Professional Documents
Culture Documents
Karabin
Alcoa Laboratories,
Alcoa Center, Pa. 15069;
Stress Component Indices for
formerly, Westinghouse Waltz Mill Site
Elbow-Straight Pipe Junctions
E. C. Rodabaugh
E. C. Rodabaugh Associates, Inc., Subjected to In-Plane Bending
Hilliard, Ohio 43026
Through use of the computer code BENDPAC, a number of solutions for elbows
attached to straight pipes are produced. This study focuses on the stresses at the
J. F. Whatham elbow-straight pipe junction. It is noted that the largest stress component is in the
Australian Atomic Energy Commission, hoop direction for h values less than 0.5, whereas for h greater than 0.5, the axial
Sutherland, Australia stress dominates. Simple expressions for stress indices in both directions are given.
Introduction
Recent analytical work by Whatham [1], resulting in the equilibrium over the pipe cross section, and £ | are self-
computer code BENDPAC, makes a parametric study of equilibrating but give the displacements. The !•% are of the
elbows feasible both in terms of time and expense. Thin- form
walled elbows subjected to one of five loading conditions can
be analyzed. Elbow end conditions may be either free,
flanged, flanged tangents, or infinitely long tangents, with dif- «= £a«/("0)
ferent conditions at each end.
The purpose of this work is to study elbows subjected to in- where f(nff) represent harmonic functions, either sine or
plane bending. Most of the elbows which are analyzed are con- cosine. For the straight pipe, the solutions, if, are simply the
nected to infinitely long straight pipe tangents, but one set of result of a straight pipe subjected to a bending moment. The
problems investigates elbows connected to finite tangents with H are solved through a system of eight equations—three
rigid flanges. The primary focus is centered on both major equilibrium, three force displacement relationships, and equa-
stress components (hoop and axial) at the junction of the tions relating the moment and rotation to the other forces and
elbow and straight pipe, which is commonly a location for a displacements.
weld. Simple design guidelines are suggested for both stress Expressions for if are of the form
components over a wide range of elbow geometries.
Figure 1 describes a number of the parameters associated
with the model. Definitions for the terms in the analysis are
^=tcje-^t^jf(^)
y'=i
given in the Nomenclature. where f(n&) are the appropriate harmonics, fly are the eigen-
values, and ri= Roc/r for the curved pipe and i\ = l/r for the
Background straight pipe; C, A, if are all complex; for in-plane bending,
The analytical model used to obtain the results in this study J=2N. Effects from opposite ends of the curved pipe are
is presented in a series of papers [1-3]. The eight unknowns of
the problem, three displacements, three forces, the in-plane
moment, and corresponding rotation are of the form
* = £' + ?
where £" are the solutions for an unterminated pipe or bend,
and the if represent the solution for a self-equilibrating stress
system which decays exponentially with distance from the
curved-straight pipe junction and flange. The unterminated
bend unknowns consist of
Contributed by the Pressure Vessels and Piping Division for publication in the
JOURNAL OF PRESSURE VESSEL TECHNOLOGY. Manuscript received by the
Pressure Vessels and Piping Division, September 25, 1984; revised manuscript
received April 11, 1985. Fig. 1 Problem description
L/R
h
R/r
0.0476 2 3 5 2 2 2
0.0714 2 3 5 3 3 3
0.143 2 3 5 3 3 3
0.191 2 3 5 2 2 2
0.286 2 3 5 3 3 3
0.715 2 3 5 3 3 3
0.954 2 3 5 5 5 5
1.43 2 3 5 5 5 5
Table 2 oH/S„
Problem
set no.
u5 9 9
B B
a» a
, | B
• "
lo
a -
0
HOOP STRESS
J °-51— a
B
a
B
a • a
B
B B
a .
a
a- HOOP STRESS
b B - AXIAL STRESS
B- AXIAL STRESS
t B I
« 0 = 90° , R/r=5 B
a a a a
1.0/0.1 a„ = 90° , FINITE TANGIENTS B
1.0/0.1 ".. I a "
a a B • 0.5 a a
+ a a a
0.5 9 B B
a B
0.1 1 1 a B
0.1 |
Fig. 2 Maximum stresses at the elbow end (« 0) for the 90-deg 0.01 0.1 1.0 10
elbows attached to long tangents
Fig. 3 Maximum stresses at the elbow end (a = 0) for the 45-deg,
Tables 2 and 3 present the maximum stresses at the welds 180-deg and those elbows with finite tangents
(oc =0) in the hoop and axial directions, respectively. These
stress values are normalized by S„, which is equivalent to the most highly stressed section within the elbow at the same cir-
nominal bending stress in a straight pipe. Also given in the cumferential location. Dhalla looks at the case ofoc0 = 90 deg,
tables are the location of maximum stress (6 value, inside (I) or R/r = 3, /J = 0 . 1 3 , and L/r = 4 (approximating the infinite
outside (O) surface) and the sign of the stress. This same infor- tangent). Carry-over factors of 0.48 in the hoop direction and
mation is portrayed in Figs. 2 and 3; but in these figures, the 0.55 in the axial direction are reported. In the present analysis,
normalization factor is amc, where amc = C2S„ and a comparable problem (h = 0.14) gives carry-over factors of
C2 = 1.95/h2/3. The open points in these figures are for 0.52 and 0.55 in the hoop and axial directions, respectively.
problems whose thickness ratios exceed the limits of thin shell This provides analytical confirmation from a second indepen-
theory (t/r> 0.3) as established by Watham [2]. dent source.
Comparisons of these numbers with the results obtained by Figures 2 and 3 exhibit some common trends. For the cases
Rodabaugh, et al. [5] shows a general agreement. In most with infinitely long tangents, aH/amc is relatively constant and
cases, the stresses obtained in this analysis are greater than ranges from 0.5 to 0.6 for values of h less than 0.5. For larger
those obtained in the Rodabaugh work. Contributing to the values of h, aH/omc decreases. The axial component always is
differences in the results are the solution technique (the finite substantially smaller than the hoop component for small h
element method is used in [5], and the stress at the weld values. It increases slowly at first and then surpasses the hoop
represents the average of the surrounding elements) and the component in the vicinity of h = 0.5. This same switch in max-
fact that rigid flanges are present in the Rodabaugh analysis at imum stress direction from hoop to axial at h = 0.5 is also
L/r values of four to five (this should nearly approximate the noted in [5]. The effect of the nearby flange is seen in the last
situation of long tangents present in problem groups one set of plots in Fig. 3. For h values less than 0.5, the hoop stress
through five in this analysis). Despite these differences, the is notably less than in any of the five previous cases. This is
two analyses exhibit agreement to within 20 percent for small due to the restriction on ovalization imposed by the flange.
h values and to within 5 percent for large h values. Since both ends of the elbow are attached to pipes with nearby
Another finite element analysis looks at carry-over factors rigid flanges, both ends have reduced hoop stresses, but in
for one specific case (Dhalla [6]). A carry-over factor is equal cases where the flange is only present at one end of the elbow
to the ratio of the stress at the elbow end to the stress at the [7], the carry-over factor is enhanced at the end where there is