You are on page 1of 227

F E A T O O L S ™ U S E R M A N U A L

FEATools™ v3.2

PAULIN RESEARCH GROUP

FEATools™ User Manual

© Paulin Research Group – 2018 All Rights Reserved


F E A T O O L S ™ U S E R M A N U A L

FEATools™ v3.2* User Manual

 2018 – Paulin Research Group


11211 Richmond Avenue • Suite 109
Houston, TX 77082
Phone +1 281.920.9775 • Fax +1 281-920-9739

www.paulin.com

* - FEATools™ is a collection of individual FEA-based programs developed by Paulin Research Group


For the purposes of tracking and support, the version number used for FEATools™
Is the version number of the CAESAR II FEA Translator program incorporated into FEATools™

© Paulin Research Group – 2018 All Rights Reserved


F E A T O O L S ™ U S E R M A N U A L

Table of Contents

Foreword, Copyright Notices and Version 3.2 New Features................................... 3


Version 3.2 New Features ......................................................................................... 4
General Guidelines:……… ...................................................................................... 12
Components of FEATools™ Version 3.2 Translator: .............................................. 14
Using FEATools™ Version 3.2: Key Points ............................................................ 23
Starting the Software (FEATools™ Version 3.2) ................................... 23
Editing Branch Connection Data ............................................................ 24
Running CAESAR II after Converting the model .................................... 40
Automatic Bends with Staunchions ........................................................ 41
Checking Pipe Shoes ............................................................................. 46
Entering Vessel Nozzles ........................................................................ 53
Enter a Rectangular Header Box ........................................................... 60
Comparing Results ................................................................................ 68
Laser Scanning and Point Cloud Manipulation: ...................................... 77
Chpater 1 Introduction to FEATools™ .................................................................... 82
Why should you use FEATools™? ........................................................ 98
Typical questions addressed by FEATools™ ........................................ 99
Chapter 2 The CAESAR II FEA Translator ............................................................ 101
Starting & Running the CAESAR II FEA Translator – ........................... 102
Quick Start ............................................................................................ 102
Detailed Input Documentation .............................................................. 107
Detailed Output Documentation ............................................................ 112
File Storage ........................................................................................... 120
Chapter 3 Theory and Discussion of the CAESAR II FEA Translator..................... 122
Recommendations for CAESAR II Version 6.10 Users......................... 132
Recommendations for PCLGold ........................................................... 134
Detailed Discussion of the CAESAR II FEA Translator Input Options... 145
ASME Section III Guidance for Sustained Stress Indices ..................... 157
Chapter 4 CAESAR II FEA Translator Discussion Examples ................................. 160
Chapter 5 CAESAR II FEA Translator Example Model Files & Discussion ............ 180

1
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

Brief Descriptions of the Example Model Files ..................................... 181


Detailed Coverage of the Example Model Files .................................... 183
Example Model File #1 – WRC 329 Example ...................................... 183
Example Model File #2 – S303 Example .............................................. 185
Example Model File #3 – bypass B13 Example ................................... 187
Example Model File #4 – Small bore pipeline example No. 1 .............. 189
Example Model File #5 – Small bore pipeline example No. 2 .............. 192
Example Model File #6 – Manifold piping example .............................. 194
Example Model File #7 – “Jim Wilcox” Complex Load Case Example . 196
Example Model File #8 – Pump Discharge Manifold Example ............. 202
Example Model File #9 – WRC 329 “Schneider” Out-of-plane Moment
Example ................................................................................................ 207
Example Model File #10 – Spring Hanger Design Example ................ 209
Example Model File #11 –API661 Square Header Box Example ........ 211
Example Model File #12 –Piping Technology and Products Shoes
Example ................................................................................................ 214
Example Model File #13 –Hillside Nozzles in Vessel ........................... 216
Example Model File #14 –Bend Trunnion Example ............................. 219
Example Model File #15 –Simple Intersection Example ...................... 221

2
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

Foreword, Copyright Notices and Version 3.2 New Features


Typical Automated Finite Element Models Available in FEAToolsV3:

Rectangular Header Boxes and Heaters Flat Heads

Pipe Shoes w/ Rings Head Openings Hillside Nozzles

Nozzles on Cones Steel on Bends Trunions on Bends with Pads

Trunion Supports and Available Finite Element Head Types

3
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

Version 3.2 New Features

Full implementation of B31J SIFs,


ASME B31J Flexibilities and SSIs for Branch
Connections
WFI & Bonney Forge Weldolets
Integrally Reinforced Weld On
Straight & Taper Bore Olet Bores
Fittings
Atypical Deficient Bevel Olets
Pad widths
Weld Thicknesses
Refractory
Hillside offset
Branch Connections Include: Stiffening Rings
Trunion Geometries (no hole in run)
Control of attached lengths
Heat exchanger boundary
conditions
Welding Tees Per EN10253 A & B
Spheres
Elliptical heads
Nozzles in
Dished heads
Conical heads
Flat heads
All Vessel Nozzles Automatically
Inserted into CAESAR II Model
Consideration of BEND ANGLE
Attached Straight Length
Bends Refractory
Stiffening Rings
Varying Circumferential Thickness
Round
With or without repads
Trunions on Bends
I-Beams, Channels, Angles, WF
Shapes, Crosses or Flat Plate
SIFs, Flexibilities and Allowable
API 661 Header Box <or>
Loads
Rectangular Vessel Nozzle
Obloswage Nozzle Shapes

4
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

Consideration of Pass Partition


Plate
Tubesheet simulation
Clamps, Guides, and Limit Stops
Flexibilities and Allowable Loads
PTP Pipe Shoe Library
Checks CAESAR II Restraint Loads
Allowables include SHOE and PIPE
Input and Edit Point Clouds
Cloud Comparisons to CAESAR II
Laser Scanning
Cloud Stand Alone Piping Eval
Tools
Individual Intersections can be
Modeled using FEA.
Various universal stiffness overrides
are provided to determine which
flexibilities control the system
behavior.
CAESAR II can be run from
FEATools
Review multiple C2 Ouput at same
time
Administrative Features Generate Difference Tables and
Plots
2 and 3 Dimensional Graphs
NozzlePRO Access to Vessel
Models
NozzlePRO Access to Branch
Models
Compare Multiple CAESAR II
Results
B31J, More Realistic, and
Conservative SSI Options.

V3.2 of FEATools introduced a variety of new capabilities and improved


workflow features when paired with CAESAR II Version 8.0 or later.

A more detailed summary of the V3 New Features is provided below:

5
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

ASME B31J more applicable data for B31 piping was released in
February of 2017. SIFs, Flexibility Factors, modeling methods, and
sustained stress guidance for branch connections are included in Table 1-
1 of B31J. Low-load test procedures for flexibility factors are provided in
B31J along with twice elastic slope (TES) methods to find sustained load
stress factors. The SIFs, SSI guidance, and k-factors are for in-plane,
out-of-plane and torsional degrees of freedom. Pressure i-factors also
implied in the SE equations are not provided in B31J but are avialable
from FEAToolsV3 when finite element analyses are performed. The
pressure SIF is available for B31 branch connections, all vessel nozzles,
and for API 661 header boxes – again thru FEATools V3.

Integrally Reinforced Welded On Fittings (IRWOFs).


The initial SIF error for d/D ratios between 0.5 and 1 was initially found by
R.W.Schneider who worked for Bonney Forge at the time and was
investigated further by E.C.Rodabaugh whose results were published in
WRC 329. (The Markl fatigue test machine from WFI in Houston run by
Kahn in Oklahoma to produce WRC 329 results is now in the PRG lab in
Houston Texas. (The machine was shipped and installed at the University
of Oklahoma, where Kahn conducted the additional fatigue tests.) As part
of that effort, Kahn found that some integrally reinforced weld-on fittings
had higher SIFs than unreinforced fabricated tees! When conducting
research for ST-LLC 07-02/STP-PT-073/B31J PRG found welding tees,
contoured inserts, and integrally reinforced welded on fittings that did not
appear to follow any known pressure design guidance for either area
replacement or Bildy pressure design. It appeared that these IRWOFs
were prepared to have the welded-on appearance of current designs, but
did not provide any of the welded area replacement. PRG fabricated two
of these fittings in a 3x4 size and repeatedly pressure tested the fitting.
Surprisingly the smaller fitting repeatedly passed the MSS-SP97 burst
test requirement. It was difficult, however, to get experienced welders to
fit-up and weld the olets as requested. Without strict control experienced
welders (at K&H Fabrication in Smithville Texas that prepares many of
the PRG fatigue and collapse specimens), would apply more weld where
they knew it should be placed, in spite of the weld contour and bevel of
the fitting. In one case, the welders burned thru the bottom edge of the
olet in the circumferential plane to create the contour that was expected.

6
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

A greater concern exists for large IRWOFs where it will not be as easy,
(or as economic), for the welders to deposit large areas of additional
weld. To address these possibilities for users, FEATools V3 has a finite
element simulation of these “atypical” olets. Users may select the
atypical olet, Bonney Forge tapered olets, WFI straight thru olets, or
copies of the Bonney or WFI geometries. Where it is not known if
sourced olet suppliers provided appropriately contoured olets for welding,
the user should select “atypical” olets to get what is believed to be
appropriatly large SIFs for bigger fittings where intended weld sizes are
not provided. These olets can be recognized by the fact that they cannot
be welded in accordance with B31.3 Fig. 328.5.4F. (See the figure
below.)

Branch Connection Additional Data. For typical B31 branch


connections FEATools V3 allows the user to provide information to the
finite element model that was not part of the test data set correlation. For
thin, stainless steel piping systems, the fillet weld size can often be equal
to or greater than the thickness of the pipe, and in this case can reduce
both the SIF, k-factor, and SSI for the branch connection. FEAToolsV3
permits the user to enter the weld thickness for pad or unreinforced
fabricated tees. Hillside branch connections, laterals, and trunions
(where the hold is not removed from the run pipe can also be simulated,
and the SIFs, k-factors, and SSIs put directly into the CAESAR II model.
SIFs from the finite element models also include i-factors for axial and
pressure load components, (not available in B31J). FEATools V3 branch
connection models also support the influence of internal refractory on the
SIF, and k-factors.

7
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

Vessel and Exchanger Models: Pressure vessel and


exchanger nozzles in the piping system can be directly modeled from
FEATools. The FEAToolsV3 modeler constructs NozzlePRO models for
nozzles in cylinders, spheres, elliptical, dished, conical, and flat pressure
vessel heads. Flexibilities, SIFs, and Allowable loads from these models
can be automatically used in the CAESAR II postprocessing of the results
in FEAToolsV3. The user can also use FEPIpe to alter the resulting
model and use the allowable load from the user-tailorable finite element
model back in the CAESAR II calculation. (FEPipe, NozzlePRO,
FEBend, FE661, etc., all use a standard internal file architecture to pass,
SIF, k-factor, allowable load, and superelement data between software
applications. Boundary conditions present in heat exchanger models can
be easily simulated from FEAToolsV3 so that increased strength but
reduced flexibility found in these nozzles can be properly included in the
CAESAR II model.

8
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

Bends: Bend geometries are oversimplifed in the current Code SIF and
k-factor equations and in the B31J equations. ASME Section III has had
updated bend expressions via the N319 code case since 1983. The
N319 code case most significant contribution is to include the effect of the
bend angle on the SIF and k-factor. This effect can be significant when
it’s recognized that as the bend angle gets smaller the behavior of the
bend approaches that of straight pipe the SIF and k-factor both approach
1. This is particularly significant for large d/t pipe where the k-factor and i-
factor are both large and constant in the current code, unchanging as a
function of the bend angle:
Bend i-factor = 0.9 (r/t)2/3 (r/R)2/3
B31J did not include the N319 approach since it did not include the
influence of attached flanges, (or varying thicknesses). The flanged end
correction Table D300 in B31.1 and B31.3 Chart B was developed only
for 90 degree bends, and so it was felt that before the N319 approach (or
something like it) could be included in B31J that the flanged end problem
for varying angles would have to be addressed.
FEATools V3 addresses this problem by using very high order ovalization
and warping curved finite element solutions to include the effect of both
angle, attached straight length and varying thicknesses – both
circumferentially and in the bend, (i.e. the bend has one thickness, while
the attached straight pipe has another thickness. Current Code
approaches don’t address this problem either, where erroneous results
can be produced when the high stress moves from the parent metal
sidewall of the bend to the girth butt welded end of the bend. FEATools
finite element approach addresses this problem along with the angle and
attached straight length problems.
FEATools also recognizes when supports are attached to bends and
automatically generates finite element models for bends with trunions and

9
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

the model attached straight pipe. All shell finite element results are
stored in libraries so that repeated runs are not required. (The bend with
trunion finite element solution only has to be produced one time.) Where
structural steel is attached to the bend in place of the trunion, the user
can specify the cross sectional shape of the steel and FEAToolsV3 will
generate the SIF and k-factors for this bend configuration. (SIFs can
often be greater for irregular steel shapes than for uniformly curved
intersection shapes like trunions.)
The bend can also be lined with refractory. Reduced stiffnesses and k-
factors are automatically computed and applied in the CAESAR II model
due to the significant reduction in ovalization due to the refractory.

Piping Technology and Products (PTP) Pipe Shoe


Library. Depending on the combination of the loads and the thickness
and temperature of the pipe, the high stress at a pipe shoe can be either
in the shoe, or in the pipe. The Piping Technology and Products Library
is designed so that the load is limited by the weakest of the components
and the determination is made automatically for all pipe shoes and all
load cases in the model. Stiffnesses and flexibilities are also computed
for the shoe/pipe combination and printed. Where torsional loads must
be constrained, the analysis often shows that stiffening rings are required
to addequately strengthen the cross section. Supports with end plates,
clamps and integral or nonintegral repads can be generated.
When needed, the user can take a copy of the PTP library and enter
reduced allowables for the pipe so that the effect of temperature can be
considered in the allowable load calculation.
Using NozzlePRO saddles of almost any geometric proportion can be
modeled, and the allowable loads and stiffnesses for the model also
generated.
Laser Scanning: Stress analysis of an existing facility should start
with laser scans of operating and if possible initial conditions of the piping
and vessels. The scan should include all steel in the vicinity used for
support of the pipe or that might possibly used for future supporting.
Vessels, heat exchangers and pumps should be scanned along with any
major connecting lines to critical equipment in the area. Well made,
registered point clouds give the analyst the opportunity to take any
measurement to within a few mm, to cut sections at any place in the
plant, and thru any vessel, pipe, or piece of equipment. Software

10
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

evaluation of collected sections of the point cloud can be used to estimate


ovalization, peaking, squareness, liftoff, initial movement, curvature and
moments in straight pipe subject to bending. The locations of supports on
bends or at any point on the pipe can be verified, along with spring
positions, and most gaps that are greater than about 2mm.
Laser scanned operating positions of the piping system can be aligned
with the CAESAR II model of the piping system and differences noted and
corrections made to the model.
Each of these functions is available in the FEAToolsV3 graphic
processor.

Administrative Tools: Pipe stress analysis involves first building


the model, and then second – evaluating the results of the calculation and
changes made to the model. The administrative comparitive tools in
FEATools V3 are used to make it easier for users to compare the results
of earlier to recent runs. Multiple runs can be compared at the same time
and graphs, tables and charts of the differences between runs can be
generated. Multiple images and plots can be shown on single or multiple
graphical windows so that different displacements, forces and moments,
or stresses can be compared graphically, tabularly, or geometrically at the
same time on multiple monitors.
Universal overrides are also provided for the varieties of stiffnesses
available in FEAToolsV3 to determine which set has the most significant
effect on system behavior. For example, if pump loads are very sensitive
to bend k-factors, then recognizing that bend thicknesses are often not
well controlled should be a part of the evaluation, and possibly
purchasing/scheduling options. “Every aspect of the project must be
setup to succeed.”

11
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

General Guidelines:

The FEAToolsV3 user can interact with the piping system model in ways not
before possible. Technologies can be called on to more accurately represent
bends, (current code approaches do not consider the bend angle in the k-factor
calculation even though this calculation has been in the Section III Code for
several decades), bends with staunchions or steel supports, vessel connections,
rectangular header boxes, and pipe supports (considering the strength of the
shoe and the pipe as a unit).

In an economic environment where stress analysis of piping systems must be


performed faster and more reliably, FEAToolsV3 permits multiple outputs to be
viewed at the same time and comparison tables generated automatically so that
users can see quickly which model is best. Inclusion of the latest flexibilities and
more accurate stress factors permits for a rapid reevaluation of the system so
that the effect of newer stress factors and flexibilities can be used to
demonstrate that a system satisfies the code requirements.

The screens and figures below provide a quick introduction to new FEAToolsV3
features. Help is included with each and examples are included thru the text
that should help users with design and input guidelines.

Users should note that B31J/STP-PT-073, WRC 329, and EPRI approaches all
include parameter ranges that are shown when the PRGiK calculation screen is
used. Each correlation method has a different set of parameter limits. For
bends, the limits are associated with the parameter h and D/T. The most
constrictive limit for most correlations remains D/T = 100. (h = TR/r2).

FEA and HyperDOF elements do not have high end D/T limits and tend to be
conservative when the D/T is in excess of 100 and below 8. In the very high D/T
ranges fabrication errors can produces differences in test results that are of the
same order as analysis and boundary condition errors. It is believed that large
D/T pipe and vessel constructions are more sensitive to low cycle fatigue issues
than either pressure capacity or high cycle fatigue issues.

All D/T ranges are senstive to high cycle fatigue issues, and so fabrication
standards generally address these concerns.

For low cycle D/T problems, stresses are high (above twice yield), and cycle, and
generally there is some manual, multi-pass welding involved that increases the
error scatter.

12
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

For most external load problems in piping and pressure vessel design in the
higher D/T ranges:

1)The finite element method provides conservative results. (This is any


properly used finite element analysis using any commercially available
software.)
2)Analysis should be used with increased inspection where services are
critical.
3)A preferred vendor list should be available for material, fabricated
products, fabrication/construction, and inspection, where good fabrication
practices can assure the designer that predicted results are conservative and
that the deviation from the mean will be small.

(Where any normal or severe cyclic service is involved, repairs, (even during
initial fabrication), should be avoided. Good inspection practices go hand-in-
hand with good fabrication practices. The objective when preparing the User’s
Design Specification (VIII-2 2.2.2) is to provide any “additional requirements
[that] are appropriate for the intended vessel service”.

13
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

Components of FEATools Version 3.2 Translator:

FEATools Version 3.2 lets the user:

→ Change the properties of any standard B31-type piping branch


connection from FEAToolsV3. (This includes increasing weld sizes, and/or
specifying attached pipe lengths, pad width sizes, olet manufacturing
approach, welding tee likely body sizes, and branch-trunions for support.)
→ Specify FEA models for nozzles in vessels or heat exchangers,
including cylindrical shells, heads, cones, and flat plates.
→ Evaluate Bend Details not in Code Equations, including the bend
angle, component thicknesses, attached pipe or steel, refractory and
external rings.
→ Include FEA models of rectangular header boxes (API 661)
→ Evaluate all pipe shoe/saddle type supports for all load cases for all
nodes
→Postprocess multiple CAESAR II Outputs and Point Clouds at the same
time graphically, tabularly, or using 2 and 3D Charts.

Tabs describing each of these functions are listed along the top of the
translation panel. (See the figure below.)

14
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

When the user has specified any additional information for a component
in the CAESAR II model, FEATools shows the item with a symbolic gold
representation of the geometry. In the image below in gold we see radial
and hillside pressure vessel nozzles, bend with staunchions, and bends
with small angles. Updated k-factors, SIFs and SSIs are provided for each
item in gold.

15
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

Each “advanced model node” can also be uniquely labeled:

In the image below trunions, small angled bends, rectangular header


boxes, and hillside nozzles in heads are shown. The trunions are
essentially intersection geometries without the hole in the run pipe
removed.

16
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

The model below shows a perfect application for small angled bends. K-
factors for the bends below will be very close to 1 since the angle is so
small. K-factors from B31.3 will be for 180 deg. bends since this is the
fixed default for B31.3.

Because of the usual large size of point clouds editing is done on a


reduced resolution image, and then once editing is completed, the full
17
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

resolution is returned for comparisons and measurements. A low


resolution clipped staunchion support is shown below along with the
CAESAR II model it will be compared with.

A small 2” diameter piping system in the PRG lab was scanned by Mr.
Joey LaFranca of Leica Geosystems in Houston. The scanned model is on
the left and the CAESAR II model is on the right in the FEAToolsV3 image
below.

18
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

Common points are selected between the two models, (point cloud on
the left).(Note the clamp and calipers at the base of the point cloud
model used to check the precision of the scanning.)

Points can be interactively weighted by the user to make sure that


directional accuracy is maintained. (Users are encouraged to watch the
30 min. training video.) Since model lengths are often entered in error,
reweighting and point selection helps the user identify all model errors in
one pass of review.

For the model shown above, the comparison is poor when vessel head
flexibilities are not included in the model. When FEAToolsV3 isused and
the cone and spherical head flexibilities are used, the model is much
more accurate.

19
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

The model without flexibilities is on the top, and with flexibilities is on the
bottom. The largest error drops from 50mm to 5mm when flexibilities are
included in the model. Calculated moments at the boundary condition
ends drop by equal magnitudes.

20
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

Measurements can be made from the point cloud to verify measurements and
calculations made in the field.

21
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

After cutting sections thru the pipe, the ellipse calculation tools help the
user determine the accuracy of ovalization measurements at elbow and
support locations.

The phrase “the Leica Laser sees all” is often heard after Leica laser scans
are reviewed. Small bore pipe interference, supports lifting off, supports
not properly replaced after maintenance, and field modifications not
reflected on any drawings are often noticed in the laser scan images.

The registered point cloud permits the software user to section and orient views of the
system and take measurements from location to location that are virtually impossible to
take in the field (because something is in the way). There is never anything in the way
of a registered laser scanned point cloud.

22
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

When working with an operating piping system, the elevation of a vessel nozzle should
never again be calculated alone. When large diameter pipe and vessels interact high up
on a column, the actual vertical displacement of the vessel nozzle can have a significant
effect on the stress and support condition in the attached large line. In many cases,
estimates of vessel nozzle displacements are excessive, resulting in stress predictions
that are too high. Using accurate displacements from laser scans can show that
operating conditions easily satisfy Code limits.

Using FEATools Version 3.2: Key Points

Starting the Software (FEATools Version 3.2)

The yin-yang icon from the desktop or from the CAESAR II ribbon starts V3 of FEATools:

The Version 3 menu appears below:

The final build number and installer shown in the bottom left hand corner will likely be
later than the versions shown.

23
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

The user must have a valid license for NozzlePRO to start, and a valid FEAToolsV3 license
to run finite element or related applications. A CAESAR II license is required to access
the basic translator features and the CAESAR II and Scan results. (Upper left and bottom
left yellow buttons in the menu above.)

Multiple instances of the translator and the output processor can be


started so long as each instance accesses data in a separate working
folder.

Editing Branch Connection Data

FEAToolsV2 and FEAToolsV3 automatically adjust the SIFs, k-factors and SSIs for B31
type branch connections using more applicable data developed from FEA results or from
correlation results. The correlation results used by default are from ASME B31J.

In the model below, each branch connection is labeled.

FEAToolsV3 allows the user to turn off the use of more applicable data for branch
connections universally, for SIFs only (ignoring k-factors), or for any branch connection
individually. Similarly, the FEAToolsV3 user can activate more applicable data on an
intersection by intersection basis. (Although this is not recommended. In the old days
this would be done because of the time it took to replace a single intesection with
updated SIFs and k-factors, with FEATools, all the intersections can be replaced faster

24
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

than the time it used to take to replace only a single intersection.) FEA results are
stored in a library and the library is accessed each time the model is needed so any
single finite element model is only run one time.

There are 30 branch connections in the model shown above and 75 bends that are
modified in FEAToolsV3.

The input data for any of the intersections can be changed by pressing the:
button. A list of all intersections is shown, and the user can edit any or
all of their properties or intersection types. The scrollable intersection list for the 30
welding tees in this model is shown below.

To change the welding tee at node 340 to a light tee – scroll the right column down to
the bottom and select “Light” as the thickness rating override.

25
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

A light tee will generate SIFs, k’s and SSIs for some very thin tees that have been found
during inspections. Note that these tees may not satisfy B16.9 burst test requirements
when the d/D ratio > 0.5, and almost certainly would not satisfy B16.9 burst test
requirements when d/D=1.

The different intersection types available are shown below:

26
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

When FEA is selected, the best available model for each intersection will be generated
and analyzed. When the correlation method is selected, all welding tee types default to
the B31J correlation, (unless another correlation method is selected), also all olet types
default to the B31J olet correlation unless another correlation method is selected.

Note that EN-10253-2 Type A tee input includes an “X” factor which is the percentage of
the design pressure that can be accomodated by the tee.

A variety of different input parameters can be entered for any single or group of tees.
Ring and refractory input for tees are shown in the panel below.

SIF, k-factor and SSI overrides are provided for each tee also thru the “More” button:

27
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

Users can override any individual SIF, SSI or k-factor for any B31 branch connection
identified in the model.

B31 Intersection types and parameters can only be changed or specified where a B31
type intersection already existed in the CAESAR II model. For example, the FEAToolsV3
user may change a CAESAR II pad reinforced intersection to a welding tee, but cannot
specify a branch connection where there was not already a branch connection (of some
type) in the CAESAR II model.

Example: It was noted in the following model, that a 4x22 B16.9 welding tee was used
for the branch connection in the CAESAR II model.

28
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

Then it was pointed out that B16.9 doesn’t list 22x4 welding tees, and so the question
became, “what fitting (if any) was actually used?” We wanted to know if an inspection
team needed to go out into the field and track each of these intersections down to see
what they actually used?

An unreinforced fabricated tee was run in place of the welding tee to compare the
current B31.3 unreinforced fabricated tee to B31J to a finite element analysis. Those
comparisons are shown below.

Intersection Node = 20 Type = Unreinforced, Fabricated UFT

Run/Header Outside Diameter = 22.000


Branch Outside Diameter = 4.500
Run/Header Thickness = 0.500
Branch Thickness = 0.237

Nozzle Transition Length = 1.000


Nozzle Angle (Degrees) = 0.000
Modulus Used = 29500000.

Branch Flexibility Factors (kax, kin, kout, ktor)


1.979 4.595 9.504 0.287

Header Flexibility Factors (kax, kin, kout, ktor) - ALL RIGID

BRANCH SIFs RUN/HEADER SIFs


B31.3 B31J FEA B31.3 B31J FEA
In 2.592 2.491 2.969 5.469 2.367 1.253
Out 3.298 3.874 4.707 6.959 0.927 0.365
Tors 1.000 0.241 0.564 1.000 0.757 0.863
Axial - - 6.634 - - 1.301
Press - - 7.035 - - 2.943

The run SIF from B31.3 that is not reduced by t/T matches the finite element result.
B31J is a little higher than the B31.3 result for the branch using t/T. Running the PRGiK
calculation below, we can see that B31J matches closely the FEA work of Wais and

29
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

Rodabaugh, (although we know that the Wais/Rodabaugh correlations did not consider
the t/T problem addressed in B31J.)

When looking at the iob equation in B31J:

It can be seen that the Sketch 2.3 note for a stress factor multiplier when t/T<0.85, and
d/D is less than 1, and D/T is greater than 25 applies well to this intersection. The t/T
ratio vs. the factor given in the note is shown below. The maximum however, is seen to
occur around a t/T equal to 0.45.

Correlation equations for other methods are shown in the PRGiK spreadsheet below.

30
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

To verify the FEA result shown here, NozzlePRO can be used and a shell or brick model
analyzed. It would be reasonable to run the shell model first, and adjust thickness and
known stress concentration parameters to see what has an effect on the predicted SIF,
however a volumetric (brick) model can also be run easily in NozzlePRO (or FEPipe) for
this geometry. The NozzlePRO results for this geometry are shown below:

do=4.5; t=0.237, Do=22; T=0.5; d=(4.5-0.237)=4.263; R=(22-0.5)/2=10.75;D=22-


0.5=21.5. d/D=4.263/21.5 = 0.1983 (below Widera, but within Wais). D/T=43;
t/T=0.237/0.5=0.474, so this is below 0.85. When t/T is used it’s value should be 0.85 in
the iob equation:

(0.038+2(d/D)+2(d/D)2-3.1(d/D)3) = 0.48907; (R/T)2/3 = 7.732; (t/T)=0.85.; the SIF


without adjustment for t/T should be: 0.48907 x 7.732 x 0.85 = 3.214. The note to
Sketch 2.3 applies since t/T is less than 0.85, d/D is less than 1, and D/T=43, which is
greater than 25.

The multiplier is (0.75(t/T)-0.89(t/T)2 + 0.18)(D/T)0.34 = 1.205; This multiplier when used


with the calculated SIF gives an adjusted SIF of: 3.214 x 1.205 = 3.874. The FEA
calculation shows that the iob for this intersection is 4.707.

The largest stress without weld geometries is found by integrating thru the wall.
NozzlePRO and FEPipe do this automatically for branch connections. (The NozzlePRO
geometry for this model can be run and analyzed with little interaction from the user.)

The result is shown below:

The maximum integrated values thru the critical SCF in the nozzle is shown below.

31
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

REGION = 504

Node Sxx Syy Szz Txy Tyz Txz (Global)


14379 8361.4 218015.2 -4027.0 -271.0 -3.4 10982.6
14382 -38008.1 84796.9 -7934.9 -243.9 0.7 17390.3
14385 -77608.7 -26470.6 -17695.7 -224.7 6.1 28717.7
14400 -116201.6 -146216.2 -20317.9 -216.3 12.5 32499.5
14403 -168001.3 -303503.3 -33755.7 -238.2 20.2 53722.5

Node Sxx Syy Szz Txy Tyz Txz (Local)


14379 -3632.6 8361.4 217620.9 -10961.4 -733.8 -9349.0
14382 -7770.0 -38008.2 84632.0 -17364.6 -976.9 -3906.6
14385 -17710.8 -77608.7 -26455.5 -28682.7 -1435.2 363.6
14400 -20540.6 -116201.6 -145993.5 -32461.4 -1586.2 5290.4
14403 -34233.3 -168001.4 -303025.6 -53664.6 -2502.8 11341.7

Membrane Bending M+B(out) M+B(in)


SXX -16238.60 -13395.40 -29634.00 -2843.20
SYY -77909.74 -78924.47 -156834.22 1014.73
SZZ -32630.15 -232916.12 -265546.28 200285.97
Txy -27705.46 -17671.59 -45377.05 -10033.87
Tyz -1404.13 -726.02 -2130.15 -678.11
Txz 685.93 9268.11 9954.04 -8582.19

SUM 73288.68 198228.53 219805.59 202526.70

For the branch: Z=(/32)(Do4-Di4)/Do = 3.2144 cu.in.


M=1e4 ft.lb. M/Z = 37,332 psi. (This is the nominal stress used in the NozzlePRO calc
due to the applied moment.)
i=M+B/(M/Z); per approach in EPRI 110996, when the weld is not included in the model.

The maximum M+B stress is 219,805 psi from the thru-wall integrated results above,
and so a rough volumetric estimate of the SIF due to out-of-plane loading thru the
branch is be 219805/37332 = 5.89. The 5 node thru the thickness brick model shows a
SIF of 5.89. The mesh could be further refined, a fillet could be added, boundary
conditions could be altered to see their involvement, etc., (These alterations would have
to be done in FEPipe.) Essentially, the conclusion that would be drawn is that the
“higher” FEA calculated SIF shown for the UFT is likely not unreasonable. There are not
many tests for these geometries. Fatigue life for branch connections can depend on the
formation of multiple cracks during cycling, and so affects of thickness, or weld quality
and size can have strong influences also. The higher 4.7 value seems reasonable for the
time being. Tests are being conducted in 2017 at PRG using various t/T ratio branch
connections to attempt to quantify further this effect.

We know from WRC 329 that olets can actually have SIFs greater than unreinforced
fabricated tees, and so we’d like to run the poorest quality olet that has been come
across to see if it would possible produce a code stress violation. The worst anticipated
quality comes from the straight bore Atypical olet. These are not copies of standard
industry olets.

32
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

The FEAToolsV3 result for the atypical olet is shown below:

Notice how the FEA model of the atypical olet gives an iob equal to 3.544 while the B31J
olet gives a slightly higher SIF than the B31.3 Olet, i.e. 1.717 to 1.488. Even poor olets
are better than unreinforced fabricated tees in this service since the discontinuity at the
junction where t/T << 1 is removed by the extra thickness of the olet. With FEAToolsV3
it is easy to change the type of interesection and to rerun the finite element analysis to
get an estimate of the strength of different fittings. For this fitting the out of plane
strengths are recorded as follows:

33
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

Component iob Notes


t/T at peak of error. Brick
and shell fea models
UFT FEA 4.707
compare. Need test in this
parameter range.
UFT B31J 3.874
UFT B31.3 3.298
UFT Wais/Rodabaugh 3.606
Olet B31.3 (current) 1.4881
Olet B31J 1.7168
Bonney Forge (Tapered) 1.463 Larger footprint – lower SIF
WFI (Straight) 1.718
Olet Straight Bore 2.029
(Copy)
Olet Tapered Bore 1.788
(Copy)
4.404 Missing Bevel – no defined
Olet Atypical Straight
weld size
4.127 Missing Bevel – no defined
Olet Atypical Tapered
weld size

From the calculations above, even the atypical olets should provide adequate support
for the branch since the t/T<<1 junction is removed and replaced by the much thicker
body of the olet. The most likely olet to be purchased in bulk from a non-preferred
supplier is the Straight, Atypical, olet, since this is the least expensive to manufacture
and stock. The CAESAR II model for this FEAToolsV3 calculation is UNKNOWNTYPE.C2.

When looking at some of the comparison runs we see can see that the atypical olet FEA
run results can double the stress. This is seen immediately in the three-dimensional
plot window since the maximum stress ratio for each case is shown adjacent to the case.
(See the image below.) We can also see that FEATools adjusted the two bend
stiffnesses due to the close proximity of the valves for each and the intersection model.
(We know this by noting the gold representation of the “advanced geometry” at both
bends, and at the branch connection being studied in this example.)

34
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

Olet geometries are generated automatically when the user picks the type of olet to be
simulated. Users can change any of the dimentions if the measurements are available.
The definitions of particular parts of the olet are given below:

35
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

Typical partially filled olets are shown below:

These are similar to atypical olets in that less than design material is placed around the
olet, and particularly in the longitudinal plane. To simulate this effect the user can enter
a negative value for the weld cap length in the circumferential length on the special olet
form: .

The above sketches hopefully help the user recognize welded-on fittings that do not
satisfy recognized commercial geometries that have been routinely tested and that are
shown in B31.3 and B31.1.

36
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

3x4 Olet Tests by Woods and Rodabaugh (WRC 392) And Partial Olet Test by PRG

When reviewing OLET inputs and output for detailed olet geometries, the
following diagram can be used to get an idea bout the profiles of olet
geometries.

37
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

Atypical Olets Have the following weld profiles:

Placed on a fitting, the atypical olets appear as shown below:

38
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

39
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

Running CAESAR II after Converting the model

There are multiple ways to start CAESAR II for running, model editing, or checking. In
the main translator form below, the CAESAR II drop down allows the user to open
CAESAR II for the jobs currently identified, or in the current working folder.

Adjacent to the input and translated CAESAR II job names on the right are the familiar
CAESAR II “running man” icons. If a current CAESAR II license is present CAESAR II can
be started AND RUN using these buttons.

Also on the very bottom of the screen (see above), the “Convert →Run →Compare”
button will run the translator, then will run CAESAR II, then will do a job comparison and
then start the comparison processor.
40
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

Automatic Bends with Staunchions

V3 looks thru the CAESAR II model for trunions on bends. In the model below there is a
+Y support at the midpoint of the bend at node 49, and

At node 59, the pipe d/D problem for branch connections identified in WRC 329 moves
to bends with trunions. In this case the t/T ratio is 1, and the d/D ratio is less than 0.5 so
we don’t expect the Schneider effect to amplify the trunion side loads, but we do expect
a possibly high SIF on the trunion attached to the bend.

The +Y support at node 49 must be some type of trunion or plate so that the support
can be attached at the midpoint. The bend at 49 should have some type of stress
intensification factor.

There are no branch connections in the model so FEAToolsV2 would not change the
geometry. FEAToolsV3 will recognize the bends with staunchions at both nodes 49 and
59 and use the finite element method to find SIFs, k-factors, and SSIs. It will also detect
the bends and 20, 30 and 40 and see if they have consistent thicknesses, sufficient
attached lengths, etc. THIS IS ALL DONE BY DEFAULT. The bend reports for this model
are shown below:

NOTES:
1)Only the bend at 20 is printed since 30 and 40 are equal to 20.
2)A default trunion is assumed at TIPT=50 since a support is there, with no trunion data.
The user can enter data for 50 in FEAToolsV3 tab: .
3)The FEA ii for node 20 is 3% higher than the B31.3 value, while the FEA k-factor for
node 20 is 9.6% lower. This is the error corrected in B31J for bends.
4)Node 50 with the trunion shows a SIF doubling due to the trunion attachment.
41
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

5)The d/D<0.5 trunion between 59 and 80 shows a SIF of about 3.


6)At node 60 there is not enough straight pipe attached before the nozzle to allow the
full bend flexibility to be developed. Both the effect of the trunion and the attached
length will be used in the updated CAESAR II model.
Equivalent Bends
Bend at: is SAME as BEND:

30 20
40 20

Bend TIPT = 20
Bend Ftg OD = 24.000 in. Bend Ftg Thk = 0.250 in.
Bend Radius = 36.000 in. Bend Angle = 90.000 deg
Enough Str Pipe In/Out no BC Effect

Bend Bend Bend Bend


Method (ii) (io) (ki) (ko)
B31.3 No Flanged Ends 5.64 4.70 25.85 25.85
B31.3 Single Flanged 3.56 2.97 16.34 16.34
B31.3 Double Flanged 2.25 1.88 10.33 10.33
EPRI Adjusted Data 5.63 3.39 20.37 19.59
B31J Adjusted Data 5.64 4.70 22.72 22.72
N319-3 Adjusted Data 5.63 3.39 20.37 19.59
Return HyperDOF values 5.83 3.84 23.36 21.57
-------------------------------------------------------------------------------------------

Bend TIPT = 50
Bend Ftg OD = 24.000 in. Bend Ftg Thk = 0.250 in.
Bend Radius = 36.000 in. Bend Angle = 90.000 deg
Trunion OD = 18.000 in. Trunion Thick = 0.250 in.
Enough Str Pipe In/Out no BC Effect Bend ATTACHMENT is Present

Bend Bend Bend Bend TRUN TRUN TRUN TRUN


Method (ii) (io) (ki) (ko) (ii) (io) (ki) (ko)
B31.3 No Flanged Ends 5.64 4.70 25.85 25.85 1.00 1.00 1.00 1.00
B31.3 Single Flanged 3.56 2.97 16.34 16.34 ... ... ... ...
B31.3 Double Flanged 2.25 1.88 10.33 10.33 ... ... ... ...
EPRI Adjusted Data 3.32 3.39 22.40 6.00 1.18 1.00 1.00 1.00
B31J Adjusted Data 9.91 4.70 22.72 22.36 5.48 4.93 1.00 1.00
N319-3 Adjusted Data 9.91 3.39 20.37 19.27 5.48 4.93 1.00 1.00
Return Shell FEA values 9.05 4.06 19.61 21.51 3.73 2.53 ... ...
-------------------------------------------------------------------------------------------

Bend TIPT = 60
Bend Ftg OD = 24.000 in. Bend Ftg Thk = 0.250 in.
Bend Radius = 36.000 in. Bend Angle = 90.000 deg
BC Removal Len = 111.733 in. Str Length in = 111.733 in.
Str Length out = 41.000 in. Num Trunions = 1
Trunion OD = 12.750 in. Trunion Thick = 0.250 in.
Attached Lengths will affect bend Bend ATTACHMENT is Present

Bend Bend Bend Bend TRUN TRUN TRUN TRUN


Method (ii) (io) (ki) (ko) (ii) (io) (ki) (ko)
B31.3 No Flanged Ends 5.64 4.70 25.85 25.85 1.00 1.00 1.00 1.00
B31.3 Single Flanged 3.56 2.97 16.34 16.34 ... ... ... ...
B31.3 Double Flanged 2.25 1.88 10.33 10.33 ... ... ... ...
EPRI Adjusted Data 3.32 3.39 22.40 6.00 1.18 1.00 1.00 1.00
B31J Adjusted Data 9.91 4.70 22.72 22.36 5.48 4.93 1.00 1.00
N319-3 Adjusted Data 9.91 3.39 20.37 19.27 5.48 4.93 1.00 1.00
Return Shell FEA values 5.97 3.32 20.71 17.51 2.68 2.85 ... ...

When the translation is completed an updated CAESAR II model is placed in the working
directory with the first model. The default new model name has “-FEA” appended to the
original model name:

42
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

Once translated, the user can review the output from the original model, re-run the
original model, or run the newly created model. This last option is the most usual one
selected, and to do this, the user should press the 2nd running man icon:

When the run completes, the output button is enabled: , and the
comparison button is enabled: . The multicolored output review
buttons imply the “comparison” of two or more models. The default in this case is that
the results of the old model and the new model should be compared.

When the comparison form appears, immediately the displacements, restrain loads,
element forces and moments, and stresses can be compared. Common element stress
comparisons for this job are shown below:

Finite element model stresses are higher since SIFs were applied to the bend trunion
elements. If we go back and look at the “new” – translated model for the trunion on the
bend, we can see that FEAToolsV3 has generated SIFs and SSIs for the trunion.

43
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

Since the highest stress often involves the loads on the trunion an accurate estimate of
the SIFs on the trunion can be important. Note that the trunion SIFs and the bend
flexibilities and SIFs were generated automatically to use more applicable data in
FEAToolsV3.

The trunion at bend 50 defaulted to an 18” pipe because the user only entered a +Y
support at the midpoint of the bend. The user can change any of the defaults on a per
bend basis by entering: , clicking on the “add” data tab: , and then
entering the node number (or numbers if the data entered should apply to multiple
bends). To change the trunion diameter to 12” and the trunion wall thickness to 0.25”
at node 50 the user should start by adding a bend data line for node 50:

Any aspect of the bend can be changed here. The attachment type should be changed to
trunion so that the user can enter data for the trunion. (It could be changed to any steel
cross section shape also if the attachment was a steel shape.)

The “more” button on the right opens an additional data form for the bend as shown
below. The user can enter comments, or fill in any of the available input fields. User
entered data here is considered an override of data from any other source.

Notice that there is a “more” button on the “more” panel. This form is shown on the
following page also.

44
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

In the second form the user can specify refractory thickness and modulus of elasticity, or
ring locations on the bend and attached straight pipe. The thickness of the bend can
also be changed at the intrados and extrados to see how this affects SIF and k-factors.

The FEAToolsV3 user can use the bend forms to modify only those properties of the
bend that should change from run-to-run, all other properties may be left to default.

45
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

Checking Pipe Shoes

Most piping systems running thru refineries, power, and gas plants in pipe racks are
supported along some part of their lengths outside of the rack, and along most of their
length inside the rack by pipe shoes. The shoes may be fully welded to the pipe, skip
welded, partially welded in areas, or clamped. For vertical support only, the method of
attachment is not too important, (unless there are thermal problems with the shoes, i.e.
hot pipe in cold environment, or transient thermal conditions). Where there are guides
or limit stops, to restrict movement, lateral and axial loads, and torsion are often
applied to the pipe shoe and then to the pipe.

Pipe Shoe Loads that Produce Torsion on the Pipe

Most support providers give allowable loads for the pipe shoes, but
cannot consider whether the shoe and loading will overstress the pipe.
When pipe shoes are welded directly to the pipe without pads, or full
encirlement support (clamps), the stress distribution in the pipe and the
pipe shoes roughly follows the same guidance as for branch connections.
When tpipe/tshoe < 1 then the high stress is in the pipe. When tpipe/tshoe > 1
then the high stress is in the shoe. Also, the allowable load in the shoe
generally follows the same trend as the allowable stress in the pipe when
considering temperature. The allowable load in the shoe drops the same
or more usually, than the allowable load in the pipe when the pipe is hot.

The Paulin Research Group and Piping Technology and Products have
worked together to address the joint stress problem for the pipe and the
shoe with a simple model library whose allowable loads can be compared
to support loads at any restraint in the piping system. Friction loads,
guide loads, and axial stops will all be checked against the allowable
loads from the finite element calculation for the pipe shoe and pipe.

46
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

Shoe/pipe combinations are provided for standard wall and XS pipe walls
at ambient temperatures. For different pipe walls or higher temperatures
(where the allowables are lower) a procedure is provided for the user to
generate an additional library of shoes and allowable loads. The
allowable loads on a pipe shoe should be a function of the pipe
thickness and temperature.

Problem Addressed by Pipe Shoe Allowable Development and Library

To Check Loads on Pipe Shoes using PTP Catalog:


Step 1: Identify the line size and locations where PTP supports should be
placed. In the following system pipe shoes should be placed at 150, 170,
270, 230, 235, 5000, 5010, 5040, 5070, 5100, 5110, 5125, 5140, 5190
and 5200. The support at 5000 is a limit stop and guide. Guides are also
placed at 5110 and 235, 5200, and 170.

47
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

For pipe sizes larger than 6” the Fig. 100 is recommended, and so Fig. 100 will be used
for all 12” pipe for vertical support only. The Fig. 400 will be used for all supports where
horizontal and axial loads are possible. Note that supports with slide plates (Fig. 200’s),
and guides (Fig. 300’s) can also be purchased directly and installed.

When the type of shoe desired is needed, it is best to make a list of what will be
installed and where: An example list for this system is shown below.

# Support Type Nodes/Mark No.


1 Fig. 100 150(MK1004-1), 170(MK1004-2a), 270(MK1004-4),
12” Dia. Vert. Only 230(MK1004-5), 235(MK1004-7)
2 Fig. 400 5000 (MK1005-10), 5010(MK1005-8), 5040(MK1005-23),
12” Dia. 5070(MK1005-24), 5100(MK1005-4), 5110(MK1005-42),
Guides and Limit Stops 5125 (MK1005-43), 5140 (MK1005-44), 5190 (MK1005-47),

48
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

5200 (MK1005-48)

Once the nodes for supports are identified, and the runs for any CAESAR II model have
been made, the support evaluation can be performed.

Click on the Supports/Nozzles tab at the top of the Translation form:

Supports are identified by support type, pipe diameter and wall thickness. Press the plus
button ( ) to add rows for an individual type of support. The following line will be
added each time the plus button is pressed.

The NozzlePRO file should point to the PTP support type of interest. For PTP libraries,
click on the Lib button and select the PTP Library:

And then the supports for standard wall pipe:

And then the PTP-FIG100-to-300-12std supports.

Add the nodes for the Fig100 pipe shoes and then repeat the process for the Fig. 400
pipe shoe and the nodes where they will be applied. The final input is shown below:

49
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

The calculator button on the left generates allowable pipe shoe, NozzlePRO saddle, or
user defined nozzle allowable load reports for the original model. The calculator button
on the right generates allowable reports for the new (translated) model.

The allowable reports summarize any overstress, and collect the maximum load types
for each node for all load cases. For the report example shown here the summary
report tells that the maximum pipe shoe overstress is 210%.

Friction, guide, limit stop, and vertical weight loads are all summed and evaluated for
the node entered.

When the design is completed, the report can be given to the structural engineer for
beam sizing, and for pipe shoes: PTP can be contacted for support acquisition and for
any additional help getting the right supports for the application.
Piping Technology & Products has a large fabrication facility in Houston, Texas
fabricating varieties of fixed pipe supports, expansion joints, constant and variable
support spring hangers, snubbers, etc.
PTP has a large, full time technical staff in Houston to help with design problems
anywhere around the world.

50
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

The available Piping Technology and Products Shoes that are available in the current
database are listed in the tables below:

51
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

52
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

Entering Vessel Nozzles

In the model below the pipe is 14” x 0.375” wall and 20” x 0.375” wall. There is a B16.9
welding tee at node 30. The tee was specified in CAESAR II and so FEAToolsV3 will
identify node 30 from CAESAR II and use more applicable data. There is also a hillside
nozzle connection at node 80 and a radial nozzle connection in the same vessel at node
170.

These are straight nozzles, so all we need to do to produce a finite element model for
node 80 and 170 is enter the node numbers into the vessel tabs. This input is shown
below:

This is the only input to produce both the radial and the hillside finite element models
and results. When the model is “converted” ; the output: graphical tabs
let the user review the finite element models and stress distributions due to the
directional loads: (Jobname = MK10010.C2.)

53
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

All of the SIF direction loads are shown on default. The user may view each one-at-a-
time by clicking on the direction of interest and then zooming and rotating the image.
Single directional images can be viewed by selecting the direction of interest from the
drop-down window in the bottom right of the form. The images may be animated

Radial Nozzle at 170 Hillside Nozzle at 80

Vessel nozzles are usually specified either:


1) on a CAESAR II model of the vessel and the attached pipe, or
2) at a restraint where the vessel side of the nozzle is not included in the model.
These model types are shown schematically below:

54
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

In the model on the left, boundary conditions are placed at node 20, AND node 20 is
called out as the “nozzle” node. In the model on the right, boundary conditions are on
the bottom of the vessel (top of the skirt), at node 40, AND node 20 is called out as the
“nozzle” node.
The biggest difference between the two models, is that for the “left” model, FEAToolsV3
does not know the diameter or thickness of the vessel, and so the user must enter the
vessel diameter and thickness since these are needed to determine the flexibility and
strength of the nozzle. In the right model, the user has defined the entire vessel in the
CAESAR II input, and so the user must only identify node 20 as a vessel node and
FEAToolsV3 will identify the geometry of the vessel and nozzle.

The same concept for vessel nozzle node identification applies for vessel heads, as
shown in the sketch below.

On the left, the CAESAR II model stops at the surface of the head. On the right the
CAESAR II model continues thru the vessel to the point of zero displacement. In both
cases, the nozzle node number is 5. In the left model the user must also provide the
55
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

head type, vessel diameter and thickness, while for the right model only the head type
needs to be provided.

An example head model and orientation vector for offset nozzles is shown below:

In general, FEAToolsV3 will properly identify the nozzle and vessle properties, but the
user can always enter known data.

Returning to the example model MK10010.C2, the output from the run for the two
vessel nozzles is shown below.

56
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

The only difference between the two runs is the offset. One of the nozzles is a hillside
connection on the vessel, and the other is a radial connection on the same vessel. A
table showing comparison of the key data is shown below.

# Item Radial Hillside Ratio


1 Axial SIF 14.995 8.226 1.822
2 Inplane SIF 8.226 2.305 3.568
3 OutPlane SIF 11.049 5.807 1.90
4 Torsional SIF 0.711 4.961 0.2
5 Pressure SIF 22.446 30.709 0.73

1 Axial k-Factor 2.557 1.334 1.916


2 Inplane k-Factor 10.198 3.002 3.397
3 Outplane k-Factor 18.561 5.849 3.173
4 Torsional k-Factor RIGID 7.535 Inf

For this particular offset geometry, the axial, inplane, and outplane SIFs are half of the
radial value in the hillside geometry because of the larger footprint. The torsional SIF is
5 times greater in the hillside geometry because torsion about the hillside nozzle tends
to produce high bending in the long-sliver like opening. The pressure SIF is also 1.5
times greater for the hillside connection because there is a larger opening in the
pressurized shell. Flexibility (k) factors tend to go up and down with the SIFs, although
not proportional, and this is what we see with the k-factors for the hillside. Using radial
nozzles for hillside connections (at least with this geometry) can result in errors in
excess of 300%.

57
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

Note that all advanced items in the model above are sketched in gold. There are
elbows, elbows with trunions, and the hillside and radial vessel nozzle. Evaluated
allowable loads from NozzlePRO for the overloaded radial nozzle are shown below, and
accessed from: .

58
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

Standard geometries FEAToolsV3 will identify for nozzles in cones are shown below

Example FEA models for these head types are shown below:

A few other “offset” definitions are shown below:

59
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

Enter a Rectangular Header Box

A rectangular header box requires a minimum amount of data, but permits the user to
thoroughly describe the box properties so that a more accurate stiffness and allowable
load table can be developed. The API 661 input can be specified at the end of a single
element or at the intermediate node where a nozzle comes in contact with the header
box. The user may include structural steel in the model, but the header box stiffnesses
should be inserted beween the end of the nozzle and the start of the rigid element
running from the surface of the header box to the centerline of the header box. (This
geometry interacts with the CAESAR II model the same way that the vessel model
interacts with the CAESAR II model.)

Thermal bowing: When there is a significant temperature gradient from the top to the
bottom of the header box, thermal bowing will occur. The thermal bowing in the
header box only can be simulated by applying a bending moment at each end of the
header box. The moment is “thermal” in origin and does not produce a stress unless
there is a resistance to the moment. CAESAR II won’t know that, and so the large stress
produced by the moment must be removed from the calculation manually. The bowing
calculation is important though, (especially in the case where the cooler flow cycles),
since it can induce high stresses in the piping or the heater body that will otherwise not
be accounted for. Depending on how the user wants to evaluate the system, the
bowing load can be calculated as a separate load case and judgment made if thermal
relief (rerouting) is needed when the stress in the pipe due to thermal bowing alone is
calculated.

Steps:

1) Compute the thermal bending moment using the equations below.

 =  x (Ttop – Tbot )
S = /64 x (Do4 – Di4 ) <or> r3t
R=d/
M=ExS/R

60
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

2) Apply the moments to the ends of the header box at the two support points. (See the
figure below which represents a thermal profile with hot fluid in the top, and cold fluid
out the bottom.)

3) Setup the CAESAR II load cases that include the bending moments that induce the
bowing moment shown above. Note that the bending moment is applied to each end of
the header box in opposite directions.

4) Ignore the stress in the header box elements, for the bowing load case or remove
M/Z from the stress, where M is the thermal bending moment applied and Z is the
section modulus of the header box.

5) The stress in the pipe and at the header box junction will be accurate and due to the
thermal bowing. If the stress is too high, then the coupling between the nozzles is too
large.

To model the header box in CAESAR II:

Structural steel members can be used to model the header box, but pipe elements are
often used and the moment of inertia about the horizontal axis replicated. The
approach used for this is:

r3t = BH3/12 - bh3/12;

where: r=H/2; B=outside width of header box base; b=inside width of header
box base; H is the outside height, and h is the inside height. From the above
equation, the equivalent pipe thickness is calculated.

The temperature difference from the top to the bottom should not be greater than 200
F (110C), (otherwise API 661 requires that the header box be split.)

61
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

Jobname = “AIRCOOL047A plus M.C2”

For this example, the height of the outside of the header box is 24”, and the equivalent
thickness of pipe is 0.893 in. The design temperature difference from the top to the
bottom is 143F. To calculate the bending moment:

 =  x (Ttop – Tbot ) = 6.6e-6 in/in/degF x 143F = 0.0009438 in/in.


S = /64 x (Do4 – Di4 ) <or> r3t = (3.14)(11.553)(0.893) = 4326.5 in4.
R = d /  = (24-0.893) / 0.0009438 = 24483 in.
M = E x S / R = 29e6 psi x 4326.5 in4. / 24483 in. = 5124731 in.lb.
 (Artificial) due to Moment = M/Sxc = 5124731/4326.5x12=14214psi.

This moment (5124731 in.lb.) is applied as shown in the figure below:

The CAESAR II entered temperature for the header box is the average of the top and
bottom temperatures. Load cases for this model are shown below. The bowing
moments (F2) were added to the initial operating case, and a separate bowing only case
was setup to see what stress and displacement the bowing produces by itself. The
operating temperature in is 384F. The temperature drops 143F to 241F, so the
temperature of the header box should be (241+384)/2 = 312.5F

The nozzles in this system are 10” and 150#. The allowable loads from API 661 are:

62
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

The nodes in the model at the 661 header boxes are shown below. Nozzles into the
header box are 80-50 and 70-60.

The loads on 50-to-80 from the stress calculation are:

As seen in the plot, node 50 is at the header box. Loads shown above are much too
large. Mx=78545 ft.lb., and the allowed in pane is 2250 in.lb. 78545/2250 = 35 times
too high.

When we run the FEA solution for the header box and for the bends and fabricated tee
junction, the displacement comparison between the two systems appears below. More
flexibility usually translates into increased flexibility.

63
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

The 80 to 50 element becomes 80-to-51 and the loads drop significantly as the
displacements increase. The loads below are seen to satisfy the API 661 allowables.
The piping is stiff and tight, and a little bit of compliance from the header box in the
model drops the loads significantly.

The input for this model is shown below: Data for the nozzle will be taken from the
CAESAR II model. The user would only need to enter it here if there was a need to
override what was in the CAESAR II model.

64
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

When clicking on and the intersections, bends with staunchions,


and nozzle connections will be shown in the output panel as shown below.

When selecting Node 50, or 60 the API 661 header box model will appear for the user to
inspect:

Additional notes from API 661 regarding nozzle loads on coolers. The cummulative
loads acting on the header box must also be calculated – so this would mean getting the

65
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

inlet and outlet loadings. The API661 program accepts input of all these forces and
moments to see if they add or subtract from each other.

66
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

It is not uncommon to see allowable increases as shown below, and where loads exceed
perceived allowables the manufacturer should almost always be contacted.

67
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

Comparing Results

FEAToolsV3 provides a variety of output manipulation features that enhance the users
ability to quickly review multiple output files. Multiple graphics windows can be
displayed and any number of output files opened and compared.

Comparison routines automatically compare any number of models with any number of
load cases. Users are encouraged to learn the comparison and filter tools so that they
can see when model changes are effective and don’t have derrogatory effects on other
parts of the model. Filter information can be stored and retrieved so that groups of
nodes – for example at pump or equipment nozzles can be viewed quickly for all load
cases.

2D and 3D plots are also generated so that all the stresses for all the load cases for all
the nodes can be shown on one plot.

As a function of total system memory, all output files in a single folder can be read into
the comparison tool and processed simultaneously.

The thermometer and measurement tools permit pulling stresses and displacements off
the same plot simultaneously. Animation of solution load cases and scaling is always
available, and point clouds of the system can also be read in and compared at any time.

Some of the more useful buttons and tools are highlighted below:

- From the main translation menu – will add


requested more applicable data to the new CAESAR II model, run both the old and the
new CAESAR II models and then opens

68
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

-
Original and new models can be plotted, analyzed in CAESAR II, and the CAESAR II
output reviewed from the main translation screen.

From the FEAToolsV3 translator menu the user can open and change any CAESAR II
model, or can build new CAESAR II models.

- When Associated Files are available from the


Tools menu, the finite element models associated with each unique item in the model is
available for output review and rerunning. These are copies of the original files that can
be changed and modified for information, and that won’t change the results from the
translators analysis.

69
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

- when clicked the original and translated file


output will be opened and compared. The comparison tool can compare outputs from
any number of jobs. When is checked, the two outputs will be opened and the
following panel shown:

Using “File” any number of output files can be loaded for review at one time. The
comparison processor will compare any number of models at one time. Users can read
in five jobs for example, and do comparisons of any combination of those five. The max
2D charts shows maximum values of stresses, reaction loads, displacements, etc., for all
nodes. When hovered over each chart shows the node wherethe value is taken. All
charts are zoomable.

Any number of graphics windows can be brought up and moved to alternate monitors.
Multiple plots can be shown and reviewed at the same time.

70
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

Typical comparison plot of displacement resultants. Any number of output job plots can
be compared. In the plots above the original model is in blue, and the FEA model is in
orange.

On the left – the range describes the node list in the order entered by the user. On the
right the results are sorted by largest difference to smallest difference. The user can
hover over the plot to see which node is plotted. Each plot can be zoomed when the
cursor is in the range or ordinate space.

Charts below the comparison plots contain each of the plots in a tabular format so that
actual values can be compared. By looking at the right table above, we can see that the
biggest displacement difference occrs atnode 28 in Case 4. The displacement resultant
varies from 0.338” to 0.475”.

71
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

In the 2D charts tab, the “Select” dropdown lets the user plot all stresses for all nodes
for all load cases in a single 2D plot. This same plot can be produced in 3D in the 3D
plot tab:

The variety of plots and tables in 3D are shown below. Any of the displacement,
reaction, force/moment, and stress plots can be shown

72
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

Multiple thermometer tools can be used with any or all 3D graphics displayed.
Coordinates, displacements, stresses and closest node are all displayed on the tool.

73
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

There are a significant number of display options available in the File:Options menu
when the graphics screens are displayed. Rendering, Navigation, View, Symbols, Model,
Tools, Labels, and Hover are the menu items used most often. Navigation permits
multiple models to be locked and rotated together

74
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

Changing and Copying Models:

There are two models involved in most FEATools functions. The initial model is
the original CAESAR II file, the 2nd model is the modified CAESAR II file that
contains additional elements at B31 intersections, additional restraints for
branch connection flexibilities, and additional k-factors and i-factors for bends
and vessel/heat exchanger/661 heads and boundary conditions.

At times it is not unusual to want to convert initial models using different


options. For example, FEA or STP-PT-073 can be used to develop flexibilities and
SIFs. SIFs can be used and flexibilities can all be assumed rigid for example as an
additional option.

The changes to any model are stored with the C2 file for that model. Let’s say
for example, that we wanted to specify a vessel head at node 100, and we want
the diameter and thickness to be 120” and 1.5” respectively. The information
describing the head, its location and size are stored in a MiM XML data structure
with the C2 file for the job. This is shown in the sketch below:

There is MiM “extra” data stored in Example001.C2 after FEATools has run. The
“extra” data changes are made to the Example001-FEA.C2 file, but the
Example001-FEA.C2 file does NOT contain any of the MiM extra data. (MiM
extra data in the Example001-FEA.C2 file would describe changes that should be
made to that file.)

If the user wants to change the vessel head thickness from 1.5 to a corroded
value of 1.387”, then the user would:
a)Make a copy of Example001.C2 that contains the MiM data.
b)Using FEATools open Example001.C2 and go to the “vessels” tab.
c)Make the thickness change.

75
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

d)Re-Convert the Model.

Often when this is done, the user wants to save the original run of the model so
that the outputs can be compared. In this case the user would follow the
procedure below:

76
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

Laser Scanning and Point Cloud Manipulation:

Point clouds are also shown in the file list. (See the menu below.) Point clouds can be
plotted and manipulated by themselves, or with CAESAR II, NozzlePRO or PCL output.
When registered point clouds are first brought in they are usually much larger than
what’s needed and are too large to process conveniently. In these cases an editing
operation is usually the first step required. The 20 to 40 Gigabyte file is read in a low
resolution mode and edited. Multiple steps in the process are stored so that the user
can go back to any intermediate step in the editing process.

The FEAToolsV3 editing wizard helps the user thru the low resolution editing process.
After the editing is completed in low resolution mode, full resolution is restored by the
wizard when the final clipped point clouds are used.

The Point Cloud Wizard can be found from the file menu item, by right clicking on the
loaded point cloud name, or at the top of the 3D viewer:

The Point Cloud files appear as shown below from the “file” menu:

77
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

In the model below, the point cloud appears (and can be manipulated) adjacent to the
CAESAR II model.

Registered point clouds read by FEAToolsV3 are PLY – binary or ascii,


and PTS, in ascii. The PTS file was developed by Leica and is a publically
used format.

The Point Cloud and the Picture:

Point clouds can be presented many ways. Photorealistic images map


parts of photographs onto triangular maps drawn between the laser
points. The photorealistic images give the best representation of the
geometry visually, but lose the actual measurement resolution associated
with the geometry, and require more computing power – although most of
the compute horsepower required today resides on the graphics memory
and is removed from the multi-cpu cores that actually drive the compute
functions. Ultimately though, all images must be retrieved from a hard
disk and must pass thru one of the central CPU’s, and so large,
sophisticated, photorealistic images, although spectacular to look out,
ultimately slow down our desired task, and at least this first version of the
scan technology uses only colored points. Perhaps as hardware

78
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

continues to improve in terms of graphic representations (for gaming),


engineering applications will also improve and we can use the
photorealistic images to get even better precision in our point
identification. For the time being, we will use the points only, and we
believe, as you see below, and will experience during use, the points are
sufficent for most piping and vessel related activity.

Even point clouds without triangular fan connectivity and mapped photo
images can be hundreds of gigabytes. Usually the larger the better,
regardless of the size of the volume scanned. (Although sometimes, the
scanner can be too aggressive and take more images, at a much higher
resolution than what is needed, but this is the exception.)

These large point clouds must first be clipped to the region of interest.
There must ALWAYS be a region of interest, and anything over about 5-
to-10,000,000 points will often slow down a computer. With virtual
memory and 64 bit processing, huge data sets can be targeted and read,
but processing can be so slow as to make them difficult if not impossible
to use.

PRG Embedded Scan Tool:

In most PRG graphical applications the scan manipulation capability has


been installed. At PRG we’ve found a large number of applications for
the technology now that we have it available for almost any problem we
deal with.

The PRG scanner software is available in a module format that plugs into
most of the PRG 3D viewer applications in the anticipation that it will be
updated frequently and new uses will be found regularly.

The PRG scan tool, sets the target object limit at 5 million points,
although the reader can set this to any physical limit they desire.

The PRG scan wizard will walk the user thru five clip-processing steps:
1) Identify the file
2) Import a low resolution version of the file for clipping

79
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

3) Clip the file in possibly multiple steps. (Each step must be kept by the
program so that “undoing” an overly aggressive clip is possible.
4) Export the clipped model as a separate geometry/file to be retrieved
separately as needed.
5) Load any clipped model for use and exit the wizard (back to the
previous PRG graphic operation.)

When the Wizard is selected, the available point clouds are displayed as
shown in the following form:

When the desired file is selected, its properties are displayed and text
discribing the recommended processing approach is printed on the right
of the screen. The STEP SIZE IS ONLY USED TO REDUCE THE
RESOLUTION FOR CLIPPING LARGE MODELS. Once clipping is
completed, the FULL MODEL RESOLUTION IS RETURNED

80
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

The user can change any of the available inputs (the darkened text), and
press the green “LOAD” button at the bottom of the screen. Note that
“Previous” and “Next” buttons are also available for the user to step thru a
process that has already been executed.

The first time a huge scan file is loaded, a variety of parameters are
calculated for the model and this may take a few minutes. A slide bar is
displayed that calculates the remaining time for the model to be read and
evaluated. (See below.)

A low resolution image for a provided “registered” point cloud is shown


below. Registration can include all points, or only a particular “volume” of
points as shown below. We recommend that all scanned points be

81
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

retained. It is easier to clip than to rescan. A separate several page


document is available from PRG that covers recommended steps to take
when scanning.

The low resolution (5,000,000) points in the attached scan keeps the
points shown in the image below so that users can get a sense of what
can be seen with five million points.

Part of the extracted point cloud and the CAESAR II model is shown
below. Once the point cloud is clipped and saved at full resolution, the
interaction with the model can begin.

82
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

Interaction with the model generally involves:

1)alignment.
2)measurement
3)extraction

The alignment process itself involves relatively simple technology applied


in unique ways to extract particular information from piping systems.
Aligment for larger piping systems is typically performed multiple times
depending on the relative calculations between the systems that need to
be made.

The simplest alignment concept involves getting the two models oriented
the same in space. This involves picking similar points on both models to
use as basis points for the alignment. Often points selected are at exactly
the same place in the pipe surface normal direction, but are not quite at
the same place in the “in the surface of the pipe” direction. For this
reason alignment points may be weighted, with the weighting based on
which coordinate direction is the most accurate.

83
© Paulin Research Group – 2018 All Rights Reserved
F E A T O O L S ™ U S E R M A N U A L

The www.paulin.com web site should be consulted for updates and


training informaiton regarding the latest use of scanning technology.

84
© Paulin Research Group – 2018 All Rights Reserved
1
Chapter
F E A T O O L S ™ U S E R M A N U A L

Introduction to FEATools™
This chapter will provide the user with a brief
overview of what programs are included in FEATools™
and a summary of what each program does and how
it is typically used.

F EATools™ is a collection of programs to enhance the CAESAR II and PV Elite


analysis and design capability by providing state-of-the-art finite element
solutions, correlations and comparisons for common piping and pressure
vessel components. FEATools™ is based principally on ASME Section VIII Division 2
Part 5 and the B31 Piping Codes. Examples and “how to” discussions are provided
later in this manual.
The FEATools™ consists of the following programs/modules:

• CAESAR II FEA Translator


• PRGik / Criticality Calculator / Flaw Detection Calculator
• FESIF
• FE107
• FETee
• FEBend

Each of these programs is described briefly on the following pages, but the bulk of
this user manual is dedicated to the CAESAR II FEA Translator. The other programs
(PRGik, FESIF, FE107, FETee, & FEBend) have been in use for some time as part of
other PRG products and are documented elsewhere. The reader is encouraged to
refer to that documentation for more information on these programs.

© Paulin Research Group – 2018 All Rights Reserved 82


F E A T O O L S ™ U S E R M A N U A L

FEATools™ Program Summary Descriptions

CAESAR II FEA Translator

The CAESAR II FEA Translator (Version 3) is a program that the user runs to
“convert” or “translate” their CAESAR II model file in order to use more applicable
data for piping system branch connections, bends, pressure vessel connections
and pipe shoes, based on the finite element method, industry correlations and
other more recent test data. The CAESAR II FEA Translator generally only requires
the click of a single button to use. The user interface exists primarily to let the user
to select an input CAESAR II model file and to specify any non-standard translation
options.

Understanding everything the CAESAR II FEA Translator does, and how it can be
used is another matter and these topics are discussed in more detail later in this
user manual.

At a high level, the CAESAR II FEA Translator does the following:

• Reads a CAESAR II model file and writes out a (“translated”) CAESAR II


model file
• Converts each standard B31 intersection (branch connection) by replacing
the intersection’s nodes with a new set of nodes (and restraints, if needed)
• Replaces OR supplies more applicable Stress Intensification Factors (SIFs or
i-factors) and flexibility factors (k-factors) for the converted
intersection/branch connection.
o These updated SIFs and k’s are based on a complete FEA analysis of
the intersection/branch connection geometry or optionally on
modern correlation equations.
• Allows the user to change properties of the branch connections provided
by CAESAR II. The B31 Intersection types can be changed and additional
data can be provided for each intersection. A wide variety of contoured
welding tees is provided to the user in the finite element option so that the
critical system user can be sure that whatever tee happens to be
purchased that the analyses will envelope its characteristics.
• Updates k-factors and SIFs for bends and bends with staunchions. (2016
and earlier B31 Code equations in Appendix D do not address bend angle,
thickness variations, trunnions, or attached pipe lengths – that can
materially affect stiffnesses and stresses in bends.
• The user can specify nozzles at locations in pipe-vessel models and the
vessel finite element model will be constructed based on the local CAESAR

© Paulin Research Group – 2018 All Rights Reserved 83


F E A T O O L S ™ U S E R M A N U A L

II geometry and stiffnesses and SIFs automatically included in the CAESAR


II model.
• API 661 Nozzle Connections can also be entered into the CAESAR II
geometry automatically by the user.
• All pipe shoes can be automatically checked for allowable loads.
• Compares output from two or more CAESAR II runs.
• Permits running and rerunning of CAESAR II from the FEATools interface.
The translated CAESAR II model file can be used as a standard CAESAR II model,
with a more accurate model of the flexibilities and stresses at branch connections,
bends, 661 header box connections, and nozzle/vessel connections. A few
diagrams representing a simplified description of how the standard intersection
models are “translated” are given on the following page.

The following table shows typical restraint load drops when flexibilities are added
to the piping models using FEATools.

© Paulin Research Group – 2018 All Rights Reserved 84


F E A T O O L S ™ U S E R M A N U A L

© Paulin Research Group – 2018 All Rights Reserved 85


F E A T O O L S ™ U S E R M A N U A L

Simplified Schematics of the CAESAR II FEA Translator Model Changes

Essentially the same modifications are made for pressure vessel heads and shells, 661 header
boxes or other rectangular vessel connections, bends and bends with trunions or steel
attachments – and evaluations are made for pipe shoe loadings for all load cases for all nodes
automatically. (A Piping Technology and Products catalog of pipe shoe models and allowables for
standard wall and XS pipe is included.)

© Paulin Research Group – 2018 All Rights Reserved 86


F E A T O O L S ™ U S E R M A N U A L

PRGik

Provides an instant evaluation of i-factors and k-factors for ASME B31.3 - Appendix
D branch connections. PRGik helps the designer/analyst determine whether the
piping system will be affected by overly conservative or under conservative SIFs or
k’s. An example of the PRGik User Interface is shown below.

PRGiK also makes sustained stress factor calculations using more applicable data
and per B31J:

© Paulin Research Group – 2018 All Rights Reserved 87


F E A T O O L S ™ U S E R M A N U A L

PRGiK also computes updated SIFs and k-factors for bends so that users can see
the potentially large difference between the currently produced k-factors by the
Codes and more applicable data in Section III and as produced by PRG hyperdof
bend calculations.

© Paulin Research Group – 2018 All Rights Reserved 88


F E A T O O L S ™ U S E R M A N U A L

PRGik also compares calculated stresses to predicted failure stresses to provide a


more complete understanding of the safety factor actually being used. The PRGiK
user simply clicks on the “Pipe Stress Evaluation” button on the main user
interface and then follows the instructions on the form. An endurance curve like
the one shown on the following page is produced.

This result provides the user with a quick visual evaluation of how close the design
stresses are to the several available endurance curves.

The table below shows the effect of the different intersection models on the
natural frequencies of small bore branch connections.

© Paulin Research Group – 2018 All Rights Reserved 89


F E A T O O L S ™ U S E R M A N U A L

FESIF

FEA for fabricated branch connections used in both piping and pressure vessels.
FESIF uses FEA methods to compute SIFs and k’s for a single standard piping or
pressure vessel geometry with just a few pieces of input data.

The main User Interface for FESIF is shown below.

By entering just a few values the user can quickly perform an FEA analysis of the
model and extract SIFs and flexibility factors for both the run and the branch, or
the vessel shell or head.

The analysis results are provided in both tabular and graphical format for the user
to easily incorporate into their other design and analyses. An example of FESIF
output is shown below.

© Paulin Research Group – 2018 All Rights Reserved 90


F E A T O O L S ™ U S E R M A N U A L

SIFs, flexibilities, allowable loads and plots are provided from FESIF.

© Paulin Research Group – 2018 All Rights Reserved 91


F E A T O O L S ™ U S E R M A N U A L

FE107

FE107 performs FEA for fabricated branch connections to compute stresses,


flexibility factors, and allowable loads per ASME Section VIII – Division 2. The
calculated results are also compared to WRC-107 and WRC-297 results for the
same component.

The main User Interface for FE107 is shown below.

By entering just a few pieces of input/geometry data for either a vessel or piping
connection the user can quickly perform a full FEA analysis of the model.

The results provide a side by side comparison with the values predicted by WRC
107 (nozzles on heads) and WRC 297 (cylinder on cylinder) for the same geometry.
An example of this output is shown below.

© Paulin Research Group – 2018 All Rights Reserved 92


F E A T O O L S ™ U S E R M A N U A L

The following types of reports/results are available in FE107

© Paulin Research Group – 2018 All Rights Reserved 93


F E A T O O L S ™ U S E R M A N U A L

FETee

FETee performs FEA modeling and analysis of individual industry standard


contoured/welding tees, (extruded and B16.9). FETee also includes an ASME B16.9
database for welding tees. FETee computes i-factors and k-factors for these
components if loads are not given, and stresses if loads are provided. If not
available, automatic crotch thickness profiles are automatically generated based
on run and branch thicknesses.

The main User Interface for FETee is shown below.

After the user has entered the welding Tee’s geometry, FETee performs an FEA
analysis of the model. When the FEA analysis is complete, both tabular and
graphical results are produced. The output results include calculated allowable
loads for standard load cases.

© Paulin Research Group – 2018 All Rights Reserved 94


F E A T O O L S ™ U S E R M A N U A L

An example of the FETee results “window” is shown below.

The Tabular results panes include a large number of results, organized in such a
way that they could be easily included in other design/analysis reports.

© Paulin Research Group – 2018 All Rights Reserved 95


F E A T O O L S ™ U S E R M A N U A L

FEBend

FEBend performs FEA calculations of SIFs, flexibilities, stresses, and allowables for
bends with or without stanchions. Stanchion supports on bends can be round
pipe or one of several standard structural shapes. FEBend can also evaluate
stresses due to axial temperature gradients in the attachment, and allows for
inclusion of local pad reinforcement on structural or circular attachments.

The main User Interface for FEBend is shown below.

The user can build the model of the bend with a support by specification of just a
few parameters (Attachment Type, Support Type) and can perform the FEA
analysis with only a little extra data. The support types for structural
attachements/supports is shown in the drop-down list in the graphic above.

Once the model is defined, and the FEA analysis is completed, FEBend provides a
rich set of results and reports.

Like NozzlePRO and FETee, FEBend provides both tabular and graphical results in
the main window of the program once the analysis results are ready. The user can

© Paulin Research Group – 2018 All Rights Reserved 96


F E A T O O L S ™ U S E R M A N U A L

navigate the results areas to reiview or copy any results of interest. Also, the
entire result set (both tabular and graphical) can be copied to an external
document or printed. An example of the FEBend window with results displayed is
given below.

© Paulin Research Group – 2018 All Rights Reserved 97


F E A T O O L S ™ U S E R M A N U A L

Why should you use FEATools™?


Using FEATools™ will:

• Bring FEA or more applicable data to bear on piping and pressure vessel
design/analysis problems involving pad reinforced intersections, laterals,
hillside nozzles, pipe on bends, structural steel attached to bends, B16.9
tees, extruded tees, or any standard connection involved in cyclic service.

• Produce automatically improved (more applicable) CAESAR II models of


branch connections in the entire piping system with the press of a single
button.
• Typically reduce calculated equipment loads, and forces and moments on
intersections.
• Remove D/T limits, and improve stress calculations for known non-
conservatisms in the Code. (See B31.3 Appendix D Note 11.)
• When fatigue is a concern plot calculated stresses, or stresses computed
using up to five SIFs versus commonly used allowable and failure curves
from B31, ASME, Markl and Hinnant/Paulin.
• Provide a cyclic pressure design capacity. (Pressure SIFs are developed
from finite element models.)
• Automatically include vessel and heat exchanger flexibilities to produce
more realistic loads
• Improve stress analysis throughput and model review capability
• Read in and compare laser scanned point clouds to CAESAR II model
results, improving the accuracy of the model and ability to actually predict
what loads exist on equipment and pipe.

In general, any time a vessel designer, piping designer, or stress analyst is dealing
with tight piping systems or critical branch or equipment connections that
produce high calculated stresses or high calculated loads, the user can apply
FEATools™ to determine if the Code calculations are being grossly over or under
conservative. Applying the FEATools™ technologies takes very little time and can
provide great value in terms of material, time and cost savings.

Attention: We URGE anyone reading this document to go to the examples


provided in Chapters 4 & 5 to convince themselves of the impact this
technology can have on piping stress analysis results. See the results and
think about how this can apply to the work that you do every day!

© Paulin Research Group – 2018 All Rights Reserved 98


F E A T O O L S ™ U S E R M A N U A L

Typical questions addressed by FEATools™


• Is my piping system really overstressed?
• Are my pump loads accurate?
• Do I really need to reroute this pipe?
• How does refractory influence stress in the pipe and loads on nozzles
• Do stiffening rings influence bends in large D/T systems.
• If I have a lot of 45 deg bends in my system, will the Code angle error effect
my calculations?
• I’m not sure where the olets came from – what’s the worst I can expect?
• Where can I get the best olet available? (Lowest SIF – higher weight)
• Are my stresses for this branch connection conservative?
• What am I, as the designer, responsible for?
o B31.3-2010 Table D300 Note (11) “The out-of-plane … SIF for a reducing branch
connection with [0.5 < d/D < 1.0] may be non-conservative. … Selection of the
appropriate SIF is the designer’s responsibility.”
• Will the pump really be OK?
• B31.3 Appendix D Note 1 says that I am responsible for selection of the
appropriate SIF. I know some of the current Code SIFs are wrong. Should I
do anything special for my particular tee?
• Should I be using a torsional i-factor?
• Am I using more applicable data?
o B31.3-2010 Table D300 Note (1): “Stress intensification and flexibility factor data …
are for use in the absence of more directly applicable data. … Their validity has
been demonstrated for D/T ≤ 100.”
• How close is this stress to failure?
o Does the low number of design cycles help?
• Are all my intersection models reasonable?
• Are the flexibility factors for all branch connections really 1.0?
• Does an API 661 header box provide any flexibility?
• Will the weld on my stainless steel branches change the SIF?
• I have a hillside nozzle and the pressure cycles – will it be OK?
• Are any of the pipe shoes overstresses or are they overstressing the pipe?
• Can I use 100% of the Code allowable stress?

© Paulin Research Group – 2018 All Rights Reserved 99


F E A T O O L S ™ U S E R M A N U A L

B31.3 Appendix D Table Showing all Flexibility Factors = 1.0

© Paulin Research Group – 2018 All Rights Reserved 100


2
Chapter
F E A T O O L S ™ U S E R M A N U A L

The CAESAR II FEA


Translator
This chapter provides information on what the
CAEASAR II Translator program does and how to use it

T He CAESAR II FEA Translator is the culmination of 20+ years of PRG work


in the development and testing of FEA technology, and at the same time
it can be applied by the user to a CAESAR II model in a matter of
minutes. This chapter provides detailed step-by-step information on how to
use the program and provides some theory and background on what the
Translator is doing “behind the scenes” so that you have a good general
understanding of how your CAESAR II model is being transformed (translated).

By using the CAESAR II FEA Translator more applicable Pipe, Bend, Vessel, Heat
Exchanger or API 661 SIFs and flexibility factors (for nozzles or branch connections)
are automatically inserted into your CAESAR II models. UFTs, RFTs, Olets, Welding
Tees, and Contoured Tees are evaluated using more applicable data to permit the
determination of more accurate stresses and more realistic loads on the pipe and
equipment (pumps).

The CAESAR II FEA Translator reads in a CAESAR II input file (.C2), and writes out a
new, updated CAESAR II input file (.C2) as well as output reports that can be
included in other CAESAR II reports if applicable. The original CAESAR II input file is

© Paulin Research Group – 2018 All Rights Reserved 101


F E A T O O L S ™ U S E R M A N U A L

preserved, untouched, so that you can always revert back to it at a later time if
needed.

Starting & Running the CAESAR II FEA Translator –


Quick Start

S tart the CAESAR II FEA Translator program from CAESAR II or from the
FEATools™ program “launch” panel shown below:

After selecting the translator, the screen on the following page will appear.

Enter the input job name. A default output job name will be entered
automatically. Change the default output name if desired. Select how you
would like the standard branch connections to be “converted” (General
Options) and then press the green “convert” button to get a “reviewed” and
converted CAESAR II input file. The default procedure translates branch
connections, bends and bends with supports only. (Bends with attached pipe
or restraints on the curvature are assumed to be trunion supported bends,
and FEA or correlation models used to produce more appropriate SIFs and k-
factors. The automated processing for bends will produce a unique k-factor
and i-factors for all bends where there is a variation in either the bend angle,
thickness (as compared to the attached pipe), or the length of attached pipe.
(FEAToolsV3 determines the amount of attached straight pipe that’s needed
to remove any boundary condition effect for the bend, and if that length is
available, then the standard bend k- and i-factors are used. If that length is
not available, then some stiffening and reduction in the stress on the bend is

© Paulin Research Group – 2018 All Rights Reserved


102
F E A T O O L S ™ U S E R M A N U A L

present. The output processor for each bend shows the B31 code calculated
value (simplified) and the finite element calculated value.

The folder icon or “file” menu can be used to search for <name>.C2 or
<name>._A files. Once a job name is selected the user may plot the job using
the plot ( ) button, or run the job using the “run” icon (if the button is
“hot”): . Note that a “hot” button is dark: and a deactivated button is
“grayed” out: . Grayed out run buttons mean that the job has not been
adequately prepared or checked yet. Once the job has been run, the output
button may be pressed to review the output for the job.

If a job is selected that has been previously analyzed, then the user can plot
the job, rerun it, or look at the output. An example input where all buttons
are available is shown below:

The green “Convert” button on the bottom right of the form starts the
conversion/translation process. The button to the left of the “convert” button

© Paulin Research Group – 2018 All Rights Reserved


103
F E A T O O L S ™ U S E R M A N U A L

will covert → run → and compare the original and new model results after
they have both been run.

Once the convert and run buttons have been pressed, the View output
comparison button can be pressed and the FEAToolsV3 output comparison
program will be started and the output from the original “non-converted”
model will be compared with the output from the “converted” model.

(The run comparison can be used for any two CAESAR II model files that have
already been run, or for any number of CAESAR II model files. The limit on
the number of jobs that can be compared at one time is the memory
available on the computer.)

The translator will run for several minutes depending on how many finite
element models must be created and analyzed. If improved intersection
correlations are used instead of the FEA analysis, only a few seconds would be
required to convert the entire model. A dialog box and small beep will sound
when the job is complete. When done, an output <name>.C2 file will be
written with more accurate SIF and flexibility models for the intersections in
the job. The new input file will contain the same load cases as the original
input file and can be read in and analyzed any time after the FEATools™
conversion completes.

When the first intersection is analyzed, the graphics tab will be opened and
five SIF plots will be shown. They are for the axial, in-plane, out-of-plane,
torsional, and pressure loadings on both the branch and the header/run. The
node number for the intersection is centered on the bottom of the picture.
The 3d models are interactive. The user can click on the models and rotate,
zoom or pan. The buttons to the right of the graphic image have hover-help,
and can be used to start or stop the animation and rotation. When more than
one intersection model has been analyzed the combo-box at the bottom of
the screen contains the node number for each intersection analyzed. The user
may select any intersection in this list to review the high stress location and
distorted shape for that particular intersection. High stress locations show
where failure due to an individual loading is most likely to occur and help the
user validate that the FEA result is accurate. The multiple tabular comparisons
made by FEATools also helps the user know that the FEA result is accurate. An
example plot for run i-factors and k-factors is shown below.

© Paulin Research Group – 2018 All Rights Reserved


104
F E A T O O L S ™ U S E R M A N U A L

Detailed tabular reports are available in the “Text Output” tab when all
analyses are completed. Charted output showing the computed SIFs are
produced during the run. More detailed information is available in the
following “How to Use” Section.

The .C2 file output from the translator contains the enhanced CAESAR II
model with more applicable SIF and flexibility data ready to run.

Many thousands of models and loading conditions have been run thru NozzlePRO
finite element templates and so a degree of confidence comes from the shear
number of models successfully analyzed. Unusual geometries, trunions and pads
pushed up against the edges of discontinuities, etc., can result in mildly or
significantly distorted finite element meshes. This is one reason that FEATools,
NozzlePRO, AxiPRO and FEPipe automatically compare the finite element solutions
to established, although less accurate hand solutions for the same or similar
problems. Bend finite element solutions are compared to existing B31 solutions
and to other correlation equations so that comparisons can be drawn.

© Paulin Research Group – 2018 All Rights Reserved


105
F E A T O O L S ™ U S E R M A N U A L

Intersection finite element calculations are compared to existing B31 solutions and
to correlation equations from STP-PT-073/B31J.

Nozzle solutions can also be generated from WRC, correlation or other solutions
inside FEAToolsV3.

Additionally the user can review the finite element mesh and exaggerated
displaced shapes to be sure that loading directions and stress distributions are
developed as expected.

The translator output text report details every change made to the model and
compares new k-factors, SIFs and SSIs to values that would otherwise be used as
an additional check to the user.

© Paulin Research Group – 2018 All Rights Reserved


106
F E A T O O L S ™ U S E R M A N U A L

Starting & Running the CAESAR II FEA Translator – Detailed


Input Documentation

A fter installing FEATools™, you should either have a shortcut on your


desktop (if this was selected during the installation process) or you can
start FEATools™ via the typical Windows application initiation. In either
case, you should be able to locate the FEATools™ program and Icon that looks like
the following:

- the Icon is labeled “StartFEATools”

FEATools™ (and the CAESAR II FEA Translator) can also be started from within the
CAESAR II program. Details on how to do this will be included with the Intergraph-
supplied documentation for CAESAR II.

Clicking on the StartFEATools Icon or the FEATools™ option found in the Start
control will start a program whose user interface appears below:

Clicking on the button labeled “CAESAR II FEA Translator” starts the CAESAR II FEA
Translator program. The CAESAR II FEA Translator program user interface (GUI) is
shown on the following page.

© Paulin Research Group – 2018 All Rights Reserved


107
F E A T O O L S ™ U S E R M A N U A L

Main User Interface for the PRG CAESAR II FEA Translator program

The Input Tab in the upper left is selected when the program is first started. The
selections the user can make on the Input tab are described on the following
pages.

© Paulin Research Group – 2018 All Rights Reserved


108
F E A T O O L S ™ U S E R M A N U A L

Main CAESAR II FEA Translator program Input Tab Values

Input >>

Under the input chevrons (>>) are options for input files and files to retain.

Select this radio button if the file you want to convert/translate is in the form a
CAESAR II “neutral” file – a file with an extension of “.CII”. This method is not
recommended since it is not the most convenient way to run the translator.

Select this radio button if the file you want to convert/translate is a “native”
CAESAR II file – a file with an extension of “.C2” or “._A”. This method is the
default and is recommended since it is the easiest way to run the translator.

If you have selected the “Convert CAESAR II Input File” radio button above, then
you can check the “Keep Intermediate Input Neutral File” checkbox to tell the
CAESAR II FEA Translator program to keep (preserve) the intermediate input
CAESAR II Neutral file that is created in the conversion process. This file is
generally not needed unless the model is to be moved to the PCL or some other
program.

This edit box is where to specify the name of the CAESAR II input file that is to be
converted/translated. The file can be located using the Browse button shown.

Output >>

Under the output chevrons (>>) the user may select the version of the CAESAR II
file to be used for outputting the desired result.

© Paulin Research Group – 2018 All Rights Reserved


109
F E A T O O L S ™ U S E R M A N U A L

General Options

This area of the GUI allows the user to specify how the CAESAR II FEA Translator is
to convert or translate the input CAESAR II file/model. The details for each option
are covered below.

FEA

If the user selects this option, all standard SIFs and k-factors for B31 intersections
found in the input CAESAR II model will be “updated” to improved values based on
an FEA analysis of the branch connection.

ASME 07-02/STP-PT 073/B31J

If the user selects this option, all standard SIFs and k-factors forB31 intersections
found in the input CAESAR II model will be “updated” or “converted” to improved
values based on correlation equations derived from a variety of sources including
FEA analysis. (The original improvements to the B31.1/B31.3 equations are taken
from WRC 329 with updates from EPRI 110996, ASME ST-LLC 07-0 and other
documents.

Translate Only

If this option is selected the CAESAR II FEA Translator does NOT update the model
with any new or updated SIFs or k’s. It simply translates one input file to another
input file of a different version.

In the General Options area of the GUI, the user can specify the Minimum Element
Length (mm) to be used. If a value is entered, the CAESAR II FEA Translator will not
produce a CAESAR II element that is shorter than the allowed minimum length.
The minimum element length is usually on the order of 0.1 inches, although
smaller or larger values may be entered. The field can be left blank and the
minimum element length will be taken to be 2.54 mm. (0.1 inches)

© Paulin Research Group – 2018 All Rights Reserved


110
F E A T O O L S ™ U S E R M A N U A L

Intersection Options

For those that are interested in having more control over the conversion process
and want to control some of the more technical features of the program, the user
can click on any of the overrides to see what effect certain adjustments to the
model will have. This can be particularly important for pump and critical
equipment loads where we want to know how sensitive the solution is to
properties of the mathematical model that we can only estimate at the design
stage.

For the critical equipment loads, the user is encouraged to change the universal
overrides and then to compare model results to see how different assumptions
affect results. Multiple variations can be compared at one time and filters can be
used to only list results for nodes of interest. In general, we would prefer lower
equipment loads and higher stresses thru the system.

Detailed Options, Stress Intensification Factor options, Sustained Stress options,


and the Minutia all control selections that can be made by the user on an
individual basis.

The logic used to decide how to improve each piping component is based on a
hierarchical arrangement of switches detailed in the figure below:

© Paulin Research Group – 2018 All Rights Reserved


111
F E A T O O L S ™ U S E R M A N U A L

The minutia tab allows the user to decide how to apply attached pipe lengths to
the intersections and bends on a global basis. Generally allowing the program to
try and select the appropriate attached length is the best approach. If for a
particularly critical intersection the user wishes to change the programs approach
then control is available globally from the Minutia tab, or locally for each branch
connection, bend, nozzle, etc. Also here the user can ask for pressure stiffened k-
factors for bends and branch connections. Until CAESAR II permits k-factor
adjustment for pressure for user generated k-factors only a single user defined k-
factor can be used. The k-factor adjustment based on angle, and attached lengths
are considered much more significant k-factors. It is anticipated that at some
point in the future a variety of k-factor adjustments based on pressure and
deflection will be available for each piping component.

Note that most of the switches described above as options are also available
individually for each branch connection, bend, 661 nozzle, or other vessel nozzle.

Starting & Running the CAESAR II FEA Translator – Detailed


Output Documentation

O nce the user has clicked the light-green “Convert” button on the bottom
right of the translator screen or the Convert → Run → Compare button
also on the bottom left of the screen, the CAESAR II FEA Translator will
begin the process of converting the CAESAR II file. Depending on the options
selected, if this is the first time the job has been run, then FEA models may need to
be run and solved for each unique intersection, nozzle, and bend in the model.
This may take from one to several minutes. A progress bar shows the as the
solution continues. Finite element models only need to be run once. Data
calculated is stored in a library to be recalled for all later use.

Under TOOLS:FEA PREFERENCES the user can tell the software where the finite
element results should be stored. If only stored in the current folder, then these
results would only be available for jobs run from this folder. If stored for the user,
on the network, or on the Internet, the user will have have more global access to

© Paulin Research Group – 2018 All Rights Reserved


112
F E A T O O L S ™ U S E R M A N U A L

current models run, or any other models stored in the same place. The FEA
preferences panel is shown below:

Note that this screen also lets the user specify the size of the cache used to store
FEA results. Each beam stiffness model result takes about 5K bytes. When stress
superelements are stored, about 5 Mb are needed for each model. The
user/company must decide how each library will be used and can provide access
and space as necessary. (Stress superelements are not available in FEAToolsV3 at
this time.)

“Translate Only” Option

If the “Translate Only” option is selected on the Input Tab the user will simply get a
confirmation when the translation (to the version selected) is complete and a
summary of system statistics. This will occur almost immediately.

© Paulin Research Group – 2018 All Rights Reserved 113


F E A T O O L S ™ U S E R M A N U A L

In this case, there will be no Chart or Graphical output for review.

Tabular Output for “Translate Only” Option

“07-02”/STP-PT 073/B31J Option

If the “07-02” option is selected on the Input Tab, again, the user will see a
confirmation message when the conversion is complete, as shown below, which
will also appear almost immediately.

© Paulin Research Group – 2018 All Rights Reserved 114


F E A T O O L S ™ U S E R M A N U A L

Using this option, however, the Tabular output under the “Text Output” Tab
contains detailed information on what intersections, bends and nozzles in the
model were updated. The more applicable SIFs and k’s are shown for each unique
intersection, bend, and nozzle and they are compared to existing B31.3, Section III,
and EPRI intersection SIFs and k-factors.

FOR FLAT, SPHERICAL, ELLIPTICAL, AND DISHED VESSEL HEADS –


correlation equations are also available, but these correlation
equations are NOT as well developed as ASME, EPRI and WRC
calculations and should be used with caution. Where the system is
dependent on these values, an FEA analysis should be run to verify
the result.

Important Note: Before going on to the FEA Option, it is worth noting the
key differences between the “07-02”/STP-PT 073/B31J
option and the “FEA” option. In both cases the model
branch connections are transformed using identical
modeling techniques (additional nodes and constraints
added to provide a model intersection that takes
advantage of the new flexibility model being used),
however in the case of the “FEA” option, the i-factors
and k-factors are calculated based on an FEA
evaluation of the component. For the “07-02”/STP-PT
073/B31J option, the i-factors and k-factors are
calculated based on correlations equations from a
variety of sources including tests and finite element
data and are likely NOT as accurate as results from a
finite element analysis.

© Paulin Research Group – 2018 All Rights Reserved 115


F E A T O O L S ™ U S E R M A N U A L

“FEA” Option

If the user selects an FEA translation of the CAESAR II model a lot more
information will be displayed during and after the translation is complete. The FEA
translation will take a minute or two for each unique intersection in the model that
must be analyzed. If the intersection was analyzed previously it will be stored in
the FEA library and reused. If all branch connections in the model have been
previously analyzed on the user’s computer then the FEA translation will occur
almost immediately after the green convert button is pressed.

For the FEA Option, as soon as the user clicks on the “Convert” button, two (2)
“pop up” windows will appear that look like the following windows when there are
FEA geometries that are not in the library:

These windows will show progress on the FEA analysis of individual branch
connections in the model that must be analyzed.

The bottom portion of the main GUI will display the progress of the conversion as
shown below. If the user wishes, the conversion process can be stopped before it
is complete by clicking on the “Stop” button.

While the CAESAR II FEA Translator is converting the standard B31 connections, it
is writing FEA results for each unique connection geometry/configuration to a

© Paulin Research Group – 2018 All Rights Reserved 116


F E A T O O L S ™ U S E R M A N U A L

database for later use. As a result, the first time a CAEASAR II model is converted it
may take 1 to 2 minutes per branch connection. Subsequent runs will re-use these
FEA results and run times will be almost instantaneous.

This is a key feature of the CAESAR II FEA Translator. Most piping work flow
requires numerous reruns of a slightly modified system. Since the FEA branch
results are stored in a library, most runs that include the FEA results are made
instantaneously.

While the FEA analysis and conversion of each standard B31 branch connection is
taking place, the CAESAR II FEA Translator places results in the Output Tabs of the
main User Interface. The user may see the program automatically select the
Graphical Output Tab and display the type of results shown below.

The contents of this Tab will be dynamically updated as the conversion is taking
place. When the conversion is complete, the user will see a confirmation dialog
like the following:

© Paulin Research Group – 2018 All Rights Reserved 117


F E A T O O L S ™ U S E R M A N U A L

At this point the user may select any of the Output data Tabs to review data for
the entire conversion, or for individual branch connections that were updated.

The results will be in tabular, chart, and graphical form. The purpose of this
information is twofold; first, the user can review the data from the translation,
which includes a thorough presentation of new SIFs and flexibilities to determine if
a significant change in the calculated stresses and forces is expected. (If the SIFs
from the FEA analysis agree with B31.3 and the k-factors are all close to 1 then the
updated model will have essentially the same output as the original model.
Second, the user can take any of the tabular or graphical output data and use it in
other reports to document the changes made to the CAESAR II model during the
translation process.

The contents of the Text Output and Chart Output are largely self-explanatory. The
user should, however refer to the later section of this user manual that covers the
theory and detailed technical explanations of the conversion process in order to
aid in the interpret these results. The Text Output tab provides detailed
conversion information for the model and each converted branch connection. The
Chart Output tab provides charts that contain SIF values for both the branch and
header for each branch connection converted.

The tabular output report should be reviewed in detail at least one time to
be sure branch connections are properly interpreted and notes of model
inefficiencies are reviewed.

The Graphical Output tab allows the user to interact with and view each branch
connection that was translated/converted. When the FEA Option is selected and
the conversion is complete, the following controls show up at the bottom of the
main user Interface:

These controls allow the used to pick a specific branch connection AND a specific
SIF (Axial, Inplane, Outplane, Torsional, Pressure). When selected, the contents of

© Paulin Research Group – 2018 All Rights Reserved 118


F E A T O O L S ™ U S E R M A N U A L

the graphical output window are updated to show the 3D Mesh plot with dynamic
stress maps (colored) for the FEA model that was built and analyzed for that
branch connection and specific SIF (or SIFS if “All” is selected).

An example of this output, for a specific Node (branch) and SIF, is shown below for
the out-of-plane loading through the run where the d/D ratio is less than 1. From
the plot, the reader can see that there is little stress concentration in the branch
connection area when the bending moments are applied through the run in the
out-of-plane direction.

The controls on the upper right side of the window allow the user to control the
dynamic behavior of the image.

Clear all results from the Graphical Output Window


Stop “Live” feed for graphical results to window while conversion is
is taking place.
, Turn on or off Animation (stress and displacement) of model

© Paulin Research Group – 2018 All Rights Reserved 119


F E A T O O L S ™ U S E R M A N U A L

Turn on or off dynamic rotation of model


Reset all graphical control settings to program defaults

The user can work with each (or all) converted connection(s) to get a feel for how
large the high stress zone is, where any failure would occur, and why the current
B31.3 Code might be in error. (For out-of-plane loads through the run for
example, 2014 and earlier B31.3 Codes via Appendix D use the i-factor for the out-
of-plane load on the branch for all d/D ratios, producing gross over-conservatism
for run pipe loads when the d/D ratio is small.)

File Storage
In the “preferences” panel at the top of the CAESAR II FEA Translator main
form, or in the NozzlePRO options form, the user will find access and control
for all of the data files used and produced by the translator, NozzlePRO and
the PCL. The options form button to access the FEA Preferences panel is
shown below:

The file storage preference panel is shown below.

To evaluate or clear existing file storage, the user should enter this form, press
on the “Explore” link, and then delete all zip files found in the “Explore” folder,
and the file prgdb5.bin.

© Paulin Research Group – 2018 All Rights Reserved 120


F E A T O O L S ™ U S E R M A N U A L

The standard folder contains precompiled database runs that are delivered
with the software. Users should not remove or otherwise change files in this
folder.

Note: The “Clean” function does not perform any function at this time since
superelements are not installed and used in FEAToolsV3 at this time.

© Paulin Research Group – 2018 All Rights Reserved 121


3
Chapter
F E A T O O L S ™ U S E R M A N U A L

Theory and Discussion of


the CAESAR II FEA
Translator
This chapter provides a theoretical and technical
explanation of the conversion process. The next
Chapters (4 & 5) focus on example uses of the
CAESAR_II FEA Translator.

T he user should by now have a good understanding of how to use the


CAESAR II FEA Translator to successfully convert their standard CAESAR II
model into a more accurate CAESAR II using improved SIFs and k-factors.

Once the user returns to CAESAR II, they can read in and analyze the converted
model and then inspect the analysis result to see what kind of impact the updated
flexibility and SIF factors had on the behavior of the piping system.

This chapter of the user manual is for users who want to gain a better
understanding of what the program is doing “behind the scenes”. It will attempt
to provide the theory and technical information needed to properly explain the
conversion process.

For users that learn better by example, the next chapter focuses on “Discussion
Examples” (Chapter 4) and provides real world models along with technical
discussions of the key improvements made to the model in the conversion
process.

How it Works:

The CAESAR II FEA Translator scans the CAESAR II input file and determines if
there are standard models of B16.9 welding tees, pad reinforced tees,

© Paulin Research Group – 2018 All Rights Reserved 122


F E A T O O L S ™ U S E R M A N U A L

unreinforced tees, sweepolets, extruded tees, or weldolets. It also scans for


bends, and in particular for bends with restraints or other elements applied to
the bend curvature since this will be indicative of a bend with a trunion. If any
of these tee types or bends are properly defined in the piping system the
model of the tee or bend can be improved by the addition of more applicable
branch and run i-factors and by the incorporation of more appropriate k-
factors to develop the proper stiffness of the intersection. The WRC 329
Figure 15 example of a common piping system shows how inclusion of the
proper branch connection flexibility in the piping model can result in more
than an 800% reduction in the calculated bending moment at the tee. (This is
covered in Example #1 in the next chapter.)

When a conversion is complete, the user can take the enhanced CAESAR II
input data file (<name>.C2) and analyze it immediately in CAESAR II, or run the
new (or original) CAESAR II model from the translator screen, and/or compare
the output from any number of CAESAR II models at once from the translator.
Using FEAToolsV3 is much more convenient than the V2 version, since the
CAESAR II model can be converted, run, and the outputs compared by pressing
a single button. If the model is sensitive to the intersection, nozzle, or bend
SIFs or flexibilities, the output results will be different from the original run.
Overall, loads in the system will likely be lower, and displacements in the
vicinity of branch connections will be higher. Depending on the piping
configuration, the increase in displacements may cause pump loads or stresses
to increase. This is considered realistic however, and the analyst should
proceed accordingly. WRC 329 includes a description of a number of
weaknesses in the current B31.3 piping code (2012) that are addressed in the
enhanced intersection model. Some of the noted weaknesses and the
improvements to them include:

1) Consideration of SIF maximum at d/D of approximately 0.7. (Addresses


concern in Note 11 of B31.3 Appendix D.)
2) Computes more applicable torsional SIFs
3) Computes more applicable pressure SIFs
4) Corrects small d/D run SIF error.
5) Eliminates potentially non-conservative effective section modulus.
6) Evaluates SIFs for piping where D/T > 100.
7) Corrects integral weld-on fitting equations (Olets)
8) Replaces k=1 error for all branch connection types.
9) Incorporates rigid element intersection model for branch connections.

B31.3 Appendix D Note 1 states that use of more applicable data is the
“designer’s responsibility,” and that the Appendix D i-factors are to be used in
the absence of more applicable data. WRC 29, other more recent documents,
© Paulin Research Group – 2018 All Rights Reserved 123
F E A T O O L S ™ U S E R M A N U A L

and FEATools™ provides considerably more applicable data automatically in


most common situations.

Bends are the most common component in piping systems that incorporate
flexibilities and current Code bend flexibility factors do not include the length
of the attached pipe, the angle of the bend, any variation in thicknesses,
refractory ovalization stiffening, or the fact that the bend might have an
attached trunion or structural steel member for support. FEAToolsV3 allows
the user to consider these conditions.

More applicable data is most useful when:

• The piping system is in cyclic service


• The piping configuration is looped and tight
• The piping system includes pipe with large D/T ratios
• Pump loads are sensitive to intersections, bends, or nozzles in the
vicinity of the pump
• Stresses are high in the run and d/D for branch connections << 1
• Stresses in olets must be accurately calculated
• Branch connections are fabricated and have d/D ~ 0.75
• Supports are present on the curvature of a bend
• Pipe is connected to vessel heads or shells
• Hillside or lateral configurations are present
• Refractory stiffens bends and tees
• External rings are present on large d/t piping systems
• Systems are glass lined (high stress failure in a glass lined piping system
is often due to a single incorrectly calculated load or stress).

A “properly defined” intersection (pipe branch connection) exists in a CAESAR


II piping model where there are two, colinear header pipes framing into the
intersection point and a single branch pipe. The header pipes must have the
same diameter and wall thickness. Intersections that don’t satisfy these
criteria are not processed. Laterals up to 60 degrees are processed.

Vessel nozzles can include rigid elements to the center of the vessel along with
parts of the vessel model, but don’t need to. Vessel nozzle models can stop at
an anchor or displacement where initial displacements are specified. A number
of vessel anchors are shown in examples in other parts of this document.

A report is printed that details how the model has been changed, and
compares the updated SIF values to values from ASME ST LLC 07-02/STP-PT
073/B31J, and from B31.3.

© Paulin Research Group – 2018 All Rights Reserved 124


F E A T O O L S ™ U S E R M A N U A L

The i-factors and k-factors used in the updated CAESAR II models can be taken
from either a finite element analysis of the intersection or from correlation
equations recommended in WRC 329 and ASME ST-LLC 07-02/STP-PT
073/B31J. The finite element analysis produces five stress intensification
factors for the branch pipe and five stress intensification factors for each run
pipe framing into the intersection. ASME ST LLC 07-02/STP-PT073/B31J
produces three i-factors for the branch pipe and three i-factors for the run
pipe framing into the intersection. Pressure and axial SIFs are not included.
The finite element i-factors are for axial translations, pressure, in-plane, out-
of-plane and torsional moments. The ASME ST-LLC 07-02/STP PT 073/B31J i-
factors are only for in-plane, out-of-plane and torsional moments. CAESAR II
Versions 6.1 and earlier use only ii and io for the B31.3 SE calculation, and
includes input for only Ia and It for the SL calculation. The CAESAR II user must
rely on Code assumptions for other i-factors. Where torsion or pressure cycles
in a piping system, the stress calculations in FE107, NozzlePRO, or PRGiK can
be used to include the effect of each more applicable i-factor.

All later CAESAR II versions allow the user to enter it, and ia for SE, for a more
applicable B31.3 analysis. Changes in the 2014 version of B31.3 resulted in an
alteration of the stress equation nomenclature and the removal of Appendix P,
and so it is unknown at this time how the Code would recommend that
pressure be added to the cyclic stress analysis. (Appendix P in older versions
of B31.3 included pressure explicitly in the discussion. Later versions of B31.3
have not.) [This was written as B31.3 (2016) is about to be publically released.]

© Paulin Research Group – 2018 All Rights Reserved 125


F E A T O O L S ™ U S E R M A N U A L

B31.3 SIF Application Table

ASME ST
SIF LLC 07-
FEA CAESAR II (SE) CAESAR II (SL) PRGiK1
Direction 02/STP-PT-
073/B31J
(Use io for (Use 1.0 or
Axial YES NO
Tees) value entered)
In-Plane YES YES YES (Use 0.75 ii) YES
Out-of-
YES YES YES (Use 0.75 io) YES
Plane
(Use 1.0 or
Torsion YES YES (Uses 1.0) YES
value entered)
(Use io if App.
Pressure YES (Use 1.0) NO
P)
Note (1) PRGiK gives the user a stress calculation screen that can be used to
develop stress intensification factors, AND can be used with calculated values of
ia and ip to calculate to more accurately calculate stresses. Future versions of
FEATools are planned to include the stress superelement formulation used in the
PCL to properly compute the stress throughout the branch connection so that
SRSS combinations of stress maximum values are no longer needed.

Tabular Reports (Text Output Tab)

When an incomplete or improperly defined intersection is encountered in the


CAESAR II model, it is ignored and a note is placed in the intersection list.
During testing of the CAESAR II translator a number of client models were
processed that had numerous incomplete intersection descriptions. Many of
the issues involved errors in geometry or cases where SIFs were placed at
incorrect locations in the vicinity of the tees.

CAESAR II users are cautioned that accurate placement and definition of i-


factors must be undertaken carefully. When an intersection is not defined, the
in-plane orientation is taken as a default orientation with respect to the
element centerline and the Y axis. The in-plane default orientation is found by
crossing the positive element axis into the Y axis. If the element axis is along
the Y axis, then the default in-plane axis is taken to be along the X axis.

At the end of the tabular report, a list of all the encountered intersections is
provided and the evaluation status is given. If the intersections were
processed, the modified node numbering is given along with the restraints and
elements used.

© Paulin Research Group – 2018 All Rights Reserved 126


F E A T O O L S ™ U S E R M A N U A L

Comparisons of FEA results with ASME ST LLC 07-02/STP-PT 073/B31J and


equivalent B31.3 results are provided in each FEA report and give the designer a
sense of the accuracy of the calculation. ST-LLC 07-02/STP-PT 073 was an
alignment project, and as such relies on experiment and previous SIF trends as well
as finite element correlations. The FEA results are based solely on finite element
models of intersections developed following the guidelines provided in EPRI TR-
110996 by Wais and Rodabaugh.

Some key parts of the tabular reports are described below:

At the top of each report is a description of the options and assumptions used
in the development of the models. This section should be reviewed at least
once for each model.

For each branch connection a description of the geometry and type of branch
connection is provided.

© Paulin Research Group – 2018 All Rights Reserved 127


F E A T O O L S ™ U S E R M A N U A L

The basic SIFs, flexibilities and stiffnesses needed to define the intersection are
printed after the input data. The branch and run (header) stiffnesses are
intended to be used simultaneously in the beam model of the piping system
and so effective stiffnesses are printed where they are different from a single
load and component stiffness. The FEA output is given in the example above.
Axial and pressure i-factors are given for both the branch and run sides of the
intersection.

Values less than 1.0 may be printed for flexibility factors or for SIFs to indicate
to the user when the i-factor may be low for a particular branch connection.
Run side i-factors for reduced welding tees can often be less than 1.0 since the
welding tee body can be considerably heavier than the attached nominal pipe.
In all cases though, where i-factors are less than 1.0, a value of 1.0 is used in
the translated CAESAR II model.

For analysts that wish to compare and scrutinize the i- and k-factor results the
following section of the report is provided.

When the piping system will undergo more than 3125 cycles and f=1.2, or
when the piping system will undergo more than 7000 cycles and torsional
loads or pressure stresses are high in a B31.3 analysis, the user is
recommended to use the FEA i-factors and the PRGiK stress evaluation to be
sure to include all fatigue causing mechanisms in the Se calculation.

Where CAESAR II does not include particular i-factors in its stress evaluation,
PRGiK can be used to develop the SIF and compare it to allowables.

© Paulin Research Group – 2018 All Rights Reserved 128


F E A T O O L S ™ U S E R M A N U A L

The above portion of the report is printed for each branch connection and
allows the user to compare SIFs and stiffnesses from the different methods. In
the report above equivalent B31 factors are compared to those from ASME ST
LLC 07-02/STP-PT 073/B31J and from FEA.
The in-plane branch SIF for B31.3 ST LLC 07-02/STP-PT 073/B31J and FEA agree
well: 3.67, 3.96 and 3.535. For the out-of-plane i-factor the FEA SIF is 1.57
times greater than either the B31.3 or ST LLC 07-02/STP-PT 073/B31J SIFs. This
is likely because ST LLC 07-02/B31J correlations were strongly influenced by
other early finite element results and correlations, and the small number of i-
factor tests for size-on-size components in the out-of-plane direction.
The PRGiK spreadsheet, also a part of FEATools™, has an i-factor calculator
that provides pad reinforced i-factors from EPRI TR-110755. Those results are
below:

From this PRGiK evaluation it can be seen that iob from EPRI TR-110755 for this
pad reinforced intersection is 5.89. The results obtained for all of the methods
shown are given in the table below:
out-of-plane
Source i-factor Over B31.3
(branch)
B31.3 4.567 1
ST LLC 07-02 4.567 1
EPRI 110755 5.887 1.287

© Paulin Research Group – 2018 All Rights Reserved 129


F E A T O O L S ™ U S E R M A N U A L

FEA 7.162 1.568


For welded-on integral fittings, called “olets”, a typical design equation was
used to generate the basic geometry of the fitting since thicknesses and tapers
are not explicitly prescribed in any MSS or B31 standards. The dimensions for
typical olet-type geometries are not published by manufacturers.

The model implementation methods incorporated into the PRG CAESAR II FEA
Translator were selected to facilitate incorporating finite element model data
into beam model solutions and to produce consistent solutions over the entire
parameter range of interest. The interaction of large D/T components that
ovalize must be closely evaluated when these components are within about
0.5D1.4T-0.4 of each other in the axial direction along the pipe. FEATools™
converts each well-formed branch connection it encounters in the piping
model.

Errant – atypical olet geometries and taper bored olets are based on tests and
geometries evaluated at PRG. These olet geometries are known to be sold,
but the design geometry is not established by any known testing besides that
done at PRG, and the PRG tests were based on expected design sizes. These
olets should give larger SIFs than any other olet, but are intended to give
realistic, values for the poorest olet geometry available commercially.

For olets in particular, the experience of the welder, the procedure, the
conditions present when the weld was made, any repair required, and the
level of inspection probably have as much do to with the ultimate fatigue
capacity of the olet as the geometry.

© Paulin Research Group – 2018 All Rights Reserved 130


F E A T O O L S ™ U S E R M A N U A L

i-factor Locations:

There are typically three i-factor locations on a branch connection. When d/D
> 0.5 there are four. The three i-factor equations are shown in the figure
below.

When d/D > 0.5 the additional location at the intersection of the branch and
run centerlines is also added to the evaluation and is included in the CAESAR II
model.

When d/D < 0.5, the i-factor is developed at the surface of the run pipe since
this is the location of the crack in a valid Markl-type cyclic bending test. When
d/D > 0.5, the i-factor is developed along the penetration line weld, which may
be closer to the centerline of the run pipe, and so the i-factor location for
branch loads when d/D > 0.5 is taken as the centerline intersection of the run
and branch pipes. This guidance is taken from NB-3683.1(d).

© Paulin Research Group – 2018 All Rights Reserved 131


F E A T O O L S ™ U S E R M A N U A L

Recommendations for CAESAR II Version 6.10 Users


These recommendations are based on the overall experience of the Paulin
Research Group (PRG). PRG’s experience may not apply in any particular case.
It is the user’s responsibility to assure that any recommendations apply to the
particular situation under evaluation. Much industrial fatigue experience has
been obtained from cyclic service where corrosion or chemical damage has not
been effectively quantified. Guidelines provided here apply to essentially non-
corroded pipe.

When the actual number of load cycles including pressure is less than 1000,
D/T < 40, material properties are not time dependent, torsional stresses are
relatively low, and SE is in all cases < 0.8xSA, then approaches used in default
CAESAR II setups are typically adequate for load evaluations involving stress in
the piping system. In some cases the existing B31.3 Codes are over-
conservative, but that jeopardizes costs, and generally not safety.

To appropriately address B31.3 Code requirements, safeguarding might be


required when the number of actual cycles exceeds 3250, when torsional loads
and pressure are significant, when corrosion is present that could affect the
ability of the pipe to undergo small plastic strain in highly stressed areas
without cracking, when the D/T ratio is greater than 50 and there are size-on-
size openings, or openings where d/D is approximately 0.7, and/or when S E >
0.8SA.

Also, in certain situations, (for example when the d/D ratio is small and run
loads are large), B31.3 2012 and earlier Code guidance for run loads can
produce overly-conservative stresses that result in a need to unnecessarily
reroute pipe. In these situations, the designer should recognize that a more
rigorous analysis is required to prevent a wasteful expenditure. The
FEAToolsV3™ CAESAR II FEA Translator properly adjusts i-factors to correct for
these Code errors, as does the guidance in B31J/STP-PT 073.

The B31.3 CAESAR II user is responsible for using adjusted approaches when
evaluating pipe loading conditions where pressure, axial, or torsional loads on
branch connections could contribute to fatigue failure. In this situation a
proper application of Appendix P should be considered when the 2012 and
earlier B31.3 Code is used. The SIFs ii and io for the branch and the run should
be verified using PRGiK, ST-LLC 07-02/STP-PT 073/B31J or finite element
analysis using any general purpose finite element code, NozzlePRO, FEATools,
or FESIF. When the D/T ratio is greater than about 50, elastic FEA analyses is
thought to become increasingly conservative. Bildy and Koves factors printed

© Paulin Research Group – 2018 All Rights Reserved 132


F E A T O O L S ™ U S E R M A N U A L

in the PRGiK spreadsheet give some estimate of the degree of conservatism,


although these were intended to be used for primary loading. The Bildy and
Koves factors were developed from pressure burst test data and may or may
not apply to other loading conditions or failure mechanisms although they
have been used for other load types. (See WRC 497 Part 1.)

When applying B31.3 Appendix P for 2012 and earlier Code years, CAESAR II
makes the assumptions given for the inclusion of pressure in the stress
calculation for SE. These assumptions can result in overly-conservative stress
evaluations and should be used carefully. As the D/T ratio in piping
components exceeds 40, the guidance of 0.75i for fabricated branch
connections and sustained loading likely becomes overly conservative.
Collapse tests have shown internal pressure to increase, reduce or not change
the collapse load due to an externally applied moment. In some specially
fabricated bimetallic or clamped joints, pressure may, however, significantly
reduce the sustained moment capacity. For ferrous branch connections and
elbows this is typically not the case. The use of elastic factors for internal
pressure, even for membrane stresses, to evaluate external loads in large D/T
branch connections can result in overly conservative results. In large D/T
branch connections, buckling is thought to be a consideration when the D/T
ratio exceeds 50-to-60 and in this case, designers might consider providing
sufficient restraint to prevent excessive displacement that might lead to the
local collapse of the pipe due to a loss of the load carrying capacity. (Ref.
NUREG/CR-0261 p.18)

When the user is concerned, a single application of FEATools™ employing the


FEA method is recommended. If stresses or critical loads change by a
significant amount in a critical part of the model some additional consideration
is warranted. The results from the FEATools™ model are considered more
applicable than existing Code calculations.

© Paulin Research Group – 2018 All Rights Reserved 133


F E A T O O L S ™ U S E R M A N U A L

Recommendations for PCLGold

PCLGold for SE:

It can be very difficult to relate one i-factor to another without consideration


of t/T, d/D and D/T ratios. The table below gives the i-factors from a
NozzlePRO finite element analysis for two different intersection types.

Item Do do T t d/D D/T t/T iib iob itb iab ipb


24x24 24 24 0.5 0.5 1 47 1 6.3 9.2 8.0 8.2 6.5
24x4 24 4.5 0.5 0.237 0.18 47 0.474 3.1 5.0 1.0 9.9 7.7
Item Do do T t d/D D/T t/T iir ior itr iar ipr
24x24 24 24 0.5 0.5 1 47 1 4.1 2.0 7.2 6.0 6.5
24x4 24 4.5 0.5 0.237 0.18 47 0.474 1.3 1.0 1.0 1.3 2.9

The SIF ia can be thought to include ip and iax. The recommendation from
Appendix P that ia = io in lieu of more applicable data appears only viable in a
limited number of cases, and so doesn’t seem a reasonable alternative for a
general purpose solution. The ia = io recommendation would be considered
valid if each ratio in the following table was approximately equal to 1.0.

1 2 3 4
Item iab/iob ipb/iob iar/ior ipr/ior
24x24 0.89 0.71 3 3.25
24x4 1.98 1.54 1.3 2.9

Where the ratio in the above table is > 1.0 then using io for ia is not
conservative. Where the ratios in the above table are less than 1.0 then using
io for ia is conservative. When values in the above table are 1.0 then ia = io for
the respective side of the branch connection and the Appendix P guidance is
reasonable.

Conclusions should not be drawn from limited parameter range examples. The
above table suggests that it is difficult to produce reasonable relationships
using ia=io both for pressure and for run i-factors. The issues described above
when ior is overly conservative for d/D<<1 further increases the difficulty. For
this reason, the PCL does not set ia = io as a default and it is not done in
FEATools by default.

© Paulin Research Group – 2018 All Rights Reserved 134


F E A T O O L S ™ U S E R M A N U A L

Users may either enter iax or ip manually in the SIF and K data sheet, or they
may let the FEA processor compute them automatically. If they are not
otherwise entered, values of 1.0 will be used, since iax and ip are not well
defined in the Code.

For B31.3 users, an accurate value of it is probably the most important SIF to
develop and use properly, and for this the user may use either ST-LLC 07-
02/STP-PT 073/B31J or FEA options for branch connections to have these
values computed and entered automatically in the PCL. The torsional stress St
in the 2010 version of B31.3 does not use a SIF. The torsional stress in the
2012 version of B31.3 uses a torsional SIF of one. The 2010 and 2012 versions
of Paragraph 319.4.4 are shown on the following page for reference and
comparison.

The B31.3 guidance is that it = 1.0 in lieu of more applicable data, as described
in 319.3.6. As can be seen in the table above for the 24x24 branch connection
itb and itr can be on the order of 8 and 7 in accordance with FEA calculations.

The comparison table below (taken from the PRGik program) shows a wider
range of values for it, but still values as high as 7.23 for Section III itb.

© Paulin Research Group – 2018 All Rights Reserved 135


F E A T O O L S ™ U S E R M A N U A L

2010 B31.3 Paragraph 319.4.4 -

2012 B31.3 Paragraph 319.4.4 -

© Paulin Research Group – 2018 All Rights Reserved 136


F E A T O O L S ™ U S E R M A N U A L

PCLGold for SL:

Starting with 2010, B31.3 included equations for stresses due to sustained
loads which also included stress index factors for nominal stresses. The
sustained stress failure criteria are collapse, or excessive displacement beyond
what is calculated using an elastic analysis.

© Paulin Research Group – 2018 All Rights Reserved 137


F E A T O O L S ™ U S E R M A N U A L

For cylinders and bends there are equations and test data indicating that the
sustained stress indices can be based on the same general dimensionless
parameters used for stress intensification factors, i.e. d/D, D/T and t/T.

Further guidance to establish the sustained stress indices can be taken from
Section III. A more detailed discussion of the Section III approaches is given in
a separate section below, but to summarize:

• For bends I = 0.963 i


• For UFTs up to d/D=0.5 Ib = 1/3 i Ir=0.57 i
• For Welding Tees Ib = 0.8i Ir = i

For unreinforced tees, PRG has recommended I = i0.5, for branch connections
up to D/T=60, and from D/T=60 to D/T=100 PRG uses an interpolation up to I=i
at D/T=100.

Using the data set from NUREG 5856, PRG now recommends for unreinforced
tees that:

I = (t/T)2/3 (i)1/3 for t/T>1 and,


I = (i)1/3 for t/T ≤ 1.

The PRG rationale is based on the concept that as the i-factor at branch
connections increase, the bending stress increases along with the local extent
of the high stress state due to the curvature at the nozzle-shell penetration.
Each of these effects reduces the overall tendency for the branch connection
to plastically deform in a gross manner and so the sustained stress index (I)
should not increase at the same rate as the stress intensification factor (i) with
(R/T)x.

The test data shows that above D/T of about 60, the exponential affect
diminishes. This is believed to be the range where local buckling begins to
control the collapse mechanism and where the elastic analysis for fatigue is
progressively conservative.

© Paulin Research Group – 2018 All Rights Reserved 138


F E A T O O L S ™ U S E R M A N U A L

The latest sustained stress indices recommended by PRG are plotted with the
NUREG/CR-5856 data below:

Out-of-plane Collapse Moment Correlation

In-plane Collapse Moment Correlation

© Paulin Research Group – 2018 All Rights Reserved 139


F E A T O O L S ™ U S E R M A N U A L

PRG recommends Ib = 0.8i and Ir = 1.0i for contoured tees, including B16.9,
sweepolets, and extruded tees.

In the absence of more applicable data, PRG uses the i-factor reductions
described above for all i-components, i.e. iax, ii, io, ip and it.

The value of 0.75 ii for approximate long radius bends is shown in the chart on
the following page.

B31J recommends the following although data suggests that more applicable
data for heavier wall welding tees could produce even lower exponents, (i.e.
exponents of 0.23 in place of 0.5, but for the time being the following
recommendations are thought reasonable and conservative.)

© Paulin Research Group – 2018 All Rights Reserved 140


F E A T O O L S ™ U S E R M A N U A L

© Paulin Research Group – 2018 All Rights Reserved 141


F E A T O O L S ™ U S E R M A N U A L

Using the 3/20/2013 PRG Correlations with ST-LLC 07-02 i-factor Equations for
size-on-size UFTs:

© Paulin Research Group – 2018 All Rights Reserved 142


F E A T O O L S ™ U S E R M A N U A L

This suggests that a significant percentage of standard size-on-size


unreinforced fabricated tees have within one to one-half the strength of
straight pipe when exposed to in-plane bending.

The same table generated for out-of-plane loads on the branch (below) show
that branch connections are more sensitive to out-of-plane loads for size-on-
size branch connections.

© Paulin Research Group – 2018 All Rights Reserved 143


F E A T O O L S ™ U S E R M A N U A L

The test approach used for most Section III configurations reported by Moore
and Rodabaugh included pinning or fixing both run side ends. One might
expect that a slight non-conservatism would result from these methods, and
that a more “Code-consistent” approach might have been to fix only one end
of the run when loading the branch, leaving the opposite run end free.

For a size-on-size 4” standard wall unreinforced branch connection, the i-


factors for a pinned and free run end are shown in the table below.

ii io
1End Fixed 1 End
2.037 2.143
Pinned
1 End Fixed 1 End
2.669 3.390
Free
Ratio 2.669/2.037=1.31 1.582

The table results suggest that for loads applied through the branch with a
single run end free, for an in-plane sustained stress test, the i-factor from
correlated results should be multiplied by 1.31, and for an out-of-plane
sustained stress test, the i-factor from correlated results should be multiplied
by 1.582.

Collapse for a 4” standard wall carbon steel unreinforced fabricated tee with a
yield strength of 45ksi, would occur at an applied moment of:

Mc = d2 t Sy / (1.31 x 1.288 ) = (4.5-0.237)2 x 0.237 x 45,000 / (1.31 x 1.288) =


114,869 in.lb. (9572 ft.lb.)

For a 22” moment arm the load at the end of the branch would be found from:

F x 22 = 114,869 in.lb.; F = 114,869/22 = 5221 lb.

© Paulin Research Group – 2018 All Rights Reserved 144


F E A T O O L S ™ U S E R M A N U A L

Detailed Discussion of the CAESAR II FEA Translator Input


Options

The following options are available for model translation:

• FEA/ASME 07-02/STP-PT 073/B31J or “Translate Only” radio button.


The user must select the source for the improved intersection data that
will be used in the updated CAESAR II model. The default is FEA. In this
case, all standard intersections, bends, or entered nozzles will be
evaluated using FEA or the best alternate option and updated in the
improved output model. If the model has 20 branch connections, and
only three unique branch connections, the translator will analyze only
the three unique geometries. Depending on the machine speed, the
FEA option generally requires about one minute per unique model
geometry. The ASME 07-02/STP-PT 073/B31J (correlation) conversion
occurs almost instantaneously since this option only requires the use of
correlation equations. If the translate only option is selected, the
output is written out to the selected CAESAR II version and file with no
model translation. A lot of bends of different sizes in the model may
slow the model down. Bends without trunions or steel attachments
take about 20-to-40 seconds per bend to analyzed using hyper degree
of freedom elements. Bends with trunions or steel attachments take
about 60 sec. per bend to analyze.

• The minimum element length text cell allows the user to override the
minimum element length setting used for the translation. The model
translator will not produce a CAESAR II element that is shorter than the
allowed minimum length. This is usually on the order of 0.1 inches,
although smaller or larger values may be entered. The field can be left
blank and the minimum element length will be taken to be 2.54 mm.
(0.1 inch.)

© Paulin Research Group – 2018 All Rights Reserved 145


F E A T O O L S ™ U S E R M A N U A L

• The”Detailed Options” <or> “Default Options” tab includes the


Contoured Tee option (for FEA analysis only), and several additional
default options for SIFs and flexibilities for the various components
evaluated. The EN-10253-A/B tees provide the most control over the
tee types from a computed pressure capacity point-of-view. Type B
tees should provide the minimum size such that the tee satisfies the
burst pressure requirement for the tee, i.e. min wall, min strength
attached pipe, will always fail before any min wall, min strength
welding tee, (providing the welding tee and pipe have the same
minimum strengths.) Type A tees have a fraction of the straight pipe
pressure capacity given by the factor “X”.

Hot formed tees are based on typical hot-formed tee processing and
available material volumes for oversized pipe used to make the tees.

The thicknesses of various B16.9 contoured tees have been measured by PRG
in Houston. In a small percentage of cases, the thickness of the stamped B16.9
welding tees measured throughout the body of the tee was found to be less
than the nominal thickness of the attached straight pipe, (but within 87.5% of
the nominal wall thickness). B16.9 requires that the manufacturers of tees
satisfy the design requirement of B16.9 either by burst test or mathematical
analysis. Most of the very thin walled B16.9 tees measured, (called light tees
above), are in ductile, alloy materials. These tees are typically easy to deform
and have a high material cost, and so it is economically beneficial for the
manufacturer to produce the thinnest tee possible with the specified B16.9
markings.

Thinner tees have higher SIFs, but also higher k-factors, (although i-factors and
k-factors do not increase or decrease proportionately.) It cannot be assumed

© Paulin Research Group – 2018 All Rights Reserved 146


F E A T O O L S ™ U S E R M A N U A L

that because the k-factor is higher, loads on the tee will be lower, and so the
thinner wall is acceptable. The presence of light wall B16.9 stamped tees
introduces a concern that the B16.9 intended margin of safety against
pressure boundary failure (burst) is not provided in the thin tees. Tests of
small, carbon steel tees made in a similar manner to produce thin bodies
indicates that this concern is warranted and that the burst pressure of the
“thin” tees might be as much as 33% lower than the intended B16.9 burst test
pressure.

The FEATools™ Code user is recommended to run Medium tees for most
applications.

Where the system is very sensitive to local stiffnesses, (i.e. where the pipe is
short and stiff and there are a number of tees that make up the configuration,
(for example the B31.3 S303 piping configuration), heavy and light tees can be
run in separate models in addition to the medium tees to make sure that an
acceptable solution is not affected by an unknown tee thickness. Where the
tee thickness has a significant impact on a critical stress or load result then
further auditing is warranted so that tees with the necessary thicknesses are
purchased and installed.

Thin-walled Tee Burst Tests Conducted at PRG in Houston

A reduction of 33% in the burst pressure equates to a real separation between


design and failure of 2 if the safety factor against the tensile strength for the
material allowable is 3. (See B31.3 Para. 302.3.2 (d) (1).) It can be argued that
the B31.3 intended separation between burst and design for B16.9 tees, (given
the B16.9 burst test criteria which is based on the tensile strength of the tee),
is three for B31.3 designs, and 2.4 for VIII-2 Part 5 designs.

© Paulin Research Group – 2018 All Rights Reserved 147


F E A T O O L S ™ U S E R M A N U A L

A B16.9 burst test requires that the test fail at a pressure greater than 2SuT /D,
where Su is the tensile strength of the fitting. If the separation between Su
and the allowable is 3, (B31.3 Para. 302.3.2 (d)(1)), then theoretically, if the
pipe and the fitting are exactly matched in terms of material strength, when
the design pressure produces a stress equal to the allowable, the pressure is
three times removed from the minimum pressure needed to burst the fitting
on the pipe.

MSS SP-97 burst tests (olets) require that the tensile strength of the attached
pipe be used in the burst pressure equation.

The opportunity to enter the axial stress intensity factor (Ia) for sustained
loading first appeared in CAESAR II in the 6.1 version. The axial stress
intensification factor (i), which is different from the axial sustained stress
intensity factor (I), has been in PRG software since 1998, and so the
technology and development procedures used to produce axial and pressure
SIFs for fatigue are well established.

PRG software produces two SIFs for axial stress producers in pipe components.
One is for the pressure, while the other is for the axial external force loading in
the piping system. PRG believes that the axial stress intensification factor in
the B31.3 Code will eventually either be dropped or modified to accommodate
the potentially large difference that can exist between the pressure and axial
load i-factors and the interaction between pressure and external load
sustained stresses. There is no clear relationship between Ia and ia.

PRG will continue to produce separate i-factors for pressure and axial loads.

The CAESAR II translator does not by default insert Ia or It in the 6.1 version of
CAESAR II since these intensity factors are not well defined, and when they
are, (either by plastic analysis or test), it is expected that they would have to
be entered along with Ii, Io, and Ip.

When the user manually uses ia and ip, they are recommended to use ip for ia
since the pressure load is typically far more damaging than the axial external
force.

It is also common to use the pressure i-factor in place of the axial load i-factor
even in the case where the axial external load i-factor is greater. The reasons
for this are:

1) Piping systems almost always have pressure stresses.

© Paulin Research Group – 2018 All Rights Reserved 148


F E A T O O L S ™ U S E R M A N U A L

2) For the first 50 years of B31-type piping analysis, axial external


loads have been ignored in all systems for expansion stress
evaluation. Since pressure almost always exists in an operating
case, and may be a significant contributor to the stress,
selecting a higher i-factor for axial loads that likely do not exist,
and applying the axial load i-factor to pressure – which does
exist, may produce an excessively conservative solution for
larger D/T piping systems, or for pipeline applications where
pressure stresses are permitted to be relatively high.
3) Pressure and axial SIFs account for stresses that have maximum
peak values at different locations in the piping system although
the stress produced by the axial stress in the B31.3 Code adds
directly to the bending stress in the expression for SE. Using
poorly correlated combination techniques can result in
excessive estimates of the stress in a system when multiple
stress components exist. Pressure and axial peak stresses are
not additive because the maximum values from each do not
occur at the same point in the pipe branch connection
geometry.

In summary, it is the PRG opinion that the pressure i-factor is the most
important i-factor to accurately evaluate for a fatigue analysis of a branch
connection, and so the user is recommended to use ip for ia when manual
calculations need to be made to satisfy Appendix P, or similar intentioned
requirements, or when fatigue analysis of the piping system is important.
(Alternatively the user can employ the PRGiK stress calculation portion of the
spreadsheet which gives the user a variety of ways to combine component
stresses into a single stress intensity to be compared against the

The CAESAR II Version 6.1 Ia SIF is for sustained loads only. If the user wishes
to enter the axial sustained stress index, a unique value should be developed.

Options for sustained stress indexes may include 0.75 ia, (ia)0.5, or ia. The
relationship between collapse and fatigue has some common basis but
variations begin to occur around D/T = 50, where local buckling begins to be a
concern. More information on the relationship between I and i is provided
later in this chapter.

Including the pressure (or the axial load) i-factor in piping analyses routinely
will require some adjustment on the part of the user so that errant conditions
are not produced. When the d/t ratio is small, there is not much affect.
When the d/t ratio gets larger, there can be a significant effect, and the user
© Paulin Research Group – 2018 All Rights Reserved 149
F E A T O O L S ™ U S E R M A N U A L

must determine if the Code is providing excessive conservatism, or realistic


results.

CAESAR II allows the user to enter:


1) The fitting outside radius for branch connections
2) The fillet leg length
3) The crotch radius for contoured tees
4) The pad thickness for reinforced fabricated tees or saddles.

These inputs are also used in the translator to adjust SIF and k-factors in the
following way:

1. The fitting outside radius (rp) can be used for unreinforced fabricated
tees or OLETs to set the outside diameter of the nozzle or barrel of the
branch at the point where it contacts the run pipe surface.
2. The fillet leg length can be used for unreinforced, pad reinforced and
olet type junctions. For the OLET type junction the fillet weld length is
the approximate leg length of the olet cover fillet. For pads and UFTs
this is the fillet leg length along the run and branch at the junction and
for pads is the fillet leg length along the edge of the repad.
3. The crotch radius is used for B16.9 tees, extruded tees and sweepolets.
4. The pad thickness is used for pad or saddle reinforced branch
connections. The pad width defaults to ½ the branch diameter.

As mentioned in WRC 329 on page 22 in a similar situation regarding small


bore branch connections which have not yet been addressed by the Code, “…
the Code requirements are obviously silly; the piping analyst should use his
judgment …”. (The small bore branch connection issue mentioned on page 22
of WRC 329 is addressed by both the FEA approach and the ST LLC 07-02/ST-PT
073/B31j approach in FEA Tools modifications of the CAESAR II models.)

© Paulin Research Group – 2018 All Rights Reserved 150


F E A T O O L S ™ U S E R M A N U A L

If pressure and axial loads cycle, the Code user is encouraged to perform a
NozzlePRO, PRGiK evaluation or to perform manual calculations to include
pressure appropriately in the fatigue analysis.

Torsional Stress Intensification Factor:

Also recommended in WRC 329 for B31.3 is the use of the torsional stress
intensification factor, (WRC 329, p.32 Para. 5.0(5).) The Code safety factor
follows the mean failure line by what is intended to be a consistent margin for
ferritic materials (See B31.3 Equation (1c) and notes.) starting when f=1.2,
where f=6N-0.2. (Ref. B31.3 Eq. 1c.). Using B31.3 Eq. 1c, the cycle count where
the consistent margin commences can be found from: 6N-0.2 = 1.2; or when N
= 3125 cycles. When the number of design cycles is greater than 3125 the user
has a roughly two times separation between design stress and mean failure,
where failure is defined as a thru-wall crack at the end of the design life.

As Rodabaugh pointed out in WRC 329, the torsional i-factor can be


approximated using the WRC 329 Eq. 42: it=(r/R)io. Subsequent trials have
shown the relationship to be more complex than the WRC 432 Eq. 42, but Eq.
42 gives the designer an estimate of the error introduced when torsional i-
factors are ignored. The simple example below illustrates the need for an
appropriate torsional i-factor for SE.

The CAESAR II user has the WRC 329 simplified torsional i-factor available in
the CAESAR II control options. When the system cycles, the CAESAR II user is
encouraged to activate this option.

© Paulin Research Group – 2018 All Rights Reserved 151


F E A T O O L S ™ U S E R M A N U A L

Torsional i-factor Example:

Assume io = 0.9(R/T)2/3 and it=(r/R)io as recommended in WRC 329 Eq. 42. If


the torsional stress computed using the 2010 version of B31.3 that does not
include the torsional i-factor in the SE equation is 1/2 of the allowable stress
for a design cycle life of 3200 cycles, for what branch diameter in a 10” x
0.375” wall run pipe produce a real stress that would be expected to have 50%
probability of having a thru-wall crack at the end of the design life?

Answer:

The mean radius for the run pipe is: R = (10.75 – 0.375)/2 = 5.1875 in.

At 3200 cycles the separation between the allowable stress and the mean
failure stress is about 2, so if the i-factor = 4, then a stress that’s one half of
the allowable without an i-factor, would be equal to twice the allowable with
an i-factor: 4 x 0.5 = 2. So in this case, an it=4 is sought.

“it” = 4 = (r/5.1875)(0.9)(5.1875/0.375)2/3; r = 4.

For an 8x10 or larger 0.375” wall branch connection, i.e. a size-on-size


intersection, when the number of cycles is greater than 3125 cycles, and the
stress calculated using the 2010 Version of B31.3 is equal to or greater than
one-half of the allowable, it is expected that at the end of the design life a
thru-wall crack will exist in the component.

The Code user is always recommended to turn on the torsional stress i-factor.
Torsional stress i-factors are relatively easy to calculate, are not too subject to
interpretation, produce high stresses at the toes of fillet welds in locations
where welds are prevalent (somewhat unlike pressure stresses), and can be
easily included in the analysis procedure.

High Stress due to High Stress due to Configuration Inducing


Torsion Out-of-Plane Bending Branch Torsion

© Paulin Research Group – 2018 All Rights Reserved 152


F E A T O O L S ™ U S E R M A N U A L

Torsional loads on branch connections can be produced whenever moments


are developed in the piping system. Whether the moment is torsional or
bending is a function of the orientation of the run pipe with respect to the
branch. (See the rightmost figure above.)

The CAESAR II Version 6.1 input for the torsional I factor is only available for
sustained stresses.

All versions of CAESAR II after 6.1 permit a full suite of SIFs, not including the
pressure SIF. In this case, the pressure SIF should be used in place of ia for
the reasons discussed above.

Whereas this value should likely be entered, one of the largest omissions in
the B31.3 Code prior to 2012 involves the assumption that it=1 for the SE
calculation, and not for sustained stresses.

The default is to use a value of 1.0 for the torsional stress indices for sustained
loads. The sustained stress torsional stress index is only available in CAESAR II
Version 6.1 and later. The B31.3 Version 2012 includes it in the expansion
stress equation, (SE), and so does later versions of CAESAR II.

The intersection attached pipe length selections from the Minutia Options tab
are shown below:

There are four choices for how the attached pipe lengths are to be used for
the translated models.

• 2D Length – default. About two diameters of straight pipe are attached


to each end of the intersection for the FEA modeling and for ST-LLC 07-
02 considerations.
• Widera Lengths. These attached pipe lengths are based on WRC 497,
Widera recommendations.
• Attached Length from the model. These lengths are based on the
element lengths adjacent to the branch connection of interest in the
actual CAESAR II piping model. If the attached length option is
selected, users can enter short lengths in the CAESAR II model adjacent
to branch connections and have them considered in the branch
connection i- and k-factor calculation. (Short attached pipe lengths

© Paulin Research Group – 2018 All Rights Reserved 153


F E A T O O L S ™ U S E R M A N U A L

affect both ST-LLC 07-02 and the FEA k-factor calculation at the
intersection. “Short” in an ST-LLC 07-02 sense, is 0.1D1.4T-0.4. )
• PRG Attached length. This length is taken as 0.5D1.4/T0.4.

The 2D length option for branch connections provides an effective, efficient


running FEA model and is recommended.

It is recognized that selecting the length from the C2 model is probably more
realistic for attached pipe. Finding the attached straight pipe is used as the
default for bends, since the attached pipe can have such a strong influence on
the flexibility of bends.

Widera showed that boundary condition lengths may need to be larger than
2D to remove the boundary completely from the branch connection stress
influence. This occurs when D/T is large however, and in this case the elastic
analysis tends to be conservative and so selecting shorter lengths tends to
adjust for the higher D/T conservatism in the elastic analysis.

The 2D length was used often prior to 2000, and appears in the following note
to Appendix D-300 in B31.3:

Many of the original tests on branch connections and bends were conducted
on smaller diameter fittings. D/T ratios for the tests were relatively small and a
2D removed boundary was adequate. When Widera began running finite
element models using larger D/T and d/D models, it became clear that when
the D/T ratio gets larger, an L=2D criteria for boundary condition removal is
not adequate to find the highest local elastic stress. The stress due to external
loads is local and located in the cat’s-eye region of the circumferential plane
for out-of-plane loadings.

It is often thought that beyond D/T=50, concern for local buckling in certain
geometries also exists. Buckling shapes are developed in bends and
intersection by default in the elastic deflection, and so elastic buckling due to
external loading is seldom a concern in elbows or branch connections where
there is always zero or internal pressure.

Buckling is a different matter in straight pipe when the d/t ratio is greater than
50. In these cases, in the situation where the uniform moment is constant
along a significant section of the pipe, ovalization in that section can occur,
and local collapse of the cross section can occur. As of 2016, much elastic

© Paulin Research Group – 2018 All Rights Reserved 154


F E A T O O L S ™ U S E R M A N U A L

buckling testing of this type on common carbon and stainless steels with four
point bending setup has not been conducted.

The last checkbox in the “Minutia” tab determines how restraint book-keeping
is performed when the new CAESAR II model is generated.

After a valid intersection in the CAESAR II model is encountered it is broken


down as shown in the figure below in accordance with the guidance provided
in ST-LLC 07-02/STP-PT 073/B31J, NB-3686.4 and NB-3686.5.

As part of the translation process, six restraints with Connecting Nodes must
be placed at three different points on each branch connection to describe the
local flexibility of the connection. In-plane, and torsional connections are the
most important stiffnesses for the header when d/D > 0.5, and in-plane, out-
of-plane and torsional connections are important for the branch.

Because there are a large number of restraints required for each intersection,
they are difficult to review manually. Typically, some are selected at random
and checks are made to be sure that the stiffnesses are applied to the model
correctly. The CAESAR II model input and the text output from the translator
describes in detail how each intersection was modified to assist with any
checking.

A stiff, or rigid, element is inserted between the intersection of the branch and
run centerlines and the surface of the run pipe. The local branch flexibility
occurs at this point in any intersection geometry. This relationship is shown
graphically in the figure below.

© Paulin Research Group – 2018 All Rights Reserved 155


F E A T O O L S ™ U S E R M A N U A L

Activation of the “Put Branch Connection Restraints at End” checkbox places


all the restraints for each branch connection in the model at the end of the
input. When the added restraints are placed at the end of the input they do
not clutter up the restraint list so that other restraints in the model can be
easily checked.

In some cases, there are not a sufficient number of elements to contain all the
restraints needed for each intersection in the model. When this happens,
additional short, rigid elements with zero weight are added to the end of the
element list. These zero weight rigid elements are connected to a single
intersection in the model and have no effect on static or dynamic results.

© Paulin Research Group – 2018 All Rights Reserved 156


F E A T O O L S ™ U S E R M A N U A L

ASME Section III Guidance for Sustained Stress Indices

The objective in this section is to derive sustained stress factors used in


Section III for comparison with the recommendations in B31 and those by PRG.

Section III NB-3652 Equation 9 can be loosely rewritten as:

B1 (PD/2T) + B2 (M/Z) < 1.5Sm

B1 would be used for the sustained stress index pressure term and B2 is used
for the sustained stress index bending term.

If Sm is approximately equal to Sh, then,

B1 (PD/2T)/1.5 + B2 (M/Z)/1.5 < Sh

If an expression can be found such that B2/1.5 = C0 i, then C0 then is the


multiplier of the i-factor for sustained load conditions.

For elbows in NB-3683.7;

B2 = 1.3 / h2/3, but not less than 1.

In the B31.3 Code, ii for elbows is 0.9 / h2/3. The h factor for NB-3683.7 is the
same as the h-factor for B31.3.

B2/1.5 = 1.3/1.5 / h2/3 = C0 i = C0 x 0.9 / h2/3.

Cancelling the h2/3, leaves

C0 = 1.3/1.5/0.9 = 0.963.

To match the Section III NB-3683.7 evaluation, (when Sm=Sh), the constant on
the i-factor would have to be 0.963.

© Paulin Research Group – 2018 All Rights Reserved 157


F E A T O O L S ™ U S E R M A N U A L

For fabricated branch connections in NB3683.8, B2b and B2r are given as:

B2b = 0.5C2b, and B2r = 0.75C2r.

If i for branch connections is equal to i = C2K2/2, and K2b for the fabricated
branch connection of interest is 2 and K2r is 1.75 per NB-3683.8(d), then a
similar equality can be set up for fabricated branch connections to find C0b and
C0r.

For the branch side:

B2b/1.5 = C0b i;
B2b/1.5 = C0b x C2b K2b / 2
= C0b x C2b x 2 / 2
= C0b x C2b

Replacing B2b with 0.5C2b;

0.5C2b/1.5 = C0b x C2b.

Dividing by C2b;

C0b = 0.5/1.5 = 1/3 = 0.333

For the run side:

B2r/1.5 = C0r i;
B2r/1.5 = C0r x C2r K2r / 2
= C0r x C2r x 1.75/2
= 0.875 x C0r x C2r

Replacing B2r with 0.75C2r:

B2r/1.5 = 0.75C2r/1.5 = 0.875 x C0r x C2r.

Dividing by C2r;

C0r = 0.75/1.5/0.875 = 0.57

© Paulin Research Group – 2018 All Rights Reserved 158


F E A T O O L S ™ U S E R M A N U A L

For welding tees in NB-3683.9(a),


B2b = 0.4 (R/T)2/3, and B2r = 0.5(R/T)2/3.
In (b) and (c),
C2b = C2r = 0.67(R/T)2/3, and
K2b=K2r = 1.0.
For the branch side:
C2b x K2b / 2 = 0.67(R/T)2/3 x 1.0 / 2
= 0.9/h2/3
where h=4.4T/R
So
B2b / 1.5 = 0.4/1.5(R/T)2/3
= C0b x 0.335(R/T)2/3.
Dividing by (R/T)2/3,
C0b = 0.4/1.5/0.335 = 0.796
For the run side:
0.5/1.5(R/T)2/3 = C0r x 0.335(R/T)2/3
Dividing by (R/T)2/3,
C0r = 0.5/1.5/0.335 = 0.995
Using Section III NB-3600 as a reference, sustained stress indices can be
developed as above, and are summarized as follows:

Component i-factor
Constant
Elbow 0.963
Fabricated Tee Branch
0.3331
Side
Fabricated Tee Run Side 0.571
Welding Tee Branch Side 0.796
Welding Tee Run Side 0.995
1 for d/D < 0.5

Note: Additional theory and recommendations for use of the CAESAR II FEA
Translator, specifically as it applies to small branch connections on large
headers, is provided in Discussion Example number 5 in the next chapter.

© Paulin Research Group – 2018 All Rights Reserved 159


4
Chapter
F E A T O O L S ™ U S E R M A N U A L

CAESAR II FEA Translator


Discussion Examples
This chapter focuses on “Discussion Examples”.
Example models are presented, converted using the
CAESAR II FEA Translator, and the results discussed.
The next chapter (Chapter 5) provides details on a
number of example model files that are distributed
with FEATools™.

T his chapter provides five discussion examples for the user to gain a better
understanding of different piping scenarios that can benefit from the use of
the CAESAR II FEA Translator. The hope is that the user will be able to
anticipate commonly found situations in a piping design/system where conversion
using the CAESAR II FEA Translator will make a significant change to the model and
the results.

The first example provides a detailed summary of the conversion process and
steps. The five discussion examples included in this chapter are summarized
below.

• S303 Example (also covered in detail in Chapter 5)

• Heater Piping Example

• Schneider 1978 from “Overview of Structural Design of Piping Systems”

• Figure 15 Example in WRC 329 (also covered in detail in Chapter 5)

• Small Branch Connection Example (with recommendations)

© Paulin Research Group – 2018 All Rights Reserved 160


F E A T O O L S ™ U S E R M A N U A L

Discussion Example # 1: S303 Example

The S303 example can be found in Appendix S of B31.3 2010 and 2012. The
model describes a piping system that experiences two thermal states. Each
thermal state heats and cools opposite legs of the symmetric, looped piping
system producing reversing thermal bending moments. The maximum stress
range occurs due to the different signed moments produced by each
temperature case. The S303.C2 file is installed with FEATools™. This model is
represented in the graphics below.

© Paulin Research Group – 2018 All Rights Reserved 161


F E A T O O L S ™ U S E R M A N U A L

When the model is analyzed, the CAESAR II the output shows an overstressed
node when the cycle range is from one thermal state to another. The analyst
recognizes a tight piping configuration without inherent flexibility. All the
straight sections contain valves, and there are no bends. The overstressed
report from CAESAR II is shown below.

© Paulin Research Group – 2018 All Rights Reserved 162


F E A T O O L S ™ U S E R M A N U A L

Since this overstressed area is due to the artificial k-factor = 1 in the B31.3
Appendix D, a more applicable k-factor can be generated using the FEATools™
program. FEATools™ can be started from CAESAR II, or from the desktop icon.
The only necessary input is the job name. The input job name and folder is
identified automatically if started from CAESAR II. There are a variety of
options, but the defaults accommodate most situations. In this case, the user
must only verify that the job name is correct and then press the Convert
button.

FEATools™ CAESAR II FEA Translator Input Screen

Notice that the suffix –FEA is appended to the CAESAR II job name. When the
translator completes, the name S303-FEA.C2 will exist in the working folder.
This CAESAR II input file will contain more applicable models for all
intersections in the job. Each intersection replaced will be described in the
text output, printed by the CAESAR II FEA Translator.

During the run, a series of run status dialogs are displayed. When the first
intersection has been processed, the graphical output for that intersection is
displayed. Examples of the status dialogs for S303 are given below.

The dialog on the left shows the progress for all nodes/branches for which FEA
models are being created and analyzed. The dialog on the right shows the
progress for a specific node/branch.

© Paulin Research Group – 2018 All Rights Reserved 163


F E A T O O L S ™ U S E R M A N U A L

When a single FEA model for S303 is available, the translator will open the
graphics, display mode and the screen will appear as shown below.

Since there are 5 SIFs for the branch, and 5 SIFs for the header/run, the plots
will appear in groups of 5. The label for the large model featured in the
middle of the screen appears on the bottom center of the screen. The axial,
in-plane, out-of-plane, and torsional SIFs are shown in smaller models around
the larger pressure SIF model in the middle in the screen image above. The
user can click on any of the smaller, surrounding models to bring them to the
center of the screen. The buttons on the upper-right corner of the viewing
screen can be used to manipulate the models.

When there are multiple intersections in a single model, the drop down box in
the bottom right corner of the screen can be used to select an individual node
and SIF direction for review.

Models are added to the list as they are processed. When all the models are
processed, the CAESAR II input file is translated, and then the “translation
successful” dialog is shown on the screen:

© Paulin Research Group – 2018 All Rights Reserved 164


F E A T O O L S ™ U S E R M A N U A L

When the success dialog is cleared, the user will notice that there are four
tabs in the upper left corner of the screen.

The first tab: “Input” contains the input file names and options for performing
the conversion. The “Chart Output” tab contains the charted results for each
intersection processed. An example charted output tab with results is shown
below: Charted output for each intersection is produced as soon as the finite
element run for the intersection is completed.

The tabular output contains applicable notes and SIF and k-output for each
branch connection. At the bottom of the tabular results is the list of node
numbers and how they were altered (new nodes added to the model) to
incorporate the improved branch connection model. An example of the
Tabular (“Text Output”) data generated for the S303 Example is given in the
following pages.

Note: This discussion example is also covered in detail in the next chapter
that discusses example model files distributed with FEATools™.

© Paulin Research Group – 2018 All Rights Reserved 165


F E A T O O L S ™ U S E R M A N U A L

Text Output for S303 Example

PRG - CAESAR II - Enhanced Branch Connection Modeling

Replace rigid branch connections with


flexible models based on FEA.

Attached lengths based on approx. two diameters of straight


pipe on each end of the branch connection.

Intersection Node = 20 Type = Welding Tee (B16.9 or Equiv)

Run/Header Outside Diameter = 24.000


Branch Outside Diameter = 24.000
Run/Header Thickness = 0.375
Branch Thickness = 0.375

Branch Flexibility Factors (kax, kin, kout, ktor)


0.837 8.234 18.657 19.267

Header Flexibility Factors (kax, kin, kout, ktor)


1.818 5.691 0.000 14.547

Advanced Branch/Nozzle SIFs:


(ii)= 4.701 (io)= 8.093 (it)= 7.116 (ia)= 6.058 (ip)= 5.236

Advanced Run/Header SIFs:


(ii)= 4.066 (io)= 1.769 (it)= 6.339 (ia)= 5.723 (ip)= 5.236

BRANCH SIFs RUN/HEADER SIFs


B31.3 STLLC07-02 FEA B31.3 STLLC07-02 FEA
In 3.417 3.291 4.701 3.417 3.278 4.066
Out 4.222 4.189 8.093 4.222 1.659 1.769
Tors 1.000 4.189 7.116 1.000 3.391 6.339
Axial - - 6.058 - - 5.723
Press - - 5.236 - - 5.236

BRANCH Stiffnesses RUN/HEADER Stiffnesses


B31.3 STLLC07-02 FEA B31.3 STLLC07-02 FEA
Axial RIGID RIGID RIGID RIGID RIGID 17370296.
In RIGID 15623077. 7858626. RIGID 29768204. 14882372.
Out RIGID 11451394. 3720215. RIGID RIGID RIGID
Tors RIGID 18570542. 2197982. RIGID 45826952. 5822506.

Intersection Node = 40

Is identical to the intersection at node 20.


Please see results written for that intersection.

Number of Intersections Detected = 6

Where i-factors are less than 1.0 a value of 1.0 is used.


Where k-factors are less than 1.0, a rigid stiffness is
for that branch stiffness for that direction.

Intersection Els Brch Branch Run/Header


Node Type Attch Angle OD Thk OD Thk STATUS

20 B16.9 or Equiv 3 90 24.000 0.375 24.000 0.375 Processed OK


30 B16.9 or Equiv 3 90 20.000 0.375 24.000 0.375 Processed OK
40 B16.9 or Equiv 3 90 20.000 0.375 24.000 0.375 Processed OK
340 B16.9 or Equiv 3 90 20.000 0.375 24.000 0.375 Processed OK

© Paulin Research Group – 2018 All Rights Reserved 166


F E A T O O L S ™ U S E R M A N U A L

Restraint Nodes RUN #1 RUN #2 BRANCH


Node Type Node CNode Node CNode Node CNode
20 B16.9 or Equiv 21 20 22 20 23 24
30 B16.9 or Equiv 31 30 32 30 33 34
40 B16.9 or Equiv 41 40 42 40 43 44
340 B16.9 or Equiv 341 340 342 340 343 344

Modified Junction Elements


At At
Junction /----- Original Junc ----/ /--------- New Junc ------/

20 20 -To- 31 -> 20 21 -To- 31 -> 21


20 -To- 41 -> 20 22 -To- 41 -> 22
10 -To- 20 -> 20 23 -To- 24 -> 24

30 21 -To- 30 -> 30 21 -To- 31 -> 31


30 -To- 35 -> 30 32 -To- 35 -> 32
30 -To- 210 -> 30 34 -To- 33 -> 34

40 22 -To- 40 -> 40 22 -To- 41 -> 41


40 -To- 45 -> 40 42 -To- 45 -> 42
40 -To- 110 -> 40 44 -To- 43 -> 44

340 340 -To- 345 -> 340 341 -To- 345 -> 341
340 -To- 321 -> 340 342 -To- 321 -> 342
140 -To- 340 -> 340 343 -To- 344 -> 344

-- End of Sample Report

Once the translation is completed the CAESAR II model S303-FEA.C2 is


available for analysis. The FEA models are stored and so any future translation
that needs these models will retrieve them from the library cache.

When the improved CAESAR II model is run, the moments and stresses drop
approximately 70%, and the model is no longer overstressed. The stress
results for the improved CAESAR II model are shown below.

© Paulin Research Group – 2018 All Rights Reserved 167


F E A T O O L S ™ U S E R M A N U A L

A schematic of the solution approach is shown in the diagram below.

© Paulin Research Group – 2018 All Rights Reserved 168


F E A T O O L S ™ U S E R M A N U A L

Discussion Example # 2 Heater Piping Example


The following piping system was analyzed using the k=1 default branch-
connection stiffnesses. The piping is relatively hot with D/T ratios around 40.

The standard CAESAR II model was run through the CAESAR II FEA Translator
to produce more applicable flexibility and stress models of the piping system.
As can be seen in the table below, some of the loads were reduced by 281%
while other loads were increased by 242%.

When the model geometry and results are reviewed, additional flexibility is
provided at 15 and 25. Higher displacements occur at node 20, and this will
increase loads at 30, and may increase the loads at 75 depending on the
interaction of the displacement and increased flexibility.

© Paulin Research Group – 2018 All Rights Reserved 169


F E A T O O L S ™ U S E R M A N U A L

Discussion Example # 3 Schneider 1978 Example

In 1978 R.W. Schneider recognized the potential effect of branch connection


flexibilities on stresses and developed the following example. Schneider’s
1978 branch connection flexibility model is identical to the analytical model
recommended in ASME ST LLC 07-02 and is similar to the model used in ASME
Section III NB3200 for branch flexibilities.

Schneider shows how in this simple single intersection piping system, the load
distribution in the three legs of the branch connection changed when
flexibilities were included in the analysis. One load was reduced by 41% while
the branch load went up 21%.

© Paulin Research Group – 2018 All Rights Reserved 170


F E A T O O L S ™ U S E R M A N U A L

Discussion Example # 4 Figure 15 Example in WRC 329

Everett Rodabaugh provides the following example in Section 4.9 (Figure 1) of WRC 329:

30" x 0.375"

240"
Unreinforced
Fabricated
Tee 10
12.75" x 0.375"
Y
Without considering the branch
connection flexibility of the X Tambient=76F 120"
Z
12.75”x30” fabricated tee at node Thot=500F
15, the out-of-plane (Z) bending
moment at node 15 is 29,193 ft.lb. Anchor 5
Including the branch connection
flexibility reduces this bending
moment to 3,141 ft.lb., nearly
1/10th the original moment.

When the CAESAR II model of this piping system is translated using the FEA
option and the model is brought back into CAESAR II, the MZ loads on the
junction at node 15 drop by almost 90% (to nearly 1/10th the originally
calculated moment.

Stated another way: the model that does NOT take into account the branch
connection flexibilities calculates an MZ Moment that is ~9.3 times (930%)
larger than the calculated MZ moment you get when the branch connection
flexibility is used in a more applicable model!

WRC 329 points out that such large reductions are not typical, but that large
changes in piping system behavior can be experienced when appropriate
flexibilities are included in the models.

Note: This example is also covered in detail in the next chapter that covers
example model files distributed with FEATools™.

© Paulin Research Group – 2018 All Rights Reserved 171


F E A T O O L S ™ U S E R M A N U A L

Discussion Example # 5 Small Branch Connection Example

An excerpt from WRC 329, page 22, regarding the small d/D issue is given
below.

For small branch connections the B31.3 Code gives a reduced i-factor for the
branch, but uses a size-on-size i-factor for the run pipe regardless of the size of
the branch. The figure shown above illustrates the potential problem with this
approach. The issue is discussed in more detail on page 22 of WRC 329, and in
NUREG/CR-0261 July, 1978.

© Paulin Research Group – 2018 All Rights Reserved 172


F E A T O O L S ™ U S E R M A N U A L

The out-of-plane i-factor for all fabricated, unreinforced branch connections is


given by the equation i=0.9/h2/3, where h = T/r. This results in an out-of-plane,
run i-factor for all d/D unreinforced branches of 0.9(R/T)2/3; where R is the
mean radius of the run pipe and T is the thickness of the run pipe.

For unreinforced branch connections, the branch diameter or thickness is not


used in the run pipe i-factor equation in the 2010, 2012 or earlier versions of
B31.3 Appendix D.

This B31.3 Code guidance provides that a 10,000 ft.lb. out-of-plane run
moment on a 24x24” size-on-size intersection produces the same stress as a
10,000 ft.lb. out-of-plane run moment on a 24x4” intersection. (Ref.
NUREG/CR-0261 p.58.)

The figure below shows the out-of-plane run moments in red.

Meshed Model of 24” x 24” & Shaded Image of 24” x 24” &
24” x 4” Branch Connection 24” x 4” Branch Connection

Finite element results, ASME Section III NB/NC 3600 correlations, and
guidance from ASME ST LLC 07-02 (see below) suggest that the stresses in the
size-on-size, 24”x24” intersection are three to six times larger than the stress
in the 4”x24” intersection depending on the run pipe wall thicknesses. This
introduces a potentially over conservative issue with the 2012 version of
B31.3. Equation 17 in Paragraph 319.4.4 in the 2012 version incorporates a
portion of the P17 Equations.

Equation 17 in the 2012 version of B31.3 is given below with its nomenclature.

© Paulin Research Group – 2018 All Rights Reserved 173


F E A T O O L S ™ U S E R M A N U A L

The axial SIF ia will now be taken to be equal to io when listed in Appendix D
for branch connections, and so for the 2012 version, for small bore branch
connections and for axial loads in run pipe, where there was no axial stress
included explicitly in earlier versions of the base Code, the axial stress is now
explicitly included per Eq. 17, and is multiplied by a potentially excessive out-
of-plane run stress intensification factor.

For all d/D branch connections, the ia per Eq. 17 is given in the table below.

© Paulin Research Group – 2018 All Rights Reserved 174


F E A T O O L S ™ U S E R M A N U A L

It can be seen that for a 24” Schedule 10 pipe, with a 4” branch, the axial
nominal stress in the 24” run pipe will be multiplied by an ia = io = 11.8. For a
24” schedule 10 pipe with any sized unreinforced branch, the axial nominal
stress will be increased by more than an order of magnitude when Eq. 17 in
the 2012 version of B31.3 is used and ia = io per the note to Eq. 17 or more
applicable data is employed.

After reading the notes in the 2012 version of B31.3 it can easily be
understood that Eq. 17 should not include stresses due to pressure in the Sa
calculation in Equation 17, although this seems contradictory to guidance in
B31.3 Paragraph 301.10, and considering:

1. Guidance in Appendix P would suggest that it should since Appendix P


explicitly includes pressure in Eq. P17a and b.
2. Guidance given in NB-3653.2 would suggest that it should. (See NB-
3653.2 and P17 Equations below.)
3. Pressure tends to cycle more than temperature in most piping systems
where thermal cycling is a concern.
4. Pressure will more likely exist in a piping system when external
moments won’t.

© Paulin Research Group – 2018 All Rights Reserved 175


F E A T O O L S ™ U S E R M A N U A L

Pressure cycling can be considered less important if it is assumed that ip = io,


and that the nominal maximum pressure stress PD/2T is allowed to be equal
to the nominal maximum out-of-plane bending stress Mo/Z, since the peak
pressure stress in this case would always be half the peak external load stress,
i.e.

(ip)(PD/4T)max allowed = half (io)(Mo/Z)max allowed


The Appendix P guidance when pressure is included in Equation P17a or P17b
should be used with care since large pressure SIFs in excess of two, (ia=ip=io)
for small d/D pipe, suggest that normal pressure design for small d/D
intersections does not satisfy the Appendix P allowables even when the
external loads are zero. One could argue that when pressure cycles, the B31.3

© Paulin Research Group – 2018 All Rights Reserved 176


F E A T O O L S ™ U S E R M A N U A L

Code user must consider it elsewhere than Equation 17, even though the
Appendix P Equation 17 user does consider it. The following example
illustrates:

Example: A 2” Schedule 160 branch stub in, is welded to a 24x0.375” run


pipe to be used as a pressure tap. Since the line is filled with a
hydrocarbon, a client standard requires that the branch be
analyzed for flexibility per Para 319.4. Sc=Sh = 20,000 psi. The
operating pressure for the run pipe is 450 psi. The pressure
term for So or Se for Equation P17a or P17b is iaFa/Ap; which for
pressure will be ia=io, and Fa/Ap=PD/4T, and so the pressure
stress will be:

(0.9)(R/T)2/3 (PD)/(4T)

= (0.9)(31.5)2/3 (450 x 24)/(4 x 0.375)


= 8.977 x 7200
= 64,632 psi.

If the allowable is from P1b, then

SEA = 1.25 f (Sc+Sh),


which might be: (1.25)(1.2)(20,000+20,000)
= 60,000 psi.

The spreadsheet below from the PRGiK program (included with FEATools™)
illustrates how the io value for the run pipe in a small d/D intersection can be
large when compared to other recommended values. The situation is worse
for unreinforced fabricated tees (UFT’s), but also exists to a lesser extent for
other branch connection types.

© Paulin Research Group – 2018 All Rights Reserved 177


F E A T O O L S ™ U S E R M A N U A L

The potential over-conservatism of the run io factor for B31.3 is shown in the
following table as a comparison against other references.

Run B31.3 ior Compared to other References


Reference iref / iB31 Ratio
ASME Section III 4.27
FEA SIF 8.976
ST-LLC 07-02 8.599

WRC 329 on page 22 states that “… in effect, the Code requirements are
obviously silly; the piping analyst should use his judgment and ignore the vent
or drain line in his design analysis.” WRC 329 used vent and drain lines to
illustrate the small d/D problem, but the ior problem exists for many d/D
branch connections < 0.5.

Refer to the spreadsheet on the following page that includes data from EPRI
TR-110996 for the 10x24” branch connection.

© Paulin Research Group – 2018 All Rights Reserved 178


F E A T O O L S ™ U S E R M A N U A L

The Section III values must be used with care when comparing to directionally
dominant SIFs. Notice the comparison in the table below:

Reference iir ior Notes


ST-LLC 07-02 3.1925 1.00
B31.3 6.9825 8.977
Largest i-factor used
Section III 3.5075 3.5075
for all directions
EPRI TR-110996 1.811 1.000

It is thought that small bore branches were not often included in the analysis
of large bore pipe in the past, but the improvement in computing capacity in
the last decades has facilitated the increase in model size, which resulted in
the inclusion of small branch pipes on larger headers pipe stress models.
B31.3 Para 322.3.2(b) can be assumed to instruct the analyst to include the
branch in the analysis, since it
is recognized in the industry
that small bore branch
connections do often fail. It is
hoped by the industry that
where failures occur,
additional analysis will
eliminate the problem.

© Paulin Research Group – 2018 All Rights Reserved 179


5
Chapter
F E A T O O L S ™ U S E R M A N U A L

CAESAR II FEA Translator Example Model Files &


Discussion
This chapter presents detailed discussion of the
Example model files distributed with FEATools™.
These example model files can be converted using the
CAESAR II FEA Translator program. The user can verify
the before and after results for themselves and gain a
better understanding of both the conversion process
and the updated model results.

T en example problems are delivered with the FEATools™ software. The


models are provided in a CAESAR II Version 6.1 C2-file format. In this
chapter, detailed discussion and program output for each of the example
models is provided.

Note: All of the example model files can be found in the “Public\Documents”
folder on the computer on which the software was installed. The
location of this folder may differ slightly depending on the version of
the operating system. For Windows 7 and Windows 8, the location is
likely the following:
installation_drive:\Users\Public\Documents\PRG
For Windows XP users this will likely be:
Installation_drive:\documents and settings\all users\documents\PRG

© Paulin Research Group – 2018 All Rights Reserved 180


F E A T O O L S ™ U S E R M A N U A L

Brief Descriptions of the Example Model Files


1) WRC329.C2 – Figure 15, WRC 329 example model showing a moment
reduction of almost 90% when branch connection flexibilities are included
in the piping system.
2) S303.C2 – Multiple thermal condition model from B31.3 Appendix S, Para
303. This very tight, looped piping system is an excellent candidate for
more applicable branch connection models of B16.9 welding tees. Before
the translation, the system shows to be overstressed when cycling from
one thermal condition to another. After translation, the same thermal
stress range shows to be less than 20% of the B31.3 Code allowable. The
system moments are controlled by the branch connection flexibilities.
3) bypassB13.C2 – Static slug load simulation that shows how including
flexibilities at branch connections does not always reduce stresses. Added
flexibility permits additional displacement due to slug loading, and smaller
diameter, attached nozzles are more heavily loaded.
4) BRG401.C2 – Small bore pipeline branch connection model that
demonstrates the inaccuracy in typical modeling approaches for small bore
branch connections when little external flexibility is provided. The
example satisfies B31.3 Code compliance when a typical model is used, but
is overstressed when more applicable modeling is employed.
5) BRG402.C2 – Small bore pipeline branch connection model that
demonstrates the B31.3 Code error highlighted by WRC 329 on p.22. The
excessive conservatism in the B31.3 calculation for the out-of-plane run
stress intensification factor is illustrated. WRC 329 cautions that the
designer must exercise judgment in this situation. The typical model
shows the run pipe to be overstressed, while more applicable i-factors
shows the run pipe stresses are satisfactory.
6) ManifConn.C2 – Small inter-manifold piping system shows how loads can
be redistributed when more accurate flexibility factors are included in the
piping system.
7) JWILCOX2.C2 – Model recommended by Mr. Jim Wilcox of CodeCad. This
intellectually challenging model illustrates how the number of load
combinations possible in a piping system can increase exponentially with
the number of operating conditions, and shows that including proper out-
of-plane i-factors for d/D<1 branch connections can significantly reduce
stresses. Load cases before translation exceed allowables, while after
translation, loads and i-factors are reduced, and each load case satisfies
the Code allowable.

© Paulin Research Group – 2018 All Rights Reserved


181
F E A T O O L S ™ U S E R M A N U A L

8) pumpV10.C2 – A simple two pump piping system demonstrates how the


translation process is not limited to D/T<100 and can be used for nozzles in
cylindrical shells to provide more appropriate boundary conditions for load
sensitive equipment. Loads on the simple piping system are reduced by
30% when more accurate flexibilities are included in the model. To be
used in conjunction with pumpV10RG.C2 for comparison and analysis of
different modeling techniques for systems of this type.
9) Outplane.C2 – Simple model designed to demonstrate the “Schneider”
effect that resulted in the writing of WRC 329. Out-of-plane moments on
run pipe are also illustrated in this simple branch connection and the
change in stress due to more applicable i-factors is observed.
10) SPRGDES3.C2 – Piping system that shows how spring hanger design can be
affected by a more accurate simulation of the pipe stiffnesses. Spring
design travel doubles after more applicable flexibility factors are used for
branch connections in the model.
11) AIRCOOL047A Plus M.C2 –
12) Line1004-1005.C2 –
13) MK10010.C2 –
14) THREEBENDS.C2 –
15) UNKNOWNTYPE.C2 –

Each of the Example model files is presented and discussed in detail in the
remainder of this chapter.

© Paulin Research Group – 2018 All Rights Reserved


182
F E A T O O L S ™ U S E R M A N U A L

Detailed Coverage of the Example Model Files

Example Model File #1 – WRC 329 Example

Model Overview & Description


Running the delivered file in CAESAR II the user should get an MZ operating
moment on the pipe at node 15 of 29,197 ft.lb. After running through the
CAESAR II FEA Translator, the operating MZ moment on the pipe at the
intersection drops to 4,725 ft.lb., a drop of almost 84%.
Stated alternately, the “un-translated” calculated MZ Moment is ~6 times
larger than the calculated MZ moment you get from the “translated” model!
A more applicable “translated” intersection model is included for the small
d/D intersection (node 15) as shown in the figure below on the right.

Model Analysis Details

Model Files: wrc329.C2 (Original CAESAR II model file)


wrc329-FEA.C2 (Translated CAESAR II model file)
wrc329.dom (PCLGold model file)

Model Notes: None. Please see overview.

Model Illustration:

© Paulin Research Group – 2018 All Rights Reserved


183
F E A T O O L S ™ U S E R M A N U A L

CAESAR II Results (Original CAESAR II Model):

Verified CAESAR II Results (Translated Model):

© Paulin Research Group – 2018 All Rights Reserved


184
F E A T O O L S ™ U S E R M A N U A L

Example Model File #2 – S303 Example

Model Overview & Description


This is the example S-303 in the B31.3 Appendix S 2012 Version. This example
demonstrates the condition where the maximum stress range can occur in
between two different operating conditions. This model is also described in
the discussion example #1 in the previous chapter.

The S303 problem is comprised of little straight pipe, no elbows, and six
welding tees. The model is planar, and so the largest moment is in the MY
direction. The highest operating moment for the two thermal cases occurs at
the node 20 welding tee. The user can see that the moment changes sign
from one thermal case to the next and that the model shows to be
overstressed for one or more range load cases.

After running the model through the CAESAR II FEA Translator, the operating
moment MY at node 20 drops and the load case range is no longer
overstressed.

Model Analysis Details

Model Files: S303.C2 (Original CAESAR II model file)


S303-FEA.C2 (Translated CAESAR II model file)
S303.dom (PCLGold model file)

Model Notes: Load Case 10 in the CAESAR II S303.C2 example model


describes the worst range case for fatigue analysis.
After running the model through the CAESAR II FEA
Translator, the operating moment MY at node 20 drops,
and the load case range 10 is no longer overstressed.

© Paulin Research Group – 2018 All Rights Reserved


185
F E A T O O L S ™ U S E R M A N U A L

Model Illustration:

CAESAR II Results (Original CAESAR II Model):

Verified CAESAR II Results (Translated Model):

This is often the way that FEATools™ is run. If the user suspects that errant i-
factors or overestimates of the stiffnesses are causing an overstress, then
once the red text is observed, the user will immediately translate and rerun
the model. If the translated model is no longer overstressed (as is the case
here), then the user knows that only how the intersections are modeled has a
significant effect on the stress calculations for this system.

© Paulin Research Group – 2018 All Rights Reserved


186
F E A T O O L S ™ U S E R M A N U A L

Example Model File #3 – bypass B13 Example

Model Overview & Description


In this model a slug load occurs at the lower elbow of the 20” gas piping
system as shown in the diagram of the piping system below. The slug load
causes the pipe leading into the vessel nozzle to experience displacements
along the line that the slug load acts. The typical CAESAR II branch connection
model assumes k=1 for each of the branches framing into the manifold.

When the branch junctions are rigid in the analysis each branch will pick up
more of the slug load than it should, artificially “protecting” the vessel nozzle
at node 145. When the model is run using the CAESAR II FEA Translator, the
summed moment loads on the vessel nozzle increase from 81,329 ft.lb. to
142,527 ft.lb., an increase of 75%. The model includes the 52,000 lb.
occasional slug load statically applied at the elbow node number 10 in the +X
direction. The statically applied slug load was calculated to be 26,000 lb., and
was multiplied by a dynamic load factor (DLF) of 2 and then applied to the
pipe.

The model also includes the finite element analysis of a branch connection
lateral.

Model Analysis Details

Model Files: bypassB13.C2 (Original CAESAR II model file)


bypassB13-FEA.C2 (Translated CAESAR II model file)
bypassB13.dom (PCLGold model file)

Model Notes: None. See Overview

Model Illustration(s):

© Paulin Research Group – 2018 All Rights Reserved


187
F E A T O O L S ™ U S E R M A N U A L

CAESAR II Results (Original CAESAR II Model):

Verified CAESAR II Results (Translated Model):

The second model results shown above are from the translated model that includes the
branch connection flexibilities. When the series of manifold branches are no longer rigid,
they carry less of the slugging load, permitting more displacement and allowing the vessel
nozzle to be loaded.

© Paulin Research Group – 2018 All Rights Reserved


188
F E A T O O L S ™ U S E R M A N U A L

Example Model File #4 – Small bore pipeline example No. 1

Model Overview & Description


When the standard CAESAR II model is run for this example the stresses show
to be ~63% of the allowable. When the model is ran through the CAEASR II
Translator, the stresses increase and show to be ~124% over the allowable. In
this case, the standard B31-type model of the small bore branch connection is
not adequate for an accurate flexibility analysis.

When small bore pipe is attached to a large diameter header the stiffnesses
may be critical if the small bore pipe configuration does not have much
flexibility. In this case, the default model of the intersection may be non-
conservative.

Model Analysis Details

Model Files: BRG401.C2 (Original CAESAR II model file)


BRG401-FEA.C2 (Translated CAESAR II model file)
BRG401.dom (PCLGold model file)

Model Notes: None. See overview.

Model Illustration:

This condition can also occur whenever the d/D ratio << 0.5 and the branch
pipe off the run has little flexibility. Many piping failures occur at branch
connections that are inherently not flexible and controlled by the
displacement of the larger branch pipe. The diagram at the end of this
example intends to illustrate more clearly this condition.

© Paulin Research Group – 2018 All Rights Reserved


189
F E A T O O L S ™ U S E R M A N U A L

CAESAR II Results (Original CAESAR II Model):

Verified CAESAR II Results (Translated Model):

The situation in this model is illustrated in the figure below. The typical beam-
type piping model is shown in the middle. Small bore flexible pipe is modeled
from the centerline to the first elbow. In most cases this assumption is not so
far off that large errors in the calculated stresses result. The recommended
ASME Section III and ASME ST LLC 07-02 model is shown on the bottom right.

© Paulin Research Group – 2018 All Rights Reserved


190
F E A T O O L S ™ U S E R M A N U A L

In the model on the bottom right, the pipe length from the centerline to the
surface is rigid, (or very stiff). The artificial flexibility of the branch pipe from
the centerline intersection to the surface is NOT added to the model. For axial
loads in the horizontal run in this configuration, adding the rigid element from
the centerline to the surface removes the majority of bending flexibility in the
system and causes the stress to more than double. The FEATools™ model
includes the correct stiffness model at this intersection and includes a rigid
element from the centerline to the surface with the local flexibilities of the
branch connection at the surface between the end of the rigid element and
the beginning of the branch pipe. If the k factor for branch connections of this
type are close to 1, then including branch connection flexibilities can actually
increase the stiffness of the piping system, like what has happened in this
example.

© Paulin Research Group – 2018 All Rights Reserved


191
F E A T O O L S ™ U S E R M A N U A L

Example Model File #5 – Small bore pipeline example No. 2

Model Overview & Description


In this example, a more advanced (FEA) analysis showed that the compressor
skid branches didn’t have enough flexibility. (The default intersection model
was non-conservative) The client rerouted the small bore compressor piping
to provide more flexibility and added extra guides to the large bore run pipe.
These improvements to the pipe the layout protecting the small bore
compressor pipe caused the run pipe to become over-stressed. (This is the
same situation as seen in the previous Example #4.)

When the designer added guides to the run pipe to reduce axial movements
on the compressor branches (see figure below), he increased the MY moment
in the pipeline, and the excessively conservative B31.3 run pipe io SIF caused
the run pipe to show to be overstressed (incorrectly).

When using STANDARD intersection models for small d/D branch


connections the Code is often grossly over-conservative due to the
assumption that the run-side out-of-plane i-factor for small d/D branch
connections is the same as the branch out-of-plane i-factor for a size-on-size
intersection. WRC 329, p.22 has called this error “obviously silly”, but
without noticing the dilemma, the designer of this system thought that his
modifications caused the run pipe to be overstressed. The CAESAR II FEA
Translator corrects this i-factor error in the run pipe.

When the new piping configuration was run through the CAESAR II FEA
Translator and then reanalyzed in CAESAR II, the model was shown to be 29%
of the allowable.

Model Analysis Details

Model Files: BRG402.C2 (Original CAESAR II model file)


BRG402-FEA.C2 (Translated CAESAR II model file)
BRG402.dom (PCLGold model file)

© Paulin Research Group – 2018 All Rights Reserved


192
F E A T O O L S ™ U S E R M A N U A L

Model Illustration:

CAESAR II Results (Original CAESAR II Model):

Verified CAESAR II Results (Translated Model):

© Paulin Research Group – 2018 All Rights Reserved


193
F E A T O O L S ™ U S E R M A N U A L

Example Model File #6 – Manifold piping example

Model Overview & Description


In this example, a large 24” manifold connection piping system contains one
branch connection. The pipe attached to the intersection is relatively short
and attaches to what is considered fixed anchors. This is a typical
configuration where flexibilities at the branch connection might have a large
effect on the calculated forces and moments. By using the CAESAR II FEA
Translator the loads in the system become redistributed when the local
branch connection flexibilities are included in the piping system model.
The calculated moments for both cases are summarized below. These types of
redistributions are typical. The largest moment at 45 drops 19% while the
second largest moment increases 25%. The impact of the of the load
redistribution due to an improved modeling of the stiffnesses may be
determined by which anchor is the most sensitive to applied moments, i.e. if
the pump is at node 5, then including flexibilities helped. If the pump is at
node 25 then including flexibilities hurt the system. In either case, a more
accurate analysis resulted and should be used appropriately by the designer.

Model 5 MX 5 MY 5 MZ 25 MX 25 MY 25 MZ 45 MX 45 MY 45 MZ
Standard
Branch
49507 -14752 -35581 -15401 68824 -2347 85604 -19615 -41622
Model
(k=1)
FEA More
Applicable
34368 -12336 -36343 -9168 85874 996 69299 -11289 -18411
Branch
Model
% of
Standard
69% 84% 102% 60% 125% 70% 81% 58% 44%
Branch
Model

Model Analysis Details

Model Files: MANIFCONN.C2 (Original CAESAR II model file)


MANIFCONN -FEA.C2 (Translated CAESAR II model file)
MANIFCONN.dom (PCLGold model file)

Model Notes: A large 24” manifold connection piping system with a


relatively “tight” configuration and connections
originally treated as inflexible anchors. (No flexibilities
taken into account)

© Paulin Research Group – 2018 All Rights Reserved


194
F E A T O O L S ™ U S E R M A N U A L

Model Illustration:

CAESAR II Results (Original CAESAR II Model):

Verified CAESAR II Results (Translated Model):

© Paulin Research Group – 2018 All Rights Reserved


195
F E A T O O L S ™ U S E R M A N U A L

Example Model File #7 – “Jim Wilcox” Complex Load Case


Example

Note: This example discusses a model that is improved using the CAESAR II
FEA Translator. It also shows further improvement when another PRG FEA
Product (PRG’s advanced pipe stress analysis program – PCLGold™) is applied
to the problem.

Model Overview & Description


Mr. Jim Wilcox of CodeCAD proposed the model used in this example, offering
insight into B31 Code fatigue analysis. The three branch lines shown below
come from identical heat exchangers that operate in random sequences. The
three possible temperature scenarios in the non-symmetric system represent
a total number of permutations equal to binary “111” + 1 = 7 + 1 = 8. (1 is
added for the case where all three branch lines are cold, (0, 0, 0), and the
remainder of the line is hot. There are eight possible thermal configurations.
The highest system stress range may occur in between any of the two possible
individual thermal states, and so the number of possible stress range cases
that must be evaluated is some number, considerably greater than 8.

© Paulin Research Group – 2018 All Rights Reserved


196
F E A T O O L S ™ U S E R M A N U A L

Once the number of operating cases is known, the number of possible unique
range cases is given by the expression below. The number of operating cases
is the number of load cases that can exist that are not equal to the ambient or
“installed” load case. The number of range cases computed, includes the
ranges from the ambient, (non-operating) case, to each of the operating cases,
and the number of possible ranges between the different operating cases.

With eight operating cases, there can be (82 + 8 )/2 = 36 permutations of load
combinations that must be investigated on a node-by-node basis. For each
node in the model, the user must investigate the 36 range cases to find the
maximum stress range present in the system for that node to confidently find
the maximum stress range.

Finding the maximum damage due to each of the ranges can be more difficult
when each operating case does not cycle the same number of times. Let’s use
a simple three load case problem to illustrate this condition. If operating case
1 cycles 100 times, and operating case 2 cycles 500 times, and operating case
3 cycles 1000 times, then if the maximum range occurs between cases 1 and 3,
then the case 3 to case 1 stress range exists for 100 stress occurrences, which
leaves 900 cycles for case 3, and 500 cycles for case 2. Next would be the 500
cycles for case 2 to case 1, if that is the next largest cycle. Finally there would
be the last 400 cycles from case 3 to ambient.

The cycles in in this example would be:

100 cycles: Case 1 to Case 3.


500 cycles: Case 2 to Case 3.
400 cycles: Case 3 to Ambient.

© Paulin Research Group – 2018 All Rights Reserved


197
F E A T O O L S ™ U S E R M A N U A L

If for another node in the system, the largest cycle ranges are always between
ambient and the thermal condition, then the cycle stress ranges would be:

100 cycles: Case 1 to Ambient


500 cycles: Case 2 to Ambient
1000 cycles: Case 3 to Ambient

For each node the maximum ranges must be evaluated since the stress in a
particular operating case can be different from node-to-node in the model.

This model also shows the complicating effect of pressure in the range
calculations. When pressure is included in the range calculation, the largest
range is often from zero load, to the operating case + pressure. When
pressure is not included, cases developed due to the different signs of the
moments can be greater and often show up more frequently in the moment
combinations.

PCLGold makes all the load case permutations automatically. An example is


shown below. For the nodes 80.3 to 90.0 the 36 possible load case
permutations are shown for each elemental node on the left, and the 28 load
cases are shown on the right.

© Paulin Research Group – 2018 All Rights Reserved


198
F E A T O O L S ™ U S E R M A N U A L

When the pressure can cycle without external loads, it is important that the
user include the pressure only variation in the stress range calculation for
fatigue. For the element output shown above it can be seen that for the node
80.3, the highest stress range is for cases 4-to-16, while for node 90, the
highest stress range is between cases 16 and 22. The stress remains relatively
high for a large number of the load case ranges, suggesting that their
evaluation is important to properly determine the fatigue life. (If stress
increases by 20%, life, in terms of cycles, may be reduced by 400%.)

For CAESAR II, it is not difficult to setup the range cases and perform the
fatigue analysis for the interpretation of B31.3 Eq. 1d so that all ranges can be
evaluated regardless of the number of times they appear. The cycle count for
each appearance of a possible range case can be computed and taken to be
the maximum number of cycles in the range. Using this approach the CAESAR
II result shows the model to be greatly overstressed for individual ranges, and
so combinations of the ranges are not required yet.

Once the model is run through the CAESAR II FEA Translator, the loads drop
significantly due to the added flexibility and the correct i-factors on the run
and branch pipes. After the translation of the model, none of the range cases

© Paulin Research Group – 2018 All Rights Reserved


199
F E A T O O L S ™ U S E R M A N U A L

show to be overstressed. Now the individual ranges should be properly


combined using cumulative damage. This is done automatically in the PCLGold
program, but is a manual process in CAESAR II. Since there are so many
ranges, the user can consider translating the model to the PCL, and running it
there to determine which of the cases are most significant. Once the user
knows which ranges are the most important, they can be setup and analyzed
in CAESAR II.

Model Analysis Details

Model Files: JWilcox.C2 (Original CAESAR II model file)


JWilcox-FEA.C2 (Translated CAESAR II model file)
JWilcox.dom (PCLGold model file)

Model Notes: None. See overview.

© Paulin Research Group – 2018 All Rights Reserved


200
F E A T O O L S ™ U S E R M A N U A L

Model Illustration:

CAESAR II Results:

Load Cases Before Load Cases After


Translation Translation

© Paulin Research Group – 2018 All Rights Reserved


201
F E A T O O L S ™ U S E R M A N U A L

Example Model File #8 – Pump Discharge Manifold Example

Model Overview & Description


In this example a model of a relatively typical small pump discharge manifold
is analyzed. The model file PUMPV10RG.C2 demonstrates the most common
way CAESAR II users would simulate this system. The vessel nozzle shown can
be modeled several ways and the effect of each approach is presented below
in terms of the pump loads.

The summary tables below show the loads calculated using the different
modeling techniques. (based on PUMPV10 files)

PUMPV10 standard and translated results


Fx Fy Fz Mx My Mz

Standard
Fx Fy Fz Mx My Mz

Advanced
PUMPV10RG (rigid) results
Fx Fy Fz Mx My Mz

Rigid

A small image of the system is shown below. Let’s review the Mz moment
since it will be the moment most influenced by the nozzle flexibility.

The “standard” model uses a rigid element from the surface of the vessel to
the centerline of the vessel and the Mz moment at the nozzle is 7087 ft.lb.
When flexibilities are included the moments reduced, but only to 6737 ft.lb.

© Paulin Research Group – 2018 All Rights Reserved


202
F E A T O O L S ™ U S E R M A N U A L

The “standard” model was then reanalyzed and a 9977 ft.lb. moment
produced. What is the difference between these models and why are the
moments different? The answers to these questions can be found in the
figures below. Figure (A) shows the basic geometry and Figure (B) the
centerline modeling approach. Local nozzle flexibilities are placed at Point
“B”. In the “standard” model, no flexibilities were included and the rigid
element was “rigid” with respect to the entering pipe. This is shown in Figure
(C). In the “rigid” model, no flexibilities were included, but the rigid element
was “rigid” with respect to the vessel. In this case, the relative “rigid”
stiffnesses assigned by the programs had an impact on the solution. This most
often occurs with vessel models. User’s must be sure that the rigid elements
from the surface to the centerline are “rigid” with respect to the vessel, or run
pipe. This can also occur when d/D << 0.5.

The results show that accidental flexibility provided by the element that was
rigid with respect to the pipe was about equal to the flexibility provided by the
CAESAR II translator. The CAESAR II translator inspects all pipe in the vicinity
of the branch connection and assigns the proper “rigidity”, thicknesses and
diameter to appropriately simulate the flexibility of the tee.

© Paulin Research Group – 2018 All Rights Reserved


203
F E A T O O L S ™ U S E R M A N U A L

Model Illustrations for PUMPV10 model file

Model Illustrations for PUMPV10RG (Rigid) model file

Model Analysis Details


Model Files: PUMPV10.C2 (Original CAESAR II model file)
PUMPV10-FEA.C2 (Translated CAESAR II model file)
PUMPV10.dom (PCLGold model file)

Model Notes: None. See overview.

© Paulin Research Group – 2018 All Rights Reserved


204
F E A T O O L S ™ U S E R M A N U A L

Model Illustration:

CAESAR II Results (Original CAESAR II Model):

Verified CAESAR II Results (Translated Model):

© Paulin Research Group – 2018 All Rights Reserved


205
F E A T O O L S ™ U S E R M A N U A L

Model Files: PUMPV10RG.C2 (Original CAESAR II model file)


PUMPV10RG.dom (PCLGold model file)

Model Notes: None. See overview.

Model Illustration:

CAESAR II Results (Original CAESAR II Model):

© Paulin Research Group – 2018 All Rights Reserved


206
F E A T O O L S ™ U S E R M A N U A L

Example Model File #9 – WRC 329 “Schneider” Out-of-plane


Moment Example

Model Overview & Description


The original reason for developing WRC 329 was to address the so-called
“Schneider” effect on reduced branch connections. The “Schneider” effect is
the increase in stress due to an out-of-plane moment about the branch when
the d/D ratio is about 0.7. This effect was first reported by R.W. Schneider in
the 1970’s and is the basis for Note 11 to B31.3 Appendix D.

The branch connection used for this problem is shown in the model diagram
below. It contains the out-of-plane branch moment of concern, but it also
contains an out-of-plane moment about the run pipe. This is a 10”x12”
intersection, where both pipes have a wall thickness of 0.375”. The default
CAESAR II model shows the operating case calculated stress on the branch
(due to the out-of-plane moment on the branch) to be 55,212 psi. The
operating calculated stress on the run pipe due to the in-plane moment on the
run is 39,941 psi.

The in-plane i-factor for the run and the branch is 4.625, and the out-of-plane
i-factor for both the run and the branch is 5.833. The d/D ratio for this
intersection is (10.75-0.375)/(12.75-0.375) = 0.838.

Once the user translates the model and uses the more applicable out-of-plane
stress intensification factor the stress on the branch jumps to 66,161 psi. The
stress on the run pipe drops to less than 10,000 psi. The large drop in stress is
due to the out-of-plane run moment applied. Stresses due to out-of-plane
loads through the run are known to be poorly evaluated by the Code for
reduced branch connections. This was the reason for Note 11 in B31.3
Appendix D.

The stress intensification factor for the run moment is less than 2, not more
than 5.0 as produced by the 2012 version of B31.3. The out-of-plane run
moment SIF of 1.044 can be considered to be more applicable data.

Model Analysis Details

Model Files: Outplane.C2 (Original CAESAR II model file)


Outplane-FEA.C2 (Translated CAESAR II model file)
Outplane.dom (PCLGold model file)

© Paulin Research Group – 2018 All Rights Reserved


207
F E A T O O L S ™ U S E R M A N U A L

Model Notes: None. See overview.

Model Illustration:

© Paulin Research Group – 2018 All Rights Reserved


208
F E A T O O L S ™ U S E R M A N U A L

Example Model File #10 – Spring Hanger Design Example

Model Overview & Description


This example demonstrates how the design travel can almost double when
branch connection flexibilities are included in the piping system that includes
hanger design.

Model Analysis Details

Model Files: SPRGDES.C2 (Original CAESAR II model file)


SPRGDES.dom (PCLGold model file)

Model Illustration:

d The general variable spring hanger design problem is one of unbalance.


Variable springs are designed to carry the weight of the piping system
through the thermal travel with as little load variation as possible. The load
variation for hot or cold load design is always k for variable springs where k
is the stiffness of the spring and  is the travel. This load is carried by the
system in the installed case for hot load design and in the cold case for cold
load design. This collected unbalanced load at each spring design location can
push the piping system around more when the system is more flexible. For
branch connection flexibilities, the added flexibility isn’t provided by more

© Paulin Research Group – 2018 All Rights Reserved


209
F E A T O O L S ™ U S E R M A N U A L

pipe, and so the weight of the system is the same whether the translator is
used or not. The free thermal travel of the pipe in the vertical direction is
often the same too, and so in practice, the unbalanced load due to spring
design doesn’t vary significantly. When there are a large number of branch
connections as in the system shown above, the system stiffness can change
considerably when the translator is run, generally dropping, and the spring
travel generally increases. The spring travel is usually a critical design
parameter since the spring travel is limited in the spring, and the greater the
travel, the greater the unbalance on the system.
Users of tight piping systems, or systems whose stiffness are otherwise
affected by branch connection flexibilities are urged to run the spring design
with both the translated and not translated models and make sure the
selected springs, loads and travel are the same. Results from the above
model are shown below.
When the CAESAR II translator was used, the calculated spring travel increased
from 1.05” to 1.62” inches.

CAESAR II Results (Original CAESAR II Model):

Verified CAESAR II Results (Translated Model):

© Paulin Research Group – 2018 All Rights Reserved


210
F E A T O O L S ™ U S E R M A N U A L

Example Model File #11 –API661 Square Header Box


Example
Model File: AIRCOOL047A PLUS M.C2 (Original CAESAR II model file)

© Paulin Research Group – 2018 All Rights Reserved


211
F E A T O O L S ™ U S E R M A N U A L

© Paulin Research Group – 2018 All Rights Reserved


212
F E A T O O L S ™ U S E R M A N U A L

© Paulin Research Group – 2018 All Rights Reserved


213
F E A T O O L S ™ U S E R M A N U A L

Example Model File #12 –Piping Technology and Products


Shoes Example
Model File: LINE1004-1005.C2 (Original CAESAR II model file)

© Paulin Research Group – 2018 All Rights Reserved


214
F E A T O O L S ™ U S E R M A N U A L

© Paulin Research Group – 2018 All Rights Reserved


215
F E A T O O L S ™ U S E R M A N U A L

Example Model File #13 –Hillside Nozzles in Vessel


Model File: MK10010.C2 (Original CAESAR II model file)

© Paulin Research Group – 2018 All Rights Reserved


216
F E A T O O L S ™ U S E R M A N U A L

© Paulin Research Group – 2018 All Rights Reserved


217
F E A T O O L S ™ U S E R M A N U A L

© Paulin Research Group – 2018 All Rights Reserved


218
F E A T O O L S ™ U S E R M A N U A L

Example Model File #14 –Bend Trunnion Example


Model File: THREEBENDS.C2 (Original CAESAR II model file)

Graphical Output

© Paulin Research Group – 2018 All Rights Reserved


219
F E A T O O L S ™ U S E R M A N U A L

© Paulin Research Group – 2018 All Rights Reserved


220
F E A T O O L S ™ U S E R M A N U A L

Example Model File #15 –Simple Intersection Example


Model File: UNKNOWNTYPE.C2 (Original CAESAR II model file)

Graphical Output

© Paulin Research Group – 2018 All Rights Reserved


221
F E A T O O L S ™ U S E R M A N U A L

© Paulin Research Group – 2018 All Rights Reserved


222

You might also like