Академический Документы
Профессиональный Документы
Культура Документы
Journal of Pressure Vessel Technology Copyright © 2010 by ASME DECEMBER 2010, Vol. 132 / 061201-1
Downloaded 20 Jan 2011 to 144.177.100.5. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
relationship is given by
0 共 − 0兲
= + 共3兲
E0 Et
For a bilinear material model, Eq. 共2兲 can be rewritten as
LHS ª 冕冉0
ⴱy
0 共 − 0兲
E0
+
Et
冊 1
d − 0共ⴱy − 0兲 − 共ⴱy − 0兲共1
2
− 0兲
冉 冊 冕冉 冊
f
0 共 f − 0兲 0 共 − 0兲
RHS ª + 共 f − ⴱy 兲 − + d
E0 Et ⴱy
E0 Et
Equating LHS= RHS,
ⴱy = 共1 − 兲0 +  f ⫾ 冑共 − 1兲共 f − 0兲 共4兲
where  = E0 / Et. Using Eq. 共4兲, the equivalent yield strength for
bilinear hardening material can be obtained.
Fig. 1 Illustrative determination of yⴱ
2.2 Ramberg–Osgood Material Hardening Model. The
Ramberg–Osgood material model can be written as
the shaded areas 共i.e., A1 = A2兲, the strain hardening curve can be
represented by an equivalent elastic perfectly plastic model with
the yield strength of ⴱy . Therefore, ⴱy can be determined by fol-
=
␣0
+
E0 E0 0
冉 冊 n
共5兲
lowing equation: where ␣ is the dimensionless material constant, and n is the strain
冕
hardening exponent.
ⴱy
1 Simplifying Eq. 共2兲 after using Eq. 共5兲, we get an expression
d − 0共ⴱy − 0兲 − 共ⴱy − 0兲共1 − 0兲
2 that leads to the equivalent yield strength, *y ,
0
冕
f Aⴱ2 ⴱ
y − B y + C = 0 共6兲
= f 共 f − ⴱy 兲 − d 共2兲 where A, B, and C are expressed in terms of material properties
ⴱy
A=1
where 共ⴱy , ⴱy 兲 is equivalent yield strength point, 共 , 兲 is an arbi-
trary point on a strain hardening curve, and f is the related stress
to the cut-off strain, f . It should be mentioned that 1 can be
calculated using the linear relation 1 = ⴱy / E0.
冉
B = 2 f +
␣nf
n−1
0
冊
Different material models such as the BH material and RO 2␣nn+1
f
relationship have been studied herein. These material models can C= + 2f
represent true stress-strain material curves within the small strain 共n + 1兲共n−1
0 兲
regions 关22兴. Therefore, the portion of the curve, which is used for Usually the value of cut-off strain, f is available as a material
determining the equivalent yield stress of equivalent elastic per- parameter. Therefore, in order to calculate f , the following equa-
fectly plastic material model, is limited to these regions only by tion needs to be solved:
choosing the cut-off strain f to be at 0.05.
n−1
0 E0n−1
0 f
2.1 Bilinear Hardening Material Model. A schematic plot nf + f − =0 共7兲
of a bilinear material model is shown in Fig. 2. The stress-strain
␣ ␣
Once f is known by using Eq. 共6兲, the equivalent yield strength
of the Ramberg–Osgood material model can be calculated.
Downloaded 20 Jan 2011 to 144.177.100.5. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
ing plastic flow can be described as s̄0ij = m0s0ij, in which m0 is the 3.3 The m␣-Tangent Method. Seshadri and Adibi-Asl 关29兴
statically admissible upper-bound limit load multiplier. The von introduced the concept of the “reference two-bar model” to relate
Mises yield criterion at limit state is given by a general configuration to the reference two-bar structure. The
scaling equations were proposed to estimate the limit load in a
3 given component using linear elastic analyses. More recently, Se-
f共s̄ij兲 = s̄ij0 s̄ij0 − 2y 共8兲
2 shadri and Hossain 关30兴 further extended the reference two-bar
and the general form of associated flow rule can be expressed as model and proposed the so-called m␣-tangent method, which en-
冉 冊
ables estimation of limit load using a single linear elastic analysis.
f The method has explicit dependency on the upper-bound multi-
˙ ij = where ⱖ0 共9兲
sij plier, m0, and the classical lower-bound multiplier, mL. The upper-
bound multiplier, m0, depends on the entire stress distribution in a
Mura et al. 关26,27兴 proposed the concept of “integral mean of
component or structure, whereas mL depends on the magnitude of
yield” in the context of variational formulation. The integral mean
maximum stress.
of yield criterion can be expressed as
The m␣-tangent multiplier can be readily evaluated by using the
冕 VT
0关f共s̄ij0 兲 + 共0兲2兴dV = 0 共10兲
following equation:
m0
Assuming a constant flow parameter 0 and 0 = 0, for linear elas- m␣T = 共16兲
1 + 0.2929共 − 1兲
tic relationship, the following expression can be obtained 关25兴:
where = m0 / mL.
y冑VT y冑VT As discussed in Ref. 关30兴, for components where ⬎ 1 + 冑2,
冑冕 冑兺
m01共=m0兲 = ⇔ N
共11兲
redistribution of secondary stresses could occur along with peak
共eq兲 dV
2
共eq兲2k ⌬Vk stresses. In such cases, the value of m0 is not constant during the
VT k=1 blunting of peak stresses, and there is a gradual reduction in its
magnitude. The choice of reference volume plays an important
where N is the total number of elements, y is the yield strength,
role in finding out the correct estimates of limit loads. Generally,
ek and ⌬Vk are the equivalent stress and volume of elements k,
for the component for which ⬎ 1 + 冑2, finding the limit load
respectively, and VT is the total volume of the component. The m01
requires the proper estimation of reference volume as these com-
limit load multiplier has been shown to be greater than classical
ponents have localized plastic regions being developed leading to
upper-bound limit load multiplier 关25兴.
larger dead volumes. In the m␣-tangent method, the complete vol-
3.1.2 Multiplier m02. Equation 共11兲 implies that the calculation ume is used when calculating limit loads. In this paper, the plastic
of m01 is based on the total volume VT. If plastic collapse occurs reference volume corrections are employed to correct the upper-
over a localized region of the structure m01 will be significantly bound multiplier.
overestimated. To overcome this problem, Pan and Seshadri 关28兴
proposed a new formulation for evaluating m0, namely, m02.
Based on the deformation theory of plasticity, the flow rule can 4 Plastic Reference Volume Method
be expressed as
The plastic reference volume is a subvolume of the component,
eij = sij 共12兲 which actively participates in plastic action at failure, whereas
where eij and sij are the deviatoric strain and stress, respectively. dead volume is the subvolume that does not participate. As shown
Therefore, can be defined as in Fig. 3, this method of finding the plastic reference volume of
general component involves integration of the m0 versus V̄ curve.
3 ¯ After the first linear elastic FEA, the elements are sorted in
= 共13兲
2 ¯ descending order of the equivalent stresses i.e., 1eq ⬎ 2eq ⬎ 3eq
where ¯ = 冑共3 / 2兲sijsij is the effective stress and ¯ = 冑共2 / 3兲eijeij is ⬎ ¯ ⬎ neq 共similar to the procedures discussed in Refs.
关5,6兴兲.The corresponding subvolumes are calculated. The upper-
the effective strain. Substituting Eq. 共13兲 into the integral mean of
bound multiplier m0 is given by the following equation:
yield criterion, the m02 limit load multiplier can be obtained as 关28兴
冑冕 共eq/eq兲dV 冑兺 N
共eq/eq兲k⌬Vk
冑冕 冑兺
VT k=1
m02 = y ⇔ y N
共14兲
eqeqdV 共eqeq兲k⌬Vk
VT k=1
Downloaded 20 Jan 2011 to 144.177.100.5. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
y冑VT have large dead volumes because of their geometry 共i.e.,
冑兺
m0 = n
共17兲 notched components兲. In such cases, m0共VRp兲 will result in
upper-bounded values.
2k Vk
k=1
Using Eq. 共17兲, the m0 values for different subvolumes is calcu- 5 Elastic Modulus Adjustment Procedure
lated as shown below:
An alternative method for estimating limit load is the EMAP.
y冑V1 The aim of EMAP is to establish inelasticlike stress fields by
m 共V=1兲 =
0
共18兲
冑21V1 = mL modifying the local elastic modulus in order to obtain the neces-
sary stress redistribution 关31兴. Numerous sets of statically admis-
y冑V1 + V2
sible and kinematically admissible distributions can be generated
m0共V=2兲 = 共19兲 in this manner, which enable the calculation of both lower- and
冑21V1 + 22V2 upper-bound limit loads.
The elastic modulus of each element in the linear elastic finite
In a more generalized form element scheme is systematically modified using the following
y冑V1 + V2 + ¯ + Vk equation:
冉 冊
m0共V=k兲 = 共20兲
冑21V1 + 22V2 + ¯ + 2k Vk ref
i q
Ei+1 = Ei 共24兲
If the k in Eq. 共20兲 is the last element in the sorted elements of a ieq
component, then m0共V=k兲 = m0共VT兲. Equation 共20兲 can be rewrit- where q is the elastic modulus adjustment parameter, ref is the
ten in terms of subvolume ratios as follows: reference stress, eq is the equivalent stress, and i is the iteration
index 共i = 1 for the initial elastic analysis兲.
冑
y
冑兺
m0共V兲 = n
共21兲 n
2k Vk
VT
兺 V
k
2
k k
k=1
ref
i
= 共25兲
These m values are then plotted against the subvolume ratio 共i.e.,
0 VT
V̄兲. A schematic of m0 versus V̄ plot is shown in Fig. 3. As each Equation 共24兲 describes how the elastic modulus at a location with
of the subvolume has a corresponding m0 value, the reference the equivalent stress eq 共e.g., the von-Mises equivalent stress兲 is
volume, which is also a subvolume will have a corresponding m0 updated from the ith to the 共i + 1兲th elastic iteration. This proce-
known as reference multiplier, m0共VR兲. dure continues until suitable convergence of a subsequent iteration
As can be seen from Fig. 3, whenever the subvolume ratio is is achieved. In this paper, the value of q is consistently assumed to
close to 0, m0共V兲 = mL and when the subvolume ratio reaches 1, be 0.1 when using EMAP, ensuring a slow but less fluctuating
m0共V兲 = m0共VT兲. Therefore, the reference multiplier should satisfy convergence trajectory.
both these conditions and also have to achieve all the intermediate
m0 values at different subvolume ratios. This can be accomplished
6 General Procedure
by taking a small segment of m0 versus V̄ curve and integrating
it over the entire range of subvolume ratios. In this section, a general procedure is outlined in a step by step
The value of m0共VRp兲 can be obtained by the following equa- manner to determine the lower-bound multipliers for components
tion: undergoing strain hardening using the proposed method.
冕
1 N for a given model with the prescribed loading and boundary
0
m共兲d = mL + 兺 共V̄
k=2
k − V̄k−1兲 ⫻ 共m0k − mL兲 共23兲
•
conditions.
The elements in the component are sorted in descending
order of the equivalent stress values.
The point C, which is the point of intersection of the m 共VRp兲 line 0
• The m0 value is calculated using Eq. 共14兲 for the compo-
with m0 versus V̄ curve, will give the value of plastic reference nent, as is discussed in Sec. 5, and mL using Eq. 共15兲.
volume factor V̄Rp. This reference volume factor will further be • Then these values of m0 are plotted against V̄ and m0共VRp兲
used to determine the modified multiplier, as is explained in Sec. is calculated using the Eq. 共23兲.
4.1. • Depending on the value of m0 / mL, the components are ap-
4.1 Categorization of Components. General components are propriately grouped.
categorized into two groups depending on the initial m0 / mL val- • For the components for which m0 / mL ⬍ 1 + 冑2, the value of
ues. The value of m0 / mL is an indication of presence of dead m0共VRp兲 is a lower-bound multiplier.
volume in a component. • For the components whose m0 / mL ⬎ 1 + 冑2, the value of
mT␣共VRp兲 should be calculated using Eq. 共16兲.
1. If m0 / mL ⱕ 1 + 冑2, then the components are well designed,
• The above steps are repeated until the converged or near
and have negligible dead volume, in such cases, m0共VRp兲 converged solution is obtained.
will be lower-bound values.
2. If m0 / mL ⬎ 1 + 冑2, then the components will have dead vol- In Sec. 7, the plastic reference volume approach is used to find
ume. In this category, components are again divided into two the lower-bound limit loads of some general components. The
groups. First, ones that develop flaws or defects during op- examples are so chosen that each one falls into one of the groups
eration 共i.e., cracked components兲 and, second, those that as classified earlier.
Downloaded 20 Jan 2011 to 144.177.100.5. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
Fig. 4 Thick walled cylinder: „a… geometry and dimensions and „b… typical
finite element mesh with loading
Table 1 Comparison of various multipliers of thick walled cylinder for various material hard-
ening models „LEFEA…
Fig. 5 Variation of m0„VRp… and m␣T „VRp… with iterations for bi- Fig. 6 Variation of m0„VRp… and m␣T „VRp… with iterations for
linear hardening thick cylinder „EMAP… Ramberg–Osgood hardening thick cylinder „EMAP…
Downloaded 20 Jan 2011 to 144.177.100.5. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
Table 2 Comparison of various multipliers of thick walled cylinder for bilinear material hard-
ening model „EMAP…
Table 3 Comparison of various multipliers of thick walled cylinder for Ramberg–Osgood ma-
terial hardening model „EMAP…
The comparisons of various multipliers with different material • Ramberg–Osgood material model:
hardening models for first iterations are presented in Table 1.
Variations of m0共VRp兲 and mT␣共VRp兲 with different iterations for
E0 = 211 GPa, 0 = 250 MPa, ␣ = 1.69, n = 8.60
bilinear hardening material model are presented in Fig. 5. Varia- The stress at the cut-off strain calculated using Eq. 共7兲 is
tions of m0共VRp兲 and mT␣共VRp兲 with different iterations for f = 361.97 MPa.
Ramberg–Osgood hardening material model are presented in Fig. Various coefficients of Eq. 共6兲 is calculated and given
6. The multiplier values at different iterations are summarized in below:
Tables 2 and 3. A = 1, B = 21101.72, C = 6.74 ⫻ 106
7.2 Compact Tension (CT) Specimen. A compact tension The equivalent yield strength calculated using Eq. 共6兲 is
specimen, as shown in Fig. 7, with a width W = 100 mm, height ⴱy = 324.33 MPa
H = 125 mm, thickness t = 3 mm, and crack length a = 46 mm is
subjected to a tensile load of P = 10 kN. The material is assumed The comparisons of various multipliers with different material
to be elastic perfectly plastic. The modulus of elasticity is speci- hardening models for first iterations are presented in Table 4.
fied as 211 GPa and the yield strength is assumed to be 250 MPa. Variations of m0共VRp兲 and mT␣共VRp兲 with different iterations for
Due to symmetry in geometry and loading, only a half of the plate bilinear hardening material model are presented in Fig. 8. Simi-
is modeled using Plane82 elements with plane stress consider-
larly, variations of m0共VRp兲 and mT␣共VRp兲 with different iterations
ation. Singularity elements are used at and around the crack-tip.
for Ramberg–Osgood hardening material model are presented in
The following strain hardening material properties are consid-
Fig. 9. The multiplier values at different iterations are summarized
ered:
in of Figs. 8 and 9 are given in Tables 5 and 6.
• Bilinear hardening material model:
7.3 Indeterminate Beam. An indeterminate beam 共Fig. 10兲
E0 = 211 GPa, 0 = 250 MPa, ET = 0.015 ⫻ E0, f with length L = 508 mm and height h = 25.4 mm is modeled. It is
subjected to uniformly distributed load of P = 1.0 MPa. The ma-
= 0.05
terial is assumed to be elastic perfectly plastic. The modulus of
The equivalent yield strength calculated using Eq. 共4兲 is elasticity is specified as 206.85 GPa and the yield strength is
ⴱy = 327.54 MPa. assumed to be 206.85 MPa. The beam is modeled using Plane82
Fig. 7 CT specimen: „a… geometry and dimensions and „b… typical finite
element mesh with loading
Downloaded 20 Jan 2011 to 144.177.100.5. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
Table 4 Comparison of various multipliers of CT Specimen for various material hardening
models „LEFEA…
Fig. 8 Variation of m0„VRp… and m␣T „VRp… with iterations for bi- Fig. 9 Variation of m0„VRp… and m␣T „VRp… with iterations for
linear hardening CT specimen „EMAP… Ramberg–Osgood hardening CT specimen „EMAP…
Downloaded 20 Jan 2011 to 144.177.100.5. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
Table 7 Comparison of various multipliers of indeterminate beam for various material hard-
ening models „LEFEA…
Fig. 11 Variation of m0„VRp… and m␣T „VRp… with iterations for Fig. 12 Variation of m0„VRp… and m␣T „VRp… with iterations for
bilinear hardening indeterminate beam „EMAP… Ramberg–Osgood hardening indeterminate beam „EMAP…
E0 = 206.85 GPa, 0 = 206.85 MPa, ␣ = 2, n = 8.47 bilinear hardening material model are presented in Fig. 11. Varia-
The stress at the failure calculated using Eq. 共7兲 is f tions of m0共VRp兲 and mT␣共VRp兲 with different iterations for
= 301.43 MPa. Ramberg–Osgood hardening material model are presented in Fig.
Various coefficients of Eq. 共6兲 are calculated and given 12. The multiplier values at different iterations are summarized in
below: Tables 8 and 9.
A = 1, B = 20685.59, C = 5.50 ⫻ 106
The equivalent yield strength calculated using Eq. 共6兲 is
8 Discussion
ⴱy = 269.65 MPa. From initial elastic analysis for any component, m0 / mL can be
calculated. Depending on the m0 / mL ratio, the component will fall
The comparisons of various multipliers with different material into one of the categories as explained earlier. For the compo-
hardening models for first iterations are presented in Table 7. nents, which fall under first category, m0共VRp兲 will be taken as the
Variations of m0共VRp兲 and mT␣共VRp兲 with different iterations for lower-bound multiplier, and for the components, which fall under
Table 8 Comparison of various multipliers of indeterminate beam for bilinear material hard-
ening model „EMAP…
Downloaded 20 Jan 2011 to 144.177.100.5. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
second category, mT␣共VRp兲 will be taken as the lower-bound mul- E0 = 200 GPa, 0 = 300 MPa, ET = 0.02 ⫻ E0, f
tiplier. It can be concluded that if plastic reference volume correc-
tion is employed mT␣ multiplier will always be a lower-bound
value. Comparing the results, it can be concluded that the multi-
= 0.05, =
E0
ET
冉 冊
= 50
pliers obtained by using the equivalent elastic perfectly plastic The stress at the failure point is
冉 冊
material models using equivalent yield strength 共which includes
strain hardening effect兲 are greater than the ones obtained by us- 0
f = ET ⴱ f + 0 − = 494.00 MPa
ing the regular yield strength. 
The equivalent yield strength is calculated using Eq. 共4兲,
9 Conclusion
ⴱy = 397.50 MPa
A simple approach is discussed in this paper to determine
equivalent yield strength of a material model with strain harden- • Ramberg–Osgood material model:
ing. Two different strain hardening material models, namely, the E0 = 200 GPa, 0 = 300 MPa, ␣ = 1.33, n = 8.60
bilinear hardening and Ramberg–Osgood models, are investi-
gated. The estimated yield strength along with limit load multipli- The stress at the failure point is calculated using Eq. 共7兲,
ers 共using the reference volume approach兲 are used to determine f = 433.69 MPa
the more appropriate limit load of a component with strain hard-
ening material. The applicability of the proposed procedure is Using Eq. 共6兲, various coefficients are calculated:
demonstrated through several numerical examples. The estimated A=1
limit loads are in good agreement with the ones obtained from
nonlinear finite element analysis.
Acknowledgment
冉
B = 2ⴱ f +
␣nf
n−1
0
冊
= 19999.19
Downloaded 20 Jan 2011 to 144.177.100.5. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
Method,” Eur. J. Mech. A/Solids, 27, pp. 195–209. 71, pp. 93–106.
关20兴 Adibi-Asl, R., and Seshadri, R., 2009, “Simplified Estimation Method for In- 关26兴 Mura, T., and Lee, S. L., 1963, “Application of Variational Principles to Limit
elastic Energy Release Rate,” Proceedings of the 12th International Confer- Analysis,” Q. Appl. Math., 21共3兲, pp. 243–348.
ence on Fracture 共ICF-12兲, Ottawa, Canada. 关27兴 Mura, T., Rimawi, W. H., and Lee, S. L., 1965, “Extended Theorems of Limit
关21兴 Hosford, W. F., 2005, Mechanical Behavior of Materials, Cambridge Univer- Analysis,” Q. Appl. Math., 23, pp. 171–179.
sity Press, New York. 关28兴 Pan, L., and Seshadri, R., 2002, “Limit Load Estimation Using Plastic Flow
关22兴 Huh, N. S., Kim, Y. J., Choi, Y. H., and Yang, J. S., 2004, “On Relevant Parameter in Repeated Elastic Finite Element Analysis,” ASME J. Pressure
Ramberg-Osgood Fit to Engineering Nonlinear Fracture Mechanics Analysis,” Vessel Technol., 124, pp. 433–439.
ASME J. Pressure Vessel Technol., 126, pp. 277–283. 关29兴 Seshadri, R., and Adibi-Asl, R., 2007, “Limit Loads of Pressure Components
关23兴 ASME, 2007, “Rules for Construction of Nuclear Power Plant Components, Using the Reference Two-Bar Structure,” ASME J. Pressure Vessel Technol.,
Division 1, NB—Class 1 Components and Appendices,” Boiler and Pressure 129, pp. 280–286.
Vessel Code, American Society of Mechanical Engineers, New York. 关30兴 Seshadri, R., and Hossain, M. M., 2009, “Simplified Limit Load Determination
关24兴 ASME, 2007, “Rules for Construction of Pressure Vessels, Division 2, alter- Using the m␣T -Method,” ASME J. Pressure Vessel Technol., 131, pp. 021213.
native Rules,” Boiler and Pressure Vessel Code, American Society of Mechani- 关31兴 Adibi-Asl, R., Fanous, I. F. Z., and Seshadri, R., 2006, “Elastic Modulus
cal Engineers, New York. Adjustment Procedures—Improved Convergence Scheme,” Int. J. Pressure
关25兴 Seshadri, R., and Mangalaramanan, S. P., 1997, “Lower Bound Limit Loads Vessels Piping, 83, pp. 154–160.
Using Variational Concepts: The m␣-Method,” Int. J. Pressure Vessels Piping, 关32兴 ANSYS, Inc., 2008, ANSYS Ver. 11.0 Online User Guide.
Downloaded 20 Jan 2011 to 144.177.100.5. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm