Elastic Stability Of Columns

Introduction:

Structural members which carry compressive loads may be divided into two broad categories depending on
their relative lengths and cross-sectional dimensions.

Columns:

Short, thick members are generally termed columns and these usually fail by crushing when the yield stress of
the material in compression is exceeded.

Struts:

Long, slender columns are generally termed as struts, they fail by buckling some time before the yield stress in
compression is reached. The buckling occurs owing to one the following reasons.

(a). the strut may not be perfectly straight initially.

(b). the load may not be applied exactly along the axis of the Strut.

(c). one part of the material may yield in compression more readily than others owing to some lack of uniformity
in the material properties through out the strut.

In all the problems considered so far we have assumed that the deformation to be both progressive with
increasing load and simple in form i.e. we assumed that a member in simple tension or compression becomes
progressively longer or shorter but remains straight. Under some circumstances however, our assumptions of
progressive and simple deformation may no longer hold good and the member become unstable. The term
strut and column are widely used, often interchangeably in the context of buckling of slender members.]

At values of load below the buckling load a strut will be in stable equilibrium where the displacement caused by
any lateral disturbance will be totally recovered when the disturbance is removed. At the buckling load the strut
is said to be in a state of neutral equilibrium, and theoretically it should than be possible to gently deflect the
strut into a simple sine wave provided that the amplitude of wave is kept small.

Theoretically, it is possible for struts to achieve a condition of unstable equilibrium with loads exceeding the
buckling load, any slight lateral disturbance then causing failure by buckling, this condition is never achieved in
practice under static load conditions. Buckling occurs immediately at the point where the buckling load is
reached, owing to the reasons stated earlier.

The resistance of any member to bending is determined by its flexural rigidity EI and is The quantity I may be
2
written as I = Ak ,

Where I = area of moment of inertia

A = area of the cross-section

k = radius of gyration.

The load per unit area which the member can withstand is therefore related to k. There will be two principal
moments of inertia, if the least of these is taken then the ratio

Is called the slenderness ratio. It's numerical value indicates whether the member falls into the class of columns
or struts.
Euler's Theory : The struts which fail by buckling can be analyzed by Euler's theory. In the following sections,
different cases of the struts have been analyzed.

Case A: Strut with pinned ends:

Consider an axially loaded strut, shown below, and is subjected to an axial load „P' this load „P' produces a
deflection „y' at a distance „x' from one end.

Assume that the ends are either pin jointed or rounded so that there is no moment at either end.

Assumption:

The strut is assumed to be initially straight, the end load being applied axially through centroid.

In this equation „M' is not a function „x'. Therefore this equation can not be integrated directly as has been done
in the case of deflection of beams by integration method.

Though this equation is in „y' but we can't say at this stage where the deflection would be maximum or
minimum.

So the above differential equation can be arranged in the following form

 Let us define a operator

D = d/dx

2 2 2
(D + n ) y =0 where n = P/EI

This is a second order differential equation which has a solution of the form consisting of complimentary
function and particular integral but for the time being we are interested in the complementary solution only[in
this P.I = 0; since the R.H.S of Diff. equation = 0]

Thus y = A cos (nx) + B sin (nx)

Where A and B are some constants.

Therefore

In order to evaluate the constants A and B let us apply the boundary conditions,

(i) at x = 0; y = 0

(ii) at x = L ; y = 0

Applying the first boundary condition yields A = 0.

Applying the second boundary condition gives

From the above relationship the least value of P which will cause the strut to buckle, and it is called the “ Euler
Crippling Load ” Pe from which w obtain.

The interpretation of the above analysis is that for all the values of the load P, other than those which make sin
nL = 0; the strut will remain perfectly straight since
 y = B sin nL = 0

For the particular value of

Then we say that the strut is in a state of neutral equilibrium, and theoretically any deflection which it suffers will
be maintained. This is subjected to the limitation that „L' remains sensibly constant and in practice slight
increase in load at the critical value will cause the deflection to increase appreciably until the material fails by
yielding.

Further it should be noted that the deflection is not proportional to load, and this applies to all strut problems;
like wise it will be found that the maximum stress is not proportional to load.

The solution chosen of nL =  is just one particular solution; the solutions nL= 2, 3, 5 etc are equally valid
mathematically and they do, infact, produce values of „Pe' which are equally valid for modes of buckling of strut
different from that of a simple bow. Theoretically therefore, there are an infinite number of values of P e , each
corresponding with a different mode of buckling.

The value selected above is so called the fundamental mode value and is the lowest critical load producing the
single bow buckling condition.

The solution nL = 2 produces buckling in two half – waves, 3 in three half-waves etc.
If load is applied sufficiently quickly to the strut, then it is possible to pass through the fundamental mode and to
achieve at least one of the other modes which are theoretically possible. In practical loading situations,
however, this is rarely achieved since the high stress associated with the first critical condition generally
ensures immediate collapse.

struts and columns with other end conditions: Let us consider the struts and columns having different end
conditions

Case b: One end fixed and the other free:

writing down the value of bending moment at the point C

2 2 2
Hence in operator form, the differential equation reduces to ( D + n ) y = n a

The solution of the above equation would consist of complementary solution and particular solution, therefore

ygen = A cos(nx) + sin(nx) + P. I

where

P.I = the P.I is a particular value of y which satisfies the differential equation

Hence yP.I = a

Therefore the complete solution becomes

Y = A cos(nx) + B sin(nx) + a

Now imposing the boundary conditions to evaluate the constants A and B

(i) at x = 0; y = 0

This yields A = -a
 (ii) at x = 0; dy/dx = 0

This yields B = 0

Hence

y = a cos(nx) + a

Futher, at x = L; y = a

Therefore a = - a cos(nx) + a or 0 = cos(nL)

Now the fundamental mode of buckling in this case would be

Case 3

Strut with fixed ends:

Due to the fixed end supports bending moment would also appears at the supports, since this is the property of
the support.

Bending Moment at point C = M – P.y

Thus,

Case 4

One end fixed, the other pinned

 In order to maintain the pin-joint on the horizontal axis of the unloaded strut, it is necessary in this case to
introduce a vertical load F at the pin. The moment of F about the built in end then balances the fixing moment.

With the origin at the built in end, the B,M at C is given as

Also when x = L ; y = 0

Therefore

nL Cos nL = Sin nL or tan nL = nL

The lowest value of nL ( neglecting zero) which satisfies this condition and which therefore produces the
fundamental buckling condition is nL = 4.49radian
 Equivalent Strut Length:

Having derived the results for the buckling load of a strut with pinned ends the Euler loads for other end
conditions may all be written in the same form.

Where L is the equivalent length of the strut and can be related to the actual length of the strut depending on
the end conditions.

The equivalent length is found to be the length of a simple bow(half sine wave) in each of the strut deflection
curves shown. The buckling load for each end condition shown is then readily obtained. The use of equivalent
length is not restricted to the Euler's theory and it will be used in other derivations later.

The critical load for columns with other end conditions can be expressed in terms of the critical load for a
hinged column, which is taken as a fundamental case.

For case(c) see the figure, the column or strut has inflection points at quarter points of its unsupported length.
Since the bending moment is zero at a point of inflection, the freebody diagram would indicates that the middle
half of the fixed ended is equivalent to a hinged column having an effective length L e = L / 2.

The four different cases which we have considered so far are:

(a) Both ends pinned (c) One end fixed, other free

(b) Both ends fixed (d) One end fixed and other pinned
 Comparison of Euler Theory with Experiment results

Limitations of Euler's Theory :

In practice the ideal conditions are never [ i.e. the strut is initially straight and the end load being
applied axially through centroid] reached. There is always some eccentricity and initial curvature present.
These factors needs to be accommodated in the required formula's.

It is realized that, due to the above mentioned imperfections the strut will suffer a deflection which
increases with load and consequently a bending moment is introduced which causes failure before the Euler's
load is reached. Infact failure is by stress rather than by buckling and the deviation from the Euler value is more
marked as the slenderness-ratio l/k is reduced. For values of l/k < 120 approx, the error in applying the Euler
theory is too great to allow of its use. The stress to cause buckling from the Euler formula for the pin ended
strut is

A plot of e versus l / k ratio is shown by the curve ABC.

Allowing for the imperfections of loading and strut, actual values at failure must lie within and below line CBD.

Other formulae have therefore been derived to attempt to obtain closer agreement between the actual failing
load and the predicted value in this particular range of slenderness ratio i.e.l/k=40 to l/k=100.

(a) Straight – line formulae :

The permissible load is given by the formulae

Where the value of index „n' depends on the material used and the end conditions.

(b) Johnson parabolic formulae : The Johnson parabolic formulae is defined as

where the value of index „b' depends on the end conditions.

(c) Rankine Gordon Formulae :

Where Pe = Euler crippling load

Pc = Crushing load or Yield point load in Compression

PR = Actual load to cause failure or Rankine load

Since the Rankine formulae is a combination of the Euler and crushing load for a strut.

For a very short strut Pe is very large hence 1/ P ewould be large so that 1/ P ecan be neglected.
Thus PR = Pc , for very large struts, P e is very small so 1/ P e would be large and 1/ P ccan be neglected ,hence
PR = Pe

The Rankine formulae is therefore valid for extreme values of 1/k.It is also found to be fairly accurate for the
intermediate values in the range under consideration. Thus rewriting the formula in terms of stresses, we have

Where and the value of „a' is found by conducting experiments on various materials. Theoretically,
but having a value normally found by experiment for various materials. This will take into account other types of
end conditions.

Therefore

Typical values of „a' for use in Rankine formulae are given below in table.

Material y orc Value of a

2 Pinned ends Fixed ends

MN/m
Low carbon 315 1/7500 1/30000
steel
Cast Iron 540 1/1600 1/64000
Timber 35 1/3000 1/12000

note a = 4 x (a for fixed ends)

 Since the above values of „a' are not exactly equal to the theoretical values , the Rankine loads for
long struts will not be identical to those estimated by the Euler theory as estimated.

Strut with initial Curvature :

As we know that the true conditions are never realized , but there are always some imperfections. Let
us say that the strut is having some initial curvature. i.e., it is not perfectly straight before loading. The situation
will influence the stability. Let us analyze this effect.

by a differential calculus

Where „ y0' is the value of deflection before the load is applied to the strut when the load is applied to the strut
the deflection increases to a value „y'. Hence

The initial shape of the strut y0 may be assumed circular, parabolic or sinusoidal without making much
difference to the final results, but the most convenient form is
 where C is some constant or here it is amplitude

Which satisfies the end conditions and corresponds to a maximum deviation „C'. Any other shape could be
analyzed into a Fourier series of sine terms. Then

Boundary conditions which are relevant to the problem are

at x = 0 ; y = 0 thus B = 0

Again

when x = l ; y = 0 or x = l / 2 ; dy/dx = 0

the above condition gives B = 0

Therefore the complete solution would be

 Since the BM for a pin ended strut at any point is given as

M = -Py and

Max BM = P ymax

Now in order to define the absolute value in terms of maximum amplitude let us use the symbol as „^'.
 Strut with eccentric load

Let „e' be the eccentricity of the applied end load, and measuring y from the line of action of the load.

Then

2 2 2
or (D + n ) y = 0 where n = P / EI

Therefore ygeneral = ycomplementary

= Asin nx + Bcos nx

applying the boundary conditions then we can determine the constants i.e.

at x = 0 ; y = e thus B = e

at x = l / 2 ; dy / dx = 0

Hence the complete solution becomes

y = A sin(nx) + B cos(nx)

substituting the values of A and B we get

Note that with an eccentric load, the strut deflects for all values of P, and not only for the critical value
as was the case with an axially applied load. The deflection becomes infinite for tan (nl)/2 = ∞ i.e. nl =  giving

the same crippling load . However, due to additional bending moment set up by deflection, the strut
will always fail by compressive stress before Euler load is reached.

Since
 The second term is obviously due the bending action.

Consider a short strut subjected to an eccentrically applied compressive force P at its upper end. If
such a strut is comparatively short and stiff, the deflection due to bending action of the eccentric load will be
neglible compared with eccentricity „e' and the principal of super-imposition applies.

If the strut is assumed to have a plane of symmetry (the xy - plane) and the load P lies in this plane at
the distance „e' from the centroidal axis ox.

Then such a loading may be replaced by its statically equivalent of a centrally applied compressive force „P'
and a couple of moment P.e
 1. The centrally applied load P produces a uniform compressive stress over each cross-section as
shown by the stress diagram.

2. The end moment „M' produces a linearly varying bending stress as shown in the figure.

Then by super-impostion, the total compressive stress in any fibre due to combined bending and compression
becomes,

Distribution of shear stresses in circular Shafts subjected to torsion :

The simple torsion equation is written as

This states that the shearing stress varies directly as the distance „r' from the axis of the shaft and the following
is the stress distribution in the plane of cross section and also the complementary shearing stresses in an axial
plane.

Hence the maximum strear stress occurs on the outer surface of the shaft where r = R

The value of maximum shearing stress in the solid circular shaft can be determined as

From the above relation, following conclusion can be drawn

(i)  maxm  T

(ii)  maxm  1/d

3

Power Transmitted by a shaft:

In practical application, the diameter of the shaft must sometimes be calculated from the power which it is
required to transmit.

Given the power required to be transmitted, speed in rpm „N' Torque T, the formula connecting

These quantities can be derived as follows

Torsional stiffness: The torsional stiffness k is defined as the torque per radian twist .
 For a ductile material, the plastic flow begins first in the outer surface. For a material which is weaker in shear
longitudinally than transversely – for instance a wooden shaft, with the fibres parallel to axis the first cracks will
be produced by the shearing stresses acting in the axial section and they will upper on the surface of the shaft
in the longitudinal direction.

In the case of a material which is weaker in tension than in shear. For instance a, circular shaft of cast iron or a
0
cylindrical piece of chalk a crack along a helix inclined at 45 to the axis of shaft often occurs.

Explanation: This is because of the fact that the state of pure shear is equivalent to a state of stress tension in
one direction and equal compression in perpendicular direction.

0
A rectangular element cut from the outer layer of a twisted shaft with sides at 45 to the axis will be subjected to
such stresses, the tensile stresses shown will produce a helical crack mentioned.

TORSION OF HOLLOW SHAFTS:

From the torsion of solid shafts of circular x – section , it is seen that only the material at the outer surface of
the shaft can be stressed to the limit assigned as an allowable working stresses. All of the material within the
shaft will work at a lower stress and is not being used to full capacity. Thus, in these cases where the weight
reduction is important, it is advantageous to use hollow shafts. In discussing the torsion of hollow shafts the
same assumptions will be made as in the case of a solid shaft. The general torsion equation as we have
applied in the case of torsion of solid shaft will hold good
Hence by examining the equation (1) and (2) it may be seen that the  maxm in the case of hollow shaft is 6.6%
larger then in the case of a solid shaft having the same outside diameter.

Reduction in weight:

Considering a solid and hollow shafts of the same length 'l' and density '' with di = 1/2 Do
 Hence the reduction in weight would be just 25%.

Illustrative Examples :

Problem 1

A stepped solid circular shaft is built in at its ends and subjected to an externally applied torque. T0 at the
shoulder as shown in the figure. Determine the angle of rotation 0 of the shoulder section where T0 is applied ?

Solution: This is a statically indeterminate system because the shaft is built in at both ends. All that we can find
from the statics is that the sum of two reactive torque TA and TB at the built – in ends of the shafts must be
equal to the applied torque T0

Thus TA+ TB = T0 ------ (1)

[from static principles]

Where TA ,TB are the reactive torque at the built in ends A and B. wheeras T0 is the applied torque

From consideration of consistent deformation, we see that the angle of twist in each portion of the shaft must
be same.

i.e a =  b =  0
 using the relation for angle of twist

N.B: Assuming modulus of rigidity G to be same for the two portions

So the defines the ratio of TA and TB

So by solving (1) & (2) we get

Non Uniform Torsion: The pure torsion refers to a torsion of a prismatic bar subjected to torques acting only
at the ends. While the non uniform torsion differs from pure torsion in a sense that the bar / shaft need not to be
prismatic and the applied torques may vary along the length.

Here the shaft is made up of two different segments of different diameters and having torques applied at
several cross sections. Each region of the bar between the applied loads between changes in cross section is
in pure torsion, hence the formula's derived earlier may be applied. Then form the internal torque, maximum
shear stress and angle of rotation for each region can be calculated from the relation

The total angle to twist of one end of the bar with respect to the other is obtained by summation using the
formula
If either the torque or the cross section changes continuously along the axis of the bar, then the  (summation
can be replaced by an integral sign ( ∫ ). i.e We will have to consider a differential element.

After considering the differential element, we can write

Substituting the expressions for Tx and Jx at a distance x from the end of the bar, and then integrating between
the limits 0 to L, find the value of angle of twist may be determined.

Instructional Objectives:
At the end of this lesson, the students should have the knowledge of:

 Stresses in thick cylinders.

 Lame’s equation for radial and circumferential stresses.
 Distribution of radial and circumferential stresses for different boundary conditions.
 Methods of increasing elastic strength of thick cylinders by prestressing.
Stresses in thick cylinders
For thick cylinders such as guns, pipes to hydraulic presses, high pressure
hydraulic pipes the wall thickness is relatively large and the stress variation across the
thickness is also significant. In this situation the approach made in the previous section is
not suitable. The problem may be solved by considering an axisymmetry about z-axis and
solving the differential equations of stress equilibrium in polar co-ordinates. In general
the stress equations of equilibrium without body forces can be given as

r 1 r rz r 

   0
r r  z r
(1)
r 1  z 
  2 r 0
r r  z r
zr 1 z z  zr
   0
r r  z r

0 and this gives
For axisymmetry about z-axis 
r rz r 
  0
r z r
r z  (2)
 2 r 0
r z r
zr z  zr
  0
r z r
In a plane stress situation if the cylinder ends are free to expand z = 0 and due to
uniform radial deformation and symmetry τrz = τz = τr = 0. The equation of equilibrium
reduces to
 r r 
 0
r r
This can be written in the following form:
r
  

(3)
r 
r r

If we consider a general case with body forces such as centrifugal forces in the case of a
rotating cylinder or disc then the equations reduce to
r r 
 2 r 0 which may be written as
r r

r r
r
 2r2 0 (4)
It is convenient to solve the general equation so that a variety of problems may be solved.
Now as shown in figure- 9.2.1.1, the strains r and  may be given by
ur 1 r   since  0
r    (5)
r E z

 
r u rr 
u
r


1
   (6)
 r
r E 
 r

u
u  r r
ur
r
r

' '
A A B B
'
A
 A



B
B'

9.2.1.1F- Representation of radial and circumferential strain.

Combining equation (5) and (6) we have
 
r  r r
1     r 0 (7)
r r
Now from equation (4) we may write
 2 2 r
r r 2 r r 22r and combining this with equation (7) we
r
may2 arrive at
  r
  2

r r 3  3   r 0 (8)

r 2
r

For a non-rotating thick cylinder with internal and external pressures pi and po we
substitute = 0 in equation (8) and this gives
2

r r

r 3 0 (9)

r2 r
A typical case is shown in figure- 9.2.1.2. A standard solution for equation (9) is
r = c rn where c and n are constants. Substituting this in equation (9) and also
combining with equation (3) we have
r c1 c2

r2
c2
 (10)
 c
 1
r2
where c1 and c2 are constants.

ro

ri
 pi po

9.2.1.2F- A thick cylinder with both external and internal pressure.

Boundary conditions for a thick cylinder with internal and external pressures pi and po
respectively are:
at r = ri r = -pi
and at r = ro r = -
po

The negative signs appear due to the compressive nature of the pressures. This gives
po pi 
2 2 2 2

pi ri poro r r 2 2
c  c2  i o
1 2 2

ro ri ro ri

The radial stress r and circumferential stress  are now given by
po pi  1
2 2 2 2

pi ri poro  r r
   i o
r 2 2 2 2 2

ro ri ro ri r (11)

2 2 2 2
po pi  1
pi ri poro ri ro
  
 2 2 2 2 2

ro ri ro ri r

It is important to remember that if  works out to be positive, it is tensile and if it is

negative, it is compressive whereas r is always compressive irrespective of its sign.
Stress distributions for different conditions may be obtained by simply substituting the
relevant values in equation (11). For example, if po = 0 i.e. there is no external pressure
the radial and circumferential stress reduce to
2  2 
pi ri 1 
  ro
r 2 2 2


ro ri   r

(12)
2  2 
pi ri 1 
  ro
  2 2 2

ro ri  r 

 
The stress distribution within the cylinder wall is shown in figure- 9.2.1.3.
 ro

ri 

r pi

9.2.1.3F- Radial and circumferential stress distribution within the cylinder

wall when only internal pressure acts.

It may be noted that r +  = constant and hence the deformation in z-direction is
uniform. This means that the cross-section perpendicular to the cylinder axis remains
plane. Hence the deformation in an element cut out by two adjacent cross-sections does
not interfere with the adjacent element. Therefore it is justified to assume a condition of
plane stress for an element in section 9.2.1.
If pi = 0 i.e. there is no internal pressure the stresses r and  reduce to
2 2
 
porro
  1 
r 2 2 i
2

ro ri r 
  
(13)
  2
 2 
poro  ir 1
 2 2 2

ro ri  r 

 
The stress distributions are shown in figure-9.2.1.4.
 po

ro

ri

r (negative)
(negative)

9.2.1.4F- Distribution of radial and circumferential stresses within the

cylinder wall when only external pressure acts.

Methods of increasing the elastic strength of a thick cylinder by

pre-stressing

In thick walled cylinders subjected to internal pressure only it can be seen from equation
(12) that the maximum stresses occur at the inside radius and this can be given by

2 2

rmax  pi  p r r

o i

rri i 2 2
(max)
rri
ro ri
This means that as pi increases  may exceed yield stress even when pi < yield.
Furthermore, it can be shown that for large internal pressures in thick walled cylinders
the wall thickness is required to be very large. This is shown schematically in figure-
9.2.2.1. This means that the material near the outer edge is not effectively used since the
stresses near the outer edge gradually reduce (Refer to figure- 9.2.1.3).
 pi

t
9.2.2.1F- A schematic variation of wall thickness with the internal pressure in
a thick walled cylinder.

In order to make thick-walled cylinders that resist elastically large internal pressure and
make effective use of material at the outer portion of the cylinder the following methods
of pre-stressing are used:

1. Shrinking a hollow cylinder over the main cylinder.

2. Multilayered or laminated cylinders.
3. Autofrettage or self hooping.

1. Composite cylinders
An outer cylinder (jacket) with the internal diameter slightly smaller than
the outer diameter of the main cylinder is heated and fitted onto the main
cylinder. When the assembly cools down to room temperature a composite
cylinder is obtained. In this process the main cylinder is subjected to an
external pressure leading to a compressive radial stress at the interface.
The outer cylinder or the jacket is subjected to an internal pressure leading
to a tensile circumferential stress at the inner wall. Under this condition as
the internal pressure increases the compression in the inner cylinder is first
released and then only the cylinder begins to act in tension. Gun barrels
 are normally pre-stressed by hooping since very large internal pressures
are generated.
Here the main problem is to determine the contact pressure ps. At the
contact surface the outer radius rsi of the inner cylinder is slightly larger
than the inside diameter rso of the outer cylinder. However for stress
calculations we assume that rso = rsi rs (say). The inner and outer

cylinders are shown in figure- 9.2.2.2.

ro rso
rsi

ri

ps

ps

Jacket or outer cylinder Inner cylinder

9.2.2.2F- Dimensions and the pressures at the contact surface of the
internal and outer cylinders.

For the outer cylinder the radial and circumferential stresses at the contact
surface may be given by
2  2 

psrs  1 o p

r
 
r rrs 2 2  2  s

ro rs  rs 
2  2 
ps rs  1ro 
 
 rrs
2 2  2 
ro rs  rs 
In order to find the radial displacements of the cylinder walls at the
u 1
contact we consider that     r . This gives the radial
r E

displacement of the inner wall of the outer cylinder as

 2 2 
ps rs ro  rs  
u  
r1
E r 2 r 2 

 o s 
Similarly for the inner cylinder the radial and circumferential stresses at
the outer wall can be given by
  p
r rr s
s

2 2
rs
 p  ri
 rrs s 2 2

rs ri
And following the above procedure the radial displacement of the contact
surface of the inner cylinder is given by
 2 2 
ps rs rs  ri   
u 
r2
E 
2 2

 s i 
The total interference  at the contact is therefore given by
 2 2 2 2 

ps rs ro  rs  rs  ri 
  .
E r 2 r 2 2 2
r r
 o s s i 

This gives the contact pressure in terms of the known variables as follows:
E
ps  2 
 2 2 2
 r  r  rs  ri 
sr  o2 s2 2 2
ro  rs rs ri 
 
The combined stress distribution in a shrink fit composite cylinder is made
up of stress distribution in the inner and outer cylinders and this is shown
in figure-9.2.2.3.
 r ro r
s s
rs
ps

ri


+
=
r
 
 r
r ri

9.2.2.3F- Combined stress distribution in a composite cylinder.

Residual circumferential stress is maximum at r = ri for the inner cylinder

and is given by
(max) 
2

 2psrs
2 2
rri
rs ri
Residual circumferential stress is maximum at r = rs for the outer cylinder
and is given by
2 2

 p r r
o s
s 2 2
(max)
rrs
ro rs
 Stresses due to fluid pressure must be superimposed on this to find the
complete stress distribution.

2. Multilayered or Laminated cylinder

The laminated cylinders are made by stretching the shells in tension and
then welding along a longitudinal seam. This is shown in figure- 9.2.2.4.
 Welded junctions

weld

weld

9.2.2.4F- Method of construction of multilayered cylinder

3. Autofrettage
In some applications of thick cylinders such as gun barrels no inelastic
deformation is permitted. But for some pressure vessel design satisfactory
function can be maintained until the inelastic deformation that starts at
inner bore spreads completely over the wall thickness. With the increase in
fluid pressure yielding would start at the inner bore and then with further
increase in fluid pressure yielding would spread outward. If now the
pressure is released the outer elastic layer would regain its original size
and exert a radial compression on the inner shell and tension on the outer
region.
This gives the same effect as that obtained by shrinking a hoop
over an inner cylinder. This is known as Self- hooping or Autofrettage.
This allows the cylinder to operate at higher fluid pressure. For a given
autofrettage fluid pressure a given amount of inelastic deformation is
produced and therefore in service the same fluid pressure may be used
without causing any additional inelastic deformation.
 Summary of this Lesson

Stresses and strains in thick cylinders are first discussed and Lame’s equations are
derived. Radial and circumferential stress distribution across the wall thickness in thick
cylinders have been illustrated. Methods of increasing elastic strength of a thick cylinder by
prestressing are then discussed. Interface pressure and displacement during shrinking a hollow
cylinder over the main cylinder have been expressed in terms of known variables. Finally
multilayered or laminated cylinders and autofrettage are discussed.