Documentos de Académico
Documentos de Profesional
Documentos de Cultura
1860
ic h .
ic h .
548
ELASTICITY OF P IP E BENDS
549
5
Before proceeding along independent lines an extensive
search was made for existing literature, which disclosed th a t very
little had been written on the subject. However, the following
articles are of special interest:
(A) Formanderung und Beanspruchung federnder Ausgleichrohren (Strains
and Stresses in Expansion Bends), by Prof. A. Bantlin, Zeitschrift des Vereines
deutscher Ingenieure. Jan. 8, 1910, p. 43.
(B) Expansion of Pipes, by Ralph C. Taggart, Trans. Am. Soc. C. E.,
paper No. 1167, D ec. 1910.
(C) Elasticity and Endurance of Steam Pipes, by C. E. Stromeyer, Engi
neering, June 19, 1914, p. 857 (from a paper read before the Institution of Naval
Architects).
550
ELASTICITY OF P IP E BENDS
(Z>) Pipe Bends, Their Growing Use and Efficacy, The Valve World,
Oct., 1915 (published by Crane Co.).
(E) The Design of Pipe Bends for Expansion in Pipe Lines, by J. G. Stew
art, Power, M ay 10, 1921, p. 742.
6 Reference (A) deals with the comparison of physical tests
and calculated results for bends similar to our double-offset expan
sion U-bends. The results are tabulated for comparison and ex
planation of discrepancies rather than for use by a designer laying
out'pipe lines.
7 Reference (B ) treats principally of the use of straight pipe
and fittings in making up expansion loops. The results are in gen
eral similar to those of the present authors for straight lengths of pipe.
8 Reference (C) describes a series of failures of steam pipes
in service, due to repeated strains beyond the elastic limit. A set
of formulas similar to those presented in this paper was worked out
to show what elongations could be cared for without eventually
producing failure.
9 Reference (D) is a report of physical tests on pipe bends
made by the Mechanical Experts D epartm ent of Crane Company,
and published in their organization paper, The Valve World. A series
of curves deduced from their experimental data and giving the
amount of expansion required to produce a fiber stress of 15,000 lb.
per sq. in. in different-shaped bends was included for the use of those
interested in pipe designs.
10 Reference (E ) gives a method of computing fiber stress
similar to th a t of Reference (C), and a chart for reading the elonga
tion corresponding to 15,000 lb. per sq. in. stress for different bends.
11 I t is worth comment th at while the formulas which follow
were worked out independently of Reference (C) and before Refer
ence (E ) was published, all three sets are practically identical. The
comparison of theoretical formulas with actual test results, and the
rather complete design graphs worked out in this paper, take the
subject a step beyond the point reached by earlier publications.
The fact th at none of the previous investigators has published all
of the mathematical steps leading up to his final formulas, will make
the derivations given by the authors of special interest to anyone
having occasion to work with the formulas.
12 Table 1 is a summary of such data published in the above
references as were in shape to reduce to a common form for comparison
with the work of the present authors. I t will be noted th a t the
formulas of References (C) and (E) agree very closely with those of
551
this paper, while Reference (D), which has been widely quoted dur
ing the past seven years, differs considerably from the others except
for the double-offset expansion U-bend. The following extract is
quoted from Reference (D ):
A 180-deg. or U-bend has twice the expansive value of a 90-deg. or quarterbend of the same size and radius, and an expansion U-bend, four times the ex
pansive value of a quarter-bend or twice that of a U-bend. A double-offset
expansion U-bend has five times the expansive value of a quarter-bend, two and
one-half times that of a U-bend, and one and one-fourth times that of an expan
sion U-bend.
C
K
S
D
R
E
I
F
A
e f in it io n s
of
Sym bols
= stress index (num erical values given in le ft half of tab le). (F orm ulas 3, 7, 9, 11.)
* flexibility index (num erical values given in rig h t half of tab le). (F orm ulas 2, 6, 8, 10.)
= m axim um bending stress in pipe wall, lb. per sq. in.
= outside diam eter of pipe, in.
= m ean radius to which ben d is form ed, in.
= modulus of elasticity (30,000,000 lb. per sq. in. for steel).
= m om ent of in ertia of section, ta k e n norm al to center line of pipe of which bend is form ed, i n . 4.
= force causing deflection of bend, lb.
= deflection of flange of ben d = elongation of pipe line due to change in tem p eratu re, in.
= to ta l expansion to be absorbed b y bend, i.e., to ta l expansion calculated for a s tra ig h t ru n of
pipe w ith length equal to distance betw een anchors which occur nearest to bend under
consideration.
552
ELASTICITY OF P IP E BENDS
553
than indicated by calculation, which of course assumes a pipe with smooth walls.
If this view is correct, the phenomenon of the great elastic deflection of the ex
pansion bends is in perfect accord with the purpose for which such an expansion
bend is placed in long steam lines, and the greater elasticity yielded by the abovementioned waves should be regarded as a valuable property possessed by the
expansion bend.
In order to determine by experiment the correctness of the assumption
regarding the effect of the waves to prove it by calculation would necessitate
very complicated and involved equations of little value a cast-iron lyre-shaped
pipe was made in which the waves or folds on the compressed side of the bend
were absent. The pipe, of about 200 mm. (7.87 in.) inside diameter and about
18 mm. (0.7 in.) wall thickness, had the same bend radius as the steel pipe. It
was furnished by Sulzer Bros., Winterthur, and represented an outstanding ex
ample of foundry technique when we consider the difficulty presented by the pro
duction of a test bend of this complicated design and of such dimensions. The
deflections of this cast-iron pipe were determined by the same method and by
the same equations as were employed in the case of the steel pipe.
From these experiments with the cast-iron pipes, together with those con
ducted with the solid steel lyre-shaped bodies, we are led to conclude that the
large deflections of the steel pipes must be attributed chiefly to waves or folds
upon the compressed side of the pipe.
A N A L Y S IS
554
ELASTICITY OF P IP E BENDS
body, the movement of its free end will be proportional to the force
acting, within the elastic limit of the material.
17
Consider any pipe bend, as in Fig. 1, showing a quarterbend, with force F acting as shown. Flange A 'is fixed; flange B is
free to move in the direction in which the force is acting. Let A be
the total deflection of flange B measured in the direction of F, and
caused by the force as it changes from 0 to F; then, considering the
problem on an energy basis, the total work done by F will be 1-FA,
or the average force times the distance moved. The work done by
F is also equal to the internal work of distortion in the bend. To
obtain an expression for the work of distortion, consider an element
of the bend of length .ds (Figs. 1 and 2). Let M = bending moment
a t this section. The unit bending stress S existing at a distance y
from the neutral axis N -N is M y / 1 , 1 being the moment of inertia
of the section. The total stress 2<S on a fiber of cross-section da
located a t this distance from the neutral axis is M yda/I. Let I be
the elongation of this fiber and E the modulus of elasticity; then
556
ELASTICITY OF P IP E BENDS
557
T H E U -B E N D
21
Consider Fig. 4, with a force F acting along the line A B
as shown. Flange A is fixed. A is the deflection of flange B measured
in the direction of F as before.
558
ELASTICITY OF P IP E BENDS
Substituting,
THE
E X P A N S IO N
T J-B E N D
22
This bend is more complicated than the ones already con
sidered and m ust be treated in a slightly different manner. The
F ig . 5
E x p a n s io n U - B e n d
559
24
The maximum unit bending stress S in the expansion Ubend occurs a t C. Let D be the outside diameter of the pipe,
THE
D O U B L E -O F F S E T E X P A N S I O N
U -B E N D
560
ELASTICITY OF P IP E BENDS
27
Now consider the portion CD, and assume it fixed at
C. A2 is the deflection of B due to bending in CD only, and is measured
in the direction of F.
F i g . 7 Q u a k t e r -B e n d
and
561
T a n g e n t , F o b c e A x ia l
562
ELASTICITY OF P IP E BENDS
FE I
Q U A R T E R -B E N D A N D T A N G E N T , F O R C E P E R P E N D IC U L A R
33
Consider Fig. 8 with force F acting as shown, parallel to
AC. Flange A is fixed. Let A be the total deflection of flange B,
measured in the direction of F. Let Ax be the deflection of the
flange B due to bending in the portion CB and A2 the deflection due
to bending in the portion AC, then A = Ax + A2.
563
4 h ii
37
Lastly, consider Fig. 9. Force F acts as shown, parallel to
AC. Flange A is fixed. A is the total deflection of flange B, measured
564
ELASTICITY OF P IP E BENDS
3E l
565
GENERAL
566
ELASTICITY OF P IP E BENDS
D E S IG N
GRAPHS
569
of the pipe, the value of the stress for any size of pipe m ay be ob
tained after finding the stress in an 8-in. pipe. The proportion be
tween stresses can be made directly on the diagram by following
the radial lines diverging from its lower right corner.
S a m p l e P r o b l e m : A 12-in. double-offset expansion U-bend having a radius
of 90 in. is to take up an expansion of 1 in. Required, the maximum bending stress
in the bend.
From the intersection of A = 1 in. and R = 90 in., extend horizontally to
the vertical line under 8-in. pipe, which is referred to as the base line, and
then obliquely (with a line passing through lower right-hand comer) until under
12-in. pipe on the upper scale. The result is 2040 lb. per sq. in.
Cn
ELASTICITY
OP
PIPE
BENDS
571
diagram s the pipe size refers to th e size of pipe used in th e bend and
has n o th in g to do w ith th e size of pipe in th e re st of th e line.
M ETHOD
OF TEST
50
T h e m eth o d u sed in d eterm in in g ex p erim en tally th e re
lation betw een force an d deflection is show n in Fig. 15, w hich is a
p h o to g rap h of th e te s t set-u p for a 6-in. double-offset expansion
U -bend. T h e left end of th e b end w as held rigidly in place, while
the rig h t end was deflected b y m eans of a h y d rau lic jack . T he
am ount of deflection was m easu red w ith a steel tap e , an d th e force
causing deflection was com p u ted from th e readings of a pressure
gage a tta c h e d to th e reservoir b e n e a th th e ram of th e jack. T he
far end of th e bend was held in place b y securely b o ltin g it to the
F ig .
15
S e t-U p
fok
T kst
of
E l a st ic it y
sio n '
of
D ouble-O ffset
Expan
U -B e n d
572
ELASTICITY OF P IPE BE N D S
51
T h e p ressu re re q u ire d to overcom e in itial friction before
a n y m o v em en t of th e b end to o k place, was determ in ed roughly by
lig h tly w edging th e p o in t of a pencil b etw een th e m ovable head an d
th e s ta tio n a ry b a rre l of th e jack , so th a t it w as su p p o rted hori
zo n tally , a n d th e n carefully p u m p in g th e ja c k a n d n o tin g th e pres
su re a t w hich th e pencil w as released, a v e ry slight m ovem ent of
th e h ead being sufficient to le t th e pencil fall. Before th e gage
p ressure w as recorded th e end of th e b end was jo lted u p and down
so as to relieve a n y b in d in g actio n th a t m ig h t be present.
F i g . 16
T e s t S e t -u p w it h R am E x t e n d e d A g a in s t t h e E x p a n s io n B e n d
52
In m easu rin g th e deflection, one end of th e steel ta p e was
hooked over th e h ead of a sm all stove b o lt screw ed in to th e to p of
th e p ipe in th e b en d n e a r th e fixed flange. M easu rem en ts were
ta k e n to a sm all chisel m a rk on th e free end of th e bend n e a r th e
flange. I n th is w ay errors in read in g th e deflection due to a m ove
m e n t of th e en d su p p o rts w ere elim inated. T h e distance betw een th e
ends of th e b e n d w as m easu red first w ith no force applied to th e free
end, an d th e n w ith th e h ead of th e ja c k in different positions ac
cording to th e p ressure applied, n o tin g th e gage pressure on th e
ja c k each tim e. B efore each read in g w as ta k e n th e ja c k was lined
u p so t h a t th e line of actio n of th e force passed th ro u g h th e cen ter
of th e fixed flange. T h e free end of th e bend was also jo lted u p and
573
F i g . 17
F orce
v s.
D i s p l a c e m e n t o f a 10- i n . E x p a n s i o n U - B e n d T e s t
R esults
port and hold in place, the line of action of the force passing through
the pipe to which the bend was connected. All pressure gages were
calibrated on a gage tester before being used. Low-reading gages
were used for the small forces to insure greater accuracy.
54 Fig. 16 is a photograph showing the jack with the ram
extended against the bend. The 6-in. scale shown above the flange
was for photographic purposes only and was not used during the
tests. Fig. 15 is the same set-up in the released position.
55 In calculating the readings taken, the frictional force was
subtracted each time from the total force acting, as given by the
corrected gage pressure. This gave the true force causing deflec
tion. The total deflection was obtained by subtracting the corre
sponding tape reading from the reading when in the released position.
574
ELASTICITY OF P IP E BENDS
RESU LTS OF TESTS
F i g . 18
F orce
v s.
is p l a c e m e n t
of
esults
6 -i n . E x p a n s io n U -B e n d T es t
575
from the formulas with one exception, this exception being the
10-in. expansion U-bend shown in Fig. 17. The radius in this case
was 6 diameters. The pipe in this bend was the largest size tested,
all others being 6-in. or smaller. I t is relatively more difficult to
bend large pipe than small pipe, so as the pipe size increases the
minimum allowable radius (expressed in pipe diameters) also has to
be increased. One American m anufacturer states th a t the shortest
radius to which a 22-in. standard pipe can be bent is 6 diameters,
F ig .
19
F o r c e v s . D is p l a c e m e n t
of
6 -i n . E x p a n s io n U - B e n d T e s t
R esu lts
576
ELASTICITY OF P IP E BENDS
thickness of the pipe wall was the one used. The pipe wall was
measured with special calipers a t several points a t each end of
the pipe, and the average value used in the calculations.
59 The expansion U-bend shown in Fig. 18 was one purchased
especially for test and had not previously seen service. In this
case there is a very satisfactory agreement between test and formula.
60 The expansion U-bends shown in Figs. 19, 20 and 21 were
bought for the auxiliary superheated-steam line a t the Connors
Creek power house and had not seen service at the time of test.
F ig .
20
F o r c e v s . D is p l a c e m e n t o f a 6 -i n . E x p a n s io n U - B e n d T e s t
R esu lts
577
F ig .
21
F o r c e v s . D is p l a c e m e n t o f a 6 -i n . E x p a n s io n U - B e n d T e s t
R esu lts
578
ELASTICITY OF P IP E BENDS
are used in practice, the designer will always be on the safe side.
His bends may or m ay not be more flexible than his calculations
would indicate, depending upon how the bends were made, but he
need not fear th at forces or stresses in excess of his calculated values
will exist.
64
All of the bends were tested while cold. Since the phy
sical properties of mild steel are not altered appreciably up to 700
or 750 deg. fahr., which are about the maximum temperatures to
be expected in a steam line with present practice, the relation be-
F ig . 22
F obce
v s.
D is p l a c e m e n t o f a 6 - i n . D o u b l e -O f f s e t E x p a n s io n
TJ-B e n d T e s t R e s u l t s
tween force and deflection should be the same whether the pipe
is hot or cold.
65
The test method used was a rather crude one, yet in sev
eral cases the agreement between calculated and test values is
remarkably close. I t will be noted th at in some cases there is a
tendency for the force to become disproportionately larger as the
displacement increases. This was doubtless due to the binding of
the jack as the flange tended to swing to one side, and the flanges
to assume positions a t an angle to the pipe line. Considering the
problem on a purely academic basis, there should be a hinge or
knife-edge connection a t both flanges in the test set-up, in order
to eliminate the errors in testing due to this movement of the flanges;
then, so long as the line of action of the force passes through A B ,
579
Figs. 5 and 6, the results should agree with the formulas if the
deflection is small and the inner side of the bend is free from waves
or folds in the pipe. I t should also be noted th a t in deriving the
formulas no allowance was made for a change in shape of the bends
as they take up expansion. Actually, the height of the bend, and
therefore its flexibility, increases as the bend is deformed, so th a t
for large deflections the actual force will be less than the calculated
force. However, as the deflections are relatively small in actual
practice, this will not affect the results seriously.
66
In checking a piping layout to determine if excessive fiber
stresses exist anywhere in the pipe wall, the longitudinal bursting
F ig . 23
Fobce
v s.
D is p la c e m e n t o f a
D o u b le - O f f s e t E x p a n s io n
U - B e n d M a d e f k o m 4 - in . S t a n d a r d P i p e T e s t R e s u l t s
stress due to the pressure of the steam should be added to the bend
ing stress given by the formulas. This is especially im portant
when working with high steam pressures.
67
Nothing has been said about the expansion which can be
compressed back into the pipe by the forces set up a t the anchorages.
In each case it has been assumed th at the expansion bend takes
cafe of all of the expansion and th a t none of it is compressed back
into the straight pipe. Actually, of course, a small part of the
expansion is taken up by compression, the amount being directly
proportional to the length of the pipe and to the force set up, and
580
ELASTICITY OF P IP E BENDS
F ig . 24
F orce
v s.
D is p l a c e m e n t o f a n E x p a n s i o n U - B a n d M
3 j - i n . B o il e r T u b e T e s t R e s u l t s
ade
fr o m
DISCUSSION
581
582
E LA STICITY OF P IP E BENDS
DISCUSSION
F ias. 25, 26
and
27
583
C o m p a r is o n o f S q u a r e L o o p s a n d P i p e B e n d s
584.
ELASTICITY OP P IP E BENDS
to each other, and so secure a flexibility without the use of the elabo
rate bends described in the paper.
A valuable addition to the paper would have been some dis
cussion of resistance of piping to torsional stresses, as it is sometimes
difficult to arrange piping to give flexibility in every direction with
out recourse to its torsional elasticity.
The general arrangement of steam piping to power units demands
careful thought, not only to provide against expansion of the piping
due to heat, but th a t small misalignment of flange may be compen
sated for in every direction.
R o b e r t C r a m e r said th a t it was gratifying to learn th a t the
authors had made an attem pt to compare directly theoretical and
experimental results. He noted in the paper, however, a point of
theoretical nature which should be considered when test results
were compared with calculated results. In the diagram giving the
basis of the theoretical investigation, Figs. 1 and 2, the neutral fiber
of the cross-section of the bend was assumed to pass through the
center of the circle which was not true in the case of a pipe bend. It
was a well-known fact, he said, th at the neutral fiber in curved beams
did not pass through the center of gravity of the cross-section as it
did in straight beams, and the peculiarity was more pronounced the
shorter the radius of the beam compared with the magnitude of the
cross-section.
For this reason, he said, it would pay the authors to pursue a
line of more rigid theoretical investigation of the deviation of the
experimental results from the theoretical results in the case of the
shorter bends. He said th a t he noted in making comparison between
the experimental and calculated results th a t in some cases the two
checked closely while in others the variation was very marked.
I t was not the slight variation to be expected from experimental or
observational errors but one which would suggest th a t the authors
should look for some error in their calculations.
The authors considered the deflecting force always to be at
right angles to the flange on which it was applied with the exception
of the case of the quarter bend, in which they had distinguished be
tween two cases, one with the force acting a t right angles to the
flange and the other with a force acting in the plane of the flange.
In the actual case, he said, the problem was really more complicated
than would appear from the authors investigation as it was fre
quently necessary to twist the flange in order to make a tight joint
and this involved another source of strain.
DISCUSSION
585
586
ELASTICITY OF P IP E BENDS
tions taken, 80 the small change of this angle a t bending. Let the
outer fibre ab after bending take the position aj)\ and its small dis
placement towards the neutral axis be denoted b y /. The exten
sion of the fiber is
I = aib j ab = a J i axci (ab a ^ )
. . . .
[a]
DISCUSSION
587
The first member on the right represents the unit elongation due to
the rotation of the cross-section bd in respect to the cross-section ac.
The second member on the right side gives us the effect of the dis
tortion of the cross-section. I t is easy to see th a t this effect is very
substantial. If we take, for instance, R + r = 60 in., and put / =
0.02 in. only, we obtain
588
ELASTICITY OF P IP E BENDS
->
F x = kF
........................ [21]
DISCUSSION
589
tR
We see that in cases where - r is small, the real maximum stress is
r2
590
ELASTICITY OF P IP E BENDS
BEND
C O N S ID E R E D A S IN IT IA L L Y
CURVED BEAM
In a pipe bend of radius R (see Fig. 29) the neutral axis N ' N '
of any radial cross-section is situated a t a distance h measured toward
E n la rg e d
S e c + io n C~C
^ A
where A = RZ ( - f i , )
\y.+ .R J
................................................
- - [28]
DISCUSSION
591
da = element of area
y = vertical distance from central axis to element of area
A = hda = area of section.
Applying the formula for A ' to a pipe bend (see Fig. 29) then:
da = trdd
y = r sin 6
592
ELASTICITY OF P IP E BENDS
F ig . 30
R e la tio n o f
a n d N o m i n a l P i p e S iz e s
a - V B2 - r2 - R
R = rad iu s of pipe bend
r m ean radius of pipe
Sizes 2 in. 12 in. are s tan d a rd weight pipe
Sizes 14 in. 20 in. are O .D . pipe
in. thick
M axim um fiber stresses of Fig. 12 are to be increased b y percentage - 100 per cent.
r
t m u st be ta k e n as m ean radius of pipe an d n o t as nom inal radius.
DISCUSSION
593
pipe. The authors have not tested any loops of this kind, but the
formulas worked out in their paper and given in connection with
Figs. 7 ,8 and 9 apply to such loops. Fig. 31 has been added to furnish
a comparison between square loops and pipe bends occupying about
the same space. An obvious disadvantage of the square loops is
th at they have four more joints per loop than a pipe bend (provided
the bend can be made from one piece of pipe) and th a t they introduce
two sharp 90-deg. elbows in place of one easy 180-deg. bend.
EFFECT OF INTERNAL PRESSURE AND TEM PERATURE
594
ELASTICITY OF P IP E BENDS
F ig . 32
S p e c ia l U -B e n d M
a d e fr o m
Qu arter B ends
and
angents
595
DISCUSSION
Consider Fig. 33 with a force F acting along the line AG. Flange
A is fixed.. A is the total deflection of flange G measured in the
F ig . 33
S p e c ia l E x p a n s io n U - B e n d M
T angents
ade
fro m
Qu a rter B en d s
and
596
ELASTICITY OP P IP E BENDS
T=arL
where
T
q
r
J
where
F = force, pounds
and
Li = distance in inches of line of action of force F from center
line of pipe under torsion.
g _ shearing stress
shearing strain
G = modulus of shearing elasticity (11,500,000 lb. per sq. in.
for steel)
-
Shearing strain =
Li
6 = angle of twist
L = length of pipe under torsion, inches.
DISCUSSION
597
598