Sunteți pe pagina 1din 7

Ceramic Grinding Temperatures

Rajadasa R. Hebbar,* Srinivasan Chandrasekdr,** and Thomas N. Fdrris*


School of Industrial Engineering and School o f Aeronautics and Astronautics, Purdue IJniversity, Wcst Lddyette, Indiana 47907

Workpiece and abrasive-tip temperatures are measured during the grinding of silicon nitride, zirconia, sapphire, and Ni-Zn ferrite with diamond abrasives. The measurements are carried out using a multiple-element infrared sensor having a time constant of -7 ps and a spot size as small as 30 pm in diameter; the temperature measurements were made for both single-point and full-wheel grinding. A simple analytical model is used to predict the abrasive-tip and work-surface temperatures, the subsurface temperature of the workpiece, and the cooling of the abrasive grains between cutting. The predictions of the model agree reasonably well with experiments. The model requires as inputs only the grinding forces, grinding process variables, and the physical properties of the abrasive and workpiece materials. It is shown that, in view of the excellent thermal properties of the diamond abrasive, the surface temperature of the ceramic workpieces is essentially independent of the thermal properties of the ceramics.

and cooling, or by diffusion wear of diamond at high temperatures when grinding ferrous materials.' Thermal phenomena thus play a key role in the mechanics of the grinding process. The measurement of peak temperatures generated during the grinding process is critical to an understanding and prediction of the various phenomena occurring in the wheel and workpiece and in the development of analytical models for ceramic grinding temperatures and stresses. Of course, control of these temperatures would be valuable in the production grinding of ceramic components.
11. Background

I.

Introduction

H E machining of ceramic materials is commonly accomplished by grinding with diamond abrasive wheels. During grinding, high temperatures are generated at the interface between the wheel and the workpiece as well as in the work subsurface, due to frictional heating and localized plastic deformation. The high temperatures are an important source of damage on the machined surface. I-' Firstly, the transient temperatures generated by grinding contribute to residual stresses and microcracking on the ground surface. Secondly, the localized temperatures cause warping of the component, especially when it is of small size and has a large surface-tovolume ratio. Thirdly, they can induce phase transformations on the ceramic surface. Such phase transformations occur, for example, during the localized heating of magnetic ceramics such as ferrites above the Curie temperature4 and the grinding of transformation-toughened zirconia. In the latter case, the phase transformations are a consequence of the transient thermal and mechanical stresses produced by grinding. Grindinginduced damage alters the mechanical, magnetic, and electrical properties of the ground ceramic. Furthermore, the transient temperatures prevailing at the abrasive grain tip during grinding contribute to wheel wear. Wheel wear can occur by thermally induced degradation of the bond holding the diamond abrasives together on the wheel, by graphitization of the diamond particles when heated above 1200C in air,' by microfracture of the abrasive grain as a consequence of repeated heating

R . 0. Scartcrgood~conlributiiig cditor

Manuscript No 196144. Receivcd DecembcrS, 199l,approvcd June I I, 1092. 'This work wah partially hupportcd by thc National Science Foundation (NSF) through Grant Nos DDM 90579 I 6 (Prograni i n Manufacturing Proccrses and Systerns, Dr. B Krdrncr, Director) and MSS 9057082 (Surface Engincering and Tribology Program. Dr. J . Larsen-Basse, Director) *Member, Anicrican Ccrarnic Socicty. ;School or Induhtrial Engineering. 'School 0 1 Aeronautics and Astronautics.

The diamond particles in grinding wheels used in ceramic grinding are typically 20-200 p m in size and irregularly shaped. They also cool rapidly after they leave the grinding zone. Therefore, the high temperatures generated at the diamond particle tip exist over small areas and for periods of time of the order of 3 0 4 0 ps. Any sensor for measuring the peak temperature must therefore be able to sample hot spots as small as 20 p m in diameter and have a time constant 10 ps or less. In the past, thermocouples have been embedded at various depths in the workpiece to measure the workpiece temperature. Such thermocouple probes have a poor frequency response and spatial resolution and therefore cannot accurately estimate the peak grinding temperatures .* Mayer and Shaw' were the first to obtain the temperature of a freshly ground steel surface by measuring the infrared (IR) radiation emitted from it using a lead sulfide detector. Later Kops and Shawl" photographed the radiation emitted by the wheel and the chips on an 1R-sensitive film to qualitatively study the grinding temperature distribution around the wheel. Ueda et al." used an infrared sensor with an indium antimonide (InSb) detector to measure the wheel-surface and work-subsurface temperatures during the grinding of steel. All the above IR grinding temperature measurements were obtained during the grinding of steels. Furthermore, the techniques assumed that the target was either a black body with an emissivity of 1 or a gray body (constant emissivity) with an emissivity value which was directly measured by a calibration procedure. While the gray body assumption is reasonable at a given temperature over a small range of wavelengths, the use of directly measured emissivity values in the temperature measurement procedure can lead to considerable error in the measured temperature. '' This is because the target emissivity during grinding could be quite different from its value during the calibration experiments. For example, if the infrared sensor was focused on the abrasive particle, the particle's emissivity would be influenced by any chip fragments adhering to it. Hence the use of an appropriate emissivity compensation algorithm which does not require a direct measurement of the target emissivity is desirable for accurate IR temperature measurements. Grinding temperatures were measured during the grinding of electronic ceramics such as Ni-Zn ferrite and sapphire by Chandrasekar et ~ 1 . and Farris and ChandrasekaP using a multiple' ~ element infrared sensor. The sensor used a fiber-optic lens assembly to pick up the radiation from the target (diamond abrasive) and focus it on the detector plane. The sensor had a

2142

spot size of 40 pin diameter and a time constant of 6.8 ps. A direct calibration procedure was, however, needed to compensate for the emissivity. The temperature measurements were conducted during single-point grinding, i.e., grinding with a single-diamond indenter mounted on the periphery of an aluminum wheel, and for grinding with a complete resin-bonded diamond wheel. The target, being the abrasive particle(s) on the wheel, was observed through a hole in the ceramic workpiece just after it completed the grinding action. The prescnt study describes experimental measurements of the abrasive and workpiece temperatures during the dry grinding of structural ceramics-tetragonal zirconia polycrystal (TZP) and hot-pressed silicon nitride-and, for comparison, electronic ceramics-Ni-Zn ferrite and sapphire. A modified version of the multiple-element infrared sensor used by Chandrasekar et ui. was used to carry out these measurements. The modifications enable emissivity compensation to be conducted using a multiwavelength technique which did not require a direct measurement of target emissivity. Furthermore, the sensor optics were improved to enable a minimum spot size of 30p m diameter to be sampled by the sensor. No temperature measurements were carried out during wet grinding, as the scattering of the infrared radiation by the coolant would lead to considerable coniplicationr while interpreting the measurements. An analytical model based on the work of Shawl5 was used to calculate the abrasive-t ip temperature and the workpiece subsurface temperature, and the predicted values are compared with the experimentally measured temperatures. The consequences of such localized temperatures in the ceramic workpiece and diamond indenter are discussed.
111. Experimental Procedure

emitted light was measured over a narrow band o f wavelength ccntered at O.Y, 1 . I , and 1.3 p m , and the temperature was determined by comparison with black-body radiation. The width of the wavelength band was 0.2 p m , i.e., ? 0.1 p m about each of these three wavelengths. For this purpose a special spectrometer was constructed using the InSb detectors which could simultaneously detect weak radiation signals at three different wavelengths. The experimental configuration for doing this is shown in Fig. 1. The light emitted from the target was collected by a flexible fiber-optic lens assembly, from which a beam divider with filters selects a narrow beam ( 2 0 . I pm) of radiation centered around the three wavelengths of interest. The beams impinged on the cooled detectors, and the output of the detectors was amplified and recorded in a digital storage oscilloscope. The radiation emitted by a heated target at temperature T is a function of wavelength and is given by Plancks Law as

where E , is the monochromatic emissive power at wavelength A, C , and C2are constants, and F ~ , ? the emissivity of the taris get at temperature T and wavelength A. If the target is a black body, then E , , ~ = I . The power radiated over a narrow band of wavelength (Ax) centered at A is obtained by integrating E , over this range as

The multiple-element infrared sensor used in the experiments consisted of eight indium antimonide (InSb) detectors (Santa Barbara Research Corporation and Barnes Corporation) and a high-speed thermal monitor (Vanzetti Systems Corporation). The detectors were mounted on a Dewar flask which was cooled by liquid nitrogen to a temperature of 77 K . Indium antimonide responds to radiation in the 0.65-5.3-pm range and can be conveniently used in the photovoltaic mode to give response times as low as I p s . When cooled with liquid nitrogen, the performance of indium antimonide improves significantly, due to lower thermal noise and also longer wavelength sensitivity. The detector elements were coupled to fiber-optic lens assemblies which transmitted the radiation from the source to the detector plane. With the multiple-element sensor, temperature measurements could be made simultaneously at four different locations. The smallest spot size achievable with the present arrangement was 30 pm; that is, the detector could sense hot spots of 30-pm diameter. The detectors were calibrated statically and dynamically. Static calibration was performed by exposing it to a standard infrared source. Dynamic calibration was performed as follows: ( I ) A camera shutter was placed between a standard (reference) infrared source and the detector. The shutter opening was controlled to expose the source for periods as short as 100 ps. (2) The detector was exposed to a standard (reference) infrared source through a hole near the periphery of a spinning aluminum disk. This configuration is similar to that of the highspeed grinding experiment. The time constants of the detector elements were determined as the time taken for the response to reach 63% of the final valuc for a step temperature input. Each detector element was calibrated separately, and the responses of the InSb elements were qualitatively similar to each other. The time constant of the slowest InSb element was 6.8 ks, while the Vanzetti highspeed detector had a time constant of 7.4 p s . Typically, infrared temperature sensing is carried out by measuring the radiative power emitted by the target over a range of wavelengths. In our measurements, the radiative power of the

where K is a constant dependent on the geometric radiation collection and transmission characteristics of the fiber-optic lens assembly system. Typically exp(C,/AT) >> I . Therefore,

In the present experiments, since the radiative power values were measured around three closely spaced wavelengths, can be assumed constant over this range of wavelengths. This assumption enables the temperature to be determined from the power measurements without explicitly determining the emissivity. The radiative power values measured in the three wavelength windows were normalized by dividing by the power value around the wavelength of A = 0.9 p m . The normalized power values were then compared with the normalized blackbody radiation spectrum to obtain the temperature of the target. This technique and its variants have been used by other investigators, notably Weichert and Schonert to obtain the temperature rise ahead of a propagating crack and Gulino et u1. to measure the temperature rise at asperity contacts. The technique has the advantage that, by making radiative power measurements at three closely spaced wavelengths and taking the normalized power values, emissivity compensation is achieved to a first order without making a direct measurement of the target emissivity.

IV. Results
(1) Single-Point Grinding Temperatures Figure 1 is a schematic of the experimental arrangement used in the temperature measurement studies. A conical diamond indenter with a 90 apex angle and with a hemispherical tip of radius 15 p m was mounted on the periphery of an aluminum disk and slid against the ceramic work surface. The relative sliding velocities between the indenter and the ceramic varied between 25 and 37 m/s, which are surface speeds commonly encountered in surface grinding. The multiple-element infrared sensor monitored the radiation from the diamond indenter tip just after it ground and passed across the 2-mmdiameter hole in the workpiece. At a disk peripheral velocity of

2744

Journcrl of the Arnericun Cerurnic Society-Hebbur

et ul.

Vol. 75, No. 10

Fig. 1.

Schematic of the measurement system for grinding temperatures

35 mis, the fiber-optic lens assembly was exposed to radiation from the diamond tip for approximately 20 p s , which is about 3 times the time constant of the lnSb detcctors. Thus it was possible to obtain a direct measurement of the diamond-tip temperature. Four different ceramics-tetragonal zirconia polycrystal (TZP), hot-pressed silicon nitride (Si,N,), single-crystal aluminum oxide (sapphire), and polycrystalline Ni-Zn ferrite-were used as workpiece materials. The ceramic workpieces were plate specimens -43.75 mm long X 12.5 mm wide X 5 mm thick with the surface to be ground being of rectangular cross section (-43.75 mm X 12.5 mm). In the case of sapphire, the e-axis was oriented perpendicular to the surface being ground. The mechanical and thermal properties of the ceramics are given in Table I.

1 8oo.r

0 0 0

1500.
0

1400.L

5
Event

Fig. 2. Distribution of measured temperatures in the single-point grinding of zirconia in which the experimental conditions for the eight measurements were identical (depth ofcut, 10 pn; table velocity, 23.4 mni/s; wheel velocity, 32 ids; diamond tip radius, 15 km).

Table 1.

Properties of Work Materials and Diamond Abrasive*


Si,N, SaoDhirc Ferrite 1070 Steel Diamond

Zirconia

(YZ-1101'

E (Gpd)
1 ,

k(W/(m."C)) c(J/(g."C)) PWm') kpc ' H (GPa)

210 0.24 2.2 0.63 6.1 8.45 12

300 0 26 33 0.72 3.22 76.5 16.7

390 0 23 35 0.95 3.90 129.7 19.6

191 0.2 8.7 0.71 5.3 32.7 7.3

203 0.26 47 0.432 7.84 159.2

1000

0.2
1000

0.525 3.5 1837.5 88

* E is Youngs modulus. Y ih the Poisson ratio, iis thermal conductivity, c is specilic h u t . 11 i:, mash dcnsity, and H IS hardness. 'YZ- I 1 is a tctragonal zircmia polycrystd 0 manutacturcd hy Norton Company. 'The value:, listed arc the product of I , p. and ( ' tahulatcd ahovc

Table 11.

Single-Point Grinding Temperatures*


~~

V = 2 5 mi\
Malcri.il

h "
~

b'

A'

Ternperdure ("C) V=12m/\ t A'


~~ ~

V 17 m/a : E A'

Zirconla 0064 SI,N, 0 17 bcrritc 0 12 Smnhirc 0 2 I

1260i51 1320 l60I+74 1494 1133-60 1 I I0 14.52?49 12.55 570230 480 620?35 542 6 9 0 t 3 0 583 920 i65 1060& 45 1270tX0

Table I1 shows the measured values of the diamond-tip temperature during the single-point grinding of the ceramic materials. The measurements were made by observing the diamond tip through a hole in the workpiece. The depth of cut used in the experiments was 10 p m , and the table velocity was 23.4 m d s . The values reported in Table I1 are the average and the standard deviation of measurements made from 6 to 10 nominally identical contacts. Because of local variations in the nature of each contact, the temperature of the diamond tip varies from one contact to another even though the grinding conditions are the same; a typical measured distribution of the diamond-t ip temperature over eight nominally identical contacts is shown in Fig. 2 for the grinding of tetragonai zirconia polycrystal (TZP). The experimentally measured temperature of the diamond grain in Table I1 increases with increasing sliding velocity between the tip and the ceramic surface. The highest temperatures were reached during the grinding of zirconia, followed in order by silicon nitride, sapphire, and nickel zinc ferrite. As noted in the Introduction, graphitization of diamond in air has sometimes been observed to occur at a temperature of 1 200"C, and this temperature is exceeded in the present study during the single-point grinding of zirconia, silicon nitride, and sapphire. However, it should be pointed out that the graphitization referred to occurred when millimeter-sized diamond crystals were heated in air for extended times.'

(2) Decay o Abrasive-Tip Temperature f In order to measure the decay of the diamond-tip temperature after it grinds and leaves the workpiece, fiber-optic pickup probes were mounted at 90" intervals along the arc of rotation

2745

2ooo1
S
8 100
k-

E
0.
0.

5.

10. Time (milliseconds)

15.

20.

i'
151

g 5

U .

250.

500.

750.

1000.

1250.

1500

OO.

Temperature (C)

Fig. 3. Decay of the diamond-tip temperature during single-point grinding (depth of cut, 10 pm;table velocity, 23.4 mmis; diamond-tip zirconia with wheel radius, I5 pm). Experimental measurements: ( 0 ) zirconia with wheel velocity of 25 m/s; (W) velocity of 32 m/s; (0) Si,N, with wheel velocity of 32 m/s; (0) with wheel velocity of Si,N, 25 m/s; (0)ferrite with wheel velocity of 37 mis. Solid lines are derived analytically by using Eq. (14).

Fig. 4. Distribution of wheel temperature in full-wheel grinding o f Si,N, (depth of cut, 12.5 pm; table velocity, 23.4 mmis; wheel velocity, 32 m/s; 220 grit).

of the wheel. The tip of the fiber-optic probes was kept at a distance of about 3.5 mm from the diamond tip. The probes picked up radiation incident from the tip when it passed in front of them and transmitted this onto the lnSb detectors. Figure 3 shows the experimentally measured temperature of the diamond tip at various times after grinding zirconia, Si,N,, and Ni-Zn ferrite. During the zirconia and Si,N, measurements, the wheel was spinning at 3450 rpm, while it was spinning at 3000 rpm for the ferrite measurements. The data points in Fig. 3 therefore correspond to just after the grain leaves contact, and 90", 180", and 270" of revolution after contact. The cooling of the abrasive tip is seen to be rapid in Fig. 3; indeed, in all the measurements shown in Fig. 3, the tip cools down to almost room temperature before it begins cutting again, i.e., within one revolution of the wheel.

250.

500.

750. 1000. Temperature (C)

1250.

1500

Fig. 5. Distribution of wheel temperature in full-wheel grinding of zirconia (depth of cut, 12.5 p m ; table velocity, 23.4 mmis; wheel velocity, 32 mis; 220 grit).

(3) Full-Wheel Grinding Temperatures The production grinding of ceramics is carried out using a complete grinding wheel consisting of diamond abrasives held together by a resin, vitrified, or metallic bond. Measurements of the grinding wheel temperature were therefore made during the full-wheel surface grinding of TZP, Si3N,, and Ni-Zn ferrite using resin-bonded diamond wheels. For comparison, the wheel temperature was also measured during the grinding of 1070 carbon steel. The thermal radiation from the wheel, i.e., bond and abrasive, was again monitored through a hole in the workpiece as in Fig. I . The diamond grinding wheels were trued using a rotary truing device (Norton Company) and dressed with an aluminum oxide abrasive stick to maintain concentricity to within 2.5 p.m. When grinding with a full wheel, different areas of the wheel surface are heated to different temperatures during the cutting process. This is because no two cutting points on the wheel surface remove the same volume of material. Furthermore, since the sensor measures the temperature of hot spots having a diameter of at least 30 p m , the grinding temperature of the full wheel cannot be characterized by a single number. Instead the measurements of the wheel temperature are best represented by a histogram. Figures 4-7 show such histograms of the temperature of hot spots on the wheel surface obtained from measurements over a part of one revolution of the wheel. These observations were made by recording and storing the output voltage of the infrared detector in a digital storage oscilloscope using a sampling time of 2 p s . The average wheel surface temperature obtained from a weighted average of the temperature distributions in Figs. 4-7 is given in Table 111. The highest wheel temperature was measured during the grinding of heat-treated 1070 carbon steel, followed in order by silicon nitride and zirconia (TZP). Note that the wheel temperature listed in Table 111 for the grinding of Ni-Zn ferrite is for grinding with a different grit size (320 grit) diamond wheel and for a different grinding condition, the

750. 1000. Temperature (C)

1250.

1500

Fig. 6. Distribution of wheel temperature in full-wheel grinding of 1070 carbon steel (depth of cut, 12.5 pm; table velocity, 23.4 mm/s; wheel velocity, 32 mis; 220 grit).

p
a

2 30

"1
1

; i
g 10
U

0 0.

250. ~

~ 500. ~

~750. ' 1000. ~ ~ Temperature (C)

1250. "

' 1500 ~

'

Fig. 7. Distribution of wheel temperature in full-wheel grinding ol ferrite (depth of cut, 20 pm; table velocity, 23.4 mm/s; wheel velocity, 32 mis; 320 grit)."

depth of cut in this case being 20 p m as opposed to 12.5 p m for the other three work materials. The full-wheel temperatures listed in Table 111 are generally lower than the single-point grinding temperatures of Table 11. This is because any cutting point of the wheel is removing only

2746

.lorirritil

01 the American Ceramic Society-Hehhur

pt

al.

Vol. 75, No. 10

Table Ill.
Matcrial

Average Wheel Surface Temperature in Full- W heel Grinding*


'li.inpcraturc ("C)

Si,N, Zirconia Steel Ferrite

692 592 72 I 620

-a

*220 grit SILC dianiond whecl. 32 inA surlacc velocity. 12.5-pni depth 01 cut b r Si,N,. Lirconia. and sfeel: 320 grit SILC d i i i i i i ~ i i dwheel ;ind 2O-pin depth o cuf f lor Icrrtte.

a fraction of the material removed by the diamond indenter in single-point grinding. (4) Subsurface Workpiece Temperatures Small blind holes, approximately 700 kin in diameter and extending to variotts depths below the surface bcing ground, were drilled from the back side of thc Si,N,, ferrite, and steel workpieces. The holes were polished with a diamond paste to make the wall surfaces smooth. Fiber-optic pickups were embedded in the hole to pick up the thermal radiation from the bottom face ofthe blind hole as the grinding wheel approached the holc. The peak temperature signal was recorded during this approach. Thc purpose was t o measure the temperature at different points in the subsurface of the workpiece and thereby obtain the temperature gradient in the workpiece during grinding with a full diamond wheel. Figure 8 shows the measured temperature at various points in the subsurface of the ceramic and steel workpieces. These are averages from five similar experiments. The depth below the surface is the distance below the freshly ground surface. The temperature gradients near the surface in Fig. 8 are the highest in Ni-Zn ferrite and lcss so in the case of Si,N, and 1070 carbon steel. In fact, tt'c gradients near the surface are about the same in Si,N, and 1070 carbon steel. No subsurface temperature measurements were carried out in zirconia and sapphire, as it was very difficult to drill clean blind holes in these materials. It should be pointed o u t that no corrections were made in the measurements for the infrared radiation entering the fiber-optic lens assembly from the lateral faces of the hole. The measured values of the subsurface temperature are therefore somewhat of an overestimation of the true subsurface temperature.

Fig. 9. Schematic of the moving heat source model used

in

the tem-

perature calculations.

( I ) Single-Point Grinding Temperatures Chandrasekar et 01. I' showed that a moving heat source model'x could be used to accurately calculate the abrasive-tip temperature and the surface temperature of the workpiece in singlc-point grinding. We briefly describe this analysis and apply it to analytically estimate the single-point grinding temperatures which were measured in this study. Figure 9 shows the schematic of the moving heat source model used in the calculations. The moving heat source model approximates the workpiece as a semiinfinite solid whose boundary is insulated away from the heat source. The heat source has a length of 20 in the sliding direction and is of infinite extent perpendicular to the sliding direction. It is moving across the solid with a velocity V and is assumed to generate heat at a rate o f 4 per unit area per unit time (Fig. 9). The solution of the temperature distribution for this problem involves modified Bessel functions. If the Peclet number
2 vu L=--->20
cy

(3)

V.

Analysis

We describe a simple modcl to analytically estimate the abrasive-tip temperature ( = surface temperature of workpiece) from a knowledge of the material properties of the abrasive and workpiece, and a measurement of grinding forces. From this temperature, both thc subsurface temperature of the workpiece and the decay of the abrasive tip temperature after grinding arc also approximately calculated.

where cy = k/(pc) is the thermal diffusivity of the solid, then the flow of heat parallel to the surface can be neglected with respect to the velocity of the moving heat source. This approxiination is appropriate for the single-point grinding experiments where the abrasive grain is moving with a velocity of 30 mis. The differential equation and the boundary conditions governing the one-dimensional transient temperature distribution are

lim T
I
i l

00Or

400

+200 0.

1 ;
T(z,O) = 0

(4)

where T = T ( z , t )is the temperature of the workpiece at depth z (Fig. 9), t is time, and R is the fraction of the generated heat that flows into the workpiece. The solution to this boundary value problem is (Section 2.9 of Carslaw and JaegeP)

10

15

20

25

30 - 35 - 40

Depth Below Sufiace (microns)

Fig. 8. Subsurfacc workpiece temperature during full-wheel grind) ing. Table velocity, 23.4 mmA and wheel velocity, 32 m/s. ( ( 0ferrite (mcasured. 320 grit, 20-pm depth o f cut); (0) (measured, 220 Si,N, grit, 12.5-pm depth of cut): (0) steel (measured, 220 grit. 1070 12.5-pm depth of cut).) The solid lines are calculated values for the

(5)
where erfc is the complementary error function. On the surface

subsurface temperature, obtained by matching the experimentally meaaured tcmpcratures at the surface (aee text for description).

October 1992

Ceramic, Grinding Tcmnpcratitues

2747 Sinele-Point Grinding Forces*


Normal force (N)
Tmgi'ntial force IN 1

Assuming that at a particular point on the surface the heat is applied beginning at x = - a , the time of heating for a point on the surface is t = ( a - x)/V, and the temperature at a point on the surface is

Table IV.
Material

-/ a(u -x)
d=
The average surface temperature is

Zirconia Si,N, Ferritc

2.0
4.5 0.7

0.70
0.95

0.20

*Dcpth of cut. 10 pm; tdhk vclocity, 23 4 mmis; wheel velocity. 32 mls forrirconia and Si,N, and 37 m/b h f c r r i t c ; diamond-tip radius, I S pm.

(7)
which form is consistent with that of a similar analysis given by Ramanath and Shawl' for full-wheel grinding. Equations (4) and (7) can be written for the abrasive grain as well as the workpiece if R is replaced by 1 - R and the properties of the abrasive are used. Writing such an equation and setting the average temperature of the grain equal to the average temperature of the workpiece under conditions of thermal equilibrium gives'"
R =
1

Jg

where the subscripts A and W refer to the abrasive and the workpiece, respectively. The concentrated contact between the abrasive and the workpiece is very similar to that observed in a microhardness test. If we assume that the contact is a circular patch of radius a , the radius of contact can be obtained as
u =

&

(9)

were measured by mounting the ceramic workpieces on a twocomponent piezoelectric force sensor (Kistler Corporation). The sensor consisted of two quartz piezoelectric crystals, one to measure the normal force and the other to measure the frictional or tangential force. A charge amplifier was used to condition and amplify the piezoelectric crystal outputs. The output of the charge amplifier was recorded on a digital storage oscilloscope. The natural frequency of the force sensor was -30 kHz in the tangential direction and -60 kHz in the normal direction, which was sufficiently high for both of the components of the single grain force to be measured. Forces were measured during the grinding of zirconia, Si,N,, and Ni-Zn ferrite. Table IV gives the average values of the measured forces in the normal and tangential directions. These were used in Eq. ( 1 1 ) to estimate the single-point grinding temperatures. Table II gives the analytically estimated temperatures obtained from Eq. (1 1) and values of R (Eq. (8))for the singlepoint grinding of zirconia, silicon nitride, and ferrite. The predictions agree reasonably well with the experimentally measured temperatures (also listed in Table 11) both in magnitude and in their variation with increasing wheel velocity. Even though the single-point grinding process is inherently threedimensional, the reasonably good predictions of the measured temperature with our two-dimensional model suggest that this model is sufficiently accurate for estimating the grinding temperature.

where N is the normal grinding force and H is the Vickers hardness of the workpiece. The heat generated per unit area per unit time is

(2) Single-Point Grain Cooling


To a first order, the cooling of the grain in air after cutting may be modeled as a lumped system cooled by forced convection. The lumping of the diamond tip is justified by its small size and high thermal conductivity of diamond. The governing differential equation for such a system is

FV q = s
where F is the tangential grinding force. Note that this model of the moving heat source does not require any knowledge of the fundamental deformation mechanism (such as brittle fracture, plastic deformation, or plowing) leading to heat generation. Substituting E q . (3) and (8-10) in Eq. (7), the average of surface temperature (Tdvg) the workpiece and the grain in single-point grinding is obtained as

refer to the temperature of the grain and ambiwhere TGand TA ent, respectively, and C is a function of the convective heat transfer coefficient and the surface-to-volume ratio of the abrasive tip. For a given abrasive tip configuration and a given set of grinding conditions, C is a constant termed the decay constant. The solution to Eq. (13) is

TG = T,,
When [(kpc),l(kpc),] >> 1, Eq. ( 1 1) becomes
-

+ (To - TA)exp( - Ct)

(14)

TdVP =

8 f l F
~

3n

Jza?'? Jm
~

and T,, is independent of the thermal properties of the work material. Equations ( 1 1) or (12) along with Eqs. (9) and (10) can now be used to calculate TdvB the normal and tangential if grinding forces are known. In our analysis, these grinding forces were measured and used as inputs. When grinding with a single-point diamond indenter mounted on the periphery of an aluminum disk, the indenter is in contact with the ceramic workpiece for only a short period of time (-50 ps or less). A dynamometer with a high natural frequency is therefore required to measure the single-point grinding forces. In the present study, normal and tangential forces

where T,, is the temperature of the abrasive just after it leaves contact with the work. The decay constant C is found from the experimental data in Fig. 3 by taking the logarithm of both sides of Eq. (14) and finding a least-squares fit for the data. The value of C thus obtained for each experimental condition is given in Table V, and the corresponding curves (Eq. (14)) are the solid lines in Fig. 3. The constant C should be independent of the work material but probably has some dependence on the wheel velocity. Indeed, the values of C in Table V for different materials are all quite close to one another, lying between 8.0 and 9.5.

(3) Subsurface Workpiece Temperatures during Full-Wheel Grinding Once the surface temperature of the workpiece in a fullwheel grinding is known, it can be treated as a constant-temperature heat source moving across the surface to obtain the

2748

Journal

01 the Americcm

Ceramic Society-Hehhar

et ul.

Vol. 75, No. 10

Table V.
Material

Value of the Decay Constant C in Eq. (14) for Single-Point Grinding*


Wheel velocity
(ds)

(lims)

Zirconia Zirconia Si,N, Si,N, Ferrite

25 32

25 32 31

9.0 8.5 8.5 8. I 9.4

-*Depth olcut. 10 pin: table velocity, 23.4 inmis; diamond-tip radius, 15 pin.

subsurface workpiece temperature. Following the analysis of single-point temperatures, the temperature again can be approximated with a one-dimensional boundary value problem given by

tures during single-point and full-wheel grinding agree well with the experimentally measured values (see Tables II and Ill and Figs. 4-8). The analyses show that a fairly detailed knowledge of the grinding temperatures of the workpiece and the abrasive can be obtained from a measurement of the grinding forces. By direct measurements shown in Fig. 9, we have demonstrated that the abrasive grain cools down to close to room temperature during the period between when it completes a cut and begins its next cut. This observation is consistent with the predictions of Kops and Shawl and Shaw. During the single-point grinding of zirconia and silicon nitride, the diamond-tip temperature was often observed to be in excess of 1200C (Table 11); this is the temperature at which diamond is thought to graphitize rapidly.5 This could cause rapid wear of the diamond abrasive. The analytical models could therefore be used to select the grinding parameters so as to keep the diamond temperature well below the graphitization range.

VOA
lim 7
. I

=
+

T,,,,

VII.

Conclusions

0 0
(15)

T(z,O)

where T,,,,is the average surface temperature of the workpiece. T,,,,could be obtained from measurement or from a suitable analytical model. The temperature distribution in the subsurface is given by (Section 2.4, Carslaw and Jaeger)

T ( z , f )= T,,, erfc
In this case, the width of the temperature source can be assumed to be the arc of contact between the wheel and the workpiece which is I = ,/%, where D is the diameter of the wheel and d is the depth of cut. Any point on the surface of the workpiece is directly exposed to the temperature source for a time period off,, = lib! Therefore, the average temperature at any depth during this heating period is

The temperatures of the workpiece and the diamond abrasive have been measured during the single-point and full-wheel grinding of ceramics using a multiple-element infrared sensor. Simple analytical models have been developed to estimate the grinding temperatures from a knowledge of the grinding forces and the mechanical and thermal properties of the work material and abrasive. The analytically estimated temperatures agree reasonably well with the experimentally measured values. The analysis shows that the workpiece temperature is independent of the thermal properties of the work material for the grinding of ceramics with diamond abrasives.

Acknowledgment:

We are grateful to Prof. M. C. Shaw of Arizona State University for discussions and critiquing this work, and for loan of the Vanrctti thermal monitor.

References
ID. B. Marshall, A. G.Evans, B. T. Khuri-Yakub, J. W. Tien, and G. S. Kino. The Nature of Machining Damage in Brittle Materials. Proc. R. Soc. Londo17. A383,461-7.5 (1983). S. Chandrasekar, M . C. Shaw, and B. Bhushan, Comparison of Grinding and Lapping of Ferrite, and Metals, J . Eiig. Ind., 109,76-83 (1987). S. Chandrasekar, M. C. Shaw, and B. Bhushan, Morphology of Ground and Lapped Surfaces of Ferrite and Metal, J . Eng. Ind., 109,84486 (1987). E. C. Snelling, Sofi Ferrite Properfirs und Applica/ions. CRC Press, Boca Raton, FL, 1969. J. E. Field (Ed.). Proprrries qfDiamond. Academic Press, New York, 1982. R. Komanduri and M. C. Shaw. New Method of Nucleating Diamonds. Nofurc, (Washington. DC), 248, 582-84 (1974). G. Shamaunder, R . R. Hebbar, S . Chandrasekar, and T. N. Farris, AbrasiveTip Temperature Measurements during the Grinding of Ceramics; in Grinding Fundumentuls und Applications, ASME PED-39. Edited by S . Malkin and J . Kovach. American Society of Mechanical Engineers. New York, 1989. J. Peters, Thermal Modela i n Grinding; in CIRP Questionnoire. Harrogate, Belgium. 1983. J. E. Mayer and M. C. Shaw, Grinding Temperatures. Luh. Eng., 13, 2127 (1957). L. Kopa and M . C. Shaw, Application of Infrared Radiation Measurements in Grinding Studies; p. 390 in Proceedings of the I Ith North American Manufacturing Research Conference, 1983. T. Ueda, A. Hosokawa, and A. Yamamoto, Studies on Temperature of Abrasive Grains in Grinding-Application of Infrared Pyrometer, J . Eng. Ind., 107, 127-33 (1985). R. V. Jones, lnsrruments and Experiences. Wiley, New York, 1988. S. Chandrasekar, T. N. Farris, and B. Bhushan, Grinding Temperatures for Magnetic Ceramics and Steel, J . Tribol., 112 [ 3 ] 5 3 5 4 1 (1990). T. N. Farris and S.Chandrasekar, High Speed Sliding Indentation of Ceramics: Thermal Effects, J . Muter. Sci., 25 [9] 4047-53 (1990). M. C. Shaw, Grinding Temperatures; pp. 304-308 in Proceedings of the 12th North American Manufacturers Research Conference, 1984. K. Weichert and K . Schonert, Heat Generation at the Tip of a Moving Crack. J . Mech. Phys. Soiids, 26, 151-61 (1978). I7R. Culino, S. Bair. W 0. Winer, and 6 .Bhushan, Temperature Medsure. ment of Microscopic Areas within a Simulated Head/Tape lnterface Using Infrared Radiometric Technique, J . Trihol.. 108, 29-34 (1986). IXH. S. Carslaw and J . C. Jaeger, Conduction uf Heat in Solids, 2nd ed. Oxford University Press, Oxford, U.K., 1959. 3.Ramanath and M. C. Shaw, Abrasive Grain Temperature at the BeginningofaCutinFineGrinding,J. Eng. Ind.. l l O [ l ] 15-18(1988). 0

Equation (17) is integrated numerically, and the results are plotted as the solid lines in Fig. 8. By comparing the experimentally measured subsurface temperatures with the analytically derived values in Fig. 8, it is seen that there is quite a good agreement between the two temperatures. This suggests that the model used to calculate the subsurface temperature is a good approximation to the actual conditions prevailing in fullwheel grinding.

VI.

Discussion

The average surface temperature generated by a single-diamond indenter when grinding ceramics has been measured. An examination of the values of R in Table I1 shows that they range between 0.06 and 0.17, If R = 0, then Eq. (12) holds and the mean temperature is independent of the thermal properties of the workpiece. To a first order of approximation, the mean surface temperature is independent of the thermal properties of the work when a ceramic is ground with a single diamond grit or a complete diamond wheel. The difference in grinding temperatures in Table I1 for different ceramics is due to the difference in specific grinding energy and mechanical properties for these ceramics. A detailed set of measurements of specific grinding energies and temperatures over a wide range of grinding conditions should resolve this issue. The simple analytical models that have been developed to estimate approximately the wheel and work surface tempera-

S-ar putea să vă placă și