Documente Academic
Documente Profesional
Documente Cultură
M. A. Shalaby
With Internal Pressure
ECG Engineering Consultants Group,
Piping Department,
Cairo, Egypt
Under In-Plane Closing
Bending Moments
M. Y. A. Younan
The American University in Cairo, The purpose of this study is to determine limit loads for pipe elbows subjected to in-
Engineering Department, plane bending moments that tend to close the elbow (i.e., decrease its radius of
P.O. Box 2511, curvature), and the influence of internal pressure on the value of the limit load.
Cairo, Egypt 11511 Load-deflection curves were obtained, and from these curves plastic collapse or
E-mail: myounan@acs.auc.eun.eg instability loads at various values of internal pressure were determined. This was
done for different pipe bend factors (h = Rt/r 2 ) using the nonlinear finite element
analysis code (ABAQUS) with its special elbow element. The limit load was found
to increase and then decrease with increasing pressure for all the elbow geometries
studied.
Introduction exist in the pipe elbows change the values of limit loads and
affect their collapse behavior. Therefore, it is very important to
Smooth pipe bends or pipe elbows are commonly regarded determine conservative values of limit loads for pipe elbows
as critical components in a piping system. They are incorporated with internal pressure under various loading conditions.
into piping systems to allow modification of the isometric rout-
The pipe bend problem has been analyzed by several ap-
ing, but more importantly pipe elbows are usually incorporated
proaches either elastic or inelastic using small and large defor-
to reduce anchor reactions. Due to their increased flexibility in
comparison to straight pipes, they allow a reduction of the reac- mation solutions with both theoretical and numerical tech-
tion forces and moments within the system as a whole by virtue niques, in addition to experimental investigations. In the follow-
of their elastic deformations. Hence, they are forced to accom- ing, the inelastic analysis of pipe elbows will be discussed
modate disproportionate displacements arising from differential through the available work in the literature.
movements. Under complex stressing of this kind, severe load- Marcal (1967) presented the first results for elastic-plastic
ing may be set up within the elbow and this may lead to strain behavior of pipe bends with in-plane bending. His collapse
concentrations and major cross-sectional deformation. moment to moment of first yield ratios ranged from about 2 for
bends of short radius to a value of about 1.55 for bends of
Due to their specific deformational behavior when exposed
beyond the elastic limit, pipe elbows are capable of plasticizing larger radius. Spence and Findlay (1973) found approximate
over large areas when a system is overstressed. Thus, they bounds on limit moments for in-plane bending by utilizing pre-
absorb considerably large thermal expansions and seismic viously existing analyses in conjunction with the limit theorems
movements in addition to cushioning transiently loaded systems of perfect plasticity. They pointed out that a bend has a signifi-
by energy dissipation as a result of plastic material flow. How- cantly lower collapse load than a straight pipe, and this load
ever, care must be taken to ensure that the collapse load is increases with increasing pipe bend factor (h). Calladine
avoided. Therefore, it is very important to determine the safety (1974) tried to find the value of pure bending moment at which
margin between incipient local yield and the load at which the full plastic bending of the bend occurs according to classical
pipe bend collapses. limit analysis. He sought a lower bound to the limit value of
If an adequate limitation of plastic deformation cannot be the moment by using a classical elastic shell analysis in conjunc-
assured, then the pipe elbow or the piping system may fail tion with the lower-bound theorem of plasticity. Goodall
sooner or later depending on the type of load imposed, because (1978a) presented the first large displacement analysis to obtain
of the decreasing resistance of the components to further plastic limit loads of thin pipe elbows under in-plane bending moments.
deformation. Several cases of damage which have occurred in His analysis was based on the lower-bound theorem of plastic-
the region of pipe elbows in various piping systems have ity, and the process of elbow buckling under in-plane closing
prompted the initiation of various investigations with regard to bending moment was examined by solving the large deforma-
deformation and failure behavior of pipe elbows. Plastic limit tion problem for a simplified yield surface. The maximum load-
analysis concepts are used for establishing allowable loads. carrying capacity obtained from his analysis was about 10 per-
They provide estimates of plastic collapse loads, i.e., loads cent lower than the limit load obtained from small displacement
above which large increases in deformation occur with small analysis. Goodall (1978b) also obtained the first lower-bound
increases in load. The low value of the limit load for pipe solution for the limit load of thin elbows under the combined
elbows relative to that of straight pipes should be sufficient effects of in-plane bending and internal pressure using small
warning not to neglect elbows when considering the design of displacement analysis. Kitching et al. (1979) presented a lower-
piping systems. The high values of internal pressure that may bound analysis, but without making restrictions on the geome-
try, so any value of radius ratio (R/r) and (h) could be used.
They showed that the limit moment is dependent upon (R/r)
Contributed by the Pressure Vessels and Piping Division and presented at the as well as on (h). Rodabaugh (1979) gave a lower-bound limit
Pressure Vessels and Piping Conference, Orlando, Florida, July 27-31, 1997, of pressure and showed that for an elbow with R/r = 3 the limit
THE AMERICAN SOCIETY OF MECHANICAL ENGINEERS . Manuscript received by the pressure is 0.8 times the limit pressure of a straight pipe. He
PVP Division, March 18, 1997; revised manuscript received August 1, 1997.
Associate Technical Editor: M. B. Ruggles.
noted that the pressure did not reduce the limit moment; indeed,
Journal of Pressure Vessel Technology Copyright © 1998 by ASME FEBRUARY 1998, Vol. 1 2 0 / 3 5
Nomenclature
D = diameter of pipe or bend P = internal pressure ; = wall thickness
E = Young's modulus R - radius of curvature a = subtended angle of bend
M = bending moment h = pipe bend factor v = Poisson's ratio
ML = limit moment r — mean radius (jy = yield stress
Model and Elements. The model used in this study was the prescribed loading. In this model, six Fourier modes with
based on the assumption of uniform bending, which means that 20 integration points around the pipe cross section were used.
the applied load is a pure bending moment and that all cross In addition, seven integration points were used through the pipe
sections deform in the same way (i.e., the effect of end con- wall. These are sufficient for the range of elbow geometries
straints is neglected). Therefore, the model was chosen as an used.
unconnected elbow subjected to combined loading of in-plane
pure bending moment and internal pressure. Loading. The load on the pipe bend has two components:
The one-dimensional cross-sectional deformation pattern ex- a "dead" load consisting of internal pressure (with closed end
pected allows very simple modeling to be adopted. Element condition), and a "live" load of pure in-plane bending moment
type ELBOW31B is a special pipe bend element with uniformly applied to the free end of the model.
deforming cross section. This element is based on an analysis The pressure is applied to the model in an initial step, and
which assumes that axial variations of both the ovalization of then held constant in the second analysis step while the bending
the cross section and the bending moment (M) are constant moment is being increased. The pressure values range from 0.0
over the axial length of the element, with the amount of ovaliza- psi to the maximum pressure that causes gross plastic deforma-
tion being dependent on the magnitude of (M), so it is ideal tion or collapse for the specific elbow geometry used in each
for this case of unconnected elbow. case. The equivalent end force, caused by the closed end condi-
tion, is applied as a follower force because it rotates with the
It is known that the elbow cross section ovalization varies in
motion of the end plane. The closed end condition was chosen
the axial direction due to the stiffening effects afforded by the
for the analysis in order to include the axial component of the
adjacent flanges or straight pipes of the elbow structure, How-
stresses caused by the internal pressure.
ever, for elbows with radius-to-thickness ratios (r/t) in the
range of 20 to 50, with reasonable angle extent (90 deg), and The elbow was subjected to in-plane pure bending moments,
with lengths of straight pipes that are typical of those normally and the loading was extended far enough in the plastic regime
encountered in practice, it seems reasonable to expect the axial in order to reach plastic collapse or instability. This large dis-
variation of the ovalization to be rather gradual and that buck- placement analysis is rotation-controlled, i.e., increments of ro-
ling will be initiated at the central section of the elbow where tation are prescribed at the free end rather than increments of
the stresses are the highest (Sobel and Newman, 1977). The moment, since it is anticipated that the collapse will be unstable.
present analysis concentrates on the behavior of this section. In these cases, displacement or rotation control can provide a
Hence, the use of special elbow element with constant ovaliza- solution even when the conjugate load (the reaction force or
tion is appropriate for such an analysis in addition to being moment) decreases as the displacement or rotation.increases.
economical. For this model, the rotation is applied to produce in-plane bend-
ing that tends to close the elbow (increase its curvature).
Moreover, this model is considered a conservative one, be-
cause end constraints increase the stiffness of the connected Verification of Model. This model was verified by compar-
elbow, and consequently increase its limit load. Furthermore, ing the same model, but with strain-hardening material with
the approximation involved in neglecting end effects is a logical that of unit length element model and shell elements model
one and agrees with what has been reported in the literature used in example 4.2.1 of ABAQUS Example Problems Manual
that internal pressure reduces the effect of end constraints. The (1995). The shell elements model is convergent with respect to
reason for this is that pressure causes the primary stresses to discretization, as the mesh convergence tests demonstrated that
be dominant; hence, the axial gradients of total (primary plus the meshes were convergent to overall response of the system.
secondary or bending) stress become small and the variation Therefore, the shell elements solution is considered exact solu-
of stress between the pipe bends and the straight pipe runs is tion, and is useful for comparison with ELBOW31B element
reduced (Hibbitt and Leung, 1985). For the case of 90-deg solution for its validation. The twelve ELBOW31B elements
bend, connected flanges only marginally reduce stresses and model predicted the same response of both the shell elements
bend deflections compared to those for an unconnected bend model and the unit length ELBOW31B element model. This
(Whatham and Thompson, 1979). Therefore, in this case where increases the confidence in the model used in this study.
high values of internal pressure are used, the approximation
appears to be quite good. Definition of Limit Loads Adopted in This Work. In this
A single element could suffice under pure bending moment, work, the term "limit load" is used as a collective term for
but the existence of internal pressure may be inconsistent with both "instability load" and "collapse load." The instability
the pure bending assumption (Boyle and Spence, 1977). Also, load is considered the maximum load in the load-deflection
the end force associated with the closed end condition makes plot where the curve approaches a horizontal asymptote. The
it better to use more than one constant bending elbow element collapse load is obtained by applying the "angle method" in
for the model. which the angle that the linear part of the load-deflection curve
The finite element model used to represent the 90-deg elbow makes with the vertical axis is measured, and a line making
consists of 12 elements of type ELBOW31B with 13 nodes double this angle with the vertical axis is drawn to intersect the
along the elbow length. The node at the fixed end is restrained load-deflection curve. The load at the intersection point is called
in all degrees of freedom; the other end node is free, except for the collapse load.
maximum load-carrying capacity of the elbow with (h < 0.5) CoBntise ~ x Ul ' ' i i i • ^ _ | « n •
XHIM i.oooc~o4 / / ! : : : : :
The only analytical solution that treats the problem of com- XHAX (.S19E-03 IS \ J ; ;
YHIM 4-407E.03 0.0*- ' ' 1 1 1 1 1
bined in-plane bending and internal pressure loading is the small VHAX l.olJE.Oi 0.00 0.01 0.02 0.03 0.04 0.05 0.06
End R o t a t i o n I tad I
displacement analysis of Goodall (1978b), which accounts for
the effect of internal pressure on the value of the limit moment Fig. 2 Moment-end rotation for an elbow with h = 0.1615, no internal
of elbows with (h < 0.5) as follows: pressure
1+ 1
h"0.1«15
*
-•
¥ 6.06*6, «,it
I 4'0M » •I
J 3.0E*6- •••!•••
2.0E+6 '
1.06*6 "i"
0.0&0- l44-- M 4 - -4*
t4-1200 1600 2000 2400 2800 32003600 4000 4400 4400 5203
H
0 400 800
Internal Pressure (psi)
¥ i6£**
= l.0E*6
50E*5
0.0E*0
0.4 0.6 0 200 400 600 800 I00O 1200 1400 1600 1800
End Rootlon (rndl Internal Pressure (psi)
Fig. 5 Moment-end rotation for an elbow with h = 0.1615, at selected Fig. 7 Variation of instability moment with internal pressure, closing
internal pressure values (0-2200 psi) case
1IIIIIIIIII
6.0Et6 -•—hlBlBbilly
• M1.3527 ! 1
• h-0.2675
6.0E+6
4t{tittn -•—CcJapse j ]••••
/
* h=0.1998 !
• h=0.1615
•g 4.0E<« JJ- ill 1 i j
•
& II I /
3 0£ 6
1 ' * 1 /
"[*jl ..-•i** ''!'!'>»! 1
I 2.0E*6 f
.A
1 _ <>--^-ityWM.jn^...,f
] i
1 --
1.0E*6
JJJ^
0.0E«O
\ 44-H • i - i • i • i -• H 4- - M i 1 i 1
0 400 800 1200 1600 2000 2400 2800 3200 3600 4000 4400 4800 5200 0.0652 0.0722 0.0967 0.1217 0.1475 0.1615 0.1996 0.2675 0.3627 0.4417
Internal Pressure (psi) Bend Factor (n)
Fig. 8 Variation of collapse m o m e n t w i t h internal pressure, closing c a s e Fig. 10 Variation of limit m o m e n t w i t h b e n d factor [h) 0 psi, closing
case
with increasing bend factor (ft) of the elbow, and that the
effect of internal pressure depends on both the diameter and and Findlay's limit equation (Eq. (1)) was 1.109 X 10 6 lb-
wall thickness of the elbow (the maximum effect is in the in. and was 1.147 X 10 6 lb-in. using Calladine's equation
case of highest diameter-to-thickness ratio Dlt). (Eq. ( 2 ) ) . The large displacement finite element solution for
The variation of collapse moment with internal pressure for the ideal plastic material gave an instability moment value
all the (h) values is shown in Figs. 8 and 9. The increase of 1.013 X 10 6 lb-in. As was expected, the theoretical limit
and then decrease of collapse moment with internal pressure is moments calculated using the limit equations of Spence and
clearly shown. Also, the peak value of collapse moment in- Findlay and that of Calladine are higher than the finite ele-
creases and shifts to the right with increasing (h). ment instability moment for the closing case. This is because
these limit equations are based on small displacements and
Variation of Limit Moment With Pipe Bend Factor (h).
do not take geometric nonlinearity or instability into account.
For the cases of internal pressure with values of 0 psi and 500
Therefore, these equations are not conservative when geo-
psi, the variation of instability and collapse moments with elbow
metric instability is concerned. On the other hand, Goodall
geometry is shown in Figs. 10 and 11.
approximate equation for the maximum load-carrying capac-
Figure 10 shows the variation of instability and collapse mo-
ity under closing moment gave a value of 1.077 X 106 lb-
ment with pipe bend factor (h) for the closing case without
in., which approaches the finite element solution.
internal pressure. This plot indicates an increase in both the
instability and collapse moment with the increase of pipe bend Another comparison was made between the finite element
factor. It should be noted from this figure that the instability small displacement solution and the theoretical small dis-
and collapse moments have nearly the same value for all elbow placement limit moment solution. In this case, both solutions
geometries under in-plane closing moment without internal do not take geometric instability into account. The small
pressure. This adds more significance to the angle method of displacement finite element analysis gave an instability mo-
determining collapse load as it predicts the same value of insta- ment value of 1.209 X 10 6 lb-in., which is closer to the
bility load. theoretical limit moment values than in the case of large
Figure 11 shows the case of in-plane closing moment with displacement finite element analysis. In fact, it is somewhat
500 psi internal pressure. In this case, there is a difference higher than the theoretical values, and this difference might
between collapse and instability moments, especially at low (h) be attributed to the approximations involved in each theoreti-
values, with the collapse load as a conservative estimate of cal limit load equation.
instability load. For the rest of elbow geometries, the large displacement
finite element instability were compared with the theoretical
Comparison With Analytical Limit Loads. It is useful limit moments calculated from Spence and Findlay, Callad-
to compare some of the finite element results with those ob- ine, and Goodall limit equations. The results are shown in
tained using the approximate equations of limit load pre- Table 2.
sented by Spence and Findlay (1973), Calladine (1974), or It should be noted that for the elbow geometries with h =s
that of Goodall (1978a), which were previously discussed. 0.09675, the finite element closing instability moment was
These equations are derived using limit theorems of plasticity greater than the limit moment calculated from limit equa-
and they assume ideal plastic material and small displacement tions, although it should be lower as it takes geometric insta-
(apart from that of Goodall, which is based on large displace- bility into account. At larger values of (h), the finite element
ment analysis). instability moment was lower. Therefore, it can be concluded
At first, the case of the elbow geometry with h = 0.1615 that the theoretical limit moments calculated from the limit
will be discussed. In this case, the limit moment using Spence
HI Hifi
— • — kiBtabiity
« h=01475
?
ZOE+6
» h=O.I217 —•—Gotflpse
• h=0D967
. h=0.0722 i
. h=0.0632
^ / 7
—™i • ^
5.0E*5-
0.0E+0- 1 — i —
0.0662 0.0722 0.0967 0.1217 0.1475 0.1615 0.1! 0.2675 0.3627 0.4417
600 1000 1200 Bend Faclw lh)
Internal Pressure (psi)
Fig. 11 Variation of limit m o m e n t w i t h b e n d factor (/>) 5 0 0 psi, closing
Fig. 9 Variation of collapse m o m e n t with internal pressure, closing c a s e case
2,06*6-
i J\
!
Elbows," International Journal of Pressure Vessels & Piping, Vol. 33, No. 2,
pp. 111-128.
Kitching, R., et al., 1979, "Limit Moment for a Smooth Pipe Bend Under In-
Plane Bending," International Journal of Mechanical Sciences, Vol. 21, pp. 7 3 1 -
1.06*6- • !« > ^ 738.
•i^*' ! Kussmaul, K., et al., 1987, "Investigation of the Plastic Behaviour of Pipe
0.06*0 ; t = 1 - " " i.—i—I—1__ Bends," ASME PVP-Vol. 127, pp. 55-66.
0.0632 0.0722 0.0967 0.1217 0.1475 0.1615 0.1998 0.2675 0.3827 0.4417
Marcal, P. V., 1967, "Elastic-Plastic Behavior of Pipe Bends With1 In-Plane
Bend Factor (h) Bending," Journal of Strain Analysis, Vol. 2, No. 1, pp. 84-90.
Rodabaugh, E. C , 1979, "Interpretive Report on Limit Load Analysis and
Fig. 12 Variation of analytical and finite element limit moments with Plastic Deformation of Piping Products," Welding Research Council Bulletin,
bend factor (h) No. 254, pp. 65-82.