Int. J. Mach.ToolsManufact.Vol. 37, No. I, pp.

101-117,1997
coj~sht O 1996.~ by ~ r Sc~acetad
~ ) Pergamon Printedin Great Bri~in.All ~
0e9o-6955/97117.00+ .00

WJ0-CO~W9

A CAD APPROACH TO HELICAL GROOVE MACHINING.

P A R T 2: N U M E R I C A L E V A L U A T I O N A N D S E N S I T I V I T Y
ANALYSIS
S. K. KANGt, K. F. EHMANNt and C. LINt
(Received 25 April 1993)

Abstract--Based on the developments in the first part of the paper, numerical algorithms for the evaluation
of the engagement models have been formulated. Computational examples are given to demonstrate the
validity of the developed models and algorithms. As a problem of pragmatic significance, the sensitivity of
the resulting hefical groove with respect to the machine setting parameters and tool profile errors has been
investigated by means of a mapping from the machine setting c~ tool profile error space to the resulting
helical profile error space. Copyright © 1996. Published by Elsevier Science Ltd

INTRODUCTION

In helical groove machining, the design of the required tool profile for the desired
helical groove or determination of the resulting helical groove by utilizing the given
tool is widely based on empirical cut and try methods. In order to resolve these
problems, the formulation of a general CAD approach for various helical groove
machining problems was developed in the first part of the paper. In this part, first,
numerical evaluation procedures are described and various illustrative examples of
inverse and direct problems with disk and axial-type tools are discussed in order to
evaluate the developed mathematical model for helical groove machining.
Furthermore, from the practical standpoint, the accuracy of a helical grooving
machine, as the measure of its ability to reach a specified tolerance level is a substantial
issue. By means of sensitivity analysis, it can be indicated which machine setting
parameters are the most important and most profoundly affect the resulting tool or
helical groove profile. However, there are very few discussions of the sensitivity (or
error) relations between the given profiles (tool or helical groove) and the associated
resulting profiles in the literature [1]. The existing results do not identify quantitative
sensitivities for machine setting parameters, which are critical indices for controlling the
accuracy of the helical grooving process. A general and quantitative setting parameter
sensitivity analysis is still missing. Consequently as the second goal of this paper, the
profile sensitivities of the resulting helical groove with respect to the machine setting
parameters will be formulated by utilizing the previously developed mathematical
models. The effects of tool profile errors on the resulting helical groove profiles are
also investigated by sensitivity analysis. Some illustrative examples are presented.
NUMERICAL EVALUATION

Before presenting specific examples, details on the numerical evaluation steps for
both the inverse or direct problem are first summarized. Equation identification num-
bers preceded by the roman numeral I (e.g. eqn (I.4)) in subsequent discussions refer
to equations given in Part I.

tDepartment of Mechanical Engineering, Robert R. McCormick School of Engineering and Applied

Science, Northwestern University, Evanston, Illinois 60208, U.S.A.

101
102 S.K. Kang et al.

Inverse problem
The specific solution steps are the following:
1. Define the various machining parameters including the machine setting parameter
(ax, ay, az, and A) and cylindrical workpiece geometric parameter (radius, R,
helix angle, 8o).
2. Determine the variation range of u in eqn (I.4), i.e. Umin ~< u ~< Umax.
3. Define the helical groove profile, eqn (1.4). The starting value of u is Umi,.
4. Evaluate the nonlinear engagement condition, eqn (I.11), with respect to para-
meter v.
5. Determine the resulting locus of contact points (in tool coordinates) by utilizing
eqn (1.23).
6. Determine the cross section of the required tool in the XrZr plane by using eqn
(I.26).
7. Steps 3 through 6 are repeated for discrete values of u in the range determined
in Step 2. The ending value of u is umax.
Step 1 supplies the necessary input parameters for the solution of the given problem.
The definition range of the helical profile, eqn (I.4), on the workpiece is determined
in Step 2. The range for u can be derived from the geometry of the workpiece. As
defined by eqn (I.4), the radial cross-section of the helical groove on the cylindrical
workpiece in the XwYw-plane is a function of a single parameter, u:

x =A(u)
y =fz(u) (I1.1)
and the envelope of the cylindrical cross-section of the workpiece forms a circle in the
XwYw-plane:
X 2 + y2 = R2 (I1.2)

where R is the radius of the workpiece. Therefore, the parameter u satisfies these two
conditions, eqns (II.1) and (II.2), simultaneously. By evaluating these two conditions,
two real (not complex conjugate) solutions with respect to u correspond to the upper
limit and lower limit, respectively (i.e. u,~i,, ~< u ~< Um~x).
In Step 3, the given helical groove is evaluated by varying the parameter u. The
starting value of u is Umin.
Based on the given helical groove surface, the engagement condition is evaluated in
Step 4. The key difficulty in the analytical solution of the inverse problem is related
to the need to solve the nonlinear equation (with respect to v) which expresses the
fundamental principle of engagement, eqn (I.11), between the cutting tool and the
helical groove. Unlike linear equations, there is no general mathematical theory which
allows the determination of the existence and the number of solutions to the
equation [2, 3]: there may be none, one, several or an infinite number. Also, unlike
for linear equations, it is not possible to envisage the use of direct methods (successive
elimination of variables) [3]. All the numerical methods will, therefore, be iterative:
starting with an initial value, either arbitrary or representing an approximation of the
solution, and the iterative procedure will be continued until the equation is satisfied
within a specified error limit.
For Step 4, the IMSL subroutine ZXMWD, which is a minimization routine for
finding the global minimum (with constraints), has been utilized. The fairly complex
nonlinear engagement model, eqn (I.11) or (1.12) is simply denoted as:

L =f(u,v) (I1.3)

The solution of eqn (II.3) with respect to parameter v under the defined constraints, can
 CAD Approach to Helical Groove Machining: Part 2 103

be obtained by numerical optimization in the least square error sense. The performance
measure is defined as the square of error, i.e.

J : [L-f(u,v)] 2 (II.4)

By minimizing this performance measure with constraints on the parameters,

"/i" "IT
- - ~ v <~ - (11.5)
2 2

almost an exact solution, v, can be obtained.

After obtaining v, the remaining two steps (5 and 6) are straightforward. By executing
Steps 3 through 6, one point on the resulting tool profile, which corresponds to one
point of the helical groove profile, is obtained. Since u starts from Umi,, the same steps
are repeated until Um~ is reached as mentioned in Step 7.
The following examples illustrate the implementation of these steps.
Grinding wheel f o r flute grinding. As the first example of helical groove surface
generation, a drill flute surface for a straight cutting edge is considered. Galloway [4]
was the first to derive the condition which the flute surface must satisfy in order to
give a straight cutting edge as viewed along the drill axis under conventional conical
grinding conditions. In this example, Galloway's equations are adopted. The equation
of the radial cross-section of the primary flute which is responsible for the generation
of the straight cutting edge is given by:

x = fl(u) = t sinv/sinu

y = f2(u) = t c o s v / s i n u (II.6)

where

v = u + t(tanSo) (cotu) ( c o t p ) / R (II.6a)

and where 2t is the web thickness, 8o is the helix angle of the drill (at the periphery),
p is the semi-point angle and u is the circumferential angle measured from the x-axis
of the drill to a point on the cutting edge. The values of u are confined to the range:

sin-~[t/R] <~ u <- w/2 (II.6b)

The profile for the secondary (non-cutting edge) flute can be arbitrarily defined in
a manner to facilitate effective removal of chips and, at the same time, to provide a
sufficient strength to the body. Here, for simplicity, this portion of the flute cross-
section will be represented by a circular arc of the form:

x = / l ( u ) = (t + r) + r c o s u (II.7)
y = f 2 ( u ) = r sinu

where r is the radius of the circular arc, and u a parameter in the range -~- ~< u ~<
-w/2.
For the evaluation of this example, the drill geometry and operating parameters are
summarized in Table 1. With these values the flute cross-section in the XMYM-plane
Table 1. Drill geometry and flute grinding parameters

Drill geometry Operating parameters

Radius (R): 10 mm ax: 125 mm

Web thickness (2t): 4 mm ay: 3 mm
Semi-point angle (p): 59* a.: 0 mm
Helix angle (8o): 30* A: 60*

HrH 37-l-ti
104 S. K. Kang et al.

is depicted in Fig. l(a). The profile section AB corresponds to eqn (II.6), and BC to
eqn (II.7).
The resulting wheel profile in the XrZr-plane for the desired flute surface is shown
in Fig. l(b). Points A, B and C are associated with the corresponding points on the
flute surface. It should be noted that, in this example, the range of u for defining the
flute surface was already determined when the desired flute geometry was defined.
Grinding wheel for a parabolic groove. As another disk-type tool case, a parabolic
cross-section, shown in Fig. 2(a), is defined in the workpiece frame by the helical
groove equation:

fl(u) = 0.1 u 2 + 0.5 u + 4.0

A (u) = u (II.8)

The blank workpiece geometry and the machine setup parameters are given in Table
2. In this case, the determination of the variation range for u should be performed by
evaluating eqns (II.1) and (II.2) by considering eqn (II.8) in order to specify the
profile on the workpiece. The range of u is obtained as:

-7.8 ~< u ~ 4 . 8 (II.9)

(a)
Desired Hute PrOfile
15 . . . . . . . . i . . . . . . . . . . . . i . . . .

10

E
"-" 0

? -5
X
_ i
-10

-15 . . . . , , , , i . . . . I . . . . I .... i . . . .

-15 -10 -5 0 5 10 15
Y-axis (ram)

(b)
Resulting Tool Profile
-114

-I18

~ -120
X

-122 .:

-124 ........... B ...... i ....

-15 -10 -5 0 5 10
Z-axis (mm)
Fig. 1.
 CAD Approach to Helical Groove Machining: Part 2 105

(a)
Desired Helical Groove
15 .... , ........ , .... , ...... ,

X -5

° I. . . . .
-10

-15 ,~ . . . . ' . . . .
. . .i. .
k . . . . . . . . . . . . . . . . . .

-15 -10 -5 0 5 10 15
Y-axis ( m m )

(b) R e s u l t i n g Tool Profile

-88 ............ ,...~..., ......

"~ -94
X

-96

-98 F. . . . . . , . . . , . . . . . . . . i ..i,
-10 -8 -6 -4 -2 0 2 4 6
Z-axis (ram)
Fig. 2.

Table 2. Workpiece geometry and grinding parameters

Workpieee geometry Operating parameters

Radius (R): 10 mm ax: 100 tara

Helix angle (8o): 32° ay: 3 ram
az: 5 mra
A: 58°

T h e l o w e r limit o f u c o r r e s p o n d s t o p o i n t A in Fig. 2 ( a ) . T h e resulting profile o f t h e

g r i n d i n g w h e e l in t h e t o o l c o o r d i n a t e f r a m e is s h o w n in Fig. 2 ( b ) .
Ball end mill machining. A s a n e x a m p l e o f a d e s i r e d helical g r o o v e s u r f a c e for
d e m o n s t r a t i n g helical g r o o v e m a c h i n i n g using an a x i a l - t y p e t o o l , a s i m p l e p a r a b o l i c
c r o s s - s e c t i o n is d e f i n e d in t h e w o r k p i e e e f r a m e b y t h e p a r a m e t r i c e q u a t i o n s :

f t ( u ) = 0.1 u 2 + 4.0
f2(u) = u (II.10)

and is shown in Fig. 3(a). The blank workpiece geometry and the machine setup
parameters are given in Table 3. By evaluating eqns (II.1) and (II.2), two real solutions,
which confine parameter u, are obtained:
106 S . K . Kang et al.

t ~
,a, Desired Helical Groove
2q
10 ............ 7. . . . . . . . . . . . . . . . . . . . . . . . . . . "....................... q

f ' t
-15 -10 -5 0 5 10 i5
Y-axis (mm)

(b)
Resulting Tool Profile
6

• "....................... i
E 2
vE
.~ 0

~< -2

-4

-6
-97 -96 -95 -94 -93 -92 -91
Z-axis (ram)
Fig. 3.

Table 3. Parameters for axial-type tool machining example

Workpiece geometry Operating parameters

Radius (R): 10 mm ax: 0 mm

Helix angle (80): 28 ° ay: 100 mm
az: 0 mm

-6.2 ~< u ~<6.2 (II.11)

The resulting profile of the axial tool is depicted in Fig. 3(b).

Direct problem
The computational procedure for the solution of the direct problem is performed in
a similar manner as for the inverse problem. The computational steps of the algorithm
can be summarized as follows:

1. Define the machining parameters including the machine setting parameter (ax,
ay, az, and A), the cylindrical workpiece geometric parameter (R, 8o) and the
tool geometric parameter (wheel radius, offset distance, primary angle, etc.),
which are associated with the given problem.
 CAD Approach to Helical Groove Machining: Part 2 107

2. Determine the variation range of u for eqn (1.2) as Umi, ~< u ~< Umax. The starting
value of u is Um~..
3. Define the tool profile, eqn (I.2).
4. Evaluate the nonlinear engagement condition, eqn (I.19), with respect to para-
meter v.
5. Determine the resulting locus of contact points (in machine coordinates) by
utilizing eqn (I.17).
6. Determine the cross section of the helical groove in the XMYM-plane using eqn
(I.28).
7. Repeat Step 3 through 6 for discrete values of u in the range determined in Step
2. The ending value of u is Um~.

After Step 1, the single valued tool profile, gx(u) and g2(u), is evaluated for the
values of u, defined in Step 2. In the direct problem case, an analytical method to
determine the range of u (i.e. the part of the tool which partidpates in generating the
helical groove) is not available. So, the range of u is empirically chosen by the user.
The engagement model between the tool and workpiece is solved with respect to v
(-~r/2 <~ v ~ ~'/2) in Step 4. The same numerical scheme (least square error method)
as for the inverse problem is applied for the solution of the engagement model. After
obtaining v, the resulting locus of contact points, eqn (I.15), and the cross-section of
the helical groove, eqn (I.28), in the XMYM-plane is determined in Steps 5 and 6,
respectively. In order to specify the whole groove profile, Steps 3 through 6 are
iterated until Um,x is reached. The following illustrative examples show the specific
implementation.
Double conical wheel with sharp edge. The equation of the radial cross-section of
an asymmetric V-shaped (double conical) grinding wheel (see Fig. 4 for the shape and
the definition of the tool geometry), which is frequently used for the generation of
the helical grooves of micro-drill flute surfaces, is defined as:

= [ - u tana + d t a n a - s, for Umin <~ /,/ < d

x = g~ L u t a n / / - d tan/3 - s, for d ~< u ~< Uma~
(i1.12)
z=g2=u

where a and/3 are the wheel inclination angles, d is an offset distance from the wheel
center line, Xr, to the v-shaped edge, and s is the maximal radius of the wheel.

X~
b

Y
d

r~

Fig. 4.
108 S. K. Kang et al.

Table 4. Workpiece/cutter geometry and grinding parameters

Workpiece Grinding wheel Operating parameters

Radius ( R ) : 10 m m Primary angle a : 45 ° ax: 8 0 m m

H e l i x a n g l e (80): 30 ° Secondary angle/3:35 ° ay: 3 m m
Offset distance d: - 2 m m az: 0 m m
Wheel radius s: 77 rnm A: 50*

For the evaluation of the resulting flute profile, the parameters given in Table 4 are
utilized. With these values the cross-section of the given tool in the XrYr-plane is
depicted in Fig. 5(a). As shown in Fig. 5(a), the tool profile is discontinuous at the
intersection of the two cutting surfaces. This discontinuity causes a numerical jump
because the sign of ~1 suddenly changes from negative to positive, and vice versa. To
prevent numerical discontinuity, a simple condition is added when the engagement
model, eqn (I.19), is solved with respect to parameter v, i.e.

If (gl - s) ~< 8, then ~'1 = 0 (II.13)

where 8 is a small positive number (e.g. 0.01).

The resulting helical groove profile in the XMY~rplane is shown in Fig. 5(b). By
(a)
Given Tool Profile
- 7 0 , . . , .... , .... ,-. ' r

-72 .....................................
E

"~ -74
7
x

-76

-781 . . , ~ .... ,, , , , J , ~ ,
-10 -5 0 5 10
Z-axis (ram)

Co)
Resulting Helical Groove

r.
.~ 0

X -5

. 5t .... ......... i,, ,; ......... !

-15 -10 -3 0 5 10 15
Y-axis (ram)

Fig. 5.
 CAD Approach to Helical Groove Machining:Part 2 109

comparing Fig. 5(a) and (b), it is noticeable that the geometry of the machined surface
on the workpiece is not simply related to the cross-sectional profile of the cutting tool
due to the non-rectilinear motion of the tool along the helical cutting path.
Double conical wheel with round edge. In practice, the edge of the double conical
grinding wheel is not perfectly sharp. Although the edge is assumed initially keen, the
shape of the edge becomes round after a few grinding operations. To alleviate the
effects of the rapid initial breakdown, in most cases, a small radius is dressed on the
edge. Therefore, a double conical wheel with a round edge, shown in Fig. 6, is
considered. The equation of the radial cross-section is formulated with reference to
Fig. 6 and summarized as:

alu + bz, for u~a. ~< u < Zl

x=gl=l-~[o2-(u-z')2+ x', forzl <~u<zz (11.14)
La2u + b2, for z2 ~< u < Um~
Z=g2=u .

The machining parameters for the evaluation are the same as in Table 4, and as an
additional parameter, the radius of the fillet is or = 2 nun. The cross-section of the
considered tool in the XrYr-plane is shown in Fig. 7(a). The resulting flute profile in
the X~Y~rplane is presented in Fig. 7(b). The effect of the round edge is clearly
shown in the resulting helical groove.
Ball end mill. As a profile of a given axial tool surface, a simple parabolic cross-
section is defined in the tool frame by the parametric equations:

x = gl(u) = u

z = gz(u) = 0.5 u z - 70 (11.15)

The cross-sectional view of the given axial tool is shown in Fig. 8(a). By utilizing
the parameters in Table 5, the resulting profile on the workpiece is depicted in Fig 8(b).

XT
XT
----~ ZT

Xmal

al=-tano, b t = d t a n o - s , a2~ tan b2=-d lanJS- s (z0,'xo) x=*,vo'--Iz-z~- + x'

¥= zeo°- ( a + p), ~,- 9o-- ~,/2
zo= (xo-l~)/az, x o = - s
zz = ~ tan~ eosa + zo, X l = o tan~ s i n a + xo
z z = o tanE cosp+ zo, x2 = o rant, sinp + xo
z' = o s i n a + z l , x' = o c o s a + x t

Fi~. 6,
110 S . K . Kang et al.

(a)
Given Tool Profile
-70

-72
A

"3 -74
t

X
-76

-78 . . . . . . i . . i i L . . . . i

-10 -6 -2 6 10
Z-axis (ram)

(b)
Resulting Helical Groove
15

10

E 5
E
'3 0
?
X -5

-10

-15
-15 -10 -5 0 5 10 15
Y-axis (ram)
Fig. 7.

PROFILE SENSITIVITY ANALYSIS

Under production conditions an unintentional deviation from the nominal setup is

common. By means of sensitivity analysis, it can be indicated which parameters most
profoundly affect the resulting profile and how large the deviations are. Sensitivity
analysis may lead the machine designer in assigning tolerances and controlling the most
sensitive parameters. Two cases of the sensitivity analysis are formulated. One is
sensitivity analysis with respect to the machine setting parameter errors for the direct
problem. The other is sensitivity analysis with respect to given tool profile errors in
direct problems.

Profile sensitivity with respect to the machine setting parameters

The cross-section of the helical groove in direct problems, for either disk or axial-
type tools, is expressed as a function of u, v, and setting parameters by referring to
Eqn (I.28):

r = r ( u , v , ax, ay, az, A) (II.16)

where r is a vector with three components x, y and z. Simultaneously, the relationship

 C A D Approach to Helical Groove Machining: Part 2 111

(a) Given Tool Profile

. . . , . . . . , . . . . ! . . . . . . . . ! . . . .

i i i ~A :

2
A

E
.~-
x
o
e~

-2
-3
B
-4
-71 -70 -69 -68 -67 -66 -65
Z-axis (ram)

(b) Resulting Helical Groove

15 .... , .... , ................

10

"-" 5
E
.~ 0
x
X -5

-10

-15 "
-15 -10 -5 0 5 10 15
Y-axis (ram)
Fig. 8.
Table 5. Parameters for axial-type tool

Workpiece geometry Operating parameters

Radius (R): 10 mm ax: 0 mm

Helix angle (80): 30* ay: 75 mm
a,: 10 mm

for the engagement between the cutting tool and the helical groove, eqn (1.19) or eqn
(1.21), is implicitly expressed in general form as:

L ( u , v , ax, ay, az, A) = 0 . (II.17)

When setting parameter errors are unintentionally introduced, the curviliner para-
meter v, which is the solution of implicit eqn (II.17), deviates from the true solution,
The sensitivity of v, therefore, with respect to setting parameter errors is expressed
by:

-aL(u,v,p)l
av
op/ l (II.18)
aLfu, v, p ) ] '
8v A
112 S.K. Kang et al.

where p = {ax, ay, az, A} and pj is the ]'th element (setting parameter) of vector p.
Therefore, parameter v is expressed virtually as an explicit function of setting parameters
as:

v = v(u, a~, ay, a~, A) = v(u, p) (II.19)

Consequently, the resulting profile, eqn (II.16), can be expressed in the form:

r = r(u, v, (u, p), p). (II.20)

An explicit and approximate error relation between the resulting profile and the
machine setting parameters can be expressed by using a first order Taylor series
expansion of eqn (II.20) in the form:

Ar = S[,(p) Ap (II.21)

where Ar, and Ap are the m x 1 error vector of the resultant cross section profile
components (m = 3), and the n x 1 error vector of the machine setting parameters,
respectively (n = 4 for disk-type tools or 3 for axial-type tools). This linear equation
can be interpreted as a linear mapping from the machine setting error space Ap to the
resulting profile error space Ar by the linear operator S[,(p), which can be also
interpreted as a sensitivity matrix:

[sz sz
~(p)= Sy Syy Sy S{J, (11.22)
L Sz~ S zay S za z S~

and each term of the m x n sensitivity matrix, S[,(p), is a scalar quantity and obtained
by the chain rule as:

Or( u , v ( u , p ),p ) = O r + _Or -~p/

Ov j = l , .., n , (II.23)
opj opj ov

where

Or _ . Ox
~p/ i~
+jOY Oz
~p~p/+k 0-p~p/, (II.23a)

Or . Ox Oy k dz
O--v= ! Ovv+ JOv + 0-v' (II.23b)

and avlap/is the quantity which is evaluated by eqn (II.18).

The sensitivity of the helical groove profile in the x-direction for the direct problem,
for example, with respect to the setting parameter a is expressed as:

S~ = °x( u' v( u ' p )'P )

OA
Ox ox Ov
= --OA+ ----av
Ox
oA = --oA+ ~v - ~-~
Ox

I]
where the relevant values for the evaluation are obtained from the Appendix as"
, (II.24)
 CAD Approach to Helical Groove Machining: Part 2 113

ax = ( U y w - ae)(UXw sinll - Uyw cosIl)

+ (MZw - a~) sinll
aA k
#x = (rx r cosa sinl~ - r y r cosll) + rxrU Yw sinA cos/l - rxr~xw sina sinll (II.24a)
Ov k
aL
= -~2(u) (k cosA - a~sinA) sinv - G(u) cosA cosv - ~l(u) (a~cosA + k sinA)

aL
av A(u) cosv B(u) sinv

The sensitivity terms of the resulting Z directional profile in eqn (11.22) are
S~ = S~ = Sl = Sz = S~ = 0 in the direct problem because the resulting profile is
detined Yonly in the XMYM-plane.
It should be noted that the sensitivity matrix can be evaluated for either the disk or
axial type problem by the same procedure. By substituting a = - 9 0 ° into eqn (II.22),
the sensitivities for the axial-type tool can be evaluated.
In the following paragraphs, illustrative examples of profile sensitivity analysis are
given.
Sensitivity of the helical groove profile. As an example of sensitivity analysis of the
resulting cross-section of a helical groove, a simple parabolic cross-section for a disk-
type tool is defined for direct problem sensitivity analysis as:

g l ( U ) = 0.5 U2 -- 20

g2(u) = U. (II.25)

For the evaluation of the sensitivity matrix, the following workpiece geometry and the
machine setup parameters are utilized: radius R = 10 mm, helix angle a = 30°, axis
distance a,, = 25 mm, ay = 4 ram, az = 3 mm, and wheel angle a = 60°.
As a result of evaluating eqn (II.23) at 200 points along the tool profile, Fig. 9(a)
through (d) are obtained. The graphs show the x, y and z directional sensitivities of
the resulting helical groove cross-section, with respect to the four machine setting
parameters. The x and y directional machine setting error dominantly affect the same
directional errors of the resulting cross-section, as shown in Fig. 9(a) and (b), respect-
ively. An error in the setting parameter az affects both the x and y directional profile
values (Fig. 9(c)). The sensitivity of the x - y cross-section of the workpiece with respect
to the setting parameter a is significantly large as shown in Fig. 9(d) in helical groove
machining. The sensitivity variation ranges are around 8 for both x and y directions
of the resulting helical groove profile.
Consequently, it can be noted that the setting parameter A is the most sensitive
parameter for direct problems, and that the directional setting parameters (ax, ay, az)
dominantly affect variations of the resulting cross-section in their own directions.
Groove profile sensitivity with respect to tool profile errors in direct problems
The purpose of this section is to identify groove profile errors due to given tool
profile errors. The sensitivity formulation follows similar procedures as in previous
derivations.
The given tool surface, denoted by eqn (I.3), is expressed as a function of the x
and y directional profile, and v as:

r=r(g~(u),g2(u),v) (II.26)

where r is a three dimensional vector with x, y and z components. Simultaneously,

the implicit model for the engagement condition, based on eqn (I.19) or eqn (I.21),
is expressed as:
114 S.K. Kang et al.

(a) (b) ~ . . ~
,['2=7 ,

! il/
a x error e f f e c t s ..... Y a, error effects
t . . . . . . . . . . . f . . . 0.4

0.9 ................ P. . . . . . . . . . . . . . . . . . . . . . . . i ............. 0.2 0.,5 •

................................................i .--'.
i ~ io B ....
04
• i [ : •B !
T, % : : : • :

0.8 ~ 0.9 ..........

~I r;............... i ............ i k'.e........ i.......... 0. 2

a i .....
y

0.7 ~ .........
~ ! .........
- ~"¢'""i
i ..................
~ ~ .............. i .............
.............. 0.2

0.6 . . . . . .
i
i . . . . . . . . . . 0.4 I).8 ".. "7:. ...... 112
40 80 120 160 200 o 411 80 120 I~} 21141
No. of points No. of points

(c)
a z error effects (d) ~ a~. e r r o r e f f e c t s
-0.2 .. . , . 0.4 • • • i - - - T 8

i°°°','-,~.
: •, b
0.3 . . . . . . . . . . ~. . . . . . . i Beg . . . . . ;--...
-0.3 " i :......... ~.~i.;.~
........... i ~ .............! ..........

" i "%. i 4
0.2 >" 2 ; '

.~ -0.4 "N 2
0.1
'= i i ~,
0
-0.5
0 / .......~ ..........i ........ i ...........
;-,.................
2
"- .~'.• ! i i

-0.6 O.I .l
O 40 80 120 160 200 O 40 811 120 160 211o
NO. o f p o i n t s No. of poinls

Fig. 9.

L (gl(u), g~(u), v) = O. (II.27)

When tool profile errors (i.e., Agl(u) and ~g2(u)) are present, the curvilinear
parameter v, which is the solution of the implicit equation of engagement, eqn (II.27),
deviates from the true value. The sensitivity of v, therefore, with respect to tool profile
errors is expresed by:

-oL(~,q)]
c91: T/ (II.28)
q~ Oqs OLfv,q)['
Ov 3
where q = {gt(u), g2(u)} and qj is the :h element of vector q. Therefore, parameter
v can be expressed virtually as an explicit function of the tool profile as:

v = v(gl(u),g2(u)) = v(q) (II.29)

Consequently, the resulting profile, eqn (II.26), can be expressed in the form:

r = r(v(q),q) (II 3 0 )

The error relation between the resulting groove profile and the given tool profile
can be expressed in the form:
 CAD Approach to Hefical Groove Machining: Part 2 115

£r = S~(q) Aq (II.31)

where Ar, and Aq are the 3 x 1 error vector of the resultant groove cross-section
profile components, and the 2 x 1 error vector of the tool profile, respectively, and
where the sensitivity S,~(q) is defined in the form:

S~(q) = / S ~ t S~2 (11.32)

LS~, S~,2

The z directional sensitivities are S~1 = S~2 = O because the resulting groove profile is
defined only in the XMYM-plane.
Each term of the sensitivity matrix, S~(q), is obtained by the chain rule as:

~,= O r ( v (aq/
u,q),q) Or
= Oq + ~ - q / 1 = 1 , 2 .
Or Ov .
(11.33)

(a) X . d i r e c t i o n a l tool
p r o f i l e e r r o r effects
1.2 . . . . . . . . . i " "

0.8
.~ 0.6

0.4 ...........................................................................................................

0.2 ..........................................~.......................~ . ~ ................................

0 .... -- ..... ~ ......... --.-%-

b : •

-0.2 . . . . ~.._..:::..~.~.;.~. ......... ;.............................................................

t , ,t [ i = • i . . . . . = . ,
-0.4
40 80 120 160 200
No. of points

(b) Y - d i r e c t i o n a l tool
. . . . p r o f i l e error effects
0.5 • . . ,, • .

0.1

.,~ -0.3

-0.7

-1.1

-1.5 • . , , , h i i , i , i , , i , ,

0 4o 8o 120 ~60 20o

No. of points

Fig. 10.
116 S . K . Kang et al.

The necessary terms for the evaluation of eqn (11.33) are presented in the Appendix.
The illustrative example is evaluated based on a simple parabolic tool profile, eqn
(11.25). As a result of evaluating eqn (II.33) at 200 points along the tool profile,
Fig. 10(a) and (b) show the x, y and z directional sensitivities of the resulting helical
groove, with respect to tool profile errors. The x directional tool profile errors
(Fig. 10(a)) dominantly affect the x directional errors of the resulting groove cross-
section. The dominant error direction of the resulting helical groove profile due to the
y directional tool profile errors is y, as shown in Fig. 10(b).
CONCLUSIONS

1. Numerical solutions were developed, based on constraints on the curvilinear

parameters for both inverse and direct problems. Computational examples have demon-
strated the validity of the developed algorithms.
2. The sensitivity of the cross-sectional profiles with respect to setting parameters
and to tool profile errors was formulated and used to identify the most sensitive
machining parameters. The grinding wheel setting parameter, A, is identified as the
most sensitive setting parameter, which should be tightly controlled to assure that the
desired tolerances of the resulting profile are maintained.
REFERENCES
[1] K. F. Ehmann, Grinding wheel profile definition for the manufacture of drill flutes, Ann. C1RP, 39
(1), 153-156 (1990).
[2] J. Ortega and W. C. Rheinboldt, Iterative Solutions of Non-linear Equations in Several Variables,
Academic Press, London (1971).
[3] Y. Gardan, Mathematics and CAD: Numerical Methods for CAD, The MIT Press, Cambridge, Massachu-
setts (1986).
[4] D. F. Galloway, Some experiments on the influence of various factors on drill performance, Trans.
ASME, pp. 191-231 (1957).

APPENDIX

In this Appendix the derivations of the required terms for the evaluation of profile
sensitivity are given.

Machine setting parameter sensitivity

For the evaluation of the sensitivity model, eqn (11.23), the derivatives with respect
to the setting parameters (ax, ay, az, A) and curvilinear parameter v are the following:

Or
- - = i (cosl~) + j ( - s i n l ~ ) + k (0) (A.1)
Oax
ar
- i (sinfl) + j (cosfl) + k (0) (A.2)
Oay

dOar-i( ~xwsinflkMyWc°sIl-)+j( MXwCOSIIkMywsinII)+k(O ) (A.3)

ar= i My w -- ay)( Mxwsin~ -- MywcOSl~) )

k + (MZw -- a,) sinl)
oA

+ i t _ / ( M y w - - ay,,--wCOS..~tMr
O + MywsinIl ) _ (MZw _ a~)cosl~]\ + k (0) (A.4)
\ k /

at"
0v k (0)

rxrUywsinit cosl-/- rXrMXw sinA sintI]

+ i (rXrCOSA sinfl - ryr cosfl
+ k J
 CAD Approach to Helical Groove Machining: Part 2 117

+ I ( rxTc°sA cosfl + ryrsin[l - rxr~xwsinAc°sfl k+ rxrMywsinAsinfl) '(A'5)

where (MXw,MyWand MZw) and (rXr, r y r a n d rzr) represent the corresponding compo-
nents of vectors given by eqns (I.15) and (I.3), respectively, and [1 = Mzw/k.
For the evaluation of eqn (II.23), the derivatives of the engagement model with
respect to the machine setting parameters and v are also required and each derivative,
based on the engagement relation, eqn (I.11), is:

OL
- ~l(u) sinX - ~2(u) sinv cosA (A.6)
Oa~
OL - ~(u) cosv (A.7)
Oay
OL
- - = 0 (A.8)
Oaz
OL
- ~:(u) (k COsA - a~ sinA) sinv - G(u) cosA cosy - ~l(u) (ax cosA + k sinA)
OA
(A.9)
OL
Ov A(u) cosv B(u) sinv. (A.IO)

Tool profile error sensitivity

The required terms for evaluating the groove sensitivity, eqn (II.32), with respect
to the tool profile errors are obtained by utilizing eqn (I.28) and expressed as follows:

Ogl = i cosy cosfl + cosA sinv sinl] - sinv k sinA (~xwsinftk - MYwcosf|)

+J(-c°svsinfl+c°sAsinvc°s~-sinvksinA(MxWc°stl+MywsinIl)-)k
+ k (0) (A.11)

cosA (UXwsin~ - Myw c o s ~ ) ~

°--r-r= i / - sinA sintI
Og2 \ k /
+ j ( - sinA cosfl cosA (UXwcosfl + Myw sin~)}\ + k (0) (A. 12)
k /

0L
-- = - ~1 sinA cosv (A.13)
Ogl
0L
-- = -g2 sinA cosv (A.14)
092

where (UXw,Uywand ~lzw) represent the corresponding components of the vector

defined by eqn (I.15) and [1 = Mzw/k.