Вы находитесь на странице: 1из 195

Release Notes

Paulin Research Group


August 1, 2014

1.0 Introduction:
This document contains 2014 update feature descriptions and detailed examples for FE/Pipe Version 7.0
and NozzlePRO Version 9.0. The programs have each been updated to the 2013 Code Year for ASME
Section VIII Division 2. Section 7.0 describes in detail the ASME Section VIII Division 2 approach for elastic
analysis. Automated models for pipe shoes, in-line pipe anchors, and vessel saddles produce allowable
loads and flexibilities and assist in the design selection. Major new feature explanations are in Section 3.0
and detailed application examples are in Section 5.0.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

Table of Contents
1.0 Introduction: .................................................................................................................................................1
1.1 FE/Pipe 7.0 and NozzlePRO 9.0 Update Feature List ....................................................................................4
1.2 NozzlePRO V.9.0 Specific Features ...............................................................................................................5
1.3 FE/Pipe V.7.0 Specific Features .................................................................................................................... 5
1.3.1 Example Files .........................................................................................................................................6
2.0 Highlighted Feature Descriptions .................................................................................................................7
3.0 New Feature Explanations: ........................................................................................................................... 9
3.1 ASME Section VIII Div 2 2013 Changes: ....................................................................................................9
3.2 Allowable Loads and SIF Calculations for Pipe Shoes, In-Line Anchors and Vessel Saddles ..................17
3.3 Measurable First Crack Appearance Calculator .....................................................................................20
3.4 3D Viewer Updates .................................................................................................................................23
3.5 Sustained Stress Indices for B31.3 2010 and Later Code Years..............................................................30
3.6 Shell Mode Calculator.............................................................................................................................33
3.7 Percentage Difference Plots ...................................................................................................................34
3.8 Seismic Load Calculator ..........................................................................................................................35
3.9 Alternate Weld Model ............................................................................................................................44
3.10 NozzlePRO Notes ..................................................................................................................................46
3.11 Updated NozzlePRO Printed Reports ...................................................................................................49
4.0 Minor Feature Additions: ........................................................................................................................... 51
5.0 APPLICATIONS:............................................................................................................................................53
5.0.1 Table of Contents ................................................................................................................................ 53
5.1 First Observable Crack in Pipe Shoe .......................................................................................................54
5.2 Pipe Shoe and Saddle Design and Analysis.............................................................................................59
5.3 Modeling of In-Line Pipe Anchors...........................................................................................................62
5.4 Drainage Tank Seismic Load Example.....................................................................................................64
5.5 Vessel Loading History ............................................................................................................................69
5.6 Horizontal Vessel Loading Example ........................................................................................................70
5.7 Hillside Nozzle Option in Tank Template ................................................................................................91
5.8 Design, Operating and Occasional Pressure Differences in VIII-2 ..........................................................94
5.9 Pressure Stiffening in Horizontal Vessels ...............................................................................................96
5.10 Effect of Rings on Stresses in Large Diameter Nozzles .......................................................................100
5.11 Saddles on Pressure Vessels and Pipe ................................................................................................101
5.11.0 Some Saddle Analysis Observations: ...........................................................................................109
5.11.1 Axial, Vertical, In-Plane, Out-Plane and Torsional Orientations ..................................................110
5.11.2 Conclusions Regarding Seismic Designs: .....................................................................................115
5.12 Brownell & Young Horizontal Vessel Model .......................................................................................116
5.13 Calculating Natural Frequencies in Horizontal Pressure Vessels........................................................119
5.14 Participation Factors Simple Example .............................................................................................139
5.15 In-Line Anchor Analysis.......................................................................................................................142
5.16 Wind Vibration....................................................................................................................................151
5.17 Pressure Stiffening of Spherical Geometries ......................................................................................156
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

6.0 Beam Load Displacement Relationships...................................................................................................158


7.0 ASME Section VIII Division 2 Elastic Stress Rules and Discussion.........................................................159
8.0 ASCE 7-10 Notes .......................................................................................................................................179
8.1 Time history plots from USGS website .................................................................................................182
8.2 Comparison of Response Spectrum and Static Seismic Loads .............................................................191
9.0 ASME VIII-2 Part 5 Plastic Analysis ...........................................................................................................194

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

1.1 FE/Pipe 7.0 and NozzlePRO 9.0 Update Feature List


1)Modifications to accommodate ASME Section VIII Division 2 2013 Code Year Changes
2)Allowable Load, SIF and Flexibility Calculations for Pipe Shoes, In-Line Anchors and Vessel Saddles
3)Measurable crack appearance calculator
4)Updated NozzlePRO Report Print Control
5)Mass Participation Factor Calculation for all Natural Frequency Calculations
6)Shell mode calculator for FEPipe and NozzlePRO
7)Seismic Load Information calculator for FEPipe and NozzlePRO
8)Percent Difference Plots for averaged and not-averaged stresses to aid convergence determination
9)Modified Load Case Options for Factored Primary Loads in VIII-2 Part 5.
10)Updated 3D Viewer Stress Labeling and Lighting Enhancement
11)3D Viewer Displacements for All Deflected Shapes
12)Zick Calculations and Estimates for in-line pipe anchors and pressure stiffening
13)Hot Formed Welding Tee Thickness Option for PCL Pipe Stress Analysis
14)Sustained Stress Index (SSI) Calculator for Sustained stress indices in B31.3 after 2010.
15)Update tabular material properties output in all FE/Pipe models
16)Liquid Dynamic Load Options
17)Hillside Nozzle Option in Tank Nozzle Template
18)Refinement in plate/vessel penetration stress calculations
19)Automated ring placement on cylindrical shells (internal or external)
20)Enhanced Stress Corrolations for Nozzles in Cylinders
21)AVI file created from viewer animation
22)Sustained Stress Indices (SSIs) and Sustained Stress Exponent Estimates on all SIF reports

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

1.2 NozzlePRO V.9.0 Specific Features


Internal Clips / Lugs
Rings on Vessel and Pipe Shoe Models (Internal or External)
Zick Calculations for Saddles Including Estimates for In-Line Pipeline Saddles and Pressure Stiffening
Allowable Load and SIF Calculations for Saddles and Pipe Shoes
Seismic Load, Shell Mode and First Measured Crack Calculator Buttons
Force Backside models in large D/T geometries
Update Report Print Control
All SIF Reports include Sustained Stress Indices and Exponent Estimates
Default Length Control Options
Additional Modifications:
1. Fix for errant message when the load transformation routine had a problem. This was caused
before plot or run without giving the user much guidance to what was the real problem.
2. Security check bypassed before plot or run.
3. Tilt angle added for structures on heads. Before the default was always 90 deg. Now the user can
pick between 0 and 90. There is no option for in between (future version if requested by users)
4. Added full 360 models for cyl-cyl brick models.
5. Added capability to free top or bottom end of cone, previously only available for cylinders.
6. Allowable loads are included in the results reports.

1.3 FE/Pipe V.7.0 Specific Features


New 3-D Viewer Features
Modal Participation Factor Calculations For Seismic Stress Evaluations
Occasional Load Combination Option Updates
Allowable Primary Stress Calculations Updated per ASME VIII-2 2013.
Averaged/NotAveraged Percent Calculation Comparison for Model Validation
Hot Formed Welding Tee Thickness Option for PCL Pipe Stress Analysis
Sustained Stress Index (SSI) added in the PRG iK spreadsheet
Tabular Material Properties in Output
Liquid Load Dynamic Model Options
Pressure Stiffening Example Models
Secondary Stress Categories for Primary and Occasional Load Cases
All SIF Reports include SSIs

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

1.3.1 Example Files


Example model files described in this document are placed in the public documents folder for the computer
system. You will need to enable write access to the folder if you want to analyze any of the models. The
actual folder name varies depending on operating system and locale but it is typically as listed below.
Windows Vista/7/8 C:\Users\Public\Documents\PRG2015\Samples
A Start menu shortcut named Sample files is created during installation in the PRG2015 program group
that will open the folder in Windows Explorer. You can access these files through the shortcut or navigate
to the folder using the environment variable (%PRG2015_Samples%).

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

2.0 Highlighted Feature Descriptions


The following images reflect a few of the features described in these update notes for both FE/Pipe Version
7.0 and NozzlePRO Version 9.0:

Ring Input for Vessels and Pipe Shoes

Allowable External Loads on Pipe Shoes and Saddles

ASME Section VIII Division 2 2013 Edition

Multiple Stress Points, Displacements and Coordinates

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

Initial Crack Prediction

B31.3 Sustained Stress Indices

Cylinder Shell Mode Calculator Hillside Nozzle in Tank

SIFs and ks for Pipe Shoes

Axial Seismic Load on Vessel Saddles

Pressure
Stiffening

VIII-2 Part 5 Discussion

Dynamic Stresses due to


Ground Motion

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

3.0 New Feature Explanations:


3.1 ASME Section VIII Div 2 2013 Changes:
The ASME, or ASME Preferences button from any program can be selected to get to the ASME options
panel. Once the panel data is modified it may be stored locally for the single model, in the current working
folder for all jobs created in the same folder, or globally as default settings. These options are available
from the bottom of the Preferences panel.

The key modified items on the pane are identified as A-D and are described below:

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

A radio button allows the user to select the 2013 version of ASME Section VIII-2. Modifications to the
occasional load factored cases have been implemented along with the use of SPL instead of 1.5Sm for the
primary load allowable. The primary factored load analysis generally only affects occasional load analyses.
The user must remember to:
a)Follow the allowable stress design (ASD) guidance in ASCE 7-10 for occasional loads.
b)Enter 70% of the seismic design load from ASCE 7-10. The ASCE 7-10 document should be reviewed
carefully so that all criteria are satisfied, and so that all factors affecting the design magnitude are properly
considered. Generally, the user only needs to remember to enter 70% of the seismic design load. If the
external loads are provided from a pipe stress analysis, in many cases the loads have already been reduced
by 70% and so the user should identify carefully how the piping loads were developed.
c)Review the estimated over strength factor where applicable from ASCE 7-10 so that geometries where
stresses are highly localized can be carefully inspected or reinforced so that local plastic deformation does
not sacrifice the ability to resist future events or undergo design fatigue loading.
B The operating pressure will always be used for operating and fatigue load cases if an operating
pressure is entered. If not, the design pressure is entered.
C The user can specify the ratio between the design and operating pressures. This is NOT the multiplier
in Table 5.3 for use with occasional load combination cases. The ASME VIII-2 Table 5.3 load cases are shown
on the following page. The cases referenced are cases 6, 7, and 8.
D By default, the 2013 VIII-2 uses k=1.0 for occasional loads per Table 5.3 (See portion of Table 5.3
below.) VIII-2 also instructs the user to use the guidelines in ASCE 7-10, which introduced the factored
loads now used for the ASD (elastic) method in VIII-2. Certain readings of ASCE 7-10 (see Chapter C2
commentary and 12.4.3.3), suggest that in some cases a k factor of 1.2 can be used with particular
occasional load combinations. The radio buttons in the D item frame allow the user to override the
default for ASME VIII-2 2013, and incorporate the use of k=1.2 in any occasional load allowable from the
allowable stress (ASD) design. ASCE 7-10 12.4.3.3 states: Where allowable stress design methodologies
are used in load combinations 5,6, or 8 of Section 2.4.1 allowable stresses are permitted to be
determined using an allowable stress increase of 1.2.

It is common to disagree about the intent of the above sentence, even after following all possible routes
from ASCE 7 Chapters 13 or 15 to arrive at Section 12.4.3.3 in ASCE 7-10. It is however, thought s important
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

10

to reach a concensus about how to proceed for any particular project to achieve the intended degree of
safety needed for that particular project.
On the General screen of the nozzles, plates and shells template, there is a primary load only analysis
drop down YES/NO selection. The default is NO.

If set to yes then only the entered primary and occasional loads will be included in a single load case. No
operating or range checks will be performed for the analysis if this option is set to YES. The purpose of this
calculation option is so that users can more conveniently evaluate the potential variety of occasional load
cases required in Table 5.3 of VIII-2 Part 5. The (8) primary load combinations in VIII-2 Table 5.3 for elastic
Allowable Stress Design analysis (ASD) are shown below:

ASME VIII-2 Part 5 Table 5.3 Excerpt

Case 6 is for uplift loading due to earthquakes or wind on buildings or vessels, and cases (7) and (8) are used
for combinations of live, snow and earthquake or wind loads. Piping and vessels generally have only two
primary load cases:
1)Design pressure + weight loads.
2)0.9xDesign Pressure + weight loads + 70% of (governing earthquake or wind).
These cases are setup automatically by FE/Pipe and NozzlePRO and each is compared against the same
primary allowable stress (SPL = greater of 1.5Sm or Sy). Note that when wind governs, the guidelines for
seismic design and materials should still be followed if there is a possibility of a seismic event occuring.
From ASCE 7-10: C2.4.1 The earthquake load effect is multiplied by 0.7 to align allowable stress design for
earthquake effects with the definition of E in Section 11.3 of this [ASCE 7-10] standard, which is based on
strength principles. Additional notes from ASCE 7-10 are found in Section 8.0 and throughout discussions in
this document.
If users want membrane and bending stresses computed and printed separately for those load cases where
an individual evaluation of membrane and bending stresses are not needed, they may activate this option
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

11

by setting the 14th item on the hopper.bin file to a 1. This can be accomplished in the local working folder
or in the installation folder as the user desires.

When the user selects the 2013 Rule Set for ASME VIII-2 additional secondary cases will be generated for
primary and occasional loads. Pl+Pb+Q < SPS will be evaluated for primary cases, and for secondary loads
that are to be evaluated as a factored load in the ASD method of VIII-2 Part 5. This essentially puts the
weight and pressure stress into a separate secondary stress category so that pressure bending stresses are
evaluated independently from operating case (thermal and external force) plus pressure bending stresses.
The Pl+Pb+Q (in/out) stresses for primary loads are labeled (SUS), while the Pl+Pb+Q (in/out) for occasional
loads are labeled (OCC).
ASME VIII-2 has always had a contradictory requirement for nozzle necks that has been subject to some
interpretation.
In Section 5.6(a)(2) for requirements for classification in nozzle necks, a PL classification shall be applied
to local primary membrane stresses that include discontinuity effects include restrained free end
dispacements (thermal). This suggests that somehow the discontinuty membrane stress at a nozzle neck
due to a thermally induced loads (the same load that al piping programs deal with using 6 dof beam
elements) should be treated as a primary stress and addressed elastically using the criteria in Table 5.3.

One would expect that this somewhat extraordinary requirement (never done in piping codes and no
failures or difficulties reported), would be reflected in Table 5.6 in the classification section of the nozzle.
Indeed, no mention of this requirement can be found there even though the location: within the limits of
reinforcement is explicitly called out. There is no definition of PL in Table 5.6 that is within the limits of
reinforcement.
One could take the position that:
a)Table 5.6 should be modified to include a PL classification for 5.6(a)(2), and that this somewhat hugely
conservative appraoch explained to the user.
b)Table 5.6 should be used and there are no PL stresses there, and so local stresses in a nozzle neck and
within the limits of reinforcement loads on the nozzle due to free end displacements (thermal) do not need
to satisfy local rules for collapse.
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

12

c)It is up to the user to decide if restrained free end displacements (strain limited thermal loads) impose a
collapse concern. This is explicitly discussed in Table 5.2 Load Descriptions where for T loads the
description is provided, [T] is the self-restraining load case (i.e. thermal loads, applied displacements). This
load case does not typically affect the collapse load, but should be considered in cases where elastic followup causes stresses that do not relax sufficiently to redistribute the load without excessive deformation.

It certainly seems here that the user must decide if the thermal loadings are adequately strain limiting or
not. For the user confused by this issue the following diagram may help clarify .

K1,2 = stiffness of the piping system


L1,2 = length from anchor to nozzle
If both systems produce forces on the nozzle that are equal, i.e. if F=F1=F2, and K2<<K1, then F2 is caused by
a system that is more flexible, and if the nozzle in system 2 plastically deforms a small amount the value of
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

13

F2 will not go down by much because the system is so flexible and therefore F2 has the characteristic of a
weight or primary load, i.e. the plastic deflection is a small part of the total displacement. On the contrary,
if the nozzle in system 1 plastically deforms a small amount the value of F1 will go down by a considerable
amount because the system is not flexible and the deflection due to the plastic deformation is a large part
of the total displacement.
For loads produced by piping systems of type 1, the loads due to free end displacements can be considered
secondary only loads, and do not need to be evaluated for collapse, or any primary loading. For piping
systems of type 2, the loads due to free end displacements (thermal) can be considered primary, and do
need to be evaluated for collapse because the load would hardly go down as the plastic deformation
(collapse) continued.
This is a very simplistic approach to the problem since the geometry of the nozzle, the direction and
magnitudes of the loadings and material properties have a considerable effect on what happens after a
local primary load becomes plastic.
The quantitative difference between the evaluations can be further simplified:
a)If a thermal induced load has a primary character, then the local membrane stress cannot exceed yield.
b)If a thermal induced load has a self-limiting character, then the local membrane+bending stress should
not exceed twice yield. In certain situations, the twice yield value can be exceeded if a simplified elasticplastic calculation is undertaken, (essentially using Ke>1 in the fatigue analysis.)
In attempting to quantify the classification of restrained displacement of attached pipe free ends, the
following logica might be used.
a)Assume the total free end displacement to be absorbed is Y.
b)Assume that an elastic analysis is performed, and at a load level that is X% of Y, the local primary
membrane stress is equal to Sy. (This stress state would describe the lower bound estimate of the point
where a deviation in the twice elastic slope line would occur. The actual deviation from linearity will start
higher since some volume of material must plastically deform before global displacements begin to behave
in a nonlinear way.) This described state is shown below:

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

14

c)If the nozzle is disconnected from the piping system, and the displacement at the nozzle due to the
thermal free end displacements (temperature) is equal to P, then if:
P(1-X/100)>Z
then it is possible that displacements due to plastic action might cause deformations that are equal to the
twice elastic slope then the free end displacements can be considered primary. Since this is a lower bound
inequality, it is thought that there does not need to be a safety factor added to this approach. When high
temperatures govern, the allowable stress should be used.
This test is usually developed with moments only, where: the displacement quantities used above, i.e. Z
and P are really the SRSS of the rotations at the nozzle connnection, and P is computed by removing any
rotational stiffnesses at the nozzle connection while keeping the translational stiffness unchanged. To
perform this evaluation the piping system and stiffnesses of the vessel connection must be available to
evaluate. An example is shown below.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

15

X = 30000/36006 = 83.3% and SRSS Rotation = 0.1411 deg.


Modifying the stiffnesses so that they are essentially zero:

The plastic rotation for the remainder of the load case is (1-0.833)x0.301 = 0.05 deg.
For this example: Z=0.1411 deg., P=0.301 deg., X=83.3%.
Since 0.05 < 0.1411 the thermal overstraining can be considered self limiting.
This example shows a condition where membrane stresses are high due to pressure. It is not thought
hugely worrisome when PL is considered a secondary stress that can cause collapse, because in these
situations Pl+Pb+Q/PL > 2 and 3Sm controls when thermal stresses are introduced. It is also beneficial that
it is generally not common that membrane stresses due to external loads add to the highest membrane
stresses due to pressure.
This is not the case when the hot allowable is considerably lower than the cold allowable. One might argue
that these are cases where creep controls the allowable and that VIII-2 elastic rules should not be used in
these areas, but it is not uncommon to use the local stress calculation in VIII-2 for these cases also. In this
situation, the same approach as above can be used to estimate the plastic collapse tendency of the
component.
The user can ask that operating local membrane stresses (P+W+T) due in part to thermal loads are treated
like primary stresses and a collapse analysis performed. the stresses inside reinforcement for the nozzle
neck (not for the parent shell) are included in this survey. The switch in the ASME preferences panel to
enact this request is shown below:

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

16

3.2 Allowable Loads and SIF Calculations for Pipe Shoes, In-Line Anchors and Vessel Saddles

SIF and allowable load generation for pipe shoes and saddles is available in NozzlePRO as an option in the
shoe/saddle design module. The user can request SIFs and allowable loads from the shoe/saddle wizard or
from the shoe/saddle input forms. Both paths thru NozzlePRO are shown below. The added SIF and
Allowable options are highlighted in orange.

There are two SIF options for both shoes and saddles. One option is for loading thru the saddle/shoe. This
is the most common option. The second option is for loading through the run pipe for shoes, or thru the
vessel only for saddles. Both options are shown schematically below.

SIF for Loads thru Run or Vessel

SIF for Loads thru Saddle or Shoe

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

17

As with typical piping configurations the SIFs developed for pipe shoes and saddles are large when the
piping D/T is large. The SIF analysis for shoes is used most often for any pipe shoe or saddle used as an
anchor, limit stop or guide that introduces any significant load to the piping system. The SIFs developed
from this analysis can be used at the locations in a piping model. Example 5.15 illustrates how this
capability is applied in detail and what might be expected. When the D/T is greater than 40 torsional SIFs
and flexibilities can become considerable. Where cyclic thermal or dynamic loads are evaluated the user is
advised to apply this capability carefully. For piping, this analysis generally shows when full encirclement
support or rings should be added to the pipe support. For vessels, this capability is used most often to
evaluate the stress due to axial loads on the vessel. Some dynamic and seismic dynamic stress capability is
also available and is demonstrated in examples 5.4, 5.5, 5.6, 5.11, 5.13, and 5.16.

Applicable SIF Locations for Pipe Shoe and Saddle Models

Pipe Shoe Options include End Plates, Side Plates, Repads and Rings

Vessel Saddle Options include Tapers, Pads, Multiple Flanges, Rings

For pipe shoes with D/T ratios greater than about 40 torsional SIFs can be considerably higher than bending
SIFs and indicate to the user when rings, pads, or full encirclement reinforcement should be used.
A number of examples are included in this documentation that use the pipe shoe and saddle wizards and
model construction options. The affect of these components on allowable loads, load distribution, presure
stiffening and natural frequencies, are discussed. Different fill levels can be simulated, and half filled
vessels or pipe can result in the largest vertical displacement and highest local stress.
From Design of Piping Systems from M.W. Kellogg:

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

18

The models simulated here assume that the shoe or saddle material is fixed either by fillet, skip, or full
penetration welds essentially around the circumference, and that if fillet or skip welds, then the weld is
sized so that any failure due to an individual loading would accure in the plate adjacent to the weld.

Note:
The models simultated in this version of the shoe modeler assume that all plates on the shoe that are in
contact with the pipe wall are in some manner strength welded to the pipe wall. This is true for nonintegral repad geometries also except that all plates are assumed to be strength welded to the repad, and
then the repad edge is strength welded to the pipe wall.
PIP standards do not require this type of weld attachment along the shoe longitudinal plate, and require
only a fillet along any endplates. Where PIP or other specifications are used that dont assure the weld is
at least as strong as the attached pate the user is responsible for insuring that the short welds will
adequately carry loads from the pipe shoe to pipe wall without failure.
Some manufacturers recommend fully welding the pipe shoe to the pipe. (See anchoragegroup.com) An
image from the Anchorage web site is shown below:

The basis for the allowable loads on the pipe shoes and saddles is based on the separation of secondary and
primary allowbles and stresses, where secondary failures are assumed to be principally due to membrane
and bending stresses, where primary failures (due to high local loads) are principally due to membrane only
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

19

stresses. The allowables for secondary stresses is SPS (3Sm or 2Sy), and the allowable for primary loads is
SPL, (1.5Sm or Sy).
Models for SIF and flexibility applications are also included in example 5.2, 5.11.1, and 5.15. Pipe stress
program users often assume that in-line pipe anchors are rigid with respect to the pipe, serving as a force
and moment firewall, preventing loads from one side of the piping system to influence loads on the other.
Frequently it is found that this assumption is invalid, (increasingly so as D/T gets larger), and so flexibility
models are also recommended for use in this case. Pipe shoe and saddle flexibility factors are normalized
to the bending flexibility of a single diameter of straight pipe. A flexibility of 30 means that 30 diameters of
straight pipe are essentically concentrated at a point spring at the shoe or saddle. Thirty extra diameters of
pipe is usually significant in the flexibility of a tight piping system. If the user is concerned about the
flexibility of the pipe shoe or saddle, they are encouraged to use the wizard to approximate the geometry,
and compute theh flexibility factors. If large flexibility factors are produced, then further evaluation may be
warranted. Example 5.15 gives a detailed example where flexibilities of the in-line anchor affect the loads
on an adjacent equipment nozzle. A typical guided pipe shoe is shown below:

3.3 Measurable First Crack Appearance Calculator


Paulin Research Group has recorded when the first appearance of cracks have appeared during testing by
the use of single channel shear wave and phased array flaw detection techniques. A correlation equation
has been developed from the test data which predicts thru-wall and first crack appearance cycle counts.
The appearance of and growth rate prediction for cracks in many PVP geometries is a function of the
secondary nature of the crack growth through the material. Far field stresses from the crack tip perspective
may relax significantly once a crack first appears. For this reason crack growth rate prediction in PVP
geometries without using J integral type crack models is though to be over-conservative. Unfortunately J
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

20

integral methods are not helpful unless an initial flaw and orientation can be postulated, and single flaws
often times to not appear alone, and so multiple cracks must be postulated for J integral evaluations to be
useful. Predictions from the Measurable Crack Calculator are based on tests conducted at the PRG facility
and correlated with equations for crack growth and failure prediction in branch connection geometries
ranging from 0.062 inches in thickness to 0.5 inches in thickness. The correlations are referenced to the
B31 stress equation (iM/Z), and to the calculated membrane and bending stress states in the material.
How to Use:
The measureable first crack calculator predicts when it might be reasonable to implement an inspection
strategy for a highly-stressed component. For stress levels due either to thermal, dynamic, weight or
pressure loads that occur at some given recurrence rate, the user can decide when during life to first
inspect the location. A more detailed example is provided in Section 5.1. The crack calculator can be
accessed from the FEPipe Tools menus in any of the templates or from the NozzlePRO optional form. The
buttons in the NozzlePRO optional form are shown below.

The crack start evaluator and other useful utility programs in the FEPipe Tools menu are shown below.

Input for the crack start evaluator is given in the following panel:

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

21

The output contains values, tables and charts that give estimated ranges for when the entered stress state
would produce cracks that are
a)measureable, (approx. 2mm deep surface flaw inside or outside)
b)half way thru wall, and
c)completely thru wall.
An example output plot is shown below. The dashed line in the center is the mean projection curve. The
orange line to the left is the curve with a one standard deviation shift lower than the mean, while the red
curve on the right is a one standard deviation shift higher. (One standard deviation includes about 20% of
the population and is based on log cycles.)

Crack appearance and growth calculations should be used carefully and with typical pressure vessel and
piping welded geometries and stress states. (They shouldnt be used for example in GRE, aluminum, or
high temperature applications.)
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

22

For the stress state entered above, the calculator suggests that the mean first appearance of a crack could
occur at 3292 cycles.

If the system cycled four times a day, then a reasonable time to do a first phased array or shear wave
inspection on the system would be found from:
4 x days = 3292; days = 3292/4 = 823;
823/365 = 2.25 years.
2.25 years, (or earlier) would be a good time to inspect the geometry. If crack appearance is thought to be
a critical and dangerous occurrence, (perhaps because a glass lining might crack), then the first inspection
might be scheduled after the lower bound first crack appearance prediction. More information is found in
the example in Section 5.1.
A variety of factors influence crack development and growth. The ASME Smooth Bar curves are based on
3mm crack growth in small axially flexed polished bar specimens. The B31 piping Codes base their SIFs on
thru-wall crack growth of as fabricated pipe products. Measurements at PRG and review of other test data
show many expected and some not expected behaviors. Factors that influence crack appearance and
growth include:
1)Membrane and bending stress ratios
2)Lambda () factor, where = (d/D)(D/T)0.5.
3)Ke (elastic followup) effect
4)Local Geometry
5)Magnitude of Loads (i.e. S > 2Sy)
6)Load Application (4pt bend vs. Markl)
7)Thickness

3.4 3D Viewer Updates

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

23

The following capabilities have been added to or enhanced in the latest version of the 3D viewer for all
instances of NozzlePRO or FE/Pipe.
Thermometer Tool
When the thermometer tool button is first pressed it goes into add thermometer mode and adds a
default thermometer dialog pointing to the highest colorized (stressed) point on the displayed image as
shown in the diagram below. The dialog box contains the coordinate on the model being pointed to, the
displacement at that point, the colorized value (usually stress), and the closest node. The current cursor
values are also shown at the bottom of the screen. Any number of thermometer tool dialog boxes can be
displayed at the same time and the user can right click on the dialog box at any time to personalize the
current data displayed and any future thermometer dialog box displayed. Examples are shown below.

The user can left-click and drag the mouse on any part of the geometry and the thermometer pointer (at
the end of the striped line) will move around the geometry showing the cartesian coordinate of the point,
the displacement at that point (if there is one), the colorized value, and the closet node to the point. The
user can change the evaluated point for any thermometer dialog box at any time by making the dialog the
current dialog. (Click on the dialog handle.)
Each subsequent time the depressed thermometer tool button (
) is pressed, another thermometer
dialog is added to the display so that users can view the results at multiple points simultaneously. The
thermometer tool pointer will follow the stress point thru animated displacement and as the model is
rotated, zoomed and clipped. The user is encouraged to experiment with each of the different viewing
options to find which option best suits particular needs.

The lighting tool has been updated to include brightness and contrast settings. This allows the user
to set the brightest bright and the darkest dark. The user can then use the light direction to move a
flashlight around the geometry to highlight an area of interest. Once the user has settings that are
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

24

preferable, the Save User Defaults button can be pressed and the setting will be saved to the registry
able for use in any future session. At any time the user can restore the factory PRG settings.
The lighting direction can be changed by clicking on the (+) in the Light Direction frame, and dragging it
around the circle. Each side of the circle is approximately 90 degrees away from the opposite side of the
circle.

The thermometer dialog boxes may be tailored by right-clicking on any dialog. The items displayed are
shown in the pop-up menu box (below). The user can turn off and on items to display. This is useful when
showing multiple high stress locations. Removing the coordinates and displacements saves screen space
and permits more selected points to be shown on a fixed screen size.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

25

The point-to locations remain attached to the selected point on the geometry as the model is rotated,
clipped, etc.
When beam models are displayed from CAESAR II, the PCL or the pipe28 template, an ISO drawing can
be created from the solid model view. Examples of the solid model and iso views are shown below:

The minima/maxima search size in the preferences panel determines how many
areas of high or low stress can be found and identified as separate discontinuity zones in the model. The
stress has to drop and then recover, i.e. maxima minima maxima for the stresses to be included in the
groupings. Highest and lowest stresses discovered in this way are listed and can be pointed to. The
collected maxima and minima are shown in the Location frame in the bottom right hand of the viewer:

The user should double-click on any of the values in the list.


The user can select multiple locations by holding down the <shift> key while making selections,

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

26

If the location arrows are not pointing in what is a


convenient viewing direction, file:preferences can be used to change from Normal Based to Origin
Based so that the arrow orientations change. The different effect is seen by comparing the same high
stress point plots above and below.

The user can send multiple images to the same viewer panel as shown
below by activating this checkbox in the file:preferences panel. In the image below, a finite element model
of the pipe shoe piping system is included with several plots of the pipe shoe itself in the same viewer panel
instance. The user can switch focus between the different images by clicking on the geometry to be
manipulated. Multiple models viewed in the same viewer can be also be organized and articulated using
the Navigation tab. Typically, if a user has a large 2nd (or 3rd) monitor, and wants to keep an open,
maximized viewer on that monitor, using the send models to existing viewer checkbox can be very
convenient.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

27

Whenever the user clicks on the thermometer tool button in the toolbar, a new thermometer tool dialog is
opened. (This is true if the thermometer toolbar is in a depressed state or not.) If a dataset has been
loaded, then thermometer dialog pointer will initially point to either the maximum or minimum peak,
selected from the peak list in the location frame. (See the screen shot above in the bottom right corner for
the location frame and the checkboxes described below.)
If the "Include Positive Maxima" checkbox is checked, then the point selected by default when the
thermometer dialog appears will be the peak with the maximum value when the thermometer tool
appears.

If the "Include Positive Maxima" checkbox is not checked, then the point selected when the thermometer
tool is selected will be the peak with the minimum value:

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

28

The maxima and minima values are selected using the following logic.

The maximums and the minimums represent individual peaks, selected as shown below.

The viewer animation can be sent to a <name>.AVI file by selecting the render option from the file menu
drop down. The user will be prompted for the file name and location. The user should be sure to choose a
suitable compression format, otherwise the avi file will be huge. Many avi formats are not supported on
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

29

different versions of Windows. AVI file formats are being replaced by MPEG formats and so the AVI file
writer will be replaced in an upcoming version.

3.5 Sustained Stress Indices for B31.3 2010 and Later Code Years
In 2010 B31.3 introduced the concept of sustained stress indices (SSIs) for sustained load evaluation in the
B31.3 piping code. This is a first step toward resolving the concept of sustained stress failures and the state
of the art of lower bound limit load evaluation using the simplified elastic analysis of piping systems around
which the flexibility portion of the Code is written. PRG has been involved in the refinement of SSI
prediction as part of the ASME ST-LLC 07-02 SIF and k-factor Alignment project and tests conducted at PRG
in Houston. One particular test, highlighting the omission of i-factors and SSI factors for torsional loads in
large D/T systems is shown in the figures below. For the D/T intersection shown, the collapse load is lower
than the B31.3 allowable load. This is due to the omission of torsional SIFs in the current version of the
B31.3 Code. The collapse load definition in this instance is the load beyond which no further increase in
force could be applied to the geometry. Gross distortion of the intersection resulted as can be seen in the
figures.

Several papers have been presented that discuss SSIs development and use:

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

30

1)Edwards, D.R., 2010 Edition of B31.3 Process Piping Code: A First Equation for Stress Due to Sustained
Loads, ASME 2010 Pressure Vessels and Piping Conference: Volume 3,Bellevue, Washington, USA, July 1822, 2010, pp. 287-294.
2)Becht, C,V., Paulin, T. Lock, W.S., Stonehouse, M., Edwards, D.R., Sustained Stress Indices (SSI) in the
B31.3 2010 Edition, PVP 2014-28267
The SIF and k-calculator can be used to get the latest SSI estimates based on results referenced in the above
papers.

For a 4x10 standard wall set of branch connections, the SSIs used currently are not very different from any
possible recommended variant. Four approaches are shown in the table. For the 4x10 standard wall tees,
the SSIs are all about the same because the SIFs are relatively low. (Note that the torsional SSIs are all close
to 1.0.)
1)Using ST-LLC 07-02 SIFs and SSI = SIFX
2)Using ST-LLC 07-02 SIFs and SSI = 0.75 x SIF
3)Using B31.3 and SSI = SIFX
4)Using B31.3 and SSI = 0.75 x SIF
Note: The exponent X used in (1) and (3) above is based on a correlation of collected test data. Actual data is used, and at no
time is the exponent X greater than 0.5.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

31

Large differences occur when high d/t branch connections are encountered. The table below compares
results for d/t=91 and d/D=1. This approximates the tested results shown above.

For the test shown in the pictures at the beginning of this section it can be stated that:
a)The use of SSI = 0.75i can be grossly over-conservative for large d/t pipe.
b)The use of SSI = 1.0 for torsion can be grossly non-conservative.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

32

3.6

Shell Mode Calculator

The shell mode calculator produces natural frequencies and mode shapes of panel ended cylinders. The
output of the calculator is shown in the following plots that show the shape associated with dynamic
natural frequencies of the specified panel ended cylinder. Beam-type modes are shown along with the
shell modes. The calculator is most often used to determine:
1)What beam type modes exist for a cylindrical shell and what natural frequency is associated with them.
2)When are shell natural frequencies lower than beam natural frequencies. (When shell natural
frequencies are lower than beam natural frequencies, shell deflections during a seismic or other dynamic
event may cause more stress than a corresponding beam mode.)
3)When shell modes may interact with pulsation or blade pass frequencies in adjoining rotating equipment.
(When shell modes are in the range of the first few harmonics (or more) of a nearby piece of rotating
equipment there may be resonance and failure at attachments or nozzles in the vicinity.)

Checking shell modes for vertical vessel sections between rings, or heads can help assure that shell modes
will not participate with dynamic excitation. An example calculation is shown in Sections 5.11, 5.13, and
5.16.
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

33

3.7 Percentage Difference Plots


Whenever averaged stress results are generated (nodal or gauss) for shell models, the averaged results are
compared with not averaged results for each node for each element in the model automatically. When the
stresses in a particular part of the model are very low, i.e. 1 or 2 psi, the % variation can be large. When the
stresses in a particular part of the model are high, the percentage variation between averaged and not
averaged results should be low to indicate that a reasonable convergence of the solution has occurred.
Various conditions in different element types can result in convergence to different values as mesh
refinement is pursued but this effect has been minimized in the FEPipe calculation. (Users of other general
purpose tools may notice that for certain element types and stress calculation methods, increasing the
mesh density causes the averaged results to converge to one value while the not-averaged results tend to
converge to another. When this occurs, the method of stress calculation and the origin and assumptions
involving the tensor components should be reviewed.)
For FEPipe or NozzlePRO, when an averaged solution completes the comparison tool can be accessed from
Advanced and then Shell Model Utilities menu as shown below. Then select Post Processing Tools,
and then Avg/Navg Comparisons.

The next menu to appear is shown below. Check the comparison box and then select either the standard
viewer or the 3d viewer. (The 3d viewer is recommended because all options are available in the
interactive processor.) The inside and outside percentage comparisons options are shown as the 1st and 4th
radio button options.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

34

The percentages are weighted by the factor MIN( 1.0, Fmul ), where Fmul = {(S)/[(0.95)(Smx)]}0.25, where S =
averaged von mises component, Smx = maximum of the averaged von mises stress components for the
specified load case. The factor Fmul is 99% or greater when S is within 9% of Smx. The weighting
reduces the percentage error for regions of the model where the stress is small. The percentage values
printed are between 1 and 1+error ratio. If the tolerance is 5%, the weighted value printed should be less
than 1.05. (All percentage values plotted will be 1.0 or above due to the weighting method used. This
makes reviewing percentage error graphical results much easier. An illustration of the method is shown in
Example 5.6. (The example is a long one, and the average vs. not averaged discussion is close to the middle
of the example discussion.)

3.8

Seismic Load Calculator

The Seismic Load Calculator gives users an estimate of the g loads developed from the ASCE 7 design rules.
These g-loads can be used directly in the saddle wizard when a full sized pressure vessel or exchanger
model is constructed.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

35

The seismic load calculation tool can also be used to compare design level spectra with ground motion
spectra from actual earthquakes. The ASCE design spectra are considerably smaller than most recognized
ground motion data. This difference and the role it plays in design are discussed in detail in the
commentary to ASCE 7-10.) Most piping system survive large magnitude earthquakes because they remain
stable after plastically deforming. When a given seismic load produces high stresses that will not undergo
plastic deformation easily the user should consider altering the configuration. This is typically found when
small lines are fixed to larger lines. In these cases, the actual displacement anticipated should likely be
evaluated, particularly in zones where seismic events are known to occur often.

The seismic load calculation screen is shown below. For static load definitions, the F button at the A
location is pressed. To get response spectrum point definitions, the F button at the B location is
pressed. The remainder of the integration and response spectra comparison tools are shown later via
examples.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

36

When the F button is pressed the following screen appears:

The Site class, Seismic Importance factor, ap, Rp and z/h are defined in ASCE 7 Chapters 11, 12, and 13. The
calculators are designed to produce results consistent with ASCE 7 Chapter 13. Chapter 13 addresses piping,
boilers and pressure vessels as shown below. Chapter 15 also addresses nonbuilding structures not similar
to buildings, that can include horizontal, saddle-supported vessels, skirt supported vertical vessels, etc.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

37

ASCE 7 should be consulted for specific applications and the extensive commentary. A brief summary of the
requirements and controlling parameters is given below.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

38

Petrochemical and industrial tanks and vessels can also fall under the guidance of ASCE Chapter 15. The
seismic applicable ASCE Chapters are:
Chapter 11 Seismic Design Criteria
Chapter 12 Seismic Design Requirements for Building Structures
Chapter 13 Seismic Design Requirements for Nonstructural Components
Chapter 14 Material Specific Seismic Design and Detailing Requirements
Chapter 15 Seismic Design Requirements for NonBuilding Structures (Similar and Not Similar to Buildings)
The basic Chapters of interest are Chapters 11, 13 and Chapter 15.
The horizontal seismic design force given in Chapter 13 comes from Eq. 13.3-1:

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

39

Fp upper and lower bounds are limited by:

From a direction point-of-view ASCE states:

The force Fp (ASCE 7-10) is the maximum directional force that is to be designed for.
The coefficient [1 + 2(z/h)] is based on a variety of measurements during earthquakes of various building
type structures and reflects the increase in response as the component is attached higher up in the
structure. Practically, h need not be less than about 15 ft. For example, if a pipe is attached to a sleeper
that is 3 ft. off grade, and the pipe is attached to the top of the sleeper, the [1+2(z/h)] component does not
need to be 3. A value of [1+2(3/15)] = 1.4. The relation [1+2(z/h)] is based on measurements of short to
medium sized structures where the roof acceleration was 3 times the ground acceleration. See C13.3.1 for
references.
ASCE 7 - C13.1.3 States that components with importance factors Ip greater than 1.0 are expected to remain
in place, sustain limited damage, and when necessary, function after an earthquake.
SDS is the spectral acceleration, short period, as determined from Section 11.4.4. The method for
determining SDS is briefly outlined in the table below:
Factors Applicable to the ASCE Seismic Method
Line

Factor

Definition

Ref

ap
Fp

Component amplification factor varying from 1.0 to


2.5.
Horizontal seismic design force

Table 13.5-1, or Table


13.6-1
13.3-1

2
3

Average roof height of structure with respect to base.

Ip

Rp

Component Importance Factor 1.0-nominal, 1.5critical


Component response modification factor that varies
from 1.0 to 12..

SDS

Short period spectral acceleration

Wp

Component operating weight

13.1.3, 13.1.4
13.1.5, Table 13.5-1, or
Table 13.6-1.
13.1.5

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

40

Height in structure of point of attachment of the


component with respect to the base. z/h does not
need to exceed 1.0

13.3.1

ASCE 7 - 13.1.7 Also states:


Where a reference document provides a basis for the earthquake resistant design of a particular type of
component, and the same reference document defines acceptance criteria in terms of allowable stresses
rather than strengths, that reference document is permitted to be used. The allowable stress load
combination shall consider dead, live, operating, and earthquake loads in addition to those in the reference
document. The earthquake loads determined in accordance with Section 13.3.1 shall be multiplied by a
factor of 0.7. The allowable stress design load combinations of Section 2.4 need not be used. The
component shall also accommodate the relative displacements specified in Section 13.3.2.
The relative story drifts from floor to floor in a building or from rack level to rack level in a plant can be
thought to be on the order of from 1 to 2 inches. See Table 12.12-1. Where piping or vessels attached to
the structure at different elevations cannot accommodate this relative lateral displacement at the support
locations, additional evaluation or design alteration may be needed.)
For Horizontal Saddle Supported Vessels
In ASCE 7 - Table 15.4-2 Seismic Coefficients for Nonbuilding Structures not Similar to Buildings, parameters
for Horizontal, saddle-supported welded steel vessels are found:
R = 3; o=2; Cd=2.5.
Paragraph 15.4.1 states that minimum seismic lateral forces will not be less than the requirements of
Section 12.8, and so there are a variety of parameters to check, some of which are a function of the natural
frequency of the vessel:
Line
1

Variable
T

Description
Fundamental period of structure per
12.8.2. See example 5.13 for
computing the natural frequency of
horizontal vessels.

TL

Long period transition period(s)


determined in Section 11.4.5. This is a
fundamental period that ranges from 4
to 16 sec. (0.25 to 0.06 Hz.) The
minimum value of TL in the US is 4 and
the maximum value is 16 s.

Response modification factor. From


Table 15.4-2 R=3 for horizontal vessels

Ie

Same as Ip seismic importance factor


based on risk category I thru iv.

Cs

Seismic response coefficient

Cs=SDS/(R/Ie)

Eq. 12.8-2

Cs(T<TL)

If T<TL ; Cs does not need to exceed the


value calculated here

Cs=SD1/T/(R/Ie)

Eq. 12.8-3

Cs(T>TL)

If T>TL ; Cs does not need to exceed the

Cs=SD1xTL/T2/(R/Ie)

Eq. 12.8-4

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

41

value calculated here.


8

Cs(min)

Cs shall not be less than

0.044SDS x Ie

Cs(min)

Cs shall not be less than

0.01

10

Cs(min)

For structures where S1 is greater than


or equal to 0.6g Cs will not be less than

0.5 x S1 / (R/Ie)

11

Seismic story shear (horizontal load)

V = Cs x W

12

Effective seismic weight per Section


12.7.2.

Eq. 12.8-6

ASCE 7 Method for Determining SDS:


Section 11.4.3 starts with values collected from the applicable seismic maps in Figs 22-1 thru 22-6. A
portion of the seismic map from Fig 22-1 is shown below:

Ss and S1 are the 0.2 and 1 s. spectral response accelerations (5 Hz and 1 Hz.)
Next the Ss and S1 values are adjusted for the Site Class (type of soil and estimated engineering properties).
The site class adjustment comes from the factors Fa and Fv.
Step

Activity
S1, Ss

Get Ss and S1 (5Hz and 1Hz response spectrum values from


USGS Figures 22-1 thru 22-6.)
Ss = MCER spectral response acceleration parameter at short
periods
S1 = MCER spectral response acceleration parameter at a
periods of 1 s.
MCER = Maximum Considered Earthquake Response

Adjust for Site class other than Site Class B


Fa Table 11.4-1
Fv Table 11.4-2

SMS = Fa x Ss
SM1 = Fv x S1

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

42

Find SDS and SD1


SDS Design earthquake spectral response at short period
SD1 Design earthquake spectral response at 1 sec. period

SDS = (2/3) x SMS


SD1 = (2/3) x SM1

ASCE 7-10 contains the following wording when relating the seismic maps in the 2010 version to those from
previous versions of ASCE 7:

For B31 piping, seismic design guidance, is given in ASME B31E-2010, Standard for the Seismic Design and
Retrofit of Above-Ground Piping Systems.
In B31E Para. 3.1, When the seismic design force is computed based on para. 13.3-1 of ASCE 7, or a similar
standard, the parameter ap shall be 2.5 and the parameter Rp shall not exceed 3.5 when applying the stress
limits of para. 3.4.
Further, The seismic loading shall be specified for each of three orthogonal directions . The seismic
design should be based on either a three-directional excitation, east-west plus north-south plus vertical,
combined by SRSS, or a two directional design approach based on the envelope of the SRSS of the eastwest plus vertical and north-south plus vertical seismic loading.
The allowable specified in B31Ea-2010 in para. 3.4 is given as:
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

43

PD/4T + 0.75i(Msus + Mseismic)/Z < MIN( 2.4S; 1.5Sy; 60ksi)


and for seismic anchor movements: FSAM/A < Sy.
B31Ea-2010 also states that Mseismic should include relative anchor movements due to earthquake loads.

3.9 Alternate Weld Model


It is a common problem for shell analysis is to properly represent both the stiffness and the stress at the
normal penetration line of a shell geometry. ASME Section VIII-2 Part 5 Annex 5A includes guidelines for
possible models that can be used to simulate the penetration line geometry. At full penetration welded
junctions cracks are expected to start at the toe of the fillet, although in practice cracking will often start
internally at defects such as cold-shorts, overlaps, etc.
ASME VIII-2 Part 5 emphasizes that the area of the shell finite elements in between the points A and C
should equal the actual penetration line area. PRG also believes that the moment of inertia of the section
between A-to-C should also be maintained.
The default PRG penetration line model has been adjusted thru the years so that results match both Markltype fatigue test results and the primary stress limits given by VIII-2 for bust tests. In general this means
stiffening the models somewhat by providing a small amount of additional area. This can be particularly
impactful when the t/T ratio is much greater or much smaller than 1, and when the tw/t or tw/T ratios are
much greater than or much smaller than 1. A more satisfactory model seems to result however when both
the area and the inertia in the penetration line are equal for the shell model and for the geometry and weld
being simulated. This can be accomplished by a variety of models, where the two most common ones are:
a)Tapered elements from A-to-B, and then from B-to-C.
b)Tapered elements from A-to-B, and then from B-to-C and then from A-to-C.
Other variations are also possible.

Since the area-inertia model is new, it was added to FEPipe and NozzlePRO as an option users may select.
In general, when it is chosen the highest bending stresses at intersections are expected to go down.
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

44

Users should recognize that it is very difficult to properly estimate the magnitude of the bending and
membrane stresses at an undefined corner even with brick finite elements. The sketch below shows the
strong gradient across the mounted strain gages in the ORNL1 test model. The boundary constraint can be
thought of as the difference between plane stress and plane strain model types. As the shell approaches
the penetration line, the increased volume of the material at the penetration provides a Poissons
resistance against contraction normal to the direction of the maximum stress.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

45

3.10 NozzlePRO Notes


1)When the model has completed running, the user can interrogate the FEPipe detailed output by clicking
on the Results button at the top of the NozzlePRO screen and selecting Review Text Results

As part of the text results the FEPipe output is available.


The 3D Plots button takes the user to the FEPipe output processing interface.

Commonly used features from this interface are shown below:

Advanced: show model origin aid setting fluid elevation


Advanced: Shell Model Utilities plot tensor values, plot % differences between averaged and not
averaged stresses
Advanced: Refractory Options plot stresses in brick models of refractory
Control:Show Displacement plot and contour shade displacements for any load case
Control:Model Verification plot properties, thicknesses, check unzipped nodes or collapsed elements.
Model verification is not available as part of the NozzlePRO output processing capability. Verification is
considered an input process.
2)Examples are included in this release note that demonstrate how the user can include pressure stiffening
in static or dynamic runs of NozzlePRO components. Including these capabilities is easier from FEPipe, but
if only NozzlePRO is available then the procedures here can be followed.
3)
When the d/D ratio is less than 0.03 NozzlePRO will consider
removing the backside from the model and using panel connections. Small d/D models can result in very
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

46

large finite element models if a sufficiently long axial length of the run or vessel is included to remove the
boundary conditions from the nozzle. Bending stiffnesses and SIFs are not affected significantly by the
boundary conditions in small d/D models but axial stiffnesses and axial SIFs may be affected because the
ovalization length of the boundary is still approximately 0.5D1.4/T0.4 regardless of the d/D ratio. Users can
force the backside of the model to be included by checking the above box in the NozzlePRO optional form.
A typical panel model with a small d/D ratio is shown below:

4)Simple piping models can be included in the NozzlePRO input to aid in the evaluation of load transfer
from the piping anchor or zero displacement point to the NozzlePRO shell model of the nozzle. An example
NozzlePRO model with piping included is shown below. This model is included in the models folder and is
called PipeSystemAndHead.nozzlepro.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

47

5)Default lengths have been organized into the following data panel:

Each default condition is defined below:


Default Lengths Prior to WRC 492 it was believed that about two diameters removed from the end of an
ovalized component was a sufficient distance to prevent the interaction of end effects. The two diameters
lengths were used, for example on a number of the larger diameter elbow flexibility tests. There are really
several criteria that establish a useful boundary condition length: a)(RT)0.5 this length is for local
meridonal bending of the shell, the donut-hole bending. The distortion associated with this length is not
associated with ovalization.
Widera Recommended Lengths In WRC 492 Widera ran a series of 33 different models and found that in
the larger d/D and D/T models four and five diameters were needed to fully remove the boundary
condtiion effect from the intersections. Wideras lengths are based on d/D ratios. For d/D < 0.333 the
length of the cylinder on either side of the branch centerline is 4D and the nozzle length is 2.5d. For d/D <
0.5 then the length on either side of the branch centerline is still 4D but the nozzle length increases to 3d. If
d/D < 0.75 then the length on either side of the branch centerline is 5D and the nozzle length is 4d. For
greater d/D ratios the length on either side of the branch centerline is 5D and the nozzle length is 5d.
PRG Recommended Lengths After looking at the ovalization problem in a bit more extensive manner,
PRG observed that ovalization lengths are a function of D/T. When T is 1/2D, then ovalization will not exist.
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

48

When T = D/100, then ovalization can be significant. The expression for approximate clearence the
boundary condition from the intersection is the length AxD1.4/T0.4. It is difficult to predict a more precise
length or to specify the constant Ax to within about 0.4 to 0.75 since the loading has some effect on the
behavior, i.e. ovalization of a bend tends to require higher values of Ax. For small d/D geometries, bending
i- and k-factors tend to be not so sensitive to the ovalization length while the axial i- and k-factors are
sensitive, but use the smaller values of Ax.
6)Internal clips and lugs Structural shapes in NozzlePRO can be inside or outside the vessel. For internal
clips and lugs, the length should be specified as a negative number as shown below.

3.11 Updated NozzlePRO Printed Reports


When pressing the print button on the NozzlePRO main menu, (below), a print browser form will appear
that shows the user how to use the system html forms printers to get complete control of each report.

Users can print each report individually or all tabular and graphical content at one time. Detailed
instructions are provided for common Internet browsers.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

49

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

50

4.0 Minor Feature Additions:


1)FE/Pipe unreinforced fabricated tee template
Feature in UFT template to provide radial restraint. This gives the user the ability to evaluate the difference
between torsional flexibility and stiffness factors. Stiffness factors for torsion provide radial restraint to the
branch, while flexibility factors for torsion do not.

This YES/NO field fixes the branch end in the transverse direction so that a torsional moment applied to the
end of the branch will cause torsional bending only without translations. The feature is used to get varying
estimates of the torsional load capacity of branch connections. This is important because B31.3 does not
provide torsional SIFs for torsional moments about the branch or run pipe. Without providing radial
constraint at the end of the branch, a torsional applied moment will produce torsional rotation along with
axial and transverse displacements of the branch end.
2)Use mean diameter for pressure areas. In the ASME Preferences panel this checkbox was added for
single nozzle shell models of tee intersections. The specific templates are :

1)Unreinforced fabricated tee


2)Pad reinforced fabricated tee
3)Hillside tee (pad or no pad)
4)Welding tee.
These templates use the checkbox T code variable NO_PRESSURE_ADJUST29 to make sure that that mean
area is used for pressure thrust instead of the ID of the geometry. The difference is shown in the diagram
below. The nozzles, plates and shell template always uses the mean area.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

51

When Pressure Acts over the Area Defined by the Nozzle Inside Diameter Pressure forces in the nozzle
axial direction in model are not balanced. The difference in total load is approximately equal to
(pi)(D)(T/2)x(pressure). The longitudinal pressure stress in the nozzle is accurate and the hoop stress in the
nozzle is not affected.
When Pressure Acts over the Mean Area of the Nozzle Pressure forces in the nozzle axial direction are
balanced. (There is no unbalanced pressure load). The longitudinal pressure stress in the nozzle is a little
larger than accurate value found using the ID. (The pressure actually acts over the inside diameter of the
nozzle, not over the diameter defined by the mean surface of the nozzle.)
3)On the optional form in NozzlePRO the user can check the Force Backside checkbox. When the d/D ratio
is less than 0.03 NozzlePRO will attempt to omit the backside of the model to improve solution speed.
Panel models are generally adequate for bending and torsional loads but may not be conservative for axial
loads when large sections of shell surround an essentially point-loaded nozzle.

When the d/D ratio is small the user can force NozzlePRO to include the backside of the model. In this case
the user should also enter the length of the parent (if a cylinder) and the location of the nozzle so that
boundary conditions can be best simulated. For large D/T geometries the influence length can be on the
order of D1.4T-0.4.
4)Hot formed tees are added to the PCL welding tee library. Hot forming is used for tee matrials that are
susceptible to cracking when being formed. The geometry of these tees is based on the method used by
manufacturers to form the tees by starting with oversized pipe, and compressing it to the appropriate
diameter prior to extruding the branch outlet. An additional average welding tee model has also been
added that likely is closer to welding tee SIFs produced by ASME ST-LLC 07-02. These tees are a small
amount thicker than the smaller of the average tees one might purchase today.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

52

5.0 APPLICATIONS:
5.0.1 Table of Contents
5.1 First Observable Crack in Pipe Shoe ...........................................................................................................54
5.2 Pipe Shoe and Saddle Design and Analysis.................................................................................................59
5.3 Modeling of In-Line Pipe Anchors...............................................................................................................62
5.4 Drainage Tank Seismic Load Example.........................................................................................................64
5.5 Vessel Loading History ................................................................................................................................69
5.6 Horizontal Vessel Loading Example ............................................................................................................70
5.7 Hillside Nozzle Option in Tank Template ....................................................................................................91
5.8 Design, Operating and Occasional Pressure Differences in VIII-2 ..............................................................94
5.9 Pressure Stiffening in Horizontal Vessels ...................................................................................................96
5.10 Effect of Rings on Stresses in Large Diameter Nozzles ...........................................................................100
5.11 Saddles on Pressure Vessels and Pipe ....................................................................................................101
5.11.0 Some Saddle Analysis Observations: ...............................................................................................109
5.11.1 Axial, Vertical, In-Plane, Out-Plane and Torsional Orientations ......................................................110
5.11.2 Conclusions Regarding Seismic Designs: .........................................................................................115
5.12 Brownell & Young Horizontal Vessel Model ...........................................................................................116
5.13 Calculating Natural Frequencies in Horizontal Pressure Vessels............................................................119
5.14 Participation Factors Simple Example .................................................................................................139
5.15 In-Line Anchor Analysis........................................................................................................................... 142
5.16 Wind Vibration........................................................................................................................................151
5.17 Pressure Stiffening of Spherical Geometries ..........................................................................................156

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

53

5.1 First Observable Crack in Pipe Shoe


Note: The crack calculator is developed from cyclic load tests at PRG on common piping and
vessel full penetration welds. The first detectable crack is assumed to be an approximately 2mm
crack in the thru-thickness direction. Tests did not include corroded or high temperature
conditions.
Problem: The stress at the tip of the pipe shoe went through 15 full hot-to-cold cycles before high
stress concentration at the anchored shoe was noticed.

The high stress level in the area of the pipe was 151,937 psi. In the pipe shoe at the edge of the
fillet, the stress was 376,134 psi. These are Pl+Pb+Q stresses which are membrane+bending
(M+B). The membrane stresses are shown in the figure below. It can be seen that the stress in
the anchor plate is almost all membrane, while the stress in the pipe wall is a combination of
membrane and bending, but mostly bending.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

54

The results from the above discussion can be tabulated:


Location

Membrane

Membrane+Bending

Shoe Endplate

362517

376134

Adjacent Pipe Wall

45203

151937

The crack processor can be opened from NozzlePRO or FEPipe. The icon in the NozzlePRO
optional form is shown below. From FEPipe it is opened from the Tools dropdown menu.

The membrane and bending stresses entered using an FSRF of 2 are shown above. Changing the
FSRF from 2 to 1.2 gives the following difference:

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

55

As suggested in the help, if the stress is computed without a stiffness or weld length modification,
then the FSRF should be 1.0. For pipe shoes and saddles constructed using the NozzlePRO
wizard, a weld leg is used as part of the stressed region determination, and so according to the
help, the value of 1.2 is most reasonable.
A measureable crack would be expected on the plate side of the pipe shoe between 105 and 226
cycles. A thru wall crack would then appear at between 607 and 2017 cycles. If the weld in the
high stress area shows undercut, starts-and-stops, or large, globular welds, then it would be
expected that a crack would be noticed in the 22 to 50 cycle range. In this case it would be
expected that the crack would begin from the outside. On the pipe shoe side, the stresses are
lower, and so even with an FSRF of 2, a measurable crack would not be expected until around 236
to 510 (or more) cycles.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

56

The stresses in this geometry are very high, but the highest values are relatively local due to the
edges of the plate punching through the sidewall of the pipe. A linearized plot of this stress
distribution is shown below. Only stresses greater than 50,000 psi are shown.

It would not be surprising if there was some permanent local bending at the support at the edges of
the pipe shoe end plates. Stresses of this magnitude caused by the thermal flexing of relatively
long runs of pipe (not, short, tight piping systems), should be treated with caution. The analysis
performed here is not recommended without a thorough review of all system characteristics that
might contribute to crack development, some of which are:
1)Pressure stresses.
2)Fluctuating pressure stresses or other external vibrations
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

57

3)Corrosion or erosion from the outside of the pipe


4)High strength material, or any material that might be difficult to weld.
5)Cracks starting from the root of the fillet.

Pressure stresses can be included in the shoe/saddle evaluation. Pressure stresses should only
be included in the crack development analysis if the pressure stress cycles along with the thermal
stress. It is not unusual for pressure to cycle more than the thermal stress when one considers
that pressure can drop and increase much faster than the thermal stress in large piping systems.
In this case pressure stresses can be evaluated by themselves.
The PRG interpretation of this stress state is that it would not be surprising if measurable cracking
did not occur until around 100 cycles, and the lack of any cracking prior to this number of cycles
should not be interpreted that the system is satisfactory. Additionally, if the weld was of high
quality, and the system did not cycle, it would not be surprising to have the system cycle 100 times
without a measureable crack appearing.
In relatively complex geometries and loading conditions crack appearances can only be predicted
within a relatively large range of cycles. The key is to have some idea when cracks might appear,
and from the stress contours, to know where to go look for them. Once a crack is detected, fitness
for service approaches can be used to determine if the crack is suited for continued service. In
most cases it likely is, as long as the loading state is well determined, i.e. the weld is not on an
offshore platform subject to potentially high, single occurance pitch and roll loads.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

58

5.2 Pipe Shoe and Saddle Design and Analysis


To build either pipe shoes or pressure vessel saddles when starting a new NozzlePRO job, click on
the Shoe/Saddle wizard icon:
or select the Shoe/Saddle Wizard from the
menu. The
Wizard will graphically/tabularly collect all needed data for the shoe or saddle geometry and
recommend defaults. When finished, the wizard will give the user the opportunity of selecting the
loading:

Most manufacturers of pipe shoes provide allowable loads on the shoes but not on the attached
pipe. A typical caution on a pipe shoe allowable load table reads, Affect of stress in pipe wall to
be determined by others.
The allowable load on the pipe shoe is indeed a function of several factors:
1)The degree of shoe/saddle support, (i.e. is the shoe or saddle anchored, guided or just
supporting vertical loads.)
2)The length of attached pipe.
3)Pressure in the pipe
4)External loads thru the run.
5)Configuration of attached pipe and thermal growth.
Pipe shoes have themselves failed due to excessive load, vibration, temperature gradients, poor
welding, overheating, cyclic loads and local collapse. Often, support allowable loads are taken
from company tables or manufacturers data based on Bijlaard/WRC107 type evaluations. These
simplified calculations often ignore:
1)Circumferential, longitudinal and torsional moments in the run pipe due to support loads
2)Stress in the support plates
3)Non-rectangular attachment profiles
4)Non-integral wear plates.
The most common stress related issues associated with support design is believed to most often
occur from:
1)Using supports for small D/T pipe on large D/T pipe.
2)Using supports as anchors without considering the affect of combined forces and moments

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

59

3)Assuming that all supports can accommodate the total suggested pipe span when high thermal
loads exist at the support in the piping system.
Maximum permitted loads:
Many pipe stress analysts ignore support concentrations all together and simply pass the loads
from the piping system at a support location to the structural engineer for supporting steel design.
The structural engineer typically does not evaluate the stress in the pipe due to structural load and
so the local stress in the pipe is often ignored. When there is not a company standard for how the
pipe shoes, stops or saddles should be fastened to the pipe, it is recommended that some design
effort should be used to size and configure welds for both external loads and for the potential of
relative thermal expansion.

The B31.3 Code (2012) states:


321.1 General The design of support structures (not covered by this Code)
321.3 Structural Attachments External and internal attachments to piping shall be designed so
that they will not cause undue flattening of the pipe, excessive localized bending stresses, or
harmful thermal gradients in the pipe wall. It is important that attachments be designed to minimize
stress concentration, particularly in cyclic services.
Where d/t ratios, loads, temperature, or cycles are high, evaluation of the local stress in the
support an the pipe can be important.
The piping analyst must decide the extent of the local evaluation that is required. Where creep is a
concern, local stresses can easily cause creep damage. In these cases local stresses should be
evaluated carefully and compared to suitable allowables such as those in ASME Section III
SubsectionNH an option included in both NozzlePRO and FEPipe.
In most other cases, only a general evaluation of the support is necessary. Assumptions involving
general evaluations often include:
1)Assuming that maximum allowable loads do not exist in each load direction simultaneously. In
this case, the maximum allowable loads are evaluated individually and any combination of load
values are kept below the individually evaluated maximums.
2)Ignoring pressure stress in the pipe wall. This is only advised when PD/2T < Sh/3, where P is
the operating pressure in the pipeline, D is the mean diameter of the pipe, T is the corroded
thickness and Sh is the hot allowable stress for the condition being evaluated. Supports are often
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

60

not evaluated unless loads are high. When loads are high, the involvement with existing pressure
stresses should be considered. (This is especially true when either the pressure, or the external
load cycles.) For high sustained stresses, the pressure can actually resist local stresses due to the
pressure stiffening effect. (Pressure stiffening effects are demonstrated in Example 5.9.)
3)Assuming that the weld-to-pipe is stronger on a per linear length of weld basis than the attached
structural plate or the pipe, i.e. any failure at the plate attachment to the pipe would occur at the toe
of the fillet thru the plate and NOT thru the weld.
4)Assuming that a fatigue analysis is not needed. Most allowable loads are based on secondary
and primary limits that are not related to fatigue. Most piping systems cycle less than 3000 times in
their entire operating life. Where the number of design cycles exceeds 3000 cycles, a more
detailed look at the peak stresses in support welds may be warranted and the use of wear plates
considered. Torsional loads are often not carried effectively by pipe supports, and where cyclic
stresses exist with high torsional loads a fatigue analysis is almost certainly warranted.
5)Assuming that wear plates can be simulated as a local increase in the pipe or vessel thickness in
some cases is not conservative. NozzlePRO models provide integral and nonintegral repad
models for clips, lugs, pipe shoes and saddles. In critical situations both model types should be
evaluated and the most conservative results used for design. (The state of wear plate contact with
the pipe or vessel outside diameter is unknown in the vessel operating condition and so running
integral and nonintegral models is thought to bound possible solutions. Where the solution is
sensitive to the integral or non-integral model type, some alternate design is recommended, or the
wearplate should be made thicker and its overall footprint on the vessel or pipe made smaller.
Large D/T pipe do not suffer point loads well and so pipe shoes without external rings can act like
point loads on large diameter pipe. Simulations including full-encirclement external rings at both
ends of the pipe shoe often show a considerable increase in support strength due to the ring and a
lowering of the local stresses.
The following examples illustrate pipe shoe and saddle applications:
5.4 Saddle modifications to accommodate axial seismic load
5.6 Comparisons of FEA vs. Zick
5.9 How pressure stiffening can affect displacements and stresses
5.13 Natural Frequency Calculations for Horizontal Vessels
5.13 Effects of Rings on Mode Shapes and Natural frequencies
5.15 Piping Example of In-Line Anchor SIFs and Flexibility Impact on Loads and Stresses
5.16 Vortex Shedding Calculation

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

61

5.3 Modeling of In-Line Pipe Anchors


In most cases, pipe stress engineers rely on existing standards, a structural support detail group, or a
support supplier to provide details for supports on a piping system.
Most supports are designed by the support provider to assure that the support is not overloaded.
Including the pipe geometry, external load magnitudes and pressure is beyond the scope of the support
design, especially considering that most support designs are prepared well before any single pipe stress
analysis. Some support manufacturers provide additional design services so that loads from a CAESAR II
calculation can be used in a specialized finite element model to determine the stresses in both the pipe and
the support. This can be a time consuming and expensive proposition.
Critical support loads can occur in any pipe geometry, but they do tend to occur more often when the pipe
d/t ratio is high and the loads are high. Pipe supports become more troublesome when the loads cycle or
the pipe is operating in the creep or cryogenic range. For highly loaded in-line pipe anchors the critical d/t
ratio starts around a d/t ratio of about 30 or what is the equivalent of standard wall 10 pipe.
The most important concerns involved in the design of in-line pipe anchors are listed below.
1)Torsional load thru pipe anchors: Since most analysis focus is on in-plane and out-of-plane loading,
torsional loads are generally ignored or overlooked. Surveys of standard piping systems and of typical pipe
anchor designs suggests that torsional loads should not be ignored. Poorly designed saddles or inline pipe
anchors often dont adequately restrain torsion and introduce high local stresses in the pipe in the
downstream vicinity of the saddle/shoe attachment points.
2)Need for full encirclement: Full encirclement for a pipe saddle/shoe comes in the form of a clamped or
welded on ring. Welded on rings are often preferred in high load, critical situations, since clamped loads
are not often considered dependable. The full encirclement increase in thickness can address a number of
high local stress issues.
3)Need for rings: For large d/t pipe any local load can cause ovalization long distances down the pipe
where the ovalization distance on either side of the support can be estimated as 0.5D1.4T-0.4, where
discontinuities separated from each other by less than this distance may interact. (The degree of
interaction is a function of the size of the discontinuity and the orientation of the load.) Six DOF pipe stress
programs do not include this ovalization in the analysis. The bending stress at supports can be similar to
the sidewall stress in a bend. The bending stress axis is usually along the sidewall of the pipe, and the
bending axis is along the axis of the pipe. This is true for weight loads, and for torsional loads thru the run
pipe. Where this bending stress is developed, external rings may be needed to hold the pipe round so that
the plane-sections assumption in beam theory is valid and to control local bending stresses at corners of
the support . As the d/t ratio increases beyond about 75, the need for external rings increases rapidly as a
function of the load and direction magnitudes. In most cases, company standards provide for external rings
at supports on large diameter pipe. (This often happens because pressures below atmospheric are possible
in the pipe and external rings are designed to accommodate some percentage of vacuum.) In other cases,

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

62

problems have occurred with anchor design for vertical loads only, as seen with Zick evaluations and rings
are provided in these instances.
4)Downstream equipment protection: Where relatively tight piping systems include in-line anchors, guides
or limit stops, the pipe analysis often assumes that the anchor, guide or limit stop is rigid. When the d/t
ratio exceeds 50, pipe anchors (depending on the design) might be very flexible and may not restrain
moments in the way a rigid anchor in a pipe stress model assumes. (Moment restraint in a piping system is
difficult to provide for almost any d/t pipe although the condition is more troublesome for larger d/t pipes.
In the large d/t case, the user is expecting that the anchor, although fixed rigidly to a concrete, or large
steel base, will limit the rotations at the anchor point to zero, when in reality the pipe locally bends around
the anchor and may behave more realistically like a pinned end connection. In the pinned connection case
moments are transferred from one side of the anchor to the other. Where rotating equipment or expansion
joints are being protected, the predicted loads and displacements are unrealistically low.
5)Temperature distribution and insulation: For hot pipe, the user should be sure that a significant
temperature gradient will not exist in close proximity to the pipe surface so that additional, often large
thermal stresses exist at structural attachment welds to the pipe.
6)Creep: Local stresses due to support attachments are often not evaluated. When the pipe operates in the
creep zone, high local/primary membrane stresses in the vicinity of supports can produce creep voids if
local stress levels and extents are high enough. ASME Section III Subsection NH rules can be used to
conservatively evaluate these stress levels.
7)For large d/t ratios (>75) if there is a possibility for partial fluid levels, high stresses can result due to the
fishbelly effect as the sides of the pipe come in at the midspan and the bottom of the pipe goes down
farther than predicted. In this loaded state, the pipe (or vessel) is susceptible to instability due to a
suddenly applied occasional load.

Pressure Stiffening and Nonlinear Analysis of Saddle Supported Vessels


Pressure stiffening will stiffen the wall of the vessel and reduce displacements due to ovalization. It is not
unusual for a vessel to displace more in the half filled condition, and then to return to the almost empty
position when fully filled and pressurized. This behavior is due to both the pressure stiffening of the shell,
and the distribution of loads that occur due to liquid filling of the vessel. The Zick analysis does not detail
either of these phenomena.
The liquid fill load distribution problem is linear elastic. The pressure stiffening problem is nonlinear and
must be solved (when needed) with an iterative procedure. Both approaches are described in example 5.9.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

63

5.4 Drainage Tank Seismic Load Example


The customer specification indicates that ASCE 7-05 Chapter 13 should be used to develop seismic loads for
the project saddle support drainage tanks. The vessel will be installed near Anchorage Alaska. The seismic
load development process is:
1)Determine Ss and S1 values for the design response values.
2)Using Ss for static loading, compute SDS and then find the design horizontal load to be used in the
allowable stress procedure of VIII-2 Part 5.

After review of the above Ss and S1 values these are considered satisfactory for both 07-10 and 07-5 ASCE
response values. (This will not always be the case.)
The client wants to use the Table 15.4-2 R=3. The results of the calculation are shown below.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

64

The calculation details are given below. The final values printed for both the horizontal and vertical
directions are already multiplied by 0.7 for use in the VIII-2 Part 5 Table 5.3 seismic load cases:
Detailed Calculations:
0.2 Second Site Coefficient
Fa = 1.00
(Table 11.4-1)
1.0 Second Site Coefficient
Fv = 1.00
(Table 11.4-2)
0.2 Second Site Adjusted Acceleration
Sms = Fa*Ss
(11.4 - 1)
= 148.81 %g
1.0 Second Site Adjusted Acceleration
Sm1 = Fv*S1
(11.4 - 2)
= 55.10 %g
0.2 Second Design Acceleration
Sds = 2*Sms/3
(11.4-3)
= 99.2067 %g
1.0 Second Design Acceleration
Sd1 = 2*Sm1/3
(11.4-4)
= 36.7333 %g
Basic Seismic Acceleration
A1 = (0.4*ap*Sds*Ip/Rp)*(1+2*(z/h))

(13.3-1)

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

65

= 124.0083 %g
Maximum Required Acceleration
A2 = 1.6*Sds*Ip
(13.3 - 2)
= 198.4133 %g
Minimum Required Acceleration
A3 = 0.3*Sds*Ip
(13.3 - 3)
= 37.2025 %g

Per section 13.3, the horizontal design acceleration is given by the basic acceleration
(A1), but may not be less than A3, and need not be greater than A2. Note that the values
are adjusted for use with the allowable stress design procedures of ASME codes per
paragraph 13.1.7.
Horizontal Design Acceleration
= 86.8058 %g
The vertical acceleration per section 13.3 is:
Av = (-0.2 * Sds)*0.7
= +-13.8889 %g

The objective is to run case (6) in Table 5.3 below for seismic. The 0.9 factor should be used on the design
pressure and 0.7E should be used for the seismic load magnitude. The 0.7 has already been applied, and so
the input to the saddle wizard model would appear:

The calculation for load case (6) and VIII-2 Part 5 2007 version shows that the saddle is overstressed. This
model is available in the models folder as anchor15.nozzlepro.
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

66

The fixed saddle model is pivoting since it is assumed that this will produce a more conservative stress
result. An option appears in the analysis to fix the saddle: Stepping back thru the saddle wizard the
checkbox for fixing rotations at the baseplate is found in the saddle geometry data inputs. This box is
checked and the model rerun. In this case the designer must make sure that bolts, slots, nuts, etc., are
provided so that in a seismic event, the fixed saddle end will not rotate.

The maximum primary membrane stress dropped from 65 to 59 ksi and the high stress moved to the
bottom, thin section of the saddle.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

67

The client wanted to try two different approaches to improve the design:
1)Use the 2013 version of VIII-2 so that a slightly higher local membrane stress might be available.

2)Increase the width of the saddle at the baseplate in the axial direction to reduce the stress.
Tapering the saddle from 25 to 45 dropped the stresses significantly as shown in the result below.

To change the above design change required only a 30 second modification to the input since the taper of
the saddle plates is a value entered directly into the saddle wizard.

Saddle Wizard Input

Tapered Saddle

Reverse Tapered Saddle

Switching to the 2013 version of VIII-2 also increased the allowable by 6ksi, since the 2013 version permits
the maximum of 1.5Sm or Sy for the allowable local primary stress.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

68

The stress in the reversed taper saddle is well below the allowable as shown below.

5.5 Vessel Loading History


A common vessel operating load history is described below:
a)Vessel is in an empty and ambient state.
b)Vessel is in an empty potentially heated state. (Allowable is the hot allowable. This may be one of the
loading conditions to be evaluated either on the startup or shutdown of the load cycle.)
c)Vessel is in various stages of filling. Generally the two of interest are half filled no pressure and filled
no pressure.
d)Vessel is filled and pressurized in the operating condition.
e)The vessel passes back thru some combination of (b) and (c) to return to condition (a).
f) The vessel is exposed to wind and seismic loading.
The Zick analysis performed by most horizontal vessel design programs evaluates the saddles subject to
weight loads and contents. A typial Zick analysis is compared to the FEA results for the same vessel in
example 5.6. For lateral loads due to wind and earthquake, the saddles are usually evaluated individually as
a beam. The finite element analysis shows how inaccurate that can be as the actual load capacity varies
circumferentially around the contact zone of the saddle.
For seismic analysis according to ASCE 7-10 Chapter 15, the fundamental period of the vessel is required.
This may be a shell mode of vibration which is not evaluated by any vessel program. Evaluating
fundamental periods of vibration are demonstrated in Example 5.13. For gas filled vessels, the vibration
modes are not affected by the contents. For liquid filled vessels the mode shapes may be influenced by the
contents, and this is a function of the tangent-to-tangent length and the d/t ratio. The distribution of the
liquid mass can be simulated in various ways since lateral movement in horizontal and vertical vessels apply
inertial load to a single side of the vessel at one time. Example XX.X shows how the different inputs can be
used to simulate the static seismic load distribution inside the vessel.
Seismic loading developed per ASCE 7 most often will result in underestimated elastic loads. Users should
look closely at the local stresses since bending and peak stresses are not evaluated in Table 5.3. The sum
of the local membrane and bending stresses due to occasional load cases will however be compared with
SPS since the bending component of the local stress due to pressure is considered secondary. Since
earthquake loads will see actual stresses above elastically calculated stress by at least o, and since the
actual fatigue stress life contribution will be much higher than o times the elastic stress due to ASCE 7
loadings, the user in a high seismic zone, might consider including a fatigue analysis of some type with
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

69

elastic seismic analysis. This evaluation should likely be based on site specific spectra, (especially for
foundation mounted vessels), or soil liquifaction spectra. The seismic load case tool
contains a small
library of time histories that can be scaled as needed to produce response spectra. Ground response
spectra can be turned into vessel stresses by using the participation factors from the seismic analysis as
shown in the simple example 5.14, or the more comprehensive example of 5.13, and then the resulting
maximum membrane and bending stresses can be used in the crack start calculator to determine if some
reasonable number of seismic cycles (100 for a typical event), would result in a measureable crack.
Pressure stiffening can increase the natural frequency of the vessel, as can external rings.

5.6 Horizontal Vessel Loading Example


The following example is taken from Brownell & Young, Process Equipment Design. The design pressure
is 90psig and the working pressure is 75psig. The detailed input steps, screens, and button selections are
included in the table in Example 5.12. Zick output for the vessel is shown below.

From VIII-2 Table 5.3 (below) the design pressure should be summed with dead and live loads for cases 1
thru 4, and 90% of the design pressure should be used for seismic and wind loading in cases 5 and 6. ASME
VIII-2 Fig. 5.1 shows that the operating pressure should be used for secondary, (ratcheting and shakedown),
and fatigue analysis. The Zick analysis can be used to evaluate Table 5.3 Cases 1 2, and 4 providing the
snow loads should be considered and can be simulated as an increase in the material or liquid density, or
applied as a uniform external load acting in a downward direction on the vessel. (In most cases, snow and
ice loads are not included except in some LNG service.)
For the current example, the Zick recommendations will be followed using the design pressure. Most
stresses computed in a Zick analysis do not include the effects of pressure in the stress but do include the
weight and distribution of the water load. Since much of the Zick approach is based on a correlation with
strain gage measurements, it can be assumed that the Zick basis is the actual distribution of the loads, e.g.
the liquid acts as an internal pressure over the OD of the vessel with its largest component at the bottom,
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

70

varying linearly to a zero liquid weight component at the top. (This distribution can significantly affect the
stress and displacement distribution in a half-filled vessel for example. Half filled vessels in some cases are
not well simulated by simply increasing or decreasing the shell density to match the net vertical load. Part
of the usefulness of the finite element method is that the user can take several approaches for the same
model, and use one to validate the accuracy of the other. Where differences in the assumptions are
significant, tests can be performed. Strain gage and displacement tests during hydrotests of horizontal
vessels are simple to conduct and add little additional to the cost of fabricating a horizontal vessel that is
already going to be hydrotested.)
Zick recommends certain distances where possible to take best advantage of the geometry of the
horizontal vessel. Some of these recommendations are included in the Table below:
Item
Saddle Location
Saddle Location

Wear Plate Width


Wear Plate Length

Ring Location

Recommendation
Distance from vessel tangent line to the saddle centerline should
be 0.4R and never greater than 20% of the tan-to-tan length.
For high D/T when the distance between saddle centerline and
the head tangent line > 0.5R, the user might consider external or
internal rings.
Should be longer than b + 1.56(RT)0.5, where b is the width (axial
length) of the saddle.
In the circumferential direction the wear plate should extend on
each side by R/10 above the horn of the saddle. Note that the
angle subtended by this distance is equal to the distance divided
by the radius of the vessel, or (R/10)/R = 1/10 radians. If the
saddle is 120 deg. (2.09 radians), the wear plate should be 2.09 +
1/10 + 1/10 = 2.294 radians (131 deg.)
Two rings at a saddle should not be closer than 1.56(RT)0.5 or
further apart than R.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

71

The Zick calculation for this vessel is given below. This calculation is included in the NozzlePRO output for
the vessel and can be found in the standard NozzlePRO output reports.
CALCULATED PROPERTIES AND STRESSES:
----------------------------------Allowable stresses given below are taken from the
recommendations made in the 1950's paper by Zick.
Effective Shell Angle at Saddle
Mean Radius used in Calculation
Hoop Pressure Stress
Longitudinal Pressure Stress

=
=
=
=

80.429
60.438
8748.001
4374.000

deg.
in.
psi
psi

The heads are close enough to the saddles to be included


as stiffeners. 30.000 <= 30.219.
LONGITUDINAL STRESSES
--------------------K1(at saddle)
K1(at midspan)

=
=

0.039
0.814

Longitudinal Membrane Stress


(At saddle)

222.992 psi

Longitudinal Membrane Stress


(At midspan)

4643.008 psi

The maximum bending at the vessel section, (resulting in a membrane


stress in the shell wall), may be either tensile or compressive. The
tension stress plus the axial stress due to internal pressure should
not exceed the allowable tensile stress times the girth butt weld
efficiency.
S(long) + PD/4t < (Sallow)(eff)
At Saddle:
(222.99+4374.00) < (13750.00)(1.000)
4596.99 < 13750.00
At Midspan:
(4643.01+4374.00) < (13750.00)(1.000)
9017.01 < 13750.00
The net longitudinal compressive stress in the shell should be the
lesser of 1/2 of the yield strength, or the value given by:
(E/29)(t/r)[ 2 - (2/3)(100)(t/r) ]
1/2 Yield
=
(E/29)(t/r)[ 2 - (2/3)(100)(t/r) ] =

18000.00 psi
13553.08 psi

At Midspan:
-269.01 < 13553.08
TANGENTIAL SHEAR STRESS

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

72

----------------------K2 =
0.862
K2w =
0.318
Max Shear Stress
Max Shear Stress
(Stress in shell

(K2
(At
(At
and

used for evaluation at equator for wear plate.)


Head ) =
4578.05 psi
saddle) =
4578.05 psi
wear plate combined.)

Max Shear @equator =


1690.43 psi
(Required for wear plate. This stress in shell only.)
The tangential shear stress should not exceed 0.8 times the
allowable stress in tension.
4578.05 < 11000.00
CIRCUMFERENTIAL BENDING STRESS AT HORN OF SADDLE
-----------------------------------------------K3 =
0.013
Compressive Stress at Horn =
Allowable Bending Stress
=

8639.69 psi
20625.00 psi

K3(for edge of wear plate) =


0.008
Compressive Stress at Saddle Edge =
13950.52 psi
EXTERNAL LOADS
-------------Zick provides a simplified analysis for external loads which assumes
that if a horizontal vessel is designed for an external pressure of
1 psi that it will successfully resist wind and other external
loads encountered in normal service. The approximate length between
stiffeners required by ASME Section VIII is approximated by the
equation: ((12)E/(52.2))(sqt(rt)(t/r)**2 )
Required unstiffened length =
Available length
=

4381.764 in.
836.146 in.

ADDITIONAL STRESS IN HEAD USED AS A STIFFENER


--------------------------------------------The heads on this vessel are close enough to the saddles to be used
as stiffeners in the design. The additional shear load carried by
the heads will induce an extra stress in the heads.
Added Head Stress =

2107.29 psi

Zick recommends combining this stress with the pressure stress in


the head and making sure that the result is less than 1.25 times the
tensile allowable stress for the head.
Note that the additional head stress of 2107.29 psi is less
than 25% of the allowable stress, so no further concern is
warranted.
RING COMPRESSION IN SHELL OVER SADDLE MIDDLE
-------------------------------------------The vertical compressive stress at the bottom point of shell
contact with the saddle is 13803.70 psi This value
is less than Zick's recommended allowed value of 18000.00 psi.
WEAR PLATE DIMENSION CHECK - ZICK
---------------------------------The wear plate provided is NOT wide enough. ................. EXCEED
Zick recommends that the wear plate be at least (B + 10t) wide. This
would require a wear plate that was at least 14.250 in. wide,
The entered plate was only 10.000 in. wide.
WEAR PLATE DIMENSION CHECK - ASME VIII-2 Part 4
-----------------------------------------------The wear plate provided is NOT wide enough. ................. EXCEED

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

73

ASME recommends that the wear plate be at least 1.56xSQRT(Rt) wide.


would require a wear plate a minimum of 17.588 in
wide, the entered wearplate was only 10.000 in. wide.
HORIZONTAL SEPARATION FORCE ON SADDLE
------------------------------------Force separating Saddle
Equivalent Stress
Zick Allowable Stress

=
=
=

41202.10 lb.
2726.92 psi
9166.67 psi

Zick analysis and other hand methods do not evaluate the stress distribution in the individual saddle flange
plates. Finite element analyses however, does show how the weight and pressure stresses vary from flange
to flange in the saddle, and in particular shows how local stresses can be developed in either the vessel or
at the baseplate, or both, when external seismic or high wind loads act on a horizontal, saddle supported
vessel.
A pressure stiffening estimation is also part of the PRG-Zick analysis based on pressure stiffening analyses
produced at PRG. The pressure stiffening analysis is intended to be overly conservative, and if more than a
24% projected influence on the stresses in the vessel is predicted, then the adjusted stresses are also
printed with an extended Zick calculation. If the solution is thought to be sensitive to pressure stiffening,
the user is recommended to do a more accurate pressure stiffening analysis using FEPipe. Pressure in large
diameter pipe or vessels often reduces stresses and increases the buckling strength due to the soccer-ball
(thin-membrane) effect. External loads and weight stresses should often be evaluated without the effects
of pressure in finite element models of saddles and pipe shoes. Worst case stresses can often occur due to
external pressure for large D/T pipe and vessels, and this case and potential buckling should be evaluated
especially where high seismic loads are anticipated and may be underestimated. (Buckling failures do not
provide the same level of safety that redundant plastic hinges can provide.) The pressure stiffening report
for the Brownell and Young vessel is included below:
PRESSURE STIFFENING ADJUSTMENT
-----------------------------Pressure in this problem may stiffen the geometry and have an
effect on calculated stresses and displacements. A rough correlation
developed by PRG makes the following predictions.
Pressure may reduce elastic calc'd displacements by:
Pressure may reduce elastic calc'd stresses
by:

1.308.
1.239.

If these reductions are significant they should be verified by a


pressure stiffened finite element analysis or test.
Hoop Pressure Stress
Longitudinal Pressure Stress

=
=

7290.001 psi
3645.000 psi

(No Change)
(No Change)

LONGITUDINAL STRESSES
--------------------Long Membrane Stress (Saddle) = 222.992 to 180.044 psi
Long Membrane Stress (Midspan) =
4643.008 psi (No Change)
TANGENTIAL SHEAR STRESS
----------------------Max Shear Stress (At Head ) = 4578.05 to 3696.33 psi
Max Shear Stress (At saddle) = 4578.05 to 3696.33 psi
Max Shear @equator = 1690.43 to 1364.85 psi
CIRCUMFERENTIAL BENDING STRESS AT HORN OF SADDLE
-----------------------------------------------Compressive Stress at Horn = 8639.69 to 6975.71 psi
Compressive Stress at Saddle Edge = 13950.52 to 0.1126369E+05 psi

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

74

ADDITIONAL STRESS IN HEAD USED AS A STIFFENER


--------------------------------------------Added Head Stress = 2107.29 to 0.1701428E+04 psi
RING COMPRESSION IN SHELL OVER SADDLE MIDDLE
-------------------------------------------Stress at Saddle Bottom Middle = 13803.70 to 11145.15 in.
HORIZONTAL SEPARATION FORCE ON SADDLE
------------------------------------Force separating Saddle = 41202.10 to 33266.70 lb.
Equivalent Stress = 2726.92 to 2201.73 psi

An in-line anchor report is also performed as part of the Zick calculation by PRG. The report is intended
to help the piping engineer designing in-line anchors or pipe shoes. Details of the in-line anchor calculation
are included elsewhere. The in-line anchor report is generated when the user selects the pipe shoe instead
of the vessel saddle model in the Wizard.
The Zick (or saddle weight and load analysis) is run automatically when the Saddle wizard is used. The steps
for the saddle wizard are shown in the following table:
Step

Screen

Notes

Select a new NozzlePRO job.

Select the Saddle Wizard

Pick the type of support. The left button is for the


pipe shoe wizard, and the right button is for the
saddle wizard. Both analysis types will produce
allowable loads and SIFs. Only the saddle wizard will
duplicate the Zick calculation.

There are a variety of saddle taper types. The saddle


tapers are used when high moments are developed at
either the vessel or the baseplate when axial or
transverse loads are applied to the horizontal vessel or
pipe.

Enter the vessel geometry.

When the geometry and load properties are defined


the analysis options form is presented. The Zick
comparative analysis is Option 1. NO OTHER
EXTERNAL LOADS OR ACCELERATIONS EXIST.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

75

When the Zick analysis is performed the longitudinal stresses at the saddle and the midpoint of the vessel
are combined with the longitudinal pressure stress. This is the only use of the pressure stresses or loads in
the Zick analysis. All other stresses are computed using the weight of the vessel and contents only.
To most easily make the Zick comparison to the FEA analysis the pressure should be left zero and the
vessel liquid depth and specific gravity should be entered so that the vessel is full of liquid.

The entry of the liquid depth will allow both the finite element analysis and the Zick calculation to use the
same weight loads. (Q) in Zicks nomenclature.
The finite element result from the Zick-type saddle analysis is shown below:

This compares with the solution data report from NozzlePRO/FEPipe that gives the net loads on the model:

Since only half of the finite element model is constructed in the Zick finite element model option, the total
vessel and liquid weight is half of the value used for the Zick analysis 200734 x 2 = 401,468 lb. ~ 401,201 lb.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

76

Saddle Load Selection Screen Zick Load Option

The same total weight load is used in both the Zick calculation and the finite element analysis. A segment
of the Zick report from above is copied below with a discussion of the stresses and magnitudes that can be
identified in the finite element results.
LONGITUDINAL STRESSES
--------------------Longitudinal Membrane Stress
(At saddle)
Longitudinal Membrane Stress
(At midspan)

222.992 psi

4643.008 psi

TANGENTIAL SHEAR STRESS


----------------------Max Shear Stress (At Head ) =
Max Shear Stress (At saddle) =
Max Shear @equator =

4578.05 psi
4578.05 psi

1690.43 psi

CIRCUMFERENTIAL BENDING STRESS AT HORN OF SADDLE


-----------------------------------------------Compressive Stress at Horn =
8639.69 psi
Compressive Stress at Saddle Edge =
13950.52 psi

Added Head Stress =

2107.29 psi

RING COMPRESSION IN SHELL OVER SADDLE MIDDLE


-------------------------------------------Ring Compressive Stress Over Saddle = 13803.70 psi

Finite element Stress Results

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

77

Membrane stress at midpoint in the vessel shell are 3260 psi (membrane) on the bottom (FEA) and 5500
psi on the top (FEA). The Zick analysis gives +/- 4643 psi. (5500 + 3260 )/2 = 4380 psi. The results from the
FEA analysis are the equivalent stresses. When there is a single component of the stress, (or one major
component), the equivalent stress is equal to the stress intensity. When the stress state is biaxial, the
equivalent stress can be approximately 10% lower than the stress intensity. The stress intensity is equal to
twice the maximum shear stress. For a single component stress, the stress intensity is equal to the value of
that component, i.e. for a component thats subject to an axial stress of 15,320 psi and no other stresses,
the stress intensity at that point will also equal 15,320 psi. For the stress state at the midspan of the vessel
the longitudinal transverse and shear stresses from the finite element model result can be plotted:

Longitudinal (6494 psi)

Tangential (-9.7 psi)

Shear (-4.5 psi)

A maximum stress intensity at the top of the vessel of 6494 psi is present. In this case the longitudinal
stress average would be (6494 + 2664)/2 = 4597 psi.
The thickness distribution (showing the edge of the wearplate) can be generated from FEPipe after the
model is constructed from the

NozzlePRO option.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

78

Thickness profile for NozzlePRO Saddle Geometry for Brownell & Young Example

To generate this plot follow the selections below from NozzlePRO:

The highest stress at the edge of the wearplate is shown in the figure below for the outside stresses. These
are the stresses that Zick correlated in his calculation.

On first inspection, the finite element result appears larger than the Zick predicted results. The Zick results
are repeated below for the saddle location:

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

79

If we assume that the Zick stresses occur at the same location in the model then the maximum shear stress
intensity is equal to:
2

2 0.5

Max Shear Stress Intensity = 2 x [(13950/2) + (4578) ]

= 16,686 psi.

The high stress at the edge of the repad near the saddle is shown below:

There is a 21,450 psi stress in this model. The location of the high stress point can be found by clicking on
the maximum value in the list in the location frame (see below).

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

80

In this view, with the arrow pointing at the high stress location it can be seen that the highest stress is in
the outer flange plate on the saddle. Zick does not calculate the stress at this point. (Islands of high
stress are segregated by the viewer and can be plotted separately. This stress organization capability
shown by the peaks in the Location frame are described at the end of Section 3.4.) The high stress in this
example model in the Brownell and Young saddle occurs in the outer flange plate on the saddle and not in
the vessel. The high stress an be pointed to by Lines and Arrows, or particles. The thermometer tool will
also by default point to the high stress location if the Include Positive Maxima checkbox is checked. (See
the viewer location frame in the above right image.)

Stepping through the high stress collected results (see the panel to the right of the plot), the next value is
the high stress in the shell of 19,587 psi. This would be compared to the maximum stress intensity from the
Zick calculation of 16,686 psi.
The stresses above were computed using a non-integral repad. These models are designed to give
conservative results and to envelope one end of the wear-plate/vessel geometry interaction. When an
integral model is run, a non-conservative solution is expected with less local bending and lower
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

81

stresses. The model result below shows this result, and gives an equivalent stress of 17,257 psi, which
compares favorably to the 16,686 psi from the Zick calculation.

Shear Stress

Longitudinal Stress

Circumferential Stress

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

82

The above calculation with an integral repad selected from the saddle wizard is shown above.

The models and the meshes are designed to produce conservative, realistic results compared to Zick and
other measured data. Maximum results for the attached model for Gauss Averaged and Nodally Averaged
results are given in the table below:
Stress
Pl
Q
M+Bi
M+Bo

Gauss Averaged
11061
20921
17782
25144

Averaged
11731
21199
18690
25445

ABSA AB-520 States:


The accuracy of the model digitization shall be indicated, either by the described use of convergence
studies or by comparison to the accuracy of previous successful in-house models. This section of the Report
must include a proposed method to verify that the model results reflect the real response of the physical
pressure equipment.
and
Plot with element stress and compare nodal (average) stress vs. element (non-averaged) stress (If the small
difference is less than 5%, the accuracy should be OK.)

Where users have to comply with ABSA requirements NozzlePRO and FEPipe can produce a weighted
comparison of the averaged and not-averaged stresses. The largest disparities of averaged and notaveraged results may occur at boundary elements where large elements are placed for efficient run times
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

83

and for boundary conditions only. Mismatches between averaged and not-averaged results can also occur
at large, instantaneous differences in thickness, (in FEPipe and NozzlePRO models) usually again at the
boundaries. These conditions are illustrated below:

Stresses are extrapolated from gauss points to node points for reporting. If the stress is constant in the X
direction and in the Y direction, then the gauss points will have the same value of the stress and the
stresses at the nodes will be the same as the stress at the gauss points. In this case, the stress at node A
from Element #1 will be the stress at node A from element #2. When these two values are averaged, the
average result will be the same as the gauss point result for each element.
A strong stress gradient across the adjoining elements is illustrated in the plot below

In this case, and partly because of the relative proximity of the gauss points and nodes, and the need to
extrapolate from the gauss points to the nodes, there are two different stresses shown at the identical
geometry point at node A.
For FEPipe usually these large differences in element size occur away from areas where there is an interest
in the stress. For regions near boundaries (or away from areas of stress interest), the stresses are generally
lower, but the averaged and not averaged percent error can behigher. This is illustrated in the table below:
Stress Area
Boundary Condition
Removed from Stress
Area of Interest
Stress at High Stress Area
of Interest

Average Stress

Not Average Stress

Percent Error

15 psi

8 psi

39%

13,432 psi

13,104 psi

2.4%

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

84

If percentage error at the the low stress location at the boundary condition removed from the area of
interst was used to qualify the accuracy of the solution described in the table, likely a good solution would
be rejected. For this reason percentage stress calculations are weighted based on the magnitude of the
stress at the evaluated location.
For this reason the NozzlePRO software weights the ratio of (Not Averaged Stress) / (Average Stress)if the
stress is less than 95% of the maximum stress in the model. (There is no weighting if the stress is in the top
5% of the stresses in the model.) The weighting is done as follows:
1)If the stress is less than 20% of the maximum stress, there is no average/not average ratio
comparison. In this case it is assumed that the stresses are so low that any variation is due most
likely to boundary condition issues.
2)If the stress is > 20% of the maximum stress and less than 95% of the maximum stress in the
model then the stress ratio is weighted by [Stress/(0.95 x Max Stress)]0.25.
To demonstrate that the 5% separation between the average and not average stress is present in critical
areas of the model the user should:
a)Find the critical load cases.
b)Find the areas of the model where the stresses are the highest
c)clip out the remainder of the model
d)Be sure there are no adjoining areas of large thickness differences
e)Plot both the inside and outside stress percentage ratios to be sure that average and not averaged
results are the same to within 5%.
The NozzlePRO models are not expected to show a compatibility of 5% in all areas of the model only in
those areas where stress accuracy is considered important.
The percentage comparisons for NozzlePRO results can be made when the run completes by taking the
following steps:
Step

Screens to Access Avg/Not Avg Stress Percentages

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

85

When the comparison checkbox is selected, and the radio button opposite VonM(Navg)out is selected,
the outside Von Mises not-averaged stress is displayed. The VonM(Avg) out selection causes the outside
Von Mises averaged stress to be plotted. The user can pick the weight percentage to plot, and then if
there are areas of the model that show higher than desired percentage errors, the individual values can be
compared to determine if indeed this is an area of interest, or simply a transition zone of small stress where
large and small elements meet.
For the Brownell and Young vessel model, case 3 is the weight and pressure case:

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

86

The high stress is inside the orange box, and this is where it is desired that the average and not averaged
stresses have converged. Zoom in and clip the model so that any unnecessary areas are removed. The
clipped portion of the model is shown below:

A typical percentage tolerance plot of this area is shown below. In the high stress area, the percentage
error is less than 0.04%. In the element size transition area, the percentage error is 9%. The percentage
error in the controlling stress area is considered more than sufficiently accurate, and the area showing a
higher percentage area is in a mesh transition zone.
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

87

The PRG estimate for reduction in stress due to pressure stiffening is 6%. To discern the effect of pressure
stiffening on the weight stresses the following procedure can be used:
1)Run the pressure case to get the stiffened elemental matrices
2)Run the weight analysis as above with the stiffened matrix to see the effect.
The steps to include pressure stiffening in the analysis are:
1)Ask to use the FEPipe editor:

on the NozzlePRO optional form.

2)Make sure the 75psig pressure is included:


3)Run the analysis:
4)When the FEPipe screen appears click on Submit & Wait:
5)Check the load case report:
. Need to know exactly what case is the pressure
case to use for stiffening. From the load case report (portion below), the 2nd case should be used:

6)Enter the General options:


7)Change fluid load option fro 2 to 3:

8)Ask to use the pressure stiffened matrix from the previous run from the load case 2.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

88

9)Click on Submit and Wait

Stress Difference due to Pressure Stiffening

Condition A

Condition B

Stress

Averaged

Averaged Stress
Stiffened

% Dif

Pl

11731

9881

18%

21199

16550

28%

M+Bi

18690

15234

22.6%

M+Bo

25445

20561

23%

Pressure stiffening will reduce the effect of the weight stresses which cause the vessel to ovalize. Only Zick
longitudinal stresses include the effect of pressure in the shell. For the compressive horn and shear
stresses it is assumed by Zick that there is no pressure stress concentration in the vicinity of the plates, and
this was apparently the case in Zicks strain gage results.
The approach shown here gives the user an estimate of the effect of pressure stiffening on the local
stresses in the shell. Just as for elbows in piping systems, when the pressure resists the ability for the pipe
to ovalize, the resulting discontinuity stress goes down.
To perform a seismic analysis, often ASCE 7 is consulted. After review of the document, locations, etc., it is
determined that the vessel must undergo a horizontal g load of 0.6gs.
(The 0.6g = 0.7E)
The ASCE 7-10 loads are maximum directional loads, so it is up to the analyst to apply the load in the proper
direction. Generally, this is either the full load in the axial direction, or the full load in the horizontal
direction.
For the seismic (or wind) load, option 3 must be used (below), and a full vessel model constructed.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

89

The vertical load is typically taken per ASCE 7 para. 13.3 and from Chapter 15 as +/- 0.2SDS Wp which is 0.12
g for this example problem. The seismic load should act with 0.9 times the design pressure as seen in Table
5.3 load combination (6). NozzlePRO makes this combination automatically per Table 5.3 as long as the
user enters 0.7 times the seismic load (E). The full value of the design pressure will act with the dead and
live loads, but only 90% of the design load should act with seismic loads per the primary load case 6. (See
Table 5.3 below.) Operating pressures (entered by the user) will be used with the secondary and peak
stress calculations.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

90

5.7 Hillside Nozzle Option in Tank Template

The pipe is ratpip2.ifu, and the tank model is rathole2.ifu.


The tank is 30 ft. in diameter with a 1 wall thickness in the bottom course. The nozzle is 14 thick with an
elevation from the base of the tank of 24. The nozzle thickness is the same as the lower course thickness
1.
The offset is described by entering the x-offset, or X Axis Skew Length text value below.

The overall nozzle angle from the Z axis (see right figure above), is the arc-tangent of the X Axis Skew
Length divided by the Nozzle Length.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

91

Note the tank wall pressure profile and the overall effect of the profile and the tank bottom boundary
condition on the tear-drop exaggerated displaced shape at the bottom of the tank. Where piping is
attached to the tank nozzle, the rotations downward at the nozzle can cause high stresses in both the tank
nozzle and pipe if not properly accomodated. The user can evaluate this condition by either
a)Analyzing the pipe and tank wall together
b)Applying the tank wall rotations to the pipe, assuming the tank wall rotations will not be affected by the
stiffness of the pipe, and then applying the pipe moments back on the tank wall.
c)If using b, then the rotational stiffness of the pipe at the attachment of the tank, can be entered into
the tank model. If changes in rotations occur, then an interaction can be expected and b should be
repeated with the rotations from step c.
The overall displacement and stress in the above model are shown below:

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

92

The rathole reinforcement should typically be evaluated with both fixed and pinned tank walls at the base
to evaluate the range of anticipated behavior. Both of these boundary condition limits are available in the
FEPipe tank nozzle template.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

93

5.8 Design, Operating and Occasional Pressure Differences in VIII-2


Local pressure stresses can often control an analysis. The difference between operating, design and
factored pressures used in the 2007-thru-2013 VIII-2 can cause confusion. The simple example below is
used to illustrate how the different pressures should be used in an analysis. The properties of the cylinder
are: OD=100; T=0.1; L=100; Free End. The pressure input for each run is 100 psi. PD/2T = 50,000 psi.

Membrane and Bending Stress Post 2007, Primary Load Case, Equivalent stress = (30.5/2)(PD)/(2T) =
0.866(PD)/(2T) = 0.866x50,000 = 43,301 psi.

Membrane and Bending Stress Post 2007, Occasional Load Case, Use 90% of the pressure stress = 0.9 x
43,301 = 38,971 psi.

The stress at the top of the model in excess of the calculated PD/2T stress is due to the crude mesh at the
end boundary condition that is constraining radial dilation of the cylindrical shell. To properly evaluate this
stress element lengths at the top of the vessel would need to be approximately half of the root(RT) =
[(50)(0.1)]0.5 = 2.23 inches. The measurement tool is used in the following figure to show the length of one
element edge. It is approximately 19 which is much to large to properly evaluate the local stress at the

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

94

discontinuity. For this example, only the general pressure stress at the center of the vessel is of interest
and so this errant boundary value will be ignored in further discussions.

Measurement Tool Showing Meridonal Element Length

Pre-2007 VIII-2 Design and Occasional Stresses due to Pressure In this case the design pressure is used for
both cases. The pressure only case is shown on the left and the occasional case is shown on the right. The
occasional case is a pressure only case.
The average stress in the vessel center is equal to the stress intensity = PD/2T = 50,000 psi.
There are several differences between the pre-2007 and the post-2007 ASME VIII-2 Code that affect
occasional, primary and other loads. The differences noted here involve
1)A shift from stress intensity to equivalent stress when going from the 2006 to the 2007 Version of ASME
VIII-2. When the stress is biaxial, the equivalent stress can be about 10% lower than the corresponding
stress intensity and that is seen here in the pressure stress state in a cylinder. The equivalent pressure
stress is 86.6% of the stress intensity.
2)Factored loads from VIII-2 Table 5.3 reduce the pressure in an occasional load case by 10% before
combining it with 70% of the earthquake load from ASCE 7.
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

95

5.9 Pressure Stiffening in Horizontal Vessels


Pressure stiffens large d/t pressure vessels and piping systems in the same way that pressure stiffens a
basketball or soccer ball. For horizontal vessels and most other PVP type geometries pressure has a
measureable effect on the displacement than the stress. Often the effect of pressure stiffening will reduce
local stresses and so the pressure stiffening effect should be used carefully to reduce stresses since in some
cases other loads may act without pressure. The stress reduction is shown in the example PRESSURE
STIFFENING ADJUSTMENT report from NozzlePRO below. The estimate shows that calculated weight
displacements are reduced by 1.44 times while the calculated weight stresses are reduced by 1.077 times,
e.g. for the vessel used in the example:
PRESSURE STIFFENING ADJUSTMENT
-----------------------------Pressure in this problem may stiffen the geometry and have an
effect on calculated stresses and displacements. A rough correlation
developed by PRG makes the following predictions.
Pressure may reduce elastic calc'd displacements by:
Pressure may reduce elastic calc'd stresses
by:

1.441.
1.077.

If these reductions are significant they should be verified by a


pressure stiffened finite element analysis or test.
Hoop Pressure Stress
Longitudinal Pressure Stress

=
=

17534.539 psi
8767.270 psi

(No Change)
(No Change)

LONGITUDINAL STRESSES
--------------------Long Membrane Stress (Saddle) = 631.984 to 586.922 psi
Long Membrane Stress (Midspan) =
3567.690 psi (No Change)
TANGENTIAL SHEAR STRESS
----------------------Max Shear Stress = 3890.57 to 3613.16 psi
Max Shear @equator = 2167.36 to 2012.82 psi

Pressure stiffening reduces the bending stresses in the models and so will most often have a bigger effect
on Q, and Pl+Pb+Q stresses than membrane (Pl) stresses. Pressure stiffening effects tend to be more
important when the Zick calculated stresses approach the magnitude of the pressure stresses in the
geometry.
Pressure stiffening will provide little help when high stresses are located in the saddle flange plates. (The
Zick analysis does not evaluate stress in saddle flange plates.)
Whenever the user is concerned about the effect of pressure stiffening, a pressure stiffening run can be
performed using FE/Pipe and NozzlePRO to evaluate the magnitude of the effect on stresses and
displacements. In many horizontal vessel design situations involving pressure and weight, the pressure
stress governs the thickness but does not significant contribute to local stress, and for this reason a Zick
analysis of the saddle is adequate. When local bending stresses due to external loads are of the same
order as hoop pressure stresses in a vessel design, the effects of pressure stiffening, full, empty and half
filled vessels without pressure should likely be evaluated. When stiffening is significant, it is usually due
only to pressure. Other physical geometries can produce stiffening (the guitar string for example), but
pressures resistance to cross sectional ovalization is the most common in piping and pressure vessel
geometries. Pressure stiffening will have approximately the same effect on weight stresses as it does on
stresses due to external loads (seismic or pitch and roll). Either effect can be evaluated using the
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

96

NozzlePRO Saddle Wizard. (Sections 3.8, 5.4, 5.11.2, and 8.0 are provided to discuss the details of applying
Seismic and Wind loads to finite element models in accordance with ASCE 7.)
Procedure for Pressure Stiffening in NozzlePRO Saddle Geometries:
The following steps are used to evaluate pressure stiffening in a NozzlePRO saddle or pipe shoe geometry.
(The approach can be used for NozzlePRO nozzle type geometries also.)
1)Run the regular elastic analysis as usual
2)Review the Load Case Report from the elastic analysis to decide which case should be used as the
stiffening case. Generally this is case 1 or 2 the first case with pressure.
3)In the FEPipe General screen select the Include Stress Stiffening option.
4)Enter the stiffening case in the Load Case for Stiffening text cell.
5)Rerun the analysis (Resubmit). The results from this second run will include the effects of pressure
stiffening on the model. Generally the displacements and stresses will be smaller.
A more detailed procedure for running pressure stiffened models is given in sections 5.9, and 5.17. A
typical displacement deflection result is shown below. With pressure stiffening the vessel sagging is
reduced by 42%. (Load cases used here are not the same as those used in the summary at the bottom of
this section.)

Displacements (in.) No Pressure Stiffening (0.2496)

Y
Y Displacement (in.) WITH Pressure Stiffening (0.1434)

The height of the saddle does not influence the Zick calculation, but can have a significant effect on the
stress in the saddle and the vessel when seismic and wind loads are included in the calculation.
The model included in the installation for this vessel is BrownYoung.nozzlepro.
Pressure Stiffening Results Summary:
Brownell and Young Full Vessel Model with design pressure for stress stiffening pass. The load cases and
stress below are developed using the following ASME preference switches:
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

97

These two switches were selected so that the secondary bending due to primary loads will be evaluated in a
separate Pl+Pb+Q load case (the 2013 code year), and so that weight and weight+pressure stresses will be
evaluated separately so that they can be compared. In this case, the weight case is analyzed with the vessel
empty.
Brownell and Young Stress Summary Different Weight, Pressure and Head Cases + Stiffening Effects

PL
Qb
PL+Pb+Qi
PL+Pb+Qo
Pl
Qb
PL+Pb+Qi
PL+Pb+Qo
Pl
Qb
PL+Pb+Qi
PL+Pb+Qo
Pl
Qb
PL+Pb+Qi
PL+Pb+Qo

Load

Stress (psi)

W
W
W
W
W+P+Head
W+P+Head
W+P+Head
W+P+Head
W+Head
W+Head
W+Head
W+Head
W+1/2Head
W+1/2Head
W+1/2Head
W+1/2Head

1834
3284
2909
3953
12960
20062
21418
19425
11980
21347
18978
25581
7747
4457
8125
8125

Stress
(psi)
(Stiffened)
1508
2504
2313
3138
12539
14973
17622
15548

% Dif

19.5
26.9
22.8
22.9
3.3
29.0
19.4
22.2

Max

Brownell and Young Vessel Displacement Summary

Load

Vert Disp (in.)

Vert Disp (in.)


Stiffened

-0.07746

-0.04534

Stiffened Shape

%Dif

52.3

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

98

W+P+Head

W+Head
W+1/2Head

-0.4866

-0.2833

-0.50632
-0.3034

52.8

Notes: Pressure stiffening can impact the fatigue life due to pressure by introducing an additional varying
bending component to the stress. In this case the W W+P case shows both the change in membrane
stress due to pressure and the change (reduction) in bending stress due to pressure. Although one can
imagine that the change has a slight improvement on fatigue life, i.e.
W(M+B)=2909W+P (no stiffening) = 21418, compared to: W(M+B)=2909 W+P(stiffened)=17622
For these calculations the design pressure is used with a 120 liquid hight (full) and a liquid specific gravity
of 0.956. There is no seismic component to this load result using a full model.
The PRG estimate of pressure stiffening from the Zick model is shown below. The displacement estimated
effect is 36.9% - and the value from the analysis is 52%. The estimated effect on the pressure is 28.6% and
the change from the pressure stiffening analysis is a maximum of 29%. Overall, for this model, the PRG
pressure stiffening estimates from correlation equations appear reasonable.

Vessel Weight Only No Pressure

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

99

Half Full No Pressure

Note that themaximum membrane + bending stress in this model occurs when the vessel is filled with liquid
and NOT pressurized. This tends to suppor the Zick approach for pressure, and the approach where shoe
and saddle allowables are based on the nonpressurized stress state.

5.10 Effect of Rings on Stresses in Large Diameter Nozzles


The results in the tables below show the effect of adding rings in the vicinity of nozzles in a cylindrical shell,
(i.e. heat exchanger). The reduction in SIFs for different geometries of rings is shown below. These types of
comparsions are easy to run since the SIFs and k-factors are produced from NozzlePRO automatically, and
can give the user an estimate of the effect a local ring will have on nozzle stresses and stiffnesses.
Ring Application: Shell=30OD x 0.5Thk; Nozzle=18OD x 0.5Thk; Rings=4Height 5 Off Nozzle
SQR(RT) = 2.71 The axial (meridonal) distance between the nozzle and the flange is 5. When compared to
the donut bendiing distance: 5/2.71 = 1.84 times (RT)0.5. There are two influence lengths however, and
the second is the ovalization length, which is governed by the length 0.5(D)1.4T-0.4 = 0.5 x (30)1.4 x 0.5-0.4 = 77
in. The rings are certainly within the distance that should affect and SIF or k-factors that are affected by, or
cause ovalization of the shell.
Item

No Rings

SIFaxial
SIFinplane
SIFoutplane
SIFtorsion
SIFpressure
Kaxial
Kinplane
Koutplane
Ktorsion

13.29
4.077
13.64
1.868
2.660
2.695
7.261
29.76
2.337

Rings = 0.5
Thk
9.21
3.87
9.53
1.706
2.69
2.54
6.04
17.5
2.14

Rings = 1.5
Thk
7.98
4.025
8.80
1.637
2.709
2.48
5.68
15.17
2.054

B31.3
6.69
8.59

Ring=0.5
Dist=RT0.5
7.96
4.11
8.26
1.63
2.71
2.5
5.344
15.14
2.08

Ring=1.5
Dist=RT0.5
6.4
4.4
7.2
1.5
2.56
2.43
4.72
12.3
1.98

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

100

Ring Application: Shell=20OD x 0.5Thk; Nozzle=18OD x 0.5Thk; Rings=4Height 5 Off Nozzle


SQR(RT) = 2.2
Item

No Rings

SIFaxial
SIFinplane
SIFoutplane
SIFtorsion
SIFpressure
Kaxial
Kinplane
Koutplane
Ktorsion

8.844
3.988
9.299
4.243
2.717
3.191
8.028
18.243
9.511

Rings = 0.5
Thk
7.02
3.46
6.4
3.73
2.55
3.14
6.933
10.828
8.241

Rings = 1.5
Thk
6.94
3.29
5.85
3.543
2.489
3.138
6.557
9.531
7.740

B31.3
5.14
6.52

Ring=0.5
Dist=RT0.5
7.057
3.257
5.813
3.545
2.597
3.142
6.512
9.844
7.753

Ring=1.5
Dist=RT0.5
6.99
3.422
5.192
3.323
2.529
3.132
5.995
8.471
7.080

These results suggest that rings in the vicinity of nozzles can reduce the stress due to certain loads by up to
about 40%, and can increase the stiffness by about two times. There is certainly a potential for rings to
increase the load magnitude more than reducing the stress magnitude. (It is for this reason that ASME ST
LLC 07-02 does not reduce the SIF when flanges or other rigid components are placed next to branch
connections. ST LLC 07-02 does, however reduce the flexibility.)

5.11 Saddles on Pressure Vessels and Pipe


Saddles on pressure vessels and pipe, just like in-line pipe anchors (See Section 5.3, and 5.15), may not
rigidly anchor the remainder of the run or horizontal vessel to the foundation. Significant flexibility may
exist between the body of the horizontal vessel (or pipe) and the foundation.
These flexibilities are computed and printed when the user asks for SIFs thru the saddle or shoe.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

101

When computing natural frequencies of the vessel for first period determination, the flexibility of the
saddle should be used if the k-factors are much greater than 1.
For the Brownell and Young vessel, the flexibility calculations are shown below:
The default model for saddles is shown below. This is to isolate the saddle from the boundary condition,
and would be OK for vessel models where there are no rings and where the head is far enough removed
from the saddle so that there is no strengthening. (Zick might suggest that the default model is good to use
when A/R > 1.)
Factors

Default Geometry
Saddle in Middle

Saddle Close to
Rigid Flat Plate

Saddle Close to
Elliptical Head

Kin
Kout
Ktor
Flex in
Flex out
Flex tor

79058872
5544674
42126972
22.95
327.4
43.08

82776352
5676154
183945664
21.92
319.77
9.868

74861312
5583076
82384856
24.24
325.10
22.032

The example in section 5.13 demonstrates how flexibilities should be inserted into a CAESAR model for the
piping system. For vessel models and natural frequency calculations, the model is essentially the same and
is shown in the sketch below.

The members A-B and G-H are elements that have section properties equal to the cross sections of the
saddle. Members C-D and J-K are short beam members that have the cross sectional properties of the
vessel. Rotational stiffnesses produced by NozzlePRO should be inserted between D and E, and between K
and L.
The calculated natural frequencies for the shell model from FEPIpe are below:

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

102

Note: If participation factors are not printed after a dynamic solution, the user should increase the available
memory. This is done thru solution preferences, and then by entering the maximum memory to any
FE/Pipe process.

The mode shapes for each frequency can be collected in the animated viewer as shown below:

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

103

The beam model should be developed to match the beam modes shown in the shell solutions.
Corresponding mode shapes are numbered. The first beam modes is mode 8. There is still participation
with some shell deformation, but mode 8 has far more of a beam shape (no lobes along the cross section
all other modes have modal lobes in the cross section plane.)
We expect the first two modes to participate with lateral and vertical earthquake excitation. These
frequencies are at approximately 7 Hz. The first beam mode occurs at 18.9Hz in mode 8. We expect modes
1, 2 and 8 to participate most with a seismic event. (Their mode shapes can easily be imagined to shake
when an earthquake dynamically displaces the saddles.)
The first natural frequencies from the beam solution are shown below:

Beam Model WITH Saddle Flexibilities

Beam Model Without Saddle Flexibilities

The guidelines for horizontal vessels from ASCE 7-10 are:

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

104

It seems reasonable when the natural frequencies of interest are governed by shell modes that the more
rigorous analysis shown here is reasonable.
As mentioned in Sections 3.6, 5.6, and 5.13, the cylindrical shell mode analyzer can provide an almost
instantaneous evaluation of shell frequencies and shapes. For the saddle-to-saddle centerline length,
frequencies and mode shapes for this vessel are shown below:

For evaluating modal components and resulting stresses the participation factors can be used with modal
stress values from FEPipe to evalute the maximum stress in each mode due to ground excitation.
The beam-model SIFs printed from the flexibility/sif analysis are shown below:

The saddle/vessel orientations for the SIFs are shown below along with the first beam mode.
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

105

This mode corresponds to an out-of-plane displacement and so the M+B (secondary) stress multiplier is 220
from the SIF table above. The stress at the 8 end of the vertical element from 9 to 8 at the centerline
should be used with the secondary stress intensifier. The secondary stress intensifier is identical to C2 from
Section III, produce a local membrane+bending stress from the nominal beam bending stress (M/Z).
This mode is in the Y direction. If the Y direction g-load at the first mode frequency of 21Hz is 0.27gs.
For mode shape 1 (the first FEPipe load case from the eigensolution for the beam model of the vessel), the
stresses are shown below:

The bending stress at node 8 is 156 psi. To find the maximum bending stress for this location for the mode
from the beam analysis the following calculation is made:
(Stress_beam) x (Intensifier) x (Participation Factor) x (g-load from spectrum)
A portion of the frequency and participation factor report for the beam model is shown below:
Natural Frequencies
Mode:
Mode:
Mode:

1
2
3

Frequency:
Frequency:
Frequency:

21.4447 Hz.
21.8962 Hz.
42.3480 Hz.

Participation Factors
Factor Units (per g)
Mode
1
2

X Direction
0.00002
0.00000

Y Direction
0.02476
0.00000

Z Direction
0.00000
0.02422

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

106

To find the M+B stress for the beam model using the shell SIFs the stress in the first mode would be:
(156)(220)(0.02476)(0.27) = 229 psi.
At the center of the vessel at node 30 the bending stress for the mode is high in the beam model, but the
stress intensification is low. At the center:
CASE

Type: OCCASIONAL

NODE

Axial

B31.3 (2008)
Sin

Sout

Stor

SL

Scode

Sall

2.0
3.0

2996.
2996.

0.
0.

165.
156.

0.
0.

3161.
3152.

3161.
3152.

24000.
24000.

9.0
8.0

2996.
2996.

0.
0.

157.
156.

0.
0.

3152.
3152.

3152.
3152.

24000.
24000.

4.0
5.0

0.
0.

4.
161.

0.
0.

0.
0.

4.
161.

4.
161.

24000.
24000.

5.0
20.0

5.
5.

143.
19297.

0.
0.

0.
0.

148.
19302.

148.
19302.

24000.
24000.

20.0
30.0

4.
4.

19486.
27391.

0.
0.

0.
0.

19490.
27395.

19490.
27395.

24000.
24000.

30.0
40.0

3.
3.

27391.
19486.

0.
0.

0.
0.

27394.
19489.

27394.
19489.

24000.
24000.

40.0
6.0

1.
1.

19297.
143.

0.
0.

0.
0.

19298.
144.

19298.
144.

24000.
24000.

6.0
7.0

0.
0.

161.
4.

0.
0.

0.
0.

161.
4.

161.
4.

24000.
24000.

(27391)(0.02476)(0.27) = 183 psi.

The same approach can be used with the shell eigensolution. The frequency and participation factor report
for the shell modes is included below:
Natural Frequencies
Mode:
Mode:
Mode:
Mode:
Mode:
Mode:
Mode:

1
2
6
7
8
9
10

Frequency:
Frequency:
Frequency:
Frequency:
Frequency:
Frequency:
Frequency:

6.8124
7.7213
18.4273
18.4783
18.9511
22.4452
24.0617

Hz.
Hz.
Hz.
Hz.
Hz.
Hz.
Hz.

Participation Factors
Factor Units (per g)
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

107

Mode
1
2
6
7
8
9

X Direction
0.00000
0.02652
0.00000
0.00168
0.02121
0.00762

Y Direction
0.00000
0.05987
0.00000
0.00162
0.03294
0.00919

Z Direction
0.17860
0.00000
0.00006
0.00000
0.00000
0.00000

The 8th mode is the mode of interest that compares to the beam solution. The stress solution is the same,
i.e.
(Stress_M+B) = (M+B from plot) x (Participation Factor) x (g-load)
The stresses from the 8th mode can be found from the output plot menu:
Stress_M+B = (124,597) x (0.03294) x (0.27) = 1106 psi for the stress in the saddle.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

108

A next valid question to be asked, is does the SIF load and shape simulate the stress pattern observed in the
dynamic shape? From a global perspective, as shown below, the answer should be yes. The high stresses
are in approximately the same locations, and the bending shapes are the same.

The stress at the center of the vessel can be evaluated the same way as the peak stress at the saddle. The
stress in the 8th mode shape at the middle of the vessel is shown above = 11549 psi.
The stress in a seismic event due to this mode and excitation in the vertical direction would be:
Stress_M+B = (11549) x (0.03294) x (0.27) = 102.7 psi for the stress in the saddle.

5.11.0 Some Saddle Analysis Observations:


1)The saddle modeler provides a number of tools to evaluate horizontal vessels subject to external loads
such as pitch and roll or seismic loading.
2)When the flexibility and SIF factors are high, beam solutions should as a minimum be intensified using the
stress intensifiers produced by NozzlePRO. When possible, full shell models should be run and the stresses
evaluated using the shell solutions.
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

109

3)The stresses in any mode for any ground excitation response spectrum can be calculated by taking the
stress from the PRG eigensolution and multiplying by the participation factor for the mode and the load
direction, and then by the load in gs, i.e. Stress Dynamic = Stress_mode x Participation_Factor x g_Load.
4)Using beam models without flexibilities and stress intensifiers can produce artificially load results when
the modes are intensified.
5)The analysis difficulty can be reduced when the vessel is thicker, or has a number of external stiffening
along the vessel length.
The allowable loads report from the SIF and flexibility report is shown below for this saddle. The allowed
vertical load is 482,129 lb. This load can be applied to the saddle:
a)From directly overhead if a vertical g-load of 1.0 is added to the vessel weight. When from directly
overhead, there is very little moment induced loading. (See Fig. A below).
b)By applying the load at the end of a cantilever where a very high potential moment can exist. (See Figs B
below).
c)By applying the load in the middle of a pinned end beam. Here the load is distributed to both ends of
the beam, and the moment magnitude is a function of the length of the beam. (See Fig. C.)

Fig. A Load applied from Overhead

Fig. B Load at End of Cantilever

Fig. C Load in Middle of Pinned End

The allowable vertical and horizontal loads are produced for conditions that are more like Fig. A and Fig C,
where the vertical loads are carried almost equally away from the support on both sides. The allowable
shoe/saddle load report includes the minimum allowed forces and moments summarized at the beginning.
These are followed by the sustained and secondary load conditions and allowables constrolled by the
support, and by the pipe. This lets the user determine which component of the design is controlling the
allowable load. A comprehensive set of notes follows the allowable load table which should be reviewed by
the user applying the loads.

5.11.1 Axial, Vertical, In-Plane, Out-Plane and Torsional Orientations

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

110

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

111

Note that the total weight load per saddle for this model is 200,000 lb. One would hope that the allowed
axial load is at least this big.
Load Components:
The finite calculation shows the actual distribution of the load in a saddle or pipe shoe. In the example
below, the high stress is in the saddle outer flange plates. The user must look at the magnitude of the
stress and and decide if plastically deforming the saddle plates is acceptable for origin of the given loads.

ASCE 7-10 para. 15.7.3 states that Structural members that are part of the seismic force-resisting system
shall be designed to provide the following: a. Connections to seismic force-resisting elements, , shall be
designed to develop o times the calculated connection design force.. o is the overstrength factor and
is the ratio of the estimated actual maximum design force in the structure considering the extent of plastic
deformation that is likely to occur divided by the load Fp. Fp x o is the estimated actual load is not
proportional to the displacement when plastic deformation occurs. This is an adequate approach for limit
load or collapse studies, but for fatigue, this may underestimate the maximum strain in the geometry when
the strain (E/) is computed on an elastic basis. Using Fp x o to compute fatigue damage for an ASD B31
or ASME VIII-2 analysis is similar to using the actual load in the Markl test to compute the SIF instead of the
projected elastic load. If a fatigue analysis is based on an elastic analysis, the load used must be based on
the actual displacements and an elastic model. (LRFD or strength analyses are different.) See ASCE 7-10
Section 15.7.3.
Seismic survival in ASCE 7-10 in the section on Ice Loads Atmospheric Icing states the following with
respect to fatigue contributions.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

112

In the section C13.3 Seismic Demands on Nonstructural Components


ASCE 7-10 includes Figure C13.3-4 which is very meaningful to pipe stress engineers, and defines the
relative movement of close approximation structures. Piping engineers will often include the components
A and B separately, and do not include the possibility that the A and B modes shown can act in the same, or
an opposite sense.

and in C13.6.5 mentions:

In C13.6.8 Piping Systems are mentioned:

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

113

B31E-2010s states:

Users should be aware that passing a fatigue analysis does not assure that a limit load study will also pass.
In general, it is the opinion of PRG that seismic loadings may cause collapse, and may contribute to the
fatigue life lost, and then in significant seismic events, both affects should be quantified and demonstrated
to be within Code limits. It is also the opinion of PRG that ASD and LRFD design loads may not adequately
describe the extent of fatigue damage that occurs during a seismic event.
For equipment and components, B31E also adds the following:

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

114

5.11.2 Conclusions Regarding Seismic Designs:


1)Elastic loads used for ASD design from ASCE and comparable Codes will underestimate actual maximum
strains anticipated during a design magnitude earthquake that occurs within the 2% probability window for
50 year return period.
2)Piping systems tend not to suffer apparent damage during earthquakes even though ground motions can
be significantly greater than those most commonly designed for. High shaking forces are damped due to
plastic deformation and the plastic deformation often results in lower loads throughout the piping and
structural system as members plastically deform.
3)It is likely that the vast majority of piping systems designed under ASCE, IBC, or similar standards since
2000, will survive a design seismic earthquake although may undergo significant plastic deformation at high
stress areas. The degree of plastic deformation experienced will be a function of the actual ground motions
experienced.
4)Some locations, such as southern California have regular seismic events of relatively low magnitudes
compared to design earthquakes. These events should likely be designed for on an elastic basis or by
adding damage due to fatigue from the seismic events, to fatigue from thermal, pressure, and other
cyclically occuring events. The variety of small magnitude earthquakes can be evaluated using some form of
magnitude probability distribution and their overall contribution to fatigue life determined. Most civil
projects can produce expected operating basis earthquakes ground motion and seismic anchor movement
and a probability density function. These inputs should be evaluated along with the long term design basis
earthquake used in the ASCE approaches. When it is known that there will likely be a seismic event during
the operating life of a facility, and it is desired to continue operating the facility without constraint once the
event occurs, then actual ground excitations should be used to produce response spectra for use in static or
dynamic analyses. Actual seismic events in piping systems should be evaluated with the following features
in mind.
1)When the system natural frequencies are higher than the seismic ZPA, a static analysis can be used
for the evaluation.
2)Most damage during a seismic event occurs because of the relative displacement of small and large
components that are connected together. Shaking forces have been found to be three times higher at
the top of most structures that are of a typical plant height, and so users should be aware of the
relative movement between supports at different levels of a structure, i.e level 1 moving in the +X
direction and level 2 moving in the X direction.
3)Vessels with large diameter pipe coming from the vessel that have similar natural frequencies can
experience an amplication of displacement and acceleration due to an interaction of the similar
frequencies and modes. This interaction effect should be investigated if the system is sensitive to
displacement limits. Examples of sensitivity include, but are not limited to:
a)pipe is refractory or glass lined

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

115

b)small, constrained pipe is attached to large pipe that will move, i.e. the large pipe will drag the
small pipe along with it, and the small pipe should have the flexibility to accommodate that movement.
c)natural frequencies and mode shapes of vessel or structural systems interact to provide amplified
responses to a given set of ground motions.
d)other single load equipment will be operating thru a seismic event, i.e. turbines and compressors.
Large equipment manufacturers can generally provide some idea of how much load nozzles can take
during seismic events when the equipment is running.
e)Any configuration where plastic deformation will be absorbed by a relatively small portion of a
connection or weld.
f)other sensitive items such as, bimetallic joints, clamps, or components where guillotine failures or
gross collapse can occur suddenly without prior plastic deformation.
g)short, weak branch connections without flexibility that extends between relatively moving floors
or larger lines/vessels.
Evaluating the Actual Magnitude of Plastic Strain
For vertical and horizontal vessels in ASCE 7-10 Chapter 15, o values of 2 are given. For piping, Rp values
can be as high as 12. This suggests that actual strain might be as high as the strain calculated during an
elastic analysis of the ASCE 7-10 ASD loadings multiplied by a value of between 2 and 10.
Users having actual expected ground movements can use those providing models are constructed from
grade. Worst case relative moments should be considered and included in a cummulative damage
evaluation. Where crack appearance analysis is available, when this threshold is passed for measured
seismic events, inspections of critical areas can assure that unepected damage has not occurred in the form
of undetected flaws. This is more important where multiple seismic events are expected for systems that
otherwise are susceptible to fatigue damage due to operation.
A review of any seismic analysis should be made after a seismic event occurs, and inspection of high stress
areas conducted. The PRG first measureable crack program can be used to select sites for inspection.

5.12 Brownell & Young Horizontal Vessel Model


Input for the the Brownell and Young horizonal vessel model used in Sections 5.11 thru- 5.13, is described
below. Input files for these models are included in the models folder. See para. 1.3.1 above for a
description of where these input files can be found.
Steps

Forms and Figures

Notes

1
2
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

116

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

117

10

11

12

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

118

5.13 Calculating Natural Frequencies in Horizontal Pressure Vessels


The model used in this example is the Brownell and Young model described in more detail in Section 5.11
and 5.12. A shaded image of the finite element model is shown below:

Large diameter shells often have low natural frequencies associated with shell modes of vibration. These
frequencies can be lower than the fundamental beam modes of vibration. For pipe, with smaller d/t ratios
this tends not to be true for typical pipe lengths, but for vessels, the opposite is often the case. The
magnitude of the lowest natural frequency can indicate whether the horizontal vessel can be classified as
an ASCE 7-10 rigid body. (See ASCE 7-10 Chapter 15.) For the Brownell and Young example vessel shell
modes of vibration are more easily excited by the energy band in a typical earthquake.
Dynamic analysis are run from the FEPipe interface. Many of the features shown here are available from
NozzlePRO while others require FEPipe.
To interrupt the NozzlePRO run path and use the FE/Pipe input editor, the user must check the Use FEPipe
Editor checkbox. When this box is checked, the FEPipe input processor is started after NozzlePRO is
finished building the model. This run process using FEPipe is termed the Advanced Run Path in the
diagram below.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

119

For each natural frequency computed, a mode shape and stress distribution in that mode shape is also
computed. The stress magnitudes are based on a unity normalized eigenvector and participation factors
for each of the X, Y and Z axes. The X, Y and Z axes are used so that stresses due to seismic base excitation
in those directions can be estimated on a modal basis and scaled using the design spectra g loads.
To compute natural frequencies for the vessel model the user must decide which of the mass and stiffness
distribution options should be included in the model. Generally this involves making a few runs to see
which of the model characteristics are critical for the analysis. Model options to evaluate include:
1)Pressure stiffening generally not included, but used to get some idea of how the vessel will behave
when pressurized. Pressure stiffened models should have higher natural frequencies and lower stresses
2)Liquid level Half way full models can sometimes cause more vertical displacement due to weight than
full models and so occasionally the user looking for possible worst case design conditions should include the
half full vessel model.
3)Liquid adds mass to the vessel. The liquid mass can be added by increasing the density of the vessel
plate, or using liquid model type 4 which applies additional point masses over the whetted surface of the
vessel in proportion to the pressure distribution (more on the bottom or side depending on the g
orientation) where the total is equal to the liquid weight. Users should almost certainly analyze large d/t
vessels both with and without liquid. The liquid mass load option 4 (shown below) provides the most
effective way to include liquid in the analysis of a horiontal vessel. The Pressure Load Option cell in the
FEPipe general form is used to activate this dynamic fluid option and is shown below. The option #4 is only
available in the 2014 version of FEPipe. It is used exclusively for dynamic analysis.

The checkboxes and switches to activate from NozzlePRO to run a dynamic analysis through FEPipe are
given in the following table.
Dynamic Analysis Steps from NozzlePRO in FEPipe

Steps

Forms and Figures

3
4
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

120

The solution will take about


20 minutes to compute the
10 mode shapes and
frequencies.

Coarse meshes can often produce natural frequencies and mode shapes that are sufficiently accurate for a
dynamic analysis using shell elements. These shapes may not produce highly accurate stresses, but overall
shapes and frequencies can be reasonably generated. Users that need to make numerous runs are advised
to evaluate the sensitivity of the dynamic solution to the mesh density and enter a reduced mesh multiplier
on the optional form when there is not a significant difference. The natural frequencies and participation
factors for the Brownell & Young horizontal vessel are given below.
Natural Frequencies
Mode:
Mode:
Mode:
Mode:
Mode:
Mode:
Mode:
Mode:
Mode:
Mode:

1
2
3
4
5
6
7
8
9
10

Frequency:
Frequency:
Frequency:
Frequency:
Frequency:
Frequency:
Frequency:
Frequency:
Frequency:
Frequency:

5.2004
5.8947
9.0382
10.2504
11.3815
14.0696
14.1106
14.4668
17.1359
18.3682

Hz.
Hz.
Hz.
Hz.
Hz.
Hz.
Hz.
Hz.
Hz.
Hz.

Participation Factors
Factor Units (per g)
Mode
1
2
3
4
5
6
7
8
9
10

X Direction
0.00000
0.04550
0.00000
0.00463
0.00000
0.00006
0.00298
0.03636
0.01306
0.00149

Y Direction
0.00001
0.10274
0.00001
0.01117
0.00001
0.00008
0.00296
0.05648
0.01576
0.00090

Z Direction
0.30656
0.00002
0.10629
0.00001
0.08023
0.00014
0.00003
0.00000
0.00000
0.00003

This suggests that there are a number of shell modes below the rigid body cutoff in ASCE 7-10 of 16.6 Hz.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

121

A stress state is calculated for each of the modes. To get the actual stress for the mode due to a given
direction of base excitation, the modal stress magnitudes must be adjusted by the modal participation
factor and a g load for the frequency. This is a dynamic feature added in the 2014 version of FEPipe.
Modal Stress due to Ground Motion = (g-load from Spectrum) x (Participation Factor) x (Stress)
The participation factor is defined as follows:
1)Solve eigenproblem to obtain P and W, where P is a series of eigenvectors and W is equal to the natural
frequency squared. The equation solved for a single mode is: KP WMP = 0, where K=system stiffness
matrix and M=system mass matrix. W is equal to the natural frequency squared.
The vector P from the eigensolution can be scaled by any value without disturbing the equality solved for
above, and so for convenience the eigenvector P is mass normalized (scaled) such that PT M P = 1.
The participation factor printed for each mode for each direction after a solution is found for each mode for
the three global displacements from:
(Participation Factor) = {P}T[M]{D}/w2. To use the participation factor to determine the stress from a
seismic ground or base spectrum:
1)Use the natural frequency (w)for the mode of interest and look up the response acceleration from the
spectrum in gs. Multiply this g value by the (participation factor), to get the stress scale factor from the
modal analysis. The scale factor can be used to multiply displacements, forces and stresses.
2)The actual response due to ground motion will be the summed response of multiple mode and directional
reponses. Often the response of a few key modes can provide a good estimate of the magnitude of the
stress that will result from a seismic event.
A typical response spectrum for Oakland California in the zip code 94601 is shown below.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

122

The mode shapes and frequencies for the Brownell and Young vessel are included in the tables on the
following pages.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

123

Mode

Shape

Notes

Mode 1 at 5.1321 Hz.

Mode 2 at 5.825 H

Mode 3 at 9.0453 Hz

Mode 4 at 10.2360 Hz

Mode 5 at 11.4009 Hz

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

124

6
Mode 6 at 13.3246 Hz

Mode 7 at 14.140 Hz

Mode 8 at 14.19 Hz

Mode 9 at 16.7207 Hz

10

Mode 10 at 18.3627 Hz

Boundary Conditions at the Saddles


A careful examination of each end of the vessel in the 9th mode shows how the saddles are essentially free
to rotate, i.e. are supported on pinned ends and can rotate about the base of the saddle in a direction that
is normal to the vessel centerline and in the horizontal plane. It is thought that this boundary condition is
the most conservative.
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

125

The user can ask for the saddles to be rotationally rigid to prevent this rotation.

The differences in natural frequencies are shown below.

Mode

Pinned End
Saddles

1
2
3
4
5
6
7
8
9
10

5.1321
5.8250
9.0453
10.2360
11.4009
13.3247
14.1400
14.1912
16.7207
18.3627

Fixed
Rotation
Saddles
5.2034
5.8985
9.0478
10.2583
11.4181
14.1423
14.1912
14.4721
17.1590
18.3656

The boundary condition with the saddle baseplates fixed is shown in the displaced shape image of mode #8
below. Note how the sliding saddle on the right slides in the X direction but does not rotate, while the fixed
saddle at the right does not translate or rotate. The low frequencies for this problem are not significantly
affected by the boundary condition of the saddles, but stresses due to the added moments thru the saddles
should be evaluated. (Very high local stresses that will not suffer plastic deformation well are subject to
crack-like flaw formation.)

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

126

To quickly determine how many natural frequencies are below a certain value the Sturm Sequence
property may be used. The frequency range calculation takes the same time as a single static solution and
so is considerably faster to run than the full system eigensolution. If the model does not have frequencies
in the specified range of interest then no further dynamic calculations are required. The key input for the
Sturm Sequence range calculation is highlighted in red below from the General Form in FEPipe.

The output from the range calculation is a single number and that is the count of the number of natural
frequencies that are below a given value. For the model above with the no-rotation saddles, there are
eight natural frequencies below 16.66 Hz. The output from the range calculation is shown below.

All other output from a range calculation can be ignored.


Another way to get a quick estimate of the vessel natural frequencies and mode shapes is to run the
cylinder frequency calculator available from FEPipe and NozzlePRO. Output from the cylinder frequency
and mode calculator is shown below.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

127

The comparison between the analytical shell-cylinder solution and the saddle supported vessel shows that
the principle vessel shapes

Mode

Shell Saddle
Model

Analyzed
Shell
Cylinder
Solution

1-2

5.2-5.8 Hz

9Hz

Shape

Two lobes
3

12.2/13.8 Hz

15 Hz.
Three lobes

4
5
6

17.8Hz

19Hz

First cantilever

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

128

The geometrically accurate model shell modes and frequencies have lower natural frequencies than the
calculated shell modes. If the analytical solution is below the 16.6 Hz threshold it is almost a certainty that
the the actual frequencies will be below the threshold also. Beam and shell modes from the analyzed shellcylinder solution is shown in greater detail above. These solutions take only a few seconds to produce.
Assume that beam theory is to be used to find the natural frequency for the vessel on saddles as suggested
by the commentary to ASCE 7-10 Chapter 15.7.14:

6dof Beam natural frequencies are included in the table below:


Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

129

Mode
1
3
4
6

6dof Beam
18.9 Hz
61.399 Hz
64.18 Hz
118 Hz.

Using beam degrees of freedom suggests that the vessel natural frequency is higher than 1/0.06s =
16.66Hz. This high a frequency would allow the vessel to be analyzed as if it were a rigid component.
The shell based calculations above show that shell modes of vibration are lower than beam modes for this
vessel. The 18dof beam element in FEPipe also shows the lower mode of vibration, i.e. 14.08 Hz. (See the
18dof natural frequency table below.)
Natural Frequencies (18dof Beam Modes) First 56 modes duplicates removed.
Mode:
Mode:
Mode:
Mode:
Mode:
Mode:
Mode:
Mode:
Mode:
Mode:
Mode:
Mode:
Mode:
Mode:
Mode:
Mode:
Mode:
Mode:
Mode:
Mode:
Mode:
Mode:
Mode:
Mode:
Mode:
Mode:
Mode:
Mode:
Mode:
Mode:
Mode:

2
4
6
8
10
12
14
16
17
18
19
21
22
24
26
29
30
32
33
36
38
40
42
44
46
48
50
51
53
54
56

Frequency:
Frequency:
Frequency:
Frequency:
Frequency:
Frequency:
Frequency:
Frequency:
Frequency:
Frequency:
Frequency:
Frequency:
Frequency:
Frequency:
Frequency:
Frequency:
Frequency:
Frequency:
Frequency:
Frequency:
Frequency:
Frequency:
Frequency:
Frequency:
Frequency:
Frequency:
Frequency:
Frequency:
Frequency:
Frequency:
Frequency:

14.0800
18.7420
24.4305
26.4204
31.4966
34.0848
40.1145
51.7790
58.4917
60.5603
61.4626
60.5603
62.5458
63.2897
65.7883
68.7735
76.2309
81.5816
76.6458
85.7078
90.9568
95.6112
98.7859
106.0880
117.5234
117.9767
116.8757
123.8539
127.4182
129.5341
127.4184

Hz.
Hz.
Hz.
Hz.
Hz.
Hz.
Hz.
Hz.
Hz.
Hz.
Hz.
Hz.
Hz.
Hz.
Hz.
Hz.
Hz.
Hz.
Hz.
Hz.
Hz.
Hz.
Hz.
Hz.
Hz.
Hz.
Hz.
Hz.
Hz.
Hz.
Hz.

Boundary conditions at the ends of the analytical solution can be similar to the the boundary conditions at
the ends of a head, i.e. radial constraint. When rigid end conditions are applied, the first twin lobed model
that includes ovalization at the end of one of the cylinders goes away. This can also be seen by looking at
the mode shapes from the string or nozshell template.
Natural Frequencies of shell finite element analysis of cylinders.
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

130

Mode: 1 Frequency:
Mode: 2 Frequency:
Mode: 3 Frequency:
Mode: 4 Frequency:
Mode: 5 Frequency:
Mode: 6 Frequency:
Mode: 7 Frequency:
Mode: 8 Frequency:
Mode: 9 Frequency:
Mode: 10 Frequency:
Mode: 11 Frequency:
Mode: 12 Frequency:

12 Hz. (two lobed mode)

12.7480 Hz.
12.7500 Hz.
14.0690 Hz.
14.0690 Hz.
18.7105 Hz.
21.2859 Hz.
21.2860 Hz.
24.2469 Hz.
24.2469 Hz.
24.2907 Hz.
26.2851 Hz.
26.2851 Hz.

14 Hz. (three lobed mode)

Mode 5 18.71 Hz. (first beam bending mode)


The first beam mode of the shell cylinder solution mode 5 occurs at 18.71 Hz. This is very close to the
6dof beam solution. For the lowest beam natural frequency, the 6dof beam first mode is adequate. For
potentially lower shell modes, the 6 dof beam solutions are incapable of finding the mode shapes or the
stresses.
When shell modes are the lowest modes and are excited by energy in a seismic range, the relatie
displacement in the mode can contribute to the stress state and should be evaluated. These stresses will
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

131

never be estimated by a 6dof beam solution unless some type of SIF is developed from the shell solution or
from test.
The circumferential modes for 18dof beam elements are shown in the table below for reference.
Eighteen Degree of Freedom Beam Element Shell Modes

18dof
18dof
DOF=13
Inplane Mode 1 sin(2*t)

DOF=11
sin(6*t)

cos(2*t)

cos(4*t)

18dof
18dof

18dof

DOF=10
sin(4*t)

DOF=7

18dof
DOF=8

DOF=12
cos(6*t)

18dof
DOF=9

The use of the different participation factors, mode shapes and models to evaluate dynamic stresses are
shown in Section 5.11.

A very crude estimate for the design of stiffening rings can be taken from the following equation:
nI = Qr3h/(SL2);

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

132

where: Q=weight of half the vessel, r=radius of vessel (inches), h=height of stiffening ring, S=allowable
stress (psi); L=length of total vessel; n=number of stiffening rings in half of the vessel, I=inertia of the
stiffening ring.
For the Brownell and Young vessel, assume the stiffening ring height will be 6 and there will be 2 close to
each saddle.
nI = (2)(200,000)(603)(6)/(12,000 x 8162) = 32.43 in.4. If n=2, and I=bh3/12; b=0.9
These rings can be entered into the ring description form .

Rings are described as a rectangular section. If angles or channels are used, then the base (b) and height (h)
dimensions should be selected so that the area and moment of inertia of the ring about its neutral axis are
properly simulated.
Rings are defined from the end of the support and are placed at the ring of nodes closest to the ring
location. Users should experiment with ring placement using a single ring before multiple rings are
entered. The measurement tool in the 3D viewer makes finding the correct length along the half vessel
length much easier:

Rings are added to the NozzlePRO finite element model as a constant ring of elements around the vessel
contacting the vessel shell only regardless of where the circumferential rings are placed. For example, if
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

133

rings are placed at the inner and outer edge of the saddle, the ring still will circle the vessel, and only be in
contact with the shell. For design purposes, this has shown to be an adequate approach, but the user
should be aware of this modeling assumption. If the rings are inside the vessel, a negative height should be
entered.
The differences in natural frequencies are shown below.

Mode

Pinned End
Saddles

Fixed
Rotation
Saddles

1
2
3
4
5
6
7
8
9
10

5.1321
5.8250
9.0453
10.2360
11.4009
13.3247
14.1400
14.1912
16.7207
18.3627

5.2034
5.8985
9.0478
10.2583
11.4181
14.1423
14.1912
14.4721
17.1590
18.3656

Fixed
Rotation
Saddles with
Rings
5.8069
6.6439
10.3728
10.6150
12.7016
14.6743
15.3638
15.4409
18.4727
18.4730

For this vessel and these rings it can be seen that the modes of interest are along the center of the shell
away from the saddles, and so to affect the natural frequencies associated with these modes the rings will
need to be placed along the length of the shell. There are two sets of modes that are below 16.66 Hz.
Mode 1 shown below has a single lobe between the saddles. Mode 6, also shown below, has two lobes
between the saddles, and essentially a shell node in the middle.

Case 1

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

134

Case 4

Case 6

To eliminate these modes there should be a ring put at the maximum displacement location of each mode.
For mode 6 (above), the two rings should be at the centers of the lobes while for mode 1 the single ring
should be at the middle of the vessel. This suggests the rings below.

The ring processor finds the closest ring of nodes and puts the ring at that location. When there are not
many nodes along a particular shell length the user should check the location of the ring and make sure
that its close enough to the desired location. If it is not close enough, then the mesh density can be
increased to improve the ring placement.
The updated model with the additional rings is shown below.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

135

The differences in natural frequencies are shown below.

Mode

Pinned End
Saddles

Fixed
Rotation
Saddles

1
2
3
4
5
6
7
8
9
10

5.1321
5.8250
9.0453
10.2360
11.4009
13.3247
14.1400
14.1912
16.7207
18.3627

5.2034
5.8985
9.0478
10.2583
11.4181
14.1423
14.1912
14.4721
17.1590
18.3656

Fixed
Rotation
Saddles with
Rings at
saddles
5.2228
5.9200
9.1293
10.2875
11.4799
14.1976
14.2735
14.4922
17.2267
18.3815

Fixed Rotation
Saddles with
Rings at
middle and 2nd
mode
8.879
10.544
13.445
14.7305
19.687
20.93
23.064
23.0734
24.519
26.027

The extra rings have incresed the lowest mode by 1.7 times, but still leaves it about half of the rigid body
cutoff frequency. The twin lobed mode has now shifted to 19 Hz and so only the first mode and modes
similar to the first mode are below the 16.6Hz ASCE 7-10 cutoff. Note that in the mode shapes below, the
rings are not stiff enough to significantly impact the mode shape. This is how the user can know if adquate
size is present for the rings to have an affect on dynamic response of the vessel. Rings cannot be assumed
to provide a rigid cross section just because they add adequate strength for external pressure.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

136

Mode 1

Mode 4

Mode 5

The input for these reinforcing rings is shown below. Note that half of the thickness is used at the zero
location, since two rings will be placed there. (The model is built in halves.)

It does look like mode 4 will be the first cantilever mode. No number of rings will eliminate cantilever mode
shapes. The frequency of the single vessel mode in between the saddles if the saddle section is rigid can be
found from a beam or theoretical shell analysis. Those results from the FEPipe theoretical shell calculation
are shown below with the plotted associated shapes:

Dynamic Natural Frequencies from the Shell Mode Frequency Calculator

Rings can be stiffened by increasing both their height and stiffness. The height of the ring has a cubic effect
on the stiffness, while the thickness of the ring has a scale multiple, i.e. if the ring is twice as thick, the
stiffness is twice as great. If the ring is twice as high the stiffness is 23 = 8 times as stiff.

Mode

Pinned
End
Saddles

Fixed
Rotation
Saddles

Fixed
Rotation
Saddles with

Fixed Rotation
Saddles with
Rings at middle

Fixed Rotation
Saddles with
Rigid Rings at

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

137

1
2
3
4
5
6
7
8
9
10

5.1321
5.8250
9.0453
10.2360
11.4009
13.3247
14.1400
14.1912
16.7207
18.3627

5.2034
5.8985
9.0478
10.2583
11.4181
14.1423
14.1912
14.4721
17.1590
18.3656

Rings at
saddles
5.2228
5.9200
9.1293
10.2875
11.4799
14.1976
14.2735
14.4922
17.2267
18.3815

and 2nd mode


8.879
10.544
13.445
14.7305
19.687
20.93
23.064
23.0734
24.519
26.027

middle and 2nd


Mode
18.6643
19.512
34.980
39.027
40.6357
40.6523

The very large rings turned the shell modes into essentially beam modes. The frequencies of extra heavy
rings are shown in the right column in the table above, and the associated mode shapes are shown below.

ASCE 7-10 also suggests that the weight of liquid in the shell should be considered. The weight of the fluid
can be added to the dynamic model as point masses at whetted nodes by using liquid model option 4. See
the input for this below.

Using option 4 to include the weight of the liquid will reduce the natural frequencies of the vessel even
further.
Saddle flexibilities and the affect it can have on natural frequencies is shown in Section 5.11. The effect of
an inline saddle that is not really rigid is also seen in the example 5.15.
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

138

The effect of pressure stiffening on the shell can be considered by using the pressure stiffening matrix when
performing the eigensolution. The steps to take in FEPipe from NozzlePRO are shown below:
1)First make the necessary static runs and be sure that displacements and stresses are as expected.
2)Next run the first dynamic calculation as performed above to inspect mode shapes and frequencies.
3)To find the effect of pressure stiffening on the frequency calculation:
a)Make the static run
b)Inspect the load cases to find the case of weight + pressure. (In the example model above it is case #2.)
c)Ask to stiffen the solution:
calculation.

<and>

run the dynamic

d)Submit and wait for the calculation:


e)The difference in natural frequencies is due to the pressure stiffened solution. Note that the ovalization
modes have shifted upward, and that the beam bending modes have not shifted much. (We saw by
comparing them with beam solutions, that the saddle anchors did not really fix the ends of the vessels,
and that the saddles for this geometry really more like pin the end of the vessel.
This run sequence is a two run process unless as part of the stress stiffening pass the assembled stiffness
matrix is saved. For subsequent analyses the geometry or mesh density cannot be changed, although the
loads (except pressure) CAN be changed. For all subsequent cases after the first stress stiffening case, the
stiffened stiffness matrix can be reused.

5.14 Participation Factors Simple Example


Take the following simple cantilever. OD=14, T=0.375, L=140, density = 0.283 lb/cu.in.; First and second
mode natural frequencies = 27.08 Hz.
Assume the g-load from the ground response spectrum for 27.08 Hz. Is 1g.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

139

Natural frequencies and participation factors from FEPipe are shown below:

The stress due to the first mode shape is 36,195 psi. The participation factor for an X direction earthquake
is the highest for mode 1 and is 0.02070 per g. If the base acceleration is 1g. then the stress due to an x
direction earthquake in the first mode would be:
(0.02070 per g)(1g)(36,195 psi) = 749 psi. (Note that the participation factor is found from (0.020702 +
0.000182)0.5 = 0.0207.
The same approach can be taken with a shell model of the cantilever. The shell model and stresses from
the first mode are shown below. The participation factor report appears to the right of the stress figure.
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

140

The M+B stress in the first mode of the cantilever will be (36,566)(1g)(0.02072)=757.6 psi.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

141

5.15 In-Line Anchor Analysis


Gas filter internals were breaking as the existing system underwent high thermal loads due to thermal
expansion from the supply gas coolers. An anchored loop was put in the inlet side of the filter to keep all
loads off the filter nozzle. The 10 pipe anchor was purchased from the vendor that supplied all other
supports for the line.

The allowable moment resultant on the filter nozzle was 2000 ft.lb. The resultant loads on the in-line
anchor are given for each load case in the figure below: Each resultant moment is less than 2000 ft.lb.

Moment and force resultant on filter nozzle after configuration change

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

142

Moment resultants acting on the in-line anchor location.

The maximum moment is the Z moment and it is 27880 ft.lb. For the 10 standard wall pipe, the section
modulus is Z = (3.14)(5.19252)(0.365) = 30.9 cu.in. The nominal thermal bending stress in the pipe at the
anchor would be M/Z = (27880)(12) / 30.9 = 10,826 psi.
The pipe stress analysis software showed that there was no overstressed areas in the pipe:

Using a typical approach for this problem, the stress analyst believed that the problem was solved. The
figure below showed the results of the flexibility analysis.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

143

Configuration Change Analysis Results

During the model review the project manager asked to see the detail of the in-line anchor. The anchor
detail is shown below.

The anchor forces acting on the inline anchor are well below the limits listed for the component. The
resultant forces acting on the restraint are shown in the figure below.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

144

Resultant Forces on Pipe Anchor Protecting Filter Nozzle

The actual to-scale size of the support for the 10 line as viewed down the axis of the pipe appears as
shown below:

The project manager noted how the pipe anchor looked similar to a rectangular attachment in WRC 107
and asked about the bending moments on the anchor. The FX loads at the baseplate are shown below:

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

145

FX Moments Acting on Pipe Anchor

The height of the pipe shoe is 4 so the moment at the surface of the pipe shoe for a WRC107 calculation
would be 4 x 1223 = 4892 in.lb. The MZ bending moment (torsion) transmitted thru the anchor is much
higher than the force times the moment arm however:

MZ moment acting on the inline anchor from flexibility analysis. (27870 ft.lb.) (334,440 in.lb.)

The WRC 107 analysis was conducted with the moment acting at the base of the support plus the force
acting at the base times the 4 moment arm. The membrane+bending stress from the WRC 107 analysis
(below) is 132,252 psi. and the result is shown in the panel below. This analysis is taken from the FEPipe
model GENERAL model input form.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

146

If the same moment is entered into the pipe shoe modeler in NozzlePRO the maximum stress shows to be
almost three times greater the stress predicted by WRC 107. A part of the reason for this is that the
NozzlePRO analysis included the actual configuration of the support as welded to the bottom of the pipe.
As an be seen in the figure below, the outer edges of the end plates create points of stress concentration
on the pipe.

The displaced shapes suggest that there may be considerable lateral displacement associated with a
torsional load on the support, i.e. that the in-line anchor is actually very flexible and not rigid at all. The
same tool used to interrogate the stress plots show the displacements at the pipe support also. The
floating panel used with the thermometer tool is shown in the figure below.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

147

Flexibilities are computed for the in-line pipe anchor when SIFs are requested in NozzlePRO from the
shoe/saddle wizard. The option where SIFs are produced due to load thru the run and pipe shoe are shown
below. The output from this selection is also shown below. The allowable loads give the allowable
moments that may be applied to the pipe shoe in the torsional direction.

It can be seen from the report that there is a high torsional SIF suggesting that for this combination of pipe
shoe anchor and pipe diameter-thickness, the allowed torsional moment capacity will be generally low. If
the original nominal torsional stress was 10,826 psi. Multiplying by the secondary stress index of 34.558 =
10,826 x 34.558 = 374,124 psi.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

148

The flexibility factors are also given in the SIF report:

A large torsional flexibility is seen at the pipe anchor. To get a realistic load both the SIFs and the
flexibilities need to be inserted in to the piping system for the flexibility analysis. The pipe shoe used as an
anchor is shown in the figure below (not to scale).

The baseplate will be bolted to the pedestal frame a large concrete foundation. The recommended pipe
model is shown in the sketch below. The properties of the run pipe are used between the node 1500 and
1000. The baseplate anchor is placed at node 1000. (The same model is used for guides and limit stops.)

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

149

When this model is put in the piping system, the loads on key components changed as follows:

Mz at 1000
FR at 1000
Stress at InLine
Anchor
Fr at Filter Nozzle

27877 ft.lb.
1696 lb.

Actual Inline
Anchor
FLEXIBLE
2064 ft.lb.
2543 lb.

84,225 psi

14,101 psi

886

1067 lb.

Mr at Filter Nozzle

1281 ft.lb.

4714 ft.lb.

Rigid Inline Anchor


RIGID

Item

Notes

Filter is
overloaded

The rigid anchor keeps the loads on the filter vessel down, but overloads the inline anchor. An actual
model of the inline anchor shows much higher displacements and a significant load redistribution.
The highest stress in the inline anchor is in the anchor plate, and so thicker plate, side flanges and a repad
will be used in a redesigned anchor to attempt to increase the stiffness and reduce the SIF.
The shoe width was changed from 4 to 7. A 0.365 thick repad was added. The shoe flange plate
thickness was changed to 0.5 and plates at the front and back will stay and be increased to 0.5 also.
When this model is put into the piping system flexibility analysis, the loads on key components changed as
follows:

Item

Rigid InLine
Anchor
RIGID

Actual Inline
Anchor
FLEXIBLE

Modified InLine
Anchor
FLEXIBLE 02

Notes

Mz at 1000

27877 ft.lb.

2064 ft.lb.

12053 ft.lb.

Support is picking up more


moment, but is designed to
accommodate the extra load.

FR at 1000
Stress at InLine
Anchor

1696 lb.

2543 lb.

1898 ft.lb.

84,225 psi

14,101 psi

15,310 psi.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

150

Fr at Filter Nozzle
Mr at Filter Nozzle

886
1281 ft.lb.

1067 lb.
4714 ft.lb.

570 lb.
2255 ft.lb.

Filter is lightly overloaded

The filter nozzle was overloaded by 255 ft.lb. although after contacting the manufacturer this load would be
permitted. The redesigned inline anchor provided the proper flexibilities, stiffnesses and local strength.

5.16 Wind Vibration


A vessel that is 360 from tangent-to-tangent operates in a gulf coast area. The operator would like to run
the unit up to wind speeds of 50 mph. The vessel has internal scrapers that must maintain their adjusted
shell clearance. Is there any potential for wind loads up to 50mph to excite any of the shell modes of
vibration and potentially interfere with any of the scrapers? The tolerance on the scraper radial distance is
thought to be 1/32, but is probably closer to 1/16, and then only greater scraper wear becomes a
problem. Much larger errors can jam the scraper blades in place.
Vortex shedding occurs at when the Strouhal number (0.2) is equal to fD/V, i.e. St = fD/V, where f is the
frequency of flutter, D is the diameter and V is the wind velocity. Using the Strouhal number of 0.2, the
vortex shedding expression can be rearranged to find find the maximum frequency excited by the
maximum wind velocity to design for.
f = (St)V/D = (0.2)(50 mph)(5280 ft./mile)/[(72)/(12)]ft. x 1hr/3600sec = 2.4 Hz.
Since the wind can blow at speeds less than 50mph, (although will produce lower loads and potentially
smaller radial shell displacements), it is desired to raise the natural frequency of the chrystalizer to a value
greater than 2.4 Hz. If there are frequencies below 2.4Hz.
The shell mode calculator will make the calculation for a cylindrical shell with panel ends in a few seconds
and requires only the input shown below. For the chrystallizer vessel this is the tangent-to-tangent length,
(as long as there is no intermediate radial restraint on the vessel), the diameter and the wall thickness.
Input for this vessel is shown below:

The results of this calculation is shown below:


Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

151

The lowest frequency is 28 Hz, materially higher than the highest frequency that will be excited by the 50
mph wind. The wind velocity that would excite the m=1, n=2 shell mode is:
V = fD/St = 28 Cy/Sec x 72 in. / 0.2 = 10080 in./sec. (1/12)/5280x3600 = 572 mph. The first shell mode of
this relatively short vessel will never be excited by wind.
A more accurate model including the heads and the support skirt can be constructed using FEPipe, and the
natural frequencies similarly calculated. This model is chrystal.ifu and is a child model for parent.ifu
delivered with the FEPipe. The natural frequencies from the chrystal analysis are shown below along with
the mode shapes. There are two lower modes that dont appear in the shell shapes shown above, and
those are the first cantilever supported bending mode of the vessel on the thinner skirt. These modes
appear at a lower frequency than the shell mode lowest frequency but are still below any modes that
would be excited by a 50mph wind speed. The next higher modes at 27 and 45 Hz correspond to the n=2
and n=3 modes from the cylindrical shell model above. (n is the number of circumferential lobes in a
particular mode shape, m is the number of lobes in the axial direction for a particular shape.)

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

152

A 50 mph wind will act on this vessel in a static way, there will be no vessel dynamic participation. The
displacements of the shell due to the 50mph wind can be analyzed.
It will be convervative to assume that the wind pressure will act uniformly over the length of the shell,
(although a variation could be provided).
The wind speed can be turned into a wind pressure on a flat, exposed surface by:
Wind Pressure (lb/sq.ft.) = 0.0027 x (mph)2
The wind pressure for entry to FEPipe would be 0.0027 x 502 / 144 = 0.04875 psi.
The input for FEPipe is shown below:

GENERAL parent.ifu form

make sure not still calculated modes and freqs.

The loads acting over the outside of the vessel appear as seen below:

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

153

There is no need to enter a cylinder shape factor of 0.7 since FEPipe mutiplies the wind magnitude by the
cosine of the wind direction vector and the exposed element side normal to get an appropriate distribution
of wind surface pressure on the outside shell of the vessel. The load case report shows that the occasional
load is included in case 4. The local membrane stresses in case 4 are very small and are shown in the left
figure below. When looking from the top, it does appear that the 50mph wind load will draw the sidewall
of the vessel in. For viewing purposes all deflected shape plots are scaled so that the magnitudes of the
displacements (which are generally too small to see), can be visualized. For the chrystalizer, wed like to
know how much the sidewall of the vessel moves in due to the wind load.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

154

The easiest way to find the lateral displacement in any displaced shape shown is to select the thermometer
tool
. This tool lets the user move the target of the thermometer to any location on the model. The
coordinate on the model is shown, along with the value being displayed (in this case the membrane stress
in the vessel due to the wind load), and the global displacements at the point under the cursor. For the plot
below, the displacement in the Z direction is 0.00047 in. This is much lower than the allowed tolerance on
the scrapers. The maximum displacement at the top of the tower in the X direction is 0.00558 in., still a
very small amount due to the 50mph wind.

The displacements over the entire vessel can also be plotted as shown below. The maximum X
displacement is 0.0006597. (The pointer tool location was selected by the user at a point of interest. The
point selected was close to the maximum lateral displacement, but clearly was not exactly at the point of
maximum lateral displacement.) The 50 mph wind load will not cause a blade interference with this vessel.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

155

5.17 Pressure Stiffening of Spherical Geometries


This example demonstrates the approach used when running a stress stiffened solution. Stress stiffened
solutions can be performed two ways. The first approach requires repeatedly making two runs each time a
new pressure loading or geometry needs to be evaluated:
1)Run a linear analysis with at least one case that will be used for stiffening
2)Run a second analysis and ask to use the stiffness case from the first linear analysis for stiffening.
When taking this approach, two steps are always required for an analysis. The first step establishes the
stress stiffness matrix, and the second step uses it.
The alternate approach should be used more carefully, may be quicker, and is described below:
1)Run a linear analysis with at least one case that will be used for stiffening
2)Run a second analysis and ask to use the stiffness case from the first linear analysis for stiffening. Ask
to save the stiffness matrix.
3)Run any other case that needs stiffening and reuse the stiffness matrix that was saved. The
stiffness matrix can only be reused if the non-stiffened loads change. If the stiffened load, the
geometry, or the material properties change then the stiffness matrix cannot be reused.
A sphere whose dimensions approximate a soccer ball will be used as an example. The first run will have a
200# weight load on the top of the ball with the bottom of the ball fixed. The displacements are cut in half
when the same load is applied but the geometry is pressure stiffened.

The input for the file is in soccer.ifu. The first three cases from the first analysis are shown below. Loads
and pressure were input.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

156

The weight case contains the -200# load on the top. The operating case has no load on the top.
Displacements from the three cases are shown below:

Weight Only (Y Load)

Wgt(Y Load) + Pressure

Pressure Only

The results of the elastic solution are shown below. For a linear elastic analysis the pressure and weight
cases can be run separately and then the results added. The change in displacement from the weight to the
pressure-and-weight case reflects the expansion of the sphere due to its pressurization, i.e. when
pressurized, the radius of the sphere in this problem gets larger by (P+W) (W) = (-4.295) (-4.5267) =
0.2317 inch.
Case
1
2
3

Condition
Weight Only -200#
Pressure and Weight
Pressure Only

Displacement at Top
(Y) (inch)
-4.5267
-4.295
+0.254

Step 2 Turn on stress stiffening and identify case 3 the pressure case as the load case result that
should be used to stiffen the geometry, and then rerun the job. FEPipe will bring in the first run load case 3
results and produce a stress stiffness solution using the case 3 (pressure only) loads. The displacements
from each of the weight load cases in the case 2 run will be reduced because pressure loads will resist the
downward compression of the sphere and when a pressure stiffened run is performed, the stiffness
matrix is modified to reflect the stiffening effect, and every load case analyzed in that run will be stiffened.

The results of the elastic solution and the stress stiffened solution are shown below.
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

157

Case

Condition

1
2
3

Weight Only -200#


Pressure and Weight
Pressure Only

Displacement at Top
(Y) (inch)
Elastic Solution
-4.5267
-4.295
+0.254

Displacement at Top
(Y) (inch)
Stiffened Solution
-2.069
-1.806
0.267

Note that the weight only case is influenced by the stiffening of the model due to pressure even though
pressure is not included in Case 1. This approach can be used to see how the stiffened pressure solution
affects the saddle/weight stress problem. The Zick analysis includes weight stresses due to liquid load that
are in some cases reduced due to pressure stiffening. The stiffening can be quantified by running a case
with pressure and weight loads first to get the stiffened case, then running the same model with weight an
liquid loads alone to see how the Zick type results (that do not include pressure for the discontinuity loads)
compare to the results from the Zick analysis.

6.0 Beam Load Displacement Relationships


#

Degree of
Freedom

j due to i

i,j due to i

Conditions

End (i) pinned ,


Rigid body rotation

End (i) pinned,


Rigid body rotation

Parameter of
Interest

rotation at j
j = i

displacement at j
i,j = (L)(i) = (L)(j)

i,j due to Fj

End (i) fixed,


Force at (j)

displacement at j
i,j = FL3/(3EI)

i,j due to Mj

End (i) fixed,


Moment at (j)

displacement at j
i,j = ML2/(2EI)

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

158

rotation at j
5

i,j due to Mj

i,j due to Fj

End (i) fixed,


Moment at (j)

End (i) fixed,


Force at j

[when the moment and


rotation at end j is
torsional, the equation
becomes:
i,j = 1.3ML/(EI)]

i,j = ML/(EI)

rotation at j
i,j = FL2/(2EI)

7.0 ASME Section VIII Division 2 Elastic Stress Rules and Discussion
This section contains applicable excerpts from the Code followed by further discussion or references that
apply to VIII-2 Part 5 analysis requirements. This section will introduce the updated elastic-plastic capability
in FEPipe and fold it into the elastic ASME Code design methods discussed below. Opinions are voiced
herein that may or may not apply to a particular situation and that reflect sometimes different positions
with respect to Code rules and guidelines that are subject to some interpretation or clarification.

Plastic analysis can be used for one part of the vessel and elastic analysis for another. If there are three
procedures for the same part and load, for example (see 5.2.1.1 below): a)linear elastic, b)lower bound
limit and c)plastic collapse, only one of the methods needs to be satisfied. Each other method may fail the
Code tests, but as long as one method satsifies the tests, the part and load is considered to satisfy Code
requirements.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

159

ASCE 7-10 states that factored load combinations for allowable stress design (ASD) must be used with
judgment. Notes on ASCE 7-10 and the plastic behavior required for compliance are provided in the next
section. Para. 5.1.3.2 states that where pressure exists, the effect of P=0 must be evaluated. The reason
for this is that some times pressure can reduce the stress due to other load cases. This occurs most often
with thermal loads, but can occur also with weight. Users concerned with this possibility are encouraged to
check the following boxes in the ASME preferences panel for the run.

The primary or sustained stress analysis is intended to protect against plastic collapse. When the limit load
or elastic-plastic analysis methods are used however, the need to establish service limits is also described
later in Part 5 and should be considered especially when external forces and moments are involved in the
load set. It is generally thought that when elastic approaches do not satisfy the Code requirements, that
the nonlinear methods of lower bound limits, or elastic-plastic analysis may still be used to qualify the
component. When using plastic analysis a number of crucial analytical requirements must be recognized:
1)For external loads on nozzles and an elastic-plastic analysis, service limits often form the basis for the
evaluation, and service limits in ASCE 7-10 and in VIII-2 are based on the judgment of the analyst. ASME
Section III provides a more stringent guideline for plastic load limits (twice elastic slope). See Section III
Appendix II.
2)For limit-load methods, artificially low limits can be produced since the load-deflection curve in many
nozzle load configurations shows an undulation that causes a premature interruption in the limit-load
analysis.
3)For elastic-plastic stress analysis the same premature stop of the solution may occur giving artificially low
allowable loads on nozzle connections. These conditions usually develop as the lambda value for the
nozzle-connection gets larger, where lambda = (d/D)(D/T)^0.5.
4)Factored seismic loads are susceptible to considerable judgment. ASCE 7-10 recognizes the need in a
seismic event for high ductility (plastic deformation) at connections. For welded connections, these high
loads may result in partial thru-wall cracks that are not detected immediately after the event. Users in
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

160

significant seismic areas are cautioned to evaluate seismic relative movement as well as inertial loads, and
to pay particular attention to overload factors (o) associated with an elastic load analysis. Detrimental
results may occcur when multiple vessels are connected by a single, large pipe run, and both vessels are
independently supported and may displace relative to one another. Since it is recognized that ASCE 7-10
gives reduced loads that can be used with elastic (ASD) or strength (LRFD) analyses, it is often informative
for the user to review the ground response spectra from ASCE 7-10 and the response spectra from actual
measured events in the vicinity. The seismic load processor provides a tool that allows user to load time
histories from site and to compare it with response spectra from ASCE 7. When the user believes that the
seismic analysis should not cause the system to undergo any plastic deformation, the analyst may wish to
use excitation that is both more conservative and closer to actual measured data. This might be particularly
true for those users that do not include SIFs in seismic or other primary analyses. Users can compare the
response spectra from time history shaking forces to response spectra from ASCE 7 in the seismic processor
in FEPipe and NozzlePRO. The steps for making this comparison are included in Section 8.1. In any event,
users should be sure that systems exposed to seismic events can accommodate seismic movements and
will undergo plastic behavior without suffering locally significant overstrain or brittle failures.
In Para. 5.2.2.4:

It is here recognized that generally strain controlled loads can result from thermal gradients and imposed
displacements. See the description of what to do with load-controlled and strain-controlled loads
described in the following paragraph. Assistance in assigning categories is found in Table 5.1.

Cases with strain-controlled loads represents Q type stresses (secondary) only, and will also be used to
evaluate the sum of primary and secondary stresses: Pl+Pb+Q so that simplified ratcheting (incremental)
straining is avoided. Load-controlled loads may produce Pm, PL, and Pl+Pb+Q loads.

Users should not attempt to use either strains or displacements from the limit-load analysis as an evaluator
of the collapse condition, or of the secondary characterstics of the component. If the twice elastic slope
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

161

method is applied using the elastic-perfectly plastic material model that lower bound limits will result in all
cases. (Essentially what were saying here is that 5.2.3 (b) displacements and strains can be used when
limited by a twice elastic slope analysis of the load stepping during the limit load finite element simulation.
There are several approaches when considering the twice elastic slope method. One is to apply the load
conditions and the Part 3 material model until the twice elastic slope pointis reached. Divide the twice
elastic slope load set by the safety factor (usually 2), and this becomes the allowable load set.
Another method is to take the material model for the limit load analysis and the factored loads from Table
5.4 and run the evaluation. If the factored loads are reached before the twice elastic slope is reached, then
the load set is permissible. This is reasonable, because in all cases, non-convergence would occur beyond
the twice elastic slope solution stoppage. The convenience is that the lower bound limit is not too far past
the twice elastic slope solution in most cases, and the twice elastic slope solution as described here
provides a)a realistic, displacement based limit easily understandable by the analyst, b)a means
(displacements) to evaluate serviceability, c)eliminates the need to deal with nonconvergence in a
nonlinear solution. (Displacements have meaning in most cases.)

It is interesting to compare the factored loads here to applications in other Codes. In the B31 codes for
example, and for single straight cantilevers, M/Z < (2/3)Sy represents approximately a 1.9 safety factor
against lower bound limit collapse. In ASME III, the safety factor for the same load case on a straight
cantilevered pipe has a 1.5 safety factor, the same we see in load case (1) of Table 5.4 above.

The concept of LRFD is used to set the acceptance criteria for this loading. The service criteria is also noted,
but per 5.2.3.4 (b) must be established using 5.2.4. An elastic-plastic analysis with small deflection theory
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

162

limited by a twice elastic slope analysis will certainly establish a lower bound. Arc methods for nonlinear
solution might eliminate some of the nonconservatism. Multiple methods are available in FEPipe to
address this concern. See further discussion regarding the elastic-plastic discussions.
From Para 5.2.3.5:

The yield stress in a lower bound limit analysis is taken as 1.5S. Actual material properties and multilinear
curve options should not be used.

Theoretically 5.2.4.1(b) is accurate, but in reality, care is needed for a variety of the aspects of the
compputation, including: a)the need for arc methods in the solution, b)the large strain model used, c)real
effects caused by longitudinal weld seams, etc. The user might note that acceptable designs using this
method may not have the safety margins between operating and pressure boundary failure that vessels
designed using rule based approaches enjoy, and so the user of the elastic-plastic collapse approach might
simply accept the lower safety factor in the operation of the facility. (This implies that the original
separation was not needed since the analysis is more accurate or that somehow the operation of the plant
is affected by the type of analysis performed. If the old, larger separation was needed in certain industries
due to local occasional overpressurization or combined loading events then the use of the newer
approaches would not be adviseable. )

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

163

In many cases, the service criteria governs the limit, and the twice elastic slope analysis will provide both a
reasonable service limit and satisfy the global criteria simultaneously. The question asked when running a
twice elastic load analysis is what limit to use? Since the twice elastic slope displacement will in all cases be
reached before collapse, the same load factors and material models can be used as givn in Table 5.5. As for
the lower bound limit load, the twice elastic slope approach, (or thrice elastic slope approach), offers the
same advantages for plastic collapse analysis as mentioned before for the lower bound limit load, i.e.
a)useful displacement solution, b)easier convergence, c)simultaneous service criteria satisfaction. The
twice elastic slope analysis describes a load level where the displacements as calculated by an elastic
analysis are off by two times and where the measurable strains in the part are within 1 to 2 percent,
(caused by the factored loads NOT the design loads. See Table 5.4.) This is because in a plastic analysis,
much greater strain and displacements may occur after the yield point has been reached.
Several examples are given below where the global criteria of plastic collapse is not sufficient.
1)The cantilever beam and externally loaded nozzles. (The service or local stress criteria governs.)
2)The pressure containment of a cylinder where the longitudinal seam limits the burst pressure even when
the weld joint efficiency is 1. Weld and parent material yields and strain limits are seldom perfectly
matched, and as a result real concentrations exist at those locations where total collapse (burst) is
concerned.
From Protection Against Collapse From Buckling 5.4.1.2 (c)

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

164

For most intersections and other PVP geometries imperfections are often not required since the
discontinuity itself provides the most significant imperfection when compared to out-of-roundness, etc.
Generally, load increment nonlinearities cannot be spotted when Newton Raphson methods are used and
so some time of arc-method is required so that very local snap thru effects do not cause convergence, or
other imperfections are not handled during the iteration. These local snap thru effects are noticed in
pressure loadings of large openings in elliptical heads, out-of-plane loads on nozzles in cylinders, and
opposing crushing point loads on a shallow open-ended cylindrical section.

For pressure design 5.3.1.1 seems reasonable, but when considering external loads and plastic collapse it
seems that the failure mode of 5.3 is viable and should be protected against. There are cases where
hydrostatic stress states and planar two-dimension stress states (biaxial stresses) can reduce the ability of
a local component to plastically strain due to the degree of hydrostatic stress present.

The limit prescribed L is the amount of allowed plastic straining for a given stress state once yield has been
exceeded. Note that even in the case where L goes to zero, the material may still approach yield (to within
the safety factor). The limit is intended to replicate the stress state where necking in the tensile test
occurs. In this stress state, microvoids begin to form and coalesce in the material and standard elasticplastic material models are no longer accurate.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

165

Note that the appropriate evaluation of ratcheting using an elastic-plastic solution is often very difficult to
perform for PVP geometries where stress states are often bi- or tri-axial. A specialized ratcheting routine is
being developed for this module in FE/Pipe and NozzlePRO. A standard Von-Mises flow model, even for
small rotation situations, is often not adequate for bi-axial ratcheting evaluations, and the constitutive
models of Chaboche, Ohno-Wang or others.

The elastic ratcheting requirements are the easiest to satisfy in that only Pl+Pb+Q < Sps must be satisfied.
For inelastic ratcheting behavior in a biaxial stress state see the appropriate version of the FEPipeNozzlePRO Plasticity Users Manual discussions.

The user must be very careful applying 5.5.2.2 in that most systems have not experienced successful
operation over their entire design life. Additionally, if a piece of equipment is designed for 3000 cycles, and
survives 3000 cycles, but has a 95% thru-wall crack at the end of 2999 cycles, it could be argued that this
does not comprise successful experience. Successful experience can only be thought of as experience for a
given time frame where unexpected cracking has not occurred at any point during that time. This implies
the following:
a)Inspection of an existing system at selected high stress locations must be performed before it is known
that cracks are not present.
b)The time of operation is the only duration for which new construction can be verified. For example:
Equipment A has survived 2000 cycles but was designed for 32000 cycles. Phased array crack detection
showed no cracking at 2000 cycles. One can only state that an identical piece of Equipment B is suited by
experience only for 2000 cycles. This is clear, if at 3000 cycles, equipment A shows the development of
cracking. It is suggested that this strict interpretation of experience be applied only in the cases where
there are: a)significant unknowns, b)new designs involved, c)high stress states, d)little test data available,
e)unusual sizes (small or large), f)exposure to damaging elements (corrosion/radiation), g)difficult to define
load conditions (seismic), h)critical applications (not 2 cold potable water lines.)
c)If experience is sited as a design basis, then monitoring of the sited existing system through its design life
becomes a mandatory part of the qualification procedure. Siting a plant that has operating successfully for
six years, but that designed for 40, as a qualified design, does not seem reasonable when the plant fails and
burns at any time in the remaining 34 years due to a fatigue design flaw.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

166

Para 5.5.3 Fatigue Assessment:

The designer must determine what is a significant operating load. Loads that are not significant do not
need to be analyzed.

The Poissons ratio goes to 0.5 when a differential element is fully plastic, and there is no further volume
change due to load. The correction in VIII-2 needs only be made when Ke does not multiply the entire
stress range. Ke>1 when the stress at a section exceeds the shakedown stress of 2Sy, and there is plastic
deformation at both ends of the cycle, (but generally over only a fairly small part of the total range.)

Cycle counting can be a complex process. When there are more than two stress cycles of interest and the
equivalent stresses can be added directly and still satisfy the Code allowable, then this conservative
approach can be taken. When the equivalent stresses from different cycle counts cannot be added directly
to satisfy the fatigue requirements then cycle counting as discussed in VIII-2 Annex 5-B must be performed.

The calculation in Annex 5-D is independent of the t/T ratio. When this ratio is less than 1 the user should
use additional caution. Additionally, only peak stresses are computed in 5-D. Note that this method can be
used in lieu of a more sophisticated evaluation. If 5-D peak stress evaluations are satisfied, the peak stress
analysis from the finite element evaluation can be ignored. Primary and secondary requirements must still
be evaluated however, and generally the finite element analysis is still used for this
Para. 5.5.3.2. Ke Penalty factors.
Once twice yield is exceeded by the membrane plus bending stress the geometry and loading may be such
that increased strain on further loading is concentrated in the area of the model thats undergoing the
original plastic strains. A simple example illustrates how this can happen. The load to state 1 in the
example is elastic. The load from state 1 to state 2 introduces plastic strains in the aluminum while the
steel remains elastic. From the diagram it can be seen that the strain in the aluminum increases more
rapidly after state 1 is reached. The rate of strain increase in the aluminum in state 1 is proportional to te
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

167

increase in the steel. After state 1 is passed, this is no longer true, and the estimate of the increase in strain
over the elastically predicted strain is the ratio Ke.

Strain,steel (1) = F1/(Esteel x Area)


Strain, aluminum(1) = F1/(Ealuminum x Area)
Strain,steel(2) = F2/(Esteel x Area)
Strain,aluminum(2) = F2/(Ealuminum x Area ) + pl

Ke is the actual strain divided by the elastically calculated strain. The situation is more complicated for the
finite element branch geometry shown in the image, but the approximation for the plastic strain from an
elastic analysis has been made for many years using a relatively simple calculation in the Code in Eq. 5.32
(given below).
The ability to perform an accurate elastic-plastic analysis permits the user to make a numerical estimate of
the Ke factor when the secondary stress exceeds 2Sy. (Below 2Sy, on unloading the material returns to a
compressive state so long as the surrounding material remains elastic, and on subsequent cycling the
material is fully elastic and there is no more plastic straining during the cycle.) This relaxation of the initial
plastic strain to purely elastic action (shakedown) is shown in the following figure.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

168

Ke can be due to bending stress distributions, cantilevered loading conditions, (See Rodabaugh and
Scavuzzo in WRC 432), notched geometries, and varying material conditions.

The calculation of Ke in Eq. 5.39 above requires an elastic-plastic analysis. Ke is a function of the magnitude
of the load, and for a cantilever, can be seen to increase to a maximum value and then drop as the load
continues to increase.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

169

The twice yield method is a way to verify an elastic fatigue analysis in a large stress zone where its not clear
how plasticity will contribute to fatigue life. This approach is only needed when 2Sy is exceeded by the
secondary stress range (Pl+Pb+Q).

Except in extraordinary cases, likely large seismic events it is thought that for most welded piping
components, with the exception of the non-integral connections sited in 5.5.6.4, that ratcheting is not a
real concern in most piping systems. Collapse of piping system under pressure in highly loaded tests, where
reverse plastic bending provides the varying load, and pressure provides the constant load, have
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

170

demonstrated radial ratcheting behavior, but tests at PRG where moments were double what B31
considers an allowed moment did not cause continued ratcheting in bends due to the secondary nature of
plastic deformations in piping components. (The first ten or so cycles of this test demonstrated the
incremental straining predicted by the Bree diagram, but subsequent cycles showed no increases in strain,
and fatigue life was not affected.) This suggests that actual ratcheting, as a failure mechanism, may appear
either: in highly loaded seismic events when the bending moments are large, or in highly loaded thru-wall
thermal events where the thermal gradients are not adequately controlled.

Specialized ratcheting material models should be developed when biaxial or triaxial stress states are to be
evaluated for ratcheting behavior so that the dynamic recovery of back stress is accomodated properly.
The specialty models of Chaboche or others should be used when incremental strains thru multiple cycles
must be quantified and when varying and nonvarying stresses are high.

It is likely that the most often problem with ratcheting in refinery, chemical and power piping systems and
pressure vessels occurs due to pressure and unconstrained dynamic loads due to earthquakes,
waterhammer or high wind events. In general, reversed dynamic loads have some component of strain
limited behavior but are not generally thought of as strain limited. Most ratcheting tests have been
conducted with externally applied bending moments that are not strain controlled, and unfortunately it is
this quality of the tests that makes quantifying the ratcheting displacements from the existing test difficu

From the 2013 Version of VIII-2

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

171

5.6(a)(2) suggests that a PL classification should be used for loads due to the restrained free end
displacements of the attached pipe. These displacements are generally caused by the uniform heating of
the attached pipe, and so 5.6(a)(2) suggests that what are generally considered strain limited loads should
be classified as a PL load and evaluated for primary load (collapse) failure. Table 5.6 in Part 5 seems to
contradict 5.6(a)(2) by showing that within the limits of reinforcement there is no PL classification.

This however, does seem to be consistent with previous versions (prior to 2006) of VIII-2 that gave
contradictory guidance. The following seem to apply to this condition:
1)Many nozzles have been designed in pressure vessels using WRC 107/297 and FEA considering that all
thermal stresses were strain limited and secondary.
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

172

2)When the attached pipe between the nozzle and the restrained free end is long and thermally expanding
in between the restrained end and the nozzle such that if plastic deformation occurs, the load is not
diminished thru an appreciable displacement of the pipe at the nozzle, then the thermal load has a primary
character.
3)If the attached pipe between the nozzle and the restrained free end is short and thermally expanding in
between the restrained end and the nozzle such that if plastic deformation occurs, the load is diminished,
then the thermal load has a secondary character.
4)Pipe flexibility analysis NEVER considers thermal loads as primary loads. Only in the case where there is
restraint liftoff of +Y restraints, is the added weight load due to the liftoff considered as part of the primary
load case. If the liftoff is small, then even this can be considered secondary, because a small amount of
plasticity causes the support to sit back down, and the primary load to diminish.
5)There has never been a pipe failure at a B31 intersection because of the assumption that external loads
applied to branch connections due to thermal expansion were secondary.
6)The argument has been made that the piping code thermal allowable is equivalent to the VIII-2 Part 5
elastic collapse allowable. This argument is outlined below:
VIII-2: Vessel Nozzles Inside Area of Reinforcement: PL(thermal) < 1.5S ~ Sy
PL = ~ (iM/Z)/2, when local bending is low since iM/Z is a peak stress.
If PL =( iM/Z )/2, then the iM/Z allowable is 1.25(Sc+Sh) = 1.2x1.25(2/3Sy + 2/3Sy) = 2Sy.
So setting PL(thermal) < Sy is the same as (iM/Z) < 2Sy

The suggestion is that PL = 0.5 x iM/Z. Both NozzlePRO and FEPipe produce coefficients for PL and
for i. If the relationship is valid the primary load index will be equal to the peak stress index divided
by 2 since they use the same nominal stress basis.
FSRF = 1.0 and weld leg = 0. (This is the EPRI 110996 Model) An example output for a 14x0.375
nozzle in a 48x0.5 cylinder. The peak and primary ratio is about 2 for both the axial and the
outplane coefficients for the assumption that the number of cycles is low (f=1.2)

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

173

There are a number of issues to discuss regarding this subject. One is that the sustained load stress
intensification factor in the B31 Codes is 0.75i for a number of components. Additionally, the error
associated with a number of SIFs for particular geometries is 2. The subject is not a simple one, but
it does seem unusual that one Code could be suggesting that thermal loads due to attached piping
should always satisfy primary allowables, while another Code (that is specifically written for piping),
states that thermal loads need only satisfy secondary allowbles.

Load Cases for Primary Loads

There has been disagreement regarding the exact interpretation of the load cases described in Table 5.3
and how they relate to the seismic and wind loads given in ASCE 7-10. The static equation that often
governs piping and external loads in ASCE 7-10 is Equation 13.3.1. Applicable equations and excerpts from
ASCE 7-10 for reference are shown below. The site coefficient g loads (Ss) are found in the mapped
diagrams in ASCE 7-10 Figs. 22-1, 22-2, etc. and are given as g magnitudes that will produce forces in the
structure that can be used for design in an elastic analysis. The critical concerns here for seismic loading is
the following:
1)Seismic source failures are not found in well-proportioned and displacement limited piping and vessel
systems because of plastic deformation and energy dissipation due to the plastic deformation that
inevitably occurs during a seismic event.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

174

2)Where small lines connect to large lines without flexibility there are failures in seismic events.
3)For large ducts, there can be failure during seismic events in the duct walls and in and around supports
where there is not an ability to undergo the relative anchor displacements.
4)Relative displacements of anchored locations, (especially for pipe or vessels that are supported away
from grade), often cause most of the damage during a seismic event.
5)Wherever there is a connection on a vessel or pipe that cannot undergo plastic deformation, there should
be some displacement limiting mechanism in the vicinity or the ductile failure of the part should be
designed for. This can occur with clips and lugs on vessels or pipe supports that will not tolerate loads that
may be two to four times higher than the seismic design loads.
6)Where the designer is concerned that the actual load should be considered, a rough estimate of this load
can be found by taking the design load and multiplying by a suitable over strength factor o, or Rp. For
most structures, vessels and pipe, over strength factors are provided in ASCE 7. The user will have to
decide a reasonable path forward when considering values of Rp = 12 from Table 13.6-1 for ASME B31
piping.
7)Since the 1960s it is well known that earthquake survival for most structures and non-structural
components requires an ability to behave plastically. Designers should account for this known and
expected behavior. In the late 80s and 90s additional seismic measurements emphasized this behavior
even more when actual ground motion was found to be significantly greater than anticipated values used
to develop design maps.
8)Where joints will not plastically deform, they should be designed to be significantly stronger than the
attached member so that the member itself, adjacent to the joint will plastically deform in a seismic event.
(This is particularly true for high seismic areas.) A typical design example is shown below:

Some design philosophies would suggest that the section A-A should be considerably stronger than section
B-B so that plastic deformation should occur in the structural shape. Structural components (especially
those in high seismic zones), should be designed either to be highly redundant, behave elastically during a
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

175

seismic event, or behave plastically at joints so that the anticipated plastic deformation will exist without
failure of the structure. In the image above, if the supporting cantilever is a calandria support and its failure
might jeopardize life or significant property, then either the A-A section should be strong enough to endure
( x Fp) loads without failing, (providing is sufficiently accurate for the geometry in question), or the AA section should be able to support an overload moment around twice as high as the section at B-B. The
definition of failing falls to to the designer in the selection of importance factors, risk coefficients, etc. In
most cases for refineries, chemical or power plants, it is assumed that the facility could go back on line after
the event. If this is the anticipation, then the designer should also be sure that design plastic deformation
will not result in crack development and growth that would reduce the design life below an acceptable
point. This is where the measurable first crack routine is useful. The over strength factor is based on the
difference between the point where the geometry deviates from elastic behavior and where it can no
longer experience an increase in load. The over strength factor is more related to the sustained stress index
than fatigue. A rough relationship can be hypothesized based on the geometry of the item. The
relationship is:
For bends and straight sections:
o x Fp  0.75i M/Z  1.1 x 2 x Sh.
For branch connections:
o x Fp  i0.5 M/Z  1.1 x 2 x Sh.
Where o x Fp is the loading. To the right of the loading is the expression used to determine the stress, and
on the far right hand side is the allowble to be used. The above expressions suggest the states of stress,
load and allowable where the strength criteria is fully satisfied.
The value of Fp from ASCE 7-10 is given below.

Note that large values of Rp will not reduce the seismic loading on any pipe segment or component to loads
less less than the limit given in 13.3-3.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

176

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

177

Factored Loads from ASME VIII-2 Table 5.3 for allowable stress (elastic) design (ASD):

Modern loading objectives intend to achieve efficient designs with consistent probabilities of failure. This
design approach is believed to be more economical and provide a more consistent, uniform approach to
the design of structures. It is possible that these approaches for the PVP industry are as uniform as
previously used safety factor methods, since probabilities for load and stress calculations in the PVP
industry are not thought similar to those in the building construction industry. When the design process is
removed from the judgment of the designer by multiple levels of probabilities developed using assumptions
regarding general structure survival rates that include many construction types not used in the PVP
industries the result does not seem likely to result in an improved or more efficient design. Resistance
factors are supposed to evaluate ability to predict the resistance of the structure to load, and the load
factors are supposed to reflect the probability that the loads will occur in certain combinations. It seems
unlikely that general probabilities could accurately reflect the effect of SIF errors or weld quality on fatigue
failures, or the rush by the project manager to meet a schedule, or the inability to get accurate loads, soil
properties, etc. It is certainly felt that in the theoretical environment, or in the building construction
environment, where gross assumptions are necessary so that the huge variation in designs can be included,
factored designs are reasonable, especially since the end connection and fabrication details (of ASCE 7
Chapter 14) provide the most appropriate degree of safety, but for the PVP industry, it is felt that little of
benefit was achieved by changing primary load requirements in the 2007 Version of VIII-2. ASCE 7-10
recognizes its own limitations however, and does place responsibility on the user to evaluate the efficacy of
the factored assumptions.
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

178

The blending of newer assumptions with older methods, (ASD), also introduces an opportunity for error. It
may be that the use of k=1.2 should still be used along with the factored loads from ASCE 7-10, i.e.
Pl < (1.5)(1.2)(S), where Pl is based on 0.7E. See ASCE 7-10 sections C.2.4.1, 12.4.3.3, and Sect. 13.1.7. In
the early 2000s when the majority of work on the 2007 version of VIII-2 was performed, it was assumed
that to properly implement the factored loads, the occasional load factors (1.3 for B31.3, 1.2 for VIII-2)
should be removed. It is now believed by some, (and bolstered by the sections sited above), that this may
not have been needed, and that the occasional load factors should be 1.3 for B31.3 and 1.2 for VIII-2 since
the intent in their being provided is that probability of failure could be a bit higher for these occasional
loads, i.e. that the safety factor for occasionally occurring primary loads did not need to be as high as the
safety factor for primary loads that act at all times. This can be observed by noting that 0.7E is used with
the full value of the always acting primary loads (D and Ps) in Table 5.3. The 0.7 in 0.7E however is to get
from the load that would produce section collapse to the load that would produce first yield at the out-fiber
since that is thought what was desired by the appropriate use Code.
Where users believe that they can use k=1.2 for the allowable occasional load the option does exist in
FEPipe and NozzlePRO to force the use of k=1.2 for post-07 versions of VIII-2.
It is important to note that since actual loads in the structure due to earthquake may be much higher than
the value of Fp used in the member design, that the fatigue analysis may be subject to this under
evaluation, and that fatigue cracking, (or more importantly crack initiation), can more easily start in the end
connection weldments. Where seismic loading is significant, the user is encouraged to use (Sm,Sb) in the
PRG crack first prediction tool to determine if measureable cracks will exist after a design earthquake has
been experienced.

8.0 ASCE 7-10 Notes


C1.1 The user of this standard, however, must exercise judgment when applying the requirements to
other structures. Earthquake loads contained herein are developed for structures that possess certain
qualities of ductility and post elastic energy dissipation capability.
C1.3.1 Chapter 13 of this standard specifies earthquake loads and deformations that must be considered
in the design of nonstructural components and systems designated in that chapter.
C2.4.1 The earthquake load effect is multiplied by 0.7 to align allowable stress design for earthquake
effects with the definition of E in Section 11.3 of this standard, which is based on strength principles.
C11.2The basis for the mapped values of the MCER ground motions in ASCE/SEI 7-10 is significantly different
from that of the mapped values of MCE ground motions in previous editions of ASCE/SEI 7. These
differences include use of (1) probabilistic ground motions that are based on uniform collapse risk, rather
than uniform hazard; (2) deterministic ground motions that are based on the 84th percentile (approximately
1.8 times median), rather than 1.5 times median response spectral acceleration for sites near active faults;
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

179

and (3) ground motion intensity that is based on maximum, rather than the average (geometrical mean),
response spectra acceleration in the horizontal plane.

ASCE 7-10 for Piping


C13.6.8.1 ASME Pressure Piping Systems. In Table 13-3, the increased Rp values listed for ASME B31compliant piping systems are intended to reflect the more rigorous design, construction, and quality
control requirements, as well as the intensified stresses associated with ASME B31 designs. Inreased Rp
values are paired with inceased values for ap that counter act the effect of the increased Rp.
For piping go to Chapter 13 Seismic Design Requirements for Nonstructural Components.
C13.1.3 Components with importance factors greater than 1.0 are expected to remain in place, sustain
limited damage, and when necessary, function after an earthquake. These components can be located in
structures that are not assigned to Risk Category IV. For example, fire sprinkler piping systems have an
importance factor, Ip, of 1.5 in all structures because these essential systems should function after an
earthquake. The component importance factor is intended to represent the greater of the life-safety
importance of the component and the hazard-exposure importance of the structure.
C13.2.4 Flexibility. In many cases, flexibility is more important than strength For example, even a braced
branch line of a piping system may displace, so it needs to be connected to other braced or rigid
components in a manner that accommodates the displacements without failure. (Fig. C13.2-1).
C13.3.1 Seismic Design Force. Beginning with the 2005 edition of ASCE 7, significant adjustments have
been made to tabulated Rp values for certain mechanical systems. For example, the value of Rp for
welded steel piping systems is increased from 3.5 to 9. The ap value increased from 1.0 to 2.5 so the net
change is negligible because Rp/ap changes from 3.5 to 3.6.
The component amplification factor (ap) represents the dynamic amplification of component responses as a
function of the fundamental periods of the structure (T) and the component (Tp)
Where the fundamental period of the component is less than 0.06s (16.6Hz) Dynamic amplification
occurs where the period of a nonstructural component closely matches that of any mode of the supporting
structure.
C13.3.2 Seismic Relative Displacements. The equations of this section are for piping systems
connected to one structure at multiple levels or to multiple structures. Two equations are given for each
situation. Equations 13.3-6 and 13.3-8 produce structural displacements as determined by elastic analysis,
unreduced by the structural response modification factor.
C13.3.2.1 For ductile components, such as steel piping fabricated with welded connections, the relative
seismic displacements between support points can be more significant than inertial forces. Ductile piping
can accommodate relative displacements by local yielding and strain accumulation below failure levels.
(See Fig. C13.3-4 Displacements between structures.)
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

180

C13.4.1 Design Force in the Attachment. Adjustments on the Rp value used for the anchorage calculation
have been eliminated, with the exception of the upper limit on Rp of 6, which is intended primarily to
address the anchorage of ductile piping systems that are assigned higher Rp values. These higher
component response modification factors reflect the inherent ductile and over strength of ductile piping
but may result in an under prediction of the forces on the anchorage.
C13.6.2 Component Period. rigid T <= 0.06s or flexible T > 0.06s.
C13.6.7 .. Ductwork systems that carry hazardous materials or that must remain operational during and
after an earthquake are assigned a value of Ip = 1.5, and they require a detailed engineering analysis
addressing leak-tightness.
C13.6.8 Piping Systems. Table 13-3 entries for piping previously listed the amplification factor related to
the response of piping systems as rigid (ap=1.0) and values for component response modification factors
lower than in the current table. However it was realized that most piping systems are flexible and that the
amplification factor values should reflect this fact; thus, ap was increased to 1.5 and the Rp values were
adjusted accordingly such that ap/Rp remains roughly consistent with previous versions.
13.1.7 .. The earthquake loads determined in accordance with Section 13.3.1 shall be multiplied by a factor
of 0.7. The allowable stress design load combinations of Section 2.4 need not be used. Relative
displacements per 13.3.2 must be accommodated.
Table 15.4-2 Values:
Vessels, elevated tanks, bins or hoppers R=3, o = 2, Cd=2.5
Horizontal saddle supported welded steel vessels R=3, o = 2, Cd=2.5
All steel and reinforced concrete distributed mass cantilever structures , including skirt-supported vertical
vessels, single pedestal or welded steel R=2 o = 2 Cd=2.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

181

8.1 Time history plots from USGS website

The seismic load development button

gives the user the capability of:

a)Finding design horizontal and vertical g-loads for static analysis from ASCE 7-05 data tables and stored
maps
b)Comparing the ASCE response spectrum to the response spectrum generated by an analysis of a time
history record.
Users can enter their own values for SDS and SD1 if they are materially different from the ASCE 7-05 values
that are used by default. Updates to the Chapter 22 Figures for ASCE 7-10 have not been completed at the
time of this writing.
The reason for comparing the response spectrum fom ASCE 7 to an actual ground motion event is to give
the user an idea of the difference that can exist between the design values and actual values. These

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

182

differences might be particularly important when there is a non-ductile component that is part of the piping
system.
The USGS web site contains a number of meaurement records from seismic events. Accessing and
comparing those records to the design records can be accomplished by following the steps below:
1. The link below contains earthquake information from 1986 to current. The data here is mostly
from the United States and its territories, but there is also information from recording locations
from a variety of additional locations.
http://nsmp.wr.usgs.gov/nsmn_eqdata.html
2. To download the data from the USGS website, click on the desired data set. For example, 1994
Northridge 6.7 magnitude earthquake in California.
3. Select either Corrected or uncorrected data, both should be similar and should give a reasonable
approximation of the earthquake. In this example, SMA-1 Recorders is selected.

4. Download the all records from all stations.

5. Unzip the file contents for all sensors. Select a sensor. Note that some sensors are ground based
and others are located at elevations in a particular building. It is this building data that was the
source of information for the 1+2z/h term. For this example, sensor 5106 was selected.
6. File 1731A.SMC was selected for sensor # 1731-1, at ground level. Copy this unzipped file into a
folder on the hard disk. Change the file extension from *.SMC to *.TXT. These are just text files
containing the seismic recorded data.
7. Open the text file in notepad. The text file must be slightly modified in some cases to be read by the
spectrum comparison program. The total number of entries must be in the first line in the file and
it must be a negative value. This file does not have the number of entries at the top, so it must be
added.
a. Find the first data point, usually under the | symbol.

b. There are 8 numbers in each line in this case. The first data line is in line #33 in notepad.
The last line of data is in line #740 in notepad. This gives 740-33+1 lines of data. This will
give 5664 data entries. But the last line only contains 7 values, so the total number of
entries is 5663.
c. Add this number with a negative sign at the start of the file:

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

183

8. There can be any amount of data at the beginning of the data file. A $$ symbol must be added to
the line following this data and just preceeding the data to be read by the program. For the data
set in this example, the added $$ is shown below.

The second value (sampling rate) of the real number set must be greater than zero and lower than
1.7e+38 (undefined) for it to be evenly spaced. Unevenly spaced files cannot be used. The sampling
rate must be 200 samples/second if the default value is used. For many non-US records the
sampling is even, but the rate is different from 200 samples/sec. In this case, the sample rate must
be placed on the first line next to the negative of the count of the samped values. For example, in
the figure above the value of -5663 indicates that an SMC data file will be used and that there will
be a $$ before the sample data points start. (See the $$ in the figure above also.) If the sampe rate
in the file was 150 sample points per second, the first line would appear -5663 150.

If there is not an even sampling rate of 200 samples per second the SMC file read will be
terminated.

9. Save the file and close. It should be ready to be used in the spectrum comparison program.

The seismic comparison capability is found by starting the seismic load processor:
is pressed the seismic data form should be opened as shown below.

. When this button

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

184

There are two capabilities most often accessed from this form:
1)Generation of horizontal and vertical seismic g-loads for a given location
2)Comparison of response spectra and static g loads using site specific time histories.
A comparison is always generated when a Generate Plots button is pressed. The comparison gives the
user some idea how the response spectrum generated from an actual time history of the ground
motion at the site differs from the horizontal g-load used in a seismic analysis.
Perform a comparison and design static g load for a seismic event at Kiholo Bay Hawaii.
In 2006 there was a seismic event that was recorded at the Wiamea station on Oahu.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

185

The modified SMC file for the Kiholo Bay seismic record is shown below. There are 200 samples per second
so theres no additional count next needed on the first line of the file. (Note that many of these modified
earthquake records from the USGS are already modified for reading and delivered with the software in the
spectrum\atime folder.

The latitude and longitude for Kiholo Bay are shown in the map below. These can be used to perform the
seismic design lookup.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

186

When the latitude and longitude are selected and the Look up values button is pressed the values to the
right are selected.

Click the Browse button to find the time history text file for the seismic event of interest. A number of
default files are stored under the installation folder in spectrum\Atime. The file of interest is shown here:

Press Generate Plots to produce the comparison.

The following should be shown for the comparison of the Kiholo Bay earthquake record and the Kiholo Bay
seismic design response spectra. This gives the engineer a sense of what the design might be exposed to
and what design values are actually used for the earthquake. In the case below, the green line is the static
design response acceleration magnitude used, the red line is the modal response acceleration magnitude
used. The blue ine is the response values from the 7-05 seismic maps. The light green line is the response
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

187

from the 2006 time history record. The time history record is turned into a response curve and the light
green curve results. It is interesting for this particular situation how the light green line is outside of the
dark blue line at almost every point. It is difficult to perceive of the dark blue line as a 2% probability of
recurrence every 50 years. o and Rp values separate design values from values that are closer to likely
existing in the structure. It is clear here that that in some cases these values can be large. It is important
for the user to know that when an earthquake occurs, the design loads being used are multiple times
smaller than earthqauakes that will likely exist, and that plastic deformation is being counted on to
minimize the damage and allow the structure to survive without failure.

The seismic report is shown below. This report illustrates how the ASCE method can get from site ground
motion g loads of 192% to design horizontal loads of 33% for an elastic analysis. As discussed in Section 8.0,
it is recognized that the magnitudes of the loads returned are intended to be used with an elastic analysis,
and that the R, Rp and o values give some guidance on the anticipated separation between design and
actual loadings.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

188

Since it is known that earthquakes do not generally cause collapse of piping systems since they are so
ductile, it is worthwhile to note that there may be significant plastic deformation occurring during a seismic
event and that even though a collapse failure does not occur, significant fatigue life may be lost.
Seismic Design Acceleration for B31 Piping
by Paulin Research Group - Houston, TX
8/14/2014 12:06:09 AM
Seismic design accelerations for ASME B31 piping are determined in accordance with
the provisions of ASCE 7-05, Chapter 13.
-------------------------------------------------------------User Input:
Seismic Design Code
= ASCE 7-05
Site Class
= B
0.2 second Acceleration
Ss = 192.00 %g
1.0 second Acceleration
S1 = 88.00 %g
Importance Factor
Ip = 1.25
Amplification Parameter
ap = 2.50
Response Modification Factor
Rp = 12.00
Average elevation ratio
z/h = 1.00
-------------------------------------------------------------Summary of Calculations:
Horizontal Acceleration = 33.60 %g
Vertical Acceleration
= +-17.92 %g
-------------------------------------------------------------Detailed Calculations:
0.2 Second Site Coefficient
Fa = 1.00
(Table 11.4-1)
1.0 Second Site Coefficient
Fv = 1.00
(Table 11.4-2)
0.2 Second Site Adjusted Acceleration
Sms = Fa*Ss
(11.4 - 1)
= 192.00 %g
1.0 Second Site Adjusted Acceleration
Sm1 = Fv*S1
(11.4 - 2)
= 88.00 %g
0.2 Second Design Acceleration
Sds = 2*Sms/3
(11.4-3)
= 128.00 %g
1.0 Second Design Acceleration
Sd1 = 2*Sm1/3
(11.4-4)
= 58.6667 %g
Basic Seismic Acceleration
A1 = (0.4*ap*Sds*Ip/Rp)*(1+2*(z/h))
= 40.00 %g

(13.3-1)

Maximum Required Acceleration


A2 = 1.6*Sds*Ip
(13.3 - 2)
= 256.00 %g
Minimum Required Acceleration
A3 = 0.3*Sds*Ip
(13.3 - 3)
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

189

= 48.00 %g

Per section 13.3, the horizontal design acceleration is given by the basic
acceleration (A1), but may not be less than A3, and need not be greater than A2.
Note that the values are adjusted for use with the allowable stress design procedures
of ASME codes per paragraph 13.1.7.
Horizontal Design Acceleration
= 33.60 %g
The vertical acceleration per section 13.3 is:
Av = (-0.2 * Sds)*0.7
= +-17.92 %g
-------------------------------------------------------------The following details are for the geographic seismic look-up using the USGS Maximum
Considered Earthquake gridded data files.
Desired Latitude = 19.74
Desired Longitude = -155.82
------------------------------------------Seismic data for specified coordinate location:
USGS Latitude Point
= 19.74
USGS Longitude Point = -155.82
0.2 Sec. Accel. (Ss) = 192 %g
1.0 sec. Accel. (S1) = 88 %g
0.2 Sec. Accel. (SDS) = 128 %g
1.0 sec. Accel. (SD1) = 58.67 %g
------------------------------------------Results within the 1.00 mile search radius:
(4 coordinate points searched)
Minimum Ss = 138.9296 %g
Maximum Ss = 192.00 %g
Average Ss = 187.9979 %g
Minimum S1 = 60.00 %g
Maximum S1 = 88.00 %g
Average S1 = 80.99483 %g
------------------------------------------The above results are generated from 2003 USGS data and based on a 2.0% probability
of exceedance in a 50 year period.

The advanced settings button allows the user to also look up seismic design values. The F lookup
option from the advanced settings is identical to the F lookup button from the time history sheet. Both
buttons result in SDS and design g load values being displayed.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

190

8.2 Comparison of Response Spectrum and Static Seismic Loads


ASCE 7-05 and ASCE 7-10 include Equation 13.3-1 for Seismic Demand loads on Nonstructural Components.
Table 13.6-1 lists ASME B31 Piping Components and provides values for ap, Rp and o.
The ASCE 7-10 values from Chapter 22 Figures for Seismic Ground Motion maps are based on risk-targeted
probabilistic ground motions and improved information regarding fault sizes, locations and possible
recurrence periods. The values obtained from the maps for 5% damping are Ss and S1, and these are called
the MCER accelerations, where MCER stands for Rick-Targeted Maximum Considered Earthquake.
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

191

Once the acceleration (g) value is obtained from the Chapter 22 Figures, it is adjusted for site specificity and
then multiplied by 2/3. The explanation for the 2/3 factor is given in C11.4.4:

This paragraph is simplified below as interpreted by PRG:


The ASCE 7-10 design approach changes in the Chapter 22 seismic load maps are intended to provide a
more uniform collapse risk that is a 1% probability of collapse in 50 years. This is partially based on a
probabilistic estimate of the lower bound margin against collapse of 1.5 inherent in structures designed in
the standard. The uncertainty in this margin is accounted for with a collapse fragility. The MCER in ASCE 710 has been adjusted up by 1.5 so that it can be used with the same equations from previous versions that
will reduce it by 1.5 (2/3).
After these adjustments the user obtains SDS and SD1, the response spectrum design accelerations which are
then further adjusted by Ie, Ip, ap, Rp, Re, or o terms. For comparison, the time history record selected
by the user can be transformed into a response spectra. That spectra is plotted for comparison without
alteration.
Additionally, the design response spectrum using the SDS and SD1 values will also be added to the plot. This
design response spectrum is shown below.

The design g-load will then be computed from Chapter 13 Equation 13.3-1 using entered values for ap, Rp
and Ip. Equation 13.3-1 is essentially:
Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

192

Fp = (0.4)(ap) (SDS) (Weight) (Ip) (1+2z/h) / Rp

13.3-1

For piping systems ap = 2.5 and so 0.4 x 2.5 = 1. The SDS value is now multiplied by an instructure multiplier
(1+2z/h), and the Importance factor, and then divided by Rp which is the response modification factor. For
ASME B31 piping, ASCE 7-10 gives values for Rp as high as 12. B31E limits Rp to 3.5. Ip is a value of between
1.0 and 1.5, where for critical systems,Ip=1.5.
In paragraph C12.9.2, the ASCE position regarding modal response analysis is discussed:

The spectrum values are multiplied by the importance factor Ie and divided by R, (or Rp) to account for
inelastic behavior. To compare the static load, the ASCE response spectra and the response spectra from a
ground motion time history, the user should decide if the comparison should include Ie and R. The relations
presented in this section are summarized in the following image:

Each path can be used to produce a spectral value SDS. The top path is the method in ASCE 7. The body
path is used to take the time waveform of the ground motion. ASCE 7 gives a method where probabilities
are calculated, with multiple ground motions, soil models, etc. and spectral values produced. The bottom
line takes a known time history from the site and creates a response spectrum from that time history. The
desire is to compare the resulting response spectra. The spectrum at point f which is simply a response
spectrum from the time waveform, will be compared to the design response spectrum from ASCE 7 using
SDS from the ASCE 7 approach, and the response spectrum from ASCE 7 times 3/2 (where Fa = 1).

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

193

From ASCE 7 Eq. 13.3-1: Fp = (0.4)(ap) (SDS) (Weight) (Ip) (1+2z/h) / Rp, it can be seen that Rp has a significant
impact in reducing both the spectral and the static seismic demand.
ASCE 7 had to make an assumption regarding the behavior of the seismic system when undergoing plastic
response. ASCE 7 had to assume that all ASME B31 Piping Systems behave inelastically to a great degree
because of the magnitude of seismic ground motions. (It is not practical to design systems to remain elastic
during speculated earthquake events.) It is the job of the ASME B31 piping and vessel engineer to be sure
that these assumptions are valid, or to otherwise accommodate the system where this is not the case, for
example, with glass lined pipe, refractory lined systems, brittle couplings, etc.
The static and modal dynamic approaches are equivalent providing the ground accelerations are applied
uniformly to the static structure.

9.0 ASME VIII-2 Part 5 Plastic Analysis


There are two types of elastic-plastic analysis, and where serviceability limits govern there are really three
because using either the limit load analysis of Table 5.4, or the Elastic-Plastic analysis of Table 5.5, may not
provide meaningful evaluations of the plastic load carrying capacity. (In this case a twice elastic slope
analysis can be useful and is well documented.) The load multipliers for limit and elastic-plastic loads are
shown below in ASME VIII-2 Tables 5.4 and 5.5. These are factored loads that should used in the analysis
as described in the table.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

194

Table of recommended loads for performing elastic-plastic analysis of external loads and pressure on vessel
nozzles.
Elastic-Plastic Recommended Loads and Approaches for FEPipe and NozzlePRO
Users employing these methods are encouraged to also perform the typical elastic analysis to verify the
increased load permitted using the plastic approach. Users should also be sure that service limits are
satisfied and should make sure that some judgment is exercised when establishing service limits for a
particular situation.
The comprehensive plastic-large rotation solver for FEPipe and NozzlePRO will be released later in 2014.

Paulin Research Group 8.4.2014 FE/Pipe V.7.0 NozzlePRO V.9.0

195

Вам также может понравиться