FEATools

TM


Ease and Benefit of Adding FEA Results to Piping and Pressure Vessel Designs


Presenter: Tony Paulin, P.E.

There are certain well-known situations where present analysis techniques provide
both non-conservative and wildly over-conservative results.

Finite Element solutions, when properly tempered by comparisons with test data have
been shown to provide better answers for many of these cases.

Production pipe stress and pressure vessel design and analysis is usually a schedule
sensitive activity. Delays, and changes can be expensive. Waiting weeks for a
complicated analysis can be expensive.

The objective is to use the latest technology, (whether its finite element based or not)
to get the right answer with software that is familiar to the user and to keep the
project on track, cost-effective and safe.
1-Dont say a line is overstressed and needs to be rerouted if it doesnt

2-Avoid adding loops if you dont need them

3-Dont build $50,000 supports for pumps that dont need extra support

4-Avoid cracking in heavily cyclic systems

5-Dont overestimate the stress by 8 times
There are practical implications!
FEATools
TM



Summary:
1) Show common situations where FEATools (or other improved calculations)
can significantly improve accuracy of results, production and workflow.
2) Look at some examples. (So we know when we should be careful.)
3) Look at how FEATools addresses these problems.
A balanced approach
When we talk about well-known situations where present analysis techniques provide non-
conservative results, we can start with B31.3:
Piping Code Limitations:
D/T > 100
0.5 < d/D < 1.0 may be non-conservative.
6
7
8
i
ob
for t/T between 0 and 1.1
9
i
ob
for t/T between 1.1 and 100
First recorded fatigue tests for piping recorded
by Markl. Dynamic shaker tests conducted by
Blaire between 1935 to 1945.
PRG k-factor, i-factor and I-factor tests
conducted in 2013 Houston Tx.
Piping Fatigue Tests from 1935 to 2013
11
Low Carbon Steel UFT (Sketch 2.3)
GTAW Intersection Weld
D/T = 91
Model Deflection
(in.)
Stiffness Difference
(%)
No k-Factors 0.019 2631 lb/in. 69%
FEATool/CAESAR II
k-Factors
0.032 1562 lb/in. 0.5%
Measured 0.032 1553 lb/in.
There are definitely warnings given to us as designers in the Codes, the (B31.3 Appendix
D Notes for example), and there are definitely a few holes in the test data record.
In 1987 WRC 329 identified weaknesses in the Code that could be corrected. In the late
1980s ASME Section III implemented almost all of the WRC 329 recommendations in
Sections NC and ND. The B31 Codes have yet to incorporate those modifications, but the
ASME project ST-LLC 07-02 was written to provide a basis for extending the WRC 329
recommendations, and later work by EPRI, WRC and others, to the B31 Codes.
Lets take a quick look at some of the notes in WRC 329 that are yet to be incorporated into
the B31 Codes
20
Short Header Long Branch Short Header Short Branch Long Header Long Branch
21
WRC 329 p.22
d/D < 1
Run SIFs
Out-of-plane run
SIFs for both
geometries are
the same
=
Silly Overestimate of the Stress:

ST-LLC 07-02 8.9767/1.0439 = 8.599 Times TOO Big
ASME Section III 8.9767 / 2.1 = 4.27 Times TOO Big
FEA SIF 8.9767 / 1.0 = 8.9767 Times TOO Big
CAESAR II Calculates the
stress in a branch connection
using the equation:
B31.3 uses i
t
= 1 for all branch connections.
B31.3 uses i
t
= 1 today.
Unreinforced
Tees
B16.9
Welding
Tees
Out-of-plane Stress Intensification Factor (i-factor) for Two Tee Types
Basically dont use the effective section modulus (Z
e
) with
the new set of equations...

AND Fix Weldon Fitting i-factors (Olets)
?
Excerpt from Table 1, p.4.
WRC 329 pp.31,32.
M
ob
comparison of test, B31.3 and
ST-LLC 07-02 for UFTs, EXTs, and Olets.
WRC 329 Fig. 16
Markl emphasizes that
the ratio (safety factor)
between the design
allowable stress and the
mean failure curve is two.
36
WRC 107 and WRC 297 are commonly used tools.

WRC 107 cylinder-cylinder evaluations do not include the stresses in the
nozzle which can be important when t/T < 1.

WRC 297 attempted to resolve the nozzle stress problem, but is overly
conservative as suggested in WRC 335:


Pinhole
Excavation
Pressure Vessels
pp. 36 & 37.
38
39
From: PVP-61871
40
What are the errors why didnt I see them in all my years working in pipe stress?
41
WPW-Welded
WPS-Seamless
User can select tee
fabrication style. Options are
based on measurements made
by Paulin Research Group
So far the discussion has centered on errors to be corrected:
Missing k-factors for all branch connections Fig. 15 Moment/Stress error of about 9
Errant run side i-factors small d/D Tees Error 8 times too high
Omitted torsional i-factor for tees Error equal to i
o
. (For large D/T ~ 10)
Olet i-factors are off by 2.
WRC 107 No stress in the nozzle t/T<1 and t/(T+T
pad
) < 1
WRC 297 overconservative stress in the nozzle by 3.7
Markl Slope Error (5 to 3)
WRC 433 Rodabaugh/Scavuzzo and Hinnant Ke for Pipe
Overly thin stamped B16.9 welding tees

Do these issues really matter:
Is a thru wall crack at the end of the life a problem?
Errors in the stress calculations are greater when:
D/T is high (say greater than 50)
The piping or vessel system cycles. (N>3125)
The piping or vessel system operates in the creep range
Failures/cracks occur most often:
Offshore where displacements and cycles can be large and high
Any onshore batch process where cycles can be high
Dynamic, high cycle situations
Throttling conditions (high cycles)
Pipe stress by proxy (let the software or the designer do the analysis)
When Loads are Important
When stresses govern they may be off
Rotating Equipment

Summary:

Calculation errors are generally a function of D/T
Bigger errors exist as D/T > 50
When cycles > 3125 stresses should be accurate
k-factors can affect loads which can affect
rotating equipment
stresses (When stresses control FEAtools might be able to help.)
Short, stiff pipe accentuate the k-factor affect. (Is L less than kd?)
i-k Spreadsheet
FEATools
TM

WRC 329 Fig. 15 Example Piping Model
Expecting:
High k-factors for the branch
Low k-factors for the run (d/D<0.5)
B31 High i-factors for the run
Reasonable i-factors for the branch?
WRC 329 Fig. 15 Example Piping Model
Without considering the branch
connection flexibility of the 12x30
fabricated tee at point 15 the out-of-
plane (Z) bending moment at point 15
is 372,000 in.lb. Including the branch
connection flexibility reduces the
bending moment to 41,832 in.lb., a
reduction of 8.8.
49
Example No. 5 Heater Piping
Example No. 5 Heater Piping
Nozzle Load Analysis:


62 OD x 0.5 Thick
M = 35,287 ft.lb.
(423,444 in.lb.)
14 OD x 0.375 Thick
Overstressed Add a 4
x 0.375 pad
OK With
4 x 0.375 pad
62 OD x 0.5 Thick
M = 35,287 ft.lb.
(423,444 in.lb.)
14 OD x 0.375 Thick
62 OD x 0.5 Thick
M = 35,287 ft.lb.
(423,444 in.lb.)
14 OD x 0.375 Thick
FEA Agrees that the Shell is OK, but the
nozzle neck is overstressed.
Conclusions:

1)If you have at least one copy of CAESAR or PVElite you should have at least one copy of
FEATools.

2)The PRGiK spreadsheet will give you a good sense of when the B31 i-factors and k-factors
might be off compared to a variety of other methods. If all the other approaches give essentially
the same i-factor and B31 does not agree with the i-factor, then its likely that all the other
approaches provide more applicable data

3)FEATools does provide analysis of single FEA models for tees and bends with staunchions
included with the i-k spreadsheet and the CAESAR II model translator.

4) The FEATools translator will take one CAESAR II piping model and convert it into another
CAESAR II piping model that incorporates all the FEA and i/k features discussed in this
presentation.

5)Some systems are more sensitive than others. You need to decide how much FEATools means
to your work. The PRGiK spreadsheet should give you some idea. Lots of straight pipe and
elbows generally means little affect from intersection k-factors. Where theres only a little
straight pipe and a few elbows the CAESAR II model translation can result in large differences in
loads and stresses.

6)Affects are more significant when there are high D/T branch connections in the piping system.
Results are more important when the number of cycles is greater than 3000 to 7000.
Thank you for your time.
T. Paulin