Beruflich Dokumente
Kultur Dokumente
This paper is an abstract2 o f a m anuscript subm itted author's conclusions are n ot discussed in th is abstract
by Mr. Bailey a t th e request o f th e P lasticity C om m ittee nor are th e m ath em atical derivations included. The
o f th e Applied M echanics Division through its chairm an, reader interested in th ese details should consult the
Dr. Nadai. The paper is also an abridgm ent o f Mr. London paper. Footnotes have been added to th is ab
Bailey's paper read before th e In stitu tion o f M echanical stract m en tion in g references w hich are readily available
Engineers, London, November 22, 1935. The purpose o f and w hich will supplem ent and illu m in ate th is abbre
this abstract is to indicate th e wide range o f topics viated treatm en t. Comprehensive and som ew hat differ
treated fully in th e original paper and to give a u seful and en t abstracts o f th e original London paper have also
adequate sum m ary o f th is pioneer work by Mr. Bailey. appeared in “Engineering,” November 29 and December
In m any cases, because o f lack o f space, th e bases o f the 13, 1935, and in “The Engineer,” Decem ber 6, 1935.
ESIGN of parts for elevated temperatures where a con and th a t the sign in front of the parenthesis be changed in this
I t is of interest to note from Fig. 1 th a t the axial creep rate be added to a system of stress without altering the creep behavior.
becomes zero at 450 lb per sq in. which stress is exactly one-half The results of further corroborative tests are shown in Fig. 3.
the circumferential stress and is the stress system set up by Here a thin-walled tube was subjected to combined torsion and
internal fluid pressure in a cylinder. The existence of this zero tension while the maximum shearing stress was kept constant.
creep rate, together with lack of creep in the wall thickness of a The creep rates in the directions of the principal stresses are
tube in simple torsion, forms the basis of the author’s important deduced from experimental results and check fairly well with the
calculated curves. The disagreements in the axial creep rates
shown by the lower curve are very likely due to a departure from
isotropy of the material. Another figure (not included in this
abstract) shows the results of further tests on isotropic material
and the agreement is almost perfect.
One additional test was made where torque was applied to a
lead tube under internal fluid pressure with results as shown in
Table 1. The fluid pressure developed a circumferential tensile
T A B L E 1 C R E E P R A T E W H E N D IA M E T R A L C R E E P R A T E F O R
F L U ID P R E S S U R E A L O N E IS R E P R E S E N T E D B Y U N IT Y
C ircum ferential
A xial (tw ist) D iam etral
R esults of te s ts .............................. 0 .4 2 8 5 .73 6.12
R esults of c alcu la tio n .................. 0.314 7 .6 5 6 .0 5
F ig . 3 C o m b in e d T o r s io n a n d T e n s io n T e s t s o n 0 .1 1 5 P e r C e n t
C a r b o n M. S . T u b e
[M axim um shear stress = 3.12 to n s per 8q in . (constant) a t 475 C.]
A question th a t arises is: How should bending and twisting lined. The results of such an analysis are shown in Figs. 6 and 7
moments arising from loading due to therm al expansion be taken for a 20-in. diameter disk with hub, turning a t 3000 rpm. In
into account? The effect of bending and twisting moments Fig. 6 the stress distributions are shown and in Fig. 7, the dis
upon a cylinder under fluid pressure and creep has been studied tribution of the creep rate in terms of the circumferential creep
in the original paper on the basis of the general Equations rate at the bore as unity. The heavy lines are for creep con
[1], [2], and [3], In the case of torsion, it has already been seen ditions and the light lines represent elastic conditions. The
th a t the predicted behavior for a lead pipe under hydraulic effect of creep in reducing the maximum stresses th at would
pressure and torsion was confirmed by experiment. I t is con occur with purely elastic action is plainly evident.
sidered th a t the results obtained by analysis in the case of carbon
steel are no less reliable. Briefly, analysis shows th a t the
presence of stress due to the fluid pressure serves to increase
greatly creep caused by bending or twisting with the result th at
any appreciable stress due to bending or twisting at the operating
temperature would be quickly reduced by creep to negligible
F ig . 5 S t r e s s , C r e e p R a t e , a n d T e m p e r a t u r e D i s t r ib u t io n in
C y l in d e r s U n d e r I n t e r n a l P r e s s u r e
see the time for limiting creep to occur is 27 hr. This curve during the early part of the heating. Fig. 10 shows a ehain-dotted
shows th at any preliminary heating of less than 20-hr duration line A B representing a constant heating effect upon microstruc
gives an improvement in creep resistance or a stiffening. ture equivalent to 20 hr a t 650 C. Consequently, assuming that
Upon the basis of changes in microstructure due to prolonged the creep behavior is in accord with microstructure as was
found in the case of carbon steels examined, the curve A B divides
TABLE 4
100,000 hours a t the area of Fig. 10 into two regions, the area above A B being a
450 C 475 C 500 C region where the creep resistance becomes less than the initial value,
T im e of h eatin g a t 650 C, h r ................... 4 .9 2 2 .9 9 4 .1
and the area below a region where superior resistance prevails.
Failure of Low-Alloy High-Creep-Resistant Steels. The author
discusses finally the failure of low-alloy high-creep-resistant
steels. It is concluded th a t “when an alloying element is used
to increase creep resistance, th a t is, resistance of the grains to
deformation, if it does not improve the endurance of the grain
boundaries to a like extent, and the time for the full deformation
of the grains to occur becomes greater than th at for intercrystal
line separation to take place, failure by the latter mode ensues.”
Fig. 13 shows the development of intergranular failure of a
molybdenum steel under creep as revealed in a photomicrograph.
C o n c l u s io n
An attem pt has been made to deal briefly with some of the
more important factors concerned in the development of a
rational treatm ent of creep for the purpose of engineering design.
In spite of the complicated nature of the phenomenon of creep,
there is no reason why its treatm ent for design should not be
made as satisfactory and reliable as is the treatm ent used for
materials under elastic conditions. The author hopes th at the
paper may be useful as a contribution to this object and th at it
may lead to much needed investigation directed to the under
standing of creep under a general system of stress.
F ig . 12 E f f e c t U p o n C r e e p o f P r e l i m i n a r y H e a t i n g o f M o l y b
d e n u m S t e e l a t 650 C a t a S t r e s s o f 6 T o n s p e r S q I n . A cknow ledgm ents
heating, the author has also formulated Table 4 showing the Acknowledgment is due to the directors of the Metropolitan
time of heating at 650 C to produce an effect equivalent to Vickers Electrical Co., Ltd., for their permission to publish the
100,000 hr at temperatures of 450, 475, and 500 C. This table is m atter contained in the paper,
based on the relation t =
0 0 0 /D ^ w jje r e i = time in
hours; T = absolute tempera
ture, deg C; e = base of Nape-
rian logarithms; A = a constant
the magnitude of which depends
upon the character of the steel
and its history.
These values have been spotted
on the curve of Fig. 12. The
“450 C” point, for example,
shows that a preliminary heat
ing at 450 C for 100,000 hr pro
duces the same resistance to
creep as would heating for 4.9 hr
at 650 C, 4.9 representing the
abscissa of the point. From the
curve we see th a t any prelimi
nary heating for 100,000 hr below
a temperature of 475 C produces
the same stiffening as a prelimi
nary heating of 20 hr or less at 650
C. From this it may be in
ferred th a t the influence of oper
ating temperatures up to 475
C is to improve the creep resist
ance of the material. The author
shows further th a t this range
may be extended to 500 C by
making an allowance for the
improved resistance occurring F i g . 13 D e v e l o p m e n t o f I n t e r g r a n u l a r F a il u r e o f a M o l y b d e n u m S t e e l U n d e r C r e e p 1 0 0 X
T h e C reep of M etals— I I 1
By A. NADAI* and E. A. DAVIS,3 EAST PITTSBU RG H, PA.
I
F a ductile metal is deformed permanently at low temperature,
the stress a must be increased to produce further plastic
strain e. The stress under which it yields is not changed
much by increasing the speed of stretching, so that in first
approximation the stress a can be expressed by a monotonously
increasing function of the plastic strain e
In many applications it is necessary to separate the elastic For ductile metals with a high melting point, such as steel and
(recoverable) portion of a strain from its plastic or permanent copper, Equation [7] cannot be a true expression of the facts at
portion. This is the case, for example, for an elastico-viscous lower temperatures because Equation [7] does not take account
material. of strain hardening and because the plastic speeds of deformation
(b) The Elastico-Viscous Material. Of the total strain e (see it" increase much more rapidly with increasing yield stresses a
Fig. 1) than would follow from the proportional relation given in Equa
tion [5]. At sufficiently high temperatures where softening
1 For p a rt I see “ The Creep of M etals,” T rans. A .S.M .E ., vol. starts, such a relation as Equation [5] may become quite useful,
55, 1933, paper APM-55-10, p. 61. particularly for amorphous materials such as glass or materials
2 3 Research Laboratories, W estinghouse E lectric and M anufac having one im portant constituent which may behave like an
turing Com pany, E ast Pittsburgh, Pa.
Presented a t the Annual M eeting of T h e A m e r i c a n S o c i e t y o f amorphous mass.
M e c h a n i c a l E n g i n e e r s , held in New York, N. Y., Decem ber 2 to 6, Experiments on the effect of the speed of deformation4 of the
1935. polycrystalline ductile metals at comparatively low temperatures
Discussion of this paper should be addressed to the Secretary, seem to favor more nearly the logarithmical law of velocity of
A .S.M .E., 29 W est 39th Street, New Y ork, N. Y., and will be accepted
until M ay 10, 1936, for publication a t a later date. plastic yielding.
N o t e : Statem ents and opinions advanced in papers are to be
understood as individual expressions of their authors, and not those 4 See reference in p a rt I of this paper, T rans. A .S.M .E., vol. 55,
of the Society. 1933, APM-55-10.
A-7
A -8 JOURNAL OP A PPL IE D MECHANICS
(c) The Logarithmical Law of Velocity of Plastic Yielding. thought that tests on creep of copper would be of considerable
This may be assumed as interest also in connection with the industrial applications of
this metal. As copper does not age-harden (at least not appre
ciably in the qualities used ordinarily) it was thought that this
metal would be suitable for the investigation of creep under
somewhat simpler conditions than those which would probably
where a0 and u0 are two material constants dependent on the
be encountered in steels or other alloys.
temperature and on the amount the metal has been strain-
hardened by previous plastic deformation. For very small S im u l t a n e o u s S t r a in H a r d e n in g a n d C r e e p
velocities, [8] must be replaced by
A close examination indicates that copper and mild steel show
the effect of the speed of deformation even at room temperature,
and at the same time they strain-harden. In a working theory
when —Wo < u < u0.
of creep which should cover the range of temperatures beginning
(d) Power-Function Law. Some observers have suggested
with those under which strain hardening is pronounced, the yield
for the velocity law a power function, which may here be written
stress a should appear as a function dependent on the funda
in the form
mental variables, plastic strain e" and strain rate u ’ = d t’/dt.
As yet, it is uncertain whether a function
At, while true stress is a = P /A , where A is the area of the bar of the final points P% and P». Their orientation and shape in
when it carries the stress a. We have the a-e" plane therefore depend on the preceding operations.
Dividing Equation [14] by dt furnishes the rate of change of
<7 = (1 + e ' ) s .................................. [17] stress with time
Differentiating Equation [17] partially with respect to t '
At the maximum load s = s„, where sm is the ultimate strength, For this rate we have two additional expressions. From Equa
d s/de" = 0. If f" — is the corresponding strain, we have tion [16] we see that
S p e c ia l C a s e s
F io . 2 C o n d it io n o f S t r e s s -S t r a i n C u r v e s a t C o n b t a n t -S t r a in
R a t e s W h e n a S t r e s s S u r f a c e D o e s N o t E x is t
From this, the creep curve t ' = /(<) under a constant stress as
The condition for the existence of a stress surface a — /(«", u") usually observed in long-time tests can be derived. Assume
can now be stated. It is that Equation [14] must be a complete for example that
differential or that
the total strain is of the origin 0. The time t to reach a given permissible strain
ei under a given stress <r is shown by the dashed curve of Fig. 3
marked “permissible stress.” Writing E ti = a\, we see also
th at the permissible stress a is given by
The constant <t>/E in formula [6] has the dimension of a time
and may be designated as a “relaxation tim e” tr = <t>/E, so called
6 = | +ftft........................ {i\
Using as an example, the logarithmical-speed law according to
Equation [8] or
Z.----i
u " = / (< r ) = Mo e™ .................................. [ ; ]
a - —1
e = - + Unte™ ........................ [fc]
E
This family of straight creep lines is shown in Fig. 4.
F ig . 3 C r e e p a n d R e l a x a t io n C u r v e s f o r E l a s t ic o -V is c o u s
M a t e r ia l
The straight lines, for values of a > <r0, have the envelope
t — — tre Those for values of a < o-0, below the line A B,
pass through the point C.
Formula [e] furnishes the time t in which a given total and per (6) Constant Load s = Const. Assuming strains not neces
missible strain ti would be reached in a creep test made under a sarily small, from Equation [17] on account of ds = 0, we have
constant stress from Equation [196] da/dt = su” and Equation [14] becomes
For an elastico-viscous material the familiar creep curves e = f(t), (c) Constant Strain Rate u = de/dt = Const. The right side
obtained under constant stress, consist of a family of straight in Equation [20] has to be taken constant.
lines passing through a common point C in Fig. 3 situated at a (d) Relaxation of Stress Considering Strain Hardening. (The
distance OC = —tT equal to the relaxation time tr to the left Turbine-Bolt Problem.) C. R. Soderberg7 has emphasized the
JOURNAL OF A PPLIED MECHANICS A -ll
importance of considering strain hardening in various problems (e) I t may be noted th a t the cases a, b, and d previously
related to creep. One application to which he particularly called mentioned are all special cases of solving the same equation
attention is concerned with the turbine-bolt problem at high
temperatures, for which he developed the means. Suppose th at
the creep law of the bolt material is known at the temperature
to be considered. It is required to find the stress a at any time t
(service times of turbines to be considered) when the initial
stress <t\ after tightening the bolt is given. Creep occurs here
under the condition that the total strain e remains unchanged.
If £i" designates the value of the plastic strain c" at the time
t = 0 when relaxation of stress starts, we have the condition C o n s t a n t S t r a in - H a r d e n in g C o e f f ic ie n t a n d I d e n t it y
of the C r e e p a n d R e l a x a t io n C u r v e
from which it is seen that the viscosity coefficient for this law is
simply
Integrating with the initial condition t = 0 and u" = Ui", the
plastic-strain rate
If we take ^ = 0, that is, for a material that does not strain- is introduced as a variable relaxation time tr, depending on the
harden, we obtain the relation between stress a and time t stress <r, the creep curves assume the shape
A-12 JOURNAL OF A PPLIED MECHANICS
hardening clearly appears in their curvature when Fig. 6 is com It should be noted that elastic hysteresis and other circum
pared with Figs. 3 and 4. stances may have a marked effect on the decrease of stress due
(6) Relaxation for Constant Strain-Hardening Coefficient. Sup to relaxation. This is being investigated in tests on relaxation
pose again that \J/ = const, and <t> — a<,/u’, the corresponding under progress.
G e n t l e T e n s il e - I m pa c t T e s t s — C r e e p o p C o p p e r
at R oom T e m pe r a t u r e
F iq . 7 C r e e p C urve
F io . 8 R e l a x a t io n C u h v k
F io . 1 0 O b s e r v e d C re e p -T im e C u r v e s a t C o n s t a n t L o a d
D u r in g 1 0 -M in u te T e s t s o n S o f t C o p p e r a t N o r m a l T e m p e ra
tu re s
F ig . 11 O b s e rv e d C re e p -T im e C u r v e s a t C o n s t a n t L o a d F ig . 13 I n t e r p o l a t e d C re e p -T im e C u r v e s a t C o n s t a n t L o a d
D u rin g 2 4 -H o u r T e s ts o n S o f t C o p p e r a t N o rm a l T e m p e ra tu re s D u r in g 2 4 -H o u r T e s ts o n S o f t C o p p e r a t N o rm a l T e m p e ra tu re s .
A-14 JOURNAL OF A PPLIED MECHANICS
TABLE 2 R E S U L T S O F 1 0 -M IN U T E T E S T S
curves must be omitted here. However, these tests show similar
T e s t n o ......................................... 1 2 3 4 5 6 results, but at the higher stresses gave slightly lower values of
P la s tic s tr a i n eo", p e r c e n t . . 1 .5 2 2 .0 2 2 .6 6 3 .2 7 3 .9 3 4 .6 6
a o lb p e r s q i n .......................... 38 45 52 59 66 74 o-0 than were obtained in the 24-hour tests.
Two creep tests gave for the creep rates at times t the values
If the plastic-strain rate is increased ten times, according to
shown in Table 3.
Equation [8], this will produce an increase in the yield stress
which is equal to 2.303 <r0. The constant aa, therefore, is 0.434
times the increase in yield stress when the speed is increased
ten times.
Table 1 contains the observations with reference to the 24-hour
tests and shows in the last line the values for the constant <r0
computed from these creep tests.
Table 2 gives the corresponding values of <r0 derived from the
10-minute tests. Cold work (strain hardening) due to previous Sum mary
stretching of the test bars is indicated in both tables by the
Means are developed to include strain hardening when dis
values of eo" or <r.
cussing observations on creep of metals. A number of cases of
Both tables show that the constant <r0 increases with the
creep under varying conditions were considered. Tests seem to
amount of previous cold work. These values of uo are indicated
support the logarithmical-speed law for copper at room tempera
in Fig. 14. Those obtained in the much more rapid tests of
ture at the high speeds of deformation realized in these tests.
10-minute duration are higher than the values of <r0 derived from
the slower tests of 24-hour duration; an influence of the speed A cknow ledgm ents
on the values of cro therefore seems probable.
In Fig. 14 two further observation points are marked by two Acknowledgments are expressed to L. W. Chubb, director of
large circles. They were obtained from comparatively slow the Westinghouse Research Laboratories for having made these
tests at constant-strain rates. investigations possible and to Messrs. R. A. McCarty, B. A.
A number of additional creep tests were made on one copper Rose (East Pittsburgh, Pa.), N. L. Mochel, and R. C. Soderberg
bar keeping the loads constant in a hydraulic 30-ton Amsler (South Philadelphia, Pa.) for their interest in these and related
machine for several hours. On account of lack of space these questions.
T w o- a n d T h ree-D im ensional Cases of S tress
C o n cen tratio n , a n d C o m p ariso n W ith
F atig u e T ests
By R. E. PETERSO N 1 a n d A. M. WAHL,2 EAST PITTSBU RG H, PA.
This paper reports th e results o f a study o f som e two- elastic theory, i.e., mathematical analysis, photoelastic tests, or
and three-dim ensional cases o f stress d istribution w ith strain measurements.
particular reference to sh afts having fillets or transverse
holes, these being o f considerable practical im portance. 2—TWO- AND THREE-DIMENSIONAL CASES OF STRESS
To determ ine the stress-concentration factor kt in such DISTRIBUTION AND STRAIN MEASUREMENTS ON
cases, strain m easurem ents were m ade, using a specially LARGE STEEL SHAFTS
developed extensom eter w ith a gage length o f 0.1 in. The G eneral
results o f these strain m easurem ents indicate th a t for
shaft fillets in bending (three-dim ensional case) th e stress- For the purpose of establishing a correlation between theoreti
concentration factor kt is little different from th e values cal stress-concentration values and those obtained by fatigue
obtained photoelastically on flat specim ens having the tests, two cases are of special interest, particularly since consider
same r/d. ratio (a tw o-dim ensional case). A com parison able fatigue-test data involving such cases are available. These
of these values o f k.t (both for shafts w ith fillets and w ith are (1) the shaft with a fillet and (2) the shaft with a transverse
transverse holes), w ith data from fatigue tests, leads to th e hole. Although no analytical solutions have been obtained for
following observations: (1) In som e cases fatigue results either of these cases, it was felt th a t because of the similarity of
are quite close to theoretical stress-concentration values. (1) to the case of a flat bar having a notch or fillet (which has been
(2) Fatigue results for alloy steels and quenched carbon solved photoelastically) a study of mathematical and photoelastic
steels are usually closer to theoretical values th an are solutions of such cases would be of considerable interest. In ad
th e corresponding fatigue results for carbon steels not dition, it was decided to make actual strain measurements on
quenched. (3) With decrease in size o f specim en, the re large steel shafts having fillets or transverse holes, in order to
duction in fatigue strength due to a fillet or hole becom es obtain stress-concentration factors which would be directly com
som ew hat less; and for very sm all fillets or holes th e re parable with the results of fatigue tests. Since the stress peaks
duction in fatigue strength is com paratively sm all. (4) existing near such discontinuities (holes or fillets) are very
Sensitivity factors determ ined for sm all specim ens should sharp, a special short-gage-length extensometer was developed
not be applied to th e design o f m achine parts regardless of for this purpose.
size. C o m p a r is o n o f M a t h e m a t ic a l a n d P h o t o e l a s t ic S t r e s s -
1— INTRODUCTION CoNCENTRATlON FACTORS FOR N O TCHES AND F lL L E T S
H E results of studies of stress distribution in two-dimen Some interesting work on the stress distribution in notches of
r/d, r is the minimum notch radius and d is the diameter or width profile and yields values of k, slightly lower than for deep notches
of the minimum section. Curves 1 and 2 in Fig. 1 represent having the same r/d ratio.
values obtained for plane and turned notches in tension, while
curves 3 and 4 show similar values for notches in pure bending. D e s c r ip t io n o p S h o r t -G a g e - L e n g t h E x t e n s o m e t e r
F ig . 1 S t b e s s - C o n c e n t r a t io n F a c t o r s f o b N o t c h e s a n d F i l l e t s
clamped against a fillet (not shown on Fig. 4) may be clearly seen The constant of the instrument, in pounds per square inch per
in Fig. 3. millimeter reading of the microscope, was obtained by clamping
Load-strain curves obtained by means of this extensometer on it to the middle part A of the tapered portion of the shafts as
steel shafts were found practically to be straight lines within the shown in Fig. 5a and Fig. 56. The slope of this taper was so
elastic limit of the material. I t was also found th a t the instru determined as to give practically a constant stress near A . This
ment could be removed and reclamped to the test specimen and stress could be figured from the known load, shaft diameter, and
the readings easily repeated within 1V 2 per cent.
S t r a in M e a s u r e m e n t s o n L a r g e S t e e l S h a f t s W it h F il l e t s
vided by the section modulus Z = wd3/ 32 of the shaft gave the The method of clamping the extensometer to the edge of the
nominal computed stress a. The stress-concentration factor6 is hole is indicated in Fig. 8. To avoid sidewise slippage of the
then points, the clamping load was applied obliquely as indicated;
thus the knife-edges press directly on the edge of the hole at the
location of the peak stress.
In this case the stress-concentration factor was taken as
The experimental points in Fig. 7 indicate values of k, which were
determined in this manner for the three fillet tests on the fillets
shown in Fig. 5a with values of r/d = 0.5, 0.281, and 0.167,
respectively. where ffma* is the peak stress as obtained by strain measurements
Curves obtained photoelastically by Frocht4 on flat specimens and a the nominal bending stress figured from the known bending
(case b in Fig. 7), are shown for comparison the dotted curve repre moment M divided by the section modulus Z, calculated on the
senting values for ratios of D /d = 1.5 and the full lines similar basis of the net section through the hole.
ones for D/d = 2. I t will be noted th a t the points obtained by
strain measurements on the large steel shafts having ratios of D/d
= 1.5 (three-dimensional case) agree well with the photoelastic
curve7 obtained on flat specimens (two-dimensional case), both
F ig . 8 M e t h o d o f C l a m p in g E x t e n s o m e t e r a t t h e E d g e o f a
H o l e in a S h a ft
Fio. 15 F a t i g u e - T e s t R e s u l t s f o b S e v e r a l S e r i e s o f V a r i a b l e
S iz e b u t G e o m e t r i c a l l y S i m i l a r S p e c im e n s H a v i n g a T r a n s
v e r s e H o le (U p p e r C h a r t) o r a F i l l e t (L o w e r C h a r t)
tioned program are given in Figs. 11, 12, 13, and 14. Supplemen
tary data on tests of specimens with transverse holes have been
previously reported9 for the 0.45 per cent carbon steel and for
0.57 per cent carbon steel (quenched and drawn). Fio. 16 D a t a o n S e n s itiv ity o f S e v e r a l S t e e l s t o S tr e s s -C o n -
In comparing fatigue results with theoretical stress-concentra c e n t r a t i o n E f f e c t i n F a t i g u e S p e c im e n s H a v i n g a T r a n s v e r s e
H o le (U p p e r C h a r t) o r a F i l l e t (L o w e r C h a r t)
tion factors, a kf term is used, defined as
scale from zero (no stress-concentration effect) to unity (full
Je. =
endurance limit for specimens without stress concentration
-------------------------------- ----------------------------- -----------------------------------------------------------------------
endurance limit for specimens with fillet or hole10 theoretical stress-concentration effect). Values of q as derived
from the data of Fig. 15 are given in Fig. 16.
No adequate short name exists for this term, which may be con I t is also of interest to plot the kf values for holes in terms of
sidered as an “effective stress-concentration factor in fatigue.” a/d, the ratio of the hole diameter to specimen diameter, as shown
The endurance limit for specimens without stress concentration in Fig. 17, and for fillets in terms of r/d, the ratio of the fillet
has been shown to either not vary greatly with size or to decrease radius to the small diameter of shaft, as shown in Fig. 18. In
somewhat with increase in diameter. In obtaining k, values from the latter case a considerable number of values from the litera-
the data of the present tests the endurance limit obtained with
0.469-in. diameter tapered cantilever specimens was used for the 11 I t has been the au th o rs’ experience th a t the 0.469-in. tapered
cantilever specimen usually gives a somewhat lower endurance lim it
numerator of the expression for kt when data from larger speci- th an the conventional 0.3-in. diam eter specimen having a 97/«-in.
contour radius, and a value closer to th a t of large specimens of 1 in.
♦ ‘‘Model T esting as Applied to Strength of M aterials,” by R . E. and 2 in. diam eter. I t is thought, therefore, th a t the kj values for
Peterson, T rans. A.S.M .E., vol. 55,1933, pap er APM-55-11, p. 79. large specimens are n o t greatly in error, although they m ay be slightly
10 Using n et section modulus. high.
JOURNAL OF A PPLIED MECHANICS A-21
turę12'13'14'15'1®'17 are included. From a study of Figs. 15, 16, 17, McAdam’s correlation data show striking agreement. If this
and 18, a number of observations can be made: correlation has any real significance it should not be altered essen
(a) In some cases fatigue results are quite close to theoretical stress- tially for various sizes of geometrically similar specimens. The
concentration values. This is an im portant point, as a general idea authors expected th a t theoretical values would be reached for all
seems to exist, based on some of the early photoelastic results steels provided specimens were made large enough, b ut Figs. 15
which were considerably in error, that fatigue data for stress-con and 16 show th a t the fatigue-data curves for normalized 0.45 per
centration cases are always well below theoretical values, i.e., on cent carbon steel are apparently asymptotic to values considerably
the safe side for design purposes. As a consequence of the present below the theoretical. In this respect, the authors’ results do not
data, one should be careful, when applying recent theoretical seem to be essentially inconsistent with McAdam’s conclusions
data to design, not to do so with the same feeling of always having
an additional indefinite margin of safety as was formerly the case.
(b) Fatigue results for alloy steels and quenched carbon steels are
usually closer to theoretical values than are the corresponding fatigue
results for carbon steels not quenched. This is evident from Figs.
15, 16, 17, and 18. In order to visualize this tendency in Figs.
17 and 18, the carbon steels (not quenched) are indicated by
small crosses the alloy steels and the quenched carbon steels by
F ig . 18 C o m p a r is o n o f R e s u l t s f o r F il l e t S p e c im e n s
work due to machining did not extend beyond a depth of 0.003 strikingly similar. For this previously mentioned guidance over
in. and probably not greater than 0.0015 in. While these few a variety of cases, the use of the lower chart is suggested. In each
results are by no means conclusive, they do indicate th a t the chart, the lower curve represents average results for a medium
answer may not be a simple one. carbon steel (not quenched) and the upper curve represents aver
Some thought is being given to the number of grains in the age results for alloy steel and quenched carbon steels. A few
volume of material subjected to the stress peak in the case of results at hand indicate th a t a low-carbon steel (not quenched),
small fillets or holes and the probability of unfavorable orienta such as ordinary structural steel for example, would have an
tions of individual grains. average curve below th at of the medium-carbon steel. The
But regardless of the reason for the relative increase in fatigue sensitivity factor q may be applied in accordance with the follow
strength of specimens with small fillets or holes, it is nevertheless ing formula, derived directly from the definitions of the terms
an effect which exists in small machine parts as produced con
ventionally and one which, by means of the data of Figs. 15 and =
16, can be estimated for design purposes. *• q(k, - 1) + 1
(id) Sensitivity factors determined for small specimens should not
where oy = endurance strength with fillet, hole, groove, or other
be applied to the design of machine parts regardless of size. Certain
discontinuity; a, = ordinary endurance limit without stress con
design codes suggest the application of a sensitivity factor to the
centration (use value from largest diameter specimens available);
theoretical stress-concentration factor and give values of sensi
kt = theoretical stress-concentration factor (mathematical, pho
tivity factors obtained from fatigue tests of relatively small
toelastic or strain m ethod); q = sensitivity factor shown in the
specimens. Although not so stated, it can easily be assumed by a
lower chart of Fig. 16.
designer th a t these values are fixed material constants. The
error which could arise from such a concept may be very large as ACKNOWLEDGMENT
can be seen from Fig. 15, and furthermore is in the unsafe direc
tion when applied to the design of large parts. I t appears th at The authors wish to express appreciation to Dr. S. Timoshenko
until a large amount of data become available, average sensitivity of the University of Michigan for advice and to Dr. L. W. Chubb,
factors such as shown in Fig. 16 can give some guidance for de Director of Westinghouse Research Laboratories for permission
signers in cases not considered here, such as shafts with grooves, to publish these results. They also wish to acknowledge the as
threads, and keyways. I t will be noticed that the two charts of sistance of F. Cassel in connection with both the strain measure
Fig. 16, representing widely different cases (holes and fillets) are ments and the fatigue tests.
T h e R a te of G ro w th of F atig u e C rack s
B y A. V. DEFOREST,1 CAMBRIDGE, MASS.
The author discusses th e resistance o f m etal to th e Therefore, it appears th a t for engineering use a further fatigue
propagation of cracks, and points out th at in th e accepted figure of merit must be taken into consideration, namely, the re
m ethod o f fatigue testin g th e results are n ot conclusive sistance of the metal to the propagation of a crack due to a stress-
because there is no d istinction between (1) th e load and the raising condition. This property has come to be called notch
num ber o f repeated stresses required to start cracks and sensitivity. Quite evidently the moment a true fatigue crack
(2) th e load and th e num ber of reversals needed to propa has started at the bottom of a notch, the crack itself constitutes
gate th e cracks to failure. The investigation reported in the greatest stress-raiser and the sharpest of all possible notches.
the paper was undertaken to determ ine (a) th e stress and We then have three figures bearing on the fatigue strength of
num ber o f reversals required to start a fatigue crack, and metals. First, the strength of the unnotched material, a figure
(b ) the resistance o f the m aterial to th e effect o f a fatigue now well known. Second, the strength under a finite mechani
crack once it had been form ed. The tests reported were cal notch. This figure is both a measure of the sharpness of the
conducted on annealed cold-rolled low-carbon steel bars notch and of the true notch sensitivity of the metal. Third, the
with various finishes. The author presents data on th e strength of the steel by virtue of which it resists the progress of
rate o f progress o f th e cracks and on th e size o f the cracks, the natural fatigue crack, once this latter has started. I t is
the latter o f w hich was determ ined by m agnetizing the natural to suppose but so far unproved th a t the sharpness of
bars and inspecting them w ith finely divided iron oxide a crack once formed, is itself a function of the material, and
in a bath o f kerosene. therefore no distinction need be made between the notch sensi
tivity at this “natural” notch and the sharpness of the notch.
H E normal process of fatigue testing is too well known to I t is altogether probable th at the sharpness of the notch is in
in the “overnight” test approaches the same principle. Many sequently high values of stress were used. The specimens were
other thoughtful workers have the same distinction in mind. in the form of 5/s-in. diameter bars, 4 ft long, with a loaded
The method here described is simply a somewhat more elabo- length of 12 in. At intervals during the test the bars were mag-
rate manner of studying the problem. Use has been made here netized and inspected with finely divided iron oxide in a bath of
of the magnetic method for locating minute cracks by mag- kerosene.
netizing steel specimens and applying finely divided iron parti- At the higher loading figures a certain amount of heat was de-
cles, these particles usually suspended in a liquid medium. This veloped but this was not considered sufficiently serious to require
control. The figures in Table 1 were derived from tests on an-
F ig . 3 C eacks in H ig h -C a b b o n St e e l W it h M a g n a f l u x on
C b a c k s 15 X
F ig . 2 P e o p a g a t i o n o f F a t i g u e C b a c k s F b o m 0 .0 5 I n . t o F e a c -
tu b e f o e A l l S u b f a c e C o n d itio n s
(A n n e a le d c o ld -ro lle d b a rs of S .A .E . 1020 Bteel, 5/a in . i n d ia m e te r .)
method for locating the incipient fatigue crack has been markedly
successful even for cracks of the order of a few thousandths of an
inch in length. Test runs were made on 6/ s-in. diameter rotating-
beam specimens and the rate of progress of the crack, but not
the stress at which the crack would not propagate, was deter
mined. The following results are from a thesis by F. W. Mag-
nuson:6 The questions at issue were to determine whether, as
seemed probable, the rate of growth of a fatigue crack was con
stant in a given m aterial and independent of the number of cycles
required to start a crack, and whether a different metallurgical
condition of the same material would markedly change the rate F ig . 4 C backs in L o w -C a b b o n S t e e l W it h M a gnaflux on
C b a c k s 15 X
of progress of a previously formed crack. To answer these ques
tions a set of low-carbon cold-drawn specimens was tested. No nealed cold-rolled low-carbon steel bars with three different types
attem pt was made to determine absolute fatigue limits and con- of surface finish. One, a finish with coarse circumferential
‘ Thesis of F. W. M agnuson, M assachusetts In stitu te of Tech scratches from a bastard file; another, with fine circumferen
nology, 1935. tial scratches produced by polishing with number 0 emery paper;
JOURNAL OF A PPLIED MECHANICS A-25
the large deflections of circular plates. S. Timoshenko and S. 15 B. R. Seth, Philosophical Transactions, Royal Society of Lon
Way discussed the suspension bridge with stiffening truss. E. O. don, 1935.
16 R. C. J. Howland, Proceedings, Royal Society of London, series
Waters gave a three-moment equation for flat plates, M. M. A, vol. 148, 1935, p. 471.
Frocht the results of a photoelastic study of the centrally loaded 17 R. C. Knight, Philosophical Magazine, series 7, vol. 19, 1935,
beam, H. M. Westergaard an account of stress functions for p . 517.
shells, W. L. Schwalbe a solution of a torsion problem by iteration, 18 D. B. Smith and R. V. Southwell, Proceedings, Royal Society
of London, series A, vol. 143, 1934, p. 271.
and J. N. Goodier an investigation of the distribution of load on 19 B. R. Sen, Zeitschrift fiir angewandte Mathematik und Mechanik,
the threads of screws, based on extensometer measurements. A vol. 13, 1933, p. 374.
photoelastic record of impact stresses by high-speed photography 20 B. R. Sen, Philosophical Magazine, June, 1935.
was presented by Z. Tuzi and M. Nisida, of Japan. 21 I. Malkin, Schweizerische Bauzeitung, vol. 103, 1934, p. 15.
22 J . P. Den Hartog and J . G. McGivern, J o u r n a l o f A p p l i e d
B ib l io g r a p h y M e c h a n ic s , Trans. A.S.M.E., vol. 2, no. 2, June, 1935.
23 J. G. McGivern and H. L. Supper, Trans. A.S.M.E., vol. 56,
1 M . A . B io t, J o u r n a l o r A p p l i e d M e c h a n i c s , T r a n s . A .S .M .E ., 1934, paper APM-56-9, p. 601.
v o l. 2, n o . 2, J u n e , 1935. 24 C. W. MacGregor, Physics, vol. 5, 1934, p. 140.
2 M. A. Biot, Annales, Sociśte scientifique de Bruxelles, vol. 53, 25 S. C. Hollister, Journal American Concrete Institute, vol. 5,
1933, p. 189. 1934, p . 361.
3 W. Hovgaard, Proceedings, National Academy of Science, vol. 26 M . M . Frocht, J o u r n a l o f A p p l i e d M e c h a n ic s , Trans.
20, 1934, p. 17. A.S.M.E., vol 2, no. 2, June, 1935.
4 A. Wojtaszak, Philosophical Magazine, series 7, vol. 18, 1934, p. 27 J. H. A. Brahtz, Review of Scientific Instruments, vol. 5, 1934,
1099. p. 80.
5 D. F. Gunder, Physics, vol. 6, 1935, p. 38. 28 A. Goetz, ibid., p. 84.
6 J. P. Den Hartog, paper presented at the National Meeting of 29 R. D. Mindlin, ibid., p. 224.
the Applied Mechanics Division of the A.S.M.E., Ann Arbor, Mich., 30 R. W. Vose, J o u r n a l o f A p p l i e d M e c h a n ic s , Trans. A.S.M.E.,
June 18-19, 1935. vol. 2, no. 3, September, 1935, p. 99.
7 J . L. Maulbetsch, J o u r n a l o f A p p l i e d M e c h a n i c s , Transac 31 H. Neuber, Trans. A.S.M.E., vol. 56, 1934, paper APM-56-17,
tions A.S.M.E., vol. 2, no. 4, December, 1935. p. 733.
8 C. W. MacGregor, Mechanical Engineering, vol. 57, April, 1935, 32 S. Timoshenko, J o u r n a l o f A p p l i e d M e c h a n ic s , Trans.
p . 225. A .S .M .E ., vol. 2, no. 1, March, 1935.
9 J. E. Younger, paper presented at the National Meeting of the 33 L. H. Donnell, Trans. A.S.M.E., vol. 56, 1934, paper AER-56-
Applied Mechanics Division of the A.S.M E., Ann Arbor, Mich., 12, p. 795.
June 18-19, 1935. 34 F. J. Bridget, C. C. Jerome, and A. B. Vosseller, Trans. A.S.
10 H. Neuber, Ingenieur-Archiv, 1934, p. 238; 1935, pp. 133 and 325. M.E., vol. 56, 1934, paper APM-56-6, p. 569.
11 R. E . Peterson and A . M . Wahl, J o u r n a l o f A p p l i e d M e 35 D. F. Windenberg and C. Trilling, Trans. A.S.M.E., vol. 56,
c h a n ic s , Trans. A .S .M .E ., vol. 3, no. 1, 1936. (This issue.) 1934, paper APM-56-20, p. 819.
12 B. R. Seth, Proceedings, Cambridge Philosophical Society, 36 A. J. Miles, paper presented at the National Meeting of the
vol. 30, 1934, pp. 139 and 392. Applied Mechanics Division of the A.S.M.E., Ann Arbor, Mich.,
13 H. Okubo, Proceedings, Physieo-Mathematical Society of June 18-19, 1935.
Japan, vol. 16, no. 3,1934, p. 430. 37 A. Weinstein, Comptes Rendus, January, 1935.
14 B. R. Seth, Proceedings, London Mathematical Society, vol. 38 H. R. Fisher, Proceedings, Royal Society of London, series A,
37, 1934, p. 502. vol. 144, 1934, p. 609.
D esign D a ta
It is im portant th a t th e data contained in technical lished by th e Applied M echanics Division o f The American
papers be m ade readily available to designing engineers. Society o f M echanical Engineers.
In order to satisfy these needs o f industry, th is section o f The data are prepared by a su bcom m ittee o f th e Applied
th e Journal will include a concise presentation o f data and M echanics D ivision on strength o f m aterials (R. E. Peter
inform ation drawn chiefly from papers previously pub son, chairm an), under th e guidance o f S. Tim oshenko.
I n tro d u ctio n Fig. 2. For example, case 1 of this figure represents a plate
H E formulas given in this paper apply to the calculation loaded by a load P uniformly distributed along the inner edge,
F ig . 3 C u r v e s f o b O b ta in in g t h e K F a o to b s F ig . 4 C u rv e s fo b O b ta in in g t h e C F a c to rs
A-30 JOURNAL OF A PPLIED MECHANICS
(a)
F ig . 5 C i r c u l a r P l a t e s W it h o u t H o l e s
P art 3 L a r g e D e f l e c t io n s o f C ir c u l a r P l a t e s W it h o u t
H o l e s — U n if o r m L o ad
Fig. 1 Fig. 2
Sommerfeld found th a t the journal assuming perfect lubrication curves for a /0 — 0.5 and for a /0 = 0.583. On these curves,
and neglecting side leakage, would run at a very low speed in a shown in Fig. 7 of this discussion, the writer has marked the run
position almost horizontally opposite the center of the bearing, ning positions for 600 rpm. I t is interesting to note th at Bos-
as indicated by Fig. 1 of this discussion. As the speed increased, wall’s test curves follow his theoretical curves rather closely but
the running position would move downward until a low point differ slightly in th a t they lie more to the right as the speed is
would be reached. Then with a still further increase in speed, increased. I t is necessary to call attention at this point to the
the running position would rise and move to the left, reaching fact pointed out by Kingsbury11 and others, th a t the impact of
concentricity with the center of the bearing at infinite speed. the oil against the leading edge of a bearing builds up a pressure
In Fig. 1 of this discussion, the writer has marked for 200, 600, th a t makes the bearing operate as if it had a longer leading angle
and 900 rpm the probable running positions of a journal with the than th a t with which it is actually constructed. Boswall states
dimensions and clearance of Stone’s test bearing, with side leakage th a t the faces of his test bearing at the leading and trailing edges
taken account of in accordance with Kingsbury’s plate X 4. The were radial. By comparing Fig. 7 with Fig. 5 of this discussion,
viscosity of the oil is assumed to be the same as in Stone’s tests noticing the locus for 600 rpm, it will be seen th a t the impact may
for all these comparisons where the running position is marked readily explain deviations of the test curves from the theoretical
for a given speed. curves. Having noted this fact, Stone’s test results can now be
In his study of the 120-deg central bearing,7 Needs investigated examined.
the m atter so carefully as to be able to show that the position of As previously stated, Stone’s bearing had a cap symmetrically
the journal for zero speed would theoretically be at the leading located with respect to the bearing surface. In addition to this
edge of the bearing just as had been determined by Sommerfeld the leading edges of his bearing were not radial but were some
for the 180-deg bearing. A study of the writer’s charts8 en what beveled, as indicated by the writer’s Fig. 6. The effect
ables one to plot a curve for the 90-deg central bearing, as shown of the bevel a t the leading edge would be to exaggerate the in
in Fig. 3 of this discussion. The dotted extension of the lower fluence of impact. The effect of the bearing cap would be to in
end of the curve toward the right is assumed, but agrees with troduce negative pressures which would be most pronounced
the theoretical determination of Sommerfeld and Needs. On this at the leading edge of the cap. There would also be an impact
curve, as well as in Fig. 2 of this discussion, the positions for 200, pressure at the leading edge of the cap which would introduce
600, and 900 rpm are marked. still another complication. I t is possible th a t impact a t the
The influence of the leading-angle ratio a /0 upon the path of leading edge of the cap and the negative pressure, which other
the journal is indicated in the writer’s Fig. 4, which is adapted wise would act near th a t point, would neutralize each other.
from one of the writer’s papers.6 The running positions for 600 The impact at the leading edge of the working face of the bearing
and 900 rpm, are marked. I t should be noted th a t when a /0 = would bring about a more rapid rise of pressure within the film
0.6, the running position for 900 rpm is nearly horizontally op than would be the case with the radial face of Boswall’s bearing.
posite the center of the bearing, but th a t when a /0 = 0.5, the Hence, it is expected th a t the curve of Stone’s running positions
position for 900 rpm is much lower. The curves from which will be like th a t of a bearing whose a / 0 ratio is greater than 0.5.
these data were adapted are shown in Fig. 8 of this discussion On this assumption, the direction of his curve in the region of 400
in which running-position paths are shown for a 120-deg bearing to 700 rpm is readily explained; compare it with the curves in Fig.
without a cap, with the angle ratio a /0 varying from 0.48 to 5 of this discussion.
0.62. In the m atter of the running position with respect to the hori
In Fig. 9 of this discussion is shown a similar set of curves zontal line through the bearing center, Stone’s points for 200 and
drawn from the data published by Boswall9 for a 90-deg bearing 600 rpm are not far from the theoretical values indicated by the
without a cap, with a /0 ratios also varying from 0.48 to 0.62. writer’s Fig. 5, but at 900 rpm the height is much greater than
The similarity between the curves for the 90-deg and the 120- would be expected from a theoretical consideration. I t is pos
deg bearing is evident from a comparison of Figs. 9 and 8 of this sible th a t the influence of impact at the leading edges and the
discussion. vacuum of the cap would explain the high value.
From Fig. 9 of this discussion, the curves for a /0 — 0.5 and The queer turn the curve makes at 200 rpm indicates th at at
a /0 = 0.6 were chosen and reproduced in the writer’s Fig. 5 the low speeds there may be competition for first importance be
and upon these the running positions have been entered for 600 tween slight metallic contact and perfect film lubrication. Some
and 900 rpm, as determined from Boswall’s data.*'10 D ata ob metallic contact may exist at 100 rpm and may disappear at 200
tained from the writer’s previous work are marked as X in Fig. 5. rpm. If it were not for metallic contact the point for 100 rpm
The first of these, for a /0 = 0.5, lies upon Boswall’s curve for might lie to the right of the point for 200 rpm. If this were the
th a t ratio and agrees almost exactly with the point determined case, the curve would conform more nearly with the theoretical
from Boswall’s data. The value for 900 rpm and a /0 = 0.575 one shown in the writer’s Fig. 3, except for the influence of im
lies about where it would be expected. The same is true of the pact at the leading edge which is not considered in Fig. 3. The
point for 600 rpm and a /0 — 0.590 and th a t for 200 rpm and behavior of Stone’s bearing in the region of 800 to 900 rpm is
a /0 — 0.67. The theoretical work of Boswall and the writer not understood.
may therefore be said to be in practical agreement. Referring to the loops of running positions reported by Mr.
The writer next refers to Boswall’s tests for which he plotted Stone, it is suggested th a t possibly they were more apparent
than real, and th a t the magnetic properties of the device used
’ “Effects of Side Leakage in 120-Deg Centrally Supported Journal for determining the running position might vary enough to mis
Bearings,” by S. J. Needs, Trans. A.S.M.E., vol. 56, 1934, paper lead the investigator into believing th a t instability was present,
APM-56-16, p. 721. where in reality there was perfect stability. This is suggested in
8 “Graphical Study of Journal Lubrication,” by H. A. S. Howarth,
Trans. A.S.M.E., vol. 45, 1923, p. 421; vol. 46, 1924, p. 881; vol. 47, order th a t the point may be further investigated.
1925, p. 1120. There is one other m atter th a t should receive consideration
• “The Theory of Film Lubrication,” by R. O. Boswall, Longmans and th a t is the theoretical locus of running positions published by
Green and Company, London and New York, 1928, 280 pp.
10 “The Film Lubrication of the Journal Bearing,” by R. O. Bos 11 “On Problems in the Theory of Fluid-Film Lubrication, With an
wall and T. C. Brierly, Proceedings of the Institution of Mechanical Experimental Method of Solution,” by A. Kingsbury, Trans. A.S.
Engineers, London, Eng., vol. 122, April, 1932, pp. 423—570. M.E., vol. 53, 1931, paper APM-53-5, p. 59.
A-34 JOURNAL OF A PPLIED MECHANICS
Karelitz,3 based on a system of investigation proposed by Giim- plotting the running positions, whereas the accurate plot for the
bel. Those running positions do not agree nearly so well with running positions is as shown in Fig. 9. This will make Fig. 6
tests by Boswall and by Stone as does the curve drawn from investi of the discussion not exactly as Mr. Howarth shows it, but in
gations by Sommerfeld, Boswall, Needs, and the writer. The general his remarks are still applicable to the point in question.
curves shown by the writer’s Figs. 1 to 5, inclusive, and Figs. 8 Regarding the loops of the running positions, it can be definitely
to 10, inclusive, were based upon the extension of the original stated th at these loops are not due to any particular properties of
theory of journal bearings as proposed by Osborne Reynolds.12 the magnetic devices used in determining the positions, because
These studies, taken in conjunction with test results, indicate such instruments have been used extensively by the author and
th a t Reynolds’ theory is very closely in accordance with the others, and there can be no question as to their reliability in such
facts, and that approximations such as proposed by Giimbel and work. Similar loops have been found by others, notably by
followed by Karelitz and Stone, are not so reliable as references Hummell in Switzerland. A discussion of the causes of these
for interpretation of test results. loops, following the lines suggested in the author’s original paper
should well merit some investigator’s time.
M a y o D. H e r s e y . 13 Would it be possible to coordinate the
In replying to Professor Hersey’s comments, the values of the
author’s results with those of previous experimenters in so far viscosity used in the Z N / P curves were based upon measured oil
as similar bearings are concerned?14 In comparing the film- temperatures corrected for thermal drops much as Professor
thickness results given by the author with values calculated by Hersey has suggested. The ratio L / D for these bearings was
Karelitz, apparent agreement is shown in the author’s Fig. 9 somewhat under 1.5. As Professor Hersey states, the units used
with the Z N / P curve based on Karelitz’ work,3 although neither in the paper refer to the Z N / P diagrams only and not to the
the film temperature nor the L / D ratio is given by the author. general equations. The values of the coefficient of friction which
If the value of Z is taken from the measured temperature of the Professor Hersey refers to in Fig. 36 of the paper were not in
bearing metal, has allowance been made for temperature varia tended to be accurate but only to indicate general tendencies, as
tions over the cross section of the oil film?16 Differences of the the figure is labeled “characteristic curves.” The actual values
order of 10 F might be anticipated at 900 rpm on a bearing are somewhat lower, as mentioned by Professor Hersey.
9-in. in diameter.
Assuming L / D not far from unity, where L is the length or width A M e th o d of B alan cin g R e c ip ro c a t
parallel with the axis, there is excellent agreement between the
values of film thickness as given by the author in his Fig. 3a and ing M a ch in e s1
those given by S. J. Needs7 for a 120-deg bearing. However, the
values for the coefficient of friction as given by the author in his M. W. D a v i d s o n . 2 The author states in the second paragraph
Fig. 36 are nearly twice those given by Needs7 for values of Z N / P of his paper: “ Moreover, if the machine is self-contained, it is
between 30 and 50. theoretically possible to obtain a perfect balance both of the dy
Referring to page 64 of the paper by Mr. Stone, the units namic forces and couples, and of couples derived from gas forces.
given in the list of symbols are not applicable to Equations [1],
[2], and [3] which are valid in any consistent system of units but
not when centipoises, pounds per square inch, and feet per minute
are combined. The units given are evidently intended for use
with the Z N / P diagrams, rather than with the equations.
Mr. Stone’s Equation [14] is not restricted to the specific as
sumptions made, but is equally valid for fitted bearings and those
undergoing side leakage, as may be shown by dimensional reason
ing.18 The author’s condensed outline of the classical theory is
original, and should be of great help to all engineers and ad
vanced students who are finding difficulty with the mathematical
literature of the hydrodynamic theory.
A u t h o r ’s C l o s u r e
composite machine would be perfectly balanced, provided it does balance is perfectly valid, regardless of Prof. M. W. Davidson’s
not have to transm it torque to an external machine.” objections to it. The only modifying restriction th at should be
The writer believes this statement to be too broad and some made is th at the various links must be perfectly rigid, which, of
what in error, which he attem pts to show in Fig. 1 of this discussion. course, is not obtained in practice.
The diagrams in this figure represent identical single-cylinder In answer to Professor Davidson’s specific argument, the force
engines fastened head to head. The upper figure shows the en vectors shown in his upper diagram, where the engines are rotating
gines running in opposite directions, and the lower figure shows in opposite directions, do not fully represent the tru th of the situa
the engines running in the same direction, and both geared as the tion. I t is well known th a t a connecting rod may be replaced by
author suggests. The inertia effect upon the engine frame of the its dynamical equivalent, consisting of two masses, one concen
reciprocating mass is represented by P; while H represents the trated at each end, and a massless moment of inertia. This
horizontal component of the inertia (linear) of the connecting system can be made to satisfy all the dynamical specifications of
rod; V represents the vertical component of the linear inertia of the connecting rod, namely:
the connecting rod; and Q represents the turning effect upon the
frame of the angular inertia of the connecting rod. (1) The center of gravity will be in the same location
The upper diagram shows th a t all couples and forces are bal (2) The total mass will be the same
anced except the vertical components of the linear inertia of the (3) The moment of inertia about the center of gravity will be
connecting rods. The lower diagrams show the linear-inertia the same.
effects of the engines to be in balance, while the couples resulting Having substituted such an equivalent system, one weight may
from the angular inertia “and linear inertia forces” of the connect be considered attached to the piston. Its inertia force is then
ing rods are additive. Neither of the arrangements is in theo balanced perfectly by th at of the corresponding weight in the
retical balance; only the addition of cylinders can make them so. other engine. The other weight is concentrated at the crank pin,
and consequently may be perfectly balanced by the counter
J. P. D e n H a r t o g . 3 In Equation [28] the authors come to weight in each engine. Finally, the inertia couples caused by the
the conclusion that for the particular single-crank compressor massless moment of inertia will be equal in magnitude and oppo
under consideration the necessary balance weight is “consider site in sign in the two engines, so th a t these are balanced. The
ably more than ‘all of the rotating plus half of the reciprocating result is a perfectly balanced machine. This result is granted to
mass.’ ” At first such a result seems strange, but a closer in be contingent upon the proper use of counterweights.
spection of the system shows th a t the balance weight is placed Now, it is quite possible th a t with certain machines un
considerably lower on the vertical shaft than the unbalancing balanced force vectors at right angles to the center line through
piston and crank. shafts will remain. In such a case, a combination of four ma
If the counterweight were made equal to the rotating plus half chines will secure perfect balance. I t is obvious that any vector
the reciprocating weight, the horizontal force would be balanced system of forces and couples can be balanced by combining the
and the center of gravity of the machine would not move. But proper number of identical systems, properly oriented in space.
the moments of the forces about the center of gravity would not The elementary method of balancing the compressor suggested
be balanced and hence the machine would vibrate rotationally by Prof. J. P. Den Hartog is quite all right for certain simple cases.
about its center of gravity, which results in a compression of the For more complicated systems, however, it usually cannot be
supporting springs and a transmission of the vibration to the applied without sacrificing precision in the results.
foundation. It is logical in this particular case to design for Incidentally, the author and his associates consider it more
balanced moments about the center of gravity and leave the hori correct to balance moments about the mean center of percussion
zontal forces unbalanced. This will result in a horizontal vibra with respect to the piston and counterweight forces. I t is in this
tion of the center of gravity (and in fact of the whole machine) region, with a very flexible mounting, th a t the least motion exists.
but no rotation about the center of gravity will follow. The Professor D enH artog’s statement th at when “the counter
horizontal vibration can be taken by the supporting springs with weight is made equal to all the rotating plus half the reciprocating
out transmission to the foundation. The author comes to this mass. . .the center of gravity of the machine would not move” is
result correctly in Equation [28] because if each term of his ex obviously in error, for unbalanced forces remain which will oscil
pression is multiplied by the proper ratio of moment terms, we late the center of gravity. There is, however, a region about the
find center of percussion with respect to the piston inertia force which
has, not zero, but minimum motion.
</> = 0.72 Fi - + 1.48 F 2 - + 1.378 F3 -
€\ e\ e\ Further, the statement claiming th at horizontal vibration can
= 0.495 Fi + 1.018 F2 + 1.015 Fs, be taken by the springs without transmission has no justification.
The degree of transmission depends on the location and stiffness
in which Fi is the reciprocating weight and F2 + F3is the rotating of the springs, and in some cases the horizontal transmission is
weight, so that the old faithful rule is seen to work well. This much more important than the couples produced by vertical dis
method of balancing gives “no transmission to the foundation” placement of the springs. This is particularly true in refrigerator
and is the most logical one. If a minimum of kinetic energy of cabinets.
the swaying machine is desired, a compromise has to be made The author regards the close agreement of Professor Den Har-
between balanced moments and balanced forces. tog’s results as chiefly accidental, and depending upon the char
acteristics of the particular system. By shifting the spring
A u t h o r ’s C l o s u r e
mounting and changing the mass distribution about the center of
The statement concerning the theoretical possibility of perfect gravity, widely different results can be obtained by the author’s
3 A ssistant Professor of Applied M echanics, H arv ard G raduate method, whereas the approximate method proposed will give the
School of Engineering, Cambridge, M ass. Mem. A.S.M .E. same result in each case.
A-36 JOURNAL OF A PPLIED MECHANICS
kinds as the two independent solutions of Bessel’s equation. the exercise problems which were omitted entirely from the earlier
He begins with the functions of order zero, gives graphical repre and larger volume. This work follows exactly in form, spirit,
sentation of these functions and applies them in the study of and arrangement of subject m atter the lively work of Professor
symmetrical vibrations of a stretched circular membrane. In the Lemon; Chichi Lasley again supplying the numerous, instructive,
second chapter McLachlan discusses Bessel functions of higher and often humorous, line drawings.
orders and shows how these functions can be used in considering About half the printed text consists of worked examples and
nonsymmetrical vibrations of circular membranes in which a problems. As far as possible the problem m atter is related to
membrane has not only a system of nodal circles but also a situations familiar to the student. The answers to all the prob
system of nodal radii. Cylinder functions and surface spherical lems are printed on the last five pages of the book. The other
harmonics are also briefly discussed in this chapter. half of the text contains an abbreviated reminder of the general
Chapter three discusses expansions in terms of Bessel functions, basic principles of classical physics. However, the essentially
by using which the representation of Bessel functions of the first classical picture of the Rutherford-Bohr atom-structure is given
kind in the form of a definite integral is derived. The evalua and where electromagnetic radiation is described quanta and wave
tions of various integrals associated with Bessel functions are mechanics are touched upon lightly.
also given in this chapter. In chapter four a brief discussion of “From Galileo to Cosmic Rays” and its subordinate companion
hypergeometric functions, gamma functions, and Struve func “Exploring in Physics” are both attem pts a t humanizing the
tions is given. The asymptotic expansions of Bessel functions study of physics which appear to have achieved this end very
are derived and are used in discussion of the theory of Bessel loud successfully.
speaker horns.
In the two following chapters the integrals for products of two-
Bessel functions are discussed. The author prefers a purely
A risto tle , G alileo, a n d th e
mathematical derivation of these integrals. For engineers it T o w e r o f P isa
would be very instructive to have also a derivation based on a
B y Lane Cooper,
A r i s t o t l e , G a l i l e o , a n d t h e T o w e r o f P is a .
consideration of principal modes of vibration of a stretched mem Professor of the English Language and L iterature a t Cornell
brane as done by Lord Rayleigh in his “Theory of Sound.” U niversity. Cornell U niversity Press, Ith ac a, N . Y., 1935. H um
In the last chapters of the book the modified Bessel functions phrey M ilford, London. C loth, 5 3/« X 8*/i in., 102 pp., 4 line
and the Ber, Bei, Ker, and Kei functions are discussed and their drawings, $1.50.
applications are shown in the solution of a transmission-line prob R e v ie w e d b y J. O rm o n d ro y d 6
lem and a problem on the resistance of conductors to alternating
current. T )R O FESSO R Cooper’s scholarly little booklet is based on
From this brief review it is seen th a t the book covers all por -*■ two major theses. First, th a t Aristotle never said th a t two
tions of the theory which are of practical importance. I t con bodies of unequal weight would fall to the earth with unequal
tains a large number of formulas and also several numerical velocities and, second, th a t Galileo never dropped two unequal
tables which will be useful in application to practical prob bodies from the Leaning Tower of Pisa to prove th a t Aristotle
lems. I t is clearly written so th a t its reading will not be was wrong if he had said it. Minor theses are, first, th a t even
difficult for research engineers working in acoustics or in electrical if Galileo did make the discovery th a t bodies of unequal weight
engineering. The only criticism which we have to make is th at fall in identical fashions in vacuo, others had discovered this fact
the book does not contain a sufficient number of illustrative prob before him and, second, th a t Aristotle and Archimedes were intel
lems with complete derivations of equations and with a detailed lectual giants compared with the “amateurish” Galileo. These
solution of these equations. Only in working through all the theses are backed up by quotations from the works of Lucretius,
steps in the solution of such problems can a beginner familiarize Aristotle, Philoponus, Leonardo da Vinci, Cardan, Stevin, Gali
himself with the use of Bessel functions. The numerous prob leo, Renieri, Viviani, Whewell, and Dingier given both in the
lems given at the end of each chapter are, in most cases, of a original languages of publication and in English translation.
purely mathematical character and cannot be of much interest By implication, Professor Cooper gives able expression in this
to an engineer. The few technical examples which are considered book to the antagonism which the humanistic outlook has long
in the book contain only a very brief derivation of the equations, felt toward quantitative philosophy. But as an interpretation
and there is not sufficient space given for explanation of the of a phase in the history of science one feels th a t the author has
physical meaning which is behind the mathematical operations. missed the point.
Perhaps the conditions can be improved if this book is read to Let it be adm itted th a t Aristotle and Archimedes stand
gether with the book on loud speakers by the same author, in higher than Galileo in some absolute scale of genius, if such a
which, we understand, Bessel functions are extensively used in scale can be devised. B ut suppose th a t Galileo never dropped
the solution of technical problems. two weights from the Leaning Tower—th a t Newton always
avoided apple orchards—th a t Archimedes never took a bath.
One might be tempted to u tter even greater heresies and deny
E x p lo rin g in P h y sics th a t Washington ever handled a hatchet or state boldly th a t there
is no Santa Claus. None of these legends is essentially part
By Reginald J. Stephenson, R yerson Physi
E x p l o r i n g i n P h y s io s .
cal L aboratory, The U niversity of Chicago. T he U niversity of of the history of what really happened—b u t all are symbols, in the
Chicago Press, 1935. Cardboard, 6‘/s X 9'/< in., 205 pp., 2 popular mind, indicating th a t something important to mankind
photographs, 225 line drawings, $1.50. did happen.
R e v ie w e d b y J. O rm o n d ro y d 6
Professor Cooper is badly misled when he equates the isolated
statements of Galileo’s predecessor and contemporaries to the
XPLORING in Physics” is designed to be a companion vol complete intellectual process contained in the dynamical sections
ume to Professor Lemon’s introductory treatise on physics of “The Dialogues Concerning Two New Sciences.” Even when
“From Galileo to Cosmic Rays,” Professor Stephenson supplying compared with such innovators as Copernicus, Gilbert, Kepler,
11 E xperim ental Division, W estinghouse E lectric and M anu • E xperim ental Division, W estinghouse Electric & M anufacturing
facturing Com pany, South Philadelphia, Pa. Mem. A.S.M .E. Com pany, South Philadelphia, Pa. Mem. A .S.M .E.
A-40 JOURNAL OP A PPL IE D MECHANICS
Bacon, and DesCartes, Galileo is still the first completely modern tical man” who concerned himself with the very practical problem
man. His process of observation, definition, hypothesis, deduc of determining w hat temperature would result from mixing hot
tion from hypothesis, experimentation, and further deduction water with cold. His formula was the expression of no theory
completely outlined in his dialogues has given the pattern to all regarding the nature of heat, but merely an expression of common
scientific thought in the last three centuries. This is what is sense applied to simple experiment. Richmann (1711-1753),
important. If Galileo took p art in any circus tricks at the on the contrary, attem pted to rationalize Krafft’s formula and
Tower he passes them over in silence in his written work, describ argued for the superiority of his own th a t it gave results con
ing, instead, the less spectacular, but more exact process of roll sistently higher than experiment by amounts attributable to the
ing bronze balls down grooved boards. vessel for which no correction could be contained in a formula
Exact observation, measurement, description and compari comprehending only masses and temperatures. In Richmann’s
son, and the meaning of exactness and its limitations are the deductions there appear suggestions th at he actually distinguished
new ideas which the modem world has contributed to the intel heat and temperature, yet these suggestions are all too vague
lectual history of mankind. If Galileo’s share in creating this to adm it of definite interpretation.
new outlook is swallowed up by the legend of the Leaning Tower Wilcke (1732-1796) undertook his experiments on latent heat
this reviewer agrees with Professor Cooper th a t the legend must in an effort to understand why hot water poured on snow did not,
be suppressed. as he expected, convert many times its weight of snow a t a tem
perature slightly lower than 0 C into water a t a temperature
slightly above 0 C. Again with Wilcke, however, inability to
T h e D isc o v ery o f Specific a n d separate the ideas of heat and temperature led to difficulties
and forced him to conclude, paradoxical as it seemed, “th a t just
L a te n t H e a ts as snow in melting takes up 72 degrees of heat without becoming
T h e D is c o v e r y o f S p e c ific a n d L a t e n t H e a ts . B y Douglas warmer, so water in freezing must give up 72 degrees of heat
McKie, P h.D ., B.Sc., L ecturer in th e H istory and M ethods of which if imparted to another equal mass of water would produce
Science, and Niels H . de V. H eathcote, B.Sc., U niversity College, in the latter 72 degrees of heat without its being possible to
London. W ith a Foreword b y E. N . da C. A ndrade, D.Sc., Ph.D ., detect the loss of heat in the ice th at was produced by the freez
F.R .S., Quain Professor of Physics in the U n iv ersity of London.
E dw ard Arnold & Co., London, 1935. Cloth, 4 3/< X 7V< in., 155 ing of the original mass of water.” The authors conclude this,
pp., 5 plates, 6s. chapter by calling attention to the superiority of Black’s re
searches and emphasizing the fact th a t they antedated those of
R e v i e w e d b y J o h n A. G o f f 7
Wilcke by more than ten years.
'T 'H IS little book contains an interesting and important contri- In the last chapter, under the heading of “Some Immediate
bution to the history of science. In it are reviewed in careful Consequences,” the authors trace the rise and progress of calo
and objective manner the early investigations into the nature rimetry, the attem pts to determine the absolute zero of tempera
of heat which may be said to have laid the foundations of the ture, and the early investigations into the “weight of heat.”
modern dynamical theory. The reader will be impressed with the Thus they conclude their study with the classical experiments of
clearness with which the early confused and confusing ideas have Rumford which began the systematic attack on the material
been interpreted and will agree with a statement in the foreword theory of heat and paved the way for the modern dynamical
by Professor Andrade, Quain Professor of Physics in the Univer theory.
sity of London, th a t “the book owes much of its excellence to One of the most interesting paragraphs in the book is th at deal
the fact th a t throughout it is based on a study of the work of the ing with the work o£ Jean Morin, editor, after Kepler, of the
original writers, in the original language.” The reader will also Rudolphine Tables. Morin’s work (1661), which appears to be
gain new appreciation of the difficulties with which early experi the very earliest attem pt to deal mathematically with a quantita
menters were faced owing to their inability to separate the ideas tive problem in heat, is especially interesting for its bearing on
of heat and temperature and new surprise th a t in these days of thermometry. To Morin, each of the eight divisions on the
scientific thinking this inability still persists, not only in the mind thermometer of his period, below the highest point represented a
of the layman, but of physicist and engineer as well. “degree of cold,” while each division above the lowest point
The work of Joseph Black (1728-1799) is discussed a t length represented a “degree of heat.” A temperature corresponding to
in chapter 1 and, in chapter 2, all doubt as to the date and prior the mid-point of the thermometer would therefore be one of
ity of Black’s experiments and his claims to the discovery of “four degrees of heat and four degrees of cold.” In reasoning as
specific and latent heats is resolved. These chapters are of to the result of mixing equal masses of water, the one having
present interest for the references they contain to James W att, “four degrees of heat and four degrees of cold,” the other having
the famous pupil of Black and great improver of the steam engine, “two degrees of heat and six degrees of cold,” he rejected the
whose Bicentenary engineers all over the world have lately correct result because to him it represented six degrees of cold
celebrated. as having proportionately less effect on four degrees of heat than
In chapter 3 are discussed the achievements of the Baltic the four degrees of heat would have on six degrees of cold. One
workers: Krafft and Richmann in St. Petersburg, Wilcke and wonders what expression could be given the Second Law of
Gadolin in Sweden. Krafft (1701-1754) was essentially a “prac Thermodynamics in terms of Morin’s ideas of temperature and
is the more willing to accept the researches of Black as a real
7 Professor of Therm odynam ics, U niversity of Illinois, U rbana, 111. discovery of specific and latent heats.
Mem. A .S.M .E. I t is superfluous to add th a t the book is highly recommended.