chapter four

Stress categories and stress

limits

Contents
4.1 Introduction.................................................................................................
4.2 Stress intensity ............................................................................................
4.3 Categorization of stresses .........................................................................
4.3.1 Primary stress .................................................................................
4.3.2 Secondary stress .............................................................................
4.3.3 Peak stress .......................................................................................
4.4 Stress limits .................................................................................................
4.5 Special stress limits....................................................................................
4.6 Practical aspects of stress categorization ...............................................
4.7 Shape factor considerations......................................................................
References .............................................................................................................
Problems................................................................................................................

4.1 Introduction
First of all we need to define the term stress. Stress is a tensor quantity
(neither a vector nor a scalar) that depends on the direction of applied load
as well as on the plane it acts. Generally speaking, at a given plane there are
both normal and shear stresses. However, there are planes within a
structural component (that is being subjected to mechanical or thermal
loads) that contain no shear stress. Such planes are called principal planes
and the directions normal to those planes are called principal directions.
The normal stresses (only stresses in those planes) are called principal
stresses. For a general three-dimensional stress state there are always three
principal planes along which the principal stresses act. In mathematical
terms we can say that the problem of principal stresses is an eigenvalue
problem, with the magnitudes of the principal stresses being the

Copyright 2005 by CRC Press, Inc. All Rights Reserved.

 eigenvalues and their directions (normal to the planes on which they act)
being the eigenvectors. Principal stress calculations form an essential
activity for a general stress analysis problem.

4.2 Stress intensity

Let us indicate the principal stresses by
1,
2, and
3. Then we define the
stress differences by:

S1;2 ¼
1 
2 ; S2;3 ¼
2 
3 ; S1;3 ¼
1 
3 ð4:1Þ

The stress intensity, SI, is then the largest absolute value of the stress
differences, or in other words
     
SI ¼ max S1;2 ; S2;3 ; S1;3  ð4:2Þ

The computed stress intensity is then compared with the material

allowables taking into consideration the nature of the loading. The material
allowables are based on yield and ultimate strength of the material with an
implied factor of safety.
Within the context of pressure vessel design codes, the comparison of
the allowable strength of the material is always done with respect to the
stress intensities. This puts the comparison in terms of the appropriate
failure theory either the maximum shear stress theory (Tresca criterion) or
the maximum distortion energy theory (von Mises criterion). These failure
theories have been discussed in some detail in Chapter 3.

4.3 Categorization of stresses

Stresses are generally characterized as (a) primary stress, (b) secondary
stress, or (c) peak stress. In the following discussion, the primary stresses
will be denoted by P, the secondary stress by Q and the peak stress by F.
These nomenclatures also apply to the ASME Boiler and Pressure Vessel
Code.1 We will now define each of the three categories of stress.

4.3.1 Primary stress

Primary stress is any normal stress or a shear stress developed by the
imposed loading which is necessary to satisfy equilibrium between external
and internal loads. These stresses are not self-limiting. If primary stresses
are increased such that yielding through net section occurs, subsequent
increase in primary stress would be through strain hardening until failure
or gross distortion occurs. Generally primary stresses result from an
applied mechanical load, such as a pressure load. The concept of
equilibrium is based on a monotonic load and a lower bound limit load

Copyright 2005 by CRC Press, Inc. All Rights Reserved.

 (for a discussion on limit load and the lower bound theorem see Appendix
C). When the limit load is exceeded, gross deformation takes place, hence
the qualification ‘‘not self-limiting.’’
A further elaboration of primary stresses is provided in a definition by
Pastor and Hechmer where they state:

Primary stresses are those that can cause ductile rupture or a

complete loss of load-carrying capability due to plastic collapse
of the structure upon a single application of load. The purpose of
the Code limits on primary stress is to prevent gross plastic
deformation and to provide a nominal factor of safety on the
ductile burst pressure.2

Primary stresses are further divided into three types: general primary
membrane (Pm), local primary membrane (PL), and primary bending (Pb).
Quite often the concepts of general primary membrane stress and local
primary membrane stress are used interchangeably; the local primary
membrane stress representing a general primary membrane stress along a
local structural discontinuity. The rigorous definition of the general primary
membrane stress is the average primary stress across a solid section
produced by mechanical loads, and excludes discontinuities and concen-
trations. The local primary membrane stress is defined as also the average
stress across any solid section, but includes discontinuities. However, the
general primary membrane stress is one that is so distributed in the
structure that no redistribution of load occurs as a result of yielding. The
failure mode associated with the general primary membrane stress and the
local primary membrane stress are meant to be different; the general
primary membrane stress leads to gross distortion with no redistribution,
and the local primary membrane stress to excessive plastic deformation
with redistribution of load.
The primary bending is the component of primary stress proportional to
the distance from the centroid of the solid section, and is produced by
mechanical loads. This definition excludes discontinuities and concentra-
tion. The concept of bending stress is akin to the situation of beam bending,
with a neutral axis along the center line with regions of tension and
compression. The membrane stress is the component having a constant
value through the section and represents an average value.

4.3.2 Secondary stress

Secondary stress originates through the self-constraint of a structure.
This must satisfy the imposed strain or displacement (continuity require-
ment) as opposed to being in equilibrium with the external load. Secondary
stresses are self-limiting or self-equilibrating. The discontinuity conditions
or thermal expansions are satisfied by local yielding and minor
distortions. The major characteristic of the secondary stress is that it is a

Copyright 2005 by CRC Press, Inc. All Rights Reserved.

 strain-controlled condition. Secondary stresses occur at structural dis-
continuities and can be caused by mechanical load or differential thermal
expansion. The local stress concentrations are not considered for
secondary stresses. There is no need for further dividing the secondary
stress into membrane and bending categories. In terms of secondary stress
we imply secondary membrane and bending in combination.

4.3.3 Peak stress

Peak stress is the highest stress in a region produced by a concentration
(such as a notch or weld discontinuity) or by certain thermal stresses. Peak
stresses do not cause significant distortion but may cause fatigue failure.
Some examples of peak stresses include thermal stresses in a bimetallic
interface, thermal shock stresses (or stresses due to rapid change in the
temperature of the contained fluid), and stresses at a local structural
discontinuity.
Within the context of local primary membrane stress, PL, as well as
secondary stress, Q, the discontinuity effects need not be elaborated. The
structural discontinuity can be either gross or local. Gross structural
discontinuity is a region where a source of stress and strain intensification
affects a relatively large portion of the structure and has a significant effect
on the overall stress or strain pattern. Some of the examples are head-to-
shell and flange-to-shell junctions, nozzles, and junctions between shells of
different diameters or thicknesses.
Local structural discontinuity is a region where a source of stress or
strain intensification affects a relatively small volume of material and does
not have a significant effect on the overall stress or strain pattern or on the
structure as a whole.
The stress classifications for various parts of a pressure vessel are
indicated in Table 4.1 and are reproduced from the American Society of
Mechanical Engineers (ASME) Boiler and Pressure Vessel Code Sections III
and VIII.1 It can be observed that the membrane stress is considered
primary for mechanical loads. For a number of geometries and loading
situations, the bending stress is considered secondary. The bending stress is
considered primary when the net section experiences the applied bending
moment.

4.4 Stress limits

The allowable stresses (or more correctly the stress intensities) in pressure
vessel design codes such as the ASME Boiler and Pressure Vessel Code are
not expressed in terms of the yield strength or the ultimate strength but
instead as multiples of tabulated design value called the design stress
intensity (denoted for example as Sm). This value is typically two-thirds of
the yield strength of the material or for other cases one-third of the ultimate

Copyright 2005 by CRC Press, Inc. All Rights Reserved.

 Table 4.1 Classification of stresses

Vessel part Location Origin of stress Type of stress Classification

Cylindrical or Remote from Internal pressure General membrane Pm

spherical shell discontinuities Through thickness gradient Q
Junction with head or Axial thermal gradient Membrane Q
flange Bending Q½2
Any shell or head Any section across External load or moment, or General membrane averaged Pm
entire vessel internal pressure across full section
External load or moment Bending across full section Pm
Near nozzle or other External load or moment, or Local membrane PL
opening internal pressure Bending Q
Peak (fillet or corner) F
Any location Temperature difference between Membrane Q
shell and head Bending Q
Dished head or Crown Internal pressure Membrane Pm
conical head Bending Pb
Knuckle or junction to Internal pressure Membrane PL½3
shell Bending Q
Flat head Center region Internal pressure Membrane Pm
Bending Pb
Junction to shell Internal pressure Membrane PL
Bending Q½2

Notes:
½1
Q and F classification of stresses refers to other than design condition.
½2
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.
½3
Consideration should also be given to the possibility of wrinkling and excessive deformation in vessels with large diameter to thickness ratio.

Copyright 2005 by CRC Press, Inc. All Rights Reserved.

 strength. Therefore we have a factor of safety of 1.5 or 3 in terms of yield
strength or ultimate strength, respectively. It is the purpose of the design
codes that these multiples of either the yield or the ultimate strength are
never exceeded in design.
The pressure vessel design codes often make specific recommendations
on the limits depending on the conditions (or situations) of design. One
typical such classification is in terms of design, normal, and upset (levels A
and B), emergency (level C), faulted (level D) and test loadings, and
accordingly limits are set appropriately. These stress limits have been
discussed in some detail in Chapter 3. As an example, for design conditions
the limits for the general primary membrane stress intensity, Pm, the local
primary membrane stress intensity, PL, and the combined membrane and
bending stress intensity Pm (PL) þ Pb, are typically expressed as:

Pm  Sm
PL  Sm ð4:3Þ
PL þ Pb  1:5Sm

These limits are sometimes higher than the actual operating conditions.
It is the intent of the design code that the limit on primary plus secondary
stresses be applied to the actual operating conditions. For normal and upset
conditions (sometimes indicated as levels A and B), the range of primary
and secondary stresses, PL þ Pb þ Q is not allowed to exceed 3Sm, or

 
PL þ Pb þ Q  3Sm ð4:4Þ
range

A word of caution is needed here. For example, a stress limit on

some of the combination of stress categories such as Pm(PL) þ Pb, PL þ
Q needs to be carefully understood. The confusion arises because of the
tendency to denote the stress intensity in a particular category by the
symbol of that category, for example P is the stress intensity for the
primary bending stress category. However (PL þ Pb þ Q) is not the
sum of the individual components of primary membrane, primary
bending and secondary stress intensities. It is in fact the stress intensity
evaluated from the principal stresses after the stresses from each
category have been added together in the appropriate manner (that is
not by adding the stress intensities). The primary plus secondary stress
limits are intended to prevent excessive plastic deformation leading to
incremental collapse, and to validate the application of elastic analysis
when performing the fatigue evaluation. The limits ensure that the
cycling of a load range results in elastic response of the material, also
referred to as shakedown (when the ratcheting stops).

Copyright 2005 by CRC Press, Inc. All Rights Reserved.

 4.5 Special stress limits
The theories of failure have been outlined in Chapter 3, the important ones
being the von Mises and the Tresca theories. However, none of the theories
indicate any limit to the magnitude of the principal stresses, as long as their
differences are within the specified limits (stress intensity limits). For
uniform triaxial compressive stresses such a position is appropriate.
However, for uniform triaxial tensile stresses, failures have been observed
to occur, the predictions from the failure theories being that failures would
not occur at all. From the available experimental data, it seems reasonable
that limiting the mean of the principal stresses to the yield strength would
ensure an adequate safety margin against failure, that is

1 þ
2 þ
3
< Sy ð4:5Þ
3

The ASME Boiler and Pressure Vessel Code takes a somewhat

conservative estimate by limiting the mean of the principal stresses to 8/9
Sy, or

1 þ
2 þ
3 8
< Sy ð4:6Þ
3 9

with an assumed value of the design stress intensity Sm equal to two-thirds

Sy, or

2
Sm ¼ Sy ð4:7Þ
3

The limit on the sum of the principal stresses becomes equal to 4Sm or

1 þ
2 þ
3 < 4Sm ð4:8Þ

4.6 Practical aspects of stress categorization

Hechmer and Hollinger have proposed ten guidelines on the evaluation of
stresses in pressure vessels calculated by finite-element method in the spirit
of the ASME Boiler and Pressure Vessel Code.3

1. Guideline 1 establishes the relationship between the membrane and

bending stresses and the associated failure modes. The failure
modes of concern are collapse/gross distortion (Pm), plastic collapse
(Pb), excessive plastic deformation (PL þ Pb), and ratcheting and lack
of shakedown (P þ Q).
2. Guideline 2 relates to the first guideline by establishing the finite-
element assessment for each failure mode. They maintain that for the

Copyright 2005 by CRC Press, Inc. All Rights Reserved.

 general primary membrane stress, Pm, finite-element analysis is not
necessary, and simple equilibrium equations for net force and net
moment are adequate. Finite-element analysis is an appropriate tool
for evaluating the local primary membrane stress intensity, PL, the
primary membrane plus bending stress intensity, PL þ Pb, and the
primary plus secondary stress intensity, P þ Q. For bending of shell
cross-sections (as shown in Figure 4.1), simple equations as in Eqs.
(4.9) and (4.10) below are proposed

p
Pm ¼ ð4:9Þ
A
6m
Pb ¼ ð4:10Þ
t2

Here p is the load and m the bending moment per unit length of the
shell.
3. Guideline 3 defines stress classification lines (SCL) and stress
classification planes (SCP) for the purpose of evaluating membrane
and bending stresses. An SCL is shown in Figure 4.1.

Figure 4.1 Stress classification line (SCL).

Copyright 2005 by CRC Press, Inc. All Rights Reserved.

 4. Guideline 4 establishes the global locations for assessment of
stresses, and states that the general primary membrane stress
intensity, Pm, should be evaluated remote from a discontinuity;
whereas the primary membrane plus bending stress intensity, PL þ
Pb, and primary plus secondary stress intensity, P þ Q, should be
evaluated at a discontinuity.
5. Guideline 5 establishes the criteria for local locations in terms of the
SCL and SCP, and the orientation for the SCL–SCP caused by a
discontinuity or blend radius. The blend radii are normally included
in the finite-element model.
6. Guideline 6 provides the definition (rewording the code definition)
of linearized stress as the stress represented by linear distributions
which develop the same net forces and moments on a section as the
total stress distribution.
7. Guideline 7 provides procedure for calculating membrane and
bending stresses from component stresses, and not principal
stresses. Linearized stresses are based on explicit computations by
Eqs. (4.9) and (4.10) and not based on arbitrarily linearizing
individual stress components.
8. Guideline 8 indicates the method for calculating principal stresses,
stress intensities, and stress ranges. All six components of the
stresses are linearized individually and then the principal stresses
are calculated for the membrane stresses. For the bending stress only
two of the component stresses are linearized, leaving the bending
stress due to shear and through-thickness components.
9. Guideline 9 provides recommendation on the use of SCL and SCP
for various geometries and states that for most axisymmetric
situations, SCLs are appropriate. SCPs are recommended for special
cases, such as flat plate with penetrations, and are deemed
appropriate where the geometry has a well-defined plane that can
be directly related to the failure mode.
10. Guideline 10 provides six fundamental recommendations on the
evaluation of stresses by finite elements, which are:
a. The FEA modeling techniques (mesh refinement, etc.) should be
adequate for the level of accuracy needed for structural evalua-
tion.
b. The finite-element nodes should be such that the location of the
SCLs can be readily established.
c. An SCL or SCP may start or end at a singularity, because the
integration of the loads along the line or the plane may override
the effect of the singularity.
d. Along discontinuities it is desirable to use an equilibrium type of
analysis to obtain more accurate results.
e. The failure locations and hence the locations of SCL and SCP
should be established based on an overall review of the flow of
stresses.

Copyright 2005 by CRC Press, Inc. All Rights Reserved.

 f. The methods recommended above generally apply for shells
with R/t ratios greater than 4. For lower ratios of R/t, the level of
accuracy is questionable.

4.7 Shape factor considerations

The primary stress intensity in the ASME Boiler and Pressure Vessel Code
is intended to prevent uncontrolled plastic deformation and to provide a
nominal factor of safety on the ductile burst pressure. These limits are
based on the principles of limit design. The material is assumed to be
elastic–perfectly plastic. For a straight bar in tension, a load producing
yield stress, Sy, results in a collapse. If it is loaded in bending, collapse
does not occur until the yield moment has been increased by the shape
factor of the section.
The shape factor, , is defined as the ratio of the load set producing a
fully plastic section to the load producing initial yielding of the extreme
fibers of the section. The shape factor for a rectangular section in pure
bending is 1.5. The current stress intensity limits in the ASME Code rules
are based on rectangular cross sections. For combined axial and bending
loads, the load set to form a ‘‘plastic hinge’’ depends on the ratio of the
tensile and bending loads.
An interaction curve as shown in Figure 4.1 is for a rectangular section.
Note that when the membrane stress, Pm, is zero, the stress calculated
elastically from the collapse moment for bending is 1.5Sy. The factor 1.5 is
the shape factor for the rectangular cross section. It should be noted that
the current code limits are nonconservative for some sections. Such
nonconservatism arises typically for sections with shape factors lower
than 1.5 (such as an I-section with a thin web). Chattopadhyay has
recommended design equations for different types of loadings to provide
adequate safety for all combinations of axial and bending loads.4 The
proposed limits in Eqs. (4.11) to (4.13) are intended to replace the existing
ones in the ASME Code, and apply to design, level C and testing limits as
provided in Eqs (3.6) to (3.8) and shown pictorially in Figure 4.2. The
proposed limits are shown below.

1. Design condition:

Pm  Sm
ð4:11Þ
Pm þ Pb  Sm

Copyright 2005 by CRC Press, Inc. All Rights Reserved.

 Figure 4.2 ASME Code limits.

2. Testing condition:

Pm  0:9Sy
Pm þ Pb  0:9Sy ; forPm  0:67Sy
ð4:12Þ
Pm þ Pb  ½ð0:9þ0:1  Þ=ð1  ÞSy  ½ð  1Þ=ð1  ÞPm
for 0:67Sy  Pm  0:9Sy

3. Level C condition:

Pm ¼ Sy
PL þ Pb ¼ S; for PL ¼ 0:Sy
ð4:13Þ
PL þ Pb ¼ ½ð  Þ=ð1  ÞPL þ ½ð  Þ=ð1  ÞS
for Sy ¼ PL ¼ 0:Sy

where S is greater of 1.2Sm or Sy. The value of  is 1/ for full

sections identified as ones for which no abrupt changes in boundary
occur.

Copyright 2005 by CRC Press, Inc. All Rights Reserved.

 For non full sections such as the I-sections and T-sections the values of 
and  have been calculated by Chattopadhyay4 and are:

 I Section:  ¼ 1.2,  ¼ 0.55

 T Section:  ¼ 1.7,  ¼ 0.75

The procedure for obtaining  and  will be discussed later.

Figures 4.3 to 4.6 give the interaction curves for circle, diamond, I- and
T-sections respectively and have been obtained using techniques outlined
in Appendix C. These interaction curves represent the upper limits of (Pm þ
Pb) as a function of Pm. It is possible to observe certain characteristics in all
these figures. At Pm ¼ 0, the limiting value of (Pm þ Pb) is the shape factor of
the cross-section times the yield strength (Sy). Then (Pm þ Pb) increases
with Pm, reaches a peak and then drops to Sy when Pm reaches Sy. The
peak value of (Pm þ Pb) is strongly dependent on the geometry of the

Figure 4.3 Interaction curve for a circular cross-section.

Copyright 2005 by CRC Press, Inc. All Rights Reserved.

 Figure 4.4 Interaction curve for a diamond cross-section.

cross-section. Another feature that is noteworthy is that different cross-

sections can have identical shape factors, but differ significantly in the
interaction characteristics (see Figures 4.3 and 4.6).
For rectangular cross-sections, a closed-form mathematical expression
describes the interaction curve. For an arbitrary cross-section, no such
expression exists, and a large number of computations are necessary to
obtain the interaction curve at discrete values of Pm and (Pm þ Pb). For
certain sections where no abrupt changes in boundary occur, the interaction
curves may be approximated by analytical expressions involving the shape
factor, . Such cross sections may be referred to as ‘‘full’’ sections. (Note
that an I- or a T-section cannot be called a ‘‘full’’ section). Examples of full
sections include rectangular, circular, diamond and trapezoidal sections.
The interaction curve for a rectangular section as shown in Figure 4.2
can be mathematically described as

Copyright 2005 by CRC Press, Inc. All Rights Reserved.

 Figure 4.5 Interaction curve for an I-section.

2 !2 3
Pm þ Pb P m 5 þ Pm
¼ 1:541  ð4:14Þ
Sy Sy Sy

See References 5 and 6 for examples.

For nonrectangular sections, we can tentatively replace 1.5 and check the
validity of this substitution later. We therefore have
2 !2 3
Pm þ Pb P m 5 þ Pm
¼ 4 1  ð4:15Þ
Sy Sy Sy

Using the notation

Pm þ Pb
y¼
Sy

and

Pm
x¼
Sy

Copyright 2005 by CRC Press, Inc. All Rights Reserved.

 Figure 4.6 Interaction curve for a T-section.

Eq. (4.15) reads

y ¼ ð1  x2 Þ þ x ð4:16Þ

y is a maximum when dy/dx ¼ 0, which gives

1 1
ymax ¼  þ at x ¼ ð4:17Þ
4 2

Furthermore,

1
y ¼  when x ¼ 0; ð4:18Þ


Copyright 2005 by CRC Press, Inc. All Rights Reserved.

 The results from Eqs. (4.17) and (4.18) may be used to define domains of
applicability of various stress limits. The interaction curves based on exact
methods using Appendix B and those using Eq. (4.15) are shown in Figure
4.7 for the circular and diamond sections. The exact and the approximate
solutions are not too different from each other.
For sections that are not full, the above approximations to the interaction
curves cannot be used, and each point on the interaction curve has to be
obtained explicitly. For such sections, Eq. (4.18) is replaced with

y ¼  when x ¼ 0;  ð4:19Þ

The value of  has to be evaluated numerically.

With these modifications, the revised limits given by Eqs. (4.11) to (4.13)
are graphically displayed in Figures 4.8 and 4.9.

Figure 4.7 Comparison of exact and approximate interaction curves.

Copyright 2005 by CRC Press, Inc. All Rights Reserved.

 Figure 4.8 Proposed stress limits for rectangular cross-sections.

Figure 4.9 Proposed stress limits for an I-section.

Copyright 2005 by CRC Press, Inc. All Rights Reserved.

 References
1. American Society of Mechanical Engineers, Boiler and Pressure Vessel Code,
ASME, New York.
2. Pastor, T.P., and Hechmer, J.L., ASME Task Group Report on Primary Stress,
J. Pressure Vessel Technol., 119, 1997.
3. Hechmer, J.L., and Hollinger, G.L. Three-dimensional stress criteria –
summary of the PVRC project, J. Pressure Vessel Technol., 122, 105–109, 2000.
4. Chattopadhyay, S., Allowable stresses for nonrectangular sections under
combined axial and bending loads, J. Pressure Vessel Technol., 110, 188–193,
1988.
5. Seely, F.B., and Smith, J.O., Advanced Mechanics of Materials, 2nd ed., Wiley,
New York, 1952.
6. Hodge, P.G., Plastic Analysis of Structures, McGraw-Hill, New York, 1959.

Problems
1. Determine the interaction curve for a thin-walled I-beam of height h
and width h/2 and having a wall thickness of t (t<<h).
2. Determine the interaction curve for a thin-walled T-beam of height h
and width h/2 and having a wall thickness of t (t<<h).
3. Determine the interaction curve for a thin-walled circular beam of
radius R and a wall thickness of t (t<<h).

Copyright 2005 by CRC Press, Inc. All Rights Reserved.