Solutions Manual

to problems in

Failure, Fracture, Fatigue

An Introduction

by
Tore Dahlberg

Anders Ekberg

Studentlitteratur, Lund 2002, ISBN 91-44-02096-1.

This manual contains solutions to problems in the textbook

Failure, Fracture, Fatigue - An Introduction

Studentlitteratur, Lund 2002, ISBN 91-44-02096-1

At present, solutions to all problems given in Chapters 1 to 6 and

Chapters 8 and 9 are available in this document (Chapter 7 does not
contain any problems and solutions to the problems in Chapter 10
will be available later).
It is our hope that these solutions will be to some help for teachers
and students when studying the topics covered in the textbook.
Samples of examinations are available (2010) on the home page of
Solid Mechanics, IEI, at Linkping University, Sweden, address:
www.solid.iei.liu.se
under the heading "Examinations" and course "Damage mechanics
and life analysis TMHL61".
Linkping and Gothenburg in February 2010
The authors

Address to Tore Dahlberg after May 2010:

smark 6, 563 92 Grnna, Sweden
Page 1:2

Solutions to problems in
T Dahlberg and A Ekberg: Failure, Fracture, Fatigue - An Introduction.
Studentlitteratur, Lund 2002, ISBN 91-44-02096-1.

Chapter 1
Introduction, failure mechanisms
Problems with solutions
Elastic deformations
1/1.
Side view :

Top view :

Cross section:
B
t
H

H >> t
B >> t

A cantilever beam, length L, carries a force P at

its free end. The beam cross section is an ideal I
profile, i.e., the area of the flanges is width B by
height t each (B >> t), and the area of the web
can be neglected. The height is H, where H >> t,
see figure.
When the beam is loaded by the force P, the
deflection 1 becomes
P L3
H 2 Bt H 2
1 =
where I = 2 Bt =
3 EI
2
2

Thus, 1 = 2P L 3 / 3E BtH 2 = 0.67P L 3 / E BtH 2.

Side view :

It has been found that this deflection is too

large. Therefore, without increasing the weight
(the mass) of the beam, another form of the
cantilever beam cross section is tried. Material
P
is moved from the flanges at the free end and
placed at the flanges at the fixed end to give a
beam cross section (still an ideal I profile) with
a width varying with the coordinate x, see
figure. The height of the profile is still H, but
H >> t the width of the flanges is now B(x) =
B >> t 2B(1 x /L).
What deflection 2 will now be obtained when
the load is applied at the free end?
P

Top view :
x

Cross section :
2 B (1 - x / L)
t
H

SOLCHAP1.DOC/2010-01-29/TD

Chapter 1

Page 1:1

Solution:
Use the differential equation
EI(x) w(x) = M(x)
(a)
to determine the deflection w(x) of the beam. The bending moment M(x) in the
beam becomes M(x) = PL (1 x /L), and the second moment of area becomes

x H2
I = 2 2B 1 t = Bt H 2 1

L 2
L

(b)

The direction of the moment M(x) has been selected to give a positive
deflection w(x) downwards.
Entering M(x) and I(x) in the differential equation gives
PL
M(x)
PL (1 x/L)
=
w(x) =
=
EI(x) E Bt H 2(1 x/L) E Bt H 2

(c)

Integration gives
w(x) =

and

PL
x + C1
E Bt H 2

PL x 2
w(x) =
+ C1 x + C2
E Bt H 2 2

Boundary conditions at x = 0 give (notation BC for boundary condition)

BC 1:
w(0) = 0 gives C2 = 0
BC 2:

w(0) = 0

gives C1 = 0

(d)
(e)

(f)
(g)

The deflection w(x) then becomes

PL x 2
w(x) =
E Bt H 2 2

(h)

and at the free end (at the force P) one obtains

PL L 2
PL 3
2 = w(L) =
=
E Bt H 2 2 2E Bt H 2

(i)

It is concluded that the rearrangement of the width of the flanges makes the
deflection at the free end to decrease from 0.67 PL3/EBtH2 to 0.5 PL3/EBtH2,
i.e., the deflection decreases by 25 per cent.
Answer: The rearrangement of the width of the flanges causes the deflection at
the free end to decrease from 0.67 PL3/EBtH2 to 0.5 PL3/EBtH2.
Page 1:2

Chapter 1

Yielding
1/2.
The stress components in a part of a structure have been calculated to x = 120
MPa, y = 80 MPa, and xy = 60 MPa (all other stress components are zero).
For this structure a safety factor s = 1.5 with respect to yielding is required.
What yield strength Y should the material have to fulfil this condition?
Investigate both the Tresca and the von Mises yield criteria.
Solution:
First, determine the principal stresses in the material. In the xy-plane one
obtains
x + y
1, 2 =

x y 2 2
+ xy = 100 63.2 (MPa)

2
and do not forget the third principal stress: 3 = 0.

(a)

Thus, the largest principal stress is 1 = 163.2 MPa and the smallest principal
stress is 3 = 0. The effective stress according to Tresca then is eT = 1 3 =
163.2 MPa.
Using safety factor s = 1.5 on the yield limit gives
Y
163.2 =
(a)
which gives Y = 245 MPa
1.5
The von Mises equivalent stress is
vM
2x + 2y x y + 32xy = 148.3 MPa

e =

Again, using safety factor s = 1.5 on the yield limit gives

Y
148.3 =
which gives Y = 222 MPa
1.5

(b)

(c)

Thus, according to the more conservative criterion (the Tresca criterion) the
yield strength of the material should be Y = 245 MPa.
Answer: The yield strength of the material should be, according to the Tresca
criterion, Y = 245 MPa (and the von Mises criterion gives Y = 222 MPa).

Chapter 1

Page 1:3

Yielding and plastic collapse

1/3.
A cantilever beam of length L and with a
rectangular cross section (base b, height h) is
loaded by the force P at the free end.
(a) Determine the load P = Pelast the beam may
be loaded with without inducing any plastic
deformation in the beam (i.e., Pelast is the
maximum load not giving any yielding in the
beam).

Material:

stress

strain

(b) Determine the maximum force Pmax the beam may be loaded with. Assume
that the material is linearly elastic, ideally plastic with yield limit Y.
(c) Finally, determine the ratio of the load Pmax to load Pelast.
Solution:
(a) To start with, the elastic solution giving Pelast will be determined. The stress
in the beam is obtained as
Mb
=
(a)
Wb
where Mb = Pelast L, and Wb is the section modulus: Wb = bh2/6.
When = Y one obtains

Mb Y b h 2
=
Pelast =
L
6L

(b)

x
z

(b) The maximum force that may be applied to

the beam will produce a plastic hinge at the
support of the beam. The bending moment
Mf
needed to produce the hinge is

Y
0

Mf = z dA = Y z b dz +

h /2
0

h /2

Y b h 2
Y z b dz =
4

(c)

Thus Pmax = Mf /L = Ybh2/4L.

(c) The ratio Pmax /Pelast becomes Pmax /Pelast = 3/2.
Answer: (a) Pelast = Y bh 2 / 6L, (b) Pmax = Y bh 2 / 4L, and (c) Pmax /Pelast = 3/2.

Page 1:4

Chapter 1

Plastic collapse
1/4.
P

2 L /3

L /3
x

The beam AC is rigidly supported (clamped) at

end A and simply supported at end C. The beam
has length L and a rectangular cross section with
base b and height h, and it is loaded at x = L /3
by a force P.

(a) Determine the force P = P0 giving one plastic hinge on the beam (to give a
plastic hinge, a cross section must be fully plastic).
(b) Determine the force P = Pf giving collapse of the structure. (For collapse,
two plastic hinges are needed. Here they will appear at A and B.)
The material is linear elastic, ideally plastic with yield limit Y.
Solution:
A

L /3

2 L /3

MA

- 5 PL /27

8 PL /81

First determine the distribution of the bending

moment M along the beam when the beam is
C
fully elastic. Handbook formulae give, for an
elastic beam, the moment MA at support A as
RC
MA = 5PL /27. The reaction force at support C
becomes RC = P /3 5P /27 = 4P /27. The
x bending moment MB at B becomes MB =
RC 2L /3 = 8PL /81. Note that MA and MB have
different directions, see figure.

According to the conventions used here the moment MA is negative, but in

most cases only the numerical value of the moment (and not the direction) is
of interest. Therefore, only numerical values of the moments will be discussed
below. Note also that, numerically, MA > MB, implying that plastic yielding
will start at point A.
Next, determine the bending moment M = Mf needed to create a plastic hinge
on the beam. When a plastic hinge appears, half the cross-sectional area A
(A = b h) is yielding in tension and half of the area A is yielding in
compression. This gives (see Problem 1/3)
Mf = x z dA = 2
A
0

h /2

bh 2
1 h2
Y z bdz = 2 Y b = Y
2 2
4

Chapter 1

(a)

Page 1:5

(a) Question (a) can now be answered. The first plastic hinge will appear in the
beam when the largest moment (here MA) reaches the value Mf. (It is assumed
that MA > MB is valid also during yielding at A.) Thus, MA = Mf gives
(numerically)
bh 2 5PL
Mf = Y
=
4
27

27 bh 2
which gives P = P0 =

20 L Y

(b,c)

Thus, at load P = P0 = 1.35 Y bh2/L one plastic hinge has developed (at A) in
the beam.
P = P0

2 L /3
MB
RC
P > P0

A L /3

2 L /3

MA= Mf plastic hinge

MA = Mf
A

P = Pf

plastic hinges
Collapse mode
P = Pf

C
RC

During the loading of the beam, after yielding

has started at point A, the load distribution in
the structure may change. This implies that the
ratio MA /MB of the bending moments may
change during yielding at A. Therefore, confirm
that MA is larger than MB also when the hinge
appears at A, i.e., confirm that MA = Mf > MB
when P = P0.
When MA = Mf and P = P0, the support reaction
RC becomes RC = P0 /3 Mf /L, and the moment
MB at B becomes MB = 2RC L /3 = 2Y bh2/15.
This moment is less that Mf, which implies that
the beam is still elastic at B when the hinge
appears at A, i.e., at least part of the cross
section at B is elastic, because the moment at
MB is not large enough to create a plastic hinge
there.

(b) The force P can be further increased, beyond P0, because the part AC of
the beam is still elastic. When MA = Mf the beam AC acts as a simply
supported beam, supported at A and C, and loaded with a moment Mf at A and
a force P (> P0) at B. Collapse of the beam will appear when the bending
moment at B reaches Mf, i.e., when a plastic hinge is obtained also at B.
When the load P > P0 has increased so much that it produces a hinge also at B,
thus giving MB = Mf, the structure will collapse. The force P = Pf that gives a
hinge also at B can be determined in the following way: the bending moment

Page 1:6

Chapter 1

at B is MB = Mf = RCf (2L /3), which gives RC = RCf = MB /(2L /3) = 3Mf /2L, see
figure below. The loading of the full beam AC then is MA = Mf at A, P = Pf at
B, and RC = RCf = 3Mf /2L at C, see figure. The force P = Pf is still unknown.
Mf

Pf

2 L /3
R C f = 3 Mf / 2 L

Pf

A L /3

2 L /3

Moment equilibrium with respect to point A

gives an equation containing the unknown force
P = Pf. One obtains
L
L 3Mf
MA = Mf = Pf RCf L = Pf
(d)
3
3
2
from which is solved

3 15 bh 2
5
Pf = Mf = Y
(e)
R C f = 3 Mf / 2 L
2
L 8
L
Thus, after the first hinge has appeared at load P = P0 = 1.35 Y bh2/L, giving
one plastic hinge at A, the force P may be increased to P = Pf = 1.875 Y bh2/L,
at which load two plastic hinges are obtained, and the structure will collapse.
Answer: P0 = 1.35 Y bh2/L gives one plastic hinge at A, and Pf =
1.875 Y bh2/L gives plastic collapse (two hinges).
Mf

Chapter 1

Page 1:7

Plastic instability
1/5.
A0 , L0

A, L

Determine the maximum tensional force Pmax the

bar in the figure (see also Section 1.4.4) may be
subjected to. Assume that the material is
hardening due to plastic deformation and use the
stress-strain relationship
= 0 m

Material:
stress
strain

Use true stress and natural (logarithmic) strain.

The true stress is defined as
P
=
A

where A is the (contracted) cross-sectional area obtained after loading, i.e. A <
A0 when P > 0, and A depends on P.
The logarithmic strain is defined as

= ln

L
L0

where L is the length of the bar when loaded, i.e. L > L0 when P > 0, and L
depends on P.
Also, remember that plastic deformation will take place with no change of
volume.
Solution:
As in the example on plastic instability, see Section 1.4.4 in the textbook, one
has
equilibrium:
P = A
(a)
V = A0 L 0 = A L

the volume is constant:

large strain:

= ln

L
L0

Differentiate (a), (b) and (c). It gives

dP = dA + A d

(b)
(c)
(d)

dV = A dL + L dA = 0

(e)

1 dL dL

=
L/L0 L0 L

(f)

d =

Plastic instability occurs when the load cannot be increased any more, i.e.
Page 1:8

Chapter 1

when dP = 0. It gives
dA + A d = 0

(g)

A dL + L dA = 0

(h)

dL = L d

(i)

Eliminate A and L from (g), (h) and (i). It gives

A
L
d
d
=
d =
d =
; thus =
dA
dL
d
d

(j)

Introduce the material relationship = 0 m into (j). It gives

0 m = 0 m m 1 giving = m

(k)

Thus, the loading force P reaches its maximum value P = Pmax when = m.
Then
L0
Pmax = A = 0 m m A = 0 m m A0
(l)
L
The ratio L0 /L is obtained from equation (c). By use of = m, one obtains
L0
L
L
giving em = , or,
(m)
= e m
= m = ln
L0
L0
L
Finally, entering (m) into (l) gives
Pmax = 0 m m A0 e m

Alternative solution:
The loading force P may be written, by use of (a) and (c),
L0
P = A = 0 m A0 = 0 m A0 e
L
Determine the maximum value of P. One has
dP
= 0 A0 {m ( 1) e + mm 1 e } = 0
d

(n)

(o)

(p)

from which = m is obtained. Entering = m into (o) gives

P = Pmax = 0 m m A0 e m

(q)

(Verify that a maximum value, and not a minimum value, has been
determined.)
Answer: Plastic instability will occur when Pmax = 0 m m e m A 0.

Chapter 1

Page 1:9

Stationary creep
1/6.
h
h

p
o

A tube subjected to an internal pressure p is

made of two circular cylinders of different
materials. The outer cylinder has inner radius r,
wall thickness h (h << r), and is made of a
Hooke material with modulus of elasticity Eo =
207 GPa (i.e. linear elastic material with
constitutive equation = /Eo).
The inner cylinder has outer radius r, wall
thickness h, and is made of a Norton material.
Thus, before the loading has been applied the
inner cylinder fits exactly in the outer one.

The material equation of the inner cylinder reads

| | n sgn
= +

Ei n t*
where Ei is the elastic modulus of the material of the inner cylinder (Ei = 69
GPa), n and n are material parameters (n > 1), sgn means sign of and is
+1 if is positive and 1 if is negative, and t* is a reference time.
(a) Calculate the stresses in the two cylinders immediately after the pressure p
has been applied to the inner cylinder (no longitudinal stresses will appear,
only circumferential).
(b) Investigate how the stresses will change with time in the two cylinders.
Solution:

q
o

q
p
i

Page 1:10

(a) When, at time t = 0, the pressure p is applied

to the inner cylinder, a contact pressure q will
arise between the inner and the outer cylinder.
The circumferential stress i in the inner
cylinder becomes i = (p q) r /h and in the
outer cylinder the circumferential stress o
becomes o = qr /h (index i for the inner
cylinder and index o for the outer). The contact
pressure q is unknown, see figure.

Chapter 1

The contact pressure q is obtained from the deformation condition that the two
cylinders have the same radial expansion u. Thus, ui = uo. For circular
symmetry the circumferential strain equals = u /r (see equation (1.18b) in
Section 1.2.6 in the textbook) where u is the radial displacement (radial
expansion) of the thin-walled cylinder. One obtains
ui = r i = r o = uo
(a)
This gives, at time t = 0,
i o
=
Ei Eo

(b)

Entering i = (p q) r /h and o = qr /h in (b) gives

Ei
p q =q
Eo

(c)

Using the numerical values of Ei and Eo one obtains q = 3p /4 (at t = 0).

Thus, at time t = 0, immediately after the pressure p has been applied, the
circumferential stress in the inner cylinder is i = (p 3p /4) r /h = pr /4h, and
in the outer cylinder the circumferential stress is o = qr /h = 3pr /4h.
It is seen that when the pressure is applied (at t = 0), the outer cylinder carries
most of the load (q > p q) and o > i). This follows from the fact that the
outer cylinder is stiffer than the inner one (Eo > Ei). Because of the lower
modulus of elasticity of the inner cylinder, it is easier for the inner cylinder
to expand, implying that the outer cylinder (which is less prone to expansion)
has to carry most of the load. Due to creep, this stress distribution will be
changed as time goes on.
(b) How will the circumferential stresses i and o develop with time?
The deformation condition ui = uo is valid also at times t > 0. It gives
i
o
+ kni =
u i = u o giving
Ei
Eo

(d,e)

where k is a constant replacing 1 / nn in the material equation (thus, k =

1 / nn), and i > 0 gives sgn i = 1. The dot ( ) indicates differentiation with
respect to time t.
The contact pressure q between the inner and outer cylinder will now depend
on time t, thus q = q(t). The initial condition on q, at time t = 0, was
determined above, namely q(0) = 3p /4.
Chapter 1

Page 1:11

Entering i = (p q) r /h and o = qr /h into (e) gives

r n
r
r
+ k (p q) = q
( p q)

h Ei
h
h Eo

(f)

Here p = 0, because p is the constant pressure in the cylinder. Rearranging the

expression (f) gives
r n
p r n
r1 1

+
+k q
=k
(q)

h Ei Eo
h
h

or
where

q + C1 q n = C2
C1 =

k { r/h } n
r { 1/Ei + 1/Eo }/h

and C2 =

(g)
(h)

k{ p r/h } n
r { 1/Ei + 1/Eo }/h

One notices that C2 /C1 = p n ; a relation that will be used below.

The homogeneous part of equation (h) reads
q + C1 q n = 0

(i)

This equation may be separated into one part containing q only and one part
containing the time t. The solution to (i) becomes, for n > 1,
q = qhom = {( n + 1) ( C1 t + C0 )} 1 / ( n + 1)

(j)

where C0 is an integration constant to be determined from a boundary (initial)

condition.
The particular solution to (h) becomes
C2 1 / n
q = qpart = = p
C1

(k)

q = qhom + qpart = { ( n + 1) ( C1 t + C0 )} 1 / ( n + 1) + p

(l)

and the complete solution is

The boundary (initial) condition q(0) = 3p /4 gives the constant C0. One obtains
3p
q(0) = {( n + 1) C0 } 1 / ( n + 1) + p =
(m)
4
giving

Page 1:12

1 3p
1 p ( n + 1)
( n + 1)
=

C0 =
n +1 4

n +1 4

Chapter 1

(n)

Thus,
q

=(n

= (n

1 / ( n + 1)

1 p ( n + 1)

+ 1) C1 t +
+p

n +1 4

p ( n + 1)

1) C1 t +

1 / ( n + 1)

+p

(o)

Finally, one obtains

r
r
i(t) = (p q) =
h
h

(n

p ( n + 1)

1) C1 t +

1 / ( n + 1)

(p)

and
1 / ( n + 1)
qr r
r
p ( n + 1)

o(t) = = (n 1) C1 t +
+ p

h h
h
4

(q)

One notices that for t = 0 the initial stresses i = pr /4h and o = 3pr /4h are
obtained (as one should). For large values of time t, the term containing the
time t will dominate over the second term in the factor enclosed by the curly
brackets in (p) and (q). The factor in the curly brackets will then be
proportional to t 1 / (n 1), which tends to zero for large values of time t. This
implies that after a long time (for large values of time t) the expression giving
i in (p), and the corresponding term in (q), will tend to zero, and the stresses
tend to i = 0 and o = pr /h, respectively. Thus, the creep in the inner cylinder
force the outer cylinder to carry a larger part of the load. After some time the
inner cylinder does not carry any load at all (or it carries a very small part of
the load). The outer cylinder will then carry the main part of the load. The
function of the inner cylinder could then be, for example, to protect the outer
cylinder from chemical reactions, or something else.
Answer: (a) At time t = 0, the stress in the inner cylinder is i = pr /4h, and in
the outer cylinder the stress is o = 3pr /4h.
1 / ( n + 1)
(b)
r
r
p ( n + 1)

i(t) = (p q) = (n 1) C1 t +

h
h
4

and

qr r
p ( n + 1)

o(t) = = (n 1) C1 t +

h h
4

1 / ( n + 1)

r
p
h

(where i tends to zero and o tends to pr /h when time t becomes large).

Chapter 1

Page 1:13

Fracture, maximum normal stress

1/7.
The stress components in a part of a structure have been calculated to xx =
120 MPa, yy = 80 MPa, and xy = 60 MPa (all other stress components are
zero). Using the maximum normal stress criterion, investigate material failure.
The material is brittle and it has the ultimate strength Ut = 150 MPa in tension
and the ultimate strength Uc = 200 MPa in compression.
Solution:
First, determine the principal stresses in the material (and dont forget the third
principal stress). One obtains, in the xy-plane,
xx + yy
1, 2 =

xx yy 2 2
+ xy = 100 63.2 (MPa)

(a)

Thus, the largest principal stress is 1 = 163.2 MPa and the smallest principal
stress is 3 = 0 (= zz). Failure is expected, since in tension the ultimate
strength of the material is Ut = 150 MPa only.
Answer: Yes, failure is expected because the maximum principal stress in the
material is 1 = 163 MPa, whereas in tension the strength of the material is
Ut = 150 MPa, only.

Page 1:14

Chapter 1

Fracture, Mohrs failure criterion

1/8.
shear
stress

The Mohr criterion for brittle fracture is, for the

material studied here, formulated as (see figure)
+ 0.4 = 1.2Y
normal where Y is the yield limit of the material. A
stress
test specimen of this material is subjected to a
torque, which is increased until fracture occurs.

Will there be any plastic deformation of the specimen prior to the fracture?
Use the Tresca yield criterion.
Solution:

1.2 Y

C B
A

Rfra
3 O

3 Y

Pure shear gives Mohrs circle as shown in the

figure (centre of the circle at the origine O).
Determine the shear stress giving fracture, i.e.
determine the radius of the circle. Triangles
OAB and OAC give
Rfra
1.2Y
=
3Y
1.22 +32 Y

which gives Rfra = 1.1142 Y

Thus, at fracture one has fra = Rfra = 1.1142 Y.

At which shear stress = yield will yielding occur?
Yielding will occur when eTresca = Y = 1 3 = yield ( yield) = 2yield.
Thus, at yielding one has (according to Tresca) yield = Y /2.
It is concluded that yielding will occur before fracture (yielding when =
0.5Y and fracture when = 1.1142Y). This means that the material must be
deformation hardening so that shear stress = 1.1142Y can be reached. When
that stress is reached the fracture will occur.
Answer: Plastic deformation will occur because = fra = 1.114Y at fracture
whereas = yield = 0.5Y at yielding.

Chapter 1

Page 1:15

Fracture, Mohrs failure criterion

1/9.
For a brittle material the ultimate strength was determined to be 100 MPa in
tension and 400 MPa in compression. Within these limits the material is
linearly elastic. Assume that the Mohr criterion for brittle fracture is given by
straight lines in the shear stress versus normal stress diagram (as in Problem
1/8). The volume of the test specimens used to determine the ultimate strengths
was 104 mm3.
This material is used in a circular thin-walled cylinder of length L = 1 m,
diameter d = 200 mm, and wall thickness h = 5 mm. The cylinder is loaded
with a torque T. With respect to brittle fracture, determine the maximum
allowable torque Tmax the cylinder may be subjected to. The size effect
according to Weibull should be taken into account.
According to Weibull the ultimate strength of a large volume of material is less
than the ultimate strength of a small volume of the same material. (One reason
for this could be that there is a larger probability of finding a flaw in a larger
volume than in a smaller volume. The brittle fracture may start at the flaw.)
Assume that the ultimate strength of a material of volume V0 is found to be 0.
According to Weibull, the ultimate strength of the same material, but
of volume V, would be
V0 1 / m
= 0
V
For the material above, use factor m = 6 in the Weibull relationship.
Solution:
The ultimate strength in tension, Ut = 100 MPa, and in compression, Uc =
400 MPa, give the fracture limit curve according to the figure below.
The shear stress in the circular thin-walled cylinder is
T
T
=
=
Wv 2 r 2h

(a)

Denote this stress xy. Thus, xy = = T/2r2h. The principal stresses (at plane
stress, zz = xz = yz = 0) are given by
xx + yy
1, 2 =

xx yy 2 2
+ xy

where, here, xx = 0, yy = 0, and xy = .

Page 1:16

Chapter 1

(b)

O
Uc

The principal stresses thus are (re-numbered)

h
1 = , 2 = 0 and 3 =
(c)
T
Mohrs circle, as given by the principal stresses
r
1 and 3, is entered into the diagram. The new
circle will have the radius 1 (= OD in the
OD =
figure). Fracture is expected when the circle
E
R= 1
reaches the fracture limit curve. The radius of
the circle is obtained from triangles OBD and
D
C
B OBC.
O O
Determine first, however, the distance OB.
Ut From triangles OBC and OBE one obtains
OB 50
50
=
(d)
200 OB + 200
giving OB = 133 MPa.

Triangles OBC and OBD give (with OD = 1)

50 OB 50
=
1
OB

(e)

from which 1 = 80 MPa is solved. Thus = 1 = 80 MPa before any

reduction of the stress with respect to volume effects has been done.
(With no reduction due to volume effects, the allowable torque T would have
been T = Wv = 25 kNm.)
Reduce the allowable stress with respect to the volume effect. Weibull gives
1/m
V0
=
0 V

Thus,

(f)

reduced V0 1 / 6
=

1/6

104
= 0.3835 = 30.7 MPa
giving
reduced =
(g)
2100 5 1000
The allowable torque T becomes Tallow = reducedWv = 9.6 kNm.
Answer: Tmax = 9.6 kNm (and Tmax would have been 25 kNm, approximately,
if the size effect had been neglected).

Chapter 1

Page 1:17

Fracture, Mohrs failure criterion

1/10.
A circular bar made of a brittle
loaded in axial compression.
materials are available. Both
linear fracture limit curves, see
following data

Material

tan

1/2

2s

1/3

material will be
Two different
materials have
figure, with the

Which one of the materials is the better if brittle fracture (in compression) of
the bar is considered?
Solution:
A
rB rA

A
B

B
s

2s

The fracture limit of material A intersects the

axis at A = s tan = s / 2. Let the radius of
circle A be rA. Similar triangles give
rA
s /2
(a)
=
2
2
rA + s s
1 + (1 / 2)

giving rA = 0.809s. This gives UcompA = 2rA =

1.62s.

The fracture limit of material B intersects the axis at A = 2s tan = 2s / 3.

Let the radius of circle B be rB. Similar triangles give
rB
2s /3
(b)
=
rB + 2s s
22 + (2 / 3)2

giving rB = 0.925s. This gives UcompB = 2rB = 1.85s.

Material B has the largest fracture strength in (uni-axial) compression: UcompB
= 2rB = 1.85s. This material should be selected.
Answer: Material B should be selected because it has the largest ultimate
strength in compression: UcompB = 1.85s (whereas UcompA = 1.62s).

Page 1:18

Chapter 1

Fracture, Mohrs failure criterion

1/11.
Two materials, C and D, have linear fracture limit curves. The two materials
have the same ultimate strength U when loaded in shear. When loaded in
tension, however, material C has a larger ultimate strength than material D,
thus UtC > UtD.
What can be said about the two materials ultimate strengths in compression?
Solution:

U
Ut

From the figure it can be seen that the larger the

ultimate strength in tension Ut is, the more
horizontal will the fracture limit curve be, and
the smaller will the circle to the left of the
-axis be. Thus, as UtC > UtD one will obtain
UcompC < UcompD.

It can also be shown that (see below)

1
1
1
=
+
U Ut Ucomp

(a)

which gives the above result.

Similar triangles give
U
Ut / 2
1
1
1
=
=
giving

s s Ut / 2
U Ut / 2 s
and

Ucomp / 2
U
=
s s + Ucomp / 2

giving

1
1
1
=
+
U Ucomp / 2 s

(b)
(c)

Summing (b) and (c) gives (a).

Answer: For the ultimate strengths in compression one has (numerically)
UcompC < UcompD.

Chapter 1

Page 1:19

Fatigue
1/12.
A component of a machine is subjected to a repeated loading sequence. The
loading sequence is repeated once a day. The loading is alternating, and one
sequence contains the following loading amplitudes (i.e. stress amplitudes
the mean value of the stress is zero)
a = 180, 160, 140 and 100 MPa
The number of loading cycles at each stress level is
n = 15, 20, 150, and 500, respectively
The Whler curve of the material is, in this stress range, given by the relation
a = 50logN + 400 MPa
Determine the damage accumulation D due to one loading sequence, and then,
determine how many loading sequences (how many days) the component
might be used before fatigue failure is expected.
Solution: For the different stress levels, the expected fatigue life N is obtained
from the Whler curve. One obtains
(400 a) / 50

N = 10

(a)

giving, respectively,
N = 25 119, 63 096, 158 489, and 1 000 000
The accumulated damage due to one loading sequence is, by use of the
Palmgren-Miner damage accumulation rule,
15
20
150
500
1
D=
+
+
+
= 0.00236 =
(b)
25 119 63 096 158 489 1 000 000
424
Finally, the number of sequences Ns to expected fatigue failure is
Ns = 1 / D 424

(c)

Thus, the component can be used in 424 days, approximately.

Answer: Damage D due to one sequence is
15
20
150
500
D=
+
+
+ 6 = 2.36 10 3
25 119 63 096 158 489 10
Expected number of sequences (days) Ns to fatigue failure is Ns = 1 / D 424.

Page 1:20

Chapter 1

Elastic instability
1/13.
P

L, EI

(a)
First eigenmode:
x
Pcrit1
(b)
Second eigenmode:
Pcrit2
x
(c)
Constrained beam :
Pcrit2
x
(d)

It is possible to increase the buckling load of a

beam by introducing extra supports along the
beam, see Section 1.4.9 (the example on
instability). It can be shown that by use of one
extra support the buckling load of a beam at
most can be increase to the second eigenvalue
of the original beam without the extra support,
i.e. to the second critical load of the
unconstrained beam. The extra support should
then be placed at the node of the second
eigenmode of the unconstrained beam (i.e. at the
point of zero deflection of the second
eigenmode).

If the extra support is placed at that node, then the beam is constrained to
buckle according to the second eigenmode, and thereby the critical load P =
Pcrit of the constrained beam (with the extra support) equals the second critical
load of the unconstrained beam.
Determine where to place one extra support on the beam that is pinned (simply
supported) at one end and fixed (rigidly supported) at the other end, as shown
in figure (a), to obtain the largest possible critical load of the constrained beam
shown in figure (d). In figure (b) the first eigenmode w1(x) of the pinned-fixed
beam is sketched, and in figure (c) the second eigenmode w2(x) of the unconstrained beam (a) is sketched. Thus, determine the coordinate x where the
mode w2(x) has zero deflection.
Solution:
The problem will be solved by use of the differential equation for an axially
loaded beam. The differential equation reads
EI w IV(x) + P w(x) = 0

(a)

where w(x) is the deflection of the beam, EI is the bending stiffness, and P is
the force loading the beam in compression. The solution to this equation is
w(x) = C1 + C2 px + C3 sin px + C4 cos px
(b)
P/EI
where p =

.
Chapter 1

Page 1:21

Boundary conditions give the constants C 1 to C 4. One has (BC for boundary
condition)
BC 1: w(0) = 0 gives C1 + C4 = 0
(c)
BC 2: M(0) = 0

gives EIw(0) = 0

which gives

C3 p 2 sin p 0 C4 p 2 cos p 0 = 0

(d)

These two equations give C 4 = 0 and C 1 = 0, which is used in the following

boundary conditions. At the beam end x = L one has
BC 3: w(L) = 0 giving C2 pL + C3 sin pL = 0
(e)
BC 4: w(L) = 0

giving C2 p + C3 p cos pL = 0

(f)

To obtain a solution with at least one of C 2 and C 3 not equal to zero, one must
have the determinant of the system of equations (e,f) equal to zero. It gives
pL sin pL

= 0 or tan pL = pL
(g)
1 cos pL
tan pL
pL
pL
2

3
2

5
2

The roots to equation (g) are determined

numerically. From the figure it is concluded that
the first root is close to 3 / 2, the second root
very close to 5 / 2, and so on. One obtains
p1 L = 4.4934
p2 L = 7.7252 5 /2
p3 L = 7 /2

and so on.

(h)

The first root p1 L gives p1 L =

P /EI L = 4.4934

which gives

4.49342EI 2.05 2EI

P = Pcrit1 =
=
L2
L2

(i)

The second root p2 L gives p L =

P /EI L = 7.7252

2
which gives

7.72522 EI
P = Pcrit2 =
L2

(j)

The eigenmodes, i.e. the forms of the deflection of the buckling beam, become
(equation (f) gives C 2 = C 3 cos pL)
for p1 L:

Page 1:22

w1(x) = C3 { p1x cos p1L + sin p1x }

Chapter 1

(k)

for p2 L:

w2(x) = C3 { p2x cos p2L + sin p2x }

(l)

where C 3 is undetemined (and different in the different modes).

The second eigenmode becomes, by use of p2 L (p2 L in radians),

x
w2(x) = C3 7.7252 cos 7.7252 + sin 7.7252
L

(m)

Now the value of x giving a zero crossing of w2(x) can be determined. One
obtains

x
(n)
w2(x) = C3 7.7252 cos 7.7252 + sin 7.7252 = 0
L

which gives x = 0.36 L.

Thus, if an extra support is placed at x = 0.36 L, the critical load of the
pinned-clamped beam increases from Pcrit1 = 2.05 2EI / L 2 = 20.23EI / L 2 (no
extra support) to Pcrit2 = 7.72522 EI / L 2 = 59.68EI / L 2 for one optimally placed
support.
Answer: The support should be placed at x = 0.36 L.

Chapter 1

Page 1:23

Resonance
1/14.

4J

4J

L, GK

Ma

2J

L, GK

M ( x)

A shaft carries two flywheels according to the

figure. One of the flywheels is loaded with a
torque M(t) = M0 sin t. Determine the maximum torque in the shaft if = 0.75e and =
0.90e, where e is the eigenfrequency of the
structure. What happens if = e? The
moments of inertia of the two flywheels are 4J
and 2J respectively, and the inertia of the shaft
can be neglected.

Ma

Solution:
Separate the flywheels from the axle
and enter the torque Ma on the axle
2 J M( t ) and on the flywheels. (The inertia of
the axle, compared to the flywheels,
is supposed to negligible here.
Therefore the torque is the same at
the two ends of the axle.)

Introduce the rotations 1 and 2 of the two flywheels. The directions of 1 and
2 are shown in the figure.
For the flywheel 1 the equation of motion becomes
4J 1 = Ma

(a)

For the flywheel 2 the equation of motion becomes

2J 2 = M(t) Ma

(b)

The torsion of the axle is 2 1.

The moment-deformation relationship for the axle is
Ma L
2 1 =
GK

(c)

Eliminate 1 and 2 from the equations (a) to (c). That gives a differential
equation in the unknown moment Ma. One obtains
M M
M sint
a L + a + a = M(t) = 0
(d)
M
GK 4J 2J
2J
2J
Page 1:24

Chapter 1

Assume a particular solution to the differential equation (d) on the form

Ma = A sin t

(e)

Enter the assumption (e) into the differential equation (d). It gives an equation
determining the unknown amplitude A. One obtains
M0
L

3
2
+ A=
(f)

GK 4J
2J
from which is solved

A=

M0 GK

(g)

2JL ( 2 + 3GK / 4JL )

The (angular) eigenfrequency e can now be determined. The eigenfrequency

is the frequency = e giving that the amplitude A in (g) tends to infinity
(which is equivalent to that the system determinant in (f) becomes zero). One
obtains
e =

3GK
4JL

(h)

The torque in the axle can now be calculated. One obtains, from (e) and (g)
2M0
M0 GK
sint
=
sint
(i)
Ma =
2JL ( 2 + 3GK / 4JL )
3 (1 2/2e)
For = 0.75e one obtains
For = 0.90e one obtains

32
M sint = 1.52 M0 sint
21 0

(j)

200
M sint = 3.51 M0 sint
57 0

(k)

Ma =
Ma =

Finally, if = e, the torque Ma in the axle tends to infinity.

It is concluded that an excitation frequency close to a resonance frequency of
the structure might be very dangerous.
Answer: Normalised with respect to the load amplitude M0, the amplitude Ma
of the torque in the shaft is Ma /M0 = Mshaft /M0 = 1.52, 3.51 and , respectively,
for = 0.75e, = 0.90e, and = e.

Chapter 1

Page 1:25

Empty page.

Page 1:26

Chapter 1

Solutions to problems in
T Dahlberg and A Ekberg: Failure, Fracture, Fatigue - An Introduction.
Studentlitteratur, Lund 2002, ISBN 91-44-02096-1.

Chapter 2
Stresses, stress concentration, stresses at crack tip, stress
intensity factor, and fracture criteria
Problems with solutions
Stress concentration
2/1.
A thin-walled circular cylindrical pressure vessel has length L, radius a, and
wall thickness h. For a tube connection a small circular hole is opened in the
wall (and the tube is mounted). The vessel is then loaded with a pressure p.
Determine the maximum stress at the hole due to the pressure if the influence
of the tube connection is disregarded.
Solution:
tan

long

A
B
B

tan

and

long

The stresses in the pressure vessel wall are

(tangentially and longitudinally)
a
a
tan = p
and long = p
(a,b)
h
2h
Study the stress concentration at the hole. There
are four critical points at the hole where high
stress may occur: points A and B, see figure.
Two points have stress A and two points have
stress B. One obtains the stresses

A = 3 long tan = 3 p

a
a
a
p =p
2h
h
2h

(c)

B = 3 tan long = 3 p

a
a 5 a
p
= p
h
2h 2 h

(d)

The stress around the hole varies between these two values. The largest stress
at the hole thus is max = B = 5pa / 2h.
Answer: The largest stress at the hole is max = 5pa / 2h.
FMCHAP2P.DOC/2010-01-15/TD

Chapter 2

Page 2:1

2/2.
A thin-walled circular cylindrical pressure vessel has a small circular hole in
the wall. The hole will be covered with a square plate; the plate being screwed
to the wall with four screws, one in each corner. By taking stress concentration
into account, where do you want to place the screws in the pressure vessel
wall?
(This is perhaps a good example to try to understand stress concentration, but
it is not a good way to fasten the plate you should never make a hole in a
region where you already have stress concentration.)
Solution: See answer.
Answer: The screws should be mounted as far away from the stress
concentration as possible; here it will be at 45o counted from the longitudinal
axis of the pressure vessel.

2/3.
A large plate has a small elliptical hole in it. The ratio of the axes (major to
minor axis) of the elliptical hole is 2 to 1. Determine the stress concentration
factor Kt when the plate is loaded in parallel to the
(a) minor axis, (b) major axis of the ellipse.
Solution:

max
nom

nom

(a)

max
nom

nom

(b)

The maximum stress at an elliptical hole is

given by

a
max = 1 + 2 nom
(a)

b
where a and b are the half-axes of the ellipse.
(a) If the load is parallel to the minor axis, see
Figure (a), one obtains

2
(b)
max = 1 + 2 nom = 5nom

1
(b) If the load is parallel to the major axis, see
Figure (b), one obtains

1
(c)
max = 1 + 2 nom = 2nom

Answer: Stress concentration factor is Kt = 5 and 2, respectively.

Page 2:2

Chapter 2

2/4.

A plate contains an elliptical hole as shown in

the figure. The plate is supported at two
opposite sides; the support being such that all
motion in the y-direction is prevented whereas
motion in the x-direction is possible. The plate
is loaded with an uni-axial stress along the
two other sides.

Under which condition (ratio b / a) will fracture start at point A and at point B,
respectively? The material is linearly elastic (with Youngs modulus E and
Poissons ratio ) up to its brittle fracture at the ultimate strength U.
Solution:
yy

yy

Due to the prevented contraction in the y

direction (strain yy = 0), stress yy will appear in
that direction. Let xx = . Hookes law gives
1
yy = 0 = [yy (xx + zz ) ]
(a)
E
Knowing that zz = 0, one obtains
yy = xx
(b)

Stresses xx (= ) and yy = xx give rise to stresses at the elliptical hole. It is

seen that if half-axis b is large (b >> a), then the largest stress at the hole will
be at point A. Contrary, if a >> b, then the largest stress will appear at point B.
Determine the ratio a / b that gives the same stress at point A as at point B.
Using the elementary case for uni-axial stress:

a
max = 1 + 2 nom and min = nom
(c)

b
one obtains (yy = xx is used)

b
A = 1 + 2 xx xx

and

a
B = 1 + 2 xx xx

Set A = B. It gives

a
1 + 2 xx xx = 1 + 2 xx xx

a
b
Chapter 2

(d)
(e)

(f)
Page 2:3

Simplifying (f) gives

1+

b
a
=0
a
b

(g)

Let b = a. It gives
2 + (1 ) = 0

(h)

Knowing that > 0, one obtains the solution = .

Thus, when = , i.e. when b = a, one has A = B. If b > a, then the stress
at point A is the largest, whereas b < a gives that the stress at point B is the
largest.
Answer: Ratio b / a > gives fracture (largest stress) at point A and b / a <
gives fracture at point B.

Stresses at crack tip

2/5.
y

Calculate, and show in a graph, the distribution

of the maximum shear stress close to a crack tip
loaded in Mode I. The material is linearly
elastic with Poissons ratio = 0.3.

r
x
z

The stress components close to the crack tip are

KI

xx =
cos 1 sin sin
2
2
2
2
r

yy =

cos 1 + sin sin

2
2
2
2
r

KI

xy =

cos sin cos

2
2
2
2
r

KI

yz = zx = 0
and, at plane deformation,
at plane stress

Page 2:4

zz = (xx + yy )
zz = 0

Chapter 2

Solution:
Determine the maximum shear stress by use of the Tresca criterion. The
principal stresses are then needed. The maximum shear stress is
1
prs
max = (prs
(a)
max min)
2
where superscript prs stands for principal stress.
In this problem we have a plane state (either plane stress or plane deformation)
in the xy plane. Therefore, the z direction is one principal direction and the
stress component zz is one of the principal stresses.
In case of plane stress one has the principal stress in the z direction
zz = 0

(b)

and in the case of plane deformation (plane strain) one obtains the principal
stress
KI

zz = (xx + yy ) = 2
(c)
cos
2
2

Next, determine the two other principal stresses, called 1 and 2. They are
1 , 2

xx + yy
=

xx yy 2 2
+ xy

cos
=
2
2
r

KI

=
=

KI
2
r

KI
2
r

cos

cos

3 2
cos sin sin
+ cos2 sin2 cos2

2
2
2
2
2
2

2 23
2 3
cos sin sin
+ cos

2
2
2
2

1 sin
2
2

(d)

The principal stresses 1, 2, and 3 thus become, with = 0.3,

Chapter 2

Page 2:5

1 =
2 =

KI
2
r

KI
2
r

3 = 0
3 = 0.6

cos

1 + sin
2
2

(e)

cos

1 sin
2
2

(f)

if plane stress
KI
2
r

cos

if plane deformation
2

(g)
(h)

Plane stress
Now the plane stress situation will be investigated. The largest principal stress
prs
then is prs
max = 1 and the smallest principal stress is min = 3 = 0. The
maximum shear stress then becomes
1 1 KI

(i)
max = =
cos 1 + sin
2 2
2
2
2
r

It is seen that max is a function of . Find the value of that gives the largest
max. By solving
d max
=0
(j)
d
one finds that max has its largest value when = / 3 = 60 degrees = 60o (and
the smallest value of max occurs when = and it is then max = 0).
Thus, at plane stress the maximum shear stress will be found in the direction
= 60 degrees from the x axis. One has
KI
(k)
max( 60o ) = 0.65
2
r

Plane deformation
At plane deformation one still has that stress 1 is the largest principal stress.
But which one of 2 and 3 is the smallest? To investigate this, set 2 = 3,
which gives
if < 47o then 2 > 3
and

Page 2:6

if > 47o then 2 < 3

Chapter 2

(l)

Assume that the maximum value of max will be obtained for < 47o. Then
max will be given by the principal stresses 1 and 3. One obtains
1
max = (1 3)
2
=

1 KI

cos 1 + sin 0.6 cos

2
2
2
2
2
r

(m)

Next, find the value of that gives the largest value of max in (m). By solving
d max
=0
(k)
d
one finds that max has its largest value when = 75.8 degrees. But this is not
in agreement with the assumption made, namely that should be smaller than
47o. Thus, we have to repeat these calculations once again with a new
assumption:
Assume that the maximum value of max will be obtained for > 47o. Then 2
is the smallest principal stress, and one obtains
1
1 KI

max = (1 2) =
cos 2 sin
2
2
2
2
2
r

(n)

Again, determine the value of that gives the largest value of max. By solving
d max / d = 0 one finds that max has its largest value when = / 2 = 90
degrees. This is now in agreement with the assumption that should be larger
than 47o. It is then concluded that
KI
max(90o) = 0.5
(o)
2
r

Answer: The maximum shear stress is

at plane stress:

at plane deformation:

max( 60o ) = 0.65

max(90o) = 0.5

KI
2
r

KI

2
r

Relations (i) and (n) give the graphs asked for (not shown here).

Chapter 2

Page 2:7

2/6.

The stress components in front of a crack tip

loaded in Mode III are
KIII

xz =
sin
2
2
r

r
x
z

yz =

KIII
2
r

cos

xx = yy = zz = xy = 0

In which direction will the shear stress have its maximum?

Solution:
yz

xz

The shear stress on a surface in the xy-plane is

KIII

2
=

=
sin2 + cos2

xz yz
2
2
2
r

r
x
z

KIII
2
r

= max

(a)

Thus, the shear stress (= max) is the same in all directions (thus independent
of the direction ).
Answer: The shear stress is the same in all directions.

Page 2:8

Chapter 2

Stress intensity factor

2/7.
Study the following crack geometries (a) to (f) and determine the characteristic
lengths a of the cracks so that the probability of crack growth is the same for
all cracks. All cracks are loaded in Mode I with a remote stress 0. The
material is linear elastic.
(a) Central crack of length 2a in a large plate.
(b) Central crack of length 2a in a quadratical plate with side length 4a.
(c) Edge crack of length a in a large plate.
(d) Two opposite edge cracks of length a in a rectangular plate with side
length (width) 2W = 4a and height h >> W.
(e) Half-elliptical surface crack, a deep and 4a (= 2c) long, in a thick plate.
(f) Elliptical crack with principal axes a and 2a embedded in a thick plate.
Solution:
Handbook formulae give stress intensity factors KI (functions fi refer to
Appendix 3 in the textbook)
2
(a)
KI 2
a h
1 KI
KI = 0
a f1 W , W where f1 = 1.0 gives a = = 0.318
0
0
(b)
KI 2
a h
KI = 0
a f1 W , W where f1 = 1.34 gives a = 0.177
0
(c)
(d)
(e)
(f)

a
KI = 0
a f5 W
a
KI = 0
a f4 W
a
KI = 0
a f7 c
a
KI = 0
a f8 c

KI 2
gives a = 0.254
0

where

f5 = 1.12

where

KI 2
f4 = 1.17 gives a = 0.233
0

where

KI 2
f7 = 0.896 gives a = 0.397
0

where

KI 2
f8 = 0.826 gives a = 0.467
0

Answer: Ratio a / (KI / 0)2 is

(a) 0.318; (b) 0.177; (c) 0.254; (d) 0.233; (e) 0.397; (f) 0.467.

Chapter 2

Page 2:9

Fracture modes
2/8.
A thin-walled circular cylinder contains a crack.
R
The plane of the crack is inclined an angle (=
N T ) with respect to the longitudinal axis of the
cylinder, see figure. The cylinder is loaded by a
torque T and an axial tensile force N.
t

Determine a relationship between T, N and so that the crack is loaded in

Mode I only, and in Mode II only. Assume that R >> t so that the curvature of
the cylinder may be neglected when the stress intensity factor is determined.
Solution:
area = 1

xy

sin
cos
xy

yy

xx

The axial tensile force N gives the normal stress

xx in the longitudinal direction, and the torque T
gives shear stress xy, where
N
T
(a,b)
xx =
and xy =
2Rt
2R 2t
The stress yy is zero (yy = 0).

Equilibrium equations in the directions of n and n give (in the equations

below the area is written after the multiplication sign , i.e. the area equals 1
for the largest area and sin and cos , respectively, for the two smaller areas)
n 1 = xx sin sin + 0 xy sin cos xy cos sin

2T
N
cos

sin +

= xx sin2 2xy cos sin = sin

2Rt
2R 2t

and

n 1 = x sin cos 0 + xy sin sin xy cos cos

=

N
T
cos 2
sin 2 +
4Rt
2R 2t

The loading of the crack is in Mode I if n 0 and n = 0, which gives

T
N
T
R
n = 0 =
cos
2
giving
sin 2 +
=

tan 2
4Rt
N
2
2R 2t
Loading of the crack is in Mode II if n = 0 and n 0, which gives, either

Page 2:10

(c)

Chapter 2

(d)

(e)

n = 0 = sin giving = 0

(f)

or

2T
N
cos

sin +

n = 0 = sin

2Rt
2R 2t

giving

T
R
= tan
N
2

(g)

Answer: Load ratio T / N = ( R / 2) tan 2 gives Mode I loading of the crack,

and ratio T / N = ( R / 2) tan gives Mode II loading.

Linear elastic fracture mechanics (LEFM)

2/9.
In a thin-walled pressure vessel there is a risk
that a 300 mm (= 2a) long longitudinal through2R
2a
thickness crack will develop almost immediately
as the pressure vessel is loaded (due to a bad
t
welding, for example). In an approximate
l
estimation of the stress intensity factor the
curvature of the pressure vessel wall was disregarded. The stress intensity
factor KI was then determined for a crack in a plane, infinitely large plate, and
it was found that the stress intensity factor so determined was only half of the
fracture toughness KIc of the material.
What would a more accurate analysis give? Could the pressure vessel be used
without danger? The vessel is manufactured of steel, and it is loaded with a
pressure p. The conditions for linear elastic fracture mechanics are assumed to
be fulfilled.
Numerical data: pressure vessel radius is R = 0.4 m, overall length of pressure
vessel is l = 5 m, crack length is 2a, where a = 0.15 m, and the wall thickness
is t = 0.015 m.
Solution:
The stresses in the wall are
rr = 0 ,

pR
,
t

zz =

pR
, and all ij = 0
2t

(i j)

(a)

The requirements for linear elastic fracture mechanics (LEFM) to be valid are
assumed to be fulfilled.

Chapter 2

Page 2:11

Case A. Study a crack in a large flat plate. The stress intensity factor then is
pR
KI = 0
(b)
a f1(0, 1) = 0
a 1 = t
a
KI =

But

KIc
2

gives

KIc = 2

pR
a

(c)

The fracture toughness KIc of the material has now been determined.
Case B. Study a crack in a cylindrical wall. The stress intensity factor will now
be obtained from a handbook (Case 14 in Appendix 3 of the textbook). The
following conditions must be fulfilled:
R >> t, a >> t and L >> R. They are all fulfilled.
Determine . One has
=

0.15

0.4 0.015
Rt

This gives f14(1.936) = 2.343. Thus

= 1.936
KI = 2.343

(d)
pR
a

(e)

which is larger than the fracture toughness KIc and fracture will be expected!
Answer: Exact analysis (taking curvature into account) gives KI 1.17KIc,
implying that failure will occur.

2/10.
M0

2a
M0

A through-thickness crack of length 2a has been

found in a large plate. The plate is subjected to
a bending moment M0 (Nm/m) per unit length.
Determine at which moment M0max crack growth
will occur.

Numerical data: a = 0.02 m (crack length 2a), plate thickness t = 0.03 m, yield
strength Y = 1300 MPa, and fracture toughness KIc = 110 MN/m3/2.

Page 2:12

Chapter 2

Solution:
Case 13 in the Appendix 3 of the textbook gives
6M0
a
KI = 0
a f13 W where 0 = 2 and M0 in Nm/m
t
Ratio a / W = 0 gives f13 = 1. Thus

KI =

6M0
t2

(a)

(b)

Crack propagation is expected when KI = KIc if linear elastic fracture

mechanics (LEFM) can be used. One has
KIc 2
110 2
= 0.0179 m
2.5 = 2.5
1300
Y

(c)

which is smaller than a, t and W a. Thus LEFM can be used.

The fracture criterion KI = KIc gives

6M0
t2

a = KIc

(d)

from which M0 is solved. One obtains

M0 =

KIc t 2
6
a

110 0.032
6
0.02

= 0.0658

MNm
m

(e)

This value of the moment M0 gives the stress in the plate. One obtains
6M0
= 2 = 439 MPa = 0.34Y
t

(f)

Answer: Crack growth is expected when bending moment of the plate is

M0max = 66 kNm/m, giving maximum remote stress = 0.34 Y in the plate.
2/11.

b 2a
l

A cantilever beam contains a through-thickness

crack. The beam is loaded by a system of forces
11 P
d
(11P and 9P) according to the figure. Determine
d
at which value of P failure may be expected.
9P
Numerical data: a = 0.02 m (crack length 2a),
beam (plate) thickness t = 0.03 m, plate width b
= 0.08 m, d = 0.06 m, l = 3.2 m, yield strength
Y = 600 MPa, and fracture toughness KIc = 50
MN/m3/2.

Chapter 2

Page 2:13

Solution:
The cantilever beam (or plate) will be loaded in tension and in bending. The
axial force becomes N = 20P loading the plate in tension, and the bending
moment becomes M = 2Pd.
The stress intensity factor KI is obtained by superposition of the two loading
cases. One obtains (Cases 1 and 13 in Appendix 3 of the textbook)
a h 6M
a
N
KI = KIN + KIM =
a f1 , + 2
a f13

W W b t
W
bt
=

a 20P f 0.02 , 1.6 + 12Pd f 0.02

1
13

P
a 20 1.186 + 12 d 1.162
bt
t

bt

0.04 0.04

= (23.72 + 27.89)

0.04

P
a = 51.6 P
a
bt
bt

(a)

If the conditions for linear elastic fracture mechanics, LEFM, at plane strain
are fulfilled, then failure will occur when
(b)
KI = KIc
Can LEFM be used?
KIc 2
50 2
= 0.0174 m
2.5 = 2.5
600
Y

(c)

This number (0.0174 m) is less than the crack (half-)length a, plate thickness t,
and the distance b / 2 a from the crack tip to the edge of the plate, so LEFM
can used.
Thus,

KI = 51.6

P
a = K = 50 MN/m3 / 2
Ic
bt

(d)

giving
P = Pcrit =

KIc bt
51.6
a

50 0.08 0.03
51.6
0.02

= 0.00928 MN = 9300 N

Failure is expected at P = Pcrit = 9.3 kN.

Answer: Failure is expected at critical load Pcrit = 9.3 kN
Page 2:14

Chapter 2

(e)

2/12.
M0

An edge crack has been discovered in a long

beam, see figure. The fracture toughness KIc and
the yield strength Y of the material depend on
temperature as given below.

M0

a
P0

P0

(a) Over which temperature range may linear elastic fracture mechanics be
used?
(b) For the upper limit of this temperature range, determine at which loadings
P0 and M0 fracture may be expected.
Numerical data: crack length a = 0.1 m, beam height W = 0.5 m, beam width
t = 0.2 m.
K Ic
MN/m 3/2
160
140
120
100
80
60
40

MPa
700

K Ic

600
500
400
-200

-100
0
Temperature T oC

100

Figure. Fracture toughness KIc and yield strength Y as function of

temperature T
Solution:
(a) First, investigate if linear elastic fracture mechanics, LEFM, can be used.
One has
KIc 2
2.5 minimum of (a, t, W a) = a
Y
which gives

KIc

Y

a
= 0.2 (
m)
2.5

(a)

Thus, numerically, one should have KIc < 0.2Y. This condition is fulfilled if
temperature T is T 3oC, see diagram (at T = 3oC, approximately, one
obtains KIc = 100 MN/m3/2 and Y = 500 MPa).
Chapter 2

Page 2:15

(b) Superposition of the two loading cases tension and bending gives (Case 5
and 6 in the Appendix 3 of the textbook)
P0
a 6M0
a

KI = KIP + KIM =
a f5 +

a
f

6
W t W2
W
bt
=

a (1.366P + 12.902M )
0
0

0.1 6M0 0.1

P0 f5 +
f
tW
0.5 W 6 0.5
(b)

tW

Fracture will occur when

KI = KIc

(c)

At temperature T = 3oC the fracture toughness KIc is KIc = 100 MN/m3/2.

Thus, at this temperature fracture will occur when
KI =

giving

a (1.37P + 12.90M ) = K
0
0
Ic

(d)

tW

(1.37P0 + 12.90M0) =

KIc tW

100 0.2 0.5

0.1

= 17.84 MN

(e)

with P0 in MN and M0 in MNm.

Thus, at fracture one has
P0c + 9.44M0c = 13.1 MN

(f)

where P0 should be entered in MN and M0 in MNm.

Answer: Temperature T should be T 3oC. At temperature T = 3oC the
yield strength is Y = 500 MPa and the fracture toughness is KIc = 100
MN/m3/2 giving the maximum load P0 + 9.44M0 = 13.1 MN, where P0 should
be entered in MN and M0 in MNm.

Page 2:16

Chapter 2

2/13.
A thin-walled circular cylinder contains a crack
o
T forming an angle 30 to the longitudinal axis of
the cylinder, see figure. Determine at which
torque T crack propagation will be expected. In
this case, the curvature of the wall may be
neglected when the stress intensity factor is
estimated (i.e., the radius R is very large
compared to crack length a and wall thickness
t).

30 o

R
t

The equivalent stress intensity factor Ke is

Ke2 = KI2 + KII2 +

4
2
KIII
+1

Numerical data: radius R = 0.2 m, crack (half-)length a and wall thickness t are
a = t = 0.005 m, yield limit Y = 1200 MPa, and fracture toughness KIc = 50
MN/m3/2. (For a small crack in a large plate the stress intensity factor in Mode
II is KII = x y
a g , where (here) g = 1.)
Solution:
Can linear elastic fracture mechanics, LEFM, be used? One has
KIc 2
50 2
(a)
= 0.00434 m
2.5 = 2.5
1200
Y
Here a = t = 0.005 m > 0.00434 m (and W a is larger), so LEFM can be
used.
xy

xy

60 o
30

x
o

Consider an element surrounding the crack, see

figure. The stress on the element is
T
T
(b)
xy =
=
Wv 2R 2t
where Wv is the section modulus in torsion of
the cylinder.
The stress xy gives a normal stress n across the
crack, loading the crack in Mode I, and a shear
stress n, loading the crack in Mode II.

Chapter 2

Page 2:17

The stresses n and n become, with = 60o,

n = xx cos 2 + yy sin2 + 2xy cos sin =

xy

1
n = (xx yy ) sin cos + xy (cos2 sin2 ) = xy
2

(c)
(d)

In Mode I the stress intensity factor KI becomes (for a small crack in a large
plate the factor f1 becomes f1 = 1)
KI = n
a f1 =

xy

(e)

In Mode II the stress intensity factor KII becomes (for a small crack in a large
plate the factor g becomes g = 1)
1
KII = n
(f)
a g = 2 xy
a
The equivalent stress intensity factor Ke becomes
3 1
Ke2 = KI2 + KII2 = 2xy a + = 2xy a
4 4
Fracture is expected when Ke = KIc (if LEFM can be used). This gives
T
Ke = xy
a =
a = KIc

2R 2t
Thus,

T=

KIc 2R 2t

50 2 0.22 0.005
0.005

= 0.501 MNm

(g)

(g)
(h)

At torque T = 500 kNm crack propagation is expected. (The LEFM conditions

were checked above.)
Answer: Crack propagation is expected when torque T = 500 kNm.

Page 2:18

Chapter 2

Solutions to problems in
T Dahlberg and A Ekberg: Failure, Fracture, Fatigue - An Introduction.
Studentlitteratur, Lund 2002, ISBN 91-44-02096-1.

Chapter 3
Small plastic zone at crack tip
Problems with solutions
The Irwin correction
3/1.

r
a

At a circular hole (radius r = 10 mm) in a large

plate two cracks (length a = 5 mm) have been
discovered, see figure. Determine the maximum
remote stress (the critical stress) the plate may
be loaded with.
Data: plate thickness t = 20 mm, fracture
toughness KIc = 40 MN/m3/2, and yield limit Y
= 500 MPa.

Solution:
Can linear elastic fracture mechanics be used?
KIc 2
40 2
2.5 = 2.5
(a)
= 0.016 m
500
Y
The plate thickness t is large enough to ensure plane strain, but the crack
length a is too short. Therefore, use the Irwin correction of the crack length.
For plane strain (t is large enough), the correction at critical stress, i.e. when KI
= KIc, is
2
1 KIc
1 40 2
r1 =
=

= 0.00034 m
6 Y 6 500

FFFCH3P.DOC/2010-01-15/TD

Chapter 3

(b)

Page 3:1

Correction of LEFM now gives

a + r1
KI =

(a
+
r
)
f
0.00534 1.78
1 3 r + a + r =
1
= KIc = 40 106 N/m3 / 2

(c)

from which = crit = 173 MPa is solved.

Answer: The critical remote stress is, approximately, = crit = 173 MPa.

3/2.
A beam with rectangular cross section, t = 20 mm and W = 40 mm, is loaded
with a bending moment M as shown in the figure. A 12 mm deep crack (a =
12 mm) has appeared in the beam (see figure). In order to increase the
load-carrying capacity of the cracked beam, the material around the crack is
removed. It is decided that 9 mm of the beam height should be removed by
grinding. After the grinding the beam height thus is 31 mm. The crack,
however, is still there, and its depth is now 3 mm.
Material properties: KIc = 40 MN/m3/2 and Y = 900 MPa.
removed

9 mm

(a) How much has the ultimate bending moment been increased by the
grinding? (Thus, calculate the ultimate load M of the beam before grinding and
after the grinding.)
(b) The purpose of the grinding was perhaps to remove all the crack. What
bending moment can be put on the beam if 12 mm were removed (i.e. even the
crack tip has been removed)? In this case the yield limit of the material limits
the loading.

Page 3:2

Chapter 3

Solution:
Can LEFM (linear elastic fracture mechanics) be used?
KIc 2
40 2
2.5 = 2.5
(a)
= 4.94 103 m = 4.94 mm
900
s
This length, 4.94 mm, is less than t (= 20 mm), a (= 12 mm), and W a (= 28
mm), respectively. Thus, it is OK to use LEFM for the first crack length (12
mm) but not for the second (length 3 mm).

Determine the ultimate load before grinding, i.e. the ultimate bending moment
for the beam with cross section 20 mm by 40 mm and crack depth 12 mm:
Case 6 in Appendix 3 of the textbook gives
KI =

a
0.012
6M
6M
0.012
a f6 W =

f6 0.040
2
2
tW
0.020 0.040

= KIc = 40 106 N/m3 / 2

(b)

Diagram gives f6 = 1.115, and one obtains the ultimate load M = M1 = 985 Nm.
After the grinding the cross section becomes 20 mm by 31 mm, and crack
depth is 3 mm. LEFM can not be used any longer. Use the Irwin correction of
the crack length. One obtains for plane strain, when the stress intensity factor
reaches KIc,
2
1 KIc
r1 =
= 0.000105 m
6 Y

(c)

Case 6 gives
aeff
6M
0.003105
6M

a
f

0.003105
f

KIeff =

eff

6
6
0.031
W 0.020 0.0312
tW 2
= KIc = 40 106 N/m3 / 2

(d)

Diagram gives f6 = 1.057, giving M = M2 = 1227 Nm.

(b) If all the material surrounding the crack had been removed, i.e. if 12 mm of
the beam height were taken away (this was perhaps the original purpose of the
grinding operation), then fracture mechanics theory needs not be used at all. In
that case, determine the bending moment the beam can be loaded with, if
plastic deformation of the beam should be avoided. One obtains

Chapter 3

Page 3:3

max =

M W /2
M 0.028 / 2
= Y = 900 MPa
=
I
0.020 0.0283 / 12

(e)

which gives M = 2352 Nm. (Conclusion: make sure that also the crack tip is
removed, if you try to improve the load-carrying capacity of a structure by
removing a crack.)
Answer: (a) The maximum bending moment M increases from 985 Nm to
1227 Nm, thus by 25 per cent. (b) The moment can be increased to M = 2352
Nm if the full crack (with crack tip included) is removed.

The Dugdale model

3/3.

(a)

y
x

2a

A large steel plate of an elastic, ideally plastic

material (Y = 500 MPa) contains a throughthickness crack of total length 50 mm (= 2a). At
a tension test of the plate the crack was found to
start growing at the remote stress = 300
MPa.
(a) At which remote stress would a 150 mm
long crack start to grow? Use the Dugdale
model of crack tip opening displacement (a) as
criterion for crack growth initiation (CTOD
criterion). One has

8 a Y
1

(a) =
ln
E
cos ( / 2Y)

(b) For the two cases 2a = 50 mm and 2a = 150 mm, calculate the size
(length) of the plastic zone at the crack tip when the stress is such that the
crack starts to propagate.
Solution:
The first measurement, at crack length 2a = 50 mm, gives the critical value of
the crack tip opening displacement (CTOD) (a)crit. One obtains

Page 3:4

Chapter 3

(a)crit =
=

8 acrit Y
1

ln
E
cos ( crit / 2Y)

8 0.025 Y
Y
1
= 0.03382957
ln
E
E
cos ( 300 / 2 500)

(a) At which stress would a 150 mm long crack start to grow?

The CTOD criterion gives, with (a)crit just calculated in (a),

Y 8 0.075 Y
1

ln
(a)crit = 0.03382957 =
E
E
cos ( / 2 500)

(a)

(b)

Solving for gives = 184 MPa.

(b) The length of the plastic zone is obtained from
1

1
=a
cos ( / 2 Y)

(c)

For 2a = 50 mm, one obtains, with remote stress = 300 MPa,

= 0.025
1 = 0.0175325 m = 17.5 mm
cos ( 300 / 2 500)

For 2a = 150 mm, one obtains, with remote stress = 184 MPa,

= 0.075
1 = 0.0145492 m = 14.5 mm
cos ( 184 / 2 500)

(d)

(e)

Answer: (a) At crack length 150 mm the critical stress is 184 MPa, (b) at
crack length 50 mm the length of the plastic zone is 17.5 mm (for critical
remote stress 300 MPa), and at crack length 150 mm the length of the plastic
zone is 14.5 mm (for critical stress = 184 MPa).

Chapter 3

Page 3:5

3/4.

(0)

During a tension test of a large plate with a

central crack the crack opening (0) was
recorded as a function of the load. The recorded
curve deviated from a straight line. The
deviation from linearity may be explained partly
by the appearance of a plastic zone at the crack
tip and partly by crack growth during the
loading.

y
x

2a

(a) How large deviation from linearity will be expected due to the plastic zone
at the crack tip? The remote stress is Y / 2.
(b) The deviation from linearity of the recorded curve appeared to be 30 per
cent. Calculate the crack propagation during the loading (assume that
symmetry is maintained, i.e. the two crack tips move the same distance).
The material is linearly elastic, ideally plastic with yield strength Y. Plane
stress is at hand so that the Dugdale model may be used. The crack opening
displacement (COD) is
(0) 8Y 1 + sin( / 2Y)
=
ln
a
E
cos( / 2Y)
and if << Y one obtains
4 KI
(0)
=
a
E
a

2 KI
1 +

+ ..
24

Solution:
(a) Using Dugdales model, the crack opening displacement (0) becomes
(0) 8Y 1 + sin( / 2 Y)
=
ln
a
E
cos( / 2 Y)
=

8 2 1 + sin( / 4 )
ln
= 4.4888
E
cos( / 4)
E

If << Y (i.e. if Y is very large) one has

Page 3:6

Chapter 3

(a)

4 KI
(0)
2 KI
1 +
=

+ ..
a
E
a 24 Y

(b)

Using KI =
a and << Y in (b), one obtains the linear elastic solution.
(If Y tends to infinity, no yielding will occur, and the solution obtained is for
the fully elastic case.) This gives

a
(0) 4
=
(c)
{ 1 + 0} = 4
a
E
E
a
The difference between the displacements in (a) and (c) comes from the plastic
deformation at the crack tip. The deviation from linearity is
4.4888 4
= 0.1222 = 12.2 per cent
(d)
4

(b) At the experiment, the deviation from linearity was 30 per cent. The plastic
deformation at the crack tips gives 12.2 per cent deviation from linearity only,
so the remaining deviation up to 30 per cent is explained by crack growth; i.e.,
the crack has propagated during the loading of the structure.
Calculate the final crack length afinal. Using (0) = 1.3 elastic, where elastic was
calculated in (c), the Dugdale model gives
a 8 2 afinal 1 + sin( / 4 )
=
ln
(e)
(0) = 1.3 4
E
E
cos( / 4 )
from which is solved
afinal =

1.3 4
a = 1.158 a
4.4888

(f)

Answer: (a) Deviation from linearity due to plastic zones is 12.2 per cent, and
(b), crack has grown to final length afinal = 1.158a during the loading, where a
is the original crack length in the unloaded structure.

Chapter 3

Page 3:7

3/5.

During a tension test of a large plate with a

central crack the crack opening (0) was recorded as a function of the load P. The recorded
P
curve deviated from a straight line as given in
the figure. Symmetric stable crack growth was
x
2 h observed during the loading of the plate.
Calculate the crack growth during the loading.

(0)

2a
2W

0.28

P MN

(0 )
0.18 mm

The material is linearly elastic, ideally plastic

with yield strength Y. The plate is thin so that
the stress state is plane (PS). Thus, the Dugdale
model may be used.
Numerical data: crack length a0 = 0.01 m, plate
thickness t = 0.002 m, plate size h = W = 0.2 m,
yield limit Y = 600 MPa, and modulus of
elasticity E = 200 GPa.

Solution:
At fracture the remote tensile stress is

c =

Pc
2Wt

The crack opening displacement (COD) at fracture is

8 Y ac 1 + sin ( /2Y)
ln
c(0) =
E
cos ( /2Y)

(a)

(b)

which gives

1 + sin (Pc /4Wt Y) 1

ac = E c(0) 8 Y ln
cos (Pc /4Wt Y)

(c)

According to the diagram, fracture occurred when Pc = 0.28 MN and c(0) =

0.1810 3 m, which gives
ac = 0.0218 m, giving 2ac = 0.0436 m

(d)

During the loading of the plate, the crack has grown the amount
2a = 2ac 2a0 = 0.0436 0.0200 m = 0.0236 m = 23.6 mm.

(e)

Answer: The crack growth is 2a = 23.6 mm (thus, the crack has propagated
11.8 mm at each crack tip).
Page 3:8

Chapter 3

Solutions to problems in
T Dahlberg and A Ekberg: Failure, Fracture, Fatigue - An Introduction
Studentlitteratur, Lund 2002, ISBN 91-44-02096-1

Chapter 4
Energy considerations
Problems with solutions
Surface energy, stresses at crack tip
4/1.
The surface energy ws and the cohesive strength
c of a material may (approximately) be
x
determined from the following simple model of
i
the inter-atomic forces:
Assume that the force i per unit area between
i
atomic planes may be represented by the
c
function
x
5x0
x0
x x0
i = c sin
4
x0
where x0 is the distance between the planes of the atoms in the unloaded state
and x is the atomic separation when the material is loaded. The factor c is a
material parameter: the bond strength.
Also, assume that the fracture limit is obtained when the strain is 25 per cent,
i.e. when x = 5x0 /4.
i

(a) Determine the parameters , c and ws in terms of E and x0, where E is the
modulus of elasticity of the material.
(b) Compare the fracture criterion by Griffith (released energy equals surface
energy) with a stress criterion for fracture of a material with a crack. Use a
small through-thickness crack in a large plate and compare the remote stresses
giving crack growth in the two cases.

FFFCH4P.DOC/2010-01-15/TD

Chapter 4

Page 4:1

For the stress criterion, assume that the mean stress over n planes of atoms in
front of the crack tip is a measure of the loading of the material. Determine
how far away in front of the crack tip (measured in number of atomic planes)
the mean stress must be equal to the cohesive strength of the material in order
to make the stress criterion equivalent to the energy criterion by Griffith. (The
stress state may be considered as plane.)
Solution:
(a) At small displacements (i.e., when x is close to x0), Hooke s law gives
x x0
i = E = E
(a)
x0
d i E
=
d x x0

which gives

From the equation given in the problem formulation

(b)
i = c sin

x x0
x0

d i
x x0

cos
= c
dx
x0
x0

one obtains

Equations (c) and (b) give, for x = x0,

di(x = x0)

E
cos 0 =
= c
dx
x0
x0

giving

c =

(c)

From the problem formulation, see figure above, one has, when x = 5x0 /4,
d i
=0
dx
which gives
Thus,

(e)

d i(x = 5x0 /4)

5 x0 /4 x0

cos
=0
= c
dx
x0
x0

(f)

1
which gives =
and = 2
4 2

(g)

1
cos = 0
4

Enter = 2 into (d). It gives the cohesive stress (cohesive strength)

E
c =
2

Page 4:2

(d)

Chapter 4

(h)

It is noted that this stress is very high: in the order of E / 10. The ultimate
strength U of a very high-strength steel is in the order of E / 100, which means
that in practice, fracture will occur due to other mechanisms than separation of
the atomic planes.
The surface energy may now be determined from the area below the curve
i(x). One obtains
3x0 /2

2ws =
x0

3x0 /2

i(x) dx =
x0

x x0
E
dx
sin 2
2
x0
3x0 /2

x x0
E x0
=
cos 2

2 2
x0 x

=
0

E x0
2 2

(i)

where the factor 2 in front of ws is there because two new surfaces are created
when the crack appears.
Thus
E x0
[Nm / m2 ]
(j)
ws =
2
4
(b) The Griffith criterion
According to Griffith, the energy that is released when a crack propagates
should be equal to the energy needed to create the new crack surfaces. By use
of the stress intensity factor KI, this crack propagation criterion can be written,
see equation (4.34) in the textbook,
+1 2
(k)
K t a = 2ws t a
8 I
where = (3 ) / (1 + ) at plane stress, and is the shear modulus of the
material: = E / 2(1 + ). The crack growth is a (the crack length a has
increased by a) and the plate, in which the (through-thickness) crack is
situated, has thickness t. Expression (k) is thus given for the new crack area
ta created. Solving (k) for KI gives
KI =

2ws 8
=
+1

Thus, fracture will occur when

2ws 8E

2(1 + ) 1 + + 1
3

=
2ws E

KI = Kc, where Kc =
2ws E

Chapter 4

(l)
(m)

Page 4:3

Now, determine the remote stress G (superscript G for Griffith) leading to

failure. For a small through-thickness crack in a large plate, the stress intensity
factor KI is
KI = G
a

(n)

Crack propagation will occur (for plane stress conditions) when KI = Kc, giving
KI = G
2ws E
a = Kc =

(o)

Thus, according to the Griffith criterion the crack will propagate when the
remote stress is
G =

2E E x0 E
=
a 42

2ws E
=
a

x0
2a

(p)

The stress criterion

y
yy =

We now turn to the stress criterion. The stress

yy in front of the crack tip is

KI

yy =

KI

2r

2r

a
2r

(q)

where is the remote stress (far away from the crack). The coordinate r is
used here in order to avoid confusion with the distance x, or x0, between the
atomic planes.
The mean value yy of yy = yy(r) over n atomic planes is
yy

1
=
n x0 0
=

n x0

1
yy (r) dr =
n x0 0

n x0
a
1
r
]
[ 2
0 =
n x0
2

n x0

2a
n x0

a
dr
2r
(r)

At fracture the mean value yy should reach the cohesive strength c of the
material. This gives
yy =

Page 4:4

2a
E
= c =
n x0
2

Chapter 4

(s)

and

= S =

E
2

n x0
2a

(t)

where superscript S stands for stress criterion.

We have now determined the stress G giving failure according to the Griffith
criterion and the stress S giving failure according to the stress criterion. In the
case that these two criteria give failure at the same remote stress, one has G =
S. This gives
G =

x0
E
=
2a 2

which gives n. Thus n = 4/ = 1.27.

n x0
= S
2a

(u)

In conclusion, if the mean stress yy of yy(r) reaches the cohesive strength c

over n = 4/ atomic planes, then the stress criterion (as formulated here) and
the Griffith criterion will predict crack propagation at the same remote stress
. The material properties should then be as given in the equation in the
problem: i = c sin (x x0) / x0, with parameter values as calculated in
problem (a).
Answer: (a) = 2, c = E /2, ws = E x0 /42, (b) n = 4 / = 1.27.

Tensor notations, strain energy

4/2.
y

Determine the elastic strain energy density in

front of a crack tip loaded in mode III. In which
x
direction has the strain energy its maximum?
z
The elastic strain energy density u is, for
linearly elastic material,
ij ij
u=
2
The stress components in front of a crack tip loaded in mode III are
KIII
KIII

yz =
xz =
sin ;
cos
2
2
2
r
2
r

and

xx = yy = zz = xy = 0
Chapter 4

Page 4:5

Solution:
Use

u=

ij ij
2

Calculate the strains. One has, for i j,

(a)
ij =

2 ij

(b)

where is the shear modulus. Enter the stresses given in the problem, and the
strains calculated in (b) into the expression (a). It gives
2
zx
yz
zy 1 KIII

1 xz

2 sin2 + 2 cos2
u = xz
+ zx
+ yz
+ zy =
2 2
2
2
2
2
2 4
2
r

2
1 KIII
1
2

=
=
KIII
2
2
r 4 r

(c)

Answer: The elastic strain energy u is u = KIII2/4r (thus independent on ),

is the shear modulus.

Energy balance
4/3.
P

a
Y

Two rectangular bars (same cross section b by

h) of linearly elastic material (modulus of
h elasticity E) are glued together along a portion
h
of its length, see figure (a). The glue has -
b
(a) characteristics according to figure (b) (rigid,
ideally plastic up to deformation U and yield
limit Y). Determine the critical load P = Pcrit.
One has a >> h and a >> b.
(b)

Solution:
Two solutions will be given.
(a) Load control gives the potential energy
= U P
where
which gives
Page 4:6

(a)

ij ij
P
U =
dV =
Vol 2
2

(b)

= P / 2

(c)

Chapter 4

Assume that the two rectangular bars may be considered to be cantilever

beams of length a. The displacement /2 of the cantilever beam end due to a
force P then is
P a3
2 P a3
=
giving =
(d)
2 3 EI
3 EI
The potential energy of the structure becomes
P2 a3
P
=
=
2
3 EI

(e)

The energy release due to a change of the crack area is

P 2 (b a)3 P 2 3(b a)2 P 2 a 2
d
d
d
=
=
=

=
d A d (b a) d (b a) 3EI b 3 3 EI
EI b
b3

(f)

The work required to create the crack surface, i.e. the work required to
overcome the glue strength, is
Ws = A Y U ( = A 2ws)
(g)
d Ws
= Y U
dA

which gives

d d Ws
=
,

dA dA

Energy balance, namely

that gives
P = Pcrit =

(h)
gives

EI b Y U
a

P2 a2
= Y U
EI b

E b h 3 b Y U
12 a

bh
a

(i)

E h Y U

2
3

(j)

(b) Alternative solution

From

P2 a3
=
3 EI

Ws = b a Y U

and

(k,l)

one obtaines, for a crack growth a,

P2 3 a2
=
a =
a
a
3 EI

Energy balance, namely

Thus

P = Pcrit =

Answer: Critical load is

and Ws =

= Ws ,

EI b Y U
a2

gives
=

Pcrit =

Chapter 4

Ws
a = b Y U a
a

bh
a

P2 a2
= b Y U
EI

E h Y U

bh
2a
3

(m,n)

2
3

(o)
(p)

E Y
Uh

Page 4:7

4/4.
P

To determine the surface energy ws of a brittle,

linearly elastic material a test specimen, see
figure, was manufactured from the material to
be investigated. When loading the test specimen
2 h crack propagation was obtained when the force
P was P = Pmax. Determine the surface energy of
b
the material. The modulus of elasticity is E, the
cross section is b by 2h, and the crack length is
a where a >> h and a >> b.

Solution:
Two solutions will be given.
(a) Load control gives the potential energy
P
P
= U P =
P =
2
2

/2
P

(a)

Assume that the rectangular bars may be

approximated to two cantilever beams of length
a. The displacement /2 of the beam end due to
a force P is

P a3
2 P a3
=
giving =
2 3 EI
3 EI

(b)

The potential energy of the structure is

P 2 (ba)3
P
=
=
2
3 EI b 3

(c)

The energy release due to a change of the crack area A = b a is

d
P2 a2
d
P 2 3(ba)2
=
=
=
d A d (ba)
3 EI b 3
EI b

(d)

The work required to create the new surfaces is

Ws = 2ws A

Energy balance
which gives

Page 4:8

(G =)

which gives
d d Ws
=
dA dA

d Ws
= 2 ws
dA
gives

P2 a2
= 2ws
EI b

2
2
P2 a2
P 2 a 2 12 6 Pmax a
=
=
ws =
2 b EI 2 b E bh 3 E b 2 h 3

Chapter 4

(e)
(f)
(g)

Thus, the surface energy ws is ws = 6Pmax2 a 2 / E b 2 h 3.

(b) Alternative solution
P2 a3
=
3 EI

From

and

W s = 2 b a ws

(h,i)

is obtained, for a crack growth a,

Ws

P2 3 a2
=
a =
a and Ws =
a = 2 b ws a
a
3 EI
a
Energy balance, = Ws, gives
P2 a2
= 2 b ws
EI

Thus

ws =

(j,k)

P2 a2
2 b EI

(l)

Answer: The surface energy ws is ws = 6Pmax2 a 2 / E b 2 h 3.

4/5.
M

A thin rectangular bar (cross section b by h) of

b
linearly elastic material (Youngs modulus E) is
h
glued to a rigid foundation, see figure (a). The
(a) glue has a linear - characteristics according to
figure (b) (rupture at stress U when
deformation is U). The bar is loaded with a
bending moment M. Determine the critical load
Mcrit. One has a >> b > h.
(b)

Solution:
Two solutions will be given.
(a) Load control gives the potential energy = U P, which here gives
M
M
=U M =
M =
2
2

(a)

Assume that the free part of the rectangular bar may be approximated to a
cantilever beam of length a. The rotation of the beam end due to the bending
moment M is
M a M a 12
=
(b)
=
EI
E bh 3
The potential energy of the structure becomes (crack area A is A = ba)
M
M 2 (ba) 6
M2 a 6
=
=
=
2
E bh 3
E b 2h 3
Chapter 4

(where A = ba)

(c)
Page 4:9

The energy release due to a change of the crack area is

d
M2 6
=
dA
E b 2h 3

(d)

The work required to create the new surface is

1
Ws = A U U
2

d Ws 1
=
dA 2 U U

which gives

(e)

Energy balance,
d d Ws
M2 6 1
= U U

=
gives
dA dA
E b 2h 3 2

(f)

which gives
2
M 2 = Mcrit
=

U U
bh
E b 2h 3 and Mcrit =
12
2

U U E h
3

(g)

(b) Alternative solution

M2 6 a
=
E b h3

From

1
Ws = ba U U
2

and

(h,i)

is obtained

6 M2
=
a
a =
a
E b h3

and Ws =

Ws
1
a = b U U a
a
2

(j,k)

Energy balance, = Ws, gives

6 M2 1
= b U U
E b h3 2

(l)

Thus
M =M
2

2
crit

E U U b 2h 3
N N
2
3
2
dimension: 2 2 m m m = (Nm)
=
12
m m

as obtained in (g).
Answer: Critical load Mcrit is

Page 4:10

Mcrit =

Chapter 4

bh
2
3

E h UU

(m)

Solutions to problems in
T Dahlberg and A Ekberg: Failure, Fracture, Fatigue - An Introduction.
Studentlitteratur, Lund 2002, ISBN 91-44-02096-1.

Chapter 5
Determination of stress intensity factor KI
Problems with solutions
Stress intensity factor
5/1.
/2
y

/2

2h

<< h

An infinitely long strip, height 2h, is split along

half its length, see the figure (the strip thus has
a semi-infinitely long crack in it). The upper
and lower boundary of the strip have been given
a prescribed displacement /2 each. Determine
the stress intensity factor KI for this configuration. Assume plane stress at the crack tip.

Solution:
The strategy to solve this problem is the following: the stress intensity factor
KI can be determined from the energy release rate G. To find G, an energy
consideration is made. When the crack grows (say a) the strip to the right of
the crack tip becomes a shorter, and the energy in that part (i.e. in a volume
a by 2h by strip thickness t) is released. This energy is used to determine G.
First the elastic strain energy density u stored in the strip far away from the
crack tip is determined. To do this, we need to know the stresses and the
strains in the strip.
The strains in the strip (far away to the right of the crack tip) are

yy =
(a,b,c)
, xx = 0 , and zz = yy
2h
The strain xx is zero because no movements can occur in the x direction far
away from the crack tip. Also, all shear strain components are zero because of
symmetry. Hookes law gives, with zz = 0,

SOLCHAP5.DOC/2010-01-15/TD

Chapter 5

5:1

yy =

1
= { xx}
2h E yy

xx = 0 =

1
{ yy}
E xx

from which the stresses are solved:

E
yy =
1 2 2h

and xx = yy

(d)
(e)

(f,g)

The strain energy density u (Nm/m3) becomes

1
1 E 2
1

u = ij ij = (xx xx + yy yy ) =
2
2
2 1 2 2h

(h)

The energy U in a section of the strip with length a is (far away to the right
of the crack tip)
1 E 2
U = u Volume = u 2h a t =
2h a t
2 1 2 2h

(i)

where t is the thickness of the strip.

The strain energy close to the crack tip is not calculated here, because close to
the crack tip the stress field is complicated, and, as will be seen, it is not
necessary to calculate the strain energy in this region.
The energy release rate G is, at displacement control,
dU
1 U(a + da) U(a) 1 U(a) U(a + da)
d
=
=
=
(j)
G =
dA
t da
t
da
t
da
Thus, the energy release when the crack grows a distance a is U =
U(a) U(a+a). The energy release rate G then becomes (at displacement
control), by use of U from (h),
1 U 1 1 E 2
2h t
=
G=
t a t 2 1 2 2h

(k)

Finally, KI is obtained from KI =

EG

, (E = E at plane stress) giving

KI =
EG
=

h
2 2h
1

Answer: The stress intensity factor is KI = E / 2

h(1 2)

5:2

Chapter 5

(l)

5/2.
P

Use the compliance method to determine the

energy release rate G and from that, determine
the stress intensity factor KI for a double
2 h cantilever beam (DCB) specimen. Crack length
is a, thickness is b and specimen height is 2h,
b
where a >> b and a >> h. The material is linear
elastic with Youngs modulus E. Investigate the
two loading cases:

(a) Load control, with the loading force P prescribed.

(b) Displacement control, with displacement prescribed.
Assume plane stress conditions.
Solution:
The force versus displacement relationship for a cantilever beam gives the
displacement / 2 of the beam end due to the loading force P as
P a3
2 P a3
=
giving =
(a)
2 3 EI
3 EI
The compliance C of the specimen is
1 2 P a 3 2a 3
dC 2 a 2
C= =
=
giving
=
(b)
P P 3 EI 3 EI
da EI
(a) The energy release rate G due to a crack extension is, for prescribed load
P,
P 2 dC P 2 2 a 2
G=
=
(c)
2 b da 2 b EI
In the case of plane stress at the crack tip, one obtains

P 2 a 2 P a 2
3
=
(d)
bI
bh
h
It is seen that the stress intensity factor KI increases as the crack grows,
implying that unstable crack growth will be expected (at least if Kc does not
depend on crack length a).
KI2
G=
G E =
giving KI =

(b) The energy release rate G due to a crack extension is, for prescribed
displacement ,
P 2 dC P 2 2 a 2
=
G=
2 b da 2 b EI
Using

P=

3 EI
2 a3

Chapter 5

(e)
(f)

5:3

a 2 3 EI 9 EI 2
=
G=

b EI 2 a 3
4 b a4
2

one obtains

(g)

In the case of plane stress at the crack tip, one obtains

KI2
G E =
giving KI =
G=

9 E 2 I 2 3 E h
h =
3 E h
h
=
4
2
2
4b a
4a
12 2 a

(h)

Here it is seen that the stress intensity factor KI decreases as the crack grows,
implying that stable crack growth will be expected (at least if Kc does not
depend on crack length a).
Answer: Stress intensity factors are (a) KI = 2
3Pa / bh
h , and (b)
2
KI =
3 Eh
h / 4a .
5/3.
M

2h
a

Use the compliance method to determine the

energy release rate G for a double cantilever
beam (DCB) specimen. Crack length is a,
thickness is b and specimen height is 2h, where
a >> b and a >> h. The material is linear elastic
with Youngs modulus E. The specimen is
loaded with bending moments M at the beam
ends.

Solution:
The (generalised) force versus (generalised)
displacement relationship (here this becomes
moment versus rotation) of a cantilever beam
M
gives the rotation angle of the beam end due
a
to the bending moment M. One obtains
Ma
2M a
=
giving = 2 =
(a)
EI
EI
The compliance C of the specimen is
2a
dC 2
C= =
giving
=
(b)
M EI
da EI
The energy release rate G due to a crack extension is, for prescribed load M,
12M 2
M 2 dC M 2 2
=
=
G=
2 b da 2 b EI E b 2 h 3

Answer: The energy release rate G is G = 12M2 / Eb2h3.

5:4

Chapter 5

(c)

5/4.
P

2h
P
a

Use the compliance method to determine the

energy release rate G for a beam with a long
central crack, see figure. The crack length is 2a,
beam thickness is b and beam height is 2h,
where a >> b and a >> h. The material is linear
elastic with modulus of elasticity E. The beam
is loaded symmetrically with two opposite
forces P at the beam centre. The crack is
assumed to grow symmetrically.

Solution:
The force versus displacement relationship of a fixed-fixed beam (length 2a)
gives the displacement / 2 of the beam centre due to the loading force P. One
obtains
P (2a)3 1 1
P a3
=

giving =
(a)
2
3 EI 8 8
12 EI
The compliance C of the specimen is

a3
dC
a2
C= =
giving
=
(b)
P 12 EI
da 4 EI
Note that, due to symmetry the crack is assumed to grow in both directions,
which gives (in d/dA) that dA = d(2ab) = 2bda (A is crack area) (or, it can be
seen as one crack with width 2b, i.e. two crack fronts of length b each). The
energy release rate G due to a crack extension in both directions then is, for a
prescribed force P,
3 P2 a2
P 2 dC P 2 a 2
(c)
=
=
G=
2 (2b) da 4 b 4 EI 4 E b 2 h 3
Answer: The energy release rate G is G = 3P2a2 / 4Eb2h3.

Chapter 5

5:5

5/5.
q

2h

v(x)

x
q

Use the potential energy to determine the

energy release rate G for a beam with a long
central crack, see figure. The crack length is 2a,
beam thickness is b and beam height is 2h,
where a >> b and a >> h. The material is linear
elastic with modulus of elasticity E. The beam
is loaded symmetrically with two opposite
distributed loads q (q in N/m) at beam upper
and lower surfaces.

The deflection of a clamped-clamped beam loaded with a distributed load q is,

for 0 x 2a,
q
v(x) =
(x 4 4ax 3 + 4a 2x 2)
24 EI
Solution:
The energy release rate G for the beam is

1 d
(a)
2b da
The crack front length (the extension of the crack tip in the z direction) is 2b,
because here there are two crack tips and each crack font has length b.
G =

The potential energy is, see equation (4.4) in the textbook,

q v dS
= U W = u dV
boundary
Vol
1
1
=
q v dS
q v dS =
q v dS
boundary
2 boundary
2 boundary

(b)

where U = W / 2 has been used (the material is linear elastic). Expression (b)
2a

gives
1 2a

q v(x) dx +
q ( v(x)) dx
(c)
=
0
2 0

where the first term refers to the upper surface of the beam (loaded upwards)
and the second term refers to the lower surface (loaded downwards). Insert v(x)
into (c) and integrate. It gives
2a

1
q2 1
4a
5
(2a)
= q v(x) dx =
(2a)4 + 4a 2 (2a)3
0
24EI 5
4
3

2 q2 a5
=
45 EI

5:6

(d)

Chapter 5

Finally the energy release rate G can be calculated:

1 d
1 ( 2 q 2) 4 1 q 2 a 4 4 q 2 a 4
G =
=
5a =
=
2b da
2b 45 EI
9 b EI 3 E b 2h 3

(e)

Answer: Energy release rate G is G = 4q2a4 / 3Eb2h3.

Reciprocity relation
5/6.
M

2h
a

Determine the stress intensity factor for the

structure in Problem 5/3 by use of the
reciprocity relation. Use Problem 5/2 as the
known Case (1).

Solution:
Use equation (5.10) in the textbook to determine the stress intensity factor.
One has
ui(1)
4
(2)
(2)

KI =
Ti
d
(5.10)
a
( + 1) KI(1)
The stress intensity factor in Problem 5/2 is known. Thus, the factor KI(1) to use
in the right hand side of (5.10) is (from Problem 5/2)
KI(1) =

P a 2
3
bh
h

(a)

From loading case (1) (Problem 5/2) one needs to know also the (generalized)
displacement at the point where the load is going to be applied in the second
loading case. This means that one needs to know the rotation angle in Case (1)
(due to load P) at the end of the cantilever beam (where the load M is applied
in Case (2)). One obtains
Pa 2
(b)
=
2EI
Also, d/da is needed. One obtains

d 2Pa Pa
=
=
da 2EI EI

(c)

In Problem 5/3 the load, i.e. the bending moment M, is applied in two discrete
points. This means that the intergration in (5.10) along the perimeter can be
discretized. Note that the integral in (5.10) has dimension N/m. In a general
case one has Ti(2) (N/m2) times dui(1)/da (m/m) times d (m), giving
(N/m2) (m/m) m = N/m. When discretizing the integral, Ti(2)d will have the
dimension N/m. In case Ti(2) is a stress (N/m2), Ti(2)d becomes force per unit
width (width in the z direction) of the structure.
Chapter 5

5:7

Here, when the (generalized) displacement is an angle , one obtains the

dimension of d/da as 1/m. The expression Ti(2)d then must have dimension N
(unit of force) to give N/m when multiplying Ti(2)d with d/da. This implies
that the loading moment M (given in Nm in the problem) should be applied as
moment per unit width (in the z direction) when the moment is entered into
(5.10). Thus, the discretized version of (5.10) becomes, in this case,
ui(1)
4
M Pa
4
(2)
(2)

KI =
T

d
=

(d)
i
a
b EI
( + 1) KI(1)
( + 1) KI(1)
Here, at the end of the expression one has d/da from (c), next to that the
bending moment per unit width M / b, and in front of M / b the factor 2, which
is there because there are two bending moments.
Using that = E / 2(1 + ) (the shear modulus), and from equatio (4.34) that
= (3 ) / (1 + ) for plane stress conditions at the crack tip, one obtains
KI(2) =

4E (1 + ) bh
h 2 M Pa = 2
3 M

2(1 + ) 4
b EI bh
Pa 2
3
h

(e)

From Problem 5/3 is obtained

KI(1) =
EG
=

which is in agreement with (e).

E 12M 2 2
3 M
=
E b 2h 3 bh
h

Answer: The stress intensity factor is, cf. answer to Problem 5/3,

5:8

Chapter 5

(f)

KI =

2
3 M
bh
h

5/7.

Determine the stress intensity factor for the

structure in the figure by use of the reciprocity
relation. Use Problem 5/2 as the known Case
P
(1).
Case (1)
2 h Investigate some different loading cases of the
x
structure (structure and loadings are symb
metric):
a
P
(a) a general loading q(x) (N/m) along part of
the beams, i.e., the loading is q(x) for 0 < x <
q ( x)
a (note that x = 0 at the free surface and x = a
Case (2)
at the crack tip),
x
(b) the loading q(x) is constant, q(x) = q0, along
part of the beams, i.e., the loading is q0 for 0 < x
q ( x)
< a,
a
(c) what happens in Case (b) above if the factor
becomes small ( << 1) but the force q0a at
a
the beam end is a constant, say q0 a = Q?
Assume plane stress conditions, and compare with Problem 5/2.
Solution:
Use equation (5.10) in the textbook to determine the stress intensity factor.
One has
ui(1)
4
(2)
(2)

KI =
Ti
d
(5.10)
a
( + 1) KI(1)
The stress intensity factor in Problem 5/2 is known. Thus, the factor KI(1) to use
in the right hand side of (5.10) is (from Problem 5/2)
P a 2
3
(a)
bh
h
From loading case (1) (Problem 5/2) one needs to know also the (generalized)
displacement at the points where the load is going to be applied in the second
loading case. This means that one needs to know the deflection in Case (1)
(due to load P) of the cantilever beam where the load q(x) is applied in Case
(2). One obtains, with ui(1)(x) = v(x),
Pa 3
3x x 3
v(x) =
2
(b)
+
6EI
a a3
KI(1) =

Also, v/a is needed. One obtains

v(x) P
=
(2 3a 2 3x 2a + 0 )
a
6EI
Chapter 5

(c)
5:9

Equation (5.10) now gives

a
4
q(x) P (6a 2 6xa) dx
KI(2) =
2
b 6EI
( + 1) KI(1) 0

(d)

where = E / 2(1 + ) is the shear modulus, = (3 ) / (1 + ) (plane stress),

and the factor 2 is there due to the two cantilever beams (integration along two
beams). In (5.10) T(2) is stress. Dividing q(x) (N/m) by the thickness b gives the
stress on the boundary.
(b) Enter q(x) = q0 into (d). It gives
a q a
4
P
0
(2)

KI =
2
(a x) dx
(1)
0 b EI
( + 1) KI
=

q0 a P
4
1
2

a
a

(a)

2
( + 1) KI(1) b EI

Thus,
K

(2)
I

(e)

P a 2 q0 a
8

=
1
(1)
b
2
( + 1) KI EI

(f)

(c) Enter q0a = Q into (f), and then let go to zero. It gives
KI(2) =

P a2 Q
8
8 b h
h P a 2 Q
=
( + 1) KI(1) EI b ( + 1) P a 2
3 EI b

8E (1 + ) b h
h a 12 Q = Qa 2
3
2(1 + ) 4 2
3 Ebh 3 b bh
h
as obtained in Case (1) (at plane stress), see expression (a) above.
=

Answer: (a)

KI(2) =

a
4
q(x) Pa (a x) dx
2
b EI
( + 1) KI(1) 0

(b) Entering q(x) = q0 into (d) gives

KI(2) =

P a 2 q0 a
8

b
2
( + 1) KI(1) EI

(c) Enter Q = q0a and = 0 into (f). It gives

as in Case (1) (at plane stress), cf. expression (a).

5:10

Chapter 5

KI(2) =

Qa 2
3
bh
h

(g)

(d)
(f)

Superposition
5/8.
Demonstrate (no calculations needed) why the two loading cases in the figures
have the same stress intensity factor KI, cf. (j) and (k) in Section 5.2.3 in the
textbook. The plates are large. In Case (2) the stress is a compressive stress
acting on the two crack surfaces (i.e., is a pressure loading the crack
surfaces).

Case (2)

Case (1)

2a

2a

Figure Loading case (1) with remote stress and loading case (2) with same
stress acting on the crack surfaces
Solution:
In Case (2), replace the compressive stress on the crack surfaces with a
tension stress . Call this loading case for Case (2b). Superimpose loading case
(2b) on Case (1). It gives the same stress distribution in the plate (in the plane
of the crack) as in a plate with no crack at all, see figure below. This implies
that the stress intensity factors in the two cases must cancel each other, and
therefore they must be the same.

Chapter 5

5:11

Case (2b)

Case (1)

No crack

2a

2a

2a

Figure Loading case (1) with remote stress and loading case (2b) with
tension stress acting on the crack surfaces
Or, doing it in anther way: Start with a plate without crack, see the leftmost
part of the figure. The stress in the left plate is uniaxial (value ) and evenly
distributed all over the plate. Create loading case (1) by cutting a crack in the
first plate. It means that in the second plate the stress cannot be transferred
from the upper crack surface to the lower crack surface (as was done in the
first plate). This gives the loading case (1) with a certain stress intensity factor
KI(1).
Next, the stress that could not be transferred over the crack in loading case
(1) is applied on the crack surfaces of another plate. This gives loading case
(2b), see the figure. The loading case (2b) gives a stress intensity factor KI(2b).
Superimposing Cases (1) and (2b) gives the first plate with no crack. But the
first plate has no stress singularity, so one must have
KI(1) + KI(2b) = 0

giving KI(1) = KI(2b)

(a)

Finally, changing the sign of the stress in loading case (2b) gives back the
original loading case (2), and one has
KI(1) = KI(2)

(b)

Answer: Replace the compressive stress in Case (2) by a tensile stress (of
magnitude ) on the crack surfaces. This gives the Case (2b). Superimpose
Case (1) and the new Case (2b). It gives a plate with a homogeneous stress
field throughout the plate, without stress concentration. Thus, the two stress
intensity factors cancel each other, and it follows that Case (1) and Case (2)
must have the same stress intensity factor.

5:12

Chapter 5

Solutions to problems in
T Dahlberg and A Ekberg: Failure, Fracture, Fatigue - An Introduction.
Studentlitteratur, Lund 2002, ISBN 91-44-02096-1.

Chapter 6
Crack propagation under cyclic loading
Problems with solutions
Crack propagation at cyclic loading, Paris law
6/1.
Crack-like flaws may develop in most materials. These flaws should be treated
as if they were cracks. Therefore, assume that the flaws are penny-shaped (the
crack looks like a coin with a sharp egde). Determine the largest flaw
allowable with respect to fatigue crack propagation in the following materials.
The loading is cyclic and the stress range is 0.5Y (and Y is the yield limit of
the material).
(a) A low-strength steel with Kth = 7 MN/m3/2 and Y = 300 MPa, and
(b) a high-strength steel with Kth = 4 MN/m3/2 and Y = 1500 MPa.
Solution:
A penny-shaped crack in the interior of the material will give the stress
intensity factor KI = 0
a f8(a). Case 8 in Appendix 3 of the textbook gives
(terms containing a / t can be neglected here so that f8 = 0.637 is obtained)
KI = 0.637 0
a

(a)

(a) No crack propagation is allowed. Then the stress intensity range KI must
be less than (or equal to) the threshold value Kth. Using Kth = 7 MN/m3/2 and
0 = Y / 2 = 150 MPa, one obtains
3/2
KI = 0.637 150
a = 7 MN/m

(b)

giving a = 1.7 mm.

(b) Similarly, using Kth = 4 MN/m3/2 and 0 = Y / 2 = 750 MPa, one obtains
KI = 0.637 750
a = 4 MN/m3 / 2

(c)

giving a = 0.0223 mm.

FMCHAP6P.DOC/2010-01-15/TD

Chapter 6

Page 6:1

It is concluded that the high-strength steel is much more sensitive to cracks

than the low-strength steel. (One reason is that the high-strength steel is loaded
at higher stress.)
Answer: Critical crack length is (a) a = 1.7 mm, and (b) a = 0.022 mm.
6/2.
P

A double cantilever beam (DCB) specimen is

subjected to a pulsating loading. The crack
length is a, thickness is b and specimen height
is 2h, where a >> b and a >> h. The material is
2 h linear elastic with Youngs modulus E. Use
Paris law
da
b
= C (KI )n
dN
to determine the number of loading cycles
required to make the crack grow from its initial
length ai to a final length a, if

(a) the load is prescribed; the load varies between 0 and Pmax,
(b) the displacement is prescribed; the displacement varies between 0 and vmax
(in the figure one has = 2vmax).
Solution:
(a) The stress intensity range becomes (p = Pmax /b (N/m) is the load range)
KI =

2
3 p

a=

h
h
Entering (a) into Paris law gives

Separating (b) gives

2
3 Pmax
b h
h

da
= C (KI)n = C
dN
da
=C
an

(a)

a
n
2
3 Pmax

a
h
b h

n
2
3 Pmax

dN
b h
h

(b)
(c)

Integrating (c) and solving for N give

b h
1
h 1 n

(a
ai1 n )
N=
C (1 n) 2
3 Pmax
n

Page 6:2

Chapter 6

(n 1)

(d)

(b) The stress intensity range becomes (here vmax = /2 is the range of the
displacement)
3 E h 2 vmax 1

KI =
(e)
a2
2
h
n
Entering (e) into Paris law gives
2

3
E
h
v
da

1
max
= C (KI)n = C
(f)
2
dN
a
2
h

Separating (f) gives

a
N
3 E h 2 vmax n
a 2n da = C
(g)
dN
ai
0

h
Integrating (g) and solving for N give

2
1
h 2n + 1

(a
ai2n + 1)
N=
2
C (2n + 1)
3 E h vmax
n

Answer: Number of cycles required is, respectively,

n
(a)
b h
1
h 1 n 1 n

(a
N=
ai )
C (1 n) 2
3 Pmax
(b)

(h)

(n 1)

2
1
h 2n + 1 2n + 1

(a
ai )
N=
C (2n + 1)
3 E h 2 vmax
n

6/3.
A structure is subjected to a pulsating loading, i.e.. min = 0. Embedded circular
(penny-shaped) cracks are discovered in the structure. The crack diameters are
2a0, which is much smaller than the thickness of the material. The structure is
made of a material that follows the crack propagation law
da
= C (KI )n
dN
(a) Determine the largest (remote) stress range 0 the structure may be
subjected to if crack propagation is not allowed.
(b) Determine the maximum stress range that may be allowed if the structure is
loaded by N = 10 000 cycles. During the loading the crack is not allowed to
grow so large that it becomes unstable. Therefore, choose safety factor s = 1.5.
Numerical data: crack length a0 = 0.0005 m, Y = 620 MPa, Kth = 2 MN/m3/2,
KIc = 36 MN/m3/2, n = 4.0 and C = 1.1210 11 m7/MN4.

Chapter 6

Page 6:3

Solution:
(a) No crack propagation
If no crack propagation is allowed, the stress intensity range KI must be less
than or equal to the threshold value Kth. Assume that there exists a crack with
the most unfavourable direction. The stress intensity factor KI then is (Case 8
in Appendix 3 in the textbook, with a << t)
a
KI = 0
(a)
a f8 c , 0
where a = c, which gives f8(1,0) = 0.637. Thus
KI = 0.637 0
a

(b)

The stress intensity range KI becomes (0 is stress range)

KI = 0.637 0
a

(c)

No crack propagation is expected if KI Kth, which gives

0.637 0
a Kth

Solving for 0 gives

Kth
2 106
0
=
= 79 106 N/m2 = 79 MPa
0.637
0.0005
a 0.637

Thus, if 0 79 MPa no crack growth is expected.

(d)

(e)

(b) Crack propagation

After N = 104 loading cycles the crack should give a stress intensity factor
KImax = KIc / s. Determine the critical crack length ac giving this stress intensity
factor.
Assume that the crack stays circular (penny-shaped) during the propagation.
One then has
K = 0.637
(f)
a
I

Fracture will occur (if LEFM is valid) when

KIc
KImax =
s
Equations (f) and (g) give

(g)

KIc
KIc 2 1
36
2 452
1
0.6370
ac = s giving ac = 0.637 s = 0.637 0 1.5 = 2
0
0

(h)

where, here, 0 (in MPa) equals the stress range 0 (in MPa) because min = 0
(load is pulsating).
Page 6:4

Chapter 6

Now the initial crack length a0 and the final crack length ac (expressed in the
unknown stress range 0) are known. During N cycles the crack will grow
from a0 to ac. The crack propagation law (Paris law) gives
da
n
a)
(i)
= C (KI)n = C(0.637 0

dN
ac
which gives
N
da
n

=
C
(0.637

)
dN
(j)

0
n
a0 (
a ) 0
Evaluation of the integrals gives
ac1 n/2 a01 n/2
n
) N (k)
= C (0.637 0

1 n/2
Entering the values of a0, ac, C, n and N into (k) gives
1

(0.5 10 3 )
4
) 104
= 1.12 10 11 (0.637 0

1
Solving for 0 gives 0 = 324 MPa = 0.
(452/(0)2)

Finally, LEFM was used when ac was determined. Is LEFM valid?

The stress 0 = 324 MPa (= max) gives the critical crack length
452
= 0.0043 m
ac =
3242
to be compared with

(l)

(m)

KIc 2
36 2
(n)
= 0.0037 m
2.5
= 2.5
1.5 620
s Y
Thus, ac is larger than 0.0037 m so LEFM may be used (t and W a are
assumed large enough).

Note that only the crack length ac at fracture is of interest here. The fracture
criterion KImax = KIc / s was used to determine ac. Only then LEFM must be
valid, not during the crack propagation phase.
Answer: The stress range should be less than (a) 0 = 79 MPa, (b) 0 = 324
MPa.

Chapter 6

Page 6:5

6/4.

2a
t

2c

A large plate of thickness t contains an

embedded elliptical crack, see figure. Determine
the cyclic life (the number of cycles to failure)
of the plate if
(a) the load varies between 0 and ,
(b) the load varies between 0.5 and .
The influence of the mean value of the stress
intensity factor is disregarded. The crack is
assumed to keep its form during the crack
propagation (i.e., ratio a / c is constant during
the crack propagation).

The material follows Paris law for crack propagation. Use safety factor s =
1.4.
Numerical data: crack length a0 = 0.001 m and c0 = 0.002 m, thickness t = 0.10
m, yield limit Y = 1200 MPa, threshold value Kth < 6 MN/m3/2, fracture
toughness KIc = 60 MN/m3/2, and crack propagation parameters n = 4.0 and C =
510 13 m7/MN4, = 400 MPa.
Solution:
For an embedded elliptical crack the stress intensity factor is (Case 8 in
Appendix 3; a << t implies that terms containing a / t may be neglected)
a
1
KI =
(a)
a f8 c , 0 =
a f8 2, 0 = 0.826
a
The stress intensity range KI becomes
KI = 0.826
a

(b)

The initial crack length a0 = 0.001 m gives the stress intensity range
Case (a):
0.001
KI = 0.826
a = 0.826 400

= 18.5 MN/m 3 / 2 > Kth

(c)

Case (b):
KI = 0.826
0.001
a = 0.826 200

= 9.26 MN/m 3 / 2 > Kth

Thus, the crack will grow in both cases.

Page 6:6

Chapter 6

(d)

Next, determine the critical crack length ac. Failure will occur (if linear elastic
fracture mechanics theory, LEFM, may by used) when
KIc
60
KImax =
ac =
giving 0.826 400
(e)

s
1.4
which gives ac = 0.0054 m at failure (or, to be exact, the crack will start to
grow rapidly at this crack length and the failure will come after a few more
cycles).
Check the conditions for LEFM. One has
KIc 2

60 2
(f)
= 0.0032 m
= 2.5
2.5
1.4 1200
s Y
Thus, ac > 0.0032 m and LEFM can be used (also t and W a are large
enough).

Paris law now gives

da
n
a )
= C (KI)n = C (0.826 0

dN

(g)

Separating and integrating give

a

N
cr
1
n

C (0.826
da
=

)
dN

0
n
a0 (

0
a )

N=

Thus

acr1 n / 2 a01 n / 2
(1 n / 2) C (0.826 0
)

(h)
(i)

Case (a) gives:

N =
(a)

0.0054 1 0.001 1
( 1) 5 10

13

= 13 855 cycles

(j)

= 24 N (a) = 221 680 cycles

(k)

(0.826 400
)

Case (b) gives:

N

(b)

0.0054 1 0.001 1

( 1) 5 10 13 (0.826 200
)
Thus N(b) = 24 N(a) = 221 600 cycles.

Answer: Cyclic life to failure is, approximately, (a) N = 13 850 cycles, and (b)
N = 221 600 cycles (cf. Problem 6/9).

Chapter 6

Page 6:7

6/5.

A large plate of thickness t contains an elliptical

surface crack, see figure. The load is cyclic and
it varies between 0 and .
(a) Determine the relationship between the
remote stress and the number of loading
cycles N for crack propagation to depths a = 5,
10, 15 and 20 mm.
(b) Determine also the stress level that may give
unstable crack growth at the different crack
depths. The crack is assumed to keep its form
during the crack propagation. The material
follows Paris crack propagation law. Use safety
factor s = 1.5.

2c

Numerical data: crack length a0 = 0.002 m, c0 = 0.004 m, t = 0.10 m, Y =

1200 MPa, Kth = 6 MN/m3/2, KIc = 70 MN/m3/2, n = 3.75, C = 9.2210 12
m6.625/MN3.75.
Solution:
Here the stress intensity range becomes (Case 7 in Appendix 3 in the textbook,
terms containing a / t may be neglected here)
1
KI =
(a)
a f7 2, 0 = 0.896
a
Paris law then becomes
da
3.75
a )
(b)
= C (KI)n = C (0.896

dN
Separating and integrating give
a

N
final
da
3.75

=
C
(0.896

)
dN

3.75
ainitial (
0
a )

1
1 1
3.75

0.875
0.875 = ()
3.75
0.875
a
a

final
initial
C (0.896

Thus

(c)
(d)

or, rewritten,
1
1 1

0.875
0.875
3.75
0.875
a
a

final
initial
C (0.896
)

3.75 log + logN = log

which gives, for a = afinal = 5, 10, 15, and 20 mm, respectively,

3.75 log + logN = 12.44, 12.57, 12.62, and 12.64

Page 6:8

Chapter 6

(e)

(f)

(b) The critical stress at the different crack lengths is, from (a), (and if LEFM
can be used)
1
KImax = ( KI = ) 0.896
(g)
a = 1.5 KIc = 46.7 MN/m3 / 2
Check for LEFM. One has
KIc 2
2.5
(h)
= 0.00378 m
s Y
implying that one may use LEFM.
With (g), the four crack lengths 5, 10, 15, and 20 mm give, respectively,
= 416, 294, 240, and 208 MPa
(i)
Answer: (a) 3.75log + log N = 12.44, 12.57, 12.62, and 12.64 for a = 5, 10,
15, and 20 mm, respectively, and
(b) = 416, 294, 240, and 208 MPa, respectively.

6/6.
l

10 l
a

A beam with an edge crack is loaded with two

forces P = P0 sint, see figure. Determine a
t
lower limit of the cyclic life of the beam
h (assume that no other cracks are present). The
material follows Paris crack propagation law.
Use safety factor s = 1.5.

Numerical data: crack depth a0 = 0.005 m, thickness t = 0.01 m, height h =

0.10 m, length l = 0.10 m, yield limit Y = 1400 MPa, threshold value Kth = 4
MN/m3/2, fracture toughness KIc = 70 MN/m3/2, and crack propagation
parameters n = 4 and C = 110 13 m7/MN4. Load amplitude is P0 = 0.042 MN.
Solution:
Between the two forces P, the bending moment in the beam is constant and it
is Pl. The maximum value of the bending moment is P0 l and the minimum
value contributing to the stress range is zero (it is assumed that the negative
part of the moment will not contribute to the crack growth). Thus, the range of
the bending moment contributing to the stress range is P0 l.
The stress intensity factor becomes (Case 6 in Appendix 3 in the textbook)
a
6Pl

KI =

a
f
(a)

6
W
t W2
For crack length a = a0 = 0.005 m one has a / W = a0 / h = 1/20 = 0.05, which
gives f6(0.05) = 1.08.
Chapter 6

Page 6:9

Determine the crack length ac at failure. One obtains

6 P0 l
ac KIc
=
KI =

a
f
c

6
W s
t W2

(b)

Here f6(ac / W) is unknown so one has to try (to guess) a value of f6. Try
f6(ac / W) = 1.06 to see what happens (we see from the diagram that f6 decreases
in the region where a / W = 0.05). It gives, for stress and fracture toughness
expressed in MPa,
6 0.042 0.1
70

a
1.06
=
(c)
KI =
c

1.5
0.01 0.102
Solving for ac gives ac = 0.0097 m.
This value of ac gives f6(ac / W) = f6(0.097) = 1.06, which is the same as the
value assumed. Therefore, the critical crack length ac = 0.0097 is selected for
the following calculations.
Paris law now gives

4
6 P0 l
da
a
n
(d)
= C (KI) = C
a f6 W

dN
t W2

Again, we have the problem of not knowing the value of the function f6 = f6(a)
when the crack grows, i.e. when a changes. From the above calculations (and
from the form of the function f6 seen in the diagram) it is concluded that f6
varies between 1.08 and 1.06 in the crack length interval obtained here.
Therefore, to be on the safe side, select f6 = 1.08 for the whole interval (this is
a conservative estimation, as it will give a too large crack propagation rate).
Now equation (d) will give (with stresses in MPa, see the dimension of the
constant C)
6 P0 l
4
da
=C
(e)
1.08 dN

a2
t W2

or

0.0097
6 P0 l
4
da

=N C
(f)
1.08

0.005 a 2
t W2

from which N = 17 900 is solved. As we have used a too large value of f6 this
will be a lower limit of the cyclic life of the structure. The cyclic life is
expected to be larger than the life calculated. Thus, it is expected that N will be
slightly more than 17 900 cycles. (An upper limit of the fatigue life is
estimated if f6 = 1.06 is used. Then one obtains, see equation (f), Nupper =
17 900 (1.08/1.06)4 = 19 290 cycles.)

Answer: The fatigue life is expected to be slightly more than N = 17 900

cycles (but less than 19290 cycles). Critical crack length is acrit = 0.0097 m.
Page 6:10

Chapter 6

6/7.

a 2r a

In a large plate of a ships hull two cracks of

approximately the same size were discovered at
a hole, see figure. The plate is loaded in cyclic
tension with min = 0.2 and max = = 60
MPa. The loading frequency is 6 cycles per
minute. Determine the remaining life (in hours)
of the plate. It is assumed that the crack
propagation is symmetric with respect to the
hole.

The material follows Paris law for crack propagation. Use safety factor s =
1.5. The geometry function f used to determine the stress intensity factor may
in this example, and for the crack length used here, be approximated to
a
a 1/6
f = 1.42
r
r
(Thus, this is an approximation of function f3 in Appendix 3 of the textbook.)

Numerical data: r = a0 = t = 0.08 m, Y = 600 MPa, KIc = 90 MN/m3/2, n = 3,

C = 110 11 m11/2/MN3.
Solution:
In this example, the stress intensity factor may be written, approximately,
a
1.42
KI = 0
= 1.42
(a)
a f r = 0
a
0 a 1 / 3r 1 / 6
1/6
(a/r)
(Note that the function f given here is an approximation of the function f3 given
in Appendix 3 in the textbook.)
Determine the critical crack length acr. Equation (a) gives, by use of KImax =
KIc / s,
KIc
3
3
90
1
= 0.22175 m
a = acr =
=
60 0.081 / 6
s 1.42
r 1 / 6 1.5 1.42

(b)

May linear elastic fracture mechanics (LEFM) be used here? One has
KIc 2
(c)
2.5
= 0.025 m
s Y
Yes, all three measures a, t, and W a are larger than 0.025 m, so LEFM may
be used. The critical crack length is acr as calculated in (b).

Chapter 6

Page 6:11

The stress intensity range becomes = max min = 0.8, which gives
KI = 1.42
0.8 a 1 / 3 r 1 / 6

(d)

da
3
0.8 a 1 / 3 r 1 / 6 )
= C (KI)n = C (1.42

dN

(e)

Paris law gives

Thus

giving

N=

N
cr
3
da = C (1.42
0.8 r 1 / 6 ) dN

a0 a 0

(f)

acr
ln
= 204 400 cycles
3
C (1.42
0.8 r 1 / 6 ) a0

(g)

Remaining fatigue life is t = N / (6 60) = 568 h.

Answer: Fatigue life N = 204 000 cycles gives that the remaining life is
expected to be 568 hours.
6/8.

2
1

A large plate of thickness t contains a halfelliptical surface crack, see figure. The load
t consists of a repeated sequence of two cycles as
shown in the figure. Determine the expected
2c
remaining life of the plate expressed in number
of sequences to failure. It is assumed that the
crack keeps its form during the crack
propagation. The material follows Paris law.
time Use safety factor s = 1.4.

Numerical data: a0 = 0.002 m, c0 = 0.004 m, t = 0.100 m, Y = 1200 MPa, KIc

= 70 MN/m3/2, Kth = 6 MN/m3/2, n = 4, C = 110 13 m7/MN4, 1 = 200 MPa
and 2 = 400 MPa .
Solution:
A half-elliptical surface crack (Case 7 in Appendix 3 in the textbook) gives the
stress intensity factor, for a / c = 0.5,
a
KI = 0
(a)
a f7 c , 0 = 0
a 0.896 = 0.90 0
a
The range of the stress intensity factor becomes
KI = 0.90 0
a

Page 6:12

Chapter 6

(b)

Will the lower stress level 1 = 200 MPa contribute to the crack growth?
Let a = a0 = 0.002 m and 0 = 1 = 200 MPa. One then obtains from (b)
KI = 0.90 200
0.002 = 14.3 MN/m3 / 2

(c)

The threshold value is Kth = 6 MN/m3/2, giving KI > Kth.

This implies that the crack will propagate at the lower stress cycle 1, and
thereby also at the higher stress level 2.
Determine next the crack length acrit at fracture. Assume that linear elastic
fracture mechanics (LEFM) is valid. Fracture will then occur when KImax =
KIc / s. It gives
KIc
KImax = 0.90 0
acrit =
(d)

s
Fracture will occur when the highest stress is loading the structure, i.e. when
the stress is 2. For 0 = 2 one obtains from (d) (if linear elastic fracture
mechanics is valid)
KIc
2 1
1
70
2
acrit =
= 0.0061 m
=
s 0.90 2 1.4 0.90 400

(e)

Is linear elastic fracture mechanics valid?

KIc 2

70 2
2.5
(f)
= 0.0043 m
= 2.5
1.4 1200
s Y
which is less than acrit, t, and W a. Thus, linear elastic fracture mechanics can
be used.

It is now known that the crack will grow from a0 to acrit, and both cycles in the
sequence will contribute to the crack propagation. Per sequence of two cycles,
the crack propagation will be
da
n
= C (KIi )n = C (0.90 0i
a )

dNs

= C (0.90 1
a ) + (0.90 2
a )

(g)

Note that Ns stands for the number of sequences (not number of cycles).
Separation of variables gives
N

crit
s
da

n
n

= C (0.90 1
) + (0.90 2
) 0 dNs
n
a0 (
a )

Chapter 6

(h)

Page 6:13


n
n
1 n /2
acrit
a01 n /2 = 1 C (0.90
) { n1 + n2 } Ns

which gives

(i)

Entering acrit, a0, 1, 2, and the material parameters C and n in (i), and then
solving for Ns, give
(j)
Ns = 19 080 sequences
Answer: Expected fatigue life is Ns = 19 000 sequences, giving N = 38 000
cycles to failure.
6/9.

2a
t

2c

A large plate of thickness t contains an

embedded elliptical crack, see figure. Determine
the cyclic life (the number of cycles to failure)
of the plate if the load varies between 0.5 and
.
The influence of the mean value of the stress
intensity factor has to be taken into account.
The crack is assumed to keep its form during
the crack propagation (i.e., ratio a / c is constant
during the crack growth).
(This is Problem 6/4(b) once again, but now the
influence of the mean value is taken into
account.)

The material follows Paris law for crack propagation. Use safety factor s =
1.4.
Numerical data: crack length a0 = 0.001 m, c0 = 0.002 m, t = 0.10 m, Y =
1200 MPa, Kth < 6 MN/m3/2, KIc = 60 MN/m3/2, n = 4.0 and C1 = 510 13
m7/MN4, = 0.7, = 400 MPa (it is assumed that the material parameters C
and n in Paris law are the same here as in Problem 6/4).
Solution:
Following the solution to Problem 6/4, one obtains the stress intensity factor
for an embedded elliptical crack as
a a
1
KI =
(a)
a f8 c , t =
a f8 2, 0 = 0.826
a
The stress intensity range KI becomes
KI = 0.826
a
Page 6:14

Chapter 6

(b)

The initial crack length a0 = 0.001 m gives the stress intensity range
KI = 0.826
0.001
a = 0.826 200

= 9.26 MN/m 3 / 2 > Kth

(c)

Thus, the crack will grow at the stress range given.

Determine the critical crack length ac. Failure will occur (if linear elastic
fracture mechanics theory, LEFM, may by used) when
KIc
60
KImax =
ac =
(d)
giving 0.826 400
MN/m 3 / 2

s
1.4
which gives ac = 0.0054 m at failure (or, to be exact, the crack will start to
grow rapidly at this crack length, and the failure will come after a very limited
number of loading cycles after this crack length has been reached).
Check the conditions for LEFM. One has
KIc 2

60 2
(e)
= 0.0032 m
= 2.5
2.5
1.4 1200
s Y
Thus, ac > 0.0032 m and LEFM can be used (also t and W a are large
enough).

Paris law, with correction for the mean stress value according to Walker, now
gives
C1
da
n
(KI)n = C (0.826 0
a )
(f)
=

n(1 )
dN (1 R)
where stress ratio R is R = 0.5. The factor C becomes C = C1 / (1 R)n (1 ) =
1.148710 12 m7/MN4.
Separating and integrating give

N=

giving

Thus,

N =
(a)

N
cr
1
n

da
=
C
(0.826

)
dN (g)

0
n
a0 (
0
a )

acr1 n/2 a01 n/2

(h)

(1 n/2) C (0.826 0
)

0.0054 1 0.001 1
( 1) 1.1487 10 12 (0.826 200
)

= 96 500 cycles

(i)

The expected fatigue life now becomes N = 96 500 cycles, to be compared

with N = 221 700 cycles in Problem 6/4. These two numbers differ by the
factor (1 R)n (1 ) = 0.435, which means that in this case the mean value plays
an important role for the fatigue life of the structure.
Answer: Expected fatigue life is N = 96 500 cycles.
Chapter 6

Page 6:15

6/10.
0

thickness t
a

Determine the number of loading cycles

required to make the crack in the figure grow by
1 mm. The structure is loaded with stress cycles
of constant amplitude. The stress range 0 is
0 = 100 MPa. Use some different initial crack
W >> a lengths ai, say ai = 1, 5, 10, 20, 40, 60, and 80
mm.

For crack growth calculations use both the un-modified Paris law:
da
= C (KI)n
dN
and the modified version, taking the closeness to the threshold value into
account:
da
= C {(KI)n (Kth)n }
dN
Compare the results. Assume that the constants C and n are the same in the
two formulae: C = 2.510 12 (units in MN and meter) and n = 3.2. The material
has the fatigue threshold value Kth = 6 MN/m3/2.
Solution:
First investigate if crack propagation is obtained for the shortest crack length ai
= 1 mm. The intensity range for the component with the edge crack is (Case 5
in Appendix 3 in the textbook, use that W >> a)
a
KI = 0
(a)
a f5 W = 1.12 0
a
The stress range 0 = 100 MPa gives, for crack length ai = 1 mm,
a
KI = 0
0.001 = 6.28 MN/m / 2
(b)
a f5 W = 1.12 100

which is larger than the threshold value Kth = 6 MN/m3/2. It is concluded that
the crack will grow for all crack lengths ai.

Paris law gives

da
= C (KI)n
dN

where KI is obtained from (a).

Separating and solving for N give

Page 6:16

Chapter 6

(c)

N =N

(d)

a + 0.001

da

i
1
=
C ai

(1.12 100
) a
The modified crack propagation law gives
a + 0.001

1 i
N =N =
C ai
(e)

3.2

(d)

1.6

da

(e)

(1.12 100
) a 1.6 (Kth)3.2
3.2

Numerical solution (for example by use of MATLAB) of (d) and (e) gives
results according to the table below. It is noted that only when the stress
intensity range is very close to the threshold value there is a large difference
between the two results (about one million cycles). A difference was expected,
since the crack growth rate tends to zero at the threshold value, whereas the
unmodified Paris law predicts a finite crack growth rate also at the threshold
value. Already at the crack length 5 mm the difference in crack growth rate
between the two formulae is small; only a few per cent.
Making the crack grow from 1 mm to 81 mm would, according to the
unmodified Paris law, require 1.7310 6 cycles, whereas the modified version
predicts 2.8910 6 cycles. The main part of this difference (one million cycles)
depends on the slower crack growth the first millimetre.
Table: Number of loading cycles required to make the crack grow 1 mm for
different initial lengths ai of the crack. The fatigue life N(d) is calculated
with the un-modified Paris law whereas N(e) is calculated with the
modified crack propagation law given in the problem. It is seen that the
two formulae give different results only when the stress intensity range
KI(ai) is very close to the threshold value Kth (here Kth = 6 MN/m3/2).
ai (m)

0.001

0.020

0.040

0.060

0.080

N(d) (cycles) 0.6310 6 73 628 26 056 8 923

3 001

1 579

1 000

N(e) (cycles) 1.6810 6 78 105 26 592 8 985

3 008

1 581

1 000

KI(ai)
(MN/m3/2)

39.7

48.6

56.1

6.27

0.005

14.0

0.010

19.9

28.1

In the table is also given the stress intensity range KI(ai) for the initial length
ai of the crack. It is noted that the fracture toughness KIc of the material must
Chapter 6

Page 6:17

be so large that after reduction with the safety factor s the fracture toughness
should be larger than KI(ai + 0.001) = KI(0.081) = 56.5 MN/m3/2. Thus, KIc
must be such that KIc > s56.5 MN/m3/2.
Answer: The un-modified propagation law gives
N = 0.6310 6, 73 600, 26 060, 8 900, 3 000, 1 580, 1000 cycles,
and the modified formula gives
N = 1.6810 6, 78 100, 26 600, 9 000, 3 000, 1 580, 1000 cycles, respectively.

6/11.

2
1

A large plate of thickness t contains a halfelliptical surface crack, see figure. The load
consists of a repeated sequence of two cycles as
t shown in the figure. The crack is assumed to
keep its form during the crack propagation. The
2c
material follows Paris law. The crack is
detected when the crack length a0 is a0 = 0.002
m. It is decided that the crack can be allowed to
grow to the length afinal = 0.006 m before the
time crack has to be repaired.

(a) Determine the expected number of cycles (twice the number of sequences)
required to make the crack grow from a0 to afinal. Use Paris law to determine
the crack propagation cycle by cycle.
(b) What number of cycles would be obtained if the root mean square value
(i.e. Krms) were used as an equivalent measure of the stress intensity range?
Numerical data: crack depth a0 = 0.002 m, c0 = 0.004 m, t = 0.100 m, Y =
1200 MPa, KIc = 70 MN/m3/2, Kth = 6 MN/m3/2, n = 4, C = 110 13 m7/MN4,
1 = 200 MPa and 2 = 400 MPa.
Solution:
A half-elliptical surface crack (Case 7 in Appendix 3 in the textbook) gives the
stress intensity factor (for a / c = 0.5)
a
KI = 0
(a)
a f7 c = 0.896 0
a
The range of the stress intensity factor becomes
KI = 0.896 0
a
Page 6:18

Chapter 6

(b)

Will the lower stress level 1 = 200 MPa contribute to the crack growth?
Let a = a0 = 0.002 m and 0 = 1 = 200 MPa. One obtains
KI = 0.896 200
0.002 = 14.2 MN/m 3 / 2

(c)

The threshold value is Kth = 6 MN/m3/2, giving KI > Kth.

This implies that the crack will propagate for the lower stress cycle 1, and
thereby also for the higher stress cycle 2.
Check the stress intensity factor for the higher stress 2 and crack length afinal =
6 mm. One obtains
KI = 0.896 400
0.006 = 49.2 MN/m 3 / 2

(d)

Thus, at the planned repair of the crack, the stress intensity factor is KI = 49.2
MN/m3/2, which is 70 per cent of the critical value KIc = 70 MN/m3/2 (the
safety factor is s = 70 / 49.2 = 1.42).
It is given in the problem that the crack will grow from a0 to afinal. Both cycles
in the sequence will contribute to the crack growth. Then, per loading
sequence the crack growth will be
da
n
= C (KI i )n = C (0.896 0 i
a )

dNs

= C (0.896 1
a ) + (0.896 2
a )

(e)

Note that Ns here stands for the number of sequences (not number of cycles).
Separation of variables gives
afinal

a 0

da

Ns

= C (0.896 1
dNs
) + (0.896 2
)
n

0
(
a
)

(f)

which gives

n
n
1n /2
a01 n / 2 = 1 C (0.896
) { n1 + n2 } Ns
afinal

(g)

Entering afinal, a0, 1, 2, and the material parameters C and n in (g), and then
solving for Ns, give
Ns = 19 265 sequences, giving N = 38 500 cycles
(h)

Chapter 6

Page 6:19

(b) Using the rms value as an equivalent stress intensity range, one obtains
Krms =

{ (Ki )2 ni }
= 0.896
a
ni

2002 + 4002
2

= 0.896
a 316.23 MN/m 3 / 2
Thus, the equivalent stress range becomes eq(rms) = 316.23 MPa.

(i)

Paris law then gives

n
n
n
1 n /2
afinal
a01 n /2 = 1 C (0.92
) ((rms)
(j)
) N

eq

2
giving N = 52 400 cycles.
Thus, the rms value here predicts 52 400 / 38 500 = 1.36 times too many
cycles. During the 52 400 38 500 13 900 extra cycles, the crack will grow
(far) beyond the 6 mm limit where the crack should have been repaired.
Conclusion: The conclusion is that the rms value can not be used for crack
propagation calculations.

(c) Extra problem:

Use N = 52 400 cycles (26 200 sequences) to see what crack length would
have been obtained if Paris law were used as in problem (a). One obtains

n
n
1 n /2
a01 n /2 = 1 C (0.896
) { n1 + n2 } 26 200
(k)
afinal

2
giving afinal = 0.0214 m = 21.4 mm.
This crack length gives the stress intensity factor, at stress 2,
0.0214 = 92.9 MN/m 3 / 2
KI = 0.896 2
a = 0.896 400

(l)

which is much larger than the fracture toughness KIc of the material. The
structure would fail long before the 52 400 cycles had been applied.
Comment:
The importance (influence) of the high amplitude loading cycles (here 2) is
diminished when the rms value is used. Instead, use
{ (KIi )n ni } 1 / n

KI equivalent =
ni

(m)

where n is the factor in Paris law (and ni is number of cycles at each stress
level). One then obtains the equivalent stress intensity range

Page 6:20

Chapter 6

K I equivalent

{ (KIi )n ni } 1 / n

=
= 0.896
a
ni

2004 + 4004 1 / 4

3/2
= 0.896
a 341.5 MN/m

(n)

This value of the stress intensity range gives

n
n
1 n /2
a01 n /2 = 1 C (0.896
) { 341.54 } 38 500
afinal

2
which gives afinal = 0.006 m, as it should.

(o)

Answer: (a) 38 500 cycles will propagate the crack to length 6 mm, (b) using
the rms value of the stress intensity range, 52 400 cycles are obtained, which is
36 per cent too many cycles. That many cycles would propagate the crack to a
length that would be longer than the critical value of the crack length. Thus,
failure would occur during the loading if the rms value were used in the
calculations.

6/12.
The irregular loading sequence of a machine component has been analysed and
reduced to the 60 stress range cycles given in the table below. Upon inspection
of the component, edge cracks of depth a = 1 mm (or longer) will be detected
and eliminated. The maximum permissible crack length in the component is a
= 5 mm, see figure.
Stress range 0 (MPa) 285

270

200

92

Number of cycles N

12

44

thickness t
a

Determine the maximum number of loading

sequences the component may be loaded with
between two inspections. Assume that cracks of
length 1 mm may remain in the structure after
an inspection.
W >> a The material has the fatigue threshold value
Kth = 8 MN/m3/2.

Chapter 6

Page 6:21

For crack growth calculations Paris law may be used:

da
= C (KI)n
dN
where C = 2.510 12 (units in MN and meter) and n = 3.2.
Solution:
First investigate if all stress cycles will contribute to the crack propagation.
The stress intensity range for the component with the edge crack is (Case 5 in
Appendix 3 in the textbook, use that W >> a)
a
KI = 0
(a)
a f5 W = 1.12 0
a
The stress range 0 = 92 MPa gives
a
0.001 = 5.77 MN/m / 2
(b)
KI = 0
a f5 W = 1.12 92

which is below the threshold value Kth = 8 MN/m3/2. It is concluded that at

least initially the lower stress range cycles 0 = 92 MPa do not contribute to
the crack propagation.
Check also for the next stress range cycle: 0 = 200 MPa. One obtains
a
0.001 = 12.6 MN/m / 2
KI = 0
a f5 W = 1.12 200

This cycle will induce crack propagation at all crack lengths.

(c)

Determine at which crack length the stress range cycles 0 = 92 MPa will
start to contribute to the crack propagation. One has
a

a1 = 8 MN/m / 2
(d)
KI = 0
a f5 W = 1.12 92

giving a1 = 1.92 mm.

Thus, as long as the crack is shorter than 1.92 mm, only the cycles with stress
range 200 MPa and larger will make that the crack grows. When the crack has
reached 1.92 mm all cycles (all four stress ranges) will be damaging to the
structure (i.e. give crack propagation).
The crack growth must be calculated i two steps: first the growth from 1 mm
to 1.92 mm, and then the growth from 1.92 mm to 5 mm.
Paris law gives

Page 6:22

da
= C (KI)n
dN

Chapter 6

(e)

Using notation Ns for number of loading sequences, one obtains

da
n
n
= C (KI i )n = C (1.12
) {1 285n + 3 270n + 12 200n } (
a)

dNs

(f)

Separating and integrating give

0.00192
da
n

= C (1.12
) {1 285n + 3 270n + 12 200n } Ns

1.6
0.001 a

(g)

Solving for Ns gives Ns = Ns1 = 2867 sequences, giving N = N1 = 286760 =

172 000 cycles (of these 172 000 cycles 286744 = 126 100 cycles do not
contribute to the crack growth).
When the crack has reached length 1.92 mm, all stress range cycles contribute
to the crack propagation. One obtains
da
= C (KI i )n
dNs
n

= C (1.12
) {1 285n + 3 270n + 12 200n + 44 92n } (
a )
Separating and integrating give
0.005
da
n

=
C
(1.12

)
{1 285n + 3 270n + 12 200n + 44 92n } Ns

1.6
0.00192 a

(h)
(i)

Solving for Ns now gives Ns = 2254 sequences (giving N = N2 = 225460 = 135

240 cycles).
In total, 2867 + 2254 = 5121 sequences, giving 307 000 cycles may be allowed
between two inspections.
Paris law has been used for crack length a = 5 mm. The largest stress range
cycle is 0 = 285 MPa. This gives the stress intensity range
KI = 1.12 0
0.005 = 40 MN/m3 / 2
a = 1.12 285

(j)

It is concluded that after reduction with a safety factor s (s > 1) the fracture
toughness KIc of the material must be at least 40 MN/m3/2. Thus, one must have
KIc > s40 MN/m3/2.
Alternative solution:
Solve the problem by use of the equivalent stress range eq.

Chapter 6

Page 6:23

For crack length less than 1.92 mm, the equivalent stress range becomes
equivalent

{ (0i )n ni } 1 / n 1 2853.2 + 3 2703.2 + 12 2003.2 1 / 3.2

=
=

ni
16

= 223.85 MPa

Paris law gives

(k)

da
n
a )
= C (KI)n = C (1.12 equivalent

dN

Separating and integrating give

0.00192
da
n

= C (1.12 223.85
) N

n
0.001 (
a )

(l)

(m)

Solving for N gives N = 45 867 cycles. Dividing N by 16 (the number of

damaging cycles in a sequence) gives the number of sequences Ns to make the
crack grow from length 1 mm to length 1.92 mm. One obtains Ns = Ns1 =
45 867 / 16 = 2867 sequences, which is in agreement with the result obtained
above.
For crack length larger than 1.92 mm, the equivalent stress range becomes
equivalent

{ (0i )n ni } 1 / n

=
ni

1 2853.2 + 3 2703.2 + 12 2003.2 + 44 923.2 1 / 3.2

=
= 155.13 MPa

60

Separating and integrating Paris law give

0.005
da
n

=
C
(1.12

155.13

)
N

n
0.00192 (
a )

(n)

(o)

Solving for N gives N = 135 247 cycles. Dividing N by 60 (the number of

damaging cycles in a sequence) gives the number of sequences Ns to make the
crack grow from length 1.92 mm to length 5 mm. One obtains Ns = Ns2 =
135 247 / 60 = 2254 sequences, which is in agreement with the result obtained
above.
Answer: In total, Ns = 2867 + 2254 = 5121 sequences may be allowed between
two inspections, giving 307 000 cycles. For crack length a < 1.92 mm the
stress range cycle 0 = 92 MPa will not contribute to the crack propagation.

Page 6:24

Chapter 6

Solutions to problems in
T Dahlberg and A Ekberg: Failure, Fracture, Fatigue - An Introduction.
Studentlitteratur, Lund 2002, ISBN 91-44-02096-1.

Chapter 7
An introduction to fatigue

No problems in this chapter

FMCHAP7P.DOC/2010-01-29

Chapter 7

Page 7:1

Empty page.

Page 7:2

Chapter 7

Solutions to problems in
T Dahlberg and A Ekberg: Failure, Fracture, Fatigue - An Introduction.
Studentlitteratur, Lund 2002, ISBN 91-44-02096-1.

Chapter 8
Stress-based fatigue design
Problems with solutions
8/1.
A generator is rotating with a speed of 3000 rpm (revolutions per minute).
Determine the number of loading cycles N the axle is subjected to due to its
own weight. The machine is operating 7000 hours per year during 20 years.
Solution:
3000 revolutions per minute give 3000 loading cycles per minute. It gives
3000 60 7000 loading cycles per year, giving 3000 60 7000 20 = 25.2 109
loading cycler per 20 years.
Answer: Number N of loading cycles is N = 25.2 109 cycles (thus, it may be
exposed to fatigue).

8/2.
Estimate the fatigue limit in tension/compression for alternating and pulsating
loading of a steel (approximately SS 2225) with the ultimate strength U = 810
MPa. Sketch the Haigh diagram.
At an investigation of a similar material the relationships
FL = 0.468 U 50 MPa and FLP 0.85 FL

were found. Use these relations and other thumb rules to estimate the fatigue
limits (alternating and pulsating) of the material.
Solution:
The relationship FL = 0.468U 50 MPa gives, for alternating loading:
FL 0.468810 50 MPa = 380 50 MPa,
i.e., the fatigue limit FL in tension/compression at alternating loading may,
according to this relationship, be expected to fall in the range 330 MPa to 430
MPa.
For pulsating loading the relationship between pulsating and alternating
FMCHAP8P.DOC/2010-01-29/TD

Chapter 8

Page 8:1

loading, FLP 0.85FL, gives FLP = 0.85FL = 0.85 (380 50) MPa = 320
42 MPa. Thus, the fatigue limit in tension at pulsating loading is expected to
fall in the range from 280 280 MPa (from 320 42 MPa) to 360 360 MPa
(from 320 + 42 MPa).
Also, compare with other materials.
Material data for SIS 2225-03, -04 and -05 gives, for rotating bending, that
FLB should be in the range 350 MPa to 460 MPa. The engineering rule FL
0.8FLB gives that FL should fall in the range 280 MPa to 370 MPa.
Another engineering rule says that FLR (= FLB) decreases from 0.45U for
mild steel to 0.38U for the harder materials. The most unfavourable case, FLR
= 0.38U gives FLR = FLB 308 MPa, while the case FLR = 0.45U gives
FLR = FLB 365 MPa. This gives, using FL 0.8FLB, that FL should be in
the range 250 MPa to 290 MPa.
Finally, select, hopefully on the safe side, for alternating loading FL = 280
MPa and for pulsating loading FLP = 0.85FL, which gives FLP = 240 240
MPa. (In practice, fatigue limits will probably be higher than the values
estimated here.)
Answer: The fatigue limit FL at alternating loading is estimated to 280 MPa,
approximately, and the fatigue limit FLP at pulsating loading is estimated to
240 240 MPa, approximately.

8/3.
Sketch the Whler (SN, Stress-Number) curve for a normalised 37Mn Si5 steel
bar with an ultimate strength of U = 810 MPa. The bar is subjected to a
rotating bending moment. At an investigation of a similar material the
following fatigue lives were found:
FLR 0.45U yields N 106 cycles,
FLR 0.90U yields N = 1000 cycles, and
FLR = U yields N =1 cycle.
Solution and answer: See figure.

FLR / U

1.0
0.9
0.45

0 1

Page 8:2

Chapter 8

log N

8/4.
(a) Draw the fatigue diagram according to Haigh for the normalised steel SS
1650-01 subjected to loading in pure tension/compression.
(b) Draw the fatigue diagram according to Goodman for the normalised steel
SS 1650-01 subjected to loading in pure tension/compression.
Solution:
a (MPa)

200
180
m

180
590

310

+ a (MPa)
-

360
310
200
180

180

310

(a) Material data gives

FL = 200 MPa, FLP = 180 180 MPa,
Y = 310 MPa and U = 590 MPa,
from which the Haigh diagram can be
constructed, see figure.

590 MPa

(b) The same data gives the Goodman

diagram according to the second figure. It is
noted that this diagram is more complicated
to draw than the Haigh diagram (one has
both mean value m and amplitude a on the
ordinate). It does not, however, contain more
information than the Haigh diagram. (From
m now on, only the Haigh diagram will be
590 MPa used.)
Answer: See figures.

- 200

- 310

8/5.

Constant-amplitude fatigue strengths for

materials subjected to different cyclic loading
conditions are often expressed in Haigh
a
B
diagrams. In the Haigh diagram given three
A
straight lines (A, B and C) are shown. Each line
C
corresponds to constant loading conditions. The
200
loading varies sinusoidally with mean value m,
amplitude a, maximum value max (= m + a),
100
m minimum value min (= m a), and stress ratio
0
0 100 200 300 400 MPa R = min /max. Determine which one of these
five loading variables is constant in the three
cases shown in the figure.
Chapter 8

Page 8:3

Solution:
For line A one notices that the sum m + a is a constant (= 300 MPa). Thus,
line A gives that stress max is constant.
For line B one notices that ratio a / m is constant (= 2). Thus, stress ratio R is
constant.
For line C one notices that mean stress m = 200 MPa + a, giving m a =
200 MPa = min. Thus, min is constant.
Answer: Line A: stress max is constant, line B: stress ratio R is constant, line
C: stress min is constant.
8/6.
Two axles with the same geometry (but of different size) are machined from
the normalised steel SS 1650-01 (U = 590 MPa). The axle diameters are 10
mm and 100 mm, respectively. One axle was machined from a raw material
with a diameter 15 mm and the other from a raw material with diameter 120
mm. The loading is rotating bending.
Estimate the ratio of the reduced fatigue limits of the two axles if they have the
same surface finish.
Solution:
In Case (a), axle of diameter 10 mm, and in Case (b), axle of diameter 100
mm, the following reduction factors are obtained:
due to the size (volume) of the raw material (15 mm and 120 mm,
respectively): (a) (a) = 1.0, and (b) (b) 0.8,
due to loaded volume: (a) (a) = 1.0, and (b) (b) 0.9,
due to surface finish: (a) (a) = , and (b) (b) = , where is unknown, but the
same in the two cases.
One obtains for Case (a) that (a)(a)(a) = 1.0, and for Case (b) that
(b)(b)(b) 0.72
The reduced fatigue limits red
FL give
Case (b)
red,
FL
Case (a)
red,
FL

0.72 FL
= 0.72
1.0 FL

Answer: The ratio of the two reduced fatigue limits is, approximately, 0.72.

Page 8:4

Chapter 8

8/7.
Assume that the fatigue limits of polished test specimens subjected to
alternating loading in tension/compression are given by (ultimate strength is Rm
= U)
FL =

Rm
2

FL = 500 +

when Rm is less than 1000 MPa, and

Rm 1000
5

when Rm is greater than 1000 MPa.

Estimate how the fatigue limits of different materials depend on the ultimate
strength. Use some different surface roughnesses, for example a polished
surface, a machined surface and a surface with a notch (a standard notch, see
Figure 8.7).
Solution:
Factor for reduction of fatigue limit due to surface finish, and formulae
above, give
Rm = U =

500

1000

1500 MPa

250

500

600 MPa

Polished specimen
( = 1.0)

FLpol =

0.68 600 =

Machined specimen
FLred = FLpol 0.87250 = 220 0.78500 = 390

0.4 600 =

Notch
FLred = FLpol

410 MPa

0.8250 = 200 0.59500 = 295

240 MPa

It is concluded that high-strength steels are very sensitive to surface

irregularities (surface finish) and notches.
Using the reduction factor Kr, where = 1/Kr, the KTH Handbook, Sundstrm
(1998), gives for some values of the surface roughness measure Ra

Chapter 8

Page 8:5

Rm = U =

500

1000

1500 MPa

250

500

600 MPa

Polished specimen
FLpol =
Ra = 10 m

0.60 600 =

FLred = FLpol 0.90250 = 225 0.65500 = 325

Ra = 100 m

360 MPa
0.25 600 =

FLred = FLpol 0.62250 = 155 0.33500 = 167

150 MPa

Also here it is noted that the high-strength steels are sensitive to surface
irregularities (surface finish) and notches.
Answer: Factor for reduction of fatigue limit due to surface finish, and the
two formulae given in the problem, give reduced fatigue limits as given in the
tables above. It is concluded that high-strength steels are very sensitive to
surface roughness and to notches.
8/8.
The mean value of the fatigue limit of a welded 4 mm thick aluminium plate
has been found to be 92 MPa and the standard deviation of the spread of the
fatigue limit measurements is 6 MPa. Determine the stress level at which the
probability of fatigue failure is 0.1 per cent. The spread of the fatigue limit is
assumed to have a normal (Gaussian) distribution. For the normal distribution
function (x) one has
(x) =
x =

0.50
0

0.90
1.28

0.99
2.33

0.999
3.09

0.9999
3.72

Solution:
The mean value of the fatigue limit (here 92 MPa) gives the failure probability
50 per cent.
The standard deviation s is s = 6 MPa.
The failure probability 0.1 per cent is obtained at the stress level
= mean 3.09s = 92 3.096 = 73.5 MPa (where 3.09 was obtained from the
table).
This implies that in mean one plate out of 1000 plates loaded at the stress level
73.5 MPa is expected to fail due to fatigue.
Page 8:6

Chapter 8

(At stress lever mean 2.33s = 78 MPa the failure probability is one per cent,
which means that one plate out of 100 is expected to fail due to fatigue if the
plates are loaded at this stress level, and so on.)
Answer: At stress level 73.5 MPa the probability of fatigue failure is expected
to be 0.1 per cent.
8/9.
The fatigue lives of welded beams have been determined at seven (n = 7)
identical tests. Stress range 250 MPa was used. The numbers of cycles to
failure were
49 000, 61 000, 71 000, 81 000, 88 000, 110 000, and 135 000.
Determine the allowable number of cycles at the stress level given, if the
probability of failure is not allowed to exceed 1 per cent with confidence C =
0.95.
The table below gives the k-factors for different values of n and P and with
confidence C = 0.95.
Tolerance limits for the normal distribution (from Rde and Westergren (1998)).
The table gives factors k1 and k2 such that the following kind of statements can be
made:
At least the proportion P is less than x + k1 s with confidence 0.95.
At least the proportion P is greater than x k1 s with confidence 0.95.
At least the proportion P is between x k2 s and x + k2 s with confidence 0.95.
n is sample size, x is sample mean, and s is sample standard deviation.
P = 0.90

P = 0.95

P = 0.99

k1

k2

k1

k2

k1

k2

3.006

3.733

3.707

4.422

5.062

5.758

2.755

3.390

3.399

4.020

4.641

5.241

2.582

3.156

3.188

3.746

4.353

4.889

2.454

2.986

3.031

3.546

4.143

4.633

10

2.355

2.856

2.911

3.393

3.981

4.437

15

2.068

2.492

2.566

2.965

3.520

3.885

20

1.926

2.319

2.396

2.760

3.295

3.621

25

1.838

2.215

2.292

2.638

3.158

3.462

Chapter 8

Page 8:7

Solution:
First calculate the logarithm of the number of cycles N to fatigue failure. It
gives xi = log Ni = 4.69, 4.79, 4.85, 4.91, 4.94, 5.04, and 5.13
Assume that xi (i.e. log Ni ) has a normal distribution. This means that it is
assumed that the xi:s are samples taken from a normally distributed process.
Determine the mean value x of xi and the standard deviation s. One obtains
n

xi
x=
= 4.9074
n
(which gives N = 104.9074 = 80 789 cycles) and
s=

1 n
(xi x )2 = 0.1496
n 1

The probability of fatigue failure should be 1 per cent at confidence level 95

per cent. A table of statistics then gives, using number of samples n = 7,
probability P = 0.99, and confidence level C = 0.95:
k = 4.641
Allowable value of x at failure probability 0.01 (= 1 P) becomes
x = xallowable = x k s = 4.9074 4.641 0.1496 = 4.2131
Number of cycles Nallowable that can be allowed at failure probability 1 per cent
is
x

Nallowable = 10 allowable = 10 4.2131 = 16 334

Answer: Allowable number of cycles is, for 1 per cent failure probability,
Nallowable = 16 300 cycles.

8/10.
A large plate has a small elliptical hole in it. The ratio of the axes of the hole
is 2 to 1. Determine the stress concentration factor Kt when the plate is loaded
in parallel to the
(a) minor axis,
(b) major axis.
Solution:
Handbook (for example Appendix 2 in the textbook) gives, Kt = (1 + 2b / a).
Using b / a = 2 and b / a = 1/2 one obtains Kt = 5 and 2, respectively.
Page 8:8

Chapter 8

Comment: A large stress concentration implies a large probability of fatigue

failure if the loading varies with time. Consequently, notches and other
geometric irregularities should be placed so that the stress concentration is
avoided or made as small as possible (i.e., use large radii at the geometric
irregularities).
Answer: The stress concentration factor Kt is (a) Kt = 5, and (b) Kt = 2.
A ground (surface finish Ra = 3 m) axle of steel
SS 1650-01 (U = 590 MPa) has been
r = 5 mm
M
M manufactured from a circular bar of 200 mm
diameter. The axle has a change of diameter
from 180 mm to 160 mm, with the shoulder
fillet radius r = 5 mm, see figure. Determine the
maximum rotating bending moment M the axle may be subjected to if a safety
factor of 2.0 with respect to fatigue failure should be used.

8/11.

Solution:
The material has fatigue limit FLB = 270 MPa when loaded with an
alternating bending moment. The ultimate strength of the material is U = 590
MPa.
The factors reducing the fatigue limit are
= 0.80 (diameter 200 mm),
= 1 here, because the fatigue notch factor Kf will be used and this factor
takes over the role of (else, one would have had = 0.87),
and = 0.92 (ground axle with ultimate strength U = 590 MPa).
Stress concentration gives Kt = 2.25 (here D / d = 1.125 and r / d = 0.03).
The fatigue notch factor Kf becomes
Kf = 1 + q (Kt 1) = 1 + 0.84 (2.25 1) = 2.05
where q = 0.84 has been taken from Figure 8.11 in the textbook.
Using safety factor sa = 2.0 one obtains

FLB which gives nom

= 0.18 270 MPa
Kf nom
allowable =
allowable =
sa
sa Kf FLB
Using

nom
allowable =

32M
= 0.18 270 106 Pa
3
d

one obtains the bending moment

M = 19 kNm.

Answer: Maximum allowable bending moment M is M = 19 kNm.

Chapter 8

Page 8:9

8/12.
An axle has a change of diameter from 40 mm
to 30 mm with a shoulder fillet with radius r =
r = 1 mm
MT
MT 1 mm, see figure. The fillet is ground (R = 5
a
30
40
m). The material is SS 1550-01 with an
ultimate strength U = Rm = 490 MPa and a
fatigue limit (in torsion) FLT = 140 MPa.
Determine the maximum alternating torque MT the axle may be subjected to if
a factor of safety 3.0 should be used?
Solution:
The shear stress at the shoulder fillet is

shoulder = Kf nom = Kf

16MT

d 3
Figures 8.7, 8, 9, and 11, and Appendix 2 in the textbook give = 0.94, =
0.90, = 1, q = 0.70, and Kt = 1.85.
The fatigue notch factor becomes
Kf = 1 + q (Kt 1) 1 + 0.70 (1.85 1) = 1.6
One obtains
16MT 1
1
1.6
= polished = 0.94 0.90 1 140 106 Pa
3
s
3
d

which, using d = 0.030 m, gives MT 130 Nm.

Answer: The maximum allowable torque is MT = 130 Nm.
8/13.
r = 1.5 mm
P

26

30 mm

An axle has a groove (ground, Ra = 7 m)

according to the figure. The axle is subjected to
a tensile loading P = 18 12 kN. The material
is 1450-01 with U = 430 MPa, Y = 250 MPa,
FL = 140 MPa and FLP = 130 130 MPa.

Determine the safety factor with respect to fatigue failure if the ratio of the
mean value and the amplitude of the loading stress is constant.

Page 8:10

Chapter 8

Solution:
a

140
126

Reduce the fatigue data. One obtains

red
FL = FL = 0.95 0.95 1 140 = 126 MPa

(MPa)
117 MPa

red
FLP = FLP = 0.95 0.95 1 130 = 117 MPa

0.544
1

Draw the Haigh diagram, see figure.

130 180 250 430MPa The notch (the groove) gives the stress concentration factor Kt 2.6.
The fatigue notch factor Kf becomes (notch sensitivity factor q from Figure
8.11 in the textbook)
Kf = 1 + q (Kt 1) 1 + 0.7 (2.6 1) = 2.12
Stresses in the material at the notch become:
mean value m = 2.6 Pm / A and amplitude a = 2.12 Pa / A (the cross-sectional
area A is calculated using the smallest diameter, d = 26 mm, at the notch).
m

Ratio a / m becomes

a 2.12 12
=
= 0.544
m 2.6 18

A straight line with this slope is entered into the Haigh diagram. The
intersection of this line with the reduced fatigue limit of the material gives that
the allowable mean stress m is m 180 MPa. Allowable mean value Pm of
the tensile force P can now be calculated. One has
4Pm
Kt 2 = m = 180 106 Pa, which gives Pm = Pallowable = 36.7 kN
d
Without safety factor the mean value of the froce can be Pallowable = 36.7 kN.
But, according to the problem, the applied load was P = Papplied = 18 kN.
The safety factor then is
Pallowable 36.7
s=
=
2.0
Papplied
18
Answer: The safety factor is s = 2.0, approximately.
8/14.
When testing smooth specimens (polished) made from a high strength steel
with yield limit Y = 1510 MPa and ultimate strength U = 1800 MPa, fatigue
limits at N = 107 cycles were obtained at the following stress levels:
800 400, 400 560, 0 720, 400 860, and 800 850 MPa.
Draw the Haigh diagram for notched cylindrical specimens of diameter D = 11
mm. The notch is 1.5 mm deep with a radius in the trough of 0.5 mm and it is
polished.
Chapter 8

Page 8:11

Solution:
Determine the reduction factors , , , the stress concentration factor Kt, and
the fatigue notch factor Kf.
Diameter D = 11 mm gives 0.98 1.0.
No bending or torsion give = 1.0.
Polished surface give = 1.0.
Thus = 1.0, which means that no reduction due to surface finish and
volume is needed here.
Stress concentration:
D / d = 11/8 = 1.38 and / d = r / d = 0.5/8 = 0.06 give Kt = 2.8.
Fatigue notch factor Kf = 1 + q (Kt 1) = 1 + 0.95 (2.8 1) = 2.7.
To make the Haigh diagram valid for the notched specimen, the mean value
(corresponding to a static load) should be reduced by the factor Kt and the
amplitude by the factor Kf.
The point of fatigue limit 800 400 then moves to 800 / 2.8 400 / 2.7 = 286
148 MPa. For the points given, the following fatigue limits are obtained:
286 148, 143 207, 0 267, 143 318, and 286 315 MPa.
The yield strength of the material is Y = 1510 MPa. If yielding is to be
avoided, the Haigh diagram should be limited at stress Y / Kt = 1510 / 2.8 =
540 MPa both on the abscissa and on the ordinate.
Answer: m = m / Kt, a = a / Kf and m + a < Y / Kt, where Kt = 2.8
and Kf = 2.7, which give fatigue limits 286 148; 143 207; 0 267; 143
318; and 286 315 MPa.

8/15.
The right end of the cantilever beam (see figure) is moved up and down; the
displacement upwards is 2a and downwards a. Determine the maximum
allowable value of a with respect to fatigue failure. Use safety factor s = 2.0.
r = 10

65

d = 40

200

Page 8:12

The surface of the beam is polished. The

r = 5 (mm) material has the modulus of elasticity E = 206
GPa, the yield limit 390 MPa and the ultimate
D = 50 strength 590 MPa. Fatigue limits are 270
200 mm
MPa at alternating bending and 240 240 MPa
at pulsating bending.

Chapter 8

Solution:
The force P that loads the free (right) end of the beam is obtained from
PL 3
3EI
=
which gives P = 3
3EI
L
Here = 0.5a 1.5a, L = 400 mm and I (= D4 / 64) is the second moment of
area of the circular beam cross section with diameter D = 50 mm (implying
that here the influence of the notch on the beam deflection is neglected).
At the notch the bending moment Mb, notch is obtained as
L 3EI
3EI
Mb, notch = P = 2 = 2 (0.5 a 1.5 a)
2 2L
2L
At the shoulder the bending moment becomes
3EI
3EI
Mb, shoulder = P L = 2 = 2 (0.5 a 1.5 a)
L
L
Stress concentration at the notch: Ktnotch = 1.8 and q = 0.83 give Kfnotch = 1.66.
Stress concentration at the shoulder: Ktshoulder = 1.4 and q = 0.88 give Kfshoulder =
1.35.
The stresses at the notch and at the shoulder may now be determined (as a
function of deflection a). One obtains
Mb, notch
3EI
= 2
(Ktnotch 0.5 a Kfnotch 1.5 a)
notch =
Wb, notch 2L Wb, notch
and

shoulder =

Mb, shoulder
3EI
= 2
(Ktshoulder 0.5 a Kfshoulder 1.5 a)
Wb, shoulder L Wb, shoulder

where Wb is the section modulus in bending.

Reduction factors: = 1, = 0.87, = 1
The reduced fatigue limit at alternating loading
a (MPa)
becomes
270
210 MPa
235
FLB,red = 0.87270 = 235 MPa
and at pulsating loading it becomes
m FLBP,red = 240 0.87240 = 240 210 MPa.
590 This gives a Haigh diagram according to the
m 240 390
(MPa) figure.
The loading is such that the ratio of the amplitude to the mean value is
constant. This gives at the notch:
notch
a

notch
m

1.66 1.5
= 2.77
1.8 0.5
Chapter 8

Page 8:13

Enter a straight line of slope 2.77 in the Haigh diagram. This line intersects the
reduced fatigue limit at m 82 MPa. Fatigue failure is thus expected when the
mean value of the stress is m = 82 MPa (the stress amplitude is then 2.77m =
227 MPa).
The mean value m of the stress at the notch will now be determined from the
calculated bending moment. Using the stress concentration factor Ktnotch = 1.8
on the nominal mean stress applied, and using the safety factor s = 2 on the
reduced fatigue limit m = 82 MPa determined, one obtains

Mb,mean
3EI
1
32
notch
6 1
1.8 nominal = m
which gives 1.8
= 1.8 2 0.5 a 3 = 82 10 Pa
s
2
Wb, notch
2L
d

From this a = 0.00048 m = 0.48 mm is solved.

At the shoulder is obtained:

shoulder
a

1.35 1.5
=
= 2.9
1.4

0.5
shoulder
m

Enter a straight line of slope 2.9 in the Haigh diagram. The line intersects the
reduced fatigue limit at m 78 MPa. Fatigue failure is thus expected when the
mean value of the stress is m = 78 MPa (the stress amplitude is then 2.9m =
226 MPa).
The mean value m of the stress at the shoulder will now be determined from
the calculated bending moment. Using the stress concentration factor Ktshoulder =
1.4 on the nominal stress applied to the beam, and using the safety factor s = 2
on the reduced fatigue limit, one obtains

Mb,mean
3EI
32
1
1
shoulder
1.4 nominal = 1.4
= 78 106
= m
= 1.4 2 0.5 a
3
2
s
Wb, shoulder
L
D

From this a = 0.00058 m = 0.58 mm is solved.

It is concluded that the notch is the most critical part of the structure. The
stresses at the notch will be limiting for the displacement a.
Answer: Maximum allowable displacement a at the free beam end will be a =
0.48 mm.

Page 8:14

Chapter 8

8/16.

cross section:

200
100

200 9 mm

15

holes

1m

1m

A beam with cross section HE200B is made of

the material 1312-00. The beam is simply
supported at its ends and loaded by a force P =
P0 P0 at its centre point.
Determine, with respect to fatigue failure of the
beam, the maximum allowable value of P0. Each
flange has two rows of holes (machined, Ra =
20 m) with diameter 8 mm. The holes are
drilled 100 mm from each other.
Material data: U = 360 MPa, Y = 240 MPa,
FL = 110 MPa and FLP = 110 110 MPa.

Solution:
The largest bending moment in the beam will appear at the mid-point of the
beam, and it is (use total beam length 2l)
P l
l
M=
= (P0 P0 )
where l = 1 m
2
2
The beam HE200B has the section modulus Wb = 570 10 6 m3 in bending.
If the holes had not been there, the stress 0 in the lower flange of the beam
would have been
M
0 =
Wb
Due to the holes, stress concentration will appear.
It is assumed that the holes are so widely separated that when calculating the
stress concentration factor at one hole, the disturbance of the stress distribution
from the other holes may be neglected.
Study half of the flange width. This can be done because of symmetry. Using
r / a = 4/50 and B / a = 100/50 the stress concentration factor Kt is obtained as
Kt 2.8.
The fatigue notch factor Kf becomes Kf = 1 + q (Kt 1) 1 + 0.8(2.8 1) =
2.44.
This gives the maximum stress at a hole situated where the bending moment
is the largest. One obtains
B P0 l
= (2.8 2.44)
B 2r 2Wb
where B is half the width of the flange, thus B = 100 mm.

Chapter 8

Page 8:15

(MPa)
94 MPa

110
94
2.44
2.8
110

240

Reduction of fatigue limits:

The flange thickness is 15 mm, the fatigue
notch factor Kf is used, where Kf > 1, and the
surface of the hole is machined. This give
0.95, = 1, and 0.90.
The reduced fatigue limit at a pulsating load
then becomes

red
FLP = FLP = 0.95 1 0.90 110 MPa = 94 MPa

The allowable stress amplitude at the hole is 94 MPa (with no safety factor). It
gives
0.100 P0 1
B P0 l
= 2.44
= 94 10 6 Pa
2.44
6
B 2r 2Wb
0.092 2 570 10
which gives P0 40 kN.
The stress amplitude was used as the design stress. The mean stress mean (=
2.894 / 2.44 = 108 MPa) is less than 110 MPa, implying that the straight line
giving the service stress intersects the horizontal branch of the fatigue limit
curve just slightly to the left of the breaking point at 110 MPa. Thus, it was
correct to use the stress amplitude a = 94 MPa as the design stress.
Answer: The force P = P0 P0 may be allowed to be, approximately, P =
40 40 kN (no safety factor included here).
8/17.
P

side view
32 H

A
top view
beam cross section
3h
A
4H

Page 8:16

h
H

The structure in the figure is loaded by a force

P = P0 P0. Determine, with respect to fatigue
failure at hole A, the maximum allowable value
of P0. The hole is machined (Ra = 7 m) and it
has a diameter of 20 mm. Material data (for SS
1510-00): U = 600 MPa, Y = 320 MPa, FL =
230 MPa and FLP is unknown (use thumb rule
FLP 0.85FL ).
Geometrical data: H = 200 mm and h = 20 mm
(cf. Chapter 1, Section 1.4.8).

Chapter 8

Solution:
The nominal stress at the hole becomes nom = Mz / I
where M = P32H, z = H /2 and
4H H 3 (4H 6h) (H 2h)3
I=

12
12
One obtains nom = 2125P = 2125(P0 P0) N/m2 (P in N).
The stress concentration factor Kt at the hole is Kt = 3.0 (here the case a small
hole in a large plate subjected to uniaxial loading is used). This gives the
fatigue notch factor
Kf = 1 + q(Kt 1) = 1 + 0.87(3.0 1) = 2.74
The factors reducing the fatigue limit due to surface finish and volume
become: 0.86, 0.90 and = 1.
Pulsating loading implies that the fatigue limit FLP should be used. One
obtains
FLP = 0.85 FL
By use of the reduction factors one obtains
FLPred = FLP = 0.85 FL = 0.860.9010.85230 MPa = 151 MPa
The amplitude P0 of the load gives the stress amplitude Kfnom = 2.74nom.
Let Kfnom = FLPred. It gives
2.742125P0 = 151106, which gives P0 = 26 kN.
Answer: Without safety factor, the force P may be P = 26 26 kN.

Damage accumulation, counting of load cycles

8/18.
A component of a road vehicle is subjected to a repeated loading sequence.
One sequence contains the following loading (i.e. stress) amplitudes (the
loading is alternating)
a = 200, 180, 150, and 100 MPa
The number of loading cycles at each stress level is
n = 15, 20, 150, and 3000, respectively
The Whler curve of the material is, in this stress range, given by the relation
a = 55logN + 430 MPa
Chapter 8

Page 8:17

Determine the damage accumulation D due to one loading sequence, and then
determine how many loading sequences the component might resist before
fatigue failure.
Solution:
The Whler curve a = 55 logN + 430 of the material gives the fatigue life
N at a given stress amplitude a.
( 430 a ) / 55
One obtains
N = 10
At the different stress levels the following damages are obtained
a (MPa)
200
180
150
100

Ni

ni

damage ni / Ni

15 199
35 112
123 285
106

15
20
150
3 000

15 / 15 199
20 / 35 112
150 / 123 285
3 000 / 106

Damage D due to one sequence is

15
20
150
3 000
D=
= 5.77 10 3
+
+
+
6
15 199 35 112 123 285 10
The expected number of sequences Ns to fatigue failure is

Ns = 1 / D 173

Answer: Damage D due to one sequence is D = 5.77 10 3, and expected

number of sequences Ns to fatigue failure is Ns 173.
8/19.
For a welded plate, a number of points on the Whler curve have been
determined. At the probability of failure p = 50 per cent and at the stress ratio
R = 0 the following points were obtained:
max (MPa)
Nf (cycles)

189
1104

182
3104

161
1105

131
3105

107
1106

(a) Use linear regression to fit a straight line in a maxlogN diagram to these
points. Estimate by use of linear damage accumulation (the Palmgren-Miner
rule) how many sequences (as given below) the plate might resist before
fatigue failure is expected. One sequence contains the following stress levels
and cycles:

Page 8:18

Chapter 8

max (MPa)
n (cycles)

120
100

150
50

180
20

160
40

140
60

(b) According to Jarfall (1977), see reference in textbook, the relationship

k N = constant
could give a better result at spectrum loading (i.e. loading with different,
possibly random, amplitudes). This relation gives a straight line in a
logmaxlogN diagram. Fit a straight line in a logmaxlogN diagram to the data
given and estimate by the formula so obtained the number of sequences to
fatigue failure.

Solution:
(a) Fit a straight line in a maxlogN diagram to the measured data. Let the
coordinate y be y = max and the coordinate x = logN. This gives
x=

4.47712

5.47712

y=

189

182

161

131

107

Fit the straight line y = Ax + B to the points.

Linear regression gives (n is number of points)
n x y x y
y A x
A=
= 43.034945 and B =
= 368.78089
2
2
n
n x ( x)
The coefficient of correlation r becomes
r=

n x y x y
[n x ( x) ] [n y ( y) ]

which means (r close to

= 0.98

1) that the points fall almost on a straight line.

The Whler curve of the material thus becomes

max = 43.03 logN + 368.78 MPa
( 368.78

From this is solved:

) / 43.03

max
N = 10
For the different stress levels the fatigue life and the damage are obtained as

Chapter 8

Page 8:19

max

Ni

ni

damage ni / Ni

120
150
180
160
140

603 815
121 282
24 360
71 028
207 093

100
50
20
40
60

0.165613610 3
0.412262510 3
0.820997210 3
0.563161710 3
0.289724910 3

The damage D due to one sequence becomes D = (ni / N i) = 2.2517610 3 .

Expected number of sequences to fatigue failure is 1 / D = 444.
(b) Fit a straight line in a logmaxlogN diagram to the measured data. Let y =
logmax and x = logN; one then obtains
x=
y=

4
2.276

4.47712
2.260

5
2.207

5.47712
2.117

6
2.029

This gives the line logmax = A logN + B, where A = 0.12755 and B = 2.8146
(and the coefficient of correlation r now is r = 0.9723).
The Whler curve gives

N = 10

( B log max ) / A

from which is obtained

max

Ni

ni

damage ni / Ni

120
150
180
160
140

582893
101353
24270
61108
174077

100
50
20
40
60

0.171558110 3
0.493327410 3
0.824061210 3
0.654582610 3
0.344676010 3

The damage D due to one loading sequence becomes D = (ni / N i) =

2.488210 3, which gives the expected number of sequences Ns to fatigue
failure: Ns = 1 / D 402.
Answer: Number of sequences to fatigue failure becomes (a) 444, and (b) 402,
approximately.

Page 8:20

Chapter 8

8/20.
A part of a structure is subjected to sequences of vibrations. Each sequence
consists of free damped vibrations giving rise to stresses with the stress ratio R
= 1, i.e., the mean value of each stress cycle is zero. In the point of the
material to be investigated, the stress amplitude of the first cycle in each
sequence is 0. The damping of the vibrations is characterized by the
logarithmic decrement . The logarithmic decrement is defined as =
ln ( i / i+1) where i is the amplitude of the vibration cycle i and i+1 is the
amplitude of the cycle i + 1, see figure (a). The Whler curve of the material
is, for stress amplitudes a < 0, assumed to be a straight line in a diagram with
loga on the ordinate (the y axis) and logN on the abscissa (the x axis), cf.
equation (8.2b). The slope of the curve is 1/m in the diagram. The material
has no fatigue limit. Determine the number of vibration sequences the structure
may be expected to survive. The number of cycles to failure at the stress
amplitude 0 would be N0.
log a
log

0
0

Figure (a)

Figure (b)

log N 0

log N

Solution:
The Whler curve is a straight line of slope 1/m in a log alog N diagram.
The equation of the curve may be written
1
log a = log N + C
(a)
m
The constant C is determined from the condition that the fatigue life is N0
cycles when the stress amplitude is 0. This gives
1
(b)
log 0 = log N0 + C
m
By inserting C as given by relation (b) into the expression (a), the equation of
the Whler curve takes the form
a 1
N0
log = log
(c)
0 m
N
The stress amplitude of the different stress cycles following the first stress
cycle (with amplitude 0) in each sequence will now be determined together

Chapter 8

Page 8:21

with the fatigue life at each stress amplitude. Let 1 = 0, 2 = 1 = 20,

..., i = i 0, and so on. The definition of the logarithmic decrement yields
i
i
= ln
= ln
(d)
i + 1
i
Solve (d) for . It gives = exp( ) (here exp( ) stands for e ).
Determine the cyclic life N = Ni at stress level i = i 0. The relation (c) gives
i 0 1
N0
log
= log
(e)
0
m
Ni
This relation gives Ni = N0 / i m .
The accumulated damage D due to one loading sequence becomes
1 1 1
1 m 2m 3m
1
D = + + + = + +
+
+ = (1 + m +2m + 3m + .) (f)
N0 N1 N2
N0 N0 N0 N0
N0
Use the series expansion

1
when | x| < 1
1x
This gives the damage D of one sequence of vibration as:
1
D=
N0 (1 m )
1 + x + x 2 + x 3 + . =

(g)

(h)

Let I be the total number of sequences the structure may be subjected to. The
accumulated damage then is I D. Fatigue failure is expected when the
accumulated damage becomes unity, thus when I D = 1. This gives
1
(i)
1=I
N0 (1 m )
Solve for I, which gives

1
D
Using numerical values = 0.05 (giving 0.95) and m = 8 one obtains
I = 0.33 N0
I = N0 (1 m ) =

Hence, the stress amplitude 0 alone would have given a cyclic life of N0
loading cycles. The free damped vibration after the first cycle (with amplitude
0) reduces, in this case, the fatigue life of the structure to one third of N0.
Answer: The number I of sequences to fatigue failure is expected to be
1
I = = N0 (1 m ) where = e
D
If = 0.05 and m = 8 one obtains I = 1 / D = 0.33 N0 (solution also given in the
Example in Section 8.2.3 in the textbook).
Page 8:22

Chapter 8

8/21.
At a random loading of a structure, the following time sequence was recorded
(force in kN, see also figure):
4 6 5 7 3 4 2 4 0 4 2 4 3 7 6 8 7 9 6 7 2 6 1 3 0 2 0
2 0 5 2 4 3 7 2 5 3 6 5 7 2 3 1 5 4 (= static level at rest)
9
8
7
6
5
4
3
2
1
0

P (kN)

time

Determine
(a) the distribution of the peaks (number of peaks larger than or equal to a
certain level),
(b) the distribution of the troughs (number of troughs smaller than or equal to a
certain level), and
(c) the distribution of exceedances (the load spectrum).
Solution: See answer.
Answer:
9
8
7
6
5
4
3
2
1
0

P (kN)

cumulative distribution of
troughs

exceedances

peaks
5

10

15

Chapter 8

20

25

number

Page 8:23

8/22.
At a random loading of a structure, the following time sequence was recorded
(force in kN, same sequence as in Problem 8/21, see also figure in that
problem):
4 6 5 7 3 4 2 4 0 4 2 4 3 7 6 8 7 9 6 7 2 6 1 3 0 2 0
2 0 5 2 4 3 7 2 5 3 6 5 7 2 3 1 5 4 (= static level at rest)
Identify loading cycles by use of the rain flow count method. Make a list of
those cycles whose amplitude is larger than, or equal to, 1 kN.
Solution:
This solution is given for the load sequence as it is given in the problem. It
should be noted, however, that the solution could be slightly simplified if the
sequence is rearranged so that it starts at its largest maximum or its smallest
minimum. Here, fore example, the part of the sequence from the starting point
at P = 4 kN to the minimum at P = 0 could be cut off and moved to the end of
the sequence given. If this is done, the final result will be the same, but the
problem of obtaining half cycles that must be matched to each other, see
below, will be avoided.
Now, let one drop of rain-water start to run (flow) from each maximum value
and each minimum value of the loading sequence (the starting point and the
final point here form a minimum). A total of 44 drops are needed, see the
figure.

9
8
7
6
5
4
3
2
1
0

P (kN)

17
15

16
14 20
5 7

31
23
25 27

12
8

10

22
24
4

Page 8:24

21

13
11

29 35
39
37

19

26

Chapter 8

28

43

38

41
33
32 36
30
34 42
18

44
40
time

Make a list of the run-ways of all drops. For each drop, write down its number
(the drop number), its stating point (load level in kN) and its end point (load
level in kN). It gives
Drop
No

starts stops
at (kN) at (kN)

1
2
3
4
5
6
7
8
9
10
11
12
13
14

4
6
5
7
3
4
2
4
0
4
2
4
3
7

7
5
6
0
4
3
4
2
9
2
4
3
4
6

15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30

6
8
7
9
6
7
2
6
1
3
0
2
0
2
0
5

7
7
8
0
7
6
6
2
3
1
2
0
2
0
7
2

31
32
33
34
35
36
37
38
39
40
41
42
43
44

2
4
3
7
2
5
3
6
5
7
2
3
1
5

5
3
4
2
7
3
5
5
6
1
3
2
5
4

Collect pairs of drops so that one pair forms a closed loop. Write down the two
drop numbers, the minimum value (in kN) of the closed loop and the
maximum value (in kN). One obtains
Drops
No

mini- maximum mum

value value

(1
2, 3
4, 29
5, 6
7, 8
9, 18

4
5
7
3
2
0

7)
6
0
4
4
9

10, 11
12, 13
14, 15
16, 17
19, 20
21, 22
23, 24
25, 26
27, 28
30, 31

2
3
6
7
6
2
1
0
0
2

4
4
7
8
7
6
3
2
2
5

Chapter 8

32, 33
34, 35
36, 37
38, 39
(40
41, 42
(43
(44

3
2
3
5
7
2
1
5

4
7
5
6
1)
3
5)
4)

Page 8:25

Here 20 closed loops have been identified. Due to the fact that the sequence
does not starts at the largest maximum or the smallest minimum (see the
introduction to the solution above) there will be some single drops left: drops
No 1, 40, 43, and 44. Also these drops may, however, be grouped into closed
loops in the following way:
Divide the running way of drop No 1 into two parts, 1a and 1b. It gives
Drop
No
starts stops
1a
1b
40
43
44

4
5
7
1
5

5
7
1
5
4

If we think of a continuation of the given sequence as a repetition of the first

sequence, then the run-way of the drop No 43 would continue the same way as
the drop No 1 has from 5 kN and onwards. The drop No 43 would then start at
1 kN, initially it will run to 5 kN and then further on the next roof to 7 kN.
Thus drops 43 and 1b together form half a cycle from 1 kN to 7 kN. This
half-cycle may be matched to the drop No 40 to form a full loading cycle (one
loop). The drop 1a now stops at 5 kN and it is seen that this part may be
matched to the drop No 44. The five one-way drop ways above may thus be
combined to give two closed loops. Again, the numbers of the two drops, the
minimum value (in kN) of the closed loop and the maximum value (in kN) are
written down. One obtains
Drop way minimum maximum
No
value
value
1a, 44
4
43, 1b, 40 1

Page 8:26

5
7

Chapter 8

In total 22 loops (cycles) have been obtained. These cycles may now be used
in a fatigue analysis. The ranges (twice the amplitude) of he loading cycles are
given in the tables above. Sometimes cycles of a small amplitude do not take
part in the fatigue process. If cycles with an amplitude less than 1 kN are
neglected, then 12 loading cycles are left to form the basis of a fatigue damage
analysis. These are
Drops
No
4, 27
7, 8
9, 18
10, 11
21, 22
23, 24

minimum maximum
value
value
0
2
0
2
2
1

7
4
9
4
6
3

25, 26
28, 29
30, 31
35, 40
36, 37
43, 1b, 34

0
0
2
2
3
1

2
2
5
7
5
7

Answer: Twelve cycles with an amplitude larger than or equal to 1 kN are

obtained (and ten cycles with an amplitude less than 1 kN). Minimum and
maximum values of the twelve cycles are: (0,7), (2,4), (0,9), (2,4), (2,6), (1,3),
(0,2), (0,2), (2,5), (2,7), (3,5), and (1,7).

Chapter 8

Page 8:27

Empty page.

Page 8:28

Chapter 8

Solutions to problems in
T Dahlberg and A Ekberg: Failure, Fracture, Fatigue - An Introduction.
Studentlitteratur, Lund 2002, ISBN 91-44-02096-1.

Chapter 9
Strain-based fatigue design
Problems with solutions
9/1.
A flat bar of material SAE 1045 with a
rectangular cross section has a small circular
hole at its longitudinal centre axis. The bar is
subjected to a remote alternating stress
= 300 MPa.
(a) Estimate by use of Neubers method and the Morrow equation the number
of cycles to fatigue failure. Assume that Kf = 0.9Kt.
(b) For the maximum stress and strain at the hole draw the hysteresis loop
when = 300 MPa.
(c) Determine the fatigue life if = 100 200 MPa.
Solution:
(Compare with the problem solved in Section 9.3.3 in the textbook.)
(a) Number of cycles to fatigue failure
If the stress state at the small hole in the flat bar were purely elastic, then the
stress concentration factor would have been Kt = 3. The fatigue notch factor Kf
is
(a)
K = 1 + q (K 1)
f

No information for calculation of the notch sensitivity factor q is given here.

Instead, Kf = 0.9Kt = 2.7 was given. Thus, use Kf = 2.7.
Due to the high stresses close to the hole, the material at the hole will yield
locally. Taking this into consideration, the stress concentration factor K and
the strain concentration factor K may be determined from equations (9.9a,b) in
the textbook. For low stresses, the second term in the material relation (9.2a,b)
can be disregarded. This means that here, where the stresses far away from the
hole are relatively low (i.e. mainly within the elastic range), the second term
will be small when compared to the first term. Expressions (9.9a,b) in the
textbook then become
FMCHAP9P.DOC/2010-01-15/TD

Chapter 9

Page 9:1

K =

max
max max
and K =
=

/ E

(b,c)

where, as can be seen, only the first term (i.e. Hookes law) in (9.2a,b) has
been used in (c).
The Neuber hyperbola becomes

Kf2 2
= K K =
E

(d)

The material data given earlier (see the textbok) for the material SAE 1045 and
the nominal stress amplitude = 300 MPa now determine the Neuber
hyperbola. The local stress amplitude a and strain amplitude a at the hole are
obtained as the intersection point between the Neuber hyperbola and the
material stress-strain relation for cyclic loading. One obtains

Kf22 2.72 3002

=
=
= 3.2805
a a
E
200 000

(e,f)
1/ 0.18

a a 1/n
a

a
= + =

+
a
E K
200 000 1344

This system of equations gives a = max = 499 MPa and a = max = 0.006574.
The stress and strain concentration factors K and K then become
K =

max 499
max
0.006574
=
= 1.663 and K =
=
= 4.3828
300
/ E 300/ 200 000

(g,h)

(Verify: K K = 1.663 4.383 = 7.29 = 2.72 = Kf2, as it should.)

Now the number of cycles to fatigue crack initiation (or fatigue failure) may be
determined. According to Morrow, equations (9.4) and (9.7) in the textbook,
and by use of the mean stress m = 0, one obtains
f m
1227
a =
(i)
( 2N )b + f ( 2N )c =
( 2N ) 0.095 + 1.0 ( 2N ) 0.66
E
200 000
Using a = 0.006574, the fatigue life 2N = 4600 reversals to failure is obtained,
giving N = 2300 cycles to fatigue failure.
(b) Display the hysteresis loop for the material at the hole, when the remote
loading stress is = 300 MPa
In problem (a) above the stress and strain amplitudes at the hole have been
calculated when the remote stress amplitude is = 300 MPa. This implies
that when the remote stress has its maximum ( = 300 MPa), the upper end
point of the hysteresis loop is situated at = max = 499 MPa and = max =
Page 9:2

Chapter 9

0.006574, see point A in the figure below. When the remote stress has its
minimum ( = 300 MPa), the stress and strain at the stress concentration
(i.e. at the hole) is = max = 499 MPa and = max = 0.006574. These
values give the lower end point of the hysteresis loop, see point B in the
figure.
Choose point A as starting point of the hysteresis loop, see figure below. A
change of the remote stress (the stress far away from the stress
concentration) will cause a change of stress and a change of strain at the
stress concentration (the hole). The changes and are obtained by use of
the Neuber hyperbola (9.13d) and the material relation for changes of stress
and strain (9.3a). Using = 300 MPa one obtains
2
2
2
2

= Kf () = 2.7 300 = 3.2805

E
200 000

(j,k)
1/n
1/ 0.18

=
+2
=
+2

E
2K
200 000
2 1344
From this system of equations = 0.004670 and = 703 MPa are solved.
The lower branch of the hysteresis loop, starting at point A, will pass through
the point C. The co-ordinates of point C are obtained from
C = A = 499 MPa 703 MPa = 204 MPa

(l,m)
C = A = 0.006574 0.004670 = 0.001904
For the student: verify that the change of the remote stress = 600 MPa will
take this branch of the hysteresis loop to the point B.
499 MPa

- 0.001912
203 MPa

- 0.006574
0.006574
C
B

0.001912
- 203 MPa

- 499 MPa

We now turn to the upper (i.e. the left) branch

of the hysteresis loop. At a change of stress
= 300 MPa the upper branch of the hysteresis
loop, starting at point B, will pass through point
D. The coordinates of point D are obtained from
D = B + = 499 MPa + 703 MPa = 204 MPa

D = B + = 0.006574 + 0.004670

(n,o)
= 0.001904

By use of these values (and perhaps some more)

the hysteresis loop may be drawn, see figure.

Chapter 9

Page 9:3

Comment: It is noticed that when the loop passes the point C the external loading (far
away from the hole) will give a nominal stress equal to zero, i.e. the bar has been
loaded a number of times up to stress = 300 MPa and then unloaded by the stress
= 300 MPa so that the total stress far away from the stress concentration is =
0. Thus, when the bar is unloaded (i.e. when = 0), the residual stress at the hole is
C (compressive) and the residual strain is C. The residual stress is compressive
because the material has been stretched out (by plastic deformation) at the hole and
it was too long when the bar was unloaded. On the other hand, when unloading
from the remote stress = 300 MPa to stress = 0, the residual stress at the hole
will become D (in tension) and the residual strain is D. In this case the material was
compressed at the hole so it was made too short when the bar was unloaded
from = 300 MPa to = 0. This explains why the residual stress is in tension at
point D when the remote stress is = 0. (The remote stress = 0 indicates that the
axial force in the bar is zero. This is the case also at the hole. For equilibrium to be
fulfilled at the cross section of the bar at the hole, there must be residual stresses
both in tension and in compression. This implies that residual stresses in tension
close to the hole will be balanced by residual stresses in compression further away
from the hole, and vice versa.)

(c) Fatigue life when = 100 200 MPa

The same bar as above (Kf = 2.7) will be investigated, but now the bar is
loaded to a nominal (remote) stress = 100 200 MPa. The nominal
(remote) stress will now vary between 300 MPa and 100 MPa. This
implies that the stress at the hole will vary between A (when = 300 MPa)
and a point on the hysteresis loop below C (the stress at the hole is C when
the remote stress is = 0).
Determine some more points on the hysteresis loop that is obtained when =
100 200 MPa.
499 MPa

- 0.001912
203 MPa

Select, once again, the starting point of the loop

at point A where A = 499 MPa and A =
0.006574. Determine the point on the loop (the
stress at the hole) when = 200 MPa. It
gives, as in (j,k) above,

- 0.006574
E 0.006574

F
B
Page 9:4

2
2
2
2

= Kf () = 2.7 200 = 1.458

E
200 000
0.001912
1/ 0.18
C

- 203 MPa = + 2

200 000
2 1344

- 499 MPa

Chapter 9

(p,q)

This system of equations gives = 519 MPa and = 0.002810. The point E
on the hysteresis loop is now obtained from
E = A = 499 MPa 519 MPa = 20 MPa

(r,s)
E = A = 0.006574 0.002810 = 0.003764
Finally, determine the turning point of the new loop (the stress at the hole)
when = 400 MPa (the remote stress then is = 100 MPa). It gives, as
in (j,k) above,
2
2

(
)
K
2.72 4002
f

=
=
= 5.832

E
200 000
(t,u)

1/ 0.18

=
+2

200 000
2 1344

This gives = 829 MPa and = 0.007038.

The turning point F on the new hysteresis loop is now obtained from
F = A = 499 MPa 829 MPa = 330 MPa

F = A = 0.006574 0.007038 = 0.000464

(v,w)

Using point F as starting point, some points on the upper branch of the
hysteresis loop may be determined from the stress changes already calculated.
This is left to the student as an exercise.
The number of loading cycles to crack initiation is obtained, according to
Morrow, from
f m
a =
(x)
( 2N )b + f ( 2N )c
E
The mean value m of the stress becomes m = (A + F) / 2 = (499 330) / 2 =
84.5 MPa, and the strain amplitude will be a = (A F) / 2 = (0.006574
+ 0.000464) / 2 = 0.003519. By use of these values in (x), and by use of the
parameter values given, one obtains
1227 84.5
a = 0.003519 =
(y)
( 2N ) 0.095 + 1.0 ( 2N ) 0.66
200 000
From this 2N = 23 070 reversals, giving N = 11 500 cycles, to fatigue failure is
solved. (This result may be compared with the result obtained in Section 9.3.3
for the same structure. There the remote stress = 200 200 MPa gave the
fatigue life N = 6 500 cycles to failure.)

Chapter 9

Page 9:5

Answer: (Compare with the problem solved in Section 9.3.3 in the textbook.)
(a) At remote stress = 300 MPa the number of cycles N to fatigue failure
is N = 2300, approximately,
(b) stress and strain at the hole are (end points of the hysteresis loop)
a = 499 MPa and a = 0.006574, respectively, and
(c) fatigue life is expected to be N = 11 500 cycles to failure (according to
Morrow) at remote stress = 100 200 MPa.

9/2.
In a structure a notch with stress concentration factor Kt = 2.8 has been found.
Assume that the material follows the cyclic stress-strain curve
a a 1/n

1/n
and for changes:
a = +
=
+2

E K

E
2K
where E = 210 GPa, K = 1200 MPa and n = 0.19.
The structure is loaded so that the nominal stress some distance away from the
notch varies between 50 MPa and + 250 MPa. (In addition to this stress, the
stress concentration will be added at the notch.)
Determine the expected number of load cycles to fatigue failure.
Use fatigue notch factor Kf = Kt, the reduction of the cross-sectional area at
rupture is = 65 per cent, and the ultimate strength of the material is U =
470 MPa.
Solution:
The material relation and the Neuber hyperbola give for stress and strain
ranges (stress range is = 250 ( 50) = 300 MPa). As we are going to use
the Coffin-Manson rule, only ranges are of interest here. The influence of the
mean values is disregarded. One obtains

1/n
1/0.19

+2
=
+2

=
E
2 K
210 000
2 1200
(Kf )2 (2.8 300)2
Neuber:
=
=
= 3.36
E
210 000
This system of equations gives = 650 MPa and = 0.005169.

Coffin-Mansons fatigue law reads

U
1
1
= ln
= 1.0498
= 3.5 N 0.12 + D 0.6N 0.6 where D = ln
E
1
1 0.65
Page 9:6

Chapter 9

(a)
(b)

(c)

Numerical values give

470
N 0.12 + 1.04980.6 N 0.6
210 000
from which N = 19 200 cycles (approximately) is solved.
Thus, failure is expected after, approximately, 19 000 cycles.
0.005169 = 3.5

(d)

Answer: Fatigue failure is expected after N = 19 000 cycles (approximately).

9/3.
A test specimen is subjected to a remote nominal stress = 500 200 MPa.
The specimen has a notch with stress concentration factor Kt and fatigue notch
factor Kf such that Kt = Kf = 2.8. Assume that the material follows the cyclic
stress-strain relationships (for amplitudes and changes, respectively)
a a 1/n

1/n
a = +
and
=
+2

E K
E
2K
Determine the stress (mean value and amplitude) and the strain (mean value
and amplitude) in the material at the place of stress and strain concentration.
Then use the calculated values in a low-cycle fatigue analysis to determine the
expected fatigue life N of the test specimen.
The material has the following cyclic properties:
E = 206 GPa, K = 1750 MPa, n = 0.11, Y = 850 MPa, f = 1400 MPa,
f = 0.60, b = 0.10, and c = 0.55.
Solution:
Determine the hysteresis loop obtained when the remote stress is varying
between 700 MPa and 300 MPa.
Comment: When = 700 MPa, the stress at the stress concentration should be
more than 700 MPa, of course, and less than 2.8700 = 1960 MPa, which is the
stress one should have had if the material were fully elastic.
The upper end point of the hysteresis loop is obtained from the intersection of
the Neuber hyperbola and the cyclic stress-strain relation. One obtains
Kf2 2 2.82 7002
a a =
=
= 18.65
(a)
E
206 000
a a 1/n
a
a 1/0.11

+
a = + =
E K
206 000 1750

Chapter 9

(b)

Page 9:7

From these two equations the upper end point (the turning point) of the
hysteresis loop will be found. One obtains
max = 1077 MPa and max = 0.017319
(c,d)
The stress concentration factor K and the strain concentration factor K may
now be determined (if desired). One obtains
max max
1077
0.017319
K =
=
= 1.5386 and K =
=
= 5.096
(e,f)
700
/ E 700 / 206 000
One finds that KK = 7.8418, and that Kf2 = 2.82 = 7.84, as it should.
A change = 400 MPa (i.e. twice the amplitude 200 MPa) of the remote
stress causes changes of and at the stress concentration. The changes
and at the stress concentration are obtained from the intersection of the
Neuber hyperbola and the stress-strain relation for changes. One obtains
Kf2 ()2 2.82 4002
=
=
= 6.119
(g)
E
206 000

1 / n
1 / 0.11
=
=
+2

+2

E
2K
206 000
2 1750
This gives the changes of stress and strain at the stress concentration as
= 1116 MPa and = 0.005479

(h)
(i,j)

Comment: One notices that 2.8 400 = 1120 MPa, which is the change one
would have obtained if the material were fully elastic. Here we have got the
change = 1116 MPa in (i), which implies that almost all deformation at the
notch is elastic at this change of the remote stress. The second term on the
right hand side of (h) is thus, in this case, much smaller than the first term. The
reason why we have such a large range for the elastic deformation here is that
the material is yielding in tension when the change of load is applied, and from
that starting point the elastic range is twice the yield limit of the material until
yielding starts at compression.
The lower end point of the hysteresis loop may now be determined. One
obtains
min = max = 1077 1116 MPa = 39 MPa and
(k)
min = max = 0.017319 0.005479 = 0.01184

(l)

The mean value and the amplitude of the stress and the strain at the notch can
now be calculated. One obtains

Page 9:8

Chapter 9

mean =

max + min 1077 39 MPA

=
= 519 MPa
2
2

(m)

a =

max min 1077 + 39 MPA

=
= 558 MPa
2
2

(n)

mean =
a =

max + min 0.017319 + 0.01184

=
= 0.01450
2
2

(o)

max min 0.017319 0.01184

=
= 0.0027395
2
2

(p)

The fatigue life N is obtained from the Morrow expression:

f mean
a =
(2N) b + f(2N) c
E
which gives
1400 519
0.0027395 =
(2N) 0.10 + 0.60(2N) 0.55
206 000
From this, N = 33 400 cycles is solved.

(q)
(r)

Thus, the expected number of cycles to fatigue failure is N = 33 400 cycles.

Answer: Failure is expected after N = 33 400 cycles (approximately).

9/4.
A structure has a notch with the fatigue notch factor Kf. The structure is
designed for cyclic loading, and (maximum) 10 000 loading cycles is expected
during the life of the structure. The remote loading (the loading far away from
the notch) is = 250 MPa. Determine the highest value of the fatigue notch
factor Kf that can be allowed.
Material data: The material can be considered linear elastic, ideally plastic with
modulus of elasticity E = 210 GPa and yield strength Y = 350 MPa. The
ultimate strength of the material is U = 380 MPa and the ductility D = 1.05.
Solution:
Use the Coffin-Manson rule to calculate the maximum strain amplitude
allowed. It gives
U
a = 1.75 N 0.12 + 0.5 D 0.6 N 0.6
(a)
E
Numerical values, as given above, give
Chapter 9

Page 9:9

a = 1.75

380
10 000 0.12 + 0.5 1.05 0.6 10 000 0.6
210000

= 0.0030983
(b)
The stress at the notch can not be higher than Y = 350 MPa. (In fact, it could
be slightly higher than 350 MPa, as the ultimate strength of the material is 380
MPa. Thus, some deformation hardening could take place in the material, but
this is not taken into account here.) The Neuber hyperbola
Kf2 2
=
E

(c)

gives, with max = Y = 350 MPa,

Kf2 2502
350 0.003098 =
(d)
210 000
From this the maximum value of the fatige notch factor is solved. One obtains
Kf = 1.91
(e)

Answer: Maximum allowable value of the fatige notch factor is Kf = 1.91.

9/5.
A structure has a notch with the fatigue notch factor Kf. The structure is
designed for cyclic loading, and (maximum) 10 000 loading cycles is expected
during the life of the structure. The remote loading (the loading far away from
the notch) is = 200 MPa. Determine the highest value of the fatigue notch
factor Kf that can be allowed.
stress

E
0.001667

strain
0.005

Material data: The material behaviour at cyclic

loading may be regarded as linear elastic, deformation hardening (see figure). The modulus of
elasticity is E = 210 GPa, the yield strength Y
= 350 MPa, the ultimate strength of the material
is U = 380 MPa, at which stress the strain (at
failure) is 0.005. The ductility is D = 1.05.

Solution:
Use the Coffin-Manson rule to calculate the maximum strain amplitude
allowed. It gives
U
a = 1.75 N 0.12 + 0.5 D 0.6 N 0.6
E
Page 9:10

Chapter 9

(a)

Numerical values, as given above, give

380
a = 1.75
10 000 0.12 + 0.5 1.05 0.6 10 000 0.6
210000
= 0.0030983
(b)
This strain is larger than the strain 0.001667 (the elastic limit, see figure), and
less than 0.005, so it is concluded that the Neuber hyperbola must intersect the
strain-hardening branch of the stress-strain curve given. The equation of this
branch is
= 335 + 9000 (stress in MPa)
Enter the strain = 0.003098 into this equation. It gives stress = 363 MPa.
Thus, the maximum stress = 363 MPa can be allowed at the notch.
Kf2 2
=
E

The Neuber hyperbola

(c)

gives, with max = 363 MPa,

Kf2 2002
(d)
363 0.003098 =
210 000
From this the maximum allowable value of the fatige notch factor is solved.
One obtains
Kf = 2.43
(e)

Answer: The largest fatigue notch factor that can be allowed is Kf = 2.43.
9/6.
A flat bar of a linear elastic, ideally plastic material with a rectangular cross
section has a small circular hole at its centre axis. The stress concentration
factor at the hole is Kt = 3.0 and the fatigue notch factor is Kf = 2.8.

stress
Y

strain

The bar is subjected to an alternating stress

= 200 MPa.
Use the Neuber method to estimate the stress
and the stain at the hole, and then use the
Morrow equation to estimate the number of
cycles to fatigue failure. Note that the material
is assumed to be ideally plastic also at cyclic
loading, i.e. no deformation hardening or
softening is present. The modulus of elasticity is
E = 200 GPa and the yield strength is Y = 400
MPa. Further, f= 1200 MPa, f= 1.0, b =
0.1 and c = 0.62.
Chapter 9

Page 9:11

Solution:
Due to the high stresses at the hole, the material will yield locally. The stress
concentration factor K and the strain concentration factor K may be written
max
max max
K =
and K =
=
(a,b)

/ E
(Hookes law is valid for stresses below 400 MPa, thus = / E )
The Neuber hyperbola becomes
Kf22
=
E

(c)

The material data given and the nominal stress amplitude = 200 MPa now
determine the Neuber hyperbola. The local (maximum) stress and strain
amplitudes a and a, respectively, at the hole are obtained as the intersection
of the Neuber hyperbola with the material stress-strain relation. One obtains

K 22 2.82 2002
a a = f =
= 1.568
(d,e)

E
200 000

a is unknown, but a = Y = 400 MPa

This gives a = max = 400 MPa and, from (d), a = max = 0.00392.
Now the number of cycles to fatigue failure may be determined. Using the
mean stress m = 0, Morrows formula gives
f m
1200
(f)
( 2N )b + f ( 2N )c =
( 2N ) 0.1 + 1.0 ( 2N ) 0.62
E
200 000
Using a = 0.00392, the fatigue life N = 13800 (or 13827) cycles is obtained.
a =

Answer: The fatigue life N = 13 800 cycles is expected.

9/7.
A test specimen with a notch (stress concentration factor Kt = 3.0 and fatigue
notch factor Kf = 2.5) is subjected to a remote stress = 300 MPa. The
material is assumed to be linear elastic, ideally plastic (also at cyclic loading)
with modulus of elasticity E = 210 GPa, and yield limit Y = 650 MPa, see
figure.
stress
Y

strain

Page 9:12

Determine the expected number of loading

cycles to fatigue failure of the specimen.
Data: Cross-section reduction at fracture is
= 65 per cent and fracture strength is U = 650
MPa.
Chapter 9

Solution:
The Neuber hyperbola gives (for amplitudes)
(Kf )2
a a =
(a)
where = 300 MPa
E
As the material is ideally plastic, it is concluded that the stress cannot exceed
Y = 650 MPa at the point where the stress concentration appears. Using this
(i.e. a = Y = 650 MPa) in equation (a), one obtains
2
1 (Kf )
1 (2.5 300)2
=
= 0.00412
(b)
a =
a E
650 210 000
The Coffin-Manson relationship gives, with a = 0.00412,
650
N 0.12 + D 0.6N 0.6
= 2 a = 2 0.00412 = 3.5
210 000
Where D = ln[1/(1 )] = 1.04982.
Solving (c) for N gives N = 8330 cycles.

(c)

Answer: Fatigue failure is expected after 8300 cycles, approximately.

9/8.
A flat bar of a linear elastic, deformation hardening plastic material has a
rectangular cross section. A small circular hole has been drilled in the bar at its
centre axis. The stress concentration factor at the hole is Kt = 3.0 and the
fatigue notch factor is Kf = 2.8.

stress
Y

The bar is subjected to an alternating remote

stress
= 200 MPa.
Estimate the number of cycles to fatigue failure
by use of Neubers method and Morrows
equation. Note that the material is assumed to
be linearly elastic, linearly deformation
k = E /10
hardening also at cyclic loading as shown in the
figure. Use modulus of elasticity E = 200 GPa,
strain slope k = E /10, and the yield strength = 400
Y
MPa. Further, f= 1200 MPa, f= 1.0, b =
0.1, and c = 0.62.

Chapter 9

Page 9:13

Solution:
Due to the high stresses near the hole, the material will yield locally. The
stress concentration factor K and the strain concentration factor K may be
written
K =

max

and K =

max max
=
/ E

(Hookes law is valid for stresses below 400 MPa; thus = / E in (b).)
The Neuber hyperbola becomes
Kf22
=
E

(a,b)

(c)

The material data given and the nominal stress amplitude = 200 MPa now
determine the Neuber hyperbola. The local stress and strain amplitudes a and
a, respectively, at the hole are obtained as the intersection of the Neuber
hyperbola and the material stress-strain relationship given in the figure. One
obtains

Kf22 2.82 2002

a a =
=
= 1.568
(d,e)

E
200 000

a = 360 + k a for a > Y = 400 MPa

This gives a = max = 432.5 MPa and, from (d), a = max = 0.003625.
Now the number of cycles to fatigue failure may be determined. According to
Morrow, and by use of mean stress m = 0, one obtains
f m
1200
(f)
( 2N )b + f ( 2N )c =
( 2N ) 0.1 + 1.0 ( 2N ) 0.62
a =
E
200 000
Using a = 0.003625, the fatigue life N = 17575 (or 17600) cycles is obtained.
Answer: The fatigue life N = 17 600 cycles (approximately) is expected.

9/9.
A test specimen is loaded with a repeated stress sequence with a nominal stress
according to the figure. The test specimen contains a notch with stress
concentration factor Kt = 2.7 and fatigue notch factor Kf = 2.5. Assume that the
material follows the cyclic stress-strain curve
a a 1/n

1/n
and at changes: =

a = +
+2

E K

E
2K

Page 9:14

Chapter 9

(a) Determine the stresses and strains at the notch needed for a fatigue life
analysis (i.e. determine the mean values of the stress cycles and the
amplitudes of the strain cycles at the notch). Use the rain-flow count
method for the cycle counting.
(b) Determine the expected fatigue life (the expected number of loading
sequences before failure) of the test specimen.
The material has the following cyclic
properties:
E = 206 GPa, K = 1750 MPa, n = 0.11,
Y = 850 MPa, f = 1500 MPa,
f = 0.60, b = 0.10, and c = 0.55.

Stress (MPa)

600
400
200
0

time

Solution:
(a) Stresses and strains at the notch
The mean value and the amplitude of the stress and strain at the notch are
obtained from the turning points of the hysteresis loops. Determine these
points.
The stress and strain at the notch, when = 600 MPa, are obtained from the
relationships (the Neuber hyperbola and the cyclic material relation for
amplitudes):
2
2

K 2( )2
= f = 2.5 600 = 10.922
a a
E
206 000

(a,b)
1
/
n

1 / 0.11

a a
a

a
= +

=
+
a
E K
206 000 1750

It is assumed that after a number of loading sequences, the stress and the strain
stabilize as given by the Ramberg-Osgoods stress-strain relation even if the
loading is not alternating here. Therefore, the remote stress 600 MPa is used as
an amplitude when the upper turning point of the hysteresis loop is determined.
From (a,b) the upper turning point of the hysteresis loop is found. This point is
marked (1) in the figure to the right below. One obtains
1 = 999 MPa and 1 = 0.010943
Thus, each time the remote stress reaches the level 600 MPa (points (A)
and (C) in the figure to the left), the stress and the strain at the notch become
1 = 999 MPa and 1 = 0.010943, respectively (point (1) in the figure to the
right).
Chapter 9

Page 9:15

Stress (MPa)

600

Stress (MPa)
(A) (C)

200

(A)

(E) (E)

400

(1) (3)

1000
(5)

500
0

(B)

(2)

time

0
(D) (D) (D)

Remote (nominal) stress giving

stress concentration at the notch

-500

0.010

(4)

Stress and strain at the notch

A change = 400 MPa of the remote stress, from level (A) to level (B) in
the figure, gives
2
2

(
)
K
2.52 4002
f

=
=
= 4.854

E
206 000

1/n

1 / 0.11
= + 2
=
+2

E
2 K
206 000
2 1750

This gives the stress and strain changes at the notch:

(c,d)

= 998 MPa and = 0.004865

The stress and the stain at the notch, when = 200 MPa at (B), become
2 = 1 = 999 998 MPa = 1 MPa and
2 = 1 = 0.010943 0.004865 = 0.006078
This point has been marked as point (2) in the stress-strain diagram.
After this the loading stress turns upwards the amount = 400 MPa up
to stress level = 600 MPa, see point (C). The stress at the notch then
increases = 998 MPa from 2 to 3 = 1 MPa + 998 MPa = 999 MPa = 1,
and the strain increases from 2 to 3 = 0.006078 0.004865 = 0.010943 = 1,
giving point (3) in the diagram. Point (3) coincides with point (1).
A stress change = 600 MPa of the remote stress, from level (C) to level
(D), gives

Page 9:16

Chapter 9

2
2
2
2

= Kf () = 2.5 600 = 10.922

E
206 000

1/n

1 / 0.11
= + 2
=
+2

E
2 K
206 000
2 1750

(e,f)

This gives the stress and strain changes at the notch:

= 1438 MPa and = 0.007596
The stress and the strain at the notch when = 0 ( is situated at point (D))
become (see point (4) in the figure)
4 = 3 = 999 1438 MPa = 439 MPa and
4 = 3 = 0.010943 0.007596 = 0.003347
After this two loops with stress changes = 400 MPa follow. At next
maximum, i.e. when has come to level (E), the stress and the strain at the
notch become (see point (5) in the figure)
5 = 4 + = 439 + 998 MPa = 559 MPa and
5 = 4 + = 0.003347 + 0.004865 = 0.008212
The final stress change, from level (D) back to (A), gives the branch from (4)
back to (1) in the hysteresis loop.
To summarize, these values give the following stresses and strains at the notch
(not all these values were asked for in the problem):
No of
cycles

min

max

mean

999

500

499 0.006078 0.010943 0.008510 0.002433

1 439

999

280

719 0.003347 0.010943 0.007145 0.003798

2 439

559

60

499 0.003347 0.008212 0.005780 0.002433

ampl

min

max

mean

ampl

Thus, at the notch the following is obtained:

- one cycle between the points (1) and (2) in the stress-strain diagram,
corresponding to cycle (A)-(B)-(C) for the remote stress,
- one cycle between the points (3) and (4) in the figure, corresponding to cycle
(C)-(D)-(A) for the remote stress, and
- two cycles between the points (4) and (5) in the figure, corresponding to
cycles (D)-(E)-(D)-(E)-(D) for the remote stress.
Chapter 9

Page 9:17

(b) Fatige life

The fatigue life Ni at each strain amplitude, with the corresponding mean value
of the stress, is obtained from the Morrow relationship
f mi
ai =
(g)
(2 Ni )b + f (2 Ni )c
E
which gives
0.002433 =

1500 500
(2 N1) 0.10 + 0.60 (2 N1) 0.55
206 000

giving

N1 = 63 650

0.003798 =

1500 280
(2 N2) 0.10 + 0.60 (2 N2) 0.55
206 000

giving

N2 = 20 355

0.002433 =

1500 60
(2 N3) 0.10 + 0.60 (2 N3) 0.55
206 000

giving

N3 = 195 400

The accumulated damage from one sequence is (use the Palmgren-Miner

damage accumulation rule)
1
2
1
1
+
+
=
(h)
D=
63 650 20 355 195 400 13 320
From this, S = 1/D = 13 320 sequences to fatigue failure is obtained.
Answer: One expects S = 1/D = 13 300 sequences to fatigue failure.

Page 9:18

Chapter 9

Extra problem, a version of 9/1.

9/1x.
A flat bar of material SAE 1045 with a
rectangular cross section has a small circular
hole at its centre axis. The bar is subjected to an
alternating stress = 300 MPa.
(a) Estimate the number of cycles to fatigue failure by use of Neubers method.
Assume that Kf = 0.9Kt.
(b) Draw the hysteresis loop for = 300 MPa.
(c) Determine the fatigue life if = 150 150 MPa.
Solution:
(a) Number of cycles to fatigue crack initiation
If the stress state at the small hole in the flat bar was purely elastic, then the
stress concentration factor would have been Kt = 3. The fatigue notch factor Kf
is
Kf = 1 + q (Kt 1)

(a)

No information for calculation of the notch sensitivity factor q is given here.

Instead, Kf = 0.9Kt = 2.7 was given. Thus, use Kf = 2.7.
Due to the high stresses close to the hole, the material will yield locally.
Taking this into consideration, the stress concentration factor K and the strain
concentration factor K may be determined from equations (9.9a,b). If the
second term in the material relation (9.2a,b) is disregarded (for stresses far
away from the hole the second term is supposed to be small as compared with
the first term) one obtains (9.9a,b) as
max
max max
K =
and K =
=
(b,c)

/ E
The Neuber hyperbola becomes
Kf22
=
E

(d)

The material data given earlier for the material SAE 1045 and the nominal
stress amplitude = 300 MPa now determine the Neuber hyperbola. The
local stress amplitude a and strain amplitude a at the hole are obtained as the

Chapter 9

Page 9:19

intersection point between the Neuber hyperbola and the material stress-strain
relation (for cyclic loading). One obtains

Kf2 2 2.72 3002

=
=
= 3.2805
a a
E
200 000

(e,f)
1/ 0.18

a a 1/n
a

a
= + =

+
a
E K
200 000 1344

This system of equations gives a = max = 499 MPa and a = max = 0.006574.
The stress and strain concentration factors K and K then become
K =

max 499
max
0.006574
=
= 1.663 and K =
=
= 4.3828
300
/ E 300/ 200 000

(g,h)

(Verify: K K = 1.663 4.383 = 7.29 = 2.72 = Kf2, as it should.)

Now the number of cycles to fatigue failure may be determined. According to
Morrow, (9.4) and (9.7), and by use of mean stress m = 0, one obtains
f m
1227
a =
(j)
( 2N )b + f ( 2N )c =
( 2N ) 0.095 + 1.0 ( 2N ) 0.66
E
200 000
Using a = 0.006574, the fatigue life 2N = 4600 reversals to failure is obtained,
giving N = 2300 cycles to fatigue failure.
(b) Display the hysteresis loop for the material at the notch, when the remote
loading is = 300 MPa
According to problem (a) above the upper turning point of the hysteresis loop
is situated at = max = 499 MPa and = max = 0.006574, see point A in the
figure below. The lower turning point is situated at = max = 499 MPa
and = max = 0.006574, see point B in the figure.
Choose the point A as starting point of the hysteresis loop, see figure below. A
nominal change of stress (far away from the stress concentration) will
cause a change of stress and a change of strain at the stress
concentration. The changes and are obtained by use of the Neuber
hyperbola and the material relation for changes of stress and strain. Using
= 300 MPa one obtains
2
2
2
2

= Kf () = 2.7 300 = 3.2805

E
200 000

(k,l)
1/n
1/ 0.18

=
+2
=
+2

E
2K
200 000
2 1344
Page 9:20

Chapter 9

From this system of equations = 0.004662 and = 702 MPa are solved.
The lower branch of the hysteresis loop, starting at point A, will pass through
the point C. The co-ordinates of point C are obtained from
C = A = 499 MPa 702 MPa = 203 MPa

(m,n)
C = A = 0.006574 0.004662 = 0.001912
Verify that the change of stress = 600 MPa will take this branch of the
hysteresis loop to the point B.
499 MPa
- 0.001912
203 MPa

- 0.006574
0.006574
C
B

0.001912
- 203 MPa

- 499 MPa

At a change of stress = 300 MPa the upper

branch of the hysteresis loop, starting at point B,
will pass through point D. The coordinates of
point D are obtained from
D = B + = 499 MPa + 702 MPa = 203 MPa

D = B + = 0.006574 + 0.004662

= 0.001912
(o,p)

By use of these values (and perhaps some more)

the hysteresis loop may now be drawn, see the
figure.

Comment
It is noticed that when the loop passes the point C the external loading (far
away from the hole) will give a nominal stress equal to zero, i.e. the bar has
been loaded (a number of times) up to stress = 300 MPa and then unloaded
by the stress = 300 MPa so that the total stress far away from the stress
concentration will be = 0. At the stress concentration, however, the residual
stress C and a residual strain C will remain.
(c) Fatigue life when = 150 150 MPa
The same bar as above (Kf = 2.7) will be investigated, but now the bar is
loaded to a nominal stress = 150 150 MPa. The nominal (remote) stress
will now vary between 0 and 300 MPa. This implies that the stress at the
hole will vary between C and A according to problem (b).

Chapter 9

Page 9:21

As an extra exercise the hysteresis loop between

A
A and C will be drawn. Select, once again, the
starting point of the loop at point A where A =
- 0.001912
499 MPa and A = 0.006574. Determine one
203 MPa
D
more point on the loop. Here = 200 MPa
- 0.006574
will be chosen. It gives, as in (j,k) above,
E 0.006574
2
2
2
2

= Kf () = 2.7 200 = 1.458

E
200 000
(q,r)
0.001912
1/ 0.18
C

F
- 203 MPa =
+2

200 000
2 1344

499 MPa

- 499 MPa

This gives = 519 MPa and = 0.002810. The point E on the hysteresis
loop is now obtained from
E = A = 499 MPa 519 MPa = 20 MPa

(s,t)
E = A = 0.006574 0.002810 = 0.003764
The point F on the loop (branch starting at C, see the figure) gets the
coordinates
F = C + = 203 MPa + 519 MPa = 316 MPa

(u,v)
F = C + = 0.001912 + 0.002810 = 0.004722
The number of loading cycles to fatigue failure is obtained, according to
Morrow, from
f m
(w)
( 2N )b + f ( 2N )c
a =
E
The mean value m of the stress becomes m = (A + C) / 2 = (499 203) / 2 =
148 MPa, and the strain amplitude will be a = (A C) / 2 = (0.006574
0.001912) / 2 = 0.002331. By use of these values in (v), and by use of the
given parameter values, one obtains
1227 148
a = 0.002331 =
(x)
( 2N ) 0.095 + 1.0 ( 2N ) 0.66
200 000
From this 2N = 95 600 reversals, giving N = 47 800 cycles, to fatigue failure is
solved. (This result may be compared with the result N = 6 500 cycles to
fatigue failure obtained when = 200 200 MPa, obtained in Section 9.3.1.)
Answer: (Compare with the solution given in example above)
(a) Number of cycles N to crack initiation is 2300, approximately,
(b) stress and strain at hole is a = 499 MPa and a = 0.006574, respectively
(c) Nf = 47 800 cycles (according to Morrow).
Page 9:22

Chapter 9