UC-NRLF

TJ
7 751
A/ 3 B 3

U
Dlfl

EACH PAMPHLET IS ONE UNIT IN A COMPLETE LIBRARY OF MACHINE DE-

SIGN AND SHOP PRACTICE REVISED AND REPUBLISHED FROM MACHINERY,

No. 14

DETAILS OF MACHINE
TOOL DESIGN
SECOND EDITION

CONTENTS
Elementary Principles of Cone Pulleys and Belts, by W.
L. CHENEY 3
Cone Pulley Radii, by JOHN J. HARMAN 7

Strength of Countershafts, by FRANK B. KLEINHANS *5

Tumbler Gear Design, by JOHN EDGAR 19
Faults of Iron Castings, by FORREST E. CARDULI o 26

Proportions of Machines Built in a Series of Sizes, by

STANLEY H. MOORE

Copyright 1910, The Industrial Press, Publishers of MACHINERY

49-55 Lafayette Street, New York City
 MACHINERY'S
REFERENCE SERIES
EACH NUMBER IS ONE UNIT IN A COMPLETE
LIBRARY OF MACHINE DESIGN AND SHOP
PRACTICE REVISED AND REPUB-
LISHED FROM MACHINERY

NUMBER 14

DETAILS OF MACHINE
TOOL DESIGN
SECOND EDITION

CONTENTS

Elementary Principles of Cone Pulleys and Belts, by W. L.

CHENEY 3

Cone Pulley Radii, by JOHN J. HARMAN - -

7

Strength of Countershafts, by FRANK B. KLEINHANS 15

Tumbler Gear Design, by JOHN EDGAR -

19
Faults of Iron Castings, by FORREST E. CARDULLO 26

Proportions of Machines Built in a Series of Sizes, by STANLEY

H. MOORE -
36

Copyright, 1910, The Industrial Press, Publishers of MACHINERY

49-55 Lafayette Street, New York City
-5 A
:
 CHAPTER I

ELEMENTARY PRINCIPLES OP CONE PULLEYS

AND BELTS*
Everyone knows that cone pulleys are usually made with regular
steps; that is, if it is one inch from one step to the next, it is also
one inch from the second to the third, etc., the reason being that
when the centers of the shafts on which the cones run are a fair dis-
ance apart, the belt will pass very nearly half way around that part of
each cone on which it is running, and the length of the belt will con-
sequently be approximately equal to twice the distance between the
shafts, added to half the circumference of the grade of one of the
cones on which it is running, and half the circumference of the grade
of the other cone on which it is running. As the steps are even, the
half circumference of any two grades of each cone will, when added
together, produce the same result. For example, if we had two cones,
the diameters of the several grades of which were 6, 8, 10 and 12
inches, it is evident that the sum of half the diameters taken anywhere
along the cones, as they would be set up for work, would in every case
be the same. If the diameters are the same, it follows that the
circumference must also be the same, and, of course, that half the
circumference must be the same, so that when the centers of the shafts
are a fair distance apart, and the difference between the largest and
smallest step of the cone not too great, the same belt will run equally
well anywhere on the cone, because it runs so near half way around
each grade of the two cones on which it is running, that the slight
difference is within the practical limit of the stretch of the belt.
But when the shafts are near together, and when the difference
between the largest and smallest step of the cone is considerable, the
belt is not elastic enough to make up this difference. Fig. 1 shows a
three-step cone, the grades being 4, 18, and 32 inches diameter, re-
spectively, there being a difference of 1* inches on the diameter
for each successive grade, and the step being therefore 7 inches in
each case. Of course, it is n<& likely that such a cone as this would
be made for practical use, but it is well to go to extremes when look-
ing for a principle. Now, it is evident that two cones, even if like the
one shown in the cut, were set up far enough apart, they would still
allow the belt to run very nearly half way around each grade of the
two cones, the angularity of the belt would be slight, and the length
of belt would therefore still be as mentioned above.
But (again taking an extreme case) by reference to Fig. 2, which is
intended to represent a belt running from the largest grade of one cone
to the smallest grade of the other cone, we see that the belt runs
three quarters of the way around the large pulley, and only one quarter
*
MACHINERY, April, 1895.

347524
4 No. rj~MACHINE TOOL DESIGN
of the way around the small one, the distance between the shafts in
this case being 19% inches.
The length of this belt will evidently be equal to three quarters of
the distance around the large pulley, plus one quarter the distance
around the small pulley, plus the distances A and B, which we find to
be each 14 inches. The circumference of a 32-inch diameter pulley
is 100% inches, and the circumference of a 4-inch diameter pulley is

12% inches (near enough for our present purpose) three quarters of;

100% is 75%, and one quarter of 12% is 3%; the length of a belt, then,
to go around a 4-inch pulley and a 32-inch pulley, running at a dis-
tance of 19% inches apart, is 75% plus 3% plus 14 plus 14; total, 106%
inches.
Now, let us take the middle cone, when the belt is running on two
pulleys, both 18 inches diameter (see Fig. 3), and, of course, the same

Machinery, N.T.
Flg.l Pig. 2

distance apart as before. The circumference of an 18-inch pulley is

56% inches, and half the circumference of two 18-inch pulleys is evi-
dently the same as the whole circumference of one 18-inch pulley;
the length of belt in this case will then evidently be 56% plus 19%
plus 19%; total, 96 inches. It is therefore evident that a belt long
enough to run on a 4- and 32-inch 19% inches apart, is 10%
pulley,
inches too long to run on two 18-inch pulleys 19% inches apart, and,
of course, it is therefore 10% inches too long to run on the middle
grades of such a cone as we have under consideration.
The thing to do, then, is to make the middle grades of these cones
(or the two 18-inch pulleys) enough larger than IB inches diameter
to just take up this 10% inches of belt, and if this were the only
case we had to deal with, it would be very easy to settle it by saying
that as half the circumference of two 18-inch pulleys is the same as
the whole circumference of one 18-inch pulley, we should make the two
18-inch pulleys enough larger in diameter to make an additional cir-
cumference of 10% inches; and as 3% inches is nearly the diameter of
 CONE PULLEYS AND BELTS 5

a 10%-inch circumference pulley, by making the middle

of both cones
18 plus 3% inches diameter (that is, inches diameter) our trouble
21%
would be ended in this particular case. It is easy enough to see, by
looking at Fig. 2, that the belt being obliged to go three quarters of
the way around the large pulley, is what makes it so much too long
to go around the two middle pulleys, where, of course, it goes but half
way around each. But, of course, what we want is some way of cal-
culating the diameters to turn any pair of cones, running at any dis-
tance apart.
If we were to draw these same 32- and 4-inch pulleys twice 19%
inches apart, and then three times 19% inches apart, and so on, until
we got them far enough apart so that the belt would practically run
half way around each, and should calculate the diameter of the middle
grade of the cone to fit each distance, we would probably formulate
a rule that would work for any distance apart, with this particular
cone; but as it is evident that the further apart the cones are to run,
the nearer to the nominal diameter of 18 inches must the middle of

Machinery, N.Y.
Fig. 3

the cones be turned, so also must it be evident that the less difference
between the largest and smallest diameter of the cone, the less must
also be the excess over nominal diameter of the middle of the cones.
Any method, then, of calculating such problems must take both of
these things into consideration. The nominal diameter of the middle
of any cone will be equal to half the sum of the diameters of the
largest and smallest part respectively. This is almost self-evident, and
no proof of it is necessary in this connection. What we want, then, is
some way to find out how much larger than the nominal diameter to
turn any one cone or cones to fit the conditions under which they are
to run. The following formula is the result of a thorough investiga-
tion of this subject by Prof. Rankine, and has proved itself to be prac-
tically correct in the shop, as well as satisfactory to those mathemati-
cians who are competent to criticise it. This formula is:

2 2irC
This formula translated into plain English means that the radius
of the center of a cone will be equal to the radius of the smallest
part, added to the radius of the largest part, and this sum divided by
6 No. 14 MACHINE TOOL DESIGN
2, and added to this the difference in radii between the largest and
smallest part squared, and then divided by twice the center distance
between the cones multiplied by 3.1416. That is, the first half of the
formula gives the radius at the center of a cone, when the largest and
smallest radii are known, and, of course, if the middle radius is equal
to the smallest radius added to the largest radius and the sum divided
by 2, it follows that the middle diameter is equal to half the sum of
the diameters of the largest and smallest part, respectively, as men-
tioned before. The second part of the formula allows us to calculate
how much larger than this nominal diameter to make the middle of a
cone, no matter what the size or center distance.
Applying this formula to the case of the cones shown in Pigs. 1, 2
and 3, we find the radius of the middle of the cone to be 10 6/10 inches,
or, what is the same thing, the diameter to be 21 2/10 inches, which,
in view of the extreme case under consideration, is very near the first
result obtained (21%), and shows that the formula is perfectly safe in
any case likely to occur in practice.
When this formula is reduced so as to express the numerical value
of diameters instead of radii, it takes the following form:
D+ d
Diameter at center of cone = -
-\
--(D d)-
,

2
the I2,y2 being the nearest value in plain and easy figures to which
the quantity containing w in the original formula can be reduced.
Applying this simplified formula to the cone which we have been
considering, it will be found that the middle diameter is 212/10, the
same as by the original Rankine formula.
a cone has five steps instead of three, it will be practically correct
If
to add half as much to the nominal diameters of the second and fourth
grades as was added to the middle grade, or, if it has four grades,
add two-thirds of what is found by the calculation to the second and
third grades (as there is evidently no middle grade). If more than
four or five grades, add to each grade according to the same principle.
We have so far been considering two similar cones, but it often
happens that one cone is larger than the other. In such case the
problem becomes a little longer to work, and the length of belt neces-
sary to go around each pair of steps of the cones must be used to find
the diameters; that is, starting with one end of the cone, find the
length of belt, and then calculate how much larger or smaller (as the
case may be) than the nominal diameter it is necessary to make each
of belt run properly.
grade, in order to make the same length
Prof. Rankine has worked out a formula for the length of belt also,
which, reduced to diameters,

Length of belt
is

= 20 H -----
as follows:
11D + lid
7 40
1
(D dY-
.

That the length of a belt to

is. pass around any two pulleys (and,
of

course, a cone is simply a set of pulleys) is the sum of the following

quantities: First, twice the center distance of the shafts; second, 11

times the diameter of the larger pulley, plus 11 times the diameter
 CONE PULLEY RADII 1

of the smaller pulley, and this sum divided by 7. This gives the
nominal length of belt, or what would be practically correct if the
center distance was fairly great; for the excess, the last part of the
formula must be used, which is the difference between the diameters
of the larger and smaller pulleys squared, and this result divided by 4
times the center distance.
Having found the length of belt to run on one end of the cones,
and keeping this for a starter, we can easily find how much to add to,
or take from, the nominal diameter of any other part of the cone to
make the same belt run, as explained before. If, for instance, we find
that the nominal diameters of the next grades that we try to bring the
length of belt one-half inch shorter than the first calculation, we add
enough to one or both diameters to make up one-half inch of circum-
ference, which would be about 5/32 of diameter, and this could all be
added, to one pulley, or half of it could be added to 'each pulley, as
convenient, and this would be practically correct.

CHAPTER II

CONE PULLEY RADII*

In the present chapter a method presented by Dr. L. Burmester in
his "Lehrbuch der Kinematik," for the solution of the cone pulley
problem, has been extensively treated. Dr. Burmester's method is
entirely graphical, and is exceedingly simple in application. While it
is not theoretically exact, it is, as will be shown later, much more
accurate than practice requires.
In order to bring out more clearly the points which will come up in
the case of open belts, let us first consider the simple case of crossed
belts. It is a well-known fact that in this case the only calculation
necessary in order to find the radii of the various steps is to make the
sum of the radii of any two corresponding steps a constant. This
may be shown in the following manner:
a == radius of step, driving cone.
=
A length of belt from contact on driving cone to contact on driven
cone.
& = radius of step, driven cone.
E = distance between centers of cones.
K = the constant sum of the of two
radii' corresponding steps.
8 = angle shown, Figs. and
4 5.

Then
ax + 6a K
sin 1
= '

E E
a, + 62 K
sin 02 =
E E
*
MACHINERY, September, 1905.
s No. 14 MACHINE TOOL DESIGN
Therefore 8 1 6n = = =a
constant.
Therefore the arc of contact on each pulley = 180 + 2 =a con-
stant.
A! A A.2
Also cot 6 = ! 2

a, + 6, IT a, + b2 7T
A, A 2
= If cot e =a constant.
180 + 20
But length of belt = 2 A -I X (2 IT a, +2* ftj)
360
180 + 20
= 2 AH X 27rX(a 1 + 61 )
360
in which, as has been shown above, all the terms are constants, there-
fore length of belt is constant.

Fig. 4 Fig. 5

Note: The subscript applied to the letters denotes that the letters
are used for the corresponding quantities in a special case; thus c^,
in Fig. 4, refers to a.
The may be determined graphically by the
radii of the various steps
following diagram (Fig. 6) :

Draw a horizontal line from A, and also draw AC making an angle

of 45 degrees with it. On this line lay off AS equal to the distance
between the cone centers, using any scale most convenient, bearing
in -mind, however, that the scale adopted now must be used con-
sistently throughout the diagram. At 8 erect the perpendicular TST'
U> the line ASC. From some convenient point on AC, as D, drop a
vertical equal to some known radius of the cone a, as DE, and then
 CONE PULLEY RADII 9

from E measure back on this vertical the radius of the corresponding

step on cone &, as EF, and from these points E and F draw lines
parallel to ASC. From the point G, where the line FG intersects the
line TST' drop a vertical. This will intersect the line EH in H.
Through H draw the horizontal MN, being the point where this line
intersects the line TST'. Then, distances on the line MO may be taken
to represent radii on cone a; and to find the corresponding radii on
cone 6 erect perpendiculars at the extremities of these radii, producing
them until they intersect the line TST'. These perpendiculars then
represent the desired radii. It may be shown as follows that the sum
of the two corresponding radii, as obtained from this diagram, is
always a constant, and the diagram therefore satisfies the conditions
for crossed belts.
Let MJ represent any radius on cone a, then JI represents the corre-
sponding radius on cone &.

Machinery, N.f.t

Fig. 6

The / JIO =3 / JOI =

45 degrees.
Therefore JI JO.=
Therefore MJ JI+ J =MJO + = =
MO a constant.
Dr. Burmester's diagram for open belts is a modification of the
diagram just shown, the only difference being that the line TST is
f

replaced by a curve. This curve was determined by plotting a series

of points,and after several pages of exceedingly intricate mathematics
he arrives at the astonishing result that this curve can be replaced
by a simple circular arc without any appreciable error.
The diagram is shown in Fig. 7, and may be drawn as follows:
Proceed as in Fig. 6 until the line TST' is drawn, then lay off distance
1
SK equal to AS. Next, with the center at A, and a radius equal
2
to AK, describe the arc XY, and the diagram is ready for use.
In order to give an idea of the extreme accuracy of the diagram, let
us observe the values obtained by Dr. Burmester in his calculations.
10 No. 14 MACHINE TOOL DESIGN
Let R = AK (Fig. 7).
= R = E X 1.11815172.
When e 0,

6=15, R = E X 1.11806842.
= 30, R = E X 1.11798671.
6

0=45, R = E X 1.11803397.
The value used for R in the diagram E X 1.11803397; so the maxi-
is

mum error of R occurs when = and equal to E X 0.00011775,

0, is

-V. 1'.
Machinery,

Fig. 7

which is much more accurate than the work of the most careful drafts-
man. Dr. Burmester gives values of R up to =
90, but as it is evi-
dent from Fig. 8 that it would be practically impossible to have a value
of 6 greater than 45, the writer has omitted the other values.

Figr. 8

In order to make the use of the diagram perfectly clear, let us solve
the following problems:
Problem 1. Fig. 9
Given :

Distance between centers of cones = 3' 4".

 CONE PULLEY RADII 11

Diameters of driving cone, 4", 8", 14", 20".

Diameters of driven cone, X, X, 14", X.
Required:
All diameters of driven cone.
Lay out the diagram and determine the point M
as previously
directed. Now
the radii of driving cone may be laid off as abscissas or
ordinates, whichever happens to be the more convenient, as the results

Fig. 9. Solution of Cone Pulley Problem -when Diameters of One Pulley and
Center Distance are Known

obtained will be exactly the same in either case. In this particular

problem it is evidently more convenient to lay them off as abscissas.
Then the ordinates erected at the ends of these abscissas will repre-
sent the corresponding radii of the driven cone. The problem is solved
in Fig. 9 and the following results obtained:
Results:
Diameters of driven cone, 22%", 19%", 14", and 7%".
This problem does not bring out all of the fine points of the diagram,
so let us solve a more complicated one, in which the different steps of
12 No. 14 MACHINE TOOL DESIGN
the cone are to transmit given velocities.
Problem 2. Fig-. 1O
Given:
Distance between centers of cones = 3' 4".

Maximum velocity of belt (assumed) 30 feet per second.

R. P. M. of driving cone 240. =
Required:
Driven cone to make 100, 240, 400 and 580 R. P. M..
The maximum belt speed will be attained when the belt is on the
largest step of the driving cone.
Therefore
2ira^ X 240
= 30; 20i =
12 X 60

2&! 240
But = ; 2^ =
28% 580

Now having obtained a value for a^ and & 1? the point on the dia- M
gram may be found. Next draw a line from M
as MO, inclined so that
any horizontal projection, as MN, will be to the corresponding vertical
projection, NO, as the R. P. M. of the driver are to the R. P. M. of the
driven; thus,
MN R. P. M. of driving cone

NO R. P. M. of driven cone

Also from similar triangles

MN MN'
NO N'O'
But we know that
R. P. M. of driving cone rad, of driven one

R. P. M. of driven cone rad. of driving cone

Therefore MN' equals radius of driven cone, while N'O' equals radius
of driving cone, thus making, for this case, radii of driving cone verti-
cal and of driven cone horizontal. The problem is solved in Pig. 10
and the following results obtained:
Results:
Dia. or driving cone, 28%", 25%", 20%", 11%".
Dia. of driven cone, 11%", 15%", -20%", 28%".
We have seen that the Burmester diagram is under all conditions
much more exact than is required in practice; and a more compact,
simpler, or quicker method of finding cone pulley radii could not be
desired. An experienced draftsman should be able to solve a prob-
lem than 10 minutes, while to obtain the same
like No. 2 above in less
results by an analytical method would require as many hours. Results
of sufficient accuracy can usually be obtained by making the diagram
to half scale, although there is no reason for reducing the scale, unless
the distance between centers is inconveniently large, and in that case
 CONE PULLEY RADII 13

the results do not need to be so accurate, as the belt will stand more
stretching.
Another graphical method for laying out a pair of cone pulleys is
as follows: First draw straight line A A, Fig. 11, supposed to connect
the centers of the cones to be laid out; then set off the centers of tht
cones B and C on line A A (full size is best) then bisect the distance
;

SQO

Fig. 1O. Solution ofCone Pulley Problem when Velocity Ratios, Maximum Belt
Speed, Center Distance, and R. P. M. of Driver are Known

between the centers of the cones and draw perpendicular line D E. Now
assume the size of the two cones say the largest is 25 inches and the
smallest 3 inches diameter. Then draw a line tangent to the circles,
or the line representing the inside of the belt G, which will intersect
the line D E at E, and taking the point E for a center scribe the circle
F. Then divide the circle F, commencing at the line of the belt G,
into as many parts as needed, of a length to suit the required speeds.
14 No. 14 MACHINE TOOL DESIGN
Draw the other radiating belt lines through the point E and the divi-
sions on the circle F, extending them toward the cone B, and they
will be the inside of the other belt lines. Draw circles tangent to these

.Vac!< inert, M T.
Fig. 11. Simple Graphical Solution of Cone Pulley Problem

lines. We now have all the diameters of the rest of the steps of the
cone to match the first, and the belts will correctly fit all the steps.
This is, of course, only an approximation rule. This method was con-
tributed to the June, 1905, issue of MACHINERY by John Swanberg.
 CHAPTER III

STRENGTH OF COUNTERSHAFTS*
There is scarcely a shop in existence which has not had a more or
less serious accident from a countershaft some time in its history. It
may have been caused by a heavy pulley running very much out of
balance, or the shaft may have been bent in the beginning. Possibly
the shaft was too light, or too long between hangers. The latter is
responsible for most of the trouble, and is the one with which this
discussion is principally concerned.
There are two methods in vogue for turning cones and pulleys; one
is toset the rough casting to run true on the inside, and the other on
the outside. This latter method makes a cheaper and an easier job,
but when turned, it requires an enormous amount of metal to balance
it. And here is the source of considerable trouble. We may balance
a large cone perfectly on straight edges, but that is a standing balance
only; and when the cone is put in place and speeded up to several hun-
dred revolutions per minute, it shakes, and shows that it is decidedly
out of balance. The trouble is that we have not placed the balance
weights directly opposite, or in the plane of the heavy portion of the
cone. The result is that neither weight, when rotating, has its counter-
balance pulling in the same line, and, of course, the pulley is sure to
be out of balance. All cones and all other pulleys which have a wide
face should be set to run true on the inside before turning.
A certain countershaft failed because it had been welded near the
center. The weld twisted and bent open, and some one was badly
injured by the fall. A weld in machine steel is so very uncertain that
it should never be trusted for such a purpose. The extra expense of a
new shaft would not warrant the hazard of such a risk.
In the calculations which follow, the spring of the shaft is limited
to 0.06 of an inch. There are plenty of countershafts which have been
running for years with about this much spring. Now, from the general
formula for the deflection of a simple beam, we have:
WL S

The deflection, or spring =

48 #7
in which W = the load at the center in pounds.
I/= the length between center of hangers in inches.
E= the coefficient of elasticity 29,000,000. =
7 = the moment of inertia of the cross-section of the shaft.
For a round shaft,
4
7T<Z
1 = (1)
64

*MACHINEBY, April, 1903.

16 No. 14 MACHINE TOOL DESIGN
in which d the diameter of the shaft in inches. We then have:
WL 3

= 0.06 (2)
48 EI
64 WL 3

From (1) and (2), we have 0.06, and

48 E TT d4
3 0.06 X d* 3 0.06
0. X 48 X 29,000,000 XT dl
~=
1

L ~~ X
64 W N 64 W
Jd*W 4,100,000 (3)

Fig. 12 shows a countershaft which is in actual service, and which is

known to be all right. A and B are keyed to the shaft. C and D are
24 DIA. X 6 BELT 24 DIA. X 6 BELT
WEIGHT= 30 TOTAL WEIGHT OF B, C AND D = 110
BELT

Machinery,N.Y.

Fig. 12

loose pulleys arranged for open and cross belts.

A
weighs 30 pounds, and B, C and weigh 110 pounds. The belts D
run as shown in the figure. If A
weighs 30 pounds, and the centers of
the hangers are 54 inches apart, then by taking the left-hand hanger as
the center of moment, we have 30 X 12 x X 27, when x is the weight =
at the center. Solving we find
30 X 12
x = = 13
27
In the same way, by taking the right-hand hanger for the center
moment, we find that
110 X 18
x2 = = 73
27
As to the belt pull, it is possible for a single belt to run up to 70
pounds per inch of width of belt, and a double belt can be taken at 100
pounds. As a double belt is used in this case, and as the slack side
of the belt is very loose when the tight side is pulling its maximum,
we will take the pull at the pulley A 6 x 100 600 pounds, and get- = =
ting this in terms of a load at the center, we have
600 X 12
#3 = =266
27
 STRENGTH OF COUNTERSHAFTS 17

and the downward pull is 13 + 73 + 266 = 352.

The pull at the pulley B will be 6 X 100 = 600, and by transferring
this to the center we have
600 X 18
=400
27
The resultant of these two forces will be the diagonal of the force
diagram, and is equal to 530 pounds, which is equal to in the for- W
mula. Introducing these terms in equation (3) we have
sf (2.44)*
L= v 4,100,000'
N 530
and by solving we find L = 65, which means that
for/ this condition of

loading the countershaft would be safe even with the hangers 65 inches
apart.
N!
30 DIA. X 2% BELT

Machinery} N.Y.
Fig. 13

Fig. 13 represents another countershaft taken from actual service.

It is belted as shown on the left-hand view, and is running all right,
although it looks rather flimsy, and one would consider it unsafe. Tak-
ing the moments of the weights of the pulleys and belt pull about the
right- and left-hand supports, and finding the equivalent pull at the
center, we obtain:
Weight at center due to pulleys = 148
Pull on 30-inch pulley =
2V2 X 100 = 250
250 X 30
Pull on 15-inch pulley = 2% X 100; = 208
36
Total downward pull= 606
500 X 30
Pull on 14-inch pulley = 5 X 100; = 417
36
Resultant downward pull = 189
Introducing this value of W
in equation (3) we have,

(1.75)'
L 4,100,000 = 59
189
This is considerably less than the distance between the hangers, and
it shows that it is not safe to place the hangers in this way. If the
18 No. 14 MACHINE TOOL DESIGN
belts ran as shown at the right-hand side of Fig. 13, we would then
have:
Weight due to pulleys (as before) = 148
Pull on 30-inch pulley = 250
Pull on 15-inch pulley = 208
Total downward pull= 606
Horizontal pull on 14-inch pulley = 417
From, these two forces we find a resultant of W = 736. Substitut-
ing this in (3) and solving as before, we find L 38, which is the =
greatest safe distance between hangers for this condition of loading.
There are cases where one must have an extra long shaft in order
towork in the pulleys, cones, etc., as shown in Fig. 14. Here the
downward loads amount to 820 pounds, and the pull at right angles
amounts to 360 pounds. The resultant 895 pounds W. =
30 DIA. X 6 BELT
28 DIA. X 6BELT WEIGHT OF CONE = 224
WEIGHT = 337 BELT

Machinery, N.Y.
Pig. 14

Introducing in the formula we have

(2.44)*
L= 4,100,000 = 55
895
This means that for this condition of loading, the center distance
should not exceed 55 inches, and since in this case it could not be made
as small as this, the pulleys should be arranged for a third hanger.
In every case, therefore', where the centers are so far apart as for-
mula (3) would indicate to be unsafe, a third hanger should be used.
If all the flimsy countershafts had a third hanger added to them there
1

is no doubt but that the number of accidents would be greatly dimin-

ished. In the above calculation the weight of the countershaft has

not been considered, as it is usually very small. If the belts run at
any other angle than that shown, the construction is made in exactly
the same way, using the required angle instead of a right angle, the
resultant of the two forces being used as in the formula. W
 CHAPTER IV

TUMBLER GEAR DESIGN*

Of the different mechanisms that have been used in the machine
tools of the past, one the tumbler gear could be found in some form
or other in almost every machine. Its office, in most cases, was to
'reverse the direction of the feed. Pig. 15 shows fhe usual form in
which it is found when used for this purpose. The gears A and B are
to be connected so that motion may be transmitted from one, which
runs constantly in one direction, to the other, which it is desired to

Machinery,N.Y.
Figs. 15 and 16. Examples of Tumbler Gears

run in either direction. Suppose that A is the driver and runs as

shown by the arrow. As connected, A drives B through the inter-
mediate gears D and C, B rotating in an opposite direction to A, as
shown by the arrows.
This mechanism is termed the tumbler gear, because the gears D
and C are supported in a frame which swings about the axis of either
the driving or the driven gear. In the case in hand, the intermediate
gears are carried in the frame E, which rotates about the axis of the
gear B. Some means, not shown, must be provided by which the
rocker frame may be changed from one position to the other, and
locked. Pig. 16 shows the mechanism shifted so that the motions of
A and B are in the same direction.
The tumbler gear has been used as a reversing gear ever since pres-
ent forms of machine tools were first invented. While it has always
*
MACHINERY, December, 1907.
20 No. 14 MACHINE TOOL DESIGN
given considerable trouble, it has shown up to disadvantage mostly
when applied to the modern machine with positive gear feed, where
great power has to be transmitted by it. It is the purpose here to
show where this gear may be used to advantage, and also to explain
the theory on which the principles of its design are based.
All of us have met with this mechanism in some form or other, and
may have formed an unfavorable opinion. The prejudice thus created
keeps us from fully appreciating the tumbler gear, even when properly
designed, and when used in the right place. It has been placed by
many along with the worm drive and the spiral gear as undesirable,
and to be avoided unless it is absolutely impossible to get along with-
out it. This opinion has been responsible for the adoption of many
combinations used for purposes that rightly belong in the field of the
tumbler gear, and many times, in order to avoid using this mechan-
ism, much unnecessary complication has resulted.
What are the faults of the tumbler reversing gear? That one on
So-and-So's lathe used to kick furiously when one tried to throw it
over. Then, the one used on the milling machine used to go into
mesh easily enough, but when any amount of strain was put onto
it, the teeth -used to crack and growl, showing that the tendency was

to drag the gear farther into m^sh, causing the teeth to bind on one
another and sometimes break. Let us look into the case represented
in Fig. 15. Pig. 17 shows the gear D just entering into mesh with A.
An examination of this figure shows that the tendency is for the teeth
of gear A, when they strike those of gear D, to cause the latter to
rotate about the axis of the rocker frame, should the gear B be locked
against turning. This tendency opposes the motion in the opposite
direction necessary to bring the gears wholly into mesh. In practice.
B is not locked, but it is necessary to overcome a certain amount of
resistance in order that it may be set in motion, and the presence ot
this resistance has the same effect as if the gear were locked. The
greater this resistance is, the greater is the effort necessary to bring
the gears into working position.
Examining the conditions in the case of Pig. 16, we see that the
effect would be just the opposite, that is, the gears would come into
mesh of their own accord as soon as a contact is produced between
the teeth of A and C. Practically no effort is necessary to bring the
gears into mesh, but, in order to withdraw the gear C from A, con?
siderable effort would be required. When the gears C and A are in
mesh and transmit power, the tendency for gear C is to crowd farther
into mesh with A, which' has the effect of binding the teeth. Should
the pressure of contact be sufficient, the binding tendency would cause
the motion to cease, or would break the teeth. This is one of the
points on which many have based their verdict against the tumbler
gear, and when designed so that such results are obtained, it is not
to be wondered at.
Direction of Tooth Pressure in Ordinary Cut Gears
The consideration in the design of tumbler gears in any form
first
is that of tooth pressure and its line of application. As all cut gears
 TUMBLER GEAR DESIGN 21

used in machine tools are made 14%-degree involute system, we

to the
will confine ourselves to that system. In this, the force tending to
revolve the driven gear is not a tangential force, applied as a tangent
to the pitch circle, but is a force applied at an angle of 14% degrees to
the tangent of the pitch circle, this 14%-degree line being termed the
line of pressure. In case that there may be some confusion as to the
above statement regarding the tangential force and the line of pres-
sure on the teeth, the case is graphically shown in Pig. 18. The
tangential force is equal to the twisting moment divided by the radius
of the pitch circle. This force is equivalent to that which transmits
motion between two disks by friction alone, the diameters of the disks
being equal to the pitch circles of the gears. This force is, in the
case of a gear, resolved into two component forces. One component
acts perpendicular to the tangential force and tends to force the gears
apart; the other acts in the direction of the line of tooth pressure
shown in Fig. 18. The tooth pressure thus is somewhat less than the

ARROW SHOW
TENDENCY
TO RESIST
THROWING IN

Fig. 17. Illustrating the Tendency to Fig. 18. Illustrating the Direction ol
resist Throwing-in of Gear Line of Pressure

total twisting force, and equals the twisting or tangential force multi-
plied by the cosine of 14% degrees.
Influence of Direction of Tooth Pressure on Tumbler Gear Design
To show what effect the line of pressure has upon the layout of the
tumbler gear, we will use the simple case shown in Fig. 19. In this
figure, A is the driving gear and B is the driven gear. These gears
are connected by means of the intermediate gear (7, which is carried
in the swing frame E, which, in turn, swings about the axis of A.
This mechanism is a simple case of tumbler gear, and while it is
little used, it is useful as a means for disconnecting a train of gears
when it is desired to stop the motion of the driven section. If we con-
sider gear B locked in the position shown, and exerting a turning
effort on the gear A in the direction indicated by the arrow, this effort
is transmitted by the teeth of A and C, and a pressure is produced
22 No. 14 MACHINE TOOL, DESIGN
between the teeth of B and C. two of which are shown in the cut. The
direction in which this force is applied is shown by the line of pres-
sure H K, and is is exerted in the direction of H. Since every force is
opposed by an equal and opposite force when in a state of equilibrium,
we have in this instance a force or reaction opposing the force along
the line of pressure referred to. It is this reaction that causes our
troubles. In the mechanism shown in Pig.- 19, the gear C and the link
E are free to rotate about the axis of A, and since the line of pressure
does not go through the center of gear A, the force acting along this
line tends to rotate the arm E about the axis of A, the direction of
rotation being dependent on which side of the center of A the line falls.
Thus in Fig. 19, the line falls in a position that produces a tendency
for the arm to force the gear C further into mesh with B. The twist-
ing moment thus set up is equal to the tooth pressure multiplied by
the normal distance from the axis of A. or G L.

Machinery, X.Y.
Fig. 19. Objectionable Tumbler Gear Design

If now, instead of trying to turn the gear A in the direction of the

arrow, we exert a torque in the other direction, the opposite sides of

the teeth would come into contact, and the line of pressure would be
located as shown by the dotted line H' K'. The normal distance of this
line from the axis of A is much greater than in the former case; con-
sequently the twisting moment tending to rotate the arm E about the
axis is also increased, but the direction in which the torque is applied
has changed the direction in which the reacting force along the line
of pressure acts, and, since this line falls on the same side of the
axis, the tendency of the arm is to rotate in the opposite direction, and
to separate the gears C and B. Had the line of pressure gone directly
through the axis of the gear A, where E is pivoted, the effect of any
force acting along it would have had no rotating influence upon the

tumbler gear arm. That this would be the ideal case needs not to be
mentioned, and it should be the aim of the designer to approach that
condition as nearly as possible.
 TUMBLER GEAR DESIGN 23

The tendency for the tumbler to crowd the gears into mesh might
be of some advantage were it desirable to throw them into mesh while
the gears are in motion; but in cases where any considerable amount
of power is being transmitted, a very stiff and rigid design will be
necessary for the tumbler frame and the locking device. It is also
well in such cases, when setting the locking device, to have the gears
mesh with plenty of play or backlash, so that, if there be any spring in
the frame, the gears will not be likely to bind and cramp. Should B
be the driver and run in the direction of the arrow, the line of pres-
sure would be H' K', and the pressure would be in the direction of H'.
The arm would then tend to carry the gear C out of mesh with B.

F
Fig. 2O. Correct Design of Tumbler Gear to runMn Both Directions

Should the direction be reversed, H K would be the line of pressure,

and the tendency would be to crowd the gear in.
The layout in Fig. 19 has two bad features. In the first place, the
gears have a tendency to crowd farther into mesh, which limits the
amount of power that can be transmitted, and increases the liability
of breakage of the gear teeth and of the tumbler frame, should an
overload be imposed upon the mechanism. Inaccuracy in the shape
and spacing of the teeth aggravate the above conditions. In the
second place, the mechanism should be used to transmit motion in but
one direction.
In most cases the throwing in or out of the tumbler is a secondary
matter, as it is done either when the gears are not in motion, or while
not under load, if running. In such cases it should be the aim of the
designer to overcome the objection of the crowding of the teeth into
24 No. 14 MACHINE TOOL DESIGN
mesh by having the line of pressure properly located, so that the tend-
ency is in the opposite direction. When it is desired to provide a
tumbler gear that can be run in either direction, the layout in Fig. 20
is recommended. The object in this case is to have the twisting
moment equal in either direction, and such that the gears have no
crowding tendency. The arrangement in Fig. 20 is laid out as follows:
Draw the pitch circles of the gears B and G and connect their centers
by the line D F. Through the pitch point draw a line G normal to H
DF. Then locate the gear A at some point on GH so that its pitch
circle will be tangent with that of C.

The Single Tumbler Gear

The
single tumbler gear is the basis of many of our modern rapid
change speed and feed mechanisms, and the principles treated above
apply to this as well as to the regular tumbler gear. Take the simple
case shown in Fig. 21, which shows the pitch circles of a four-gear

Machinery,N.Y.
Fig. 21. Single Tumbler Gear in a Feed Box

cone and the driver A and tumbler gear C. It is evident that only one
position of the gear C can be such that the ideal condition prevails,
that is, only when in mesh with one gear of the cone can the line of
pressure pass through the axis of the tumbler frame. Fig. 21 shows
this to be the case when C is in mesh with the gear B'. Each subse-
quent shifting of the tumbler along the cone brings the line of pres-
sure eccentric to the axis, until the position of extreme eccentricity is
reached when C is in mesh with B"". In mechanisms of this kind, it
should always be the aim of the designer to have the line of pressure
pass as close to the center of rotation of the tumbler frame as is
possible, because the locking devices used with this type of tumbler
gear are necessarily of such a design as to be quick in action, and in
consequence are not very stiff or rigid. The line of pressure should
always be made to fall on that side of the axis where the tendency
is to separate the gears rather than to bring them closer together.

When the gear C is supported in a swinging frame which does not

 TUMBLER GEAR DESIGN 25

slide in a lateral direction, but the changes are made by shifting C

along an intermediate shaft, the supporting member should be located
at the end where the line of pressure has the greatest eccentricity, as
the greatest strain comes at that end. Thus, in Fig. 21 the support
should be at the same end as B"". The diameter of the intermediate
gear C has an important effect on the location of the line of pressure.
It will be found that it should in most cases be as large as B"", in
order that the line of pressure may come right. However, no exact
rule can be given by which the diameter of G can be calculated, as it
depends greatly on the difference in the diameters of B' and B"", and
also on the diameter of A.

Rules for the Design of Tumbler Gears

What direct rule can be given that may be used as a guide, in laying
out the tumbler gear? Referring to Fig. 19, we see that the gear C
is revolving in a direction away from the axis of the tumbler at the

point of tangency of the pitch circles of C and B, and that the reacting
force tends to crowd the gears farther into mesh. Had this line of
pressure fallen on the other side of the axis of the tumbler frame, the
tendency would have been opposite in effect. When the gear C is
revolving so that a point on the pitch circle travels away from the
pivot of the tumbler, and the line of pressure falls somewhere between
the pivot point and the axis of the driven gear B, the tendency will
be to crowd. From this we therefore may formulate the following
rules:
Rule I. When
the gear about which the tumbler gear swings is the
driver, and the between the axis of that gear
line of pressure falls
and that of the driven gear, the motion of a point on the pitch circle
of the tumbler or intermediate gear, when near the contact point, must
be toward the axis of the tumbler frame. Should the direction of a
point on the pitch circle be opposite, the line of pressure must fall out-
side of that area included between the axis of the pivot gear and the
driven gear.
Referring again to Fig. 19, it is seen that should the driving gear
be B, the above rule does not apply, but may be altered to read thus:
Rule II. When the gear about which the tumbler gear swings is
the driven gear, and the line of pressure falls in the space between
the axis of this gear and that of the driving gear, the motion of a
point on the pitch circle of the intermediate gear at the contact point
must be away from the axis of the pivot gear; when the line of pres-
sure falls outside of this space, this motion must be reversed.
By following these two rules, more as a precaution than as a com-
pulsory condition, much better success may be expected in the results
obtained.
 CHAPTER V

FAULTS OF IRON CASTINGS*

POINTS FOR THE MACHINE TOOL DESIGNER

The most useful and indispensable of all the materials with which
the designer has to do, is cast iron. Of all the metals used in the con-
struction of machinery, it is the cheapest. It is the one to which we
can the most readily give the form and proportions which we desire.
It is, of all the common materials, the most easy to machine. While
it is lacking in strength and ductility, its cheapness makes it pos-

sible to use it in such ample quantity as to overcome these disad-

vantages, and in many constructions it may be so shaped and propor-
tioned, or so reinforced by other materials, as to make this lack but a
slight detriment. It is therefore a matter of interest to the designer
to learn of the various faults to which this valuable material is sub-
ject, and the best ways in which they can be avoided or minimized.
Causes of Blow-holes
Probably the one fault which spoils more castings than any other,
is the result of an outrush of gas from the materials of the cores or
the mold, into the molten iron, at the instant of solidification. If the
solidification of the iron has proceeded so far that the outrushing
gas or steam cannot bubble through it, and escape through the vents
which should be provided for the purpose, it v, ill be imprisoned in the
substance of the casting, forming one or more holes, according to the
special shape of the casting, and the quantity of the escaping gas.
These holes, which are known as blow-holes, may not be apparent on
the outside, and quite often occur in such a location that they do no
particular harm, but it is more often the case that they occur at some
point where they become apparent when the metal is being cleaned, or
where their presence weakens the casting greatly.
Steam from Moisture in Sand
The gases which cause blow-holes may come from three sources.
They may be, and generally are, caused by the generation of quanti-
ties of steam from the moisture contained in the molding sand, by the
heat of the iron. In the case of dry sand and loam castings, the
quantity of steam so generated is so insignificant, if the molds have
been properly heated, that it gives no trouble whatever. In the case
of green sand castings, however, the moisture present, and therefore
the steam generated, is quite large in amount, and special precautions
have to be taken to prevent blow-holes.
When the molten iron pours into a green sand mold, all the moisture
in the layer of sand immediately in contact with the iron will at once
be transformed into steam. The depth of the sand layer so affected
*
MACHINERY, October and November, 1907.
 FAULTS OF IRON CASTINGS 27

depends on the thickness and extent of the fiery mass to which it is

adjacent. The steam, so formed must either force its way through the
molten iron iri the form of a mass of bubbles, or else it must escape
through the sand. To facilitate its escape, the mold is vented. That
is, after the damp sand has been packed around the wooden pattern
by ramming it closely into place, a wire is thrust repeatedly into the
mold, making numerous passages for the escape of the steam and gas.
It is obviously impossible that one of these vent-holes should extend

to every point in the layer of sand adjacent to the casting, so it is

necessary that most of the steam and gas should force its way for
some small distance through the sand, before it can reach a vent-hole.
This it can only do when the sand is somewhat porous. If the sand is
too tightly rammed, it will lack the necessary porosity, and even though
it be unusually dry, and the venting carefully done, the casting will be

full of blow-holes. Cases have occurred where molds have been

rammed so hard that the castings were nothing better than shells, the
whole interior being a mass of blow-holes.
Decomposition of Binder in Cores, and Entrapping- of Air
The second cause of blow-holes in iron castings is the decomposition
of the material, generally flour or molasses, used as a binder in pre-
paring the cores, and its escape in the form of gas, into the iron, at
the instant of pouring. It is impossible to prepare and bake a core in
such a way that it will not give off large quantities of gas when the
iron is poured, and so means must be provided for the escape of this
gas. In order to do this, the cores are prepared with wax strips run-
ning through them. When the core is baked, the wax melts, leaving
passages, known as core vents, for the escape of these gases. If the
core is of such form, and so set in the mold, that the gases can
escape from these vents in an upward or sidewise direction, and leave
the mold without forcing their way through the molten iron, no blow-
holes will result.
A third source of blow-holes is the entrapping of air in certain parts
of the mold, and its mixing, on expansion, with the iron. This trouble
is due to insufficient venting of the mold, and is not a fault to which

the designer need pay any particular attention.

Dry Sand or Loam Advisable for Large Complicated Castings

In the case of large and complicated castings, it is generally advis-
able to make dry sand or loam castings, in order to avoid, as far as
possible the chance of blow-holes. When the mold is very large, it is
difficult to vent it thoroughly, and when the work on it extends over
a period of three or four weeks, it is impossible to keep the vents
from filling up; hence the general use of dry sand work for large
castings. Often, however, for the sake of economy, fairly large and
complicated pieces must be undertaken in green sand, and it becomes
a matter of importance that they be so designed that the molder will
not be compelled to invite disaster by keeping his sand too wet, or
ramming it too hard, and that there be no part of the mold which may
not be thoroughly vented.
 T
28 A o. 14 MACHINE TOOL DESIGN
Elements of Green Sand Molding-
In order that we may understand thoroughly the effect of the design
of a casting on the probability of blow-holes, it is necessary that we
review, in a brief way, the elements of green sand molding. The sand
is sprinkled with water, and thoroughly mixed and sifted, preparatory
to packing or "ramming" it around the pattern. The object of wet-
ting the sand is of course to cause it to stick when it is packed. Up
to a certain point, the wetter it is, the better it will stick, but the
molder should not wet any more than is necessary. In the same
it

way, the more tightly the sand is rammed, the better its particles will

cohere, and the more easily will the mold be handled, and the pattern
drawn. However, tight ramming and w et sand, while they make a
r

solid and easily handled mold, invariably produce blow holes, and are
therefore to be avoided.
It will be apparent then, that if a pattern be of complicated form,
or hard to draw, or if when it is drawn it leaves the sand in such a
form that the mold will easily fall together at a little jarring, the
molder will be. compelled to wet the sand more and to ram it harder
than usual. Small, deep openings, sharp fillets, and thin walls and par-
titions of sand, are especially troublesome. Not only do they make it
difficult to draw the pattern, and handle the mold, and so make exces-
sive wetting and hard ramming imperative, but they cause spots in
the mold which the venting wire is unlikely to reach. For these
reasons, they are to be avoided when possible, in any class of molding,
whether it be green sand, dry sand, or loam work, and on no account
should such work be permitted in the case of large green sand castings.
When designing a casting to be made in green sand, the designer
ought to know the position which it will occupy in the mold, when it
is poured. In general, the parts of a casting which lie in the bottom
of the mold will be the soundest, and those parts which must be
machined, or which require the greatest strength, should therefore
occupy the bottom of the mold, if possible, when the casting is poured.
Having decided which side will be down, the designer needs generally
to pay no particular attention to the configuration of the lower part
of the mold, provided only that all of the pattern can be drawn, and
that there are no sand partitions which overhang, or whose extent is
large in proportion to their thickness. To insure a sound casting, the
sand in the lower parts of the mold must be comparatively dry, and
loosely rammed. This condition of affairs is not generally hard to
attain, since all the work on the sand is done with the pattern in
place, and that part of the mold is not generally moved or handled
after the support of the pattern has been withdrawn. In the lower
part of the mold, the sand is generally supported at all points in a
very thorough manner by the sand lying under it, and so hard ramming
or wet sand is unnecessary. If, however, the pattern must be made
with loose pieces, or with sharp fillets, or must leave thin walls or
tongues of sand when it is withdrawn, the case is changed. Then
hard ramming and wet sand are almost compulsory, and the molder
 FAULTS OF IRON CASTINGS 29

is not to be blamed if he does not produce sound green sand castings.

The fault is with the designer.

The upper part of the mold must of necessity be rammed harder
than the lower part, since the sand is not supported from beneath, but
hangs from above. This is not as great a disadvantage as it might
seem to be at first sight, since the escaping gases do not have to
make their w ay through the iron, as they would if they were given
r

off by the sand in the lower part of the mold. The venting, how-
ever, must be just as thorough, and it is desirable that the sand should
be as dry as possible. The whole arrangement of the upper part of
the casting should be such that the sand may be well supported from
above. Generously rounded fillets and corners, simple surfaces, plenty
of "draft," and an absence of depending walls and masses of sand,
make the mold easy to handle, and therefore promote freedom from
blow-holes.

When Green Sand and Dry Sand Both May Be Used

It often occurs that the larger part of a casting is of simple form,
and easy to mold. A certain part of it, however, may be of a form
exceedingly difficult to mold, and therefore likely to give a great deal
of trouble. It is not necessary that the whole casting should be made
in a dry sand mold, but a core-box may be made to take care of the
difficult part of the work, even though the work would ordinarily be
done without a core. It is just as easy, and often just as desirable to
cast the external face of a casting against a core, as the internal face.
While it may not pay to do this if only one casting is wanted, if a great
many are wanted it is often the cheapest possible way of making them,
and reduces to a minimum both the work of the molder and the chance
of a spoiled casting. Often forms may be cast in this way which could
not be attempted in any other. If it is desirable to use this method of
working, the designer has it in his power to make the construction of
the core-box much simpler and cheaper than it might otherwise be, by
giving the matter a little thought.
In arranging the coring of a mold, it is always better, if possible,
to support the cores at the top. The gases formed in the core, being
light,tend to rise, and if the core is supported at the bottom only,
they tend to escape into the iron, and to bubble through it. If they
can escape at the top, they will pass off without coming in contact
with the iron. When it is impossible to support the cores at the top,
they should be so arranged that the gases can pass off at the sides,
and escape from the mold without coming in contact with the iron.
Sponginess
A second fault to which iron castings are subject is that of spongi-
ness. Sponginess is due to the formation of gas bubbles in iron, at
the instant of solidification. In all ordinary cases this is due to an
improper mixture of iron. However, if the casting is very thick at
one place, and thin at most others, it will be impossible to obtain a
mixture which will have satisfactory properties for general work, and
not be spongy at points of extraordinary thickness. It is an excellent
30 No. 14 MACHINE TOOL DESIGN
rule to allow no part of a casting to be at a greater distance from
a sand surface than 2% inches. In case this rule is strictly adhered
to, and the castings are of fairly uniform thickness no trouble will
be experienced from sponginess, except from the use of poor iron.
When, however, we are obliged to concentrate a considerable quantity
of metal at one place, and give it a greater thickness than 5 or 6
inches, either we must take care that it will be at some point where
the sponginess will do no harm, or else we must make provision to do
away with it.
The only practical method for doing this is to place a riser immedi-
ately over the heavy spot. When the metal is poured, and the riser
is full, a rod of wrought iron is inserted and worked up and down

until the metal has almost solidified. By so doing, the bubbles have
a better chance to escape, and the iron is left perfectly solid. Of
course, it is not possible to use a riser effectively in this manner,
unless it can be placed directly over the heavy spot. A riser at a
point a few inches distant is useless. The use of a riser in this way,
and for this purpose, is unnecessary when the part of the casting in
which the heavy spot occurs is subject to no particular stress, or is
not required to be tight under steam, air, or hydraulic pressure, but
nevertheless, a spongy spot is a defect in a casting, which should, if
possible, be avoided.
Shrink-holes
A third fault to which iron castings are subject is that of shrink-
holes. A shrink-hole is a cavity in a casting caused by the shrinking
away of the metal in cooling. Like sponginess, this defect is espe-
cially likely to occur in those parts of a casting which are excessively
thick. To avoid this fault, it is best to avoid sudden changes in the
thickness of a section. If the part of a casting which is unusually
thick does not have to be machined, the difficulty may be overcome
by placing in the mold at that point a piece of iron, so that the casting
will be caused to solidify at that point first, on account of the chilling
effect of the cold iron. If, however, the heavy spot in the casting has
to be machined, or if it is subject to heavy stress, this method of pre-
venting shrink-holes is to be avoided, since the chilling of the iron
makes it so hard as to be impossible to cut with a tool, and at the
same time creates stresses within the metal which weaken it. In such
a case, shrink-holes are best prevented in the same manner as has
already been described for the prevention of sponginess, namely, the
use of a heavy riser, and the working of the iron with a rod when it

is cooling.
The designer must therefore avoid heavy spots in castings, when-
ever possible, for the reason that they are likely to produce two serious
faults, sponginess and shrink-holes. He must avoid them especially
in those parts of castings which are to be machined or which are sub-
ject to heavy stresses. If they cannot be avoided entirely, in such a

case, they should be so arranged that risers may be placed immedi-

ately over them, so that a rod may be inserted into the riser, and
into the heart of the spot where the metal is thickest.
 FAULTS OF IRON CASTINGS 31

Scabbiness
A fourth fault often encountered in iron castings is that of scabbi-
ness. Although iron in the molten condition does not permeate the
sand of the mold, as water would if it were poured in, nevertheless,
on account of the great weight of the fluid, it has a considerable
erosive action on the materials of the mold. If, as it flows into the
mold, the iron eats away fillets or partitions, or scours away patches
of sand, it is obvious that the casting will not be of the proper form,
but will have its angles partly filled up, and unsightly protuberances
upon its surfaces. Such imperfections as these are known as scabs.
The sand so washed from its proper place may float on the iron, and
rise to the top of the mold, where it forms a dirty mixture, which,
when cleaned away, leaves a rough depression in the surface of the
known as a scab.
casting, also
The remedy for this trouble
.
is to avoid as far as possible sharp

fillets, and thin tongues

of sand, projecting into the mold, and to so
gate the casting that the current of iron, as it enters the mold, will
spread itself out, and not concentrate itself in any particular direc-
tion, for if it does, it will eat away the part against which it flows,
just as quickly and surely as would a current of water. In general,
proper gating is a matter which must be attended to by the molder, but
if the designer has arranged things so that proper gating is incon-

venient or impossible, the castings will almost surely be scabby.

Sand-holes
The fifth fault to which iron castings are subject, namely sand-
holes, is one which is almost invariably associated with that of scab-
biness. If the sand which has been eroded by the entering current of
iron does not rise immediately to the surface, the iron may partially
solidify before it will float to the top. As a result, it will rise till it
strikes a part of the iron which has so solidified, and will remain there,
imprisoned within the body of the casting. Sand-holes generally occur
in that part of a casting which lies near the top of the mold, and just
a little ways under the skin. They may occasionally form large cavi-
ties which seriously impair the strength of the casting, but more often
they form very small holes, which, being full of sand, destroy the edge
of any tool which may be used for the purpose of machining the cast-
ing, and leave the finished surface pitted and unsightly.
From this description of their origin, it must be apparent that the
cure for sand-holes, as far as the designer's work is concerned, is the
same as that for scabbiness. The fault may also be avoided by the
use of a. riser, so arranged that the current of iron will sweep the loose
sand out of the mold and into the riser, where it will do no harm; but
while this remedy eliminates sand-holes, it does nothing to remedy
scabbiness, which is generally the cause of sand-holes.

Floating Cores
1

A sixth difficulty often encountered in the production of sound iron

castings, is caused by floating cores. The buoyant effect of the molten
-32 No. 14 MACHINE TOOL DESIGN
iron on a core equal to about three times the weight of the core,
is
if it is solid, and very much more than that
in case the core is hollow.
Large cores are generally built-up about cast-iron skeletons known
as core frames. These core frames are roughly of the same shape as
the core, and serve to support it, and to bind it together. Were it
not for these frames, heavy cores would fall to pieces by their own
weight, and would be broken up in the process of casting, owing to
the buoyant effect of the iron. A projecting piece of core having a
volume of four cubic feet, for instance, will weigh approximately 500
pounds, and have a buoyant force of about 1,500 pounds thrusting it
upward when the mold is poured. If the core frame is not amply
strong and stiff, this force will bend the projection upward, or even
break it off entirely. Hence the necessity of making large cored cavi-
ties of such form that the cores may be rigidly anchored and thor-
oughly secured. Nor is it sufficient that provision be made for secur-
ing the cores, but they must be of such shape, and so reinforced by
the frames, that they will be stiff and strong, and not bend appre-
ciably under the tremendous forces which will act on them when they
are surrounded by the molten iron.
One of the most difficult things to cast properly is a long iron pipe
having a small diameter and thin walls. If such a pipe be cast in
the usual position, that is, lying horizontally, there will be an upward
thrust along the whole length of the core, tending to bend it. On
account of its slenderness, the central portions will be deflected up-
ward, making the walls of the pipe thinner on one side, and thicker
on the other. Often the deflection proves sufficient to thrust the core
against the side of the mold, if it is long, or, in case the thickness
of the wall is great in proportion to the length of the core, to break
it off entirely. On this account, pipes and hollow columns of cast
iron are often cast on end, thus avoiding any deflection of the core.
The same principle may be applied to many other pieces, by taking
care to so design them that long and slender cores shall have a vertical
position when the mold poured. If they have such a vertical posi-
is

tion, and are supported at both ends, they will have no tendency to
deflect one way or another, and this source of trouble may be com-
pletely avoided.
Cold Shuts
The seventh fault is as a cold shut. A cold shut is caused
known
by the imperfect uniting of two or more streams of molten iron, flow-
ing together, which are too cold to coalesce. Such a fault often occurs
on the upper side of a thin cylinder cast horizontally, when the iron
is not sufficiently hot at the instant of pouring. It there appears as
a seam in the side of the cylinder, where it is very apparent that the
metal has united imperfectly. It is not only a weak spot In the cast-
ing, along which it will readily split if called upon to sustain any
great stress, but it is a spot which will surely leak under pressure,
and which it is impossible to calk. The cause of the imperfection is
generally improper gating, or else too great thinness of metal. If the
iron is obliged to flow in thin streams for long distances, it will be
 FAULTS OF IRON CASTINGS 33

cooled very much, and probably the advancing face will be partially
solidified.Consequently, when it meets a similar advancing face of
metal, which has been similarly cooled, there is small likelihood of
their uniting properly.
The remedy is obviously to so design the casting that the metal
will not have to flow in thin streams for long distances. The arrange-
ment of gates and risers is often of great importance in minimizing
cold shuts, and if the casting is large, and at the same time has thin
walls, the designer must see that the gates may be so arranged that
the iron may quickly fill up the mold. While the arrangement of the
gating generally depends on the molder's fancy, he may often be lim-
ited by the shape of the casting, and obliged to place the gate at some
point where the iron, in flowing in, must spread itself into a thin
sheet, or pass for a considerable distance through a narrow passage.
Under such circumstances, a cold shut is hardly to be avoided.
Shrinkage Strains
The eighth and last fault is that of shrinkage strain. If we have
two pieces of iron fastened end to end, as shown in Fig. 22, one piece
being notably thinner than the other, the thinner piece will solidify
the mold, and cool some hundreds of degrees below its freezing
first in

Machine -y,N.r.
Fig. 22. Casting of such Shape as to be Subjected to Severe Internal Stresses

point, before the thicker part solidifies. As a result, the thicker part,
when cooled to air temperature, will have, or rather tend to have, a
less length than the thinner part, the reason being that at the instant
of solidification of the thicker part, both pieces had the same length,
although the thinner part was much the cooler. The thin part will
then be in compression, while the thick part is in tension, and severe
stresses will exist within the piece, which make it weaker than it
would otherwise be in most cases.
Sometimes, however, we are enabled to utilize the shrinkage stresses
to advantage. For instance, when cast iron was the standard material
for the manufacture of ordnance, guns were cast with cores through
which water was circulated, so as to cool the surface of the bore
before the outer parts solidified. When a gun is fired, it is known that
the inner layers of metal are stretched more than the outer ones. By
cooling the inner layers of metal first, shrinkage strains are produced
in the walls of the gun, causing the outer layers of metal to com-
press the inner ones. The combined effect of the shrinkage stresses
and the stresses produced by the explosion is to produce a uniform
stress throughout the walls of the guns, and so reduce the chance of
rupture.
It is not often, however, that we are able to take advantage of

shrinkage strains in this way. More often they are troublesome, caus-
34 No. 14 MACHINE TOOL DESIGN
ing work to warp in the process of machining, or causing mysterious
cracks to develop without apparent cause. Since these strains are due
to unequal rates of cooling in the different parts of the casting, the
best way to eliminate them is to so arrange the thickness of the vari-
ous parts, that the entire casting shall solidify at the same time. The
second best way is to so arrange the parts of the casting that the
unequal contraction shall not produce dangerous stresses at any point.
In order that the entire casting shall cool at a uniform rate, it is
necessary that all parts of it shall be of approximately uniform thick-
ness, and that there shall be no sudden changes of section. In order
that unequal contraction shall not produce dangerous stresses in the
metal, it is necessary that there shall be no sharp corners, and that the
various parts shall be free to expand when necessary. For instance,
a wheel or pulley with a solid rim is likely to have severe stresses

Machinery. A'. T.

Fig. 23. Method of Obviating Shrinkage Strains in Large Wheels

set up within the arms by unequal cooling, but if the hub be divided
as shownin Pig. 23, by means of a thin core, and then bolted subse-
quently, no shrinkage strains will occur, since the arms are free to
expand or contract, independently of the rim.
Shrinkage strains often become so serious that it becomes neces-
sary to make pieces in two or more parts, which it would be perfectly
possible to make, at much less expense, in one piece. Large jacketed
cylinders, for steam and gas engines, are good examples of this. When
cast in one piece, the shrinkage stresses, together with the stresses set
up by the varying temperatures incident to services, are often suffi-
cient to crack them. Were the piece shown in Fig. 22 made in two
parts, as shown in Fig. 24, there would be no shrinkage strains in
either part, although the cost of machining the surfaces which are
 FAULTS OF IRON CASTINGS
fitted together, and of putting in the bolts, would not always warrant
the construction.

Relative Economy of Simple and Complicated Castings

In conclusion, it may be well to state that most of the faults
enumerated will be more
likely to occur in a part of a complicated
casting, than in a similar part of a simpler casting. For instance, the
cylinder of a gas engine will be more likely to have some imperfec-
tion if it is cast integral with the frame, than if it is cast separately.

f
Machinery. N.r.
Fig. 24. Piece shown in Pig. 22, made in Two Parts

In the same way, the frame will be more likely to have an imper-
fection of some kind, than if it were cast separately. Assuming that
ten per cent of the cylinders or frames* would be lost if they were
cast separately, it is more than likely that fifteen per cent of the cast-
ings, having cylinder and frame cast together, would be rejected for
faults in the frame, and fifteen per cent of the remainder would be re-
jected for faults in the cylinder. In other words, twenty-eight per cent
of these castings would be rejected, against ten per cent of the simpler
forms. If more than eighteen per cent of the cost of the castings is
saved in machining, or in other ways, by casting cylinder and frame
together, it is well to do so, but if the saving is not more than sufficient
to balance the loss, it is well to make several simple forms, instead of
one complicated one.
 CHAPTER VI

PROPORTIONS OF MACHINES BUILT IN A

SERIES OP SIZES*
The problem of cost reduction forces itself, with increasing vivid-
ness upon the mind of every person who has to do with the manu-
facture of machinery. To the "small shop" people, and to those whose
product is unsystematized and whose ideas of methods to pursue are,
as yet, vague, this chapter may prove of some assistance.
*
There are three important means by which the shop product may be
systematized: By the use of formulas; by the use of tables; and by
the use of charts. As the two latter may be considered as the tabu-
lated, or graphic results of the former, we will deal only with the
formulas. In determining sizes, weights, and costs, these formulas
are generally most efficient time-savers. For convenience, formulas
in this chapter will be divided into two classes: The class used to
produce the first of a type of machine we will call fundamental; and
the class used to produce several sizes of this type of machine, empiri-
cal. Upon seeking fundamental formulas in text books and in me-
chanical engineers' pocket-books we are confronted by a diversity of
opinions and tabulated results that are, at least to a novice, a bit con-
fusing. These formulas, it is always to be remembered, have their
application in the special case under consideration, and are to be used
only as guide posts in our journey of design. It is evident to most
designers that some kind of a tentative method must, sooner or later,
be resorted to in the type design, for in nearly all machines the gov-
erning conditions soon become so numerous or indefinite as to render
a subdivision of the problem a necessity. A certain amount of judg-
ment is absolutely essential in the use of most fundamental formulas,
and discrimination is always necessary.
Graphically, fundamental formulas can be represented by curves, and
will be correct for all sizes under identical conditions, while empirical
formulas rest on no such basis and hold true for but a certain series
within certain limits. This constitutes the vital difference between
fundamental and empirical formulas. A fundamental formula is one
found through mathematical reasoning, while an empirical formula is
made up by means of trial methods.
Suppose that we have built two or three sizes of a certain type of
engine and that they are successful; we desire to put on the market
an entire line. Our sizes of this type of engine will- run from 10-inch
cylinder diameter in the smallest to 30-inch cylinder diameter in the
largest. We have built a 12-inch and a 24-inch engine and perhaps
an 18-inch. These engines were, as was imperative, tentatively de-
signed. In seeking the derivation of the empirical formula for the
*
MACHINERY, November, 1902.
 PROPORTIONS OF MACHINES IN SERIES 3?

length of the cross-head shoe, we find that on our 12-inch engine we

have given it an area of 55% square inches, and on our 24 inch engine
its area is 190% square inches. In each case the length of the shoe
was nearly twice its width, so we decide to make it so in our line
of engines; solving for the width, we have in the 24-inch engine
2x 2 = 190.125; = 1/95.0625 =
a? 9.75.

making our shoe length inches, and for the

for the 24-inch engine 19%
12-inch engine 10% inches.
To any scale in Pig. 25, perpendicular to the line NL, lay off these
shoe lengths PB and P'B' 10% inches, and 19% inches, respectively-
making the distance BE' equal to 12 inches, the difference between our
sizes 12 inches and 24 inches. Through points PP draw line 8A inter-
f

secting NL at A. At B" for our 18-inch size erect a perpendicular

5"P". Draw PF intersecting E"P" at F. Using the notation given in
the figure, we get the simple equations

Machinerv,N.T.

In this last formula many will recognize an old acquaintance the

equation for a straight line. Let us now analyze this equation. From
c
the figures it is seen that x = ih.e desired dimension and that = the
d
rate of increase in the slope of the line. If now we measure the
distances and substitute their values for c and d we may determine the
c
ratio , which we will call /.
d
c 10.5
Then / = = ==%, and
d 14
x = fy + c, or x = %y + c.

In interpreting our empirical formula x fy -f c, we have =

y = a common unit to which all other sizes are to be referred,
a? desired dimension,
38 No. 14 MACHINE TOOL DESIGN
f = a factor ofy,
c = a constant increment to be added in each case.
The unit of value y, is a bolt or cylinder diam-
as generally selected,
eter, or the capacity of themachine. Obviously, in our line of engines,
we select the cylinder diameter D as our value of y, and our unit
formula then becomes %D + c. The value of c is now determined by
direct substitution in the following manner: x being the shoe length,
we substitute for it 19% inches (its length on the 24-inch cross-head);
then
% X 24 + c = 19V
2 ;
c = i$y
2 18 =

Machinery, .V.F.
Fig. 26

Note, that while we have assigned to y and c other values, we have

not altered the relations; our formula for this particular cross-head
dimension now becomes %D + 1% inches.
For convenience in charting these sizes, some point is determined
upon as a pole about which these lines (represented by our formulas)
are drawn as vectors, the ordinate length for a particular size giving
the desired dimension. If now in the determination of other formulas
it be found, as is likely to be the case, that these lines do not all pass

24 26
Machinery, N.Y.

through a common point, it becomes necessary to select one. In well-

designed machines the intersection of these lines with the base line
will come close together, and an average of these intersections is
selected as a pole. Figs. 26 and 27 will serve to illustrate the purport
of this paragraph.
Experienced designers are well aware that the final test of any
dimension in a design is that of satisfying all fundamental calculable
 PROPORTIONS OF MACHINES 7A 'SERIES 39

conditions; nevertheless, the instances where our empirical* formulas

prove incorrect are very few indeed. With the design for our line of
engines thus systematized, let us consider what are to be the advan-
tages that will naturally result from it. In the first place, the weights
of any particular parts, or details of any size in our line of engines
may be determined prior to its design or manufacture. In the deter-
mination of weights, cubic contents, and similar processes, the use of
"differences" as applied to higher mathematics, will not only prove
an efficient time-saver, but relieve much of the drudgery attendant
upon such operations.
A brief explanation of the use of "differences" is as follows: When
we have a series of numbers connected by a regular, though not obvious
law, the nature of that law may be discovered by forming a new series
of differences between each two terms of the original series, and then

TABLE SHOWING PRINCIPLE OF THE METHOD OF DIFFERENCES

Arithmetic Solution. Algebraic Solution.

Col. Col. Col. Col. 1. Col. 5. Original Ser. Col.6. 1st Ser. Col,
|Term Col. 7. 24
4.

5
W Differences.
Ser. Dif,
8.

3rd
Dif.
12
~T y+x C
20 x+c
la" y+2x+c C
32
TT
49
x+3c
^3" c
72
"30" 2/ + 5,r flOc c
102
"38" c
140
If c
10 187 9
a?-|-7c

IT
"57"

"ST
244 u y+Sx+2Sc

312

treating tne new series (which we may call the series of first differ-
ences) in the same way, until we reach a series of differences, the law
of which is obvious. In the table above will be found both the arith-
metic and algebraic solutions of problems by "differences."
In column 1 of the table is given a series- of numbers, which we
suspect follows some definite, though not obvious law, and which we
desire to discover. We here take the differences between each two
terms in column 1 and put them down in column 2. Having proceeded
with the two orders of differences, the law becomes apparent early in
the process of determining the values in column 3. Referring again
to the table, it is evident that the next term of column 3 must be 11,
which gives 68 (57 + 11) as the next term of column 2 (the series
of first differences) and 312 (244 + 68) for the original series. Note
that this series can thus be obtained indefinitely, and that ultimately,
40 No. 14 MACHINE TOOL DESIGN
in any regular series, some one series of differences will become a
constant. on the principle of differences that calculating ma-
It is

chines are constructed to compute logarithmic tables, etc.

In the algebraic solution of such problems as involve the determina-
tion of weights and volumes, it will be necessary to calculate these
weights or volumes for the 1st, 2d, 3d, and 4th terms of our given
series. By substiting in the formulas in column 5 the numerical
value of W, which is the first term in our given series, we may equate
these expressions and our calculated values for the 2d, 3d, and 4th
terms, and determine, by simple algebraic processes, the values of
c, x, y, and ultimately, those values which we are requiring in the
original series, column 5.
The computations concerning the cost of materials logically follow
the determination of volumes and weights and are made with compara-
tive ease. However, our next problem concerning the determination
of the cost of labor is a more difficult one to solve. Formulas should
express this cost in so many cents per pound of product, including all
shop charges, and be established partially by experience and partially
by methods suggested in this chapter.
In many instances it will be found both desirable and convenient to
have this cost formula embody the unit dimension. When this is the
case, the formula is, as are most cost formulas, established by the
tentative methods to which we have just alluded. As the methods
employed in the deduction of these formulas render them purely
empirical, one or another form of expression may have to be, adopted.
However, formulas of this class usually assume the form of, or at least
may be solved into, the familiar type form
ax2 60? + c,

where a and 6 are factors of the unit dimension, and c is a constant.

For the purposes of illustration we will assume that the formula
for the cost of labor which we have established is
3
D 2
15D + 314.
2

In this form the formula gives the total cost of labor in dollars for
the size desired. The cost of labor C for our 18-inch engine would
then be
3
C = (18)
2
15 X 18 +314 = 530.
2

The computation of a final cost formula, embodying the unit dimen-

sion, is the last process in our development of shop formulas; this
formula is derived directly from those relative to the costs of material
and labor.
 RETURN TO the circulation desk of any
University of California Library
or to the

NORTHERN REGIONAL LIBRARY FACILITY

Bldg. 400, Richmond Field Station

University of California
Richmond, CA 94804-4698

ALL BOOKS MAY BE RECALLED AFTER 7 DAYS

2-month loans may be renewed by calling
(510)642-6753
1-year loans may be recharged by bringing
books to NRLF
Renewals and recharges may be made
4 days prior to due date

DUE AS STAMPED BELOW

JUL 3 2004

AUG 1 2004

DD20 15M 4-02

 YC 53944

UNIVERSITY OF CALIFORNIA LIBRARY