GRI-00/0192.07
GRI GUIDE FOR LOCATING AND
USING PIPELINE INDUSTRY RESEARCH
SECTION 7
FITTINGS AND COMPONENTS
Prepared by
J. Kiefner
Kiefner and Associates Inc.
893 High Street, Suite L
Worthington, Ohio 43085
for
GAS RESEARCH INSTITUTE
1700 S. Mt. Prospect Road
Des Plaines, Illinois 60018
GRI Project Manager
Keith Leewis
Pipeline Program
March 2001Fittings and Components
FITTINGS AND COMPONENTS
Scope
This section of the Handbook discusses pipeline fittings and components, Fittings and
components are used to join, branch, cap, change the direction of, or change the size of straight
runs of pipe. Because of their geometric configurations, components usually experience higher
stress levels than adjacent pipes. This makes them potentially more susceptible to failures if the
higher stresses are not accounted for in their design, or in the layout of the piping system.
Consequently, the engineering community determined in the early years of piping system design
that fittings and components should be studied to better define the local stresses due to internal
pressure and applied bending moments, the flexibility characteristics which affect both the
stresses in the component and the overall distribution of loadings within a piping system, and
how details of component design and fabrication influence performance.
This chapter focuses on the research associated with the evaluation of pipe fitting and
component performance from early periods using experimentation and classical mechanics to
‘more recent efforts involving finite element methods. In the context of this discussion, pipe
fittings are limited to bends, branches, and flanges. While other types of fittings such as reducers,
caps, and wyes are important components, the bulk of fitting research has focused on bend and
branch design. The terms “fittings” and “components” are often used interchangeably. In the
context of this discussion, “fittings” refers to forged one-piece elements that are installed in-line,
primarily butt-welding elbows and tees, while “components” covers both fittings and fabrications
that perform essentially the same function as fittings.
The category of bends includes forged elbows, fabricated (mitered) bends, induction
bends, and cold field bends. Field bends and induction bends are distinct from standardized bend
fittings in that they are custom or made-to-order items formed from and integrally with initially
straight tangent pipe. Though sometimes regarded as “pipe”, they serve the same purpose as an
elbow, namely to change the direction of the run pipe. Like the standard bend, their shape alters
the magnitude and distribution of stresses due to applied loadings compared to straight pipe.
Unlike standard forged fittings, considerations associated with pipemaking uniquely affect the
properties of the final product. Those properties may differ greatly from those of the adjacent72 Fittings and Components
pipe. For practical purposes, induction bends and field bends comprise special cases of in-line
components which share some elements of “pipe” as well as fitings.
‘This discussion categorizes branches as either “integrally reinforced” or “fabricated”.
Integrally reinforced branches include forged tees and extruded outlets. Extruded outlets are
similar to induction bends in bridging the definition of pipe versus fitting in that they are made
from straight pipe. Fabricated branches include unreinforced junctions of straight pipe, reinforced
branches incorporating welded-on pads, saddles, or sleeves, and branches made using weld-on or
weld-in manufactured outlets. Examples of manufactured outlets include “Weld-o-lets” (a
proprietary name) and other similar “o-let” components. The weld-on fittings are often referred
to as “integrally reinforced” because usually no additional reinforcement of the opening is
necessary because of their heavy wall. However, clearly they are nonintegral to the run pipe, so
in the context of this report they are considered a subgroup of a fabricated branch. The weld-in
branch is normally contoured like a welding tee and inserted in an opening cut in the pipe. Its
performance approaches that of a welding tee, While the integrally reinforced branches are
usable only in new construction or modification of a line that is out of service, other branch
styles may be used in those situations as well as for hot-taps in operating pipelines.
‘The category of flanges includes welding neck types, slip-on types, and those that are
integrally formed at the end of a fitting such as a bend or branch. Methods of attachment to pipe
for all components include full-penetration welds with butt-welding components, fillet welds on
certain non-integral reinforcing elements and slip-on or socket-type components, threading, or
grooving. Threaded or grooved attachments will not be discussed, the majority of attachments
being welded.
Literally hundreds of papers have been published describing the results of theoretical,
numerical, and experimental investigations into various aspects of the pressure integrity and
fatigue performance of bends and branches, so it is impossible and probably unnecessary to
attempt to review and describe the collective body of work in its entirety. Instead, attention will
be given to seminal theoretical papers that laid the groundwork for present concepts, reports on
tests or experiments that demonstrated important aspects of fitting performance, survey reports
which summarize progress and findings, and selected recent efforts that point to important
revisions or refinements to earlier efforts. Discussion is organized in terms of design and73 Fittings and Components
performance considerations by fitting type for (1) internal pressure and pressure pulsation, and
(2) cyclic bending loads and flexure.
‘The focus will be on applications important to the oil and gas pipeline industry and on
studies sponsored by the pipeline industry, though findings from research in other industries will
be covered to the extent that they may be relevant to pipeline applications. Hence, many issues
that have presented important challenges to piping engineers in other industries, such as the
detailed theoretical analysis of Class 1 nuclear piping design, the performance of stainless steel
or nonferrous components, or the behavior of fittings at high temperatures, to name just a few,
will not be covered in significant detail. The 83 references cited in this section thus represent a
small fraction of the cumulative body of work on fittings and components.
Research on Fittings and Components
‘The evaluation of pipe fittings dates back to the turn of the century when theoreticians
‘began to apply classical mechanics methods to investigate the distribution of stresses in ideal
shells due to internal pressure or applied moments. Also, the empirical principles employed in
the design of nozzles and flanges in boilers that were developed early in the century were
initially carried over directly to the design of branches and flanges, respectively, in pressurized
pipe. Substantial and rapid progress in understanding the behavior of fittings, in developing
reliable design rules, and in the standardization of manufactured components was achieved
between 1945 and 1965, Refinement and optimization continues today, often by numerical
techniques (finite element analysis), although it is important to distinguish between empirical and
theoretical results as they apply to piping.
Numerous organizations and individuals have contributed to the vast body of research
underlying our present understanding of fittings and components, so only a few are listed to give
some idea of the breadth of interests. Funding for research has been provided by fitting and
equipment manufacturers (Tube Turns, Taylor Forge & Pipe, Bonney Forge); large plant
engineering firms (M. W. Kellogg, Babcock & Wilcox, Westinghouse Electric); government
agencies (the U. S. Navy, NASA, the U. S. Atomic Energy Commission); various groups
sponsored by the fossil and nuclear power industries and the gas and oil refining and
transportation industries (the Electric Power Research Institute, the Pressure Vessel Research14 Fittings and Components
Council of the Welding Research Council, the Pipeline Research Committee of the American
Gas Association); and foreign counterparts (the British Welding Research Association, the
British Ship Research Association). Organizations whose facilities and staff have been involved
in carrying out research on fittings and components have included manufacturing and
engineering firms such as those mentioned above; universities (Pennsylvania State University,
University of Waterloo, University of Ilinois, Hinois Institute of Technology, University of
Manchester); government R&D facilities (the Naval Research Laboratory, Oak Ridge National
Laboratory); and independent R&D organizations (Battelle Memorial Institute, Southwest
Research Institute). Of course the ideas and understanding are the contribution of individuals,
who can be identified by authorship of the reports referred to herein or as named in their
bibliographies.
Research on most aspects of pipe fittings and components is a rather mature field.
However, fitting and component behavior continues to be studied using traditional simplified
empirical methods, and sometimes, more recently using detailed finite element methods on a
case-by-case (and usually proprietary) basis. Situations where finite element analysis (FEA) has
Proven useful include the development of flexibility factors for components that fall outside the
range of components for which factors have been accepted and that are too large or too expensive
to be extensively tested at a reasonable cost, or optimization of a mass-produced component in
order to achieve unit cost savings. Other research continues including an effort to verify the
application of present piping flexibility-fatigue design concepts which were developed originally
using Grade B and X52 pipe and matching components, to higher X-grade materials; extension
of the flexibility-fatigue concepts to low-cycle fatigue; consideration of specific wall thickness
requirements for pressure design of elbows; development of a standard specification for
induction bends; establishing the acceptability of minor wrinkle deformations in pipeline bends;
and the publication of standard methods for proof testing fittings, components, and joints, and
developing stress intensification and flexibility factors. However, successful worldwide
operation of fittings and components in piping and pipeline systems gives testimony to the
overall adequacy of the treatment of fittings in the piping design codes and component standards
as they have evolved in current use.75 Fittings and Components
Internal Pressure and Pressure Pulsation
Standards
Intemnal pressure is the primary loading and governs the basic design of both buried
Pipelines and above-ground piping systems, though consideration for external loads and
Alexibility remain essential factors in establishing layouts for above-ground piping. It has
Sencrally been the rule that fittings should not constitute the weakest part of the piping system
‘fom the standpoint of pressure integrity. Standards for manufactured components such as forged
butt welding tees and elbows (ASME B16.9 and MSS SP-75) rely on performance language
rather than dimensional specifications to achieve satisfactory intemal pressure capacity when
used with “matching” or “designated” pipe. The standards require proof testing of a sample
representative of the finished product to bursting at a pressure at least equal to 105 percent of a
failure pressure computed as 2S1/D, where Sis the actual tensile strength of the fitting, and s and
D are the nominal dimensions of the matching or designated pipe. The proof testis intended to
apply to similarly proportioned fittings from one-half to twice the size tested. The value of Sis
determined either by cutting out and testing material from a representative fiting, or from plate
Coupons that remain attached to the fitting as it undergoes forging and heat treatment. Periodic
retesting is not mandated by B16.9, which applies only to Grade B-equivalent material, though
some manufacturers retest routinely. Testing at the rate of one per lot or heat is required by SP-
75 in cognizance of the unique and variable characteristics of the high-strength materials it
addresses. Both standards require that fittings be able to withstand a hydrostatic pressure test in
the field without failure, leakage, or impairment of its serviceability (excessive deformation or
‘gross through-wall yielding).
‘The applicable ASME B16 and ISO committees are still working on draft standards for
induction bends. Induction bends are not “cookie-cutter” components to the extent that fittings
are. They are greatly affected by the fabricator’s skill in selecting initial materials and in
consistently controlling the heat treatment of the bend from piece to piece. Draft standards
Presently specify minimum wall thickness design to account for the theoretical stress distribution
ina curved pipe. Performance is to be assured by control of process essential variables to a
Qualified range, material propertics tests made on a sampling basis per heat, and nondestructive
‘examination of the finished product. No hydrostatic test is required.16 Fittings and Components
Proof testing of fabricated components and field bends is usually performed in-line with
the pipeline and so is limited to yielding in the carrier pipe. Consequently, design rules which
require minimum thicknesses and other dimensions, along with workmanship quality standards,
have evolved to assure that the finished component will safely tolerate the intended service and
test pressures.
Elbows
A concise summary and bibliography of theoretical and experimental studies of piping
elbows is given by Rodabaugh, Dufly, and Atterbury in a 1969 report to the A.G.A. NG-18
Committee.” Lorenz® showed that the circumferential (hoop) stress due to internal pressure is
greater than for straight pipe on the inside of a bend, and less than for straight pipe on the outside
ofa bend. Assuming a uniform wall thickness, the hoop stress is increased by a factor (the
“Lorenz factor”) of 1.25 in the inside arc (intrados) of a standard short-radius elbow, and is
decreased by a factor of 0.875 on the outside arc (extrados) of the bend. As the ratio of bend
radius to pipe radius (R/r) increases, both factors converge toward 1. If the bend subtends a small
angle as is the case with field bends, the “Lorenz effect” is reduced by restraint from the tangent
pipe.” It happens that the forging process for forming elbows from straight pipe can result in an
increased wall thickness on the intrados and decreased wall thickness on the extrados. The
thickening and thinning tends to compensate for the distribution of elastic hoop stresses arising
from the curved pipe geometry. Bending stresses are also present along the crown due to a
discontinuity in the membrane stress solution for the toroidal segment™, and due to any initial
ovality of the pipe section, though these bending stresses do not appear to affect the pressure
integrity of the elbow. Excessive ovality can affect the elbow flexibility.
Rodabaugh examined the results of initial yield and burst tests conducted on elbows and
bends by Gross" and others, finding good agreement between experimental and theoretical
results up to initial gross yielding. He also concluded that gross yielding through the wall during
a hydrostatic test should be considered an impairment of serviceability. From the burst tests,
Rodabaugh found it impossible to reliably calculate the burst pressure of a bend. Burst pressures
calculated using (Lorenz’s) elastic stresses often underpredicted the actual burst strength of
short-radius, thick-walled elbows which failed at only slightly lower pressures than those
computed for straight pipe of the same diameter, wall, and material. This was attributed to theTI Fittings and Components
fact that elbows yield locally (much as pipe with a corrosion defect) redistributing stresses in the
Process. He also found that large thin-walled, long-radius bends tended to burst at a pressure
more closely approximated as that for straight pipe divided by the Lorenz factor. While no
explanation for this was offered, one could speculate that such fittings cannot redistribute stresses
over a large enough area to alter the stress distribution required by their basic geometry to
equilibrate loads.
In evaluating the fitting standards in effect up to 1969, Rodabaugh concluded that
performance language based on proof testing to a burst pressure atleast equal to that computed
for matching straight pipe, and withstanding a field hydrotest to 100 percent of SMYS in the
attached pipe without impairment, was desirable and adequate. He also evaluated whether
minimum wall dimensions (as per Lorenz’s solution) should be specified, concluding that it
would be redundant to the performance requirements and possibly overly restrictive. He also
suggested that a mechanism be established to verify compliance to performance requirements.
‘The forging process involves heating, straining, quenching, and possibly heat treatment,
So material properties in the finished product will differ from initial properties and will vary
significantly throughout the component. It was shown in several studies by Williams for PRC“?
that high-strength fittings may exhibit large variation in local strength, toughness, or transition
‘temperature properties throughout the component and that these properties could be infetior to
the matching pipe. A purchaser should carefully consider whether a manufacturer can
consistently achieve the desired qualities in the finished product.
Induetion Bends
Relatively few reports on induction bends were found in the literature. Williams”
examined the quality and properties of five 24-inch to 42-inch induction bends fabricated from
cold-expanded pipe in a study for PRC. It was determined that significant alteration to the
mechanical properties of the original material occurs, resulting in most cases in a reduction of
Yield strength and an increase in transition temperature, The degradation of properties was
‘greatest in the unbent tangent arms. The degradation is not surprising considering that the initial
Pipe was cold-expanded. Significant irregularities in wall thickness and diameter, and ripples,
ware observed as well. In contrast to Williams’ observations, two limited studies of d-inch and
6-inch Grade B induction bends showed improved mechanical properties from the bending78 Fiutings and Components
process." Final bend properties are influenced by initial chemistry, induction heating
temperature and duration, travel speed, quench rate, tempering time and temperature, and.
magnitude of deformation. Consistent bend quality will depend on the fabricator’s ability to
maintain these factors over time.
Other papers written by manufacturers demonstrating their capabilities were found."""
‘While probably little better than sales brochures, they do generally describe the metallurgical
aspects of the process and demonstrate that induction bends can be successfully manufactured
with adequate or even outstanding mechanical strength and toughness properties in the finished
material.
Rodabaugh’s findings imply that ideally the bend would be manufactured so as to exhibit
either enhanced material strength or greater wall thickness in the finished product in proportion
to the Lorenz factor for the given bend geometry. The proposed B16 induction bend draft
standard bases its pressure rating on this requirement, though it may be unnecessary in practice,
If the ratio of bend radius to pipe radius (R/t) equals 5, the Lorenz factor is only 1.125. Hence an
induction bend with R/r25 and no attendant thickening in the intrados or enhancement of
material properties will have a pressure capacity equal to or better than equivalent pipe having a
12.5-percent wall undertolerance. This matches a common minimum requirement for some pipe.
Provided the bend is of long radius, the material in the finished product meets the minimum
strength requirements for the adjacent straight pipe, and the wall thickness in the crotch and
tangents is at least equal to that of the adjacent pipe, then the bend’s performance should at least
match the expected performance of adjacent pipe, though perhaps not the performance of the
actual adjacent pipe. This was demonstrated by the burst tests reported in References 10 and 14.
‘There, the fittings failed in the tangents about at a minimum pressure one would expect for
matching pipe, but at less than the pressure that would have been required in the proof test
required by B16.9 or SP-75.
Cold Field Bends
‘The requirement that the strength or wall thickness be increased to compensate for the
Lorenz factor is unnecessary with cold field bends because they usually subtend small bend
angles over short arc lengths such that they are effectively reinforced by the straight tangents.
‘Satisfactory performance of cold field bends in the field appears to validate this position,19 Fittings and Components
although cold bends are often made using heavier wall pipe to avoid introduction of wrinkles.
Present regulations in the U.S. state that bends must have a “smooth contour” which is often
interpreted in practice to prohibit even mild ripples.
‘The formation of mild ripples and wrinkles on the intrados of the bend becomes
increasingly likely as the pipe Dt and actual yield strength increases. A study for the PRC?
attempted to better understand how the geometry, material properties, and bending technique
interact to produce wrinkles, in response to field reports that pipe coated with fusion-bonded
epoxy (FBE) was particularly susceptible to wrinkle formation. It was thought that residual
stresses from the coating application, which involves heating, may play a role. From a practical
standpoint, it is likely that little can be done by a user to make pipe less susceptible to wrinkling,
and avoidance of wrinkles by trial and error still governs bending practices in the field.
Several studies'"*?” have indicated that mild ripples are not a concer for normal pressure
integrity, as well as pressure cycles and cyclic bending loads within limits. However, the extent
of investigation by testing or analysis was found by the ASME B31.8 Committee to be
inadequate for developing acceptance rules when it considered this issue in 1998. In fact, there is
not yet a consensus on what constitutes an acceptable mild ripple as opposed to an unacceptable
wrinkle. Interest continues in showing that mild ripples can be safely tolerated. Relief on zero-
tolerance of ripples could yield savings in construction costs through less wasted pipe and less
restrictive alignment profiles.
Fabricated Bends
Fabricated or mitered bends are no longer permitted in construction of new pipelines that
operate at stress levels greater than 30 percent of SMYS. Even without this restriction, they have
become obsolete because they cost more to fabricate than a factory bend or a cold field bend, and
they impede passage of in-line inspection tools. Nevertheless, miter (or mitre) bends remain in
service in some older systems and can be found in power and process piping as well, especially
large-diameter, low-pressure piping,
As with plain elbows, Rodabaugh™ reviewed available data on miter joints for PVRC,
including theoretical solutions» and experimental results™*”” in order to determine a miter
angle below which the joint could be treated as fit-up of a girth weld rather than as a miter. His
recommendation was more restrictive, for pipe with D/t>70, than the 3 degrees (total included710 Fittings and Components
angle) allowed by ASME B31.8. He also recommended that miters be limited to operating
pressures less than 100 psig and nonflammable, incompressible fluid contents except for vent
lines to the atmosphere. Miters will inevitably disappear from pipeline usage for cost and
operations reasons, so it is probably unnecessary to act on these recommendations.
Bends with Defects
‘Very limited information was found on the pressure integrity of bends that contain
defects. Two theoretical treatments were developed by Erdogan, et al" for the aerospace
industry. Forged butt-welding elbows containing crack-like surface defects of various sizes and
orientations were pressure tested by Reynolds™ and Schulze, et al°”. PRC sponsored a study in
which four elbows were tested, 2 containing artificial metal loss defects and 2 affected by actual
corrosion in service.” These studies appear to indicate that bends or elbows containing
corrosion could be safely evaluated using existing methods for evaluating axial flaws in straight
pipe, e.g., per ASTM STP 536, ot ASME B31G. However, the data base is quite sparse and
includes no induction bends or high-strength fittings.
Branches
Tt was already well understood by the tum of the century that an opening in a flat plate
under unidirectional loading experiences increased stresses around and near the opening. This
served as the starting point for virtually all subsequent treatments of openings in pressure vessels,
and later, pipes. The concept of locally reinforcing the opening in such manner as to replace
the metal removed for the opening was well-established by the late 1930's, though researchers
struggled well into the 1960's to understand the effect of reinforcement contour and distribution
on the distribution and magnitude of stresses. Extensive experiments were conducted involving
photoelastic models and strain-gaged full-sized steel vessels, These findings led directly to
present-day contoured nozzle designs for critical vessel applications. Some of the key documents
describing these advancements are listed as References 33 through 38 for purposes of
recognizing the individual contributors and to serve as a resource.
Few of the advancements in this area have transferred to branch design in pipe, for
several reasons. One reason is that nozzles tend to be small relative to the vessel diameter, so
much of the experimental data or theoretical assumptions do not apply where the branch is
‘greater than half the header diameter. Another reason may be that optimization of the nozzleTAL Fittings and Components
100 Subjected to In-Plane and Out-of-Plane Moment", Press. Ves. & Piping
Conf., New Orleans (June 23-26, 1988).
Fowler, J. R., Samman, M. M., and Al-Sannaa, M. S., "High-Pressure Pipe Design", PR-
201-9202, A.G.A. PRC (1993).
Fowler, J. R., and Alexander, C. R., “Design Guidelines for High-Strength Pipe Fittings",
PR-201-9320, A.G.A. PRC (1994).7-31
s1etsieg
TEST TRS TR] — TT
me
emma
em ‘eum “avenne)| peau sme tono|
x x a ser | vow [mmbonronp=L
or me
seousou
SIRS MTOM] — TT
a | x st oma mE] poy pT suey a
ey es sa aT
7 eee fee (eeeee tac men tapana sso
Sa 7
x nul x} x wera ony
A a
‘f femsens ste arteo] “out ponent
fensoducs a a "N79 ade ome se ocean
saa] aT
sxarg APL aa ran ome
’ x x] x Season ot 2a sas ae 21
ae ST a TTA —
w | x x] ox | ws tcc dey
Toma
i ‘amperes Ra a
x x | se x |x | corres | Kendo | ea eer] "OR BNeEOR| mg puede omy ace
a en
Z bir tny pera oa
x x x me} x x} x voy | mr-wodasron Jevnprs nada eee
a a anal
x x x » | x x} ox | tat | voy | cer-veiersron
T
x x x w | x x| x vow | st-wsdeuston
x x x w |x x| x vey el 7
x x x | « |x x| x vane |
ar TOTET AG
x x " 61
sa
7 : ' ee seas one,
x x wa FES maar ey
Tian a
x foo fox fox fxl «x | oa | vow | e-winsron “ay na] wraye Aree amg
‘paren
‘iar si
Coma | stra | sony |omousdea| ema |nonont amen! $9 | ssamoqannes
seem | yeas | “Aer | pong | wa | in| pose | Buemedg
‘seopsminog paw ui7-32
RT EEES SH ih WE
“par1ar9 wrk seyno pony peony sang 0)
HEMI HENNY
x x ” x] ox | 6a | oun | ismwaoun, oa
WE
Teas %
x x x x tr x x 561 rae
x x 1" x x | x | et | vow | soc virusron .
Te
aceasta |W
x x ft x] ox | oe eat,
TeUPVED WoT oa
x x ra x x |x | ae | oaven PaaS |
=
1 weet |
x we a ee
[SRT SGN Be NOG] WE
st 61 Eecerteaeg mend emg eH |
SE SR TT PHD FO TRIS SG] — TE
«px {oe pe ex va
a
«fo» | pe x « va eran
: q
x » po Lx «px ae
x [ox ton
«|» ln
xe fe { « |e fa | x xf | aw | ome
x | x x x [ox | we fw trp ens
SaraeTE TEER ar
x |x le |x x [x [om | vow Paton pmo er
saree eT at
‘Aon pms rer
x fx |e fx x] ox | wa | vow
sor uuatna
aa TERNS
: Seaoael ogee) ata
x |x | a wor | vow emmsiene| masa] asmmannanconcrasn|
a ar oy
a
mang | O20 | eer | ngrent |oenna| ea |eneommre |= | ren
wert | Sur [wt | “wat "ava | [ase |" mn | amcay | ENG
iojeawo; pus TUN7-33
510 ees
TE SarRT ESSN
ee ee ee
auto
: 1 SRT ESE
x oz en ee
= ial * x oe Air [|-DN “Sa Wargepey)
prangayy
ot SPF
: foe femeenvsie suse] pentagon
ie i i. x] x 6 ey suey duped je eeGeuy 3
*
ooo aceae |
x |x fw | x xox | wm | vow | pees
7
x fox fw x| x vov | aster tton
TTI sa
wate — |"roamaN'94] __ pmapompmaserdo ig
x |x |x fw | x x] ox | as | vor | epsemosnn | taser cons] waxige te ns wens
|11-DN 01 woday ‘sed
Tara} SRST
x | «x [x fw |e |e lel x ‘a'viment| starrpasooteas eee
Sa Treo Ts suapsaua aun Aaa ——Ty
eee eae eg x| x
SAT
x | x | « [ow | x | « [etx ‘oomnipnea eee
ee ely x] ox | esa fotmginn | ears ta z
Sein i
were " i
x « | ox | ow x] ox | om | oavsn | austen
gore sn
a TSI) AE
mum egal po
Sana vomes ama
x |x fw x |x | wea | ome | er-emecoun
x | « |x | w |x x] x | on | oma
x ee ce x |x | em | oma
@
tee | any | O2me | tea | rupee [senndsa| sna sme empeaea| eoy ag
Samoa | Semon | Sot | comms | atea’ | at fy [in| Tesmn | aeegs | MERON
mal
‘wopDoRNED pa STeH
we7-4
stp yefeg
eae ean See oat ec eee ote oe
namietiee
a0 a
x x 4 u x ¥66t OWAd 26E BND NNG UA ‘Burda 105 sxoroe4 voneayysuary|]
pe
: SS Ee
: : : al 7 2 IM soa ans 20
= = x i = ame a “tapriodling TAD] TauRsarti pa voc Bardia) — oF
: wet =
= Oe lew lak eve see see rnin aver | SS
ee ete ee
ae uae
x x had fe bad fe _o suuor pedals ‘penton 301
x [aa ae oa al
a ‘ipo oma er sa
x |x fox [x | x] x | on eeeaan pn ees og
vest
x |x |x fxw foo xf ox | ase ETE
Cpsmmeit| Seon | Came | tea | gps |seder evanog ins
‘ram samen | Sram | woes | “Aa” [na7236 Fittings and Components
Reference 1
Rodabaugh, E. C., Duffy, A. R., and Atterbury, T. J., “The Internal Pressure Capacity of
Butt Welding Elbows”, NG-18 Report 22, American Gas Association, 63 pages
(September 18, 1969) .
‘Theoretical solutions for curved pipe and physical tests on forged bends and butt welding
elbows performed by fitting manufacturers or other researchers addressing pressure capacity
were reviewed in detail. Fourteen references are cited. The study examined the theoretical
distribution of elastic stresses in ideal curved pipe, the actual distribution of strains in as-
manufactured bends, the stiffening effects of tangent
bends), the pressure-yield characteristics, and burst strengths. The contemporary fitting standards
and pipeline codes were reviewed and suggestions made to improve assurance of adequate
performance of bends in use.
Rodabaugh concluded that elastic membrane stresses and the pressure that causes
localized through-wall yielding can be calculated with adequate accuracy using Lorenz's
theoretical solution provided the bend subtends an angle greater than an attenuation angle based
on shell theory. Where the bend subtends a narrower angle, the estimated yield pressure will be
conservative, The burst pressure of a bend cannot be reliably predicted on the basis of elastic
stresses, but where the elbow is thin-walled and subtends an angle much larger than the
attenuation angle, the actual burst pressure approaches that obtained from calculated elastic
stresses,
Tr was recommended that the design of bends and elbows be governed by performance
criteria rather than specific dimensional or material property criteria. Specifically, the component
specifications, ASME/USAS (now ASME/ANSI) B16.9, and MSS SP-48 and SP-63 (superseded
by SP-75), should state that elbows be designed to withstand a field hydrostatic test to 100
percent of SMYS in the mating (“designated”) pipe without failure, leakage, yielding through the
wall, or impairment of serviceability, and that the component be designed so that actual burst is
at least equal to the calculated burst pressure of the designated pipe. Additional recommendations
were made for revisions to the contemporary version of ASME (USAS) B31.8 to remove
Janguage that could be interpreted to impose unrealistic requirements. Most of Rodabaugh’s
recommendations appear to have been adopted by the various standards in their present forms.
pipe on narrow angle bends (e.g 12 degree731 Fittings and Components
Reference 2
Lorenz, H., “Theorie der Rohrenfedermanometer” (1910) (not reviewed herein).
Reference 3
Kalnins, S., “Analysis of Curved thin-Walled Shells of Revolution”, AIAA Journal, Vol.
6, 4 pages (1968).
Reference 4
Clark, Gilroy, and Reissner, “Stresses and Deformations of Toroidal Shells of Elliptical
Cross Section”, Journal of Applied Mechanics, Vol. 74, 11 pages (1952)
References 2, 3, and 4 made significant contributions to the understanding of the behavior
of curved pipe under internal pressure. Lorenz developed a theory for the distribution of
membrane stresses due to pressure in a pressure gage Bourdon tube in 1910, of which curved
pipe of circular cross section is a special simplified case. Lorenz showed that the elastic
circumferential (hoop) membrane stress is greater on the inside of the bend (intrados), and less
on the outside of the bend (extrados), relative to straight pipe of the same wall thickness. On the
inside of the bend, the hoop stress is calculated as,
= Pr[Ra-2
t| Rid
oy
where ris the pipe radius (D/2),t is the pipe wall, and R is the radius of the bend at the neutral
axis. On the outside of the bend, the hoop stress is calculated as
_ Pr nea]
Det
For a standard elbow with a ratio of bend radius to pipe radius (R/t) of 3, the Lorenz factor is
1.25 on the intrados, and 0.875 on the extrados. For a very long-radius bend as might be the case
With ficld bends or induction bends in line pipe, the Lorenz factors on both the inside and outside
of the bend converge toward a value of 1. Lorenz’s basic results have been validated by
numerous pressure tests on elbows, though they do not include end effects of the straight pipe
tangents at each end of the elbow, which tend to reduce the effect of pipe curvature in the bend7-38 Fittings and Components
within a decay or attenuation length. In 1968 Kalnins was able to precisely evaluate these effects
ina shell of revolution which is rotationally symmetric about a curve using a special computer
program which solves a system of partial differential equations derived from shell theory.
Lorenz’s solution also did not describe the local discontinuity bending stresses that arise at the
neutral meridian which were determined theoretically by Clark, Gilroy, and Reissner, nor the
bending stresses due to rerounding of the elliptical cross section of as-manufactured elbows. For
the latter no general solution exists though approximations from straight pipe were deemed
sufficiently accurate,
Reference 5
Gross, N., “Experiments on Short Radius Pipe Bends”, Proc. Inst. Mech. Engrs., Vol. 1B,
24 pages (1952).
Gross was one of many researchers who conducted tests on pipe bends in order to
compare their strengths and flexibility characteristics under various loading conditions. Bends
‘were nominally 6.625 inch in diameter with 0.255 to 0.323 inch walls, of seamless or welded
construction. Materials were carbon steel similar in strength and chemistry to AS3 Grade B or SL
Grade B. The bend radius was 9 inches, so R/=3 approximately. The membrane and bending
siress distributions throughout the elbows subjected to intemal pressure and to applied bending
loads were determined using an array of strain gages on the inside and outside of the elbows.
Reference 6
Williams, D. N., “Examination of the Properties of a Quenched-and-Tempered Fitting”,
NG-18 Report No. 114, American Gas Association, 24 pages (April 17, 1978).
Reference 7
Williams, D. N., “Examination of the Properties of a Y65 Quenched-and-Tempered
Fitting”, NG-18 Report No. 118, American Gas Association, 21 pages (March 23, 1979).
Reference 8
Williams, D. N., “Evaluation of the Properties of High-Strength Large Diameter Pipeline
Fittings”, NG-18 Report No. 123, American Gas Association, 42 pages (May 8, 1981).7239 Fittings and Components
Detailed measurements of dimensions and material properties were made on several large
pipeline fittings. Reference 6 studied a 30-inch OD x 0.625-inch WT size-on-size straight tee
with 60 ksi minimum strength (Y60 grade). Reference 7 studied a 24-inch OD x 0.500-inch WT
size-on-size straight tee with 65 ksi minimum strength (Y65 grade). Reference 8 studied a 30-
inch OD size-on-size Y60 straight tee and a 30-inch 45-degree Y52 elbow. Some of the results
from References 6 and 7 are summarized in Reference 8.
Each fitting was found to exhibit large variation in strength and toughness properties
depending on both position and orientation in the component. Impact properties were affected by
the initial rolling direction of the plate material used to form the fitting. The fittings in
References 6 and 7 were supplied in the quenched and tempered condition. Coupons were tested
both as-received and stress-relieved. The stress relief had little effect on strength, but increased
the toughness transition temperature, The fittings in Reference 8 were supplied in the normalized
condition and exhibited more uniform properties than the quenched and tempered fittings. In
some cases, measured properties did not fall within the range of values reported by the
‘manufacturer, and were less than specified minimum levels in a few tests. Also, extensive weld
repairs in excess of what was allowed by MSS SP-75 were discovered in one tee, and extensive
surface grinding to repair mechanical damage was discovered in the elbow.
Reference 9
Williams, D.N., “Evaluation of the Properties of Induction Hot Bends Fabricated from
High-Strength Cold-Expanded Pipe”, NG-18 Report No. 142, American Gas Association,
86 pages.
Detailed measurements of dimensions and material properties were made on five
induction bends (two 30-inch OD, X65; one 24-inch OD, X52; one 24-inch OD, X60; and one
42-inch OD, X70). Material properties from the starting pipe were also measured. All five
exhibited significant changes in strength levels from the initial pipe as a result of the bending
process. In all cases but one, yield strength decreased in both the bend and the unbent tangents
and was less than the minimum specified in several tests. The transition temperature was
increased in the unbent section in all but one case, but went either way in the bend. Cyclical
variations in wall thickness and diameter, and ripples were observed in one of the bends.7-40 Fittings and Components
Williams concluded that it cannot be assumed that all pipes in a specific lot of pipes will
respond identically to the induction bend process. Minor pipe-to-pipe variations in chemistry can
significantly affect the outcome, Differences in temperature, travel speed, and quench rate
inevitably occur around the pipe circumference at any point in the bend which then affect
properties around the circumference. Perliaps most important is the need to carefully control.
‘these factors throughout the bend forming process, which can take up to an hour, and from bend
to bend and shift to shift. In two cases, adverse effects on the dimensions and properties were
blamed by the bend fabricator on malfimctions in the process temperature control during
bending. However, the imperfections observed in the final product are not easily identified
during manufacture, and there is no reason to believe that any of the bends could not have found
their way into service.
Reference 10
Rodabaugh, E. C., and Woods, G. E., "Fatigue, Moment Capacity and Burst Tests of )
Induction Bends", unpublished reports to two bend manufacturers, 35 pages (July 1992);
and 43 pages (April 2000).
This citation refers to two separate reports, having the same authors and titles, describing
the results of tests conducted for two separate bend manufacturers, in which the performance of
their induction bends were compared to Standard B16.9 forged elbows. The reports are
‘unpublished. The test specimens used in both test programs were 1.5D bends (R/r=3) made from
NPS 40 and 6 NPS 40 A106 Grade B seamless pipe having initial wall thicknesses about 5
percent heavier than specification. The wall thickness through the intrados was increased 10
percent in one manufacturer's specimens and over 70 percent in the specimens from the second
‘manufacturer. The wall thickness through the extrados was decreased by 7 to 13 percent from
the initial wall in all the specimens. Yield and tensile strengths were increased through the
induction bend compared with the unbent pipe in all specimens,
‘The induction bends performed better in all regards than the forged elbows tested in these
studies or as reported on in prior work. ‘This was attributed to the increased wall thickness and
‘material strength at the intrados of the bends. The performance of the bends was generally
consistent with theoretical relationships when adjustments were made for actual wall thickness7-41 Fittings and Components
and material strength. However, none of the burst test specimens, including the forged elbows,
met the minimum proof test pressure specified in B16.9, though the induction bends failed in the
straight tangent legs, not the bend itself. The results are representative of small diameter
Grade B bends produced by two manufacturers, and cannot necessarily be extrapolated to large
diameter bends made from high-strength pipe produced by other manufacturers.
Reference 11
Hashamoto, T., Sawamura, T., Komizo, Y., Nakate, H., and Nakatsuka, Y., “High
Strength Bent Pipe for Arctic Use”, Paper 15, NG-18/HLP Joint Technical Meeting on
Line Pipe Research, Columbus, OH, 24 pages (September 20-21, 1984).
Reference 12
Aoki, N,, Fuji-i, T., Miyamoto, H.,and Hidaka, Y.., “High-Tensile HF-ERW Pipe Bends
for Arctic Use”, ASME Pipeline Engineering Symposium, 10" Energy-Sources
Technology Conference, Dallas, PD-Vol. 6, 6 pages (1987).
Reference 13,
Graf, MK., Hillenbrand, H. G., and Niedethoff, K. A., “Production of Large Diameter
Linepipe and Bends for the World’s First Long-Range Pipeline in Grade X80”, Paper 5,
Proc. & Symposium on Line Pipe Research, A.G.A. PRC, Houston, 14 pages (September
26-29, 1993).
Reference 14
Kondo, J., Nagae, M., Ume, K., and Hirano, O., “The State of the Art of High Strength
Induction Bent Pipe”, Paper 6, Proc. 8 Symposium on Line Pipe Research, A.G.A. PRC,
Houston, 14 pages (September 26-29, 1993).
References 11 through 14 describe the metallurgical processes involved in forming
induction bends, and describes the finished product in terms of mechanical properties and
material chemistry. The papers demonstrate the manufacturers’ abilities to produce a finished
component that is compatible with high-strength, high-toughness line pipe. Considering the
sources of information, the reports should be regarded as little more than sales brochures, but
they do indicate that induction bends can be produced with desirable properties.7-42 Fittings and Components
Only Reference 14 gives the results of an actual performance test. The nominal
dimensions were 24-inch OD * 0.500-inch WT with a SD radius, though the actual wall
dimension in the crotch of the bend was not reported, The Lorenz factor for this geometry would
be around 1.05. The bend was designated as X60. The reported yield and tensile strengths of
material in the intrados was 64.5 ksi and 83.2 ksi, respectively, and in the tangent arm 65.0 ksi
and 76.9 ksi, respectively. The fitting would have yielded at a pressure corresponding to a stress
of 61.4 ksi (64.5/1.05) in straight pipe, so it would have withstood a high-pressure field
hydrostatic test without gross deformation. The fitting failed in the tangent at a pressure of 3,278
psig, about what one would expect for 24-inch OD x 0.500-inch WT pipe meeting the minimum
requirement of SL X60. Although this was 11 percent less than the pressure that would have been
required in a proof test in a B16.9 or SP-75 standard component, it certainly demonstrated
adequate performance of the finished product.
Reference 15
Williams, D. N., “Investigation of Field Bending Problems in Line Pipe”, Paper 16, NG-
I8/HLP Joint Technical Meeting on Line Pipe Research, American Gas Association,
Columbus, OH, 24 pages (September 20-21, 1984).
Field reports indicated increased susceptibility to formation of wrinkles during cold
bending of pipe coated with fusion-bonded epoxy. Materials tests and wrinkling tests were
performed on numerous samples of 8-inch NPS « 0.172-inch WT X42, 12-inch NPS x 0.203-
inch WT X42, 12-inch NPS x 0.250-inch WT X52, and 24-inch OD x 0.375-inch WT X52.
Discontinuous yielding versus progressive strain hardening was ruled out as a factor. Residual
stresses introduced by the coating process, which involves heating, were considered a potential
contributor to the problem. No specific evaluation of the effect of wrinkles on pipe integrity was
made.
Reference 16
Murray, N. W., and Bilston, P., “Rational Acceptance Limits for Field Bends in Oil or
Gas Pipelines”, Paper V-2, Proc. CANMET Intl. Conf. on Pipeline Reliability, Calgary,
13 pages (June 2-5, 1992),7-43 Fittings and Components
‘Modes of deformation that can occur during field bending are reviewed, including
ovalization, wall thickness variation, wrinkling or rippling, nonuniform bend radius (kinking).
Variation in mechanical properties and residual stresses will also occur. Fairly simple analyses
were used to suggest some limits on deformation. The adverse effect of wrinkling on coating
integrity is recognized, but no specific evaluation of the effect of wrinkles on pipe integrity was
made.
Reference 17
Bilston, P, and Murray, N. W., “Evaluation of Factors Affecting Field Bending,
Characteristics of Line Pipe”, Paper 6, PRC/EPRG 9" Biennial Joint Technical Meeting
on Line Pipe Research, Houston, 17 pages (May 11-14, 1993).
Reference 18
Bilston, P, and Murray, N. W., “The Role of Cold Field Bending in Pipeline
Construction”, Paper 27, Proc. & Symposium on Line Pipe Research, A.G.A. PRC,
Houston, 19 pages (September 26-29, 1993).
These two papers describe the same body of work. Elastic buckling theory was used to
predict a buckle wavelength. Actual buckle wavelength was somewhat less due to friction from
contact against the bending die. A prototype buckle waveform was modeled from experiments in
order to develop post-buckling analyses. This led to relationships between strain, bend
shortening, and buckle height, and relationships between maximum buckle height and die radius
and residual angle which agreed reasonably well with limited tests. Tests were made on 3.5-inch
and 16-inch OD pipe. A bend on 16-inch OD x 0.25-inch WT X42 pipe containing mild ripples
0.21 inch high was pressurized to a stress level of 115 percent of SMYS. Strain in the wrinkled
area was only 0.4 percent. The pipe burst at 140 percent of SMYS in an area remote from the
wrinkles. The results of this work evidently led to acceptance of mild ripples up to 50 percent of
the wall thickness in the Australian pipeline code.
Reference 19
Olson, R., Clark, T., and Odom, T., “Evaluation of the Structural Integrity of Cold Field-
Bent Line Pipe”, Paper 6, Proc. EPRG/PRC 10* Biennial Joint Technical Meeting on
Line Pipe Research, Cambridge, 13 pages (April 18-21 1995).7-44 Fittings and Components
Reference 20
Olson, R., Clark, T., and Odom, T., “Evaluation of the Structural Integrity of Cold Field-
Bent Line Pipe”, Paper 33, Proc. 9" Symposium on Pipeline Research, Houston, 20
‘pages (September 30 1996).
This PRCI-sponsored program was undertaken to establish criteria for bends containing,
ripples that assure that they will withstand a hydrostatic test in the field to 115 percent of SMYS,
daily pressure variations of 10 percent of SMYS, and altemating longitudinal stresses due to
seasonal temperature changes. Tests were conducted to apply unrealistically large loadings on
bends containing wrinkles that would probably be rejected. Survival would demonstrate that less
severe deformations under less severe loadings could be safely tolerated. The first paper
describes only the test program, while the second paper covers the tests in an identical manner
and also addresses analytical efforts. The reproduction of photographs of test specimens is
superior in the first paper.
‘One test specimen was a 30-inch OD * 0.300-inch WT X70 pipe with one large wrinkle
1.5 times the wall, and several smaller ripples. The second specimen was 36-inch OD x 0.385-
inch WT X65 pipe with ripples of similar magnitude to the first specimen. Both were coated with
FBE. The first specimen was hydrotested to 100 percent of SMYS, then subjected to cyclic 4-
point bending to high longitudinal stress levels until fatigue cracking developed through the wall
in the circumferential direction. The second specimen was hydrotested to 108 percent of SMYS,
then subjected to cyclic 4-point bending to high longitudinal stress levels. While it grew fatigue
cracks in the girth welds, it did not develop a fatigue failure in the ripples.
The analytical model developed by Bilston in References 17 and 18 was applied to
develop estimates of the cyclic strain levels which agreed well with strain gage readings. These
‘were then used in a strain-life analysis which did not do a particularly good job of predicting the
observed fatigue life except to qualitatively indicate that fatigue life would be expected to
decrease with increasing wrinkle severity.
Reference 21
Kiefer, J. F., and Alexander, C. R., Addendum to API Publication 1156, “Effects of
‘Smooth and Rock Dents on Liquid Petroleum Pipelines, Phase 2” (to be published).7-45 Fittings and Components
Three samples of 36-inch OD x 0.281-inch WT X52 line pipe were cold bent without a
mandrel to create intentional wrinkles of 1.7, 3.7, and 6.9 percent of the OD in depth. The
internal pressure was cycled to produce a range of hoop stress between 12 and 84 percent of
SMYS. The 6.9-percent wrinkle failed by fatigue in 1,086 cycles, the 3.7-percent wrinkle failed
in2,791 cyéles, and the 1.7-percent wrinkle survived 44,541 cycles without failure or cracking,
‘No attempt was made to derive an effective SIF or i-factor in terms of present Code allowables.
Reference 22
Rodabaugh, E. C., “Review of Data on Mitre Joints in Piping to Establish Maximum
Aolasty for Fabrication of Girth Butt Welds”, WRC Bulletin 208, 22 pages (August.
‘The purpose of this study was to define a miter angle below which the joint can be
considered equivalent to a girth butt weld for purposes of fit up rather than as a miter. Secondary
objectives were to review the validity of ASME Code equations for the stress intensification
factors used in flexibility analyses, and to review data on stresses in miters due to intemal
pressure, on which the Code gives no guidance. Both theoretical solutions and experimental data
are reviewed from 23 references, ¢.g., References 23 through 27 herein,
‘Stress concentration effects under intemal pressure were determined theoretically and
experimentally to increase with D/t and miter angle. Concentration factors of 1.5 to 8.7 on the
nominal pressure hoop stress were determined experimentally in some configurations. Fatigue-
effective stress concentration factors from pressure-fatigue tests were determined to be between
2.5 and 4.25 in relation to the strain-controlled fatigue behavior of polished round bars.
Stress concentration effects under moment loading were determined theoretically and
experimentally to increase with ratio of miter spacing to pipe radius, and with miter angle. The
Code flexibility and stress intensification factors were deemed adequate.
Rodabaugh recommended a maximum “fit-up” angle of not more than 9(U)” in degrees
(per side of the joint, or double this value as an included angle), where r=D/2 nominally. For pipe
with D/>70, this is more restrictive than ASME B31.8 which permits fit-up angles of 3 degrees.
He also recommended limitations on the use of miter bends (in ASME B31.1) to situations where
the pressure is less than 100 psig, the fluid is nonflammable and incompressible except in vents746 Fittings and Components
to the atmosphere, the number of lifetime full-pressure cycles is Jess than 7,000, the included
miter angle is less than 45 degrees, and full penetration welds are used. These suggestions are
also more restrictive than the requirements in ASME B31.8.
Reference 23
Green, A. E., and Emmerson, W. C., “Stresses in a Pipe with a Discontinuous Bend”, J.
Mech. Phys. Solids, Vol. 9, 14 pages (1961).
Reference 24
Murthy, M. V. V.., “Stresses at the Intersection of Two Cylindrical Shells of Equal
Diameter”, Jnt. J. Solid Structures, Vol. 5, 14 pages (1969).
Reference 25
Bond, M. P., and Kitching, R., “Multi-Mitred and Single-Mitred Bends Subject to
Internal Pressure”, Int. J. Mech. Sci., vol. 13, 18 pages (1971).
Reference 26
Lane, P. H. R., and Rose, R. T., “Experiments on Fabricated Pipe Bends”, Brit. Welding
J. Sune 1961).
Reference 27
Macfarlane, D. S., “Fatigue Strength of Gusseted Pipe Bends”, Brit. Welding J.
December 1960).
References 23 through 27 are discussed in Reference 22, but are listed for recognition of
their contribution. Green and Emmerson developed simplified theoretical estimates for the
stresses due to internal pressure in miter bends. Murthy and Bond and Kitching developed
successive refinements, Bond and Kitching’s theory was found to be in agreement with test
results reported by Bond and Kitching, Lane and Rose, and Macfarlane in most cases, but it gave
unconservative results in some others. The results of Green and Emmerson were always
conservative.TAI Fittings and Components
Reference 28
Delale, F., and Erdogan, F., “The Crack Problem in a Specially Orthotropic Shell with
Double Curvature”, Eng. Fract. Mech., Vol. 18, 15 pages (1983).
Reference 29
Joseph, P. F., and Erdogan,
General Loading Conditions
1987).
“Plates and Shells Containing a Surface Crack Under
”, NASA Contractor Report CR-178328, 381 pages (July
“The first report gives factors on the crack-tip stress intensity of a surface flaw in a
shallow toroidal shell with unequal curvatures, derived from a transverse shear theory. The
second document derives the crack-tip stress intensity of a surface flaw by adapting the line-
spring model to the shell curvature. They are of interest in that they offer solutions applicabie to
pressurized bends containing a crack or corrosion defect. Both documents are theoretical in
nature, Results are compared to theoretical solutions for other geometries such as straight
cylinders. Accounting for the double curvature results in a modification of the bulging factor
‘compared to that for straight pipe derived by Folias. The bulging factor is reduced in the intrados,
though the membrane stress due to pressure is greater. The bulging factor is increased in the
extrados, though the membrane stress there is lower. These are the result of the degree of
restraint produced by the shell curvatures. The changes are on the order of 5 percent for typical
elbow geometries.
Reference 30
Reynolds, M. B., “Failure Behavior of Flawed Carbon Steel Pipes and Fittings”, General
Electric Report GEAP-10236, Prepared for US Atomic Energy Commission, Contract
AT(04-3)-189, Project Agreement 37, 28 pages (October 1970).
Reynolds tested A106 Grade B pipe and matching elbow and tee fittings containing
various artificial defects. Pipe sizes ranged from 4 NPS to 12 NPS. Tees and elbows were all 6
‘NPS 80. An unflawed elbow and tee were pressurized to failure to establish a baseline. Slit-like
flaws of various lengths, depths, and orientations were introduced in pipe, elbow and tee
specimens, Pipes with axial flaws were pressurized to failure, while those with circumferential
flaws were loaded in bending to failure. Elbows were pressure tested with longitudinal flaws on7-48 Fittings and Components
the extrados and on the crown (mean bend radius), and with circumferential flaws centered on.
the crown. The defects located on the intrados of the bend had the largest adverse effect. Tees
‘were pressure tested with axial flaws located at the “hillside” position, and either axially or
transversely in the crotch. The burst pressures of the tees were unaffected by the transverse
defects at the crotch, but were affected by the other two defect arrangements.
Reference 31
Schulze, H. D., Togler, G., and Bodman, E., “Fracture Mechanics Analysis on the
Initiation and Propagation of Circumferential and Longitudinal Cracks in Straight Pipe
and Bends”, Nuc. Engrg. and Design 58, 13 pages (1980).
Tests were carried out on 2.5-inch OD 0.138-inch WT pipe and bends fabricated from a
low-alloy structural steel (0.16 C, 0.69 Mn, and 0.31 Mo) with measured yield and tensile
strengths of 48 and 71 ksi, respectively. Some of the pipes were special internally-finned types,
but several bends were of normal configuration with R/m3. Bends with slit-like longitudinal
defects of various length and depth located on the outside arc (extrados) of the bend were
pressurized to failure. Some tests superimposed a bending load producing a stress of 46 ksi. Tests
were conducted at room temperature and at 500 F. The bending load appeared to have no
influence on the failure pressure. It was concluded that existing theories for longitudinal flaws in
straight pipe could be applied to defects in elbows, provided the defect position and bend
‘geometry are accounted for in determining the hoop stress.
Reference 32
Bubenik, T., and Rosenfeld, M. J., “Assessing the Strength of Corroded Elbows”, NG-18
‘Report 206, American Gas Association, 31 pages (May 1993).
‘This study for the PRC was undertaken to evaluate whether criteria for evaluating
corrosion in straight pipe could be applied to bends. The theoretical results from Erdogan, et al
%) and the test results reported by Reynolds™ were reviewed. Four elbows with flaws were
pressure tested. Two were 12-inch Grade B elbows with machined metal-loss defects, one on the
intrados and the other on the extrados. The other two elbows were removed from pipeline service
because they contained corrosion. One was a 4-inch elbow with a short defect 70 percent of the7-49 Fittings and Components
wall in depth, located on its crown. The other was 6-inch elbow with a long defect 34 percent of
the wall in depth, located on the intrados.
‘The burst strengths of these and Reynolds flawed elbows were evaluated using ASME
B31G with the standard straight-pipe bulging factor, with or without Lorenz’s stress factor, and
considering failure at the flow stress and at the material ultimate strength (actual, not specified
minimum). Those made using the Lorenz stress factor (which could be 1 or less for some defect
locations) and the actual ultimate strength rather than the flow stress gave the best predictions
with least scatter, though 2 of 10 results were unconservative by up to 10 percent. But, it was
also shown that simply using the flow stress and no Lorenz factor was conservative. Thus, it
appears that ASME B31G could be safely applied to corrosion in a bend.
Reference 33
Waters, E. O., “Reinforcement of Openings in Pressure Vessels”, Weld. J. Res.
Supplement, 12 pages (June 1958).
This report provides a concise summary of theoretical and experimental research and
development in the design of nozzles in pressure vessels and pipe, focusing on developments
since 1930. Waters specifically recognized and discussed the unique aspects of openings in
pipelines as compared to those in pressure vessels, and discusses the studies at Battelle and Tube
‘Tums funded by A.G.A.
Reference 34
WRC Bulletin 51, June 1959,46 pages, containing: Waters, E. O., “Theoretical Stresses
Near a Circular Opening in a Flat Plate Reinforced with a Cylindrical Outlet”;
‘Hardenburgh, D. E., “Stresses in Contoured Openings of Pressure Vessels”; and Taylor,
C.E., Lind, N. C., and Schweiker, J. W., “A Three-Dimensional Photoelastic Study of
Stresses Around Reinforced Outlets in Pressure Vessels”.
Reference 35
Mershon, J. L., “PVRC Research on Reinforcement of Openings in Pressure Vessels”,
‘WRC Bulletin 77, 54 pages (May 1962).7-30 Fittings and Components
Reference 36
Hardenburgh, D. E., Zamrik, S. Y., and Edmondson, A. J, “Experimental Investigation
of Stresses in Nozzles in Cylindrical Pressure Vessels”, WRC Bulletin 89, 35 pages (July
1963).
Reference 37
WRC Bulletin 113, April 1966, 70 pages, containing: Taylor, C. E., and Lind, N. C.,
“Photoelastic Study of the Stresses Near Openings in Pressure Vessels”; Leven, M. M.,
Photoelastic Determination of the Stresses in Reinforced Openings in Pressure Vessels”;
and Mershon, J. L., “Preliminary Evaluation of PVRC Photoelastic Test Data on
Reinforced Openings in Pressure Vessels”.
Reference 38
Rodabaugh, E. C., Atterbury, T. I., Cloud, R. L., and Witt, F. J., “Evaluation of
Experimental and Theoretical Data on Radial Nozzles in Pressure Vessels”, U. S. Atomic
Energy Commission, TID-24342, 585 pages (March 31, 1966).
References 34 through 37 are key papers in the development of current understanding of
how the contour and distribution of reinforcement affects the magnitude and distribution of
stresses in the vicinity of openings in vessels. Rodabaugh, in Reference 38, provides a detailed
history, discussion, and evaluation of theoretical developments and compares results from theory
to experimental data.
Reference 39
Del Buono, A. J., Visset, P. L., and Williams, F. S. G., “Design of Hot Tap Connections
in High Pressure Pipe Lines", ASME 53-PET-31, 1953, 22 pages, and Petr. Engr., Vol.
26, No. 2 (1954).
This study clearly explained the shortcomings of design rules for openings in pressure
vessels in meeting the needs for high-pressure pipelines. The results of tests on full scale model
vessels conducted by Taylor Forge were described. The authors recommended specific design
details for branch connections in pipelines, namely increased wall thickness in the run (giving a
lower operating stress), use of contoured forged tees or full-encirclement reinforcement,
increased thickness in the branch, proper proportioning of the cover fillet at the base of the
branch connection, tapering of the reinforcement edges to avoid excessive weld size and to7-51 Fittings and Components
control stress concentrations, and avoidance of cracks, unfused areas, and sharp comers (at the
fillet toes) in the weld. The rules for branch design in the present edition of ASME B31.8 reflect
all of these recommendations.
Reference 40
Markl, A. R. C., George, H. H., and Rodabaugh, E. C., “Pressure-Pulsation Tests of
Branch Connections to Large-Diameter Pipe”, 4.G.4. Gas Supply, Transmission and
Storage Conference, Pittsburgh, 19 pages (May 9-10, 1955).
Reference 41
Markl, A. R. C., George, H. H., and Rodabaugh, E. C., “Why Branch Connections Fail”,
Pipe Line Ind., 7 pages (July 1955).
Reference 41 is a condensed version of Reference 40. Reported are the results of pressure
pulsation tests on 9 styles of branch connection. The test specimens were 22-inch OD x 0.312-
inch WT DSAW X52 run pipe with a 10-inch branch, as well as several models scaled to 12.75-
inch OD x 0.188-inch OD A106 run pipe. The pressure was cycled to produce a hoop stress
fluctuating between 50 and 90 percent of SMYS. Tests were run on (A) plain capped pipe as a
baseline, (B) an unreinforced branch, (C) an ordinary saddle, (D) a full-encirclement sleeve, (E)
an encirclement sleeve plus saddle, (F) an encirclement saddle with side seams, (G) an
encirclement saddle with seams on the crotch and back, (H) an encirclement saddle with bolted
side flanges, (J) an encirclement tee, and (K) a forged butt-welding tee. Three to five specimens
of each type were tested.
The result were ranked in order of demonstrated performance, normalized to the
performance of a saddie-reinforced tee (Type C) because it was the design all of the other branch
designs were intended to improve upon. Results were reasonably consistent for a given type of
construction. Failures attributed to welding quality were eliminated by repairing the weld and
continuing the tests. Thus the results were governed by basic geometry.
Ranking given by the authors were, by type and life factor: Type A, 5 to 15; Type K, 6 to
8; Types G and E, 6 to 15; Types H, F, and D, 0.7 to 7; Type C, 0.7 to 1.4; Type J, 0.09 to 2.5;
and Type B, 0.06 to 0.08. The poor performance of the encirclement tee (J) was unexpected.7-52 Fittings and Components
Reference 42
Rodabaugh, E. C., and George, H. H., “Design and Strength of Welded Pipe Line Branch
Consetions, Proc. ASCE, J. of the Pipeline Div., Vol. 83, No. PL1, 32 pages (March
General loadings on the branch, Code design rules, and field failures of branches are
discussed. Results are reported in detail from numerous tests performed at Tube Tums and by
others on static and cyclic pressure loading, and static and cyclic external loads on branches.
Pressure pulsation tests were performed by the authors on 24-inch OD x 0.500-inch WT header
pipe with 4-inch, 8-inch, or 12-inch extruded outlets (Type L, as extension of the nomenclature
from References 40 and 41), and with the same size fabricated branches with saddle
reinforcement (Type C) and circular pad reinforcement (Type M). The cyclic pressure
performance ranges of Types L and M branches were 6 to 10 and 0.5 to 1.7, respectively. The
additional Type C branches fell in the same range as the earlier series of tests.
Cyclic bending load tests were performed on branches having a various forms of
reinforcement, including with the run thickened, the branch thickened, saddle, circular pad in
ifferent thicknesses, and triform ribs. Stress intensification factors ranging between 1.1 and 4
were developed in accordance with Mark!'s fatigue S-N failure law, iS=245,000/N®? (refer to
Markl paper, Reference 61). It was concluded on the basis of the test data and field experience
that branch connections with integral reinforcement, large intersection radii, gradual changes in
wall thickness, and preferential thickening in highly stressed areas were preferred. Where
branches had to be of the fabricated type, full-encirclement reinforcement was recommended for
sovere service conditions, while saddles or pads were acceptable for moderate service. It was also
‘emphasized that quality of fabrication in terms of fit-up and especially welding could affect
performance of a fabrication.
Reference 43
Atterbury, T. J., Beall, L. G., McClure, G. M., VerNooy, B., and Battisto, S.,
“Experimental Stress Analysis of Several Full-Opening Reinforced Branch Connections”,
Special Report to NG-11 Branch Comnection Program, American Gas Association
(Wevember 30, 1958).7-53 Fittings and Components
Eight styles of fabricated reinforced 16-inch size-on-size tees (except one 8-inch branch) --
were studied to determine stresses under the action of intemal pressure or transverse bending
load on the branch. Pressure pulsation tests were conducted to determine pressure fatigue
performance. The run pipe was 16-inch OD x 5/16-inch WT API SL X52 pipe. The fabricated tee
styles suitable for hot-tapping included a cast steel split tee, a full encirclement saddle, a forged
split tee, 2 weld-o-let on thick-walled split sleeve, a weld-o-let on thin-walled split sleeve, a 16-
inch saddle, and an 8-inch saddle. Results were then compared to those from a forged butt
welding straight tee, which represents the preferred branch form though it cannot be used for hot-
tapping.
Stress intensification factors (SIFs) under internal pressure or extemal load were derived
from strain gage measurements. For internal pressure, the forged butt welding tee had the lowest
SIF, followed closely by the split forged tee and the cast steel tee. These three also exhibited the
lowest SIFs due to transverse bending load applied at the branch hot tap flange. The others
exhibited SIFs 2 to 3 times greater. Those with thinner branch walls also did not redistribute or
shed stresses very well after initially yielding.
‘The pressure pulsation tests cycled the run pipe between 50 and 90 percent of SMYS. The
butt welding fitting exhibited 4 to 40 times longer pressure pulsation fatigue life than the
fabricated branches. In general, the fatigue failures occurred at the points of maximum static
pressure strain, It was also determined that fabrication weld quality significantly affected fatigue
performance as well, but welding was eliminated as a factor by making weld repairs and
continuing with the tests.
It was concluded that those fittings with smoothly contoured openings and no fillets
around the opening gave the superior performance. Also, it was observed that the fillet welds at
‘the ends of the sleeves were potential fatigue concerns. Finally, those fittings having thicker
‘branch walls had lower SIFs under transverse bending loads than those with thinner walls.
Reference 44
Atterbury, T. J., McClure, G. M., Roos, C. HL, and Grover, H. J., “Branch Connections,
Report 2: Survey of Research”, A.G.A. NG-11 Report, American Gas Association, 78
pages (January 1959).7-54 Fittings and Components
Results from work to date sponsored by A.G.A. and others, some previously unpublished,
were compiled in tables giving the type of design, authors, dimensions and materials, loading,
performance, and location of failure. In general, it was found that one of three high-stress
locations limited the pressure or load carrying capabilities. These were: the inside surface at the
crotch adjacent to the cutout, the outside surface of the run pipe on the side of the connection,
and in the fillet around the end of full-encirclements if they are pressure-retaining.
Reference 45,
Atterbury, T. J., Vagins, M., and MeClure, G. M., “Branch Connections-Development of
Rules for Design”, Projects NG-11 and NG-22 Summary Report, American Gas
Association (January 30, 1961).
‘The program described in this report studied the effects of various combinations of
header, branch, and pad dimensions on the membrane and bending stresses in the fabricated
assembly due to intemal pressure. The area-replacement principle of branch design handed down
from boiler design rules for nozzles in the early part of the century were originally developed for
small values of branch to header diameters (4/D), where the stresses near the opening are
primarily membrane in nature. It had already been established that where d/D>1/2 in thin-walled
Pipes, local bending stresses can become quite large. Also, it was not possible to determine the
effects of the various dimensional parameters on these stresses. Therefore, several unreinforced
branches and various styles of reinforced branches were evaluated using strain gages, with the
objective of developing design rules that more effectively reduced stress levels in the fabrication
than the area replacement approach to branch design.
Reinforcement techniques included circular pads, stacked pads where the second pad is,
smaller in diameter than the base pad, and a pad-on-sleeve arrangement. These latter two are not
‘commonly seen in service. The study led to a proposal for revisions to the branch connection
design rules in ASME B31.8 which were somewhat more complex than the standard area
replacement technique. These more exact rules were never adopted by the Code.755 Fittings and Components
Reference 46
Vagins, M., Atterbury, T. J., and McClure, G. M., “The Analysis of Spacing
Requirements for Multiple Openings Composed of Welded Branch Connections”,
American Gas Association (April 20, 1962).
Strain gage data from fabricated branches tested for A.G.A. in 1959 and 1960 were used
to develop spacing criteria for headers with multiple openings, such as pulsation bottles or other
manifold arrangements. The data showed that the intensified stresses decay with distance from
the opening. It was assumed that the header between branches in close proximity will be subject
to stress levels that are a simple summation of the overlapping stress fields of the individual
‘branches. The rules for design of multiple openings in effect in the contemporary edition of
ASME B31.8 were evaluated in light of the results of the study and determined to be generally
safe or conservative, These rules remain in effect in the present day edition of the Code.
Reference 47
Atterbury, T. J., Rodabaugh, E. C., Vagins, M., and McClure, G. M., “Development of
Rules for Design of Pipeline Branch Connections”, Project NG-11 Summary Report,
American Gas Association (January 10, 1963).
This report summarizes findings from A.G.A.-sponsored investigations of branch
connection design going back to 1954. Many of the findings were reported in several of the
foregoing documents. These research programs have influenced the Code to the extent that
ASME B318 and B31.4 both recognize the potential for high stresses in large branches in thin-
walled pipe and address that concern by imposing special limitations or requirements on
fabrication details for large openings in pipe operating at high stress levels that do not appear in
the design rules for plant and process piping.
Reference 48
Rodabaugh, E. C., Duffy, A. R., and Atterbury, T. J., “Fabricated Branch Connections in
‘High Yield Strength Run Pipe”, NG-18 Report No. 23, American Gas Association
(March 13, 1970).
The rules in ASME B31.8, which allow pad or saddle reinforcement where d/D<1/2 were
evaluated in terms of their applicability to branches fabricated in X-60 or stronger pipe. Two
hydrostatic test failures in X60 pipelines at pad-reinforced branches, as well as results from7-56 Fiings and Components
‘experiments on branches, were considered. Rodabaugh concluded that due to the stress
concentration at the side of the saddle or pad and the possible notch sensitivity of high-strength
‘materials available at that time, the pad or saddle reinforced branch may not have the safety
margin on pressure that was intended (roughly 80 percent of the computed burst strength of the
plain pipe). A test program designed to explore the issue further was outlined.
Reference 49
Rodabaugh, E. C., “Review of Service Experience and Test Data on Openings in Pressure
Vessels with Non-Integral Reinforcement”, WRC Bulletin 166, 26 pages (October 1971) .
As the title suggests, Rodabaugh provides a comprehensive review of service failures and
test data (from PVRC, A.G.A., and others) pertaining to fabricated openings in pressure vessels
and branches in piping and gas transmission pipelines. He addressed both cyclic pressure and
‘extemal loading performance. Numerous photographs of test and service failure specimens are
presented.
It is shown that both cyclic pressure and cyclic moment loading behaviors conform to
‘Mark's fatigue S-N law derived for fittings and components. Rodabaugh concludes that
satisfactory performance under cyclic conditions cannot be related solely to burst strength alone
(.e., area replacement) and the overall satisfactory performance of the majority of fabrications
indicates that they undergo few large pressure cycles. He observes that problems are most offen
associated with the fillet welds, with initial fit-up and weld deposit contour as contributing
factors.
Reference 50
Rodabaugh, E. C., “A Review of Area Replacement Rules for Pipe Connections in
Pressure Vessels and Piping”, WRC Bulletin 335, 57 pages (August 1988).
‘Rodabaugh provides yet another comprehensive examination of the service and test
experience of branches in pipe, this time in the context of evaluating the adequacy of branch
design rules in the ASME B31 piping design codes, ASME vessel design rules, and tee fitting
design requirements. Portions of the work that are more applicable to pipe connections to vessels
‘would be relevant to reciprocating compressor pulsation bottles,751 Fittings and Components
Reference 51
Horsley, D., Nippard, F. E., and Pick, R. J., “Finite Element Investigation of the NPS 36
Western Alberta Mainline Rupture at the James River Interchange”, * Sympasion on
Pipeline Research, Houston, 8 pages (September 30, 1996).
A detailed finite element stress analysis was performed of a fabricated branch connection
that failed in service. The branch design was a 24-inch branch on a 36-inch * 0.406-inch X52
pipe reinforced by a full-encirclement saddle, operating at 782 psig. The branch apparently failed
due to aweld defect. However, there were aggravating factors including a bending load caused
by settlement, and poor fit-up between the saddle and run pipe. The effects of these factors on
local stresses at the welding defect were evaluated. It was concluded that an epoxy grout should
bbe used in the clearance under the saddle to immobilize the junction between the branch and run
pipes. These findings were verified in full-scale tests.
Reference 52
Bodman, E. and Fuhlrott, H., “Investigations of Critical Crack Geometries in Pipes”, 5*
Intl. Conf, on Struct. Mech. in Reactor Tech. (SMIRT 6), 10 pages (1980).
Reference 53,
Schulze, H. D., Brenner, U., and Fublrott, H., “Application of Fracture Mechanics for
Safety Assessments of Cracked Pipes”, 12 pages (1984).
References 52 and 53 describe a variety of loading tests on austenitic and ferritic pipe and
forged butt-welding tees containing artificial crack-like defects in various orientations. The tees
were matched to 2.375-inch OD x 0.0787 inch WT pipe equivalent to Type 321 stainless steel.
‘Surface defects were given four orientations on the tees: longitudinal in the crotch, transverse in
the crotch, aligned with the branch on the side of the tee, and angled at 45 degrees on the side of
the tee. The tees were subjected to various combinations of intemal pressure and applied bending
load, but only the tees with axial defects in the crotch were subjected to pressure only. It was
concluded that the failure pressures of the tees were within the scatter of data for straight pipe.
The authors concluded that they could be adequately evaluated using existing methods,
specifically that of Kiefner and Maxey, ASTM STP 536, 1973.758 Fittings and Components
Reference 54
WRC Bulletin 347, September 1989, 25 pages, containing: Schroeder, J., "Welded Tee
Connections of Pipes Exposed to Slowly Increasing Internal Pressure"; and Schroeder, J.,
"Flawed Pipes and Branch Connections Exposed to Pressure Pulses and Shock Waves".
The first paper compiles burst test data for various types of pipe connections, mostly
obtained from the foregoing references, and attempts to correlate the burst strength relative to
plain pipe in terms of the ratio of branch to run wall thicknesses (YT) and branch to run diameters
(WD). It was shown that the burst strength of connections normalized to plain pipe will tend to
decrease from 0.95 or 1 to around 0.8 as A/D increases above 0.5. Normalized strength increased
with VT but only in the high-d/D range. The second paper studied the effects of rapid
pressurization on both straight pipes and branches. Pipe specimens were 12.75-inch OD x 0.375-
inch WT A106 Grade B. Flawed and unflawed pipe specimens were subjected to gradual
pressure to bursting. The tests were then repeated under shockwave conditions (pressure rise of
2,000 psig in 15 milliseconds), while the strain history was recorded. The tests were then
repeated on unreinforced branch connections.
Reference 55
Rodabaugh, E. C., “Background of ANSI B16.5 Pressure-Temperature Ratings”, 37%
Midyear Meeting, API Division of Refining, New York, 37 pages (May 11, 1972).
A concise history of flange design is given. The rationale behind the pressure and
temperature ratings of present standard flanges is explained. Rodabaugh explains how and why
B16.5 flanges differ from ASME Boiler and Pressure Vessel Code flanges, as well as why
certain Class 150 and 300 flanges are more prone to leakage. Key references important to the
development of flange design are also cited, including several of the ones listed below.
Reference 56
Waters, E. O., and Taylor, J. H., “The Strength of Pipe Flanges”, Trans. ASME, Vol. 49,
No. 12, 19 pages (1927).759 Fittings and Components
Reference 57
Waters, E. O., Wesstrom, D. B., Rossheim, D. B., and Williams, F. S. G., “Formulas for
‘Stresses in Bolted Flanged Connections”, Trans. ASME, 11 pages (1937).
Reference 58
Labrow, S., “Design of Flanged Joints”, Proc. Inst. Mech. Engrs., Vol. 156, 8 pages
(1947).
Reference 59
Wesstrom, D. B., and Bergh, S. E., “Effect of Internal Pressure on Stresses and Strains in
Bolted-Flanged Connections”, Trans. ASME, 15 pages (July 1951).
Reference 60
Lake, G. F., and Boyd, G., “Design of Bolted, Flanged Joints of Pressure Vessels”, Proc.
Inst. Mech, Engrs., Vol. 171, 40 pages (1957).
The above are a few of several key references in the development of design and analysis
of bolted flanges for pressure containment in pressure vessels and piping. Papers by Waters and
Wesstrom report results of tests performed by them or others, while the other papers are mainly
analytical.
Reference 61
Markl, A. R. C., “Piping-Flexibility Analysis”, Paper No. 53-A-51, Trans. ASME, 23
pages (Feb. 1955).
Markl discusses the historical development of the flexibility factors and stress
intensification factors for individual styles of fittings. He summarizes the approximate
assumptions leading to the various degrees of accuracy of various pipe framework analysis
techniques, including mathematical matrix-based methods solved using “memory-endowed
electronic ... computing devices”. He recognizes that inelastic action and relaxation in the hot
condition will occur, and rationalizes the effects of self-springing and cold-springing on
shakedown of the piping system to elastic action. Markl then derives the allowable stress levels
for thermal expansion (or any displacement-induced) stress range found in ASME B31.1 and
B313 (and soon to be retumed to B31.8), specifically S,=f{(1.25S,+0.50S,), where S, and S, are7-60 Fittings and Components
the allowable material stresses in the cold and hot conditions, and fis a fatigue strength reduction
factor inversely related to the number of anticipated cycles, f6/N"?, He then quantifies the range
of factors of safety from both theoretical and experimental standpoints. Finally, he develops a
simple screening criterion under which no analysis is deemed necessary.
This is probably the single most important paper written on the subject of design and
analysis of unrestrained piping systems. Its relevance is undiminished by time in describing the
simplified design process embodied in piping codes in effect today, and should be required
reading for anyone who claims to be a piping stress analyst or piping designer. At one time, the
concepts clearly discussed by Markl were well understood by practicing piping engineers.
However, the implicit assumptions have been forgotten and the engineering judgement required
of the designer has been lost, even as automated push-button piping analysis and design
processes (typically developed by programmers and software developers who do not understand
the subject of piping design) have improved the designer’s ability to estimate stresses. That this
is the case is probably responsible for widespread misinterpretation of the intent of the ASME
Code for Pressure Piping.
Reference 62
The M. W. Kellogg Company, Design of Piping Systems, John Wiley and Sons, Inc., 384
pages (1941 and 1956).
Reference 63
Grinnell Corporation, Piping Design and Engineering, 7* Ed., 320 pages (1995).
References 62 and 63 provide information useful to conducting a piping flexibility
analysis. The Kellogg book is more of a technical primer, while the Grinnell book is primarily
tables and charts.
In the Kellogg book, Chapter I discusses the strength and failure mechanisms that the
designer must consider including plastic collapse, creep, fracture, and fatigue, Chapter 2
describes the design assumptions and simplifications embodied in the ASME piping code rules
and allowable stresses, most of which remain in place today. Chapter 3 reviews the effects of
fitting design details on fitting flexibility and stress intensification. Chapters 4, 5, and 6 discuss
flexibility analysis by a simplified method, by the general analytical method, and by mode!761 Fittings and Components
testing, respectively. Chapter 7 discusses expansion joints. Chapter 8 discusses purposes and
designs of various types of supports, restraints, and braces. Chapter 9 addresses vibration contol.
Several appendices provide data or mathematical derivations. In light of the ready availability of
piping analysis computer programs, Chapters 4 through 6 are somewhat dated, however the other
chapters are still highly relevant,
‘The Grinnell book compiles data on pipe materials, pipe fittings, valves, threads, and the
like. The most valuable features are the piping expansion tables which giving maximum bending
and torsional stresses, and reaction forces and moments due to thermal expansion for numerous
planar and three-dimensional piping layouts commonly found in plant facilities. These can be
used to obtain results that are equally valid to those obtained from piping analysis computer
programs (which is not to say that they are numerically equivalent, nor more nor less
conservative), within the limits of uncertainty inherent in any flexibility analysis and in the Code
criteria,
Reference 64
Markl, A. R. C., “Fatigue Tests of Piping Components, Paper 51-PET-21, Trans. ASME,
17 pages (1952).
This paper summarizes bending fatigue tests performed on a wide variety of pipe
specimens and fittings including: polished round-bar specimens machined from Grade B pipe,
plain straight pipe, straight pipe containing a girth butt weld, short- and long-radius elbows,
fabricated miter bends, forged welding tees, unreinforced branches, pad-reinforced branches,
welding neck flanges, slip-on flanges, and socket-welded flanges. Directional components were
tested both in-plane and out-of-plane,
Mark found that within a scatter band, the S-N curves of each type of component lay
essentially parallel to the round-bar S-N curve except that generally no endurance limit was
observed. When plotted on log-log paper, entire body of data was essentially parallel and could
be reasonably approximated by the formula iS=245,000 N-, He determined that the least stress
intensification in practice corresponded to straight pipe joined by a girth butt weld. In order to
avoid having the location of a butt weld control the piping design, Markl used it as the
normalizing basis, assigning it a value of 1 and adjusting the constant accordingly. Thus, for all7-62 Fittings and Components
other components, “i” is the relative fatigue-effective stress intensification of a component above
that of a girth butt weld, which itself has been found to represent an implicit factor of 2.
Interestingly, Marki’s formula can be reduced to a single-load failure law by substituting N=1/4
and i=0.5 to get S=6.5%10" psi, which elastically calculated is equivalent to a strain of 2 percent,
a fairly conservative estimate.
Mark found that the SIFs could be related to the same geometric characteristic as the
flexibility factor, such that the fatigue performance of entire families of components could be
generalized. Moreover, the greater the flexibility, the greater was the stress-intensification factor.
Reference 65
Rodabaugh, E. C., "Standardized Method for Developing Stress Intensification Factors
for Piping Components", WRC Bulletin 392, 11 pages (June 1994).
‘Neither the ASME B31 Code for Pressure Piping, nor Section III of the Boiler and
Pressure Vessel Code, both of which specify stress intensification factors, give guidance on how
such factor should be established. WRC 392 proposes a specific protocol intended to assure that
experimental procedures and interpretations of i-factors derived for nonstandard components for
Which the standard values may not be accurate are consistent with existing SIFs and allowable
stresses. Background on the development and use of i-factors is given.
‘The four basic steps to developing a valid SIF are as follows. A preliminary load-
deflection curve is generated. Cyclic bending tests with controlled displacement are performed
on multiple specimens. The intent is to produce failure in a reasonable number of cycles, which
means a certain amount of plastic deformation will take place. Hence, moments should be
extrapolated linearly from the elastic portion of the load-deflection curve. Failure is defined at
the number of cycles at which a through-wall crack is detected. The SIF is calculated from
5=245,000(SN™).
Reference 66
Rodabaugh, E. C., and Atterbury, T. J, “Flexibility and Stress Intensification Factors of
Piping Components with Moment Loading”, Project PR-3-39, American Gas Association
(October 21, 1966).7-63 Fittings and Components
Rodabaugh presents an exhaustive review and evaluation of recent (at that time)
experimental and theoretical information concerning the flexibility and stress intensification
characteristics of fittings subject to bending loads. He also provides an introduction to the
concept and use of flexibility (&) and stress intensification (i) factors for readers unfamiliar with
the subject. The & and i factors were developed to evaluate piping layout in steam power and
refinery piping that undergoes large temperature changes, While large temperature variations
typically do not occur in most gas piping facilities with the exception of the regeneration cycle
on desiccant dehydrators, they are equally applicable to other kinds of displacement-induced
loadings. These include fit-up misalignment; structural movements such as support or foundation
settlement or offshore platform movements; and vibration in compressor stations. The k factor is
the moment rotation that occurs in a component normalized to the rotation in straight matching
pipe. The i factor is derived from fatigue performance tests of representative fittings (“Mark
tests”), normalized to the performance of straight pipe with a girth butt weld. The point is made
that elastic stress intensification factors derived by theoretical analysis are roughly twice the /
factors derived by fatigue testing. Subtle but significant influences on the k and i factors are
discussed, including end effects of tangent pipe or flanges, stiffening effects of internal pressure,
and dimensional variation in Markl-test specimens,
‘Recommendations are given for revisions to the factors in the contemporary edition of
ASME B31.8. The most important of these are to incorporate factors to account for the pressure-
stiffening effect in thin-walled elbows, and to restrict the i given for B16.9 tees to those with a
crotch radius larger than one-eight of the branch or outlet diameter. These revisions are found in
the present-day code. Some other recommendations regarding refinement of k for tees, as well as
refinements directed toward less-common details do not appear to have been adopted in ASME
B318 or the other B31 piping codes.
Reference 67
Rodabaugh, E. C., and George, H. H., “Effect of Intemal Pressure on Flexibility and
‘Stress-Intensification Factors of Curved Pipe or Welding Elbows”, Paper No. 56-SA-50,
‘Trans. ASME, 10 pages (1957).
Starting with Beskin’s more accurate solution for bends, the authors derived
modifications to the available flexibility factors to account for the stiffening effect of intemal7-64 Fittings and Components
pressure. This effect is not important in heavy-walled elbows such as those used in steam piping,
but is significant in large thin-walled bends. The results are compared to tests on 30-inch OD x
0.500-inch WT elbows having a nominal bend radius of 45 inches, attached to 30-inch OD X52
pipe. Strain gages were used to measure stresses in and around the bend.
Reference 68
Kitching, R. and Bond, M. P., “Flexibility and Stress Factors for Mitred Bends Under In-
Plane Bending”, int. J. Mech. Sci, Vol. 12, 9 pages (1970).
Extensive tests and computer analysis based on shell theory were conducted in order to
refine the factors. (Mitered bends are used in British gas-cooled reactor facilities.) Good
agreement was found between the computer models and flexibility factors determined from the
tests. Theoretical stress concentration factors, divided by two to adjust for Markl’s normalizing
basis, were in good agreement with i-factors derived from bending-moment fatigue tests.
However, the overall discrepancy between the existing Code equations and test data was not
large, so the improvements appear to be incremental.
Reference 69
Rodabaugh, E. C., “Accuracy of Stress Intensification Factors for Branch Connections”,
WRC Bulletin 329, 44 pages (December 1987).
‘The general background of SIFs and flexibility of branch connections leads into detailed
and rather complicated discussion of certain unresolved issues in then (and still) present Code
ules. One important problem is the “M,, Inconsistency” wherein certain tests indicated that the
Code equations could be unconservative for out-of-plane moments applied to the branch. One
clement of the inconsistency lies in the resultant bending stress, and whether the section modulus
‘was intended to be that of the run or the branch. Another is that there is apparently a peak in
actual SIFs in the intermediate size range 0.5100. Also, the Code should state that
the effect of closely spaced branches (within two run diameters clearance) requires special
consideration. Of significance is the recommendation that i=, instead of i=(3/4)i,+1/4.
Reference 70
Woods, G. E., and Rodabaugh, E. C., “WFI/PVRC Moment Fatigue Tests on 4x3 ANSI
B16.9 Tees”, WRC Bulletin 346, 8 pages (August 1989).
Out-of- plane bending moment fatigue tests were performed on ANSI B16.9 43 reducing
tees having four different geometries in the transverse section. The results found that there is no
peak at d/D=0.75 in the effective i-factor for such tees, in contrast to what has been observed
with fabricated branches. It was concluded that the minimum crotch radius and thickness
requirements in the Code were necessary but possibly not sufficient to assure uniform
conservative behavior.
Reference 71
Atterbury, T. J, McClure, G.M., and Rodabaugh, B. C., “Fatigue of Welded Branch
‘Connections Subject to Cyclic Bending Loads”, ASME Paper No. 64-PET-38, 8 pages
(1964).
Fatigue failures at the nozzle-to-shell joint in compressor pulsation bottles, due to
vibration, led to this PRC-sponsored study of the behavior of branch connections under cyclic
loadings. Four types of branches were studied: pad reinforced, saddle reinforced, unreinforced
drawn (extruded) outlets with the shell thickness the same as in the pad or saddle reinforced
‘examples, and reinforced drawn outlets with the shell thickness the same as the sum of the pad
and shell thicknesses.
‘The headers were 16-inch OD * 0.500-inch WT X42, except that the reinforced drawn
outlets used a 1-inch wall thickness, The branches were 6-inch NPS Schedule 40 seamless Grade
B pipe. Cyclic loading was applied by an eccentric mass driven by a variable speed motor.7-66 Fittings and Components
Stresses were monitored by strain gages. Nominal stresses were between 2 and 5 ksi. The
dynamic tests were cartied out to 10” cycles. All experiments were carried out with 800 psig
internal pressure producing a circumferential stress in the header of 30 percent of SMYS.
Failures generally occurred adjacent to weldments. In the pad reinforced specimens, the
failures occurred at the toe of the branch attachment weld. In the saddle reinforced specimens,
failures occurred in any of several locations: at the toe of the saddle-to-header weld, in the branch
at the toe of the saddle-to-branch weld, in the saddle at the base of the saddle-to-branch weld, or
in the branch at the toe of the branch-to-header weld. In the unreinforced drawn outlet, the
failures occurred on the header side at the toe of the branch attachment weld. In the reinforced
drawn outlet, the failures occurred on the branch side at the toe of the branch attachment weld.
‘The authors concluded that the reinforced drawn outlet had the best performance,
followed by the pad reinforced branch, the saddle reinforced branch, and the unreinforced drawn
outlet last. However, this reviewer concluded that the reinforced drawn outlets and saddle
reinforced branches had comparable performance. Moreover, the narrow stress range of 3 to 6 ksi
from 10” down to 5x10° cycles indicates high sensitivity to stresses in each of the designs,
Specimens of the pad reinforced branches which received post-weld stress relief heat
treatment experienced roughly a half-order of magnitude improvement in fatigue life. Specimens
of the pad reinforced branches which had the fillet welds ground to a smooth contour at the toes
experienced a full order of magnitude improvement in fatigue life, essentially to an equivalent
level of performance achieved by the reinforced drawn outlet.
Reference 72
Fondriest, F. F., Atterbury, T. J., and McClure, G. M., “Study of Effects of Cyclic
Bending Loads on Performance of Branch Connections”, NG-11 Report, American Gas
Association (May 2, 1960).
Pad-reinforced fabricated riser connections at the compressor discharge suction bottles
‘were studied using strain gages to determine local stress levels in such assemblies during
‘operation. Bottles were tested in one station that had experienced no problems and in another
station that had experienced leaks due to fatigue at the riser flanges and branch connections. In
one station, both long and short riser designs were tested on the same compressor unit. Strains
due to fit-up, static pressure, and operation were recorded. It was determined that stresses due to1-67 Fittings and Components
fit-up and static pressure could be up to 10 to 30 ksi, while dynamic stresses were approximately
one-tenth of these levels. Also, fit-up stresses were greatly reduced by first bolting in the loose
flanges to the compressor, and then welding them to the risers, as opposed to welding the flanges
to the risers and then bolting the completed units to the compressor.
‘The in-situ data were then used to design a test fixture consisting of 16-inch OD x 0.500-
inch WT X42 header with two 6-inch NPS x 0.250-inch WT Grade B risers. The fixture could
induce longitudinal bending, longitudinal rotation, or transverse rotation to the header by
application of a static displacement-induced load at the riser flanges. A vibrational load was
introduced at the header using a variable speed rotating eccentric mass, and the assembly was
pressurized with water to 800 psig. Fatigue S-N curves out to 10” cycles at mean stress levels of
10, 15, and 20 ksi were developed. Failure typically occurred in the crotch of the branch at the
toe of the attachment weld. When a riser failed, the header was cut in half and mated to a new
half and the testing continued.
In the in-situ tests, when 2 fatigue crack developed leading to failure, the failing riser
began shedding dynamic loads to the other risers. In the bench tests, no significant difference in
fatigue performance was observed between the three static mean stress. This was attributed to
early yielding at the weld toe,
Reference 73
Markl, A. R. C., “Fatigue Tests on Flanged Assemblies”, Trans. ASME, 10 pages (1950).
Markl’s well-established test protocol of cyclical reversed bending was applied to several
styles of flange installed on 4-inch Grade B pipe. Flange styles included: slip-on, with undersized
and oversized fillet welds at the hub and face; slip-on, with only a hub or face weld; socket-
‘welding, with both inner and outer fillet welds; socket-welding, with only outer fillet welds;
threaded; lap-joint; ring (hubless); and welding neck.
Leakage through the gasket generally did not occur except in the lap-joint flange,
indicating that the flanges were stiff enough for their rated pressure with even large bending
loads applied. The vast majority of flanges developed circumferential cracks at the toes of fillet,7-68 Fittings and Components
‘welds (if applicable). The welding neck flanges failed at the welded attachment, The threaded
flange failed by persistent leakage through the threads,
Overall, the welding neck flange performed as well as straight pipe with a butt weld,
eaming a SIF equal to 1.00. The double-welded socket welding flange eamed i=1.15, the double-
welded slip-on or ring flanges eamed i=1.25, the single welded (at the hub) slip-on or socket
‘welding flanges eared i=1.30, the lap-joint flanges earned i=1.60, and the threaded flanges.
eamed i=2.30.
Reference 74
Bijlaard, P. P., "Stresses from Local Loadings in Cylindrical Pressure Vessels", Trans.
ASME, 12 pages (August 1955).
Reference 75
Bijlaard, P. P., "Stresses from Radial Loads and External Moments in Cylindrical
Pressure Vessels", Welding Research Supplement, 12 pages (December 1955).
In Reference 74, Bijlaard describes a theoretical solution for the local stresses and
deflections in a cylindrical shell due to applied forces and moments. His solution assumes that
the forces or moments are averaged over a rigid rectangular attachment, and that the shell
remains intact (.¢., no cut-out as for a nozzle). Elements of this solution for individual load
components, and for spherical shells as well, were developed in several prior WRC Bulletins.
Reference 75 presents results from Reference 74 reduced to curves that could be readily
used by a design engineer to determine angular rotation, and for local membrane and bending
stress produced by a radial force, and an in-plane or out-of-plane moment applied to a rigid
attachment. It is applicable to the evaluation of loads applied to a structural attachment such as a
lug or stanchion welded onto a pipe or pipeline.
Reference 76
Wichman, K. R., Hopper, A. G., and Mershon, J. L., “Local Stresses in Spherical and
Cylindrical Shells Due to Extemal Loadings”, WRC Bulletin 107 (August 1965, Revised
March 1979),7-69 Fittings and Components
Reference 77
Mershon, J. L., Mokhtarian, K., Ranjan, G. V., and Rodabaugh, E. C., “Local Stresses in
Cylindrical Shells Due to External Loadings on Nozzles — Supplement to WRC Bulletin
No. 107 (Revision 1)", Revised WRC Bulletin 297, 88 pages (September 1987).
WRC 107 presents a "cook book" approach to calculating the local shell stresses due to
applied forces or moments in all three axes. The design curves are based on Bijlaard’s analysis.
WRC 297 is a refinement of WRC 107 specifically applicable to cylindrical nozzles on
cylindrical shells based on a thin shell analysis by Steele, Stresses at the nozzle juncture not
specifically addressed by WRC 107 are covered, along with larger D/T and better detail curves
for small d/D. An appendix presents correlating results from finite element analyses. Users
should be sure to use the September 1987 Revision 1, which corrects errors in the initial August
1984 edition. References 76 and 77 are both applicable to the evaluation of external loads applied
through the branch connection on a pipe or pipeline.
Reference 78
Powell, G. H., Clough, R. W., and Gantayat, A. N,, "Stress Analysis of B16.9 Tees by the
Finite Element Method: A Progress Report”, 71-PVP-40, ASME Press. Ves, & Piping
Conf., San Francisco, 9 pages (May 10-12, 1971).
This paper describes progress toward development of a generic tee analysis program that
would automatically generate a finite element mesh of a tee from minimal user input. The basic
‘modeling considerations of element formulation, geometry definition, mesh generation, and
output in interpretable form are discussed. The problems associated with each of these issues
‘were enormous given the primitive state of FEA software at that time. The automatic mesh
‘generation and color stress contour plots taken for granted today simply did not exist. Good
comparison was obtained between calculated stresses and test results on a 126 Schedule 40
reducing tee under internal pressure and in-plane bending on the branch.
Reference 79
‘Natarajan, R., and Mirza, S., "Effect of Thickness Variation on Stress Analysis of Piping
Elbows Under Internal Pressure”, Comp. & Struct., Vol. 18, No. 5, 12 pages (1984).7-10 Fittings and Components
Reference 80
Natarajan, R., and Mirza, §., "Effect of Internal Pressure on Flexibility Factors in Pipe
Elbows with End Constraints", J, Press. Ves. Tech., Trans. ASME No. 107, 4 pages
1985).
The first paper examined the effects of wall thickening on the intrados with equivalent
wall thinning on the extrados on stresses in an elbow subject to intemal pressure or pressure plus
bending. The second paper studied the effects of end constraints on the flexibility factors in pipe
elbows subject to intemal pressure. The numerically-derived stresses were in good agreement
with experimental strain gage data from inside and outside surfaces. However, the presentation of
results is flawed by not separating membrane from bending components, and by expressing
results in terms of a stress factor that is defined only in an extemal reference.
‘Natarajan concluded in the first study that wall thickness variations had no effect on the
location or magnitude of maximum stresses, which occur as circumferential bending stresses on
the crown. The second paper demonstrated that in the presence of end constraints, pressure
further reduces the flexibility of elbows, by up to 50 percent in thin-walled elbows. This is
consistent with the findings by Rodabaugh and George®”. A similar trend was observed for
elbows with smaller included angles; however, the reduction for thin-walled materials is only
about 25 percent. None of the findings could be regarded as entirely novel with respect to the
prior theoretical or experimental work of others.
Reference 81
Fujimoto, T., and Soh, T., "Flexibility Factors and Stress Indices for Piping Components
with D/>100 Subjected to In-Plane and Out-of-Plane Moment", Press. Ves. & Piping
Congf., New Orleans (June 23-26, 1988).
The concem addressed by this work was the recognition that the stress intensification
factors and flexibility factors in ASME B31.3 were derived mainly from heavy-walled piping,
but that thin-walled components are often used in chemical plants with low temperature
operations. A series of experiments and finite element analyses were conducted in order to
determine whether the factors were unconservative for the thin-walled components, or if the rules
specified for the nuclear industry in ASME Section II, NB-3600 would be more appropriate.7 Futings and Components
One shortcoming of the paper is that the word "code" is used interchangeably for ASME
B31.3, which uses stress intensification factors, and NB-3600, which uses stress indices. Stress
indices differ from SIFs in that they represent stress factors to be used with the theoretical design
approach used uniquely with nuclear piping. The author’s lack of clarity results in confusing
conclusions. The authors also failed to show a direct and understandable comparison of the
code-calculated values as compared to the values derived from finite element analyses. Despite
these shortcomings a reasonable level of accuracy was achieved in terms of comparing
experimental and analytical results. It was observed that where D/t>100, the flexibility factors for
miter bends were higher than those calculated by ASME B31.3, and in comparison with NB-
3600, the calculated SIFs and flexibility factors for unreinforced fabricated tees were higher and
lower, respectively. Miter bends and size-on-size unreinforced fabricated branch connections are
not commonly used in pipeline facilities.
Reference 82
Fowler, J. R., Samman, M. M., and Al-Sannaa, M. S., "High-Pressure Pipe Design", PR-
201-9202, A.G.A. PRC (1993).
Reference 83
Fowler, J. R., and Alexander, C. R., "Design Guidelines for High-Strength Pipe Fittings",
PR-201-9320, A.G.A. PRC (1994).
‘The American Gas Association funded work to evaluate whether the SIFs in the ASME
piping codes are unconservative for large diameter fittings with thin walls, implying that Markl’s
work may have been extrapolated beyond its applicability. In addition to concems relating to
SIPs, discussion in the report also addressed the methods by which fittings are qualified relative
to one another according to the governing standard, MSS SP-75, Although some burst tests were
performed, the bulk of the research involved finite element analysis (FEA) using ABAQUS
clastic thin shell elements. Reference 82 analyzed size-on-size tees, elbows and concentric
reducers having diameters from 16 to 36 inches, while Reference 83 was an extension of the
previous work looking at fittings with diameters ranging from 4 inches to 12 inches. The latter
also included reducing tees in the analysis.7-12 Fisings and Components
‘The authors found that flexibility factors derived from the FEAs were generally in good
agreement with those used in the Code. They also determined that the peak stress factors
estimated by the FEAs were consistently greater than the Code SIFs, sometimes by a factor of 2.
From this they concluded that the Code SIFs are unconservative, by as much as a factor of 2.
‘The analytical work appears to be technically correct, but it is important for readers to
understand that the authors’ interpretation of the peak stress concentration factor (SCF) derived
by a theoretical or numerical analysis (which the authors refer to as “stress intensification
factors”) is not equivalent to the SIF as used in the Code, The Code SIFs are fatigue-effective
factors based on actual performance, normalized to the fatigue-effective performance of straight
Pipe containing a girth butt weld. The girth butt weld is assigned an SIF of 1 but represents an
effective built-in factor widely regarded as at least 1.5 and nominally equal to 2. This
normalizing basis is embedded in the Code SIFs and implicitly accounted for in the Code
allowable stresses. Markl himself remarked on this in his papers", and it has been pointed out
subsequently by Rodabaugh™ and others™ in documents that the authors even listed in their
bibliography.
‘The ASME B31 Mechanical Design Technical Committee reviewed the subject work and
concluded that the authors’ FEA results represented peak stresses (SCFs), not SIFs. The also
concluded that the factor of 2 difference was consistent with the authors’ interpretation of the
meaning of SIFs. If the authors’ numerical SCFs are divided by 2, they fall within the range of
scatter typically observed for experimentally derived SIFs. WRC Bulletin 392 presents a
standardized method for developing SIFs on a consistent basis with the Code. Use of the authors’
findings as they interpreted them, while unnecessary, would lead only to a more conservative
design so the only harm would be economical.
Despite the difference in interpretation discussed above, the paper correctly points out
that the design basis for fittings is inadequate to establish acceptable fatigue performance (as has
been pointed out by others, see for example Reference 49), and perhaps even pressure
performance over the full size range manufactured. In particular, the authors recommend that
fitting purchasers audit the design acceptability of the fittings and burst tests that are performed
on the respective fittings. In terms of pressure containment, fitting purchasers are encouraged to
require that the crotch thickness in elbows be at least 1.25 times that of the adjoining pipesTB Fittings and Components
(tress in the crotch region are 125 percent of the nominal stress in the elbow), and that the crotch
thickness in tees be at least 1.5t and that the crotch radius be no less than d/8. (This suggestion
for tees is already a Code requirement to use the SIF for butt welding tees instead of the higher
SIF for outlets.) While these are valuable suggestions, they create a need to measure dimensions
that are not presently standardized, In order to use these suggestions, one would have to
‘incorporate them into a purchase specification, along with acceptable tolerances and
contingencies for noncompliance,