Beruflich Dokumente
Kultur Dokumente
www.paulin.com
Chapter 1
Introduction to the PRGiK™ Calculation Tool .............................. 1
Chapter 2
Detailed PRGiK™ Discussion ....................................................... 6
Chapter 3
PRGiK™ Examples .................................................................... 17
Chapter 4
Recommended use of k-factors ................................................. 34
Chapter 5
Stress Intensification Inconsistencies in WRC 329 .................... 36
Chapter 6
Reference Data .......................................................................... 38
1
Chapter
F E A T O O L S ™ U S E R MA N U A L
Introduction to the
PRGiK™ Calculation Tool
This chapter will provide the user with a brief
overview of what the PRGiK™ calculation tool does
and when it should be applied.
Generally, some additional evaluation beyond the current (2012) B31.3 piping
code is required when:
3) When the pipe stress program of choice does not permit for pressure i-
factors or pressure stress components to be a part of the equation 17
Code stress calculation for fatigue for pressure cycling at a branch
connection. (In this case the user must find some alternative approach to
evaluate Paragraph 301.10.)
4) When the pipe stress program of choice does not permit entry of torsional
i-factors greater than 1 as part of the equation 17 Code stress calculation
for fatigue for torsional loads on branch connections.
How to Use:
Activate the PRGiK program from the main FEATools Start dialog (shown
above on the left). The main PRGiK application (shown above on the right)
should appear.
Fill in, as a minimum, the first four data items as shown below:
The user can see stress and cycle values compared to allowables or mean
failure curves.
Detailed PRGiK™
Discussion
This chapter will provide the user with a detailed
technical discussion of the PRGiK calculation tool
not typical although the system shown in WRC 329 is not exceptional.
Guidelines for use of k-factors for system evaluation are provided in
Section 2.0.
3) Calculate B31 Stress SE using equation from Appendix P including a
separation of axial, pressure and torsional SIFs. There are numerous ways
to combine the up to five stress components that describe the stress state
in a pipe. When stresses cycle and are relatively high, it is important to
make the proper combination of stresses for comparison against the
allowable to be sure the desired Code separation from failure is
maintained.
4) Determine if given loads and cycles would likely cause a fatigue failure.
Mean failure lines for several criteria are presented. Plotted results
include the failure lines, the actual number of cycles and the calculated
stress so that separation between failure and stress can be estimated.
Only the branch and run pipe diameters and wall thicknesses need to be input.
(First four items in the upper left corner of the form. All other entries will be
defaulted.)
The branch connection types used in the calculation are defined in Figure 3
below.
A discussion of how the PRGiK results for the main screen might be used is
provided in a later chapter of this manual.
Applicable Notes:
1) Section III NB and NC equations are based on the worst case SIF for any direction.
Generally this is the out-of-plane direction, and so the out-of-plane SIF is reported
for the in-plane and torsional directions and so in some cases the Section III NB
and NC in-plane and out-plane SIFs can be artificially higher than those from the
other methods and should likely be ignored.
2) ST LLC 07-02, EPRI 110996 and 110755 SIFs are based on finite element regressed
surfaces, adjusted to match fatigue test data. The first public use of this approach
was described by Wais and Rodabaugh in EPRI 110996. WRC 497 gives the sum of
membrane and bending stresses on the surface of the geometry in a manner
similar to that used in EPRI 110996. The SIF determination method used in EPRI
110996 is then used to find the SIF from the regressed finite element data in WRC
497.
3) When the background is shown in red, at least one of the equation parameters is
outside of recommended limits.
4) Wais model boundary condition lengths used in EPRI 110996 and EPRI 110755 are
generally shorter than Widera models in WRC 497 and so for larger D/T branch
connections some interaction of the boundary condition occurs, resulting in lower
flexibility and SIF factors for Wais.
5) SIF factors are adjusted when effective section modulii are used as part of the
Code B31 stress evaluation so that SIFs given are equal to (Stress)(Z/M); where M
is the moment in the branch or run, and Z is the section modulus of the branch or
run. Mb and Zb are always used together, and Mr and Zr are always used together.
6) As recommended in WRC 329 and echoed by the finite element results in EPRI
110996, WRC 497 and ST-LLC 07-02, for the run pipe, the in-plane stress
intensification factor is higher than the out-of-plane stress intensification factor.
This is a reversal of the trend observed for in-plane and out-of-plane loads
through the branch pipe, and is a reversal of the in-plane and out-of-plane SIFs for
the run pipe in the B31.3 Code.
Typical example uses are shown below with discussions and comparisons.
The Pipe Stress Evaluation button permits the user to enter loads and/or stresses
for probability of failure, safety factor and comparisons with allowables and mean
failure curves. This capability is intended to let the user know when they are close
to the allowables, and when they are far removed from allowables so that the
actual magnitude of the SIF or calculated load is not that critical a part of design.
This capability helps the designer know when to use “more applicable data”.
Fabricated tee
Sketch 2.3 - UFT
Extruded outlet
Sketch 2.4 - EXT
Once the SIFs and k-factors have been computed, stresses may be evaluated
by pressing the “Pipe Stress Evaluation” button:
A. Input and control. Entries in this section let the user test variations and
see the instantaneous affect they have on the stress, allowable or
predicted cycles to failure.
B. Load Screen. Entries in section B are used to provide load components
from the case under study for the stress evaluation.
C. Stresses computed for each load or for the load combinations are plotted
in this area and compared to a variety of allowables. The “C” Endurance
Curve Window screen defaults as disconnected and is toggled on and off.
D. Stress/Allowable Plot Control. Log-log plots can be deceiving. Various
options for displaying the stress and allowables can be selected to look at
the calculated values. In particular, the user may zoom in on the stress
and allowable and switch to linear representations to get a clearer idea of
the relationship between the stress and the various allowables.
E. Selected tabular results are available for use and review.
F. Various collective manipulation options are available to help make clearing
joint text cells, or resetting parameters easier. The analyst that often uses
this screen to compute stresses should explore this option to ease the
repeated calculation effort.
G. Various stress combination options. When combining the stresses there
are several interpretations when multiple i-factors are used. Several of the
options are available so that the user can see if there is much of a
difference in the stress between the options.
“Options” Panel
The options panel permits quick clearing and entry of commonly used loads
and i-factors and it permits the user to enter stresses directly. By default,
forces, i-factors and pipe properties are entered and the program computes
stresses, allowables, probabilities, damage factors, etc. In some situations,
the user would rather enter the stresses directly and have the downstream
properties calculated directly. The “Expose Stress” button should be pushed
for this option. When the user pushes the “Expose Stress” button, the
stresses may be input, and loads will be back calculated based on the given i-
factor and pipe cross section properties.
PRGiK™ Examples
This chapter will present a collection of examples of
the use of the PRGiK calculation tool.
Example #1:
Input:
There are six correlation equations for unreinforced fabricated tees (Sketch
2.3) in the PRGiK spreadsheet. For this branch connection, the results from the
six correlation equations are given below:
Table 1-1 iob stress intensification factors for 8x12 Schedule 20 fabricated tee.
Given that each alternative method indicates that the B31 method is low, it
seems likely that, in accordance with Note 11, stresses computed using the
B31.3 i-factors from the 2010 (and earlier) Appendix D will be non-
conservative by between 1.22 to 1.74 times. If the directional component
causing the stress to be 91% of the allowable is ONLY the out-of-plane
moment, then the increase will certainly cause the intersection to be
overstressed according to the Code.
If some of the 91% is due to other stresses at the intersection, then the
increase in the out-of-plane moment might not be enough to cause a Code
failure at the intersection. The user must evaluate each component of the
stress to determine if the increase in the out-of-plane i-factor will cause the
branch to be overstressed.
Example #2:
If the ST-LLC 07-02 out-of-plane i-factor is used, the Sout stress would increase
from 27433 to 27433x11.486/7.694 = 40,930 psi. The bending stress is the
square root of the sum of the squares of the in-plane and out-of-plane
stresses, and so:
44,700 > 42250 psi, and so the increase in the out-of-plane stress
intensification factor will cause this intersection to exceed the Code allowable
stress.
Example #3:
Section 6.0 in this document details a number of the points made in WRC 329
regarding i-factor errors.
The inplane stress intensification factor for the branch connection is compared
in Table 1-3 below, and it can be seen that what is more applicable data shows
the in-plane SIF to be about one half of the B31.3 value.
The DNV, EPRI/Wais/110996, WRC 497 and STLLC07-02 methods all agree that
the i-factor for the in-plane branch stress intensification factor for this branch
connection should be between 3 and 4 and not 6. Adjust the in-plane stress
calculation using the lower i-factor.
The total stress at the intersection is computed using Eq. 17 from B31.3 and so
the total stress will be:
The in-plane stress was too high, and the out-of-plane stress was too low, but
the final result was still an overstressed branch connection.
Example #4:
Problem: For the 12x8 Sch 20 branch connection the piping was rerouted
due to stresses on the run. The result from that calculation is
shown below. Does the pipe really need to be rerouted?
One of the inconsidtencies noted in WRC 329 is for the out-of-plane i-factor
for loads through the run pipe. For the 8x12 Sch 20 intersection evaluated in
this problem, the values of ior for the different methods are given in Table 1-2
below:
This suggests that in the B31.3 Code might be overestimating stress in the run
pipe due to an out-of-plane moment by 7.7 times. All i-factors are defined
with respect to a girth butt weld. An i-factor = 1.0 is with respect to the stress
at a girth butt weld and not the stress in a plain pipe removed from welds.
Many of the ior factors calculated by FEA for smaller d/D ratios are less than 1
since the high nominal stress is on the side of the pipe 90 degrees removed
from the branch connection. The high stress location for the ior stress is
shown in the figure below with the red star. EPRI TR 110996 and ST-LLC 07-02
agree that the out-of-plane i-factor for the run for this intersection is equal to
one. The WRC 329 comment made regarding this stress component is
included next to the figure below.
For the run the torsional SIF is at least 2.68 times greater than what’s used in
B31.3, while the ior SIF used for B31.3 is probably about 7.7 times too high,
and the iir is probably about 6.02 / 3.56 = 1.69 times too high. The adjusted
stress calculation would be:
The stresses in the run side were grossly overestimated since the run-side out-
of-plane i-factor for B31.3 is only based on size-on-size branch connections.
The pipe did not need to be rerouted due to stresses in this tee. The stress in
the run side elements were 61% off the allowable, and not 230% of the
allowable.
Example #5:
Problem: The PRGiK table for the 8x12 Sch 20 fabricated branch
connection includes k-factors for several of the references, but
the k-factors for B31 are blank – why is that? Does that have
any impact on the piping analysis?
The kob factors for branch connections in the PRGiK tables are a multiplier on
the diameter. For branch side flexibility factors, the k-factors multiply the
branch pipe nominal diameter, and for the run side flexibility factors, the k-
factors multiply the run pipe nominal diameter. The B31 Appendices for SIFs
and k-factors give k-factors equal to 1.0 for all branch connections. ASME
Section III NB3200 gives k-factors for fabricated tees when d/D < 0.5 and
methods for installing those k-factors.
The most important priority given by WRC 329 in the recommendation section
is that for the B31 Codes, that the meaningless k-factor of 1 should be deleted.
The note from WRC 329 is provided in part below:
Since WRC 329 was written, E.Rodabaugh, E.Wais, Widera, PRG, and others
have produced k-factor equations for branch connections. PRGiK compares
the most common ones. For the 8x20 Sch 20 branch connection, the kob
flexibility factor is given in the table below.
The out-of-plane flexibility factor for this branch connection is almost at least
30, and so at the surface of the branch connection, a point rotational stiffness
that is as flexible as the rotational equivalent of 30 diameters of branch pipe
should be present in the piping system.
The analyst must look at the piping configuration and decide if 30 extra
diameters of bending flexibility applied as a point rotational spring at the
intersection of the branch centerline and run surface will further reduce the
loads on the branch connection. Including flexibilities at branch connections
provides more accurate flexibility models of the piping system.
Pressing the “k-per elbow basis” button gives the following k-factors for the
8x12 Sch 20 tee.
For the kob flexibility, omitting the local flexibility factor would be equivalent
to leaving out the extra flexibility of 1.3 bends.
Example #6:
Problem: Using the output for the 8x12 Schedule 20 tee, what
conclusions can be drawn for the intersection?
The PRGiK table for the 8x12 Schedule 20 tee is shown below. Conclusions
drawn from this result are included in the table below.
Conclusions:
EXAMPLE #7:
Appendix P is the only section of B31.3 that explicitly includes pressure with
applied external loads to compute fatigue, and so the equations P17a and
P17b will be evaluated to include the pressure stress as recommended in
P319.4.4(a). In the definition of Fa = axial force, the axial force due to internal
pressure is to be included. The resulting nominal stress is found from F/A,
which for pressure in a pipe cross section is equal to PD/(4T). The stress S a is
found from ia x (F/A) or for the pressure part of the longitudinal stress will be ia
x (PD/4T).
In the definition for ia, ia is equal to io for components besides elbows and
straight pipe. For the run pipe in the example problem, io =
(0.9)(11.875/0.25)2/3 = 11.804. The design pressure for the pipeline is 175
psig.
The PRGiK spreadsheet uses existing correlations and so does not produce an
i-factor for pressure. The FESIF program that is also a part of FEATools can be
used to generate i-factors due to pressure. Two FESIF calculations are
required. One analysis is performed to obtain the i-factors for branch side
loads, and the other analysis is performed to obtain the i-factors for run side
loads.
The shell FEA analysis suggest that all the i-factors for pressure in this thick,
small connection are 1.0.
NozzlePRO lets the user transform the shell model into a brick (volumetric –
thru-thickness) model.
The maximum linearized stresses (for only the two element through the
thickness model is given below)
From EPRI 110996, for use with external loads, (and failure at the edges of the
fillets), the i-factor for use with pressure would be: ip = C2K2/2; where K2 = 1
when weld lengths are not entered and C2 is computed using the shell type
membrane and bending stress. C2 = 12,012 / 4200 = 2.86. ip = C2/2 = 1.43.
For a finer mesh (5 nodes through the thickness of the non-compatible shaped 11
noded, reduced integration brick element), the maximum linearized stresses are
essentially the same as those above, confirming this result. This would lead us to
believe that the ip should be 1.43 for pressure and not 11.804. The pressure stress
in this case would be expected to be off by 11.804/1.43 = 8.25 times.
Recommended use of
k-factors
T here are two common uses of k-factors for piping system analysis. These
are discussed in detail in this chapter.
1. Comparative – Qualitative:
The analyst computes the k-factor and then based on a qualitative evaluation
of the piping system decides whether an analysis that does not include k-
factors gives an adequate representation of the piping system. When k-
factors are close to 1 and the piping attached to the branch connection is
relatively long and flexible, then the k-factors will likely have little effect on
the calculated forces, moments and stresses. When the k-factors are >> 1
and/or the pipe attached to the branch connection is relatively short – or
otherwise not-flexible because of the support in the area, then the k-factors
may have a large effect. Also when branch connection branches or runs are
at the end of long runs of cantilevered pipe, and the maximum moment
occurs at the branch connection due to support in the vicinity of the branch
connection large differences in the calculated moments can occur. The
example in WRC 329 Fig. 15 is of this type and shows an 800% reduction in the
out-of-plane bending stress at the branch connection in the example.
2. Evaluative – Quantitative:
Stress Intensification
Inconsistencies in
WRC 329
In the forward to WRC 329 written by Sam Moore, Mr. Moore writes: “[Mr.
Rodabaugh] … identified a large number of problems with the different code’s
usage of branch connection SIFs in their design procedures.”
p.9 “… using i = 1.0 for Mt on full size outlet branch connections can lead to
inaccuracies far greater than the Mob inconsistency.”
p.12 “We would rate the relative complexity of i-factors for pipe, elbows and
branch connections by the ratios of 1:5:500. … [readers] will not find any
simple answers in this report.”
p.13 “Extruded outlets are somewhat related to ANSI B16.9 tees in that
extruded outlets, like B16.9 tees, may vary significantly between
manufacturers.”
p.19 “ [C’2bo] … suggests that the t/T variable for r/R between 0.5 and 0.95 is
not very significant, and the Code assumption that /M is independent of t is
not too bad.”
p. 22 “For run moments on branch connections with small r/R, both intuition
and Ref. 26 data indicate that the B31.3 relationship ii = 0.75io + 0.25 is at
best, reversed in relative magnitude of iir and ior, … and in effect, [the] Code
requirements are obviously silly.”
p.24 “The available fatigue test data are inadequate to even guess at the
general accuracy of Code i-factors for run moments or how they vary with R/T,
r/R, t/T, r/rp or some other parameters.”
p.24” .. values[for] Mtr indicate that the B31.3 SIF i=1.00 for Mtr is perhaps
unconservative even for r/R < 0.5.”
p.28 “The Mob tests indicate that there is a peak somewhere around 0.75.”
p.32-33 “… delete the use of ii = 0.75io + 0.25 for branch connections tees, …
[it] gives the wrong relative magnitude for Mor versus Mir, [and] it
underestimates the difference between Mob and Mib for r/R between about
0.3 and 0.95 and perhaps over-estimates the difference for r/R below 0.2 and
for r/R = 1.0.”
p.37 “[limits on the inside radius of the branch connection are] dropped
because moment fatigue tests and theory indicate that the inside corner
radius is not a critical consideration.”
Reference Data
Index