Você está na página 1de 6

A Computational Procedure for

D. Mackenzie Calculating Primary Stress for the


J. T. Boyle ASME B&PV Code
Department of Mechanical Engineering, A simple procedure for calculating primary stress consistent with the ASME B&PV
University of Strathclyde, Code is presented. The procedure is based on an iterative elastic analysis technique
Glasgow, Scotland referred to as the elastic compensation method, which invokes the lower-bound limit
load theorem to define maximum allowable static loads for the vessel. The method
is simple to implement as an automatic computational procedure in a conventional
elastic finite element program and requires minimal input from the designer.

Introduction
The ASME Boiler and Pressure Vessel Code Sections III lower-bound limit load theorem states that if for a given load
and VIII (Division 2) (ASME, 1989) postulate three distinct there exists a statically admissible stress field in which the stress
"types" of stress associated with three distinct failure mech- nowhere exceeds yield, then that load is a lower bound on the
anisms: peak, primary, and secondary stress. Peak stresses are limit load of the structure (Calladine, 1985). By definition, the
highly localized stresses which occur at local structural dis- primary stress field p is a statically admissible stress field, and
continuities and are limited by fatigue considerations. For thus meets one requirement of the lower-bound limit load
brevity, peak stresses are not considered in this paper but will theorem. If, in addition, stress field p satisfies the material
be addressed in future work. Primary stress is associated with yield criterion, the primary stress field fully satisfies the lower-
gross plastic deformation under static loading. Collapse will bound theorem and the applied load is a safe lower bound on
occur if the structure is unable to support the applied loads, the limit load of the structure. The yield criterion adopted in
and thus the primary stress field is load or equilibrium-con- ASME B&PV Code Section III and Section VIII Division 2 is
trolled. The primary stress field p is a statically admissible the maximum shear stress theory, also known as the Tresca
stress field (in equilibrium with the externally applied loads), criterion, and maximum allowable stresses are defined in terms
with maximum allowable value limited with respect to the limit of stress intensity SI (the algebraic difference between the larg-
load of the structure. Secondary stress is associated with failure est and smallest principal stress). Code stress limits are defined
due to ratcheting under cyclic loading. Secondary stress arises in terms of allowable stress, Sm. Two different primary stress
from compatibility requirements (constraint of adjacent parts limits are defined in the Code for combinations of three types
or self-constraint), and therefore the secondary stress field q of primary stress. The maximum permissible value of general
is deformation or displacement-controlled. Secondary stresses primary membrane stress intensity Pm is Sm, but for local
are self-limiting in that local yielding and minor distortions primary membrane stress intensity PL, general primary mem-
lead to stress redistribution such that failure does not occur brane plus primary bending stress intensity (Pm + Pb), and
from a single application of load. However, if the magnitude local primary membrane plus primary bending stress intensity
of secondary plus primary stress is too great, the residual (PL + Pb) the maximum is 1.5 Sm.
stresses required to ensure shakedown cannot be set up in the The tabulated values of Sm given in the Code are based on
structure and ratcheting occurs. The maximum allowable value consideration of both the yield stress and ultimate tensile
of secondary stress, taken in conjunction with the primary strength of the material such that Sm is notionally two-thirds
stress, is therefore defined with respect to the shakedown load of "design" yield strength aY (ASME, 1969). The primary
of the structure. stress limits defined in terms of design yield stress aY are,
Code assessment of pressure components is usually based therefore,
on linear elastic analysis. Essentially (neglecting peak stresses),
the elastically calculated stress field ae is assumed to be the Pm<-ffy PL<oY PM + Pb<aY PL + Pb<aY
sum of the primary and secondary stress fields
ae=p + q (1) That is, the Code restricts the primary stress to within the
Tresca design yield locus. Thus, the primary stress field fully
In pressure vessel design, the magnitude of p is limited by the satisfies the lower-bound limit load theorem, and " . . . con-
static load-carrying capacity or limit load of the structure. The servatively assures the prevention of collapse as determined by
the principles of limit analysis." (ASME B&PV Code Section
Contributed by the Pressure Vessels and Piping Division and presented at the VIII Division 2 Appendix 4-136 Applications of Plastic Anal-
Pressure Vessels and Piping Conference, Denver, Colorado, July 25-29, 1993, ysis (b)).
of THE AMERICAN SOCIETY OF MECHANICAL ENGINEERS. Manuscript received by
the PVP Division, October 14, 1993; revised manuscript received May 24, 1994. The magnitude of q (in conjunction with p) is limited by
Associate Technical Editor: W. K. Liu. the shakedown load of the structure; that is, ae=p + q (ne-

Journal of Pressure Vessel Technology NOVEMBER 1994, Vol. 116/339

Copyright 1994 by ASME


Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/ on 01/27/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use
glecting peak stress) is limited to ensure shakedown under cyclic
loading. The Code assumes shakedown occurs if the calculated
elastic stress range is less than or equal to twice the yield stress;
therefore, the limit on primary plus secondary stress intensity
(P+Q), is 2oy or 3Sm (assuming Sm = 2/3aY).
The elastic stress field ae of Eq. (1) is uniquely defined by
elastic finite element analysis, but the primary and secondary
stress fields, p and q, respectively, are not. The B&PV Code
design by analysis rules require the designer to: ensure the
maximum total stress satisfies the fatigue limit; isolate peak
stresses and ensure that the remaining primary plus secondary
stress meets the shakedown requirement; identify a conserv-
ative primary stress system and limit the maximum primary
0 / ITERATION
stress to prevent plastic collapse. To meet these requirements,
the designer must partition the calculated elastic stress field Fig. 1 Maximum stress per iteration
into primary, secondary, and peak categories, which is prob-
lematic when the design is based on 2-D and 3-D solid finite
element analysis. Here a simple systematic design by analysis (The stress used may be either stress intensity or von Mises
procedure based on the foregoing definitions of primary and equivalent stress, depending on the yield criterion used.) This
secondary stress and application of the elastic compensation procedure effectively stiffens the structure at low stress regions
method is proposed. and softens it at high stress regions, causing the stresses to
redistribute in a manner similar to failure mechanisms such as
plastic hinges. The magnitude of the redistributed stress field,
Stress Categorization by Elastic Compensation as, may increase or decrease between iterations, but over a
The elastic compensation method has been developed from number of iterations there is generally a net decrease in max-
a stress categorization technique called the reduced modulus imum stress value, as illustrated in Fig. 1.
method, originally proposed as a tool for categorizing stresses The stress, strain, and displacement fields obtained by the
in piping systems (Dhalla, 1984; Severud, 1984; Roche, 1986; elastic compensation procedure are substituted into plastic
Boyle and Spence, 1987) and later extended to more general bound theorems to obtain approximate failure loads for struc-
pressure vessel applications (Dhalla, 1987; Marriott, 1988; tures. For a given applied load set Pd, the lower-bound limit
Boyle, 1989; Seshadri, 1990; Seshadri and Fernando, 1991; load for iteration i, P/,-, is given by the expression
Seshadri, 1991). In the reduced modulus method, elastically
calculated stress is partitioned into primary and secondary (3)
components by comparing the simulated inelastic response of ' si' max
a structure with ideal models of primary and secondary be- where aY is the material yield stress, and asl the maximum
havior. In pressure vessel applications, material inelasticity is (redistributed) stress given in iteration i (Mackenzie et al., 1992,
simulated in an iterative elastic finite element analysis proce- 1993a,b). The best estimate of lower-bound limit load, P/, is
dure by systematically reducing the elastic modulus of highly the highest of the iteration limit loads; that is,
loaded elements so as to allow stress redistribution. Marriott
noted that if the maximum stress calculated in the reduced P, = max(P) (4)
modulus analysis satisfies the yield limit, the stress field sat- Primary Plus Secondary Stress. Neglecting peak stress, the
isfies the lower-bound limit load theorem and the applied load elastic stress field ae calculated in iteration 0 is subject to the
is a lower bound on the limit load of the vessel or component Code shakedown limitation
(Marriott, 1988). Seshadri proposed that elastic and modified
modulus stress fieldsin which the moduli of high-stress ele- '-"max 3o,
ments are reduced and low-stress elements increased-could Primary Stress. The primary stress constituent of ae is sub-
be used to identify statically determinate points in the structure,
ject to specified limits on Pm, PL, (Pm + Pb) and (PL + Pb).
and hence calculate approximate limit loads in a manner similar If the primary stress field cannot be distinguished from the
to reference stress methods in creep analysis. The present writ-secondary stress field, the designer must ensure that the elastic
ers have proposed a third technique for calculating limit loads stress meets the primary stress limit, which may lead to over-
based on the reduced modulus method; in particular, Mar- conservative design. However, the elastic compensation limit
riott's lower bound theorem approach and Seshadri's redis- load stress field can be used to make such a distinction. Each
tribution technique. This method is referred to as the elastic iteration of the elastic compensation procedure gives a stati-
compensation method, and can be used to obtain bounds on cally admissible stress field which can be used in conjunction
both limit and shakedown loads. with the lower-bound limit load theorem to calculate lower-
In the elastic compensation method, approximate failure bound limit loads. However, this stress field also meets the
mechanisms are simulated in an iterative elastic finite elementCode requirements for primary stress (it is an equilibrium stress
analysis by systematically modifying the moduli of elements field), and therefore can be designated as a primary stress field.
so as to cause the stress to redistribute. An initial elastic finite
The maximum primary stress due to an applied load P^ is the
element analysis is performed for an arbitrary load set, say maximum stress in the elastic compensation stress field, I asi I max.
Pd, to establish the elastic stress field. This initial solution is
Clearly, it is advantageous to base the design on the iteration
taken as iteration zero in a series of linear elastic analyses, in
giving the lowest value of lffs,lmax, (that is, the highest lower-
which the elastic modulus of each element is modified ac- bound limit load Vj). Convergence is not an issue as all the
cording to the equation iterations give equilibrium stress distributions; some are simply
better than others.
In a conventional elastic analysis, two limits are defined
(ff(/-i)) for primary stress, depending on the type of primary stress:
where subscript i is the iteration number, a a nominal stress S/max < Sm for general primary membrane stress and
value, and a(,_ i} the maximum (unaveraged) nodal stress as- S7max < 1.5S, for local primary membrane stress or primary
sociated with the element calculated in the previous iteration. membrane plus primary bending stress. In the latter case, the

3 4 0 / V o l . 116, NOVEMBER 1994 Transactions of the ASME

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/ on 01/27/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


maximum elastically calculated primary stress is allowed to factored proportionally such that when the load is equal to
reach the yield stress (assuming Sm is two-thirds aY), but an the limit load P/, the maximum stress is aY. If the stress is
adequate margin of safety (with respect to limit load) is given limited to S proportionality gives a maximum allowable load
by post-yield stress redistribution. By definition, limit load of
stresses cannot redistribute to give a lower maximum stress;
therefore, if an elastic compensation stress fieldwhich sim-
ulates the limit stress distributionis defined as the primary Oy 3
stress field, the entire stress field must be subject to the primary assuming Sm is two-thirds aY. Thus, the assumed (elastic com-
stress limit of Sm to ensure an adequate margin of safety. pensation) primary stress field and primary stress limit are
This interpretation of primary stress and the applicable stress consistent with the B&PV Code.
limit is effectively equivalent to the Code design by limit anal- The limit load calculation may be based on either the Tresca
ysis procedures. The ASME B&PV Code Section VIII Division or the von Mises yield criterion, von Mises giving a higher
2 Appendix 4-136.3 states that the limits on general primary load. The use of the less conservative von Mises criterion is
membrane stress intensity, local primary membrane stress in- proposed here as the elastic compensation method gives a con-
tensity, and primary membrane plus primary bending stress servative lower bound. (A precedent for using the von Mises
intensity need not be satisfied at a specific location if the criterion in pressure vessel design is the use of elastic-plastic
specified loadings do not exceed two-thirds of the lower-bound finite element analysis, which is usually based on the von Mises
collapse load. If 4-136.3 is applied, the maximum allowable failure surface.)
load, corresponding to primary stress or limit load require-
ments, is
Sample Problems
The design by analysis procedure described in the foregoing
* max ~ - * /
is illustrated by considering sample analyses of two cylindrical
In the elastic compensation procedure, load and stress are vessels with flat heads, one "thick" and the other "thin," as
shown in Fig. 2. Two designs are presented for each vessel; a
standard approach based on stress linearization and an alter-
native elastic compensation approach. For the purposes of
linearization, five distinct regions are identified in the vessel,
1 as defined in Fig. 3 and Table 1. The material model is linear
elasticity, with
t r*1
6-
2
10 =30E6psi cry=303psi S, = -aY=20E3 psi K = 0.3
10-

14- All analysis (and stress linearization) was performed using the
'
ANSYS finite element program (Swanson Analysis Systems
Inc., 1992) with eight node axisymmetric solid elements used
throughout.
Fig. 2 Sample vessel dimensions
Stress Linearization Analysis. The stress limits for the five
regions shown in Fig. 3 are explicitly defined in the ASME
B&PV Code and are reproduced in Table 1. Note (2) of Table
NB-3217-1 accounts for the possibility that the plate edge bend-
0 ing moments reduce the bending stress at the center of the
plate. If yielding occurs at the edge of the plate, the moment
constraint is lost and the actual stress at the center of the head
is greater than that calculated by elastic analysis (Cooper,
1973); in effect, the discontinuity bending stress behaves like
a primary stress and is classified Pb.
To apply the stress linearization procedure to the finite ele-
ment models, linearization class lines must be defined. In re-
porting an investigation of the ASME B&PV Code criteria,
Fig. 3 Nominal regions for conventional stress classification Hechmer and Hollinger (1991) made the following recom-

Table 1 Stress categories and limits for regions identified in Fig. 3


Region Description ' Stress categories Stress limits
1 Flat-head, Membrane = primary P <J
center region Bending = primary * in ^m
P,n+Pb= 1-5S,
2 Flat-head junction to Membrane = local primary PL=1.5S
transition Bending = secondary (n) PL + Q{a) = 3S,
or or
Bending = primary Pm + Pb= 1.5S,
3 Head/shell transition Membrane = local primary i \ = 1.5S,
region Bending = secondary PL + Q = 3S,
4 Shell junction to Membrane = local primary PL=l.5S,
transition Bending = secondary PL + Q=3S,
5 Cylindrical shell Membrane = primary p S
Grad thro' wall= secondary * m ^m
P, + Q = 3S,
'"' Note (2) of Table NB-3217-1 states "If the bending moment at the edge is required to maintain the bending stress in the middle to
acceptable limits, the edge bending is classified as Pb. Otherwise, it is classified as Q."

Journal of Pressure Vessel Technology NOVEMBER 1994, Vol. 116 / 341

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/ on 01/27/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


mendations inter alia on the applicability of stress linearization Thick Plate Vessel Analysis. Linear elastic analysis was
at specific regions of pressure vessels: performed for an applied pressure of 1000 psi. The finite ele-
ment mesh and applied boundary conditions are illustrated in
Linearization is "valid" in regions clearly recognized as Fig. 5. The maximum linearized stress intensity, total non-
structural elements; that is, shells of revolutions and circular linearized (primary + secondary + peak) stress intensity, and
plates. Linearization is not "valid" in transition elements, associated maximum permissible pressures Pmax
which normally serve the purpose of connecting discrete 1000 allowable
structural elements, or at sharp junctions in the model which * max or
do not represent the actual geometry. In the case of smooth
junctions, such as fillets or blends, it is recommended that for the five class lines are given in Table 2.
stresses be evaluated in the row of elements adjacent to the The maximum allowable pressure for the thick plate pressure
junction or along the nodal line at the junction. vessel is the smallest calculated maximum for the five regions
considered in the foregoing. Assuming the linearization pro-
For brevity, only five class lines are considered for each cedure is valid throughout the structure, the maximum allow-
geometry, as illustrated in Fig. 4. According to the foregoing able pressure is limited by the stress in the cylindrical shell and
recommendations, class line 3 in Fig. 4 is not valid. In the has a value of Pmax = 7.65 kpsi. However, if linearization is
sample analyses, linearization is applied in all five regions, but, assumed to be invalid in region 3, in the absence of any al-
in addition, allowable pressures are also calculated without ternative categorization techniques, the maximum allowable
applying any categorization technique, in which case the total pressure must be limited by defining the stress in region 3 as
(nonlinearized) elastic stress must be treated as primary; Pm local primary stress, subject to an allowable limit of 1.5Sm.
in regions 1 and 5 and PL in regions 2, 3, and 4. The maximum allowable pressure is therefore limited by the
stress in the transition region, and has a value of Pmax = 5.90
kpsi. Assuming Note (2) of NB-3217-1 to be valid does not
CLASS CLASS
CLASS CLASS CLASS
effect the calculated allowable pressure.
CLASS
LINE 1 LINE 2 LINE 1 LINE 2 LINE 3
LINES

Thin Plate Vessel Analysis. Linear elastic analysis was per-


L" z:::::i:\ formed for an applied pressure of 100 psi. The finite element
\ 1 CLASS 1
mesh and applied boundary conditions are illustrated in Fig.
\ LINE 4 1
t 6. The maximum linearized stress intensity, total nonlinearized
1
CLASS
LINE 4
(primary + secondary + peak) stress intensity and associated
1
1
calculated maximum permissible pressures Pmax
1
1
1
CLASS 1 100 allowable
LINE 5 |
1 CLASS * ITiaX cr T
1 LINE 5
r~~i

for the five class lines are given in Table 3.


Assuming the linearization procedure to be valid throughout
Fig. 4 Linearization class lines the structure, the maximum allowable stress is limited by the

ANSYS 4.4A1
DEC 15 1992
9l35l09
PLOT NO. 1
PREP7 ELEMENTS
TYPE NUM
TDIS
PRES
ZV -1
DIST-8.8
XT -7
YF -8

Fig. 5 Thick-plate vessel finite element model

3 4 2 / V o l . 116, NOVEMBER 1994 Transactions of the ASME

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/ on 01/27/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


1 ANSYS 4.4A1
DEC 15 1992
10l09l38
PLOT NO. 1
PREP7 ELEMENTS
TYPE NUM
TDIS
PRES
ZV -1
DIST-7.7
XF -7

i!
YF -5

liftffltfUf ~"
*N
1
1
1
1
1
|Y I
a x llailiiyii

Fig. 6 Thin-plate vessel finite element model

Table 2 Thick-plate maximum linearized and total stresses Table 3 Thick-plate maximum linearized and total stresses
Maximum pressure Maximum pressure
Region 5 / (psi) Pmax (PSi) Region SI (psi) P m ax (PSi)
P,= 1368 14620 P, = 492.7 4059
1 P, + P = 3455 8693 1 Pm + P = 5365 559
Total = 3567 5607 Total = 5377 372
P L =1917 15649 P L = 938.8 3196
PL + Q= 2558 23455 P L + 2 = 5705 1052
2 PL + P ? 0 ) = 2558 11728"" 2 P L + P 6 = 5705<0) 526 (a)
Total = 2314 12964 Total = 6283 478
P L = 605 49587 P L = 186.4 16094
3 P i + Q=3246 18484 3 P i + Q = 967.7 6200
Totaf=5088 5896 Total = 5204 576
Pz, = 1582 18963 P L = 202.5 14814
4 P i + 0=3062 19595 4 P i + Q = 757.3 7922
Total = 3185 9419 Total =1459 2056
P m = 2615 7648 P,= 171.1 11689
5 P ra + Q = 3321 18067 5 Pm + Q = 229.0 26201
Total = 3381 5915 Total = 232.8 8591
1 1
If Note (2) of NB-3217-1 valid. If Note (2) of NB-3217-1 is valid.

stress at the center of the plate and has a value of Pmax = 559 which is greater than the maximum stress at the center of the
psi. This is also true if linearization is assumed invalid in region head given by the finite element analysis. Therefore, the dis-
3, the transition region. The maximum allowable pressure cal- continuity bending stress must be considered to be primary
culated for the junction is Pmax = 576 psi, which is higher than bending stress Pb.
that for the middle of the plate. However, the validity of Note
(2) of NB-3217-1 effects the calculated allowable pressure for
thin plate vessel. If the stress at the junction is limited to Elastic Compensation Analysis. In the elastic compensa-
(PL + Pb), the maximum allowable pressure is 526 psi. The tion analyses, stress categorization is circumvented by invoking
maximum stress at the center of a simply supported circular the ASME B&PV Code Section VIII Division 2 Appendix 4-
plate of thickness h and radius a subject to uniform pressure 136.3 design by limit analysis rules. The limit load calculations
P is given by are based on the von Mises failure criterion. The secondary
3(3 + y)Pa2 stress limits conform to the Code criterion limiting the stress
intensity range to twice yield, based on the Tresca yield cri-
8h2 terion.
In this case
Thick-Plate Analysis. Applying the elastic compensation
3(3 + 0.3) 100 102 procedure (10 iterations, von Mises failure criterion), to the
= 12375 psi
8 l2 thick-plate vessel finite element model shown in Fig. 5, Eqs.

Journal of Pressure Vessel Technology NOVEMBER 1994, Vol. 116 / 343

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/ on 01/27/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


(3) and (4) give a lower-bound limit pressure of 13.26 kpsi. on stress categories for applications in which unequivocal Code
Therefore, the maximum allowable pressure according to limit guidance is not given. This paper has concentrated on using
load (primary stress) requirements is the elastic compensation method to calculate allowable loads
Pmax = 2/3P,= 8.84kpsi consistent with the ASME B&PV Code design by analysis rules
for primary stress. The elastic compensation method can be
The first yield pressure (Tresca) Py= 5.87 kpsi; therefore, the implemented as an automatic computational procedure for
maximum Code allowable pressure satisfying shakedown re- "off-the-shelf" finite element packages utilizing FORTRAN
quirements is programs or software-specific macro routines (as detailed for
Pmax = 2P y = 11.74 kpsi ANSYS by Mackenzie et al. (1994)). More complex inelastic
analysis techniques, such as elastic-plastic finite element anal-
The maximum pressure is controlled by primary stress/limit ysis, utilize actual material stress-strain relationships and may
load requirements and is Pmax = 8.84 kpsi. result in calculation of higher allowable loads. However, ine-
Thin-Plate Analysis. Applying the elastic compensation lastic analysis is more difficult to perform than linear elastic
procedure (10 iterations, von Mises failure criterion), to the analysis, requiring specialist software and design staff. The
thin-plate vessel finite element model shown in Fig. 6, the elastic compensation method offers a simple, robust alternative
lower-bound limit pressure is 914 psi. Therefore, the maximum to more advanced analysis tools.
allowable pressure satisfying the limit load (primary stress)
requirements is Acknowledgment
Pmax = 2/3P, = 609psi This research has been funded by a grant from the UK
The first yield pressure (Tresca) PY is 407 psi; therefore, the Science and Engineering Research Council. Use of ANSYS
maximum allowable pressure satisfying the Code shakedown through an educational license is also acknowledged.
(primary plus secondary) requirements is
Pmax=2Py=814psi References
ASME Boiler and Pressure Vessel Code, 1989, The American Society of
Therefore, the maximum pressure is controlled by primary Mechanical Engineers, New York, NY.
stress/limit load requirements and is Pmax = 609 psi. ASME, 1969, Criteria of the ASME Boiler and Pressure Vessel Code for
Design by Analysis in Sections III and VIII, Division 2, The American Society
Discussion and Conclusions of Mechanical Engineers, New York, NY.
Boyle, J. T., and Spence, J., 1987, " A Procedure for the Assessment of Elastic
The sample analyses presented above demonstrate the use Follow-Up in High Temperature Piping Systems," Design and Analysis of Pip-
of the elastic compensation method as a stress categorization ing, Pressure Vessels and Components, eds., W. E. Short et al., ASME PVP-
Vol. 120, New York, pp. 197-201.
tool. In a conventional design by analysis based on lineari- Boyle, J. T., 1989, "Elastic Follow-Up and the Categorization of Secondary
zation, the thick-plate vessel allowable pressure Pmax = 7.65 Stress," Proceedings, ASME PVP-Vol. 161, Honolulu, HI, pp. 47-53.
kpsi is controlled by the stress in the cylindrical shell (region Calladine, C. R., 1985, Plasticity for Engineers, Ellis Horwood Ltd, Chi-
5). However, if, as has been suggested, linearization is not chester, U.K.
valid in the transition region between the plate and the shell, Cooper, W. E., 1973, "An Introduction to the Design Procedures of the
ASME Boiler and Pressure Vessel Code," US-Japan Joint Symposium Pressure
in the absence of an alternative categorization procedure, the Vessel Technology and Pressure Component Codes, Tokyo, Japan, pp. 22-59.
maximum total (nonlinearized) stress in the transition is limited Dhalla, A. K., 1984, "Verification of an Elastic Procedure to Estimate Follow-
to the local primary stress limit of l.5S giving an allowable Up," Design of Elevated Temperature Piping, eds., R. H. Mallet and R. M.
pressure of Pmax = 5.90 kpsi. (This is clearly an excessively Mello, ASME PVP-Vol. 86, New York, NY, pp. 81-96.
Dhalla, A. K., 1987, " A Simplified Procedure to Classify Stresses for Elevated
conservative assumption and unlikely to be made in practice.) Temperature Service," Proceedings, ASME PVP-Vol. 120, New York, NY, pp.
If the proposed elastic compensation approach is used, the 177-188.
maximum allowable pressure is Pmax = 8.84 kpsi, an improve- Hechmer, J. L., and Hollinger, G. L., 1991, "Three-Dimensional Stress Cri-
ment of 15 percent over the linearization procedure and a 50- teria," Proceedings, ASME PVP-Vol. 210-2, San Diego, CA, pp. 181-191.
Mackenzie, D., Shi, J., Nadarajah, C , and Boyle, J. T., 1992, "An Iterative
percent improvement over the worst-case design. The thin-plate Elastic Analysis Procedure for Estimating Lower-Bound Limit Loads," Pro-
vessel allowable pressure is controlled by the discontinuity ceedings, ASME PVP, New Orleans, LA, pp. 129-134.
stress at the edge of the flat plate, where stress categories are Mackenzie, D., et al., 1993a, " A Simple Method of Estimating Limit Loads
clearly definedsubject to assessing the requirements of Note by Iterative Elastic Analysis I, II, & III," International Journal of Pressure
(2) of Table NB-3217-1and linearization is valid. The allow- Vessels and Piping, Vol. 53, No. 1, pp. 77-142.
Mackenzie, D., Nadarajah, C , Shi, J., and Boyle, J. T., 1993b, "Simple
able pressure given by the linearization procedure is Pmax = 526 Bounds on Limit Loads by Elastic Finite Element Analysis," ASME JOURNAL
psi, while the elastic compensation approach gives a slightly OF PRESSURE VESSEL TECHNOLOGY, Vol. 115, pp. 27-31.
higher pressure of Pmax = 609 psi, a 16-percent increase. This Mackenzie, D., Hamilton, R., and Boyle, J. T., 1994, "Using ANSYS ADPL
difference in allowable pressure is due to the limit stress dis- Macros to Calculate Limit Loads by Iterative Elastic Finite Element Analysis,"
Proceedings, 6th International ANSYS Conference, Pittsburgh, PA, Vol. 3, pp.
tributions implicitly assumed in the two procedures. Applying 14.31-14.38.
the lower-bound limit load theorem to the elastic membrane Marriott, D. L., 1988, "Evaluation of Deformation or Load Control of
plus bending type of stress distribution obtained by lineari- Stresses Under Inelastic Conditions Using Elastic Finite Element Stress Anal-
zation gives a lower limit load than the elastic compensation ysis," Proceedings, ASME PVP-Vol. 136, Pittsburgh, PA, pp. 3-9.
Roche, R. L., 1986, "Estimation of Piping Elastic Follow-Up by Using Con-
method, in which the actual (inelastic) limit state stress dis- ventional Computations," International Journal ofPressure Vessels and Piping,
tribution is simulated. Vol. 26, pp. 53-78.
Seshadri, R., 1990, "Classification of Stresses in Pressure Components Using
The vessels considered in the sample analyses have clearly the 'GLOSS' Diagram," Proceedings, ASME PVP-Vol. 186, Nashville, TN, pp.
defined Code stress categories and the designer is simply re- 115-123.
quired to partition the elastic stress into membrane and bending Seshadri, R., 1991, "The Generalized Local Stress Strain (GLOSS) Analysis-
components and apply the Code limits. Such clear guidance Theory and Applications," 25th Anniversary Volume, ASME JOURNAL OF PRES-
SURE VESSEL TECHNOLOGY, pp. 219-227.
on stress categories is not always available, and often the de- Seshadri, R., and Fernando, C. P. D., 1991, "Limit Loads of Mechanical
signer must decide on appropriate stress categories for cal- Components and Structures Using the GLOSS R-Node Method," Proceedings,
culated membrane and bending stresses. By addressing the ASME PVP-Vol. 210-2, San Diego, CA, pp. 125-134.
primary and' secondary failure modes directly by basing the Severud, L. K., 1984, " A Simplified Method Evaluation for Piping Elastic
Follow-Up," Proceedings, 5th International Conference on Pressure Vessel
design on inelastic analysis and Code guidelines, such as ASME Technology, ASME, San Francisco, CA, pp, 367-387.
B&PV Code Section VIII Appendix 4-136, stress categorization Swanson Analysis Systems Inc., 1992, ANSYS User's Manual for Revision
is circumvented, removing the onus on the designer to decide 5.0, Swanson Analysis Systems Inc., PO Box 65, Johnson Road, Houston, PA.

3 4 4 / V o l . 116, NOVEMBER 1994 Transactions of the ASME

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/ on 01/27/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

Você também pode gostar