Von Mises' Yield Criterion and Nonlinearly Hardening Rotating Shafts
Von Mises' Yield Criterion and Nonlinearly Hardening Rotating Shafts
Von Mises' Yield Criterion and Nonlinearly Hardening Rotating Shafts
Acta Mechanica
DOI 10.1007/s00707-004-0088-z Printed in Austria
Summary. A computational model is developed to estimate the stress distributions in rotating elastic-
plastic solid and hollow shafts by the use of von Mises’ yield criterion, deformation theory of plasticity and
a Swift-type hardening law. An efficient numerical solution procedure based on the shooting method and
Newton iterations is designed and used throughout this work to treat shafts with fixed and free ends. The
results of the computations are verified by comparison with analytical solutions in the elastic range as well
as with analytical elastic-plastic solutions employing Tresca’s yield criterion available in the literature. The
stresses, displacement and plastic strains are computed for nonlinearly hardening elastic-plastic solid and
hollow shafts rotating at different speeds, and the results are presented in graphical form.
1 Introduction
Elastoplastic analysis of stresses and deformations in rotating solid and hollow shafts has been
performed comprehensively in the past by the use of Tresca’s yield criterion and its associated
flow rule for ideally-plastic and linearly hardening materials. Gamer and Lance [1] obtained an
analytical solution for a linearly hardening rotating hollow shaft with fixed ends. Later, Mack
[2] obtained the closed form solution of a rotating hollow shaft with free ends. In that work, the
shaft was assumed to behave ideally-plastic. The solution for a rotating ideally-plastic solid
shaft with free ends was obtained by Mack [3]. In a later work, Gamer et al. [4] studied the
stresses in an elastic-plastic rotating solid shaft with fixed ends considering ideally-plastic shaft
material. Recently, Eraslan [5] extended the works of Mack [3] and Gamer et al. [4] to shafts
made of linearly hardening materials.
The objective of this work is to study elastic-plastic deformations of rotating solid and hollow
shafts with fixed as well as with free ends using von Mises’ yield criterion and Prandtl-Reuss’
equation. Shafts, exhibiting ideally-plastic, linear and nonlinear isotropic strain hardening in the
form of Swift’s hardening law are considered. The present study extends all of the works listed
above to include (i) von Mises’ yield criterion, (ii) nonlinear isotropic hardening, (iii) small and
large values of the hardening parameter.
An efficient and stable numerical algorithm possessing high order accuracy is designed and
used throughout to handle the nonlinearities associated with von Mises’ yield criterion. This
algorithm is based on a nonlinear shooting method using Newton iterations with numerically
approximated tangents. The results of the computations are compared with the existing ana-
lytical solutions based on Tresca’s criterion [1]–[5] for two purposes: to verify the present
130 A. N. Eraslan
computational model and to allow a comparison between the predictions of two commonly used
yield criteria. These comparisons show that there exists fairly good agreement in the predictions
of stress and displacement distributions and relatively poor agreement in plastic strain predic-
tions for shafts made of ideally-plastic and linearly hardening materials. It is also shown that for
such shafts the elastic-plastic interface predicted by Tresca’s criterion advances further than that
of von Mises’ and hence, lower fully plastic limits are predicted by Tresca’s criterion.
2 Model development
The following dimensionless and normalized variables are introduced. Radial coordinate:
r ¼ r=b, normal stress: rj ¼ rj =r0 , normal strain: j ¼ j E=r0 , radial displacement:
u ¼ uE=r0 b, angular velocity: X ¼ xbðq=r0 Þ1=2 , hardening parameter: H ¼ gr0 =E, with b
being the outer radius of the shaft, r0 the yield limit of the material, E the modulus of elasticity,
x the angular velocity, q the mass density and g the hardening parameter. The equations given
below are written in terms of these variables. For convenience, the over-bars are dropped in the
rest of the paper.
A state of generalized plane strain and small deformations are presumed. The equation of
equilibrium in radial direction,
rh ¼ ðrrr Þ0 þ X2 r2 ; ð1Þ
0
the geometric relations: r ¼ u ; h ¼ u=r, the compatibility relation
r ¼ ðrh Þ0 ; ð2Þ
and the generalized Hooke’s law
r ¼ pr þ rr mðrh þ rz Þ; ð3Þ
For purely elastic deformations pi ¼ 0. In a state of generalized plane strain z ¼ 0 ¼ constant
and from Eq. (5) the axial stress is determined as
rz ¼ 0 þ mðrr þ rh Þ: ð6Þ
At this point, it is convenient to define the stress function Y ðrÞ in terms of the radial stress as
Y ¼ rrr . From the equation of equilibrium (1) we obtain rh ¼ Y 0 þ X2 r2 . Hence, the total
elastic strains become
1
r ¼ ð1 m2 ÞY mð1 þ mÞY 0 mr2 X2 ð1 þ mÞ m0 ; ð7Þ
r
m
h ¼ ð1 þ mÞY þ ð1 m2 ÞY 0 þ r2 X2 ð1 m2 Þ m0 : ð8Þ
r
Nonlinearly hardening rotating shafts 131
The elastic equation is obtained by the substitution of r and h in the compatibility relation (2).
The result is
d2 Y dY ð3 2mÞX2 r3
r2 þ r Y ¼ : ð9Þ
dr2 dr 1m
This is a Cauchy-Euler nonhomogeneous differential equation, and its analytical solution will
be presented later.
2 1
dpz ¼ dk rz ðrr þ rh Þ : ð17Þ
3 2
In terms of increments of plastic strain components, the equivalent plastic strain increment
dEQ is given by
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 p p 2h p 2 i
dEQ ¼ dij dij ¼ ðdr Þ þ ðdph Þ2 þ ðdpz Þ2 : ð18Þ
3 3
Using plastic incompressibility dpr þ dph þ dpz ¼ 0 together with Eqs. (13) and (18), one
obtains
3 dEQ
dk ¼ : ð19Þ
2 ry
In the absence of plastic predeformation, according to deformation theory of plasticity, the
above equations for the plastic strains become
EQ 1
pr ¼ rr ðrh þ rz Þ ; ð20Þ
ry 2
EQ 1
ph ¼ rh ðrr þ rz Þ ; ð21Þ
ry 2
EQ 1
pz ¼ rz ðrr þ rh Þ : ð22Þ
ry 2
On the other hand, according to Swift’s hardening law, the relation between the yield stress and
the equivalent plastic strain can be expressed as
where
2rr rh rz drr
N1 ¼ ; ð29Þ
2ry dr
2rz rr rh
N2 ¼ ; ð30Þ
2ry
2rh rr rz
N3 ¼ : ð31Þ
2ry
Then, Eq. (27) is differentiated with respect to the radial coordinate r and combined with
Eq. (28) to obtain
drz 1 drr drh
¼ N1 N4 N5 þ N6 ry þ ðN3 N4 N5 þ N6 ry Þ ; ð32Þ
dr N8 dr dr
in which the following variables have been defined:
N4 ¼ rm
y ðm 1Þ þ 1; ð33Þ
N5 ¼ rr þ rh 2rz ; ð34Þ
N6 ¼ rm
y þ 2Hmry 1; ð35Þ
N7 ¼ rm
y þ Hry 1; ð36Þ
N8 ¼ 2N7 ry N2 N4 N5 : ð37Þ
Finally, the total strains from Eqs. (25) and (26) are placed in the compatibility relation (18)
making use of the relations (28) and (32) to give
rN1 N4 ðN8 N9 þ N5 N10 Þ ½2Hð1 þ mÞry þ 3ðrm
y 1Þðrh rr Þ
2r2 N11 X2 þ
2HN8 r2y 2Hry
ð38Þ
rN6 ðN8 þ N10 Þ drr d2 Y
rN11 2 ¼ 0;
2HN8 ry dr dr
where
N9 ¼ rr þ rz 2rh ; ð39Þ
N10 ¼ N2 N4 N9 þ N6 ry ; ð40Þ
If all the stresses rr , rh , ry and drr =dr in Eq. (38) are expressed in terms of Y and dY =dr by
using rr ¼ Y =r and rh ¼ Y 0 þ X2 r2 , Eq. (38) can be cast into the general form:
134 A. N. Eraslan
d2 Y dY
¼ Gðr; Y ; Þ: ð42Þ
dr2 dr
Equation (42) is a nonlinear two-point boundary value problem and may be solved numeri-
cally, subject to the following boundary conditions:
Y ðaÞ ¼ 0 and Y ð1Þ ¼ 0 for 0 a < 1: ð43Þ
It is noted that, while Y ðaÞ ¼ 0 holds for both a 0, in case of a solid shaft rr ð0Þ ¼ Y 0 ð0Þ
whereas for hollow shafts rr ðaÞ ¼ Y ðaÞ=a. Letting /1 ðrÞ ¼ Y and /2 ðrÞ ¼ dY =dr one may
write
d/1
¼ /2 ; ð44Þ
dr
d/2
¼ Gðr; /1 ; /2 Þ: ð45Þ
dr
Equations (44) and (45) form a system of initial value problems (IVP) and are integrated
numerically, starting with the initial conditions /1 ðaÞ ¼ 0 and /2 ðaÞ ¼ dY =drjr¼a . Since the
gradient of Y at r ¼ a is not known, a Newton iteration scheme is designed to obtain the
correct value of this gradient by requiring Y ð1Þ ¼ 0. Iterations begin with the initial estimate
/02 ðaÞ, and at the k-th iteration cycle the IVP-system is solved three times with
where D/ is a small increment of the order of /k2 ðaÞ=100. A better approximation for /2 ðaÞ is
obtained from
ð2D/ Þf1
/kþ1 k
2 ðaÞ ¼ /2 ðaÞ : ð46Þ
f2 f3
Iterations are repeated until j/2kþ1 ðaÞ /k2 ðaÞj < eT with eT being the specified error tolerance.
In the following calculations, eT is taken as 108 in double precision computing environment.
The main advantages of using this procedure are stability, rate of convergence and availability
of state-of-the-art ODE solvers for accurate integrations.
The initial value system defined by Eqs. (44) and (45) is stiff at early stages of the integration.
The robust stiff ordinary differential equation solver LSODE [7] is used to overcome the
stiffness problem. For shafts with free ends, an outer iteration loop is performed to estimate the
unknown axial strain 0 . Newton iterations similar to those explained above start with the
ð0Þ
initial estimate 0 and are repeated until the net axial force Fz vanishes, that is
Z Z 1
Fz ¼ rz dA ¼ 2p rz rdr ¼ 0: ð47Þ
a
The axial stress rz in the integrand is determined either from Eq. (27) or from Eq. (6)
depending on whether the region is plastic or elastic. Of course, in a numerical calculation it is
not possible to make the axial force exactly equal to zero. Thus, the main iterations to calculate
0 are terminated when Fz 1010 . The analytical value of 0 calculated at the elastic limit
usually provides a reasonable initial estimate to begin. A typical elastoplastic computation with
this procedure executes less than 1 second on a Pentium-IV based PC running at 1.8 GHz.
Nonlinearly hardening rotating shafts 135
3 Preliminary calculations
C1 ð1 þ 2mÞX2 r2
rh ¼ 2
þ C2 ; ð50Þ
r 8ð1 mÞ
mX2 r2
rz ¼ 2mC2 þ 0 ; ð51Þ
2ð1 mÞ
C1 rm0 1 2m 2 3
u ¼ ð1 þ mÞ þ ð1 2mÞrC2 X r : ð52Þ
r 1 þ m 8ð1 mÞ
For shafts with free ends, the net force Fz in axial direction vanishes, hence, the constant axial
strain is determined from Eq. (47) as
ð1 þ a2 ÞmX2
0 ¼ 2mC2 þ : ð53Þ
4ð1 mÞ
The elastic solution is completed by the application of boundary conditions. For a solid shaft,
the stress components must be finite at the axis of the shaft, therefore C1 ¼ 0. The surface of the
shaft is free of any traction and hence rr ð1Þ ¼ 0. This condition gives
ð3 2mÞX2
C2 ¼ : ð54Þ
8ð1 mÞ
Equations (49) and (50) reveal that at the axis of a solid shaft
ð3 2mÞX2
rr ð0Þ ¼ rh ð0Þ ¼ C2 ¼ : ð55Þ
8ð1 mÞ
On the other hand, in the case of a hollow shaft with traction free surfaces, the boundary
conditions read rr ðaÞ ¼ rr ð1Þ ¼ 0 which result in
a2 ð3 2mÞX2
C1 ¼ ; ð56Þ
8ð1 mÞ
ð1 þ a2 Þð3 2mÞX2
C2 ¼ : ð57Þ
8ð1 mÞ
The stresses for a solid shaft with fixed ends are obtained as
X2 ð3 2mÞð1 r2 Þ
rr ¼ ; ð58Þ
8ð1 mÞ
X2 3 2m ð1 þ 2mÞr2
rh ¼ ; ð59Þ
8ð1 mÞ
mX2 3 2m 2r2
rz ¼ : ð60Þ
4ð1 mÞ
136 A. N. Eraslan
Yielding commences at the axis of the shaft where stresses satisfy the inequality rr ¼ rh > rz . If
the yield stress (14) is written at this location, it becomes
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1n o
ry ¼ ½rr ð0Þ rz ð0Þ2 þ 0 þ ½rz ð0Þ rr ð0Þ2 ¼ rr ð0Þ rz ð0Þ; ð61Þ
2
which is the yield stress according to Tresca’s yield criterion. Hence, for a solid shaft both von
Mises and Tresca criteria predict the same elastic limit angular velocity. Plastic deformation
takes place as soon as ry ¼ 1. Substituting the stresses (58) and (60) in rr ð0Þ rz ð0Þ ¼ 1, the
elastic limit angular velocity X ¼ X1 is determined as
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
8ð1 mÞ
X1 ¼ : ð62Þ
ð1 2mÞð3 2mÞ
For a solid shaft with free ends, the axial strain is obtained as 0 ¼ mX2 =2 by using Eq. (53).
The axial stress, which depends on 0 , becomes
mX2 ð1 2r2 Þ
rz ¼ : ð63Þ
4ð1 mÞ
Yielding begins at the axis of the shaft where the stress state satisfies rr ¼ rh > rz . Placing the
stresses in rr ð0Þ rz ð0Þ ¼ 1, the elastic limit angular velocity is determined as
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
8ð1 mÞ
X1 ¼ : ð64Þ
3 4m
In the case of a hollow shaft of inner radius a with fixed ends, using Eqs. (56) and (57), the
stresses are determined as
X2 ð3 2mÞ ða2 r2 Þðr2 1Þ
rr ¼ ; ð65Þ
8r2 ð1 mÞ
X2 a2 ð1 þ r2 Þð3 2mÞ þ r2 ½3 2m ð1 þ 2mÞr2
rh ¼ ; ð66Þ
8r2 ð1 mÞ
mX2 ð3 2mÞð1 þ a2 Þ 2r2
rz ¼ : ð67Þ
4ð1 mÞ
Yielding begins at the inner surface of the shaft where rh > rz > rr ¼ 0. The elastic limit
angular velocity is obtained as
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
4ð1 mÞ
X1 ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi: ð68Þ
2
½3 2m þ a ð1 2mÞ ½1 mð1 mÞ
If the ends of the hollow shaft are free, the axial strain is found to be
1
0 ¼ mX2 ð1 þ a2 Þ: ð69Þ
2
Hence, the axial stress takes the form
mX2 ð1 þ a2 2r2 Þ
rz ¼ ; ð70Þ
4ð1 mÞ
accordingly, the elastic limit angular velocity is determined as
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
4ð1 mÞ
X1 ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi: ð71Þ
ð3 þ a2 Þ½3 5m þ a2 ð1 3mÞ þ m2 ð7 þ 6a2 þ 3a4 Þ
Nonlinearly hardening rotating shafts 137
Note that, based on Tresca’s yield criterion, yielding begins at the inner surface as soon as
rh ðaÞ ¼ 1, irrespective of the end condition. This condition leads to the limit for hollow shafts
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
4ð1 mÞ
X1 ¼ : ð72Þ
3 2m þ a2 ð1 2mÞ
Taking Poisson’s ratio as m ¼ 0:3, elastic limit angular velocities for solid and hollow shafts of
different inner radii are calculated and the results are presented in Table 1.
As can be seen in Table 1, Tresca’s yield criterion predicts lower elastic limit angular
velocities than von Mises’ for hollow shafts.
In this section, the results of the computations will be compared with analytical solutions in
the elastic range. Throughout these calculations Poisson’s ratio is taken as m ¼ 0:3. First, the
stresses and displacement are calculated for a solid shaft with free ends at the elastic limit
angular velocity X1 ¼ 1:76383. A numerical solution is performed for this problem starting
ð0Þ
with 0 ¼ 0:4. Computations are converged to 0 ¼ 0:466667 after 11 main iterations. At
each main iteration 3, 4 sub iterations are performed to get the converged value of /2 ð0Þ. At the
end of the 11-th iteration, /2 ð0Þ ¼ 1:33333. On the other hand, the nonzero integration con-
stant is calculated from Eq. (54) as C2 ¼ 1:33333 and the normalized axial strain is calculated
from Eq. (53) as 0 ¼ 0:466667. Numerical and analytical solutions for the stresses and
displacement are compared in Fig. 1a. Both solutions agree perfectly. In this figure, dots
represent the analytical solution, and the stress variable / is calculated from
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
iffi
1h
/¼ ðrr rh Þ2 þ ðrr rz Þ2 þ ðrh rz Þ2 ; ð73Þ
2
which corresponds to the yield stress ry in the plastic core. Note that / ¼ 1 at the elastic-plastic
border and / < 1 in the elastic region.
Next, an elastic hollow shaft with free ends having an inner radius of a ¼ 0:2 is considered.
The elastic limit angular velocity, as predicted by von Mises’ yield criterion, is X1 ¼ 1:10681
(see Table 1), and with the help of Eq. (69), the analytical solution gives 0 ¼ 0:191105. The
numerical result for the axial strain is exactly the same as that of the analytical value, and it is
obtained after 3 main iterations. Figure 1b presents numerical and analytical stress distribu-
tions. Again, dots represent the analytical solution and like in elastic solid shaft calculation,
perfect agreement between numerical and analytical solutions is obtained. These sample
calculations show that the numerical solution algorithm performs well and the computer code
that implements this algorithm functions properly.
1.6 1.2
1.4
1.0
1.2
sr
f sq
0.8
stresses and displacement
1.0
0.4 u 0.4
sr
0.2 u
sz 0.2
0.0 sz
0.0
–0.2
–0.4 –0.2
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
a radial coordinate b radial coordinate
Fig. 1. Comparison of analytical (dots) and numerical solutions for a rotating (a) solid shaft with free
ends at X1 ¼ 1:76383, (b) hollow shaft with free ends at X1 ¼ 1:10681
In this section, the results of the computations will be compared with analytical solutions for
linearly hardening and ideally-plastic shafts available in the literature by taking the material
parameter in Eq. (23) as m ¼ 1. In his earlier work [5], the author obtained analytical solutions
for elastic-plastic deformations of a linearly hardening rotating solid shaft with fixed ends and
with free ends. In that work, all stages of elastic-plastic deformations from purely elastic to
beyond fully plastic were studied using Tresca’s yield criterion and its associated flow rule. It
was shown that for a shaft with fixed ends the first stage of elastic-plastic deformation even-
tuates at X ¼ 2:63658 if the hardening parameter is taken as H ¼ 0:5. In this work, compu-
tation is performed for a solid shaft with fixed ends (0 ¼ 0) to obtain the distributions of stress,
radial displacement and plastic strain at the speed of X ¼ 2:63658 using the same hardening
parameter, H ¼ 0:5. The results of this partially plastic shaft are compared with the analytical
solution in Fig. 2. Figure 2a shows the stresses and the displacement. Dots represent the
analytical solution. As can be seen in this figure, both solutions agree well, and perfect
agreement is obtained for the radial displacement. The distributions of plastic strain are
compared in Fig. 2b. In this figure, analytical strains are drawn by dashed lines. Plastic strains
smaller in magnitude are predicted by von Mises’ criterion. Furthermore, Tresca’s criterion
advances the plastic-elastic border further than von Mises’. Since the results of partially plastic
linearly hardening solid shafts with free ends were not presented in [5], comparisons with such
shafts could not be made. However, Mack [3] presented the results of partially plastic solid
shafts made of ideally-plastic material. With the present model, ideally-plastic materials can be
simulated by taking H ¼ 106 . In Fig. 3a, calculated stresses and displacement in the elastic-
Nonlinearly hardening rotating shafts 139
3.5 0.15
3.0 0.10
eqp
0.05 eqp
stresses and displacement
2.5
sq erp
sr 0.00
plastic strains
2.0
sz
–0.05
1.5 ezp
–0.10
f
1.0 rep rep
–0.15
u
0.5 –0.20
0.0 –0.25
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
a radial coordinate b radial coordinate
Fig. 2a. Stresses and displacement, (b) plastic strains in a linearly hardening rotating elastic-plastic
solid shaft with fixed ends at X ¼ 2:63658. Dots (a) and dashed lines (b) represent the analytical
solution of Eraslan [5]
1.50
1.25 0.05
sr
sq
f
1.00 eqp
stresses and displacement
0.75 erp
0.00
plastic strains
0.50 u
0.25
sz ezp
–0.05 rep
0.00 rep
–0.25
–0.50 –0.10
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
a b
radial coordinate radial coordinate
Fig. 3a. Stresses and displacement, (b) plastic strains in an ideally-plastic rotating elastic-plastic solid
shaft with free ends at X ¼ 1:81659. Dots (a), and dashed lines (b) represent the analytical solution of
Mack [3]
plastic solid shaft with free ends corresponding to X ¼ 1:81659 are compared with the ana-
lytical solution of Mack [3] (dots). Good agreement is obtained. As can be seen in Fig. 3a, the
stress variable throughout the plastic core is calculated as / ¼ 1, which indicates ideally-plastic
140 A. N. Eraslan
1.2 1.40
sq 1.35
1.0
f
1.30
rep
stresses and displacement
0.8
u
angular velocity
1.25
0.6
1.20
0.4
1.15
0.2
sr 1.10
0.0 1.05
sz
–0.2 1.00
0.5 0.6 0.7 0.8 0.9 1.0 0.5 0.6 0.7 0.8 0.9 1.0
a radial coordinate b elastic-plastic border radius
Fig. 4a. Stresses and displacement in an ideally-plastic rotating elastic-plastic hollow shaft with free
ends at X ¼ 1:24916, (b) propagation of elastic-plastic border radius. Dots represent the analytical
solution of Mack [2]
material behavior. The plastic strains are presented in Fig. 3b in comparison with the analytical
strains (dashed lines).
Final comparisons are made with the results of Mack [2] for an ideally-plastic hollow shaft
with free ends. The stresses and displacement in a partially plastic hollow shaft of inner radius
a ¼ 0:5 rotating at X ¼ 1:24916 are given in Fig. 4a, together with the analytical solution
(dots). In Fig. 4b, the propagation of the elastic-plastic interface for this shaft with increasing
rotation speeds is presented. In this figure, dots represent the solution obtained by using
Tresca’s yield criterion given in [2]. As can be seen in Fig. 4b, the elastic-plastic border radius
predicted by Tresca’s criterion advances further than that of von Mises’, and as a consequence,
a lower plastic limit angular velocity is predicted by Tresca’s criterion. This comparison and the
ones discussed above verify the present elastoplastic model on different classes of problems and
for a wide range of values of parameters.
1.6 0.10
m = 1.0
1.4 m = 2.0
eqp
1.2 0.05
f sq
m = 0.5
1.0
stresses and displacement
0.00
0.8
plastic strains
sr
0.6
u –0.05 ezp
0.4
m = 0.5
0.2
sz –0.10
m = 1.0
0.0
m = 2.0
–0.2 –0.15
rep
–0.4
–0.6 –0.20
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8
a radial coordinate b radial coordinate
0.08
0.07 m = 2.0
0.06
erp
0.05 m = 1.0
plastic strain
0.04
m = 0.5
0.03
0.02
Fig. 5a. Stresses and displacement in a nonlin-
0.01 early hardening rotating elastic-plastic solid
shaft with free ends at X ¼ 1:9 using m ¼ 0:5
0.00 and H ¼ 0:5, (b) circumferential and axial
0.0 0.2 0.4 0.6 0.8 1.0 plastic strains for different m values, (c) radial
c radial coordinate plastic strain for different m values
The propagation of the elastic-plastic border radius with increasing angular velocity for a
linearly hardening (m ¼ 1) solid shaft with free ends is depicted in Fig. 6a using
different hardening parameters. The effect of H on the propagation of the plastic core can
easily be evaluated on this figure. Similar calculations are performed for different material
parameters using the hardening parameter H ¼ 0:5. Figure 6b presents the results of these
calculations.
The increase in magnitude of the axial strain 0 with increasing angular velocities for
linear hardening is computed using different hardening parameters, and the results of
these calculations are presented in Fig. 7a. As can be seen in this figure, the magnitude of
142 A. N. Eraslan
2.05 2.05
2.00 m = 0.5
2.00
H = 0.6
m=1
0.3 1.95
1.95 m=2
angular velocity
angular velocity
H = 0.1
1.90 1.90
rep rep
1.85 1.85
1.80 1.80
1.75 1.75
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
a b elastic-plastic border radius
elastic-plastic border radius
Fig. 6. Propagation of the elastic-plastic interface in a rotating solid shaft with free ends (a) for
different H values using m ¼ 1, (b) for different m values using H ¼ 0:5
2.1 2.1
m = 0.5
H = 0.6
1
2.0 0.3 2.0 2
0.1
angular velocity
angular velocity
1.9 1.9
e–0 e–0
1.8 1.8
1.7 1.7
1.6 1.6
0.3 0.4 0.5 0.6 0.7 0.8 0.3 0.4 0.5 0.6 0.7 0.8
a axial strain b axial strain
Fig. 7. Magnitude of total axial strain with increasing values of the angular velocity in a rotating solid
shaft with free ends (a) for different H values using m ¼ 1, (b) for different m values using H ¼ 0:5
the total axial strain decreases with increasing values of the hardening parameter H, however,
this effect is insignificant at low values of the angular speed X. The effect of the material
parameter m for H ¼ 0:5 on the magnitude of the total axial strain for the same shaft is shown
in Fig. 7b.
Nonlinearly hardening rotating shafts 143
2.965 2.045
2.035
2.960
plastic limit angular velocity
1.995
2.950
H = 0.2
H = 0.2
1.985
2.945 1.975
0.6 0.8 1.0 1.2 1.4 0.6 0.8 1.0 1.2 1.4
a material parameter m b material parameter m
1.66 1.465
1.64 1.455
H = 0.6
1.435
1.60
H = 0.4
1.425
1.58 H = 0.4
1.415
1.56
1.405
H = 0.2
1.54
1.395
H = 0.2
1.52 1.385
1.50 1.375
0.6 0.8 1.0 1.2 1.4 0.6 0.8 1.0 1.2 1.4
c material parameter m d material parameter m
Fig. 8. Variation of fully plastic limit angular velocity with m using H as a parameter for a (a) solid
shaft with fixed ends, (b) solid shaft with free ends, (c) hollow shaft with fixed ends having an inner
radius of a ¼ 0:5, and (d) hollow shaft with free ends for a ¼ 0:5
Finally, the dependence of the fully plastic limit angular speed on the material parameters H
and m in rotating solid and hollow shafts is determined computationally, and the results are
depicted in Fig. 8. Figure 8a shows the variation of the fully plastic limit angular velocity with
m using H as a parameter for a rotating solid shaft with fixed ends. The plastic limit angular
velocity increases as m decreases and/or H increases. Similar variations are obtained for a
rotating solid shaft with free ends as shown in Fig. 8b, for a rotating hollow shaft with fixed
ends having an inner radius of a ¼ 0:5 in Fig. 8c, and for a rotating hollow shaft with free ends
(a ¼ 0:5) in Fig. 8d.
144 A. N. Eraslan: Nonlinearly hardening rotating shafts
5 Concluding remarks
A computational model based on von Mises’ criterion, deformation theory and Swift’s hard-
ening law is developed to study the elastic-plastic response of rotating solid and hollow shafts
using different end conditions. Swift’s hardening law is used for its simple form, however the
model is designed in such a way that any other hardening law or polynomial strain-yield stress
relations can easily be incorporated. This paper represents an extension of the previous studies
in the theoretical analysis of elastic-plastic rotating shafts to include von Mises’ yield criterion
and nonlinear hardening material behavior. The distributions of stress and displacement for
linearly hardening or perfectly-plastic rotating shafts obtained by the present model agree well
with those obtained by Tresca’s criterion. However, plastic strains much smaller in magnitude
are predicted by the use of von Mises’ criterion. Computations for nonlinearly hardening
rotating solid and hollow shafts indicate that the stresses, displacements and strains in those are
affected significantly by the material and hardening parameters.
Acknowledgements
The author takes this opportunity to thank Prof. W. Mack at Technische Universität Wien for many
helpful discussions on the subject. Mr. T. Akiş, a PhD student in the Department of Engineering
Sciences at Middle East Technical University, has been helpful during the course of this work. His help
is appreciated. As usual, the manuscript was checked by Ms. F. Asya with great care and patience for its
style and language. The author is deeply indebted to her.
References
[1] Gamer, U., Lance, R. H.: Stress distribution in a rotating elastic-plastic tube. Acta Mech. 50, 1–8
(1983).
[2] Mack, W.: Rotating elastic-plastic tube with free ends. Int. J. Solids Struct. 27, 1461–1476 (1991).
[3] Mack, W.: The rotating elastic-plastic solid shaft with free ends. Techn. Mech. 12, 119–124 (1991).
[4] Gamer, U., Mack, W., Varga, I.: Rotating elastic-plastic solid shaft with fixed ends. Int. J. Engng
Sci. 35, 253–267 (1997).
[5] Eraslan, A. N.: On the linearly hardening rotating solid shaft. Eur. J. Mech. A/Solids 22, 295–307
(2003).
[6] Mase, G. E.: Continuum mechanics. Schaum’s Outline Series. New York: McGraw-Hill 1970.
[7] Hindmarsh, A. C.: ODEPACK. A systematized collection of ODE solvers. In: Scientific
computing (Stepleman, R. S., ed.). Amsterdam: North Holland 1983.