s22 B … sY2

C 19.5622 - 19.5621 -28.662 + 1 -28.6622 D … 13622

?

1187 … 1296
Using this theory, failure will not occur.