THICKWALL CYLINDERS

AND
PRESS FITS

25
Chapter Three
Nomenclature

A = cross-sectional area, m2
a = coefficient of linear thermal expansion, (ºC)-1
C1,C2 = integration constants
c = radial clearance
d = diameter, m
E = modulus of elasticity, Pa
l = length, m
ns = safety factor
P = force, N
p = pressure, Pa
r = radius, m
Sy = yield strength, Pa
T = torque, N-m
th = thickness, m
tl = tolerance, m
∆tm = temperature change, ºC
x,y = Cartesian co-ordinates, m
z = axial direction in cylindrical polar co-ordinates, m
α = cone angle, deg
β = body force per volume, N/m3
∆ = diametral interference, m
δ = interference or displacement, m
ε = strain
θ = circumferential direction in cylindrical polar co-ordinates, rad
µ = coefficient of friction
ν = Poisson’s ratio
ρ = density, kg/m3
σ = normal stress, Pa
ω = angular velocity, rad/s

26
http://www.technonet.co.kr/
Chapter 3 – Thick-Walled Cylinders and Press Fits

3.1 Introduction

Cylindrical elements are often used as machinery elements. Often a cylindrical shaft in a round
hole may serve a specific purpose. In designing these elements engineers are expected to specify
class of fit or specific tolerances. The tolerances, clearances or interferances are specified
according to the two tables on page 388 in the text.

Of primary concern in this course are the elements which are held together by the force caused
by elastic deformation as one element is forced into another to form the part. The theory
governing the design of these parts is developed from elastic theory applied to the stress – strain
relationships in thick-walled cylinders. A wall is considered thick if it is 1/10 the cylinder radius
or more.

3.2 Thick-Walled Cylinders

3.2.1 Stress Distribution

In thick-walled cylinders under internal and/or external pressure, the principal stresses are σr and
σt in the normal plane and σo in the longitudinal direction. The longitudinal σo is not significant
because: (1) It is equal to zero in open-ended cylinders, (2) It falls between σr and σθ and, hence,
does not enter commonly used maximum shear stress theory, (3) It is uniform across the cross-
section and along the length so that it does not influence σr and σθ. Hence, the condition of
plane stress exists.

The principal stress σr and σθ for a thick-walled cylinder under internal and/or external pressure
as shown in Fig. 1.1 are given by the following equations derived in Chapter 10 of textbook:

Fig. 3.1 Thick-walled cylinder under internal and external pressures

27
http://www.technonet.co.kr/
 2 2 2 2
ri p i − ro p o ri ro (p i − p o )
σr = 2 2
− 2 2
(3.1)
ro − ri r 2 (ro − ri )

2 2 2 2
ri p i − ro p o ri ro (p i − p o )
σθ = 2 2
+ 2 2
ro − ri r 2 (ro − ri ) (3.2)

The stress distribution is plotted in Fig. 3.2 for pi and po acting alone.

Fig. 3.2 Stress distribution

The radial displacement δr is given by

δr =
ri
(σ θ − νσ r ) (3.3)
E
where ν is Poisson’s ratio and E is the modulus of elasticity of the material.

3.2.2 Limiting Internal Pressure for Solid Cylinders

The allowable internal pressure for solid cylinders is limited not by wall thickness but by the
yield strength of the cylinder material. To illustrate this fact, assume ductile material and the
maximum shear stress theory for yielding, outside pressure po = 0.

Yielding will occur at the inner surface of the cylinder (r = ri) when
28
http://www.technonet.co.kr/
 σ r − σ θ = 2τ y

Substitute for σr and σθ from Eqs. (3.1) and (3.2) with r = ri, po = 0 results

2
ro p i
2τ y = 2 2 2
ro - ri

or

τy
ro = ri (3.4)
τ y − pi

It is seen as pi approaches τy, the material yield strength, the required radius b becomes infinite.
Hence, the allowable internal pressure can not be made arbitrarily high by making the cylinder
wall arbitrarily thick. Furthermore, it is obvious that yielding will commence when the bulk of
the material, with increasing r, is at low stress. Hence, solid thick walled cylinders use material
inefficiently. Material usage can be greatly improved by properly designed compound thick
cylinders.

3.3 Compound Cylinders

A compound cylinder is made by press-fitting one or more jackets around an inner cylinder.
This causes a residual compressive stress σθ at the cylinder bore (at r = ri) when the internal
pressure is zero. Then when the working pressure pi is applied, the resulting tangential stress is
superimposed with the residual tangential stress leaving a reduced net σθ.

3.3.1 Shrink and Press Fit Stresses

In a shrink or press fit, the residual stresses generated depend on the amount of interference.
Assume two cylinders as shown in Fig. 3.3, in which

Do - Di = 2δr = diametral interference

In order to be assembled at radius rf, the larger tube must displace outward at c by the amount uro,
the smaller tube must displace inward at rf by the amount δri. The total displacement must equal
the diametral interference, ∆, or

29
 Fig. 3.3 Interference of metal in press fit

∆ = 2(δ ro − δ ri ) (3.5)

The displacement is given by Eq. (3.3), that is, for outer tube (or disk)

rf
δ ro = (σ θ −νσ θ )
E
Substituting Eqs. (3.1) and (3.2) for po = 0, ri = rf, r = rf, pi = pc, results in

rf  ro 2 + r f 2 
δ ro = pc  2 + ν 
E  ro − r f 2 

(3.6)

Similarly, for inner tube (or shaft)

rf
δ ri = (σ θ −νσ θ )
E

Substituting Eqs. (3.1) and (3.2) for pi = 0, ro = rf, r = rf, po = pc, results in

rf  rf + ri 
2 2

δ ri = − pc − ν 
E  r f 2 − ri 2  (3.7)


If the two cylinders are of the same material, substituting Eqs. (3.6) and (3.7) into Eq. (3.5) gives,

4r f
3
 2
r −r
2 
∆= p c  2 o 2 i2 
E ( )(
 r f − ri ro − r f
2
)

30
 (3.8)

Thus the contact pressure, pc, due to interference alone, is

pc =
( 2
E∆ r f − ri
2
)(r o
2
− rf
2
) (3.9)
3
(
4r f ro − ri
2 2
)
The interference stresses can now be calculated by Eqs. (3.1) and (3.2) with the following
substitutions:

Fig. 3.4 Calculation of the interference stresses

Fig. 3.5 Stress distribution

The calculated interference stresses are shown in Fig. 3.5(a). Stresses due to pi can be easily
superposed as shown in Fig. 3.5(b).

Example 1: A 200-mm-diameter steel shaft is to have a press fit in a 500-mm-diameter cast

31
 iron disk. The maximum tangential stress in the disk is to be 35 MPa. The
modulus of elasticity for steel is 206,900 MPa and half this amount for cast iron.
Poisson’s ratio is equal to 0.3.
Find:
(a) the required diametral interference of metal.
(b) the force required to press the parts together if the coefficient of friction is
equal to 0.12 and the disk is 250 mm thick in the axial direction.
(c) the torque which the joint could carry because of the shrink fit pressure.

Solution: (a) By Eq. (3.2) or from Fig. 3.2(a): with ri = 100 mm, ro = 250 mm, and po = 0, r
= ri, σθ = 35 MPa

2 2
ro − ri 250 2 − 100 2
pc = pi = σt = × 35 = 25.3 MPa
ro
2
+ ri
2
250 2 + 100 2

For disk: By Eq. (3.6): with rf = 100 mm, ro = 250 mm

rf  ro 2 + r f 2  100 × 25.3  250 2 + 100 2 

δ ro 
= pc 2 +ν  =  + 0.3  = 0.0412 mm
E  ro − r f 2   250 − 100
2 2
5
2 .069×10
 2 
outward, that is, increase in hole radius

For shaft: By Eq. (3.7): with rf = 100 mm, ri = 0

 r f 2 + ri 2 
rf
δ ri = − p c  2 − ν  = − 100 × 25.3 (1 − 0.3) = −0.0086 mm
E  r f − ri 2 
 2.069 × 10 5

Inward, that is, decrease in shaft radius

By Eq. (3.5): ∆ = 2 (δro - υri) = 2 (0.0412 + 0.0086) = 0.100 mm

(b): Force required for assembly: F = 200 π 250 × 25.3 × 0.12 = 477,000 N

(c): Torque carried by press fit: T = 477,000 × 100 = 47,700,000 Nmm

32
3.3.2 Design of Compound Cylinders

Given: internal pressure pi, bore radius ri, and outer radius ro.
Required: optimum radius rf and optimum pressure pc.

Assume the maximum shear stress theory as a basis of the design.

σt −σr
τm =
2
Assume that for optimum design, the maximum shear stress in the inner tube, τmi, is equal to the
maximum shear stress in the outer tube, τmo.

τmi = τmo

Evaluate τmi at r = ri, superposing stresses due to pc and pi.

Evaluate τmo at r = rf, superposing stresses due to pc.
Equate τmi and τmo and obtain

2 2 2
ro (rf - ri )
2
rf (ro 2 - ri 2)
pc = pi
 rf 2 ro
2  (3.10)
 + 
r - r
2 2
ro - rf 
2 2
 f i

Substitute pc in τm and obtain

2 2
p i ro r f
τm = 2
(
r f ro − r f
2 2
)+ r (r o
2
f
2
− ri
2
) (3.11)

dτ m
Take = 0 and obtain optimum c.
d rf
r f = ri ro (3.12)

Substitute Eq. (3.12) into Eq. (3.10) and obtain optimum pc.

p i (ro − ri )
pc = (3.13)
2(ro + ri )
33
Substitute pc from Eq. (3.13) into Eq. (3.11) to obtain a convenient expression for τmax in
optimum design.
pi ro
τm = (3.14)
2 (ro - ri )

34