Professional Documents
Culture Documents
5 Stress Analysis and Pressure Vessels PDF
5 Stress Analysis and Pressure Vessels PDF
http://www.cheng.cam.ac.uk/research/groups/laser/
Synopsis
1
Thermal Effects
Torsion.
1.1
1.2
1.3
2.1
2.2
2.3
2.4
3.1
3.2
3.3
4.1
4.2
4.3
5.1
5.2
5.3
5.4
6.1
6.2
7.1
7.2
7.3
7.4
Supporting Materials
There is one Examples paper supporting these lectures.
Two good textbooks for further explanation, worked examples and exercises are
Mechanics of Materials (1997) Gere & Timoshenko, publ. ITP [ISBN 0-534-93429-3]
Mechanics of Solids (1989) Fenner, publ. Blackwell [ISBN 0-632-02018-0]
This material was taught in the CET I (Old Regulations) Structures lecture unit and was examined
in CET I (OR) Paper IV Section 1. There are consequently a large number of old Tripos questions
in existence, which are of the appropriate standard. From 1999 onwards the course was taught in
CET1, paper 5. Chapters 7 and 8 in Gere and Timoshenko contain a large number of example
problems and questions.
Nomenclature
The following symbols will be used as consistently as possible in the lectures.
E
G
I
J
R
t
T
Youngs modulus
Shear modulus
second moment of area
polar moment of area
radius
thickness
thermal expansivity
linear strain
shear strain
angle
Poissons ratio
Normal stress
Shear stress
Ongoing Example
We shall refer back to this example of a typical pressure vessel on several
occasions.
Distillation column
2m
P = 7 bara
carbon steel
t = 5 mm
18 m
External diameter
D
r
L
h
wall thickness, t
r
L
L
h
Force equilibrium
D2
P = D t L
4
if P > 0, then L is tensile
L =
PD
4t
Force balance, D L P = 2 h L t
h =
SAPV LT 2005
CFK, MRM
PD
2t
c, Radial stress
r o ( P )
D
h , L P (
).
2t
thin walled, so D >> t
so h , L >> r
so neglect r
Compare terms
SAPV LT 2005
CFK, MRM
D2
P
= h D t
4
PD
h =
4t
bulk =
W
= Y
Dt
2 EI
W= 2
L
Consider a cylindrical column. I = R3t so the compressive stress required
to cause buckling is
buckle
W
2 ED3t 1
2 ED 2
=
=
=
2
Dt
8L
Dt
8L2
buckle
2 E
=
2
8( L D)
or
SAPV LT 2005
CFK, MRM
where L/D is a slenderness ratio. The mode of failure thus depends on the
geometry:
stress
Bulk yield
Short
Long
L /D ratio
SAPV LT 2005
CFK, MRM
Vacuum vessels.
Cylindrical pressure vessels subject to external pressure are subject to compressive hoop stresses
PD
2t
Consider a length L of vessel , the compressive hoop force is given by,
h =
P D L
2
If this force is large enough it will cause buckling.
h L t =
length
SAPV LT 2005
CFK, MRM
I=
P D L 4 2 EI
=
2
( D )2
b t3 L t3
=
12
12
SAPV LT 2005
CFK, MRM
p buckle
2E t
3 D
X
d
W
There is a concentration of stress at the rod surface below XX and this value
should thus be considered when we consider failure mechanisms.
The ratio of the maximum local stress to the mean (or apparent) stress is described by a stress concentration factor K
K=
max
mean
The values of K for many geometries are available in the literature, including
that of cracks. The mechanism of fast fracture involves the concentration of
tensile stresses at a crack root, and gives the failure criterion for a crack of
length a
a = Kc
where Kc is the material fracture toughness. Tensile
stresses can thus cause failure due to bulk yielding or due
to cracking.
SAPV LT 2005
CFK, MRM
crack =
Kc 1
stress
failure locus
length of crack. a
SAPV LT 2005
CFK, MRM
2.
= E
failure
Yield
Stress
Elastic
Limit
up to a limit, known as the yield stress (stress axis) or the elastic limit (strain
axis). Below these limits, deformation is reversible and the material eventually
returns to its original shape. Above these limits, the material behaviour depends
on its nature.
Consider a sample of material subjected to a tensile force F.
2
F
1
3
1 = (1 F / A ) / E
10
2 =
1
E
;3 =
1
E
where is the Poisson ratio for that material. These effects are additive, so for
three mutually perpendicular stresses 1, 2, 3;
2
1
3
Giving
1 =
1
E
2 =
3 =
1
E
1
E
3
E
3
E
3
E
Values of the material constants in the Data Book give orders of magnitudes of
these parameters for different materials;
Material
Steel
Aluminum alloy
Brass
E
(x109 N/m2)
210
70
105
11
0.30
0.33
0.35
Puniform = K v
i.e. the volumetric strain resulting from the application of a uniform pressure. In
the case of a pressure causing expansion
so
1 = 2 = 3 = P
P
1
1 2 3 ] =
(1 2 )
[
E
E
3P
v = 1 + 2 + 3 =
(1 2 )
E
E
K=
3(1 2 )
1 = 2 = 3 =
= G
- shear strain
12
Cylindrical vessel:
Longitudinal strain
L =
PD
L
(1 - 2 ) =
4tE
L
Hoop strain:
h =
n
E
R
D
PD
=
(2 - ) =
R
D
4tE
radial strain
r =
t
3PD
1
=
r - h - L ] = [
t
4ET
E
13
[ONGOING EXAMPLE]:
L =
1
( L - n )
E
1
60 x 10 6 - (0.3)120 x 10 6
9
210 x 10
h = 0.486 millistrain
r = -0.257 millistrain
Thus: pressurise the vessel to 6 bar: L and D increase: t decreases
Volume expansion
Cylindrical volume:
D 2
Vo = o Lo
4
New volume
V =
(Do
4
(original)
2
+ D) (Lo + L)
L Do
2
1 + h ] [1 + L ]
= o
[
4
V
Define volumetric strain v =
V
V - Vo
2
v =
= (1 + h ) (1 + L ) - 1
Vo
= 1 + 2 h + h2 (1 + L ) - 1
v = 2 h + L + h2 + 2 h L + L n2
Magnitude inspection:
14
max (steel) =
190 x 10
3
=
small
9 = 0.905 x 10
210 x 10
Spherical volume:
h =
so
1
[ h - L - r ] = PD (1 - )
4Et
E
(Do + D )3 - Do3
v =
6
D 3o 6
= (1 + h)3 1 3h + 0(2)
(c)
General result
v = 1 + 2 + 3
ii are the strains in any three mutually perpendicular directions.
L = 0.114 mstrain
n = 0.486 mstrain
rr = -0.257 mstrain
v = 2n + L
new volume = Vo (1 + v)
Increase in volume =
D2 L
4
= 0.129 L
A=area
work done
L0
x
2
kx1
1
x1
x1
w = o Fdx = o k x dx =
=
Fx
2
2 1 1
x1 = Lo 1
AL o 11
W =
F1 = A 1
2
Al o 1
2 Al o 1 1
U =
1 1
2
ALo 1 1
2(ALo )
12
2E
1
[11 + 22 + 33]
2
3
2
etc
Now 1 = 1 E
E
E
In a 3-D system, U =
So U =
1
12 + 2 2 + 32 - 2 ( 1 2 + 2 3 + 3 1)
2E
3P
P
U =
(1 - 2 ) =
2E
2K
16
P 6 barg
V = 61 x 10-3 m3
Increasing the pressure compresses the contents normally test with water.
5
6 x 10
P
= - 0.273 mstrains
=
V water? v (water ) =
K
2.2 x 109
decrease in volume of water = -Vo (0.273) = -15.4 x 10 3 m3
Thus we can add more water:
Extra space = 61 + 15.4 (L)
= 76.4 L water
extra space
p=0
p=6
17
3. Thermal Effects
3.1. Coefficient of Thermal Expansion
= LT Linear
Volume
T = 10oC
L = 11 10-5
T = 500oC
Consider a steel bar mounted between rigid supports which exert stress
Heat
= T -
If rigid: = 0
so
= ET
(i.e., non buckling)
steel:
18
.
steel construction
L = 3 m . full of water t = 3 mm
D=1m
Initially un pressurised full of water: increase temp. by T: pressure rises
to Vessel P.
The Vessel
L =
PD
= 83.3 P
4t
n = 2L = 166.7 P
19
Strain (volume)
L =
= 210 x 10
= 11 x 10 6
0.3
L h
+ l T
E
E
Similarly
h = 6.75 x 10-10 P + 11 10 6 T
v = L + 2n = 15.08 x 10-10 P + 33 x 10-6 T = vessel vol. Strain
Expands
due to T
increase
Contracts
due to P
increase:
v, H2O = vT P/K
P = 13750 T
21
temperature
controllers
a=
4 mm (2 + 2 mm)
b=
10 mm
a
d
Cu
Fe
L = 100mm
Heat by T:
22
Cu
Fe
Cu
Fe
can then calculate the shearing forces [and hence the BM].
Shearing force F
compressive in Cu
Tensile
in Fe
F
F
= Fe T +
bdE cu
bdE Fe
Equating strains:
cu T -
So
1
F 1
+
= ( cu - Fe ) T
bd E cu E Fe
23
bd = 2 x 10-5 m2
Ecu = 109 GPa
cu 17 x 10-6 k-1
T = 30C
Fe = 11 x 10-6 K-1
F = 387 N
(significant force)
24
Fe
increase T
Data: cu > Fe
Balance forces:
Tensile force in steel
|FFe| = |Fcu| = F
Stress in steel
= F/AFe = Fe
= F/Acu = cu
copper
Steel strain:
FE = Fe T + Fe/EFe
= FeT + F/EFeAFe
copper strain
cu = cuT F/EcuAcu
25
Strains EQUAL:
1
1
F
+
=
Acu Ecu
Fe E Fe
1A4
442444
3
( cu - Fe ) T
d
sum of strengths
26
= 1 + 2
[RADIANS]
Hookes Law
=G
2 C
G shear modulus =
27
E
2(1 + )
2r
Gr
L
28
r
L
dr
.T = 2 r.r r
T = 2 r 2 dr
or
cf
r.dA
A
G
L
G
{J}
L
dA
so
T G
=
=
J
L
r
M E
= =
I R y
29
Gr
J = 2 .r r 2 dr =
R4
y
r
J=
D 4
32
Now
r2 = x2 + y2
It can be shown that J = Ixx + Iyy
30
Rectangular rod:
I xx
I yy
bd 3
=
12
3
db
=
12
J=
bd 2
b +d2
12
31
D = 25 mm
L = 1.5 m
Failure when = y = 95 MPa
G = 81 GPa
max
rmax
Now
J=
D 4
32
T 95 x 10 6
= =
J
0.0125
so T = 291 Nm
= 383 x 10 8 m 4
G T
81 x 109
9
= 7.6 x 10 =
From
L
J
1.5
32
291 x
45 kW
2
x 1450
60
0.025m
y
rmax
6
32
T 95 x 10 x 2 291
=
=
4
D
D - 0.0254
J
D 4.15 cm
33
T
=
r J
now cylinder
J=
so
2
D
[(D + 2t )
32
- D4
[8D t + 24 D t
32
D 3t
4T
D 3 t
2T
D 2 t
34
2 2
+ ...
CET 1, SAPV
Vectors
tensors of rank 1
r
r
F = ma
hence :
F1 = ma1
F2 = ma2
F3 = ma3
or :
Fi = mai
Tensors of rank 2
3
pi = Tij q j
i, j = 1,2,3
j =1
or :
p1 = T11q1 + T12 q2 + T13 q3
p2 = T21q1 + T22 q2 + T23 q3
p3 = T31q1 + T32 q2 + T33 q3
Axis transformations
The choice of axes in the description of an engineering problem is
arbitrary (as long as you choose orthogonal sets of axes!). Obviously the
physics of the problem must not depend on the choice of axis. For
example, whether a pressure vessel will explode can not depend on how
we set up our co-ordinate axes to describe the stresses acting on the
34
CET 1, SAPV
vessel. However it is clear that the components of the stress tensor will
be different going from one set of coordinates xi to another xi.
How do we transform one set of co-ordinate axes onto another, keeping
the same origin?
x1
x2
x3
a11
a12
a13
x2 ' a21
x3 ' a31
a22
a32
a23
a33
x1 '
xi ' = aij x j
j =1
Reverse transformation:
3
xi = a ji x j
j =1
xi ' = aij x j
xi = a ji x j
35
CET 1, SAPV
Tensor transformation
How will the components of a tensor change when we go from one coordinate system to another? I.e. if we have a situation where
pi = Tij q j = Tij q j (in short form)
j
where Tij is the tensor in the old co-ordinate frame xi, how do we find the
corresponding tensor Tij in the new co-ordinate frame xi, such that:
pi ' = Tij ' q j ' = Tij ' q j ' (in short form)
j
p' p q q'
Hence:
pi ' = aik pk
pk = Tkl ql
ql = a jl q j '
Thus we have:
36
CET 1, SAPV
pi ' = aik Tkl a jl q j '
For example:
Tij ' = ai1 a jl T1l
+ ai 2 a jl T2l
+ ai 3 a jl T3l
= ai1 a j1 T11 + ai1 a j 2 T12 + ai1 a j 3 T13
+ ai 2 a j1 T21 + ai 2 a j 2 T22 + ai 2 a j 3 T23
+ ai 3 a j1 T31 + ai 3 a j 2 T32 + ai 3 a j 3 T33
Note that there is a difference between a transformation matrix and a 2nd
rank tensor: They are both matrices containing 9 elements (constants)
but:
Symmetrical Tensors:
Tij=Tji
37
CET 1, SAPV
38
CET 1, SAPV
We can always transform a second rank tensor which is symmetrical:
Tij
Tij '
such that :
T1 0
Tij ' = 0 T2
0 0
0
0
T3
Consequence? Consider:
pi = Tij q j
then
p1 = T1 q1 ,
p2 = T2 q2 , p3 = T3 q3
39
CET 1, SAPV
Mohrs circle
Consider an elementary cuboid with edges parallel to the coordinate
directions x,y,z.
y
Fxy
y face
Fx
x
z face
Fxz
Fxx
z
x face
The faces on this cuboid are named according to the directions of their
normals.
There are thus two x-faces, one facing greater values of x, as shown in
Figure 1 and one facing lesser values of x (not shown in the Figure).
On the x-face there will be some force Fx. Since the cuboid is of
infinitesimal size, the force on the opposite side will not differ
significantly.
The force Fx can be divided into its components parallel to the coordinate
directions, Fxx, Fxy, Fxz. Dividing by the area of the x-face gives the
stresses on the x-plane:
xx
xy
xz
It is traditional to write normal stresses as and shear stresses as .
yx , yy , yz
40
CET 1, SAPV
and on the z-face we have:
zx , zy , zz
xx xy
ij = yx yy
zx zy
xz
yz
zz
Note that the first subscript refers to the face on which the stress acts and
the second subscript refers to the direction in which the associated force
acts.
yy
y
x
yx
xy
xx
xx
xy
yx
yy
and therefore:
xx xy xz xx xy xz
ij = yx yy yz = xy yy yz
zx zy zz xz yz zz
Hence ij is symmetric!
41
CET 1, SAPV
This means that there must be some magic co-ordinate frame in which all
the stresses are normal stresses (principal stresses) and in which the off
diagonal stresses (=shear stresses) are 0. So if, in a given situation we
find this frame we can apply all our stress strain relations that we have set
up in the previous lectures (which assumed there were only normal
stresses acting).
Consider a cylindrical vessel subject to shear, and normal stresses (h, l,
r). We are usually interested in shears and stresses which lie in the
plane defined by the vessel walls.
ij
ij '
So stress tensor is symmetric 2nd rank tensor. Imagine we are in the coordinate frame xi where we only have principal stresses:
0
1 0
ij = 0 2 0
0 0 3
42
CET 1, SAPV
The transformation matrix is then:
a11
aij = a21
a
31
a13 cos
a23 = sin
a33 0
a12
a22
a32
sin
cos
0
1
Then:
ij ' = aik a jl kl
cos
= sin
0
sin
cos
0
0 cos
0 sin
1 0
sin
cos
0
1 cos 2 + 2 sin 2
43
0 1 0
0
0 0 2 0
1 0
0 3
0
3
CET 1, SAPV
Hence:
1
1
( 1 + 1 ) + ( 2 1 ) cos 2
2
2
1
( 2 1 ) sin 2
2
44
Shear stresses
y,z plane
x, y plane
normal
stresses
x,z plane
y =
Pyield
A
for steel
(ii)
max
max = 1/2
occurs along plane at 45 to 1
Similarly for tensile test.
Case (ii)
2 < 0 < 1
1 - 2
M.C.
Fails when
= max
= y =
y
2
1 - 2 = y
i.e., when
C =
1
[ + 2 + 3 ]
3 1
UD =
1 2
+
2E 1
1
=
(1
12G
22 + 32 + 2 (1 2 + 2 3 + 3 1 ) 3C 2 + 6C 2
2E
2
- 2 ) + ( 2 - 3 ) + ( 3 - 1 )
M.C.
1
(1 - 2 )2 + (2 + 3 )2 + (3 - 1 )2 > 1 2y + 02 + 2y
12G
12G
{( - )
+ ( 2 + 3 ) + (3 - 1 ) > 2 2y
2
PD
= 50 N / mm 2
4t'
PD
h =
= 100 N / mm 2
2t'
L =
Calculate stresses
and 3 = r 0
Mohr's Circle
(100,)
50
100
s
(50,)
Circle construction
s = 75 N/mm2
t = (252+2)
1,2 = s t
10
Case A
(100,)
100 1
2 50
s
(50,)
Case B
(100,)
100
50
(50,)
We do not know whether the Mohr's circle for this case follows Case A or
B; determine which case applies by trial and error.
Case A; 'minor' principal stress is positive (2 > 0)
Thus failure when
max = 12 y = 105N / mm 2
11
For Case A;
max =
1
2
1
2
[75 +
(252 + 2 )]
2 = 135 252
= 132.7N / mm 2
1 = 210 N / mm2 ; 2 = 60 N / mm2
Giving
Case B;
We now have max as the radius of the original Mohr's circle linking our
stress data.
Thus
= 101.98N / mm2
Principal stresses
1,2 = 75 105
Thus Case B applies and the yield stress is 101.98 N/mm2. The torque
required to cause failure is
T = D2 t' / 2 = 32kNm
Failure will occur along a plane at angle anticlockwise from the y (hoop)
direction;
102
2 = 76.23 ;
75
2 = 90 - 2 = 6.9
tan(2 ) =
12
Example
h = 100N / mm 2
L = 50 N / mm 2
r 0 N / mm2
What torque will cause failure if the yield stress for steel is 210 N/mm2?
Mohr's Circle
(100,)
50
100
s
(50,)
Giving
1 = s + t = 75 + 252 + 2
At failure
UD =
2 = s t = 75 252 + 2
3 = 0
1
( 1 2 ) 2 + ( 2 3 ) 2 + ( 3 1 ) 2 }
{
12G
13
Or
(1 2 ) 2 + ( 2 3 ) 2 + ( 3 1 ) 2 = ( y )2 + (0)2 + (0 y ) 2
4t 2 + (s t)2 + (s + t) 2 = 2 2y
2s2 + 6t 2 = 2 y2
s 2 + 3t 2 = 2y
752 + 3(252 + 2 ) = 210 2
= 110 N / mm 2
D2 t'
T=
= 35kNm
2
which is larger than the value of 32 kNm given by Tresca's criterion - in this
case, Tresca is more conservative.
14
7. Strains
7.1. Direct and Shear Strains
Consider a vector of length lx lying along the x-axis as shown in Figure
1. Let it be subjected to a small strain, so that, if the left hand end is fixed
the right hand end will undergo a small displacement x. This need not be
in the x-direction and so will have components xx in the x-direction and
xy in the y-direction.
xy
xx
lx
Figure 1
xx =
xx
;
lx
xy =
xy
lx
xx is the direct strain, i.e. the fractional increase in length in the direction
of the original vector. xy represents rotation of the vector through the
small angle 1 where,
1 tan 1 =
xy
l x + xx
xy
lx
= xy
yy =
yy
ly
yx =
yx
ly
yx
Figure 2
yy
y
ly
as in Figure 2.
lx
ij =
ij
li
where i, j = 1,2,3
xy = ( 1 + 2 )
A'
xy
B'
C'
Figure 3b
Figure 3a
xy = +( 1 + 2 )
ij = ( ij + ji )
and since
ij = ji,
we have
ij = ji.
Note that the TENSOR SHEAR STRAINS are given by the averaged
sum of shear strains:
1
1
1
1
ij = ( ij + ji ) = ( 1 + 2 ) = ji
2
2
2
2
11
1
ij = y 21
2
1 y
2 31
1
y12
2
22
1
y32
2
1
y13
11
2
1
1
y 23 = y12
2
2
1
33 y13
2
1
y12
2
22
1
y 23
2
1
y13
2
1
y 23
2
33
where the diagonal elements are the stretches or tensile strains and the
off diagonal elements are the tensor shear strains.
Thus our strain tensor is symmetrical, and:
3
ij = ji
This means there must be a co-ordinate transformation, such that:
ij '
ij
such that :
1 0
ij = 0 2
0 0
0
0
ij ' = aik a jl kl
cos
= sin
0
sin
cos
0
0 cos
0 sin
1 0
sin
cos
0
1 cos 2 + 2 sin 2
And hence:
0 1 0
0 0 2
1 0 0
0
0
0
3
1
1
(1 + 1 ) + ( 2 1 ) cos 2
2
2
1
2
1
( 2 1 ) sin 2
2
Note, however, that on this occasion we plot half the shear strain against
the direct strain. This stems from the fact that the engineering shear
strains differs from the tensorial shear strains by a factor of 2 as as
discussed.
45
principal strain
120
/2
B
2
B
radius t
so we can write
circle, centre s,
A = s + t cos(2 )
B = s + t cos(2 + 90) = s t sin( 2 )
C = s t cos(2 )
3 equations in 3 unknowns
Using strain gauges we can find the directions of Principal strains
/2
o
X
o
E
1
E
2
E
(1+ )
qq = 2 =
2
E
1
E
o
(1+ )
E
Y
Q
pp
xx = 0
xy
2
X (0, )
xy
o
E
(1 + )
o
/2
/2
and
2 o
(1+ )
E
But by definition o = G
so
G=
o
E
=
2(1+ )
1 =
1
E
2 =
3 =
1
E
1
E
3
E
3
E
3
E
So using strain gauges you can work out magnitudes of principal strains.
You can then work out magnitudes of principal stresses.
Using Tresca or Von Mises you can then work out whether your
vessel is safe to operate. ie below the yield criteria
10