Sei sulla pagina 1di 6

Michael C.

Gibson
e-mail: m.c.gibson@cranfield.ac.uk

Amer Hameed
Anthony P. Parker
John G. Hetherington
Defence College of Management and Technology,
Engineering Systems Department,
Cranfield University at the Defence Academy,
Swindon, SN6 8LA, UK

A Comparison of Methods for


Predicting Residual Stresses in
Strain-Hardening, Autofrettaged
Thick Cylinders, Including the
Bauschinger Effect
High-pressure vessels, such as gun barrels, are autofrettaged in order to increase their
operating pressure and fatigue life. Autofrettage causes plastic expansion of the inner
section of the cylinder, setting up residual compressive stresses at the bore after relaxation. Subsequent application of pressure has to overcome these compressive stresses
before tensile stresses can be developed, thereby increasing its fatigue lifetime and safe
working pressure. This paper presents the results from a series of finite element models
that have been developed to predict the magnitude of these stresses for a range of end
conditions: plane stress and several plane-strain states (open and closed ended, plus true
plane strain). The material model is currently bilinear and allows consideration of strain
hardening and the Bauschinger effect. Results are compared to an alternative numerical
model and a recent analytical model (developed by Huang), and show close agreement.
This demonstrates that general purpose finite element analysis software may be used to
simulate high-pressure vessels, justifying further refining of the models.
DOI: 10.1115/1.2172964

Introduction

Numerical stress analysis of the relatively complex behavior of


a tube during autofrettage pressurization and depressurization requires:
a
b
c

implementation of an appropriate yield criterion


appropriate subdivision of the region
implementation of physically realistic boundary
conditions

All of these requirements are met in the work reported herein as


part of the initial development and validation of a finite element
FE procedure.
The Bauschinger effect 1 manifests itself as a reduction of
compressive yield stress as a result of prior plastic tensile strain
2. The Bauschinger effect factor is the normalized magnitude
of compressive stress during unloading at which behavior deviates
from linear elastic. In the development phase reported here, the
strain hardening during loading and unloading and Bauschinger
effect during unloading are assumed to be bilinear. A value of is
selected to ensure reyielding near the bore as autofrettage pressure
is removed.
Figure 1 shows the inner bore and outer radii ra and rb, as
well as the primary and secondary yield radii r p and rs, which
designate the extent of plastic deformation during the loading and
unloading processes, respectively.
This paper summarizes the development of a family of finite
element models aimed at creating an accurate representation of
the different end conditions used when modeling hydraulic autofrettage, rather than a precise material model of a gun steel. Instead
Contributed by the Pressure Vessel and Piping Division of ASME for publication
in the JOURNAL OF PRESSURE VESSEL TECHNOLOGY. Manuscript received December 5,
2005; final manuscript received December 19, 2005. Review conducted by John H.
Underwood. Paper presented at the Gun Tubes Conference 2005, April 1014, 2005,
Keble College Oxford, hosted by Cranfield University, RMCS, Shrivenham, Paper
No. S3P11.

Journal of Pressure Vessel Technology

a bilinear material, incorporating the Bauschinger effect reverse


yield stress set by was selected corresponding to the - plot
shown below Fig. 2.
To confirm the validity of a new model, comparison to results
obtained using alternative models is essential. Only then may the
new model be applied to new configurations.
Accordingly, the results from an FE model of a thick-walled
cylinder are here compared to results from an analytical model 3
and the Hencky programme 4. Specifically, the autofrettage and
residual stresses are scrutinized: the former referring to stresses
developed when the tube is subject to the peak autofrettage pressure; the latter to stresses that remain when the autofrettage pressure is removed.

Compared Methods

This paper compares the current FE model with other methods,


for a number of different end conditions, which are described
below. All three methods use the von Mises yield criterion, generally more suitable than even a well-developed Tresca solution,
such as that formulated by Liu 5, to predict behavior within the
plastic region. All configurations are pressurized to cause initial
plastic deformation throughout an equal proportion of the tube
wall thickness. This proportion is termed overstrain and is often
defined as a percentage of the wall thickness.
2.1

End Conditions

Open-ended: Hydraulic autofrettage in which pressure is


maintained by two frictionless pistons, with zero net
axial force in the tube itself. A gun barrel may be considered as an open-ended tube as one end is left free, and
there is zero net axial force in the tube itself.
Closed-ended: Hydraulic autofrettage in which pressure
is maintained by a cap at each end, creating a net axial
force in the tube itself. Provided calculations are performed at least one St. Venant distance from each cap,

Copyright 2006 by ASME

MAY 2006, Vol. 128 / 217

Downloaded 11 Nov 2008 to 129.5.224.57. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Table 1 Huang loading parameters


K = 2.0

K = 2.5

A1 = 990 MPa
A2 = 20.9 GPa
B1 = 1

A1 = 990 MPa
A2 = 20.9 GPa
B1 = 1

may be treated independently. To match the material used in the


model, linear strain hardening was selected. It is recognized that, following initial plastic deformation, many materials of
interest exhibit significant nonlinearity during unloading. However, the purpose of this work is to validate other aspects of the
model and the nonlinear unloading complexity is deliberately
avoided.
Although Huangs formulation is restricted to an incompressible material under plane strain, it does nonetheless allow for
rapid analysis of an autofrettaged tube. Other methods such as
Avitzur 6,7 and Bland 8 are less direct. Huangs method is
therefore well suited for comparison to, or validation of, another
method.
Table 1 contains the three parameters relating to the loading
phase. They are selected with knowledge of the materials elasticplastic loading curve and hence are the same for both wall ratios.
The elastic stresses part 0-1 in Fig. 2 are given by Eq. 3, the
plastic stresses part 1-2 in Fig. 2 by Eq. 4

ANSYS

Fig. 1 Tube geometry and yield diagram

the tube is in a plane state with z constant, but nonzero


throughout the tube wall.
Plane strain: More formally termed true plane strain, this
condition sets axial strain to zero throughout the tube
wall. This represents the midlength state of a long tube
constrained to its original length, such as a built-in pressure vessel.
Plane stress: In this condition axial stress is set to zero.
This may be used to represent a relatively thin sheet.
Accordingly, this condition is not particularly useful in
terms of pressure vessels and gun tubes, which tend to be
long. It may be used to model configurations such as
holes in aircraft skins subjected to mandrel enlargement.
Also, analytical plane stress solutions are frequently used
to normalize results for other end conditions.

The numerical modeling of each of the above end conditions is


described in later sections.
2.2 Huangs Model. Huang has developed a von Mises solution of an elastic-plastic tube. The solution is made possible by the
following simplifications:
Incompressibility
Plane strain

r + + z = 0
z = 0

= A1 + A2B1

Table 2 contains the three parameters relating to the unloading


phase, which depend on the prior plasticity in the tube. Accordingly, two sets of unloading parameters must be defined, one for
each wall ratio. These account for the fact that for a given overstrain, the bore plastic strain, on which strain hardening and the
Bauschinger effect depend, increases with wall ratio. Huangs
method assumes a uniform response to plastic strain in the initial
yield zone, as does the ANSYS model. The elastic stresses part 2-3
in Fig. 2 are given by Eq. 5, the plastic stresses part 3 in Fig.
3 by Eq. 6

UL = UL E2

Coincidentally, for this case the incompressibility condition creates stresses that are identical to the closed-end condition. The
model considers the tube material to behave linearly in the elastic
phase both loading and unloading, and either linearly or according to a power law in the plastic phase loading and unloading

= E1

UL

= A3 + A4

ULB2

2.3 Hencky Program. The Hencky program allows accurate


solution of hydraulic autofrettage for a wide range of end conditions and materials. The program was derived from a basic formulation by Jahed and Dubey 9 and further developed by Parker
et al. 4. It allows radial variation of unloading properties which
are crucial when determining residual stresses based on prior
plastic strain. This feature is beyond the capability of the ANSYS
FE model 10 employed here. Notably, the Hencky program can
simulate the often significant nonlinearity exhibited by various
candidate gun steels e.g., A723, HY180, PH 13-8Mo during unloading following initial plastic deformation.
2.4 ANSYS FE Model. Two forms of ANSYS models were created, one to simulate the plane stress condition and the other the
plane strain conditions true plane strain z = 0, plus open and
Table 2 Huang unloading parameters

Fig. 2 Material stress-strain diagram

218 / Vol. 128, MAY 2006

K = 2.0

K = 2.5

A3 = 1623 MPa
A4 = 20.9 GPa
B2 = 1

A3 = 1999 MPa
A4 = 20.9 GPa
B2 = 1

Transactions of the ASME

Downloaded 11 Nov 2008 to 129.5.224.57. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 3 Plane stress model diagram

closed ended. Both forms utilized an eight node, two-dimensional


2D, element, namely, PLANE82 which allows accurate results to
be obtained in the 2D analyses conducted.
Autofrettage pressures were determined by using the corresponding value from the Hencky program as an initial figure; the
precise autofrettage pressures for the relevant ANSYS model were
then obtained by interpolating the autofrettage stresses and iterating to give equal overstrain.
2.5 Plane Stress Model. The plane stress model Fig. 3 represents a small angle 0.5 deg of an r , section of the tube. The
plane stress element property was employed to achieve the required end condition. A range of angles and mesh sizing factors
were varied in an error analysis, to obtain a low-complexity model
still generating accurate results.
The results of the error analysis were used to develop a mesh in
which the radial dimension of the elements was varied throughout
the tube wall; smaller elements were used at the bore to properly
model plastic deformation and associated stresses.
2.6 Plane Strain Models. The plane strain models represent
an r , z section of the tube in which the longitudinal z- axis is
placed on the global ANSYS Y-axis, allowing the elements axisymmetric property to be used to reduce model size. Again, a
range of models was tested, altering model geometry and mesh
sizing factors, until a small but accurate model was obtained.
The three different plane strain conditions were implemented
by applying a variety of end constraints to the r , z section. The
true plane strain condition is achieved by specifying zero movement in the Y direction of the end surfaces AB and CD, as Fig. 4

Fig. 4 True plane strain model diagram

Journal of Pressure Vessel Technology

Fig. 5

Open-ended model diagram

shows.
Figure 5 shows the open-ended condition, which is achieved by
constraining end surface CD to zero movement in the Y direction
and coupling all nodes on the opposite surface AB to ensure
equal Y deflection, such that the surface deflection is determined
by the equilibrium condition zero net axial force.
The closed-ended condition, shown in Fig. 6, is designated by
constraining as in the open-ended condition but applying an additional axial load, Fz, to surface AB the node-coupled surface
equal to
Fz = pAF r2a

2.7 Modeling Process. Autofrettage is modeled by applying


the chosen pressure to the loading surface representing the bore
of the tube in the first load step and solving. The residual stresses

Fig. 6 Closed-ended model diagram

MAY 2006, Vol. 128 / 219

Downloaded 11 Nov 2008 to 129.5.224.57. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 7

r , z Section mesh

are then determined by removing the pressure in the second load


step and solving again. However, to model the Bauschinger effect, between the loading and unloading load steps the tube material must be altered from the default kinematic hardening model;
the Bauschinger effect is represented by changing the yield stress
of the region of the tube that underwent plastic deformationthe
primary yield zone ra r r p. This is achieved by defining two
temperature profiles for the material, one matching the initial material state and the other the deformed material state whose yield
stress has been preselected using the peak equivalent stress experienced by the tube bore. Note that the material properties are
otherwise unchanged between the two temperature profiles i.e., a
change in temperature is not being simulated; the titular change
of temperature is merely a convenient method of tailoring the
material properties to simulate the Bauschinger effect.
The mesh used for the three r , z section models is shown in Fig.
7. The plane strain condition allows considerable reduction in the
number of elements, as the section length may be kept short. The
end constraints are selected to simulate the midsection of the tube,
by maintaining plane strain conditions, making a long section unnecessary. As in the r , mesh, an error analysis was conducted to
identify an accurate but not overly fine mesh, and element sizes
were decreased at the bore.

Comparisons

The FE model was tested in two sets of comparisons, to 70%


overstrain: first, against both Huangs model and the Hencky program in the incompressible, true plane strain condition; second,
against the Hencky program in a variety of end conditions true
plane strain, plane stress, open and closed ended for a more standard material, = 0.3. For each, two wall ratios were used, K
= 2.0 and K = 2.5. As stated above, the focus of this paper is on
geometric accuracy rather than material fidelity; accordingly, the
Youngs and Tangent moduli were kept constant in loading and
unloading i.e., E1 = E2 = E, H1 = H2 = H.
The input parameters are summarized as follows refer to Fig. 2
and Tables 1 and 2 to compare:

Fig. 8 Comparison of autofrettage stresses, = 0.5, K = 2.0,


= 0.45

against those predicted by Huangs method and the Hencky Program for K = 2.0; Fig. 9 plots the residual hoop stresses from the
ANSYS model against those predicted by Huangs method and the
Hencky Program for both K = 2.0 and 2.5. In both figures, the
results from the Hencky Program and Huangs method are shown
as lines generally too close to be distinguishable and the results
from the ANSYS model are shown as symbols overlaying the corresponding plots from the other two methods.
All three principal stresses show good agreement with the
Hencky Program and Huangs model. The plots for K = 2.5
showed similar agreement and are omitted in the interest of brevity. Experience has shown that, for validation and subsequent
checking, it is important to summarize autofrettage stresses sepa-

Y = 1100 MPa
E = 209 GPa
H = E / 10= 20.9 GPa
= 0.7, K = 2.5
= 0.45, K = 2.0
The values of were chosen to give similar amounts of reyielding for both wall ratios.
3.1 Normalization of Results. The radial position is normalized using the following expression to relate it to the tube wall
thickness:
rNorm =

r ra
rb ra

The stresses are normalized with respect to the yield stress Y .


3.2 First Comparison: Huang-Hencky-ANSYS, = 0.5. The
results from the first stage of the comparison are shown below.
Figure 8 plots the autofrettage stresses from the ANSYS model
220 / Vol. 128, MAY 2006

Fig. 9 Comparison of residual hoop stresses, = 0.5, for K


= 2.0, = 0.45, and K = 2.5, = 0.7

Transactions of the ASME

Downloaded 11 Nov 2008 to 129.5.224.57. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 10 Residual hoop stresses for the open-ended tube

rately from residual stresses.


When assessing the increase in fatigue lifetime of an autofrettaged tube, the key property is bore residual hoop stress. Figure 9
shows a comparison of the ANSYS calculated values shown by
symbols against those from the Hencky Program and Huangs
method. A very close agreement can be seen throughout the tube
wall, including at the bore. This indicates the ANSYS model can
accurately reproduce results from Huangs model, when using a
bilinear material.

Fig. 12 Residual hoop stresses for the plane strain tube

quired to achieve 70% overstrain are listed in Tables 3 and 4.

Discussion

3.3 Second Comparison: Hencky-ANSYS, = 0.3. The results


from the second comparison, of the ANSYS model against the
Hencky Program for the four specified end conditions, are shown
in Figs. 1013 for K = 2.0 and 2.5. The graphs plot the residual
hoop stresses throughout the tube walls; they show the ANSYS
results as symbols overlaid on the continuous lines generated using the Hencky Program results. The autofrettage pressures re-

The key property of all the residual hoop stress plots is the
presence of secondary yielding near the bore, which ultimately
restricts the degree of prestressing possible in tubes. The reyielding depends on the autofrettage pressure due to the compressive
effects of its removal and the strength of the material altered
from initial properties by deformation during loading.
For the incompressible conditions, Figs. 8 and 9 show excellent
agreement between the ANSYS model, the Hencky program and
Huangs model. This indicates that the ANSYS model can accurately predict stresses in such a case. The comparisons between
the ANSYS model and Hencky program for the wider range of end
conditions and a more realistic Poissons ratio, are given in Figs.
1013. Again, a close match is exhibited. A slight variation may

Fig. 11 Residual hoop stresses for the closed-ended tube

Fig. 13 Residual hoop stresses for the plane stress tube

Journal of Pressure Vessel Technology

MAY 2006, Vol. 128 / 221

Downloaded 11 Nov 2008 to 129.5.224.57. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Table 3 Autofrettage pressures, K = 2.0


Pressure
MPa

End state

ANSYS

Hencky

Open ended
Closed ended
Plane strain
Plane stress

877.4
913.2
906.5
877.3

879.1
914.7
908.4
878.8

End state
Open ended
Closed ended
Plane strain
Plane stress

ANSYS

Hencky

1220
1254
1246
1224

1223
1257
1249
1227

be seen in the reyield zones ra r rs away from the bore, more


clearly visible for K = 2.0 and = 0.3; the ANSYS results show a
small decrease in the magnitude of residual hoop stresses and a
slight, commensurate, increase of reyield radius rs.
The autofrettage pressures in Tables 3 and 4 show two sets of
very similar values. The pressures for the open-ended condition
are most similar to those for the plane stress. This is not surprising
since the former represents a tube in which the axial stresses sum
to zero net force and the latter one in which all axial stresses are
zero. The autofrettage pressures for the closed-ended condition are
most similar to those for the plane strain condition. This is likewise unsurprising since they reduce to the same problem for the
case of an incompressible material. Finally, the fact that autofrettage pressure for plane strain and open ends exceeds that for plane
stress reflects the observations in 11. This could also be inferred,
using von Mises criterion, from the presence of an axial stress
which, in the near-bore region, generally varies between 0.3 and
0.5 times hoop stress, depending on Poissons ratio.
From Tables 3 and 4, it can be seen that the pressure required
for a given depth of autofrettage is affected by the chosen end
condition. This is controlled by the von Mises yield criterion that
considers all three principal stresses. The axial stress is the intermediate principal stress and is influenced by the chosen end condition; it, therefore, influences the degree of yielding. The variation between the various models in the near-bore reyield zone is
modest. The effectiveness of autofrettage lies in its ability to create compressive residual bore stresses to enhance service life by
inhibiting the growth of surface cracks and increasing the safe
working pressure of the tube.

Summary

Good agreement was observed between the results generated by


the ANSYS FE model, the Hencky numerical program and Huangs
analytical model. This demonstrates that given a correctly calculated degree of plastic strain at the bore, an accurate value of
residual stress may be predicted by the ANSYS model for a number
of end conditions. However, the bilinear stress-strain profile used

222 / Vol. 128, MAY 2006

Nomenclature
A14
B1,2
E1,2
H1,2
K

Table 4 Autofrettage pressures, K = 2.5


Pressure
MPa

here is an approximation. To obtain more realistic values of residual stress, a more accurate material model is required that properly follows the nonlinear unloading and also encompasses the
varying degree of plastic strain experienced by the material
throughout the tube wall. These enhancements will be the focus of
future work.

Norm

pAF
ra , rb
r p , rs
UL

E
+,
Y

+,
Y

material model parameters


material model exponents
loading and unloading Youngs moduli
loading and unloading reverse tangent moduli
wall ratio, rb / ra
superscript indicating a normalized value
autofrettage pressure at ra
inner and outer tube radii
primary and secondary yield radii
unloading superscript
Bauschinger effect factor, Y / +Y
Poissons ratio
elastic stress range between peak plastic strain
and onset of reverse yielding
forward and reverse yield stresses, in simple
tension
forward and reverse yield strains, in simple
tension

Acknowledgment
The first author would like to acknowledge Rosamund Gibson
and Cleveland Gibson, John Reynolds, and Darina Fierov for
their help in the preparation of the paper.

References
1 Bauschinger, J., 1881, ber die Vernderung der Elasticittsgrenze und des
Elasticittsmodulus Verschiedener Metalle, Zivilingenieur, 27, pp. 289348.
2 Milligan, R. V., Koo, W. H., and Davidson, T. E., 1966, The Bauschinger
Effect in a High-Strength Steel, ASME J. Basic Eng., 88, pp. 480488.
3 Huang, X. P., and Cui, W., 2005, Effect of Bauschinger Effect and Yield
Criterion on Residual Stress Distribution of Autofrettaged Tube, Gun Tubes
2005 Conference, Oxford, April.
4 Parker, A. P., Troiano, E., Underwood, J. H., and Mossey, C., 2003, Characterization of Steels Using a Revised Kinematic Hardening Model NLKH
Incorporating Bauschinger Effect, ASME J. Pressure Vessel Technol., 125,
pp. 277281.
5 Liu, C. K. undated, Stress and Strain Distributions in a Thick-Walled Cylinder of Strain-Hardening Material, Elastic-Plastically Strained by Internal
Pressure, NASA TN D-2941.
6 Avitzur, B., 1988, Determination of Residual Stress Distributions in Autofrettaged Tubing: A Discussion, Bent Laboratories Technical Report No.
ARCCB-MR-88034.
7 Avitzur, B., 1989, AutofrettageStress Distribution Under Load and Retained
Stresses After Depressurization, Bent Laboratories Technical Report No.
ARCCB-TR-89019.
8 Bland, D. R., 1956, Elastoplastic Thick-Walled Tubes of Work-Hardening
Material Subject to Internal and External Pressures and to Temperature Gradients, J. Mech. Phys. Solids, 4, pp. 209229.
9 Jahed, H., and Dubey, R. N., 1997, An Axisymmetric Method of ElasticPlastic Analysis Capable of Predicting Residual Stress Field, ASME J. Pressure Vessel Technol., 119, pp. 264273.
10 ANSYS 9.0, ANSYS, Inc., Canonsburg, PA, http://www.ansys.com
11 Parker, A. P., 2001, Autofrettage of Open-End Tubes Pressures, Stresses,
Strains and Code Comparisons, ASME J. Pressure Vessel Technol., 123, pp.
271281.

Transactions of the ASME

Downloaded 11 Nov 2008 to 129.5.224.57. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Potrebbero piacerti anche