Proceedings

Proceedings of PVP2005
of PVP2005
2005
2005 ASME
ASME Pressure
Pressure Vessels
Vessels and Piping
and Piping Division
Division Conference
Conference
July
July 17-21,
17-21, 2005,
2005, Denver,
Denver, Colorado
Colorado USA USA

PVP2005-71535
PVP2005-71535

CLASSIFICATION OF SHELL STRESSES

RESULTING FROM PIPING NOZZLE LOADS

Trevor G. Seipp Shiju V. P. George

Fluor Canada Ltd. Fluor Canada Ltd.
55 Sunpark Plaza, S.E. 55 Sunpark Plaza, S.E.
Calgary, AB CANADA T2X 3R4 Calgary, AB CANADA T2X 3R4
Tel: 403-537-4509 Tel: 403-537-4095
Fax: 403-537-5371 Fax: 403-537-5371
Email: [email protected] Email: [email protected]

Shawn W. Morrison
M5 Engineering Inc.
Calgary, AB CANADA
Tel: 403-831-1917
Email: [email protected]

ABSTRACT thermal loads likely follow Curve B; that is, somewhere between
In a previous paper, the authors discussed the classification Curve A (loads that cause primary stress) and Curve C (loads that
of membrane stresses caused by piping loads resulting from cause secondary stress). As there are only two options allowed by
restrained free thermal expansion. In that paper the authors the Vessel Code to classify membrane stresses resulting from
proposed that although these stresses exhibit both primary and piping thermal loads, a choice must be made between whether it is
secondary characteristics, it is more appropriate to classify the more appropriate to classify them as primary or secondary.
resulting stresses as primary.
In this paper, the authors present further FEA-based
justification of this hypothesis. First, the vessel-nozzle Curve C
intersection is modeled with 3D solid elements. Next, an elastic-
plastic analysis of the previously examined piping/vessel layout is
presented, with an examination of the final strain levels. Finally,
a shakedown evaluation is performed. Φ
The results of these evaluations show that, unless additional
analysis indicates otherwise, the membrane stresses resulting from
restrained free thermal expansion piping loads should be classified Curve B
Load

as primary.
Curve A
BACKGROUND
In George, Seipp and Morrison [9], it was proposed that it ∆
would be more appropriate to classify the membrane stresses
caused by piping loads resulting from restrained free thermal
expansion as primary. In that paper, it was discussed how
“thermal stresses” as defined in the Vessel Code [1,2] do not arise
from “thermal displacement” loads as defined in the Piping Code
[4,5]. Common practice has resulted in the categorization of shell Deflection
stresses arising from thermal displacement loads to be thermal. In Figure 1. Functional relationship between loads and displacement.
turn, these thermal stresses are classified as secondary in the Loads that develop primary stresses (Curve C), loads
Vessel Code. that develop secondary stresses (Curve A) and piping
However, as shown in the previous paper [9], membrane thermal loads (Curve B).
stresses caused by restrained free thermal expansion also have
primary stress characteristics. As shown in Figure 1, piping

1 Copyright © 2005 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 05/17/2014 Terms of Use: http://asme.org/terms

 The membrane stresses noted above are only in reference to Immediately prior to the nozzle, the pipe reduces to a NPS 12
the vessel shell, as the Vessel Code already provides for the nozzle. The layout of the pipe, along with the locations of
classification of these stresses in the nozzle. Article 4-138 [2] supports, guides, and anchors is shown in Figure 2.
refers to the evaluation of stresses in the nozzle neck. The pressure vessel is a typical tower in a refinery or
petrochemical installation. It was designed in accordance with
Within the limits of reinforcement given by AD-540.2 whether ASME Section VIII – Division 1 [1]. It is 2.44 m (96 in) ID, and
or not nozzle reinforcement is provided, the Pm classification is 14 mm (0.55 in) thick. As stated above, the nozzle is NPS 12. A
applicable to stress intensities resulting from pressure induced
57 mm (2.25 in) thick nozzle forging was chosen to meet area of
general membrane stresses as well as stresses, other than
discontinuity stresses, due to external loads and moments replacement requirements.
including those attributable to restrained free end
displacements of the attached pipe. [2] RESULTS: 3D LIMIT LOAD ANALYSIS
A 3D finite element model of the vessel, nozzle, and piping
Also, within the limits of reinforcement, a PL classification configuration was created. Solid quadratic brick elements (20-
shall be applied to local primary membrane stress intensities node hexahedral) were used. Three of these elements were used
derived from discontinuity effects plus primary bending stress through the thickness in the vessel and the pipe. This was
intensities due to combined pressure and external loads and
expected to give reasonable results. Element aspect ratios were
moments including those attributable to restrained free end
displacements of the attached pipe. [2] kept small in the region of the nozzle-shell intersection; however
they were allowed to become large away from this location. An
PURPOSE elastic-perfectly-plastic material model was used. Geometric non-
It is the intent of this paper to establish how stresses linearities were not included.
developed by thermal piping loads are to be classified, given that
the Vessel Code only offers two choices: primary or secondary.

SCOPE
A finite element model of a typical pressure vessel and
piping layout, the same as in the previous paper [9], was modeled.
The piping layout included several supports, guides, and a
directional anchor attached to a radial nozzle on a vertical vessel.
Reasonable temperatures and pressures were applied.
First, a 3D elastic-plastic model using a limit load approach
was evaluated. In this analysis, the attached piping system was
directly modeled to capture the functional relation of how the
applied nozzle loads decayed with deformation in the vessel shell.
Design temperatures and pressures were applied to evaluate the
system response in terms of strain at the nozzle-to-shell interface.
Second, the same 3D model was evaluated using a finite
tangent modulus including geometric non-linearities. Lastly, a
shakedown analysis of this system was performed.
The results of these evaluations were compared to the results Figure 3. Finite element model of the piping system and the
from the previous paper to establish if it is more appropriate to pressure vessel.
classify thermal piping load membrane stresses as secondary or
primary. Temperature, pressure, and gravity (acceleration) loads were
applied incrementally, according to standard limit load analysis
procedures. If the finite element model converged, then plastic
collapse was not expected to occur. For loads that typically
follow Curve C in Figure 1, such as pressure and gravity, lower-
bound plastic collapse behavior is well documented. To maintain
a 1.5 design margin on plastic collapse, the collapse stress used in
this analysis was set equal to the allowable stress of 124.8 MPa
(18,100 psi), which is 2/3 of the yield stress [3].
The limit load analysis converged, indicating that lower
bound plastic collapse does not occur at the design conditions.
The magnitude of the final local plastic strains indicated that there
was substantial local deformation. At the nozzle-to-shell junction,
the maximum total strain intensity was 1.43%, as shown in
Figure 4.
Figure 2. Layout of the piping system. One additional observation regarding the results of this
analysis would be that the total strain intensity at the nozzle-to-
DESCRIPTION OF THE MODEL GEOMETRY shell junction for the 3D solid element analysis is approximately
The pipe is 16” NPS, Sch. XS fabricated from A106-B 40% of what was predicted using a shell element approach.
material. The design pressure for the entire system was 1.03 MPa
(150 psi), and the design temperature was 371°C (700°F).

2 Copyright © 2005 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 05/17/2014 Terms of Use: http://asme.org/terms

 removed 10 times to determine if the plastic strain at the end of
each load application continued to increase or ultimately reaches a
final value.
The final strain for each load application is shown in Table 1.
From these results, the strains asymptotically approach a limit.
The actual strain intensity for the last step is still quite significant
at 1.16%.

Table 1. Shakedown analysis strain results.

Load Application Loading strain (%) Unloading strain
(%)
1 1.0083 0.7909
2 1.0994 0.8720
3 1.1293 0.9021
4 1.1429 0.9154
5 1.1498 0.9219
6 1.1535 0.9255
Figure 4. Total strain intensity at the nozzle-to-shell junction for
7 1.1556 0.9276
the 3D limit load analysis.
8 1.1569 0.9289
9 1.1577 0.9298
RESULTS: 3D ELASTIC-PLASTIC ANALYSIS 10 1.1583 0.9305
Using the same model from the 3D limit load analysis, the
effects of both strain-hardening and geometric non-linearities DISCUSSION
were evaluated. A bilinear kinematic material model with two In this paper, the authors chose to perform a number of non-
different non-zero tangent moduli (1 GPa and 10 GPa) was linear evaluations. Each of these evaluations increased in level of
examined. These values were considered appropriate for a engineering sophistication to better understand how membrane
Young’s modulus of 210 GPa. Temperature, pressure, gravity stresses resulting from thermal piping loads should be classified.
loads and yield stress were applied as above. The Vessel Code implies a limit on strain of 0.4% based on a
linear-elastic approach [9]. The observed strains from these non-
linear analyses were greater than this limit. Since these strains
exceed the implied linear-elastic limit, the corresponding stresses
have primary characteristics. However, neither collapse nor
ratcheting was observed. This result would imply that the
membrane stresses also have secondary characteristics. Therefore,
membrane stresses developed from thermal piping loads exhibit
characteristics of both a primary and secondary nature.
It is important to appreciate that stress classification is used
only when performing a linear-elastic assessment, such as those
presented in WRC-107 [6, 7] or WRC 429 [8]. The Vessel Code
offers only two options for classifying membrane stresses:
primary or secondary. Given that membrane stresses developed
from thermal piping loads exhibit characteristics of both a primary
and secondary nature, conservatism requires that these membrane
stresses be classified as primary if only a linear elastic evaluation
is performed. That is, if the membrane stresses pass a 1.5Sm
check, with the thermal piping load membrane stresses included,
Figure 5. Total strain intensity at the nozzle-to-shell junction for and the combined membrane-plus-bending stresses pass a 3Sm
the 3D elastic-plastic analysis. check, the Vessel Code minimum required evaluation would be
satisfied.
With the 1 GPa tangent modulus, the maximum observed As a consequence of this classification choice, recall that
strain intensity was 1.35%, not significantly different from the when this configuration was evaluated using both a WRC 107 and
perfectly-plastic model. However, with a 10 GPa tangent linear-elastic FEA approach [9], the shell stresses were
modulus, the maximum total strain intensity was noticeably lower unacceptably high. Consequently, the vessel would need to be
at 1.00%, as shown in Figure 5. Regardless of the tangent provided with additional reinforcement, or the piping
modulus used in the analysis, the magnitude of the final local configuration re-routed.
plastic strains indicates that there is substantial local deformation. The vessel engineer is always permitted to perform
additional analyses beyond the minimum required by the Vessel
RESULTS: 3D ELASTIC-PLASTIC SHAKEDOWN Code, such as those offered in this paper, to mitigate the
ANALYSIS consequences of not acknowledging the secondary characteristics
In order to evaluate whether or not the current configuration of the thermal piping loads. It should be noted that the effort to
shakes down to elastic action (i.e. does not ratchet), a shakedown perform these additional analyses may be significantly greater
analysis was performed. The thermal loads were applied and than the effort required to perform the linear-elastic assessment.

3 Copyright © 2005 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 05/17/2014 Terms of Use: http://asme.org/terms

 For example, a WRC-107 evaluation may take 15 minutes, REFERENCES
whereas to perform elastic-plastic and ratcheting analyses may 1. ASME, 2004, ASME Boiler & Pressure Vessel Code, Section
take more than 40 hours. Furthermore, there is no guarantee that VIII – Division 1, American Society of Mechanical
the additional engineering effort will result in a more favourable Engineers, New York, NY.
conclusion. Therefore, the vessel engineer must appropriately
weigh the costs and risks of all the alternatives before deciding on 2. ASME, 2004, ASME Boiler & Pressure Vessel Code, Section
a course of action. VIII – Division 2, American Society of Mechanical
Engineers, New York, NY.
CONCLUSION
In the author’s previous paper [9] it was proposed that the 3. ASME, 2004, ASME Boiler & Pressure Vessel Code, Section
membrane stresses developed from all piping loads, be classified II, Part D, American Society of Mechanical Engineers, New
as primary for the purposes of evaluating the corresponding shell York, NY.
stresses. In this paper, the authors performed a number of non-
linear evaluations to further explorer the validity of the conclusion 4. ASME 2002, ASME Code for Process Piping B31.3,
reached in that paper. American Society of Mechanical Engineers, New York, NY.
These additional evaluations demonstrate conclusively that
5. ASME 2002, ASME Code for Power Piping B31.1,
stresses caused by piping thermal loads have both primary and
American Society of Mechanical Engineers, New York, NY.
secondary characteristics.
However, the Vessel Code provides only two options for 6. Wichman, K.R., A.G. Hopper, and J.L. Mershon, March
classification of the membrane stresses caused by piping thermal 1979 Revision of August 1965, Local Stresses in Spherical
loads: primary or secondary. Based on the results from this paper, and Cylindrical Shell due to External Loadings, WRC
to satisfy Vessel Code minimum requirements, it is required to Bulletin 107, Welding Research Council, New York, NY.
classify these membrane stresses as primary if only a linear-elastic
assessment is performed. 7. Mershon, J.L., et al., August 1984, Local Stresses In
On a case-by-case basis, the vessel engineer is always Cylindrical Shell Due To External Loadings on Nozzles –
permitted to perform additional analyses beyond the minimum Supplement to WRC Bulletin No. 107, WRC Bulletin 297,
required by the Vessel Code in order to mitigate the consequences Welding Research Council, New York, NY.
of not acknowledging the secondary characteristics of the thermal
piping loads. 8. Hechmer, J.L., and G.L. Hollinger, February 1998, 3D Stress
Based on the conclusions reached in this paper, the authors Criteria Guidelines For Application, WRC Bulletin 429,
recommend that the Vessel Code [2], article 4-112(i) be modified Welding Research Council, New York, NY.
as follows:
9. George, S.V.P., T.G. Seipp, and S.W. Morrison,
…. An example of local primary membrane stress is the “Classification of Thermal Piping Loads Using Limit Load
membrane stress in a shell produced by external load and Analysis”, PVP2004-2613, Design and Analysis of Pressure
moment at a permanent support or (Add) those caused by Vessels, Heat Exchangers and Piping Components, PVP Vol.
restrained free thermal expansion of the attached piping 477, ASME, New York, NY.
at a nozzle connection.

4 Copyright © 2005 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 05/17/2014 Terms of Use: http://asme.org/terms