M. C.

Shaw The Rating of Abrasive Cutoff Wheels

Professor and Head,
Department of Mechanical Engineering, A rational approach to cutoff wheel selection is presented based on the most important
Carnegie-Mellon University,
Pittsburgh, Pa. Fellow ASME
criterion for this operation—minimum cost per cut. New parameters are identified which
play a more significant role than the conventional quantities grain size, grade (hard-
ness), and structure number normally used to characterize a wheel. The new quantities
are: a) the effective number of cutting points per square inch on the wheel periphery
(substitute for grain size); b) the volume that may be accommodated between success-
sive grains under cost optimum conditions (substitute for structure number); c) a chip
flexibility parameter (no counterpart in conventional wheel signature); d) the downfeed
corresponding to a grinding ratio of unity (this reflects wheel grade and resistance to
wear). A number of experimental examples are presented to illustrate the application of
the procedures proposed.

Introduction be realized it is important that wheel selection and the prediction

of wheel performance be put on a scientific basis. A single method
At the present time there appears to be no rational approach to should not be expected to work for all grinding operations, but
the selection of grinding wheels and of predicting their perfor- important classes of grinding will have to be identified and the
mance. Wheels are initially selected based on experience and in- procedure worked out for each class. However, the method of pro-
tuition. A satisfactory wheel is then arrived at in the workshop by cedure to be followed is quite clear from our experience in the ball
a process of trial. bearing industry. The mechanism of greatest importance to life
This was the situation in the early days of the ball bearing in- (surface fatigue in the case of the ball bearing) must first be iden-
dustry. However, bearings are selected today based on load rating tified and a mathematical design equation based on this concept
tables which enable the expected life to be predicted with a prob- established. This relationship may then be applied in the rational
ability of failure that is usually taken to be 10 percent. Means for approach to selection and the prediction of performance.
including the influence of bearing configuration (type, number of The rating procedure to be considered here in the first instance
balls, retainer design, etc.), speed, element in rotation (outer or pertains to abrasive cutoff wheels.
inner race or both), type of loading (constant, intermittent, The economics and mechanics of the abrasive cutoff operation
shock, etc.), method of lubrication, and physical environment are have been presented in two recent papers [1, 2]. 1 Items presented
included in the load rating process. This is all made possible be- in these papers that are pertinent to the present discussion are
cause bearings normally fail as a result of a single definite mecha- reviewed here.
nism called surface fatigue. As soon as this single mechanism re-
sponsible for practically all bearing failures had been identified T h e A b r a s i v e CutOff O p e r a t i o n
and the characteristics of bearings in terms of this mechanism
The single most important item in the abrasive cutoff operation
were understood, it became possible to put bearing selection on a
is the cost per cut. Since wheel wear has an important bearing on
rational basis in terms of rating tables.
cost it is of major concern. Surface finish, dimensional accuracy,
The readily available rational approach to ball bearing selec- or surface integrity which are frequently of major importance for
tion and the prediction of their performance is largely responsible other operations are relatively unimportant here.
for the wide use of ball bearings today and the fact that these Wheel wear is conveniently expressed in terms of the grinding
bearings have competed so successfully with other bearing types ratio (G) where
that are actually capable of longer life.
If the full potential of grinding as a metal removal method is to v o l u m e of w o r k r e m o v e d
G =
v o l u m e of w h e e l c o n s u m e d
Normally, cutoff wheels are operated at very high speeds—the
higher the better. A speed of 12,500 fpm is quite common and
Contributed by the Design Engineering Division of The American Soci- there is a trend to increase this. A second variable of importance
ety of Mechanical Engineers for publication in JOURNAL OF ENGINEER-
ING FOR INDUSTRY. Manuscript received by the Design Engineering Divi-
1
sion October 2,1973. Paper No. 74-DE-F. Numbers in brackets designate References at end of paper.

138 / FEBRUARY 1975 Transactions of the ASME

Copyright © 1975 by ASME
Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jmsefk/27618/ on 03/24/2017 Terms of Use: http://www.asme.o
 CD

o*
Ga-

o
z
a:
CD

Fig. 1 Schematic diagram of the abrasive cutoff operation

is the downfeed of the wheel (d, ipm). The abrasive cutoff opera- DOWNFEED, d
tion is shown diagrammatically in Fig. 1. A third variable of im- Fig. 2 Variation of grinding rates (G) with downfeed rate (d) for the
portance shown here is the length of cut (t). abrasive cutoff operation
When a cutoff wheel is operated at constant speed but at dif-
ferent values of downfeed, the grinding ratio is found to vary as
shown in Fig. 2. This curve is seen to have two branches and a
maximum point. A detailed study of the cutoff process has re-
vealed that the controlling mechanism for the left-hand branch is
C* = grains/in 2 I = work length
wheel face temperature. When the downfeed (d) is relatively low,
Ci = optimum vol/chip y _ wheel cost/cu in.
there is too much time for heat to flow into the work, and the sur-
face temperature of the wheel becomes excessive, and hence G is do = chip flexibility parameter x machine and labor cost/min
low. As d increases, G increases until a second mechanism takes di = value of d when G = 1 b = wheel thickness
over—namely chip crowding. For points on the right-hand branch The downfeed for minimum cost may be found by differentiat-
of the curve, there is insufficient space between adjacent grains to ing equation (2) relative to (d) and equating to zero. When this is
accommodate the chips generated. done it is found that
A s d increases, the volume per chip increases, and hence G de-
creases. A{y)b „ Hy)
G* d* bd* (3)
Operation on the left-hand branch of Fig. 2 should be avoided, t(x )h n(x)
due to danger of overheating the workpiece. Also, operation on
Substituting this value of G into equation (2) and simplifying
this branch of the curve is not economically attractive. Wheel
gives
consumption will be least at the point where G is a maximum
(M) and at first thought, this may be considered the best condi-
tion for operation from the point of view of cost. However, this
£1 (1 + n) (4)
h d*
would be true only if labor cost were insignificant. Since the labor
component of cost decreases with the downfeed rate (d), the cost It is thus evident that we should always operate on the right-
optimum point will always lie to the right of maximum point hand branch of Fig. 2, since this provides cuts of minimum cost.
(M). Thus, we should always attempt to operate on the right- The controlling mechanism for this portion of the curve is chip
hand branch of the curve and preferably at the value of (d) which crowding and this plays the same role as surface fatigue in ball
makes the total cost per cut a minimum. This value of downfeed tearing technology. At the cost optimum point (d*) there will be
is designated (d*) and the corresponding value of grinding ratio insufficient space between grains to accommodate the chips. It is
as(G*). therefore important that the mean volume per chip be consid-
For a given wheel, wheel speed (V) and length of cut (l), G is ered.
found to vary linearly on a log-log plot with d, in the practical re- The volume of metal removed per unit time will be
gion of operation. Thus, the general wheel wear equation may be V o l / t i m e = Ibd (5)
written
where b is the wheel width and the number of chips generated per
dG" = Ci (1) unit time will be
where n and Ci are constants. c h i p s / t i m e = VbC (6)
In reference [1] the economics of abrasive cutoff are thoroughly
discussed and it is shown that the cost per unit depth of cut will where C is the number of cutting points per square in. of wheel
be as follows for a wheel operating under constant conditions ex- surface.
cept for downfeed (d): Therefore, the mean volume per chip will be
0 x Ayb Vol/chip = —
(2) (7)
~h = ~d +
~~Gt
where When operating at the cost optimum downfeed (d*) the volume
= cost per cut, cents
it per chip will be (id*/VC*) and since chip crowding is the con-
h = depth of cut, in. trolling mechanism in the practical region of abrasive cutoff, we
x = cost of machine per minute including operator and over- might expect this to be a constant for all values o f t . Thus, the
head, cents per min design equation relating optimum downfeed rate (d*) and (I) for
d = downfeed rate, ipm a given wheel operating at a given speed (V) may be taken to be
A = workpiece area, in. 2
y = mean value of cut off wheel per unit volume, cents/cu in. id*
2 (8)
b = width of wheel, in. vc* ~
G = grinding ratio, dimensioriless where Ci is a constant equal to the optimum volume per chip.

Journal of Engineering for Industry FEBRUARY 1975 / 139

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jmsefk/27618/ on 03/24/2017 Terms of Use: http://www.asme.o

 60 of I when the downfeed (d) is very large may at first seem like a
i r strange result. However, when d is very large the cutting distance
40 to cause chip crowding will be small compared with£ . Hence,^
should have little to do with wear when d is very large. Thus, for
large d we should expect G to have the same value for all values
of I (or n since n is related to I).. Referring to equation (1), it is ev-
20 ident that this can be so only when G = 1. It is then clear that by
large d we mean a value for which G = 1. This value we call di,
(S which in turn is equal to Ci.
o" In practice, values of G that are less than one are sometimes ob-
served in abrasive cutoff. This may occur when a mechanism
z• other than chip crowding is controlling and causing premature
grain loss. However, this is an abnormal situation and a cause for
z 6 y/x =0.5-
1 the faulty wheel behavior should be evident upon investigation.
O Equation (1) may be written as follows when evaluated for the
4 cost optimum case
d*G*n =zCi = di (9)
The inverse slope of the wear curve (n) may be expressed as fol-
lows, since G = 1 when d = di
log (dt/d*)

n
(10)
.1 / I /I I log G*
8 10 t 20 40 60 180 100
Values of the cost optimum points were next determined for
d 0 = 5 . 0 d 0 = 7.5 d 0 =l4 d, = 76 each value of t by solving the following equations simultaneously.
DOWNFEED, d(ipm)
dG" = C t = d\ (equation 1)
Fig. 3 Actual values of grinding ratio (G) versus downfeed for work-
pieces of different length U). Wheel: A243-R6B (24 grain size), 20 X G* = -(2-)bd* (equation 3)
0.191 X 1, Work: AISI 1020 steel, wheel speed: 12500 fpm
n x
This was done for three values of (y/x), which is the ratio of
wheel cost per cubic in. (y) to labor and machine cost per min
(x). In many cases y and x will be about the same numerically
General Experimental Study (for example, typical values are y = 15 cents/cu in. and x = 15
In order to check this expression as a means of relating values cents/min). Therefore, one of the values of y/x selected was
of.(,. and d*, cutoff tests were run on AISI 1020 steel having differ- unity. The other two values were 50 percent above and below this
ent values of|f (% to 3 in.) using a 20 x %6 x 1 wheel having the value, respectively (i.e., y/x = 0.5 and 1.5). This approximately
designation A243 R6B (24 grain size). The nominal wheel speed represents the practical range of values. The optimum points for
was 12,500 fpm and cuts were made at several values of downfeed each of the three values of y/x employed are shown on Fig. 3.
for each value of£, and the corresponding values of grinding ratio Values of C (number of cutting points on the wheel face per
(G) were determined. These results are shown plotted in Fig. 3. unit area) were determined for each test condition by use of the
The curves are seen to be consistent with equation (1). Exponent projection method described in reference [2]. These values are
n (inverse slope of G versus d curves) is seen to be a strong func- presented in Fig. 4. Cost optimum values (C*) are also indicated
tion of cutting length (t), while di (value of d when G = 1) is the on these curves.
same for all values of I. This common value of d for all curves is The data given in Figs. 3 and 4 were next used to test the va-
designated (d-y). lidity of equation (8), which, up to this point, is purely an as-
The fact that G will have the same value (unity) for all values sumption. This was done by plotting experimental values of d*

200
e ^COST-OPTIMUM VALUES
o
/ 3#-"°*"^^ ~™»^^—
150

•1/2

'c 100
// / X s - (2i)* 2
X
/ ' y/x=I.O / ^3
y/x=l.5 ^ 0 5
50

1 1 i
10 20 30 40
DOWNFEED, d (ipm)
20 30 40 50 MO6)
Fig. 4 Variation of number of cutting points per square in. (C) with C*/J, (in 2 -min)
downfeed (d) for cuts of different length(i). Grinding conditions same
as Fig. 3. G* value not measured but assumed by interpolation Fig. 5 Variation of d* with C* V/l. Grinding conditions same as Fig. 3

140 / FEBRUARY 1975 Transactions of the ASME

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jmsefk/27618/ on 03/24/2017 Terms of Use: http://www.asme.o

 versus VC*/ I for all values of I investigated. If equation (8) is
correct, a straight line passing through the origin should be ob-
tained. The resultant plot is shown in Fig. 5. This is seen to be an
excellent straight line which, however, does not pass through the
origin. The equation for this line is seen to be

l(d*-rfn)
VC* (ID
instead of the expression initially assumed in equation (8).
The difference between equations (8) and (11) may be inter-
preted as being due to the fact that thin chips may be accommo-
dated more readily than thick chips due to their greater flexibili-
ty. Thus, the optimum volume per chip (ld*/VC*) may be some-
what greater for thin chips (large I ) than for thick chips (small
t). This may be readily seen by writing equation (11) as follows:

id* td0 50 (x|06)

C, + VC* o p t i m u m chip v o l u m e (12)
VC*
The second term on the right will be larger the larger the value of T"
t. Equation (11) may be looked on as an improved version of Fig. 7 Variation of d * with C* Vjl. Grinding conditions same as Fig. 3
equation (8) which takes relative chip flexibility into account.
In the foregoing discussion, it is evident that three design equa-
tions may be written which must be solved simultaneously in If the following quantities are known these equations may be
order to predict values for the three required unknowns d*, G*, solved for d*, G*, and n.
and n. C* = grains/in. 2
From chip volume considerations: C2 = optimum vol/chip
l(d*-d0) do = chip flexibility parameter
= C, (13) di = value of d when G = 1
VC*
I = work length
From wheel wear considerations:
V =cutting speed
log djd* y _ wheel cost/cu in.
(14)
log G* x machine and labor cost/min.
From cost optimization: b = wheel thickness
The solution for d*, G*, and n may be obtained as follows:
-)bd* (15) 1 d* may be obtained directly from equation (13) since all
quantities but d* will be known.
2 Equations (14) and (15) must be solved together for G and
re. By substituting equation (14) into (15), we obtain

60 I ^-bd*
50 G* log G* = K log G* (16)
log (dx/d*)
where K is the quantity in brackets which involves known quan-
30
tities only. Fig. 6 gives corresponding values of G* and K. From
Fig. 6 it is evident that two values of G correspond to a single
value of K, and this gives rise to some difficulty when attempting
20 a solution. In general the proper value of G will lie on the lower
branch of Fig. 6 when I' is large and vice-versa. The only known
way of ascertaining which of the two roots is the real one and
which is spurious is to refer to the plot of original data (Fig. 3).
This point will be illustrated in a subsequent example.
10
Four wheel characteristics are involved in the foregoing analy-
8 sis, three of which will be constant for a given y/x (C2, do, and
7 di). Only C* will have different values for different values of^
6 (see Fig. 4). It would be more convenient if a mean value of C*
5 were to be used. This may be done with insignificant change in
the predicted values of d*, G*, and re, provided appropriately al-
tered values of do and C2 are employed.
3
For example, Fig. 5 is shown replotted in Fig. 7 using C* =
e mean grains/sq in. in place of the variable values of C* that per-
tain for different values of I (Fig. 4). It is evident that equation
2 (13) holds equally well when C* replaces the variable values for
C* but that the slopes fC2j and intercepts (do) will have slightly
different values.
If the information in Table 1 were to be given by a wheel man-
ufacturer for a cutoff wheel-workpiece combination it would be
1 i
15 20 possible to rationally select and use the wheel best suited for cost
K
optimum performance. The manner in which this could be done is
Fig. 6 G* versus K illustrated in the following example.

Journal of Engineering for Industry FEBRUARY 1975 / 141

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jmsefk/27618/ on 03/24/2017 Terms of Use: http://www.asme.o

 Table 1 Table 2
Wheel: designation, A 2 4 3 - R 6 B ; diameter, 20 in.; thickness W h e e l : designation, A24-T6-B, d i a m e t e r , 20 in.; thickness,
(6) 0 . 1 9 0 in. 0 . 1 9 1 in.
Wheel Speed (V): 12500 fpm Wheel Speed: 12500 fpm
W o r k : A I S I 1020 steel W o r k : A I S I 1020 steel

y/x c* C 2 (in. 3) do (ipm) di (ipm) X

c* C2
(in. 3 )
do
ipm
rfi
ipm
(in.-2) (in." 8 )
0.5 119 8 0 . 0 X 10 s 12.7 76 0.5 123 6 5 . 5 X 10 8 15.5 75
1.0 135 66.5 6.0 76 1.0 132 60.4 7.5. 75
1.5 139 62. 3.5 76 1.5 125 55.5 5.0 75
Note: These values are from Figs. 4 and 7.

with each of the two values of G* on Fig. 3 is it possible to decide

which of the two values of G* happens to be the real one (i.e., the
one which clearly lies between the I = % and t = 1 lines on Fig.
Example 3)
A square bar of AISI 1020 steel measuring % X % in. is to be From equation (14)
cut. The wheel speed available is 12500 fpm. The value of y/x is
close to 1 (y = 15<t/cu in. and x = 15$/min). The wheel under log 24
consideration is that of Table 1. The values of d*, G*, n, and cost W = = 1 1 5
0^343 -
per cut are to be estimated. The workpiece is oriented such that
and from equation (4)
Substituting into equation (13) l* 15
(1 + 1.15) = 1.35 c e n t s p e r in. cut
(0.75)(d* - 6) h 24
66.5 x 10"
12(12500)(135)
d* = 2 4 . 0 i p m
i* = .675 c e n t s / c u t
The value of K from (equation 16) is
1 R Experimental Verification
(0.75)(-i|)(.19)(24)
6 90 In order to check these results, bars measuring % X were cut
* = —T/nn = - at the following downfeed rates: 20, 25, and 30 ipm. In each case a
single determination of G was made involving 10 cuts for each de-
and hence from Fig. 6 termination. The results are shown plotted in Fig. 8 where the
solid line for I = % is that predicted by use of Table 1 and the
G* = 4.5
points are for the actual experiments performed. Agreement is
In general there will be two values of G*. Only by plotting d* seen to be excellent.

60
60
i—r~r 40

40-
OMEASURED CUTTIN6 LENGTH,X ( i n
9 CALCULATED
20

o
AT

10
K

LENGTH CUT,* (in) W B

z
9 6
EX.

4 -

2 -

6 8 10 20 40 40 8 0 100
DOWNFEED, d (ipm)
dD = 5.0 7.5 15.5
DOWNFEED, d ( i p m )
d,=76 ipm
Fig. 9 Actual values of grinding ratio (G) vs downfeed (d) for work-
Fig. 8 Comparison of measured and predicted results f o r t = % in. pieces of different length, Wheel: A24-T6B, 20 X 0.191 X 1 . Works: AISI
Grinding conditions same as Fig. 3 1020 steel. Wheel velocity: 1 2 5 0 0 fpm

142 / FEBRUARY 1975 Transactions of the ASME

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jmsefk/27618/ on 03/24/2017 Terms of Use: http://www.asme.o

 •= 100

CUTTING LENGTH,t (in)

20 30 40 50 60
DOWNFEED, d (ipm)

Fig. 10 Variation in number of cutting points per square inch (C) with
downfeed (d) for cuts of different length ( ( ) . Grinding conditions same
as Figure 9

Second Wheel 1 To determine G and C at two values of I and at three

Results for an additional wheel are given in Figs. 9 to 11 and in values of d for each value oH.
Table 2. These additional values show that the same relative pic- 2 These results should be plotted as shown diagramatically
ture holds for a wheel of another hardness (T). in Figs. 13a and b. From Fig. 13a the quantity dx may be deter-
A specimen of very small t, (V*) was also included in the second mined.
wheel tests (Fig. 9). The G versus d curve for this small I did not 3 Measure n for the two curves of Figure 13a and determine
pass through point di as did all other curves. Specimens having d* and G* using the following values of y/x: 0.5, 1, 1.5, plot these
an I this small are obviously not space limited and hence belong points on Figures 13a and b as shown.
to a different family of results not covered by the foregoing analy- 4 Choose values of C* for each value of y/x.
sis. However, specimens of practical size behave in accordance 5 Plot d* against VC*/t as shown in Fig. 13c and determine
with the space-limited mechanism. do and C2.
Fig. 11 was plotted using the actually measured values of C* for 6 Use the values of do, C2, €*, and di thus obtained and com-
each test. This figure is shown repeated in Fig. 12 using the mean pute n, d*, and G* for values of t = 0.5, 1, 1.5, 2, 3, and 4, and
values of D* for each value of £ which are designated C*. The re- y/x = 0.5,1, and 1.5.
sults are again very similar. The final plot contains all of the constants describing wheel
characteristics given in Table 1 and at the same time, will enable
More Practical Approach the user to determine at a glance the value of d* to be used for a
While the method that has been outlined involving the con- given combination oft and y/x.
stants of Table 1 is a possible method of rating cutoff wheels it is The cost per cut ($) may be determined by noting n corre-
not very practical. This is particularly true in view of the diffi- sponding to the d* chosen, and substituting into equation (4):
culty of determining which value of G is the correct one from Fig.
6. It is also not realistic to expect a manufacturer to perform as t* = | j r ( l + n) (4)
many tests as were performed here for the two wheels investi-
gated. A more realistic approach appears to be the following: The user may quickly determine which of several wheels should
be used for a given job by noting the wheel corresponding to the
lowest value of <t*. It is estimated that the time required to rate a

y /x 0.5,

ii- 20-

60 MO6)
5 0 (x|0 s )

Fig. 11 Variation of d * with C*V/1. Grinding conditions same as Fig. 9

g Fig. 12 Variation d * with C*V/i. Grinding conditions same as Fig. 9

Journal of Engineering for Industry FEBRUARY 1975 / 143

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jmsefk/27618/ on 03/24/2017 Terms of Use: http://www.asme.o

 1=1
-9*—,
\
\
\y/»
\ ,1.5

o *' l.O

' '\ 1 (b)

1 1 \ 1
1 / f9

*»' s / / \

/ *» >
G=l
(a)
liL_ _

• ••.• •.?•.••.•;.••-v.

WHEEL:A24-R6-B
20«.I9IX|
W0RK-AIS1 1012
V M 2 , 5 0 0 fpm

y/> 5* C2 dp cl,
1.5 - - - -
1.0 - - - -
0.5 - - - -

f = -=<Hn)

G=l

Fig. 13 Schematic diagrams illustrating data required for cutoff wheel

analysis

wheel following the foregoing abbreviated procedure is two days. 14d shows the original experimental lines fort = \ and 1 in. to-
gether with the lines obtained from the general constants for this
Third Wheel Test wheel given in Table 3.
A third wheel of different grain size (A46-R6-B) was subjected
to the abbreviated procedure outlined above in order to test the
feasibility of the scheme. Plots corresponding to those of Fig. 13
are given in Fig. 14a-d for this wheel. From this last example it
is evident that the method is not only feasible but practical. Fig. Table 4
C* C2 do d,
X in." in. 3 /cut ipm ipm
A243-R6-B (24 grain size)
Table 3 0.5 119 80 X 108 12.7 76
1.0 135 66.5 6.0 76
Wheel: designation, A46-R6-B, diameter, 20 in. thickness, 1.5 139 62 3.5 76
0.184 in. A24-T6-B
Wheel Speed: 12500 fpm
Work: AISI 1020 steel 0.5 123 66 X 108 15.5 75
1.0 132 60 7.5 75
y c* C2 _ d0 di 1.5 125 56 5.0 75
(in.- 2 ) (in.3) ipm ipm A46-R6-B
0.5 183 11.0 X 10» 15.5 32.5 0.5 183 11 X 108 15.5 32.5
1.0 190 17.5 7.5 32.5 1.0 190 18. 7.5 32.5
1.5 193 18.0 4.5 32.5 1.5 193 18 4.5 32.5

144 / FEBRUARY 1975 Transactions of the ASME

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jmsefk/27618/ on 03/24/2017 Terms of Use: http://www.asme.o

 1 1 .,. r-. j i l l
4 0 --

20
1

i=l/2s x m

y/X
« ,5 \
VA
1

A/ \ :
1

8-
80 100 (xlO6)
6-
/ A 0.5\ £"V ,(in
lnz.-min)

/ v %*
2-

1 1 ' 1 'l / , ° H I I I
2 4 6 8 10 1 20 • 40 60 80

d0= 6.1 8.7 16.3 d,= 32.5

DOWNFEED, d, (ipm)

<
or
NG

300 10
Q
Z 8
IE
U> 6
250

200

150

10 20 30 40 d 0 = 4.5 7.5 15.0 d, = 32.5

DOWNFEED, d (ipm) DOWNFEED, d (ipm)

Fig. 14 Abbreviated wheel characteristics. Wheel: A46-R6B; 20 X

0.184 X 1. Work: AISI 1020 steel. Wheel speed: 12500 fpm. a) G versus
d curves; b) C versus d curves; c) d* versus C* Vj% curves; d) final plot

Discussion size as might be expected (from about 60 X 10~ 8 to 20 X 10" 8 cu/

By comparing the results for all three wheels the following ob- in. chip as the grain size changes from 24 to 46).
servations may be made (see Table 4 and Figs. 3,9, 14d): 4 The quantity do (chip flexibility parameter) is about the
1 The quantities C* (cutting points per sq in.), and Ci (vol same for all three wheels, for the same value of y/x. This suggests
per chip) seem to be more meaningful measures of wheel perfor- that do is a material constant, largely independent of wheel char-
mance in abrasive machining than the three related wheel pa- acteristics. Since do measures the role of chip flexibility relative
rameters: grain size, grade, and structure. to optimum specific chip volume it is only natural that this be a
2 The quantity C* seems to be about the same for wheels of work material constant. The quantity do is independent of i, and
the same grain size but increases appreciably with decrease in apparently depends only on y/x (relative wheel-to-labor cost).
grain size. 5 The quantity dt (downfeed when G = 1) appears to be es-
3 The quantity Ci (mean vol/chip) is about the same for sentially independent of wheel grade, but depends on grain size
wheels of the same grain size but decreases with decrease in grain (di goes from 75 to 33 as grain size goes from 24 to 46).

Journal of Engineering for Industry FEBRUARY 1975 / 145

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jmsefk/27618/ on 03/24/2017 Terms of Use: http://www.asme.o

 The foregoing observations are based on relatively little data and the corresponding value of cost optimum grinding ratio (G*)
and should be checked by further testing. optimum downfeed (d*) and optimum cost (£*) may be read di-
rectly from gages.
A d a p t i v e Control U n i t By use of this device the grinding wheel rating process should
An adaptive control unit has been developed at Carnegie-Mel- be significantly shortened and made less laborious.
lon University [3, 4] that should prove useful in rating grinding
wheels by the method suggested here. This device automatically Acknowledgment
measures the change in wheel diameter after a certain number of The author wishes to acknowledge a. grant from the Grinding
cuts in production (say 5) and then computes the wheel and labor Wheel Institute that supported this study and the very helpful
cost per cut using analog components and stores the sum. The criticism and advice of the Research Subcommittee of the GWI.
downfeed is then increased by a small incremental amount. After
the next series of five cuts the mean cost per cut is again comput-
References
ed and compared with the previous one. If the mean cost per cut
is less for the second series than for the first, the downfeed is 1 Farmer, D. A., and Shaw, M. C, "Economics of the Abrasive Cut-Off
Operation," TRANS. ASME, SeriesB, Vol. b89,1967, p. 514.
again increased by a small incremental amount. Otherwise the 2 Shaw, M. C, Farmer, D. A. and Nakayama, K., "Mechanics of the
downfeed is reduced by the equivalent amount. In this way cost Abrasive Cut-Off Operation," TRANS. ASME, Series B, Vol. b89, 1967, p.
optimum operating conditions will be automatically achieved 495.
after a relatively small number of cuts. 3 Gall, D. A., Sauer, W. J., and Atkins, N., "Adaptive Control of the
Abrasive Cut-Off Operation," Proc. ASTME, MR69-228,1969.
Values of specific wheel cost (y) and labor cost (x) may be 4 Gall, D. A., "Adaptive Control of the Abrasive Cut-Off Operation,"
dialed into the unit to take care of different operating conditions, Annals of CIRP, Pergamon Press, Oxford, England, Vol. 17, 1969, p. 395.

146 / FEBRUARY 1975 Transactions of the ASME

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jmsefk/27618/ on 03/24/2017 Terms of Use: http://www.asme.o