Electrical machining characteristics of cemented carbides

Electrical machining characteristics of cemented carbides

77 Wear, 116 (1987) 77 - 88 ELECTRICAL CARBIDES MACHINING P. C. PANDEY CHARACTERISTICS OF CEMENTED and S. T. JILANI ~e~artmeni of ~ecka~jcul ...

809KB Sizes 0 Downloads 17 Views


Wear, 116 (1987)

77 - 88






and S. T. JILANI

~e~artmeni of ~ecka~jcul Roorkee 24 7 667 (India)

and Industrial







~~~~~ersit~ of Roorkee,

1, 1986)

Summary Electric discharge machining (EDM) of cemented carbides is accompanied by problems such as the presence of a resolidified layer, a large tool wear ratio and thermal cracks. In this regard, selection of the appropriate operating conditions and the role of carbide composition is quite crucial. However, in the available literature conflicting reports have been made regarding the role of binders during EDM of cemented carbides. In this paper the authors have applied a single-spark model for heat transfer, in EDM, to compute the crater shape, volume of metal erosion per spark and the thickness of the resolidified layer for three different grades of cemented carbides. The experimental work includes a study of the effects of pulse parameters and the carbide composition on the rate of metal removal, relative electrode wear and the shape and size of the crater produced. The experimentally obtained response equations have been utilized for the computation of optimum pulse duration and binder content in the carbide so as to yield maximum removal rate and minimum electrode wear.

1. Introduction Cemented carbides are enjoying an ever increasing popul~ity in industry and are normally used for the manufacture of metal-forming and cutting tools and components requiring a high resistance to wear. The successful shaping of carbides into intricate shapes has mainly been possible owing to rapid advances in electric discharge machining (EDM) technology [l, 21. However, in some cases, after EDM, the carbide surfaces tend to crack [3 - 51. Lenz et al. [4] have identified the importance of EDM parameters, such as pulse duration and amplitude, from the viewpoint of crack prevention. Watson and Freer [5] observed that, whilst machining Kenna: metal carbide alloy (25 wt.% Co), cracking was always present when pulses of 4 - 500 hs were used. Others [6,7] have studied the machinability and surface characteristics of carbides in EDM. Pandit and Rajurkar [7] compared their own results with those published by the CIRP [ 81 but were 0043-1648/87/$3,50

@ Elsevier


in The Netherlands


unable to derive definite conclusions, especially as regards the role of binders (cobalt) in the spark machining of carbides. They recommend that extensive studies be conducted to understand the exact nature of the problem associated with EDM of carbides. In this paper the authors have conducted analytical and experimental investigations into the technological aspects of EDM for three different grades of carbides. A single-spark model, for heat transfer during spark machining, has been employed to compute the crater shape, volume of erosion per spark, thickness of the resolidified layer and tool wear. The experimental work includes a study of the effects of pulse duration and carbide composition on the metal removal rate (MRR), relative electrode wear (REW), shape and size of individual craters produced during the course of machining and the thickness of the resolidified layer. Experimentally obtained responses have next been employed to yield optimum values of the operating parameters.

2. Analysis

of metal removal

Figure 1 shows an idealized model of heat transfer in EDM, wherein the situation corresponds to the heating of a semi-infinite body by a disc-shaped heat source. In the absence of internal heating the heat flow is governed by A’T(x,

y, z, t) -

+ g (


(x, y, z, t) = 0


Equation (l), in conjunction with the appropriate boundary conditions, has been solved numerically by the authors and the results reported elsewhere [ 91. From eqn. (1) the temperature at instant “t” within the semi-infinite body is obtained as follows:

In order that eqn. (2) is applicable to temperatures in excess of the melting point of the electrode materials, the thermal conductivity and diffusivity, for






Fig. 1. Idealized


for heat transfer

in EDM


use in eqn. (2), have been modified as follows and in subsequent tions “a” and “c” have been replaced by a’ and c’ respectively: a’=



~ PC





Equation (2) has been solved numerically using the program EDM-2 [9]. This gives the temperature distribution in the work and tool electrodes during the passage of a single spark. The temperature values thus obtained have been employed for determining the magnitude of erosion per spark. It is assumed that the volume of the electrode material included between the surface z = 0 and the melt isotherm boundary is ejected out as debris and is equal to the metal erosion per spark. This also gives the dimensions of the crater produced at the anode and cathode and the magnitude of REW. REW is defined as average rate of tool electrode


average rate of work electrode

erosion erosion

x 100

The thickness of the resolidified layer has been computed following the procedure reported earlier [lo]. From eqn. (2) it is possible to locate the boundaries of the boiling and melt isotherms within the work electrode. The thickness of the resolidified layer is computed assuming that the metal erosion is mainly due to evaporation and the removal efficiency of material from the molten pool is very poor. These assumptions have been shown to yield reasonably accurate results [lo]. A computer program, EDM4 [lo], has been employed to compute the thickness of the resolidified layer. The resolidified layer thickness has also been determined experimentally by examining the transverse section of the machined specimen under a metallurgical microscope.

3. Analytical


Table 1 gives experimental and other relevant data pertaining to work materials for use in eqns. (2) and (3). Figure 2 shows the relationship between pulse duration and the magnitude of metal removal per pulse assuming that Pwork = 90 W. Figure 3 shows the dimensions of the crater produced during a single spark, whereas Fig. 4 illustrates the effect of pulse duration on the ratio h/d. Figures 5 and 6 give a relationship between the pulse duration, REW and thickness of the resolidified layer respectively.

4. Experimental


The experimental experiments. For this polynomial

work was planned on the basis of statistical design of purpose the response has been approximated by a


TABLE 3 Data related to tool and work


____.__~_ of carbides

Thermal conductivity (W m-r K-l) Density (kg me3) Specific heat (J kg-’ K-l) Adapted specific heat (J kg-’ K-l) Adapted diffusivity (m2 s-r) Specific melting heat (J kggr) Melting temperature (“C) Boiling temperature (“C) Compressive strength (kN mmm2) Young’s modulus (kN mmP2) Coefficient of thermal expansion (urn m-l K -I ) co (wt.%) WC (wt.%) TiC + TaC (wt.%)

Carbide grade .-______

(Widia India)




80 14800 151 247 21.8 x 10-e 25.0 x lo4 2597 5900 5.3 608 5

67 14300 210 306 15.3 x 10 6 25.0 x lo4 2597 5900 .4.4 568 5.5


6.0 94.0 -

12.0 85.0 3.0

13000 240 336 11.7 ?: 10 6 25.0 X 104 2597 5900 3.7 490 6.5

Work diameter (mm) 10 (solid cylinder) Tool - commercially pure copper machined from a cold-drawn solid rod, cylindrical, 12 mm in diameter with a 3.2 mm central hole for the dielectric supply Dielectric - kerosene (commercial grade) -.-

Fig. 2. Metal removal per pulse as a function GT-20; - * ---, GT-40; Pwork = 90 W.

of pulse duration:


, GT-10; ---,



3. Analytical



due to a single spark (Pwork

= 90 5%‘): (a) GT-10;


GT-20; (c) GT-40.




IO Pulse


4. Dependence

fpw, rk = 90 W).

50 duroflon





, _Lls

of the depth-to-diameter

ratio of the craters on the pulse duration



Fig. 5. REW in carbide GT-20; - . -. GT-40.








Fig. 6. Depth of the resolidified

Y, =

layer as a function

E b,Xi +~ biiXi2 + C




= 60 W):

( CX 10,



of pulse duration


= 90 W)


where xi (i = 1, 2, . . . , k) are the coded levels of k quantitative variables whose effects have to be studied and bo, b, etc. are the least-squares estimates of the regression coefficients. In order that the regression coefficients can be evaluated, the effects of the variables xi on the yield Y, must be studied at different levels. Levels of the variables x 1 and x2 (pulse duration and percentage of cobalt respectively) have been selected as -1, 0 and 1.


The transforming x2 are as follows:


for obtaining

values of x1 and

Levels in coded form 0 +1

Variable SC,;pulse duration (~.ls) x,; cobalt in the carbide specimen 2(11-rti -In x1 = (In00

the coded


500) + 1 20)

500 20

(wt.%) x* =

Z(C0 - 20) 20 - 6

100 12

-1 20 6


where Co is the composition (wt.%) of cobalt in the carbide specimen. For a reliable estimate of the yield each trial was repeated twice. The adequacy of the proposed model was tested by comparing the variance (Table 2). The computed values of the regression coefficients are also given in this table. The regression equations are plotted in Figs. 7 - 9.

Fig. 7. Effect of pulse duration on MRR (experimental): ~~p, GT40.


GT-10; -- .--, GT-20;

Fig. 8. REW for three grades of carbide (experimental): pulse current, 4.5 A; tool, copper; dielectric, kerosene; - * -, GT-40; - - -, GT-20; -, GT-10.







Fig. 9. R,,, as a function of pulse duration (experimental): pulse current, 4.5 A; tool, copper; dielectric, kerosene; ---, GT-10; - - -, GT-20; - . -, GT-40.



5. Discussion of results Analytical results exhibiting the effect of carbide composition (grade) on metal removal and crater dimensions are given in Figs. 2 and 3. These results pertain to machining with a single spark using rectangular pulses. For the sake of computation it is assumed that of the total pulse energy delivered during a spark 40% is transmitted to the tool and 60% to the work. Differences in the machining behaviour exhibited by the three carbides in Figs. 2 and 3 are mainly due to differences in their chemical compositions, grain size and thermal properties. These can also be seen to influence the shape and size of individual craters (Fig. 4) and the REW (Fig. 5). It, is of interest to note that for low pulse durations GT-10 gives slightly higher metal removal as compared with GT-20 and GT-40, although the former possesses higher thermal conductivity and diffusivity. This trend, however, reverses if pulses longer than 75 ps are employed (Fig. 2). For GT-10 machining, if pulses longer than 75 PS are employed, enough time is available for heat dissipation to the interior of the electrode body on account of its superior thermal ch~ac~~stics. Therefore its erosion rate decreases when pulses of more than 75 PS are employed (Figs. 2 and 3). The REW and the thickness of the resolidified layer have also been found to depend upon the carbide grade (Figs. 4 - 6). In general, the resolidified layer thickness increases with pulse duration and depends upon the carbide grade as well (Fig. 6). ~xperimen~~y obtained response equations (eqn. (4) and Table 2) have been plotted in Figs. 7 - 9. Figure 7 reveals the nature of the relationship between the MRR and the pulse duration. It can be noticed that Figs. 2 and 7 agree to a certain extent qualitatively; however, a large discrepancy exists between the absolute values predicted by the two. This is on account of the inherent weakness of the single-spark model [ll] and to the fact. that the present analysis is unable to account for the influence of the metallurgical parameters of the work electrode. The higher rate of metal removal, observed for GT-10, at smaller pulse durations (t, < 75 PS) is attributed to the absence of titanium and tantalum in the work electrode. The metal carbon bond in such a composition is of a low energy type which, during sparking, leads to extensive dec~burization of the work surface. The erosion resistance of the carbide surface thereby diminishes and also leads to rough surface formation (Fig. 9). An increase in the erosion rate of GT-40 as the pulses become longer is due to a change in the mechanism of metal removal. GT-40 is much harder, more brittle and possesses higher hot hardness as compared with GT-10 and GT-20. For this composition, at higher pulse energies (equal to if X uf X ti), the total metal removal occurs owing to the combined effects of melting, evaporation and brittle fracture. The incidence of brittle fracture in the case of GT-40 would be higher because it contains TaC and a higher percentage of cobalt. Extensive evaporation of cobalt takes place (the melting point of cobalt is 1500 “C compared with the decomposition temperature of 2600 “C!for WC) during





Fig. 10. Scanning electron microscopy photographs of a machined specimen (top view), (a) GT-10, magnification 1095X. (b) CT-20, magnification 1095x (pulse duration, 20 /A). (c) GT-40, magnification 1095x (pulse duration, 20 11s). (d) GT-40, magnification 1186x (pulse duration, 100 MS).

a spark and as a result the carbide skeleton, which is highly brittle and susceptible to brittle failure [12], is left behind. The thickness of the resolidified layer and the incidence of surface cracking can be seen to depend upon the pulse duration and carbide grade (Fig. 10). Cracking is mainly due to steep temperature gradients, local expansion and contraction and rapid fluctuations in temperature. Based on the microscopic examination of the machined specimen it is possible to grade the carbides, from the cracking point of view, in the order GT-40, GT-20 and GT-10. The surface cracking susceptibility of GT-40 is higher mainly on account of the presence of TiC and a higher cobalt content which has a higher coefficient of thermal expansion compared with WC. This also results in a reduction in the transverse rupture strength, compressive strength, impact resistance, elastic modulus and thermal conductivity. The presence of a large percentage of cobalt in CT-40 releases a large volume of cobalt vapour and is responsible for pin holes, honeycombs etc. in the machined surface. The optimum values of pulse duration and cobalt percentage for achieving a maximum MRR and minimum REW have been computed using




pulse duration

Objective (optimum

function value)


= 2.684 (mg min-l) = 0.0031 (mm3 s-l) REW (2.386%)

and cobalt

percentage Optimum

for maximum


and minimum


Pulse duration


Co content


20 (CT-40)


6 (GT-10)

eqn. (4) and the data in Table 2. Rosenbrock hill climb algorithm duration and cobalt content are tion of 250 I.CSis obtained for the [ 51, alternatively, recommend 63 alloy with a Cu-W tool.



Optimized values were computed using the [13]. The optimized values of the pulse given in Table 3. An optimum pulse duracarbide GT-40 (Table 3). Watson and Freer /.B pulse for the machining of a 25 wt .% Co

6. Conclusions A two-dimensional heat source model is used to compute the metal removal per pulse, crater shape and the depth of the resolidified layer in EDM of a range of tungsten carbides with different chemical compositions and properties. It has been shown that the presence of cobalt has a significant influence on the machining behaviour of the carbides. Tungsten carbides with a high cobalt percentage (such as GT-40) when machined electrically are more susceptible to surface cracking and surface defects, such as pin holes and honeycombs, as compared with carbides with a low cobalt percentage (such as GT-10). From the viewpoint of maximum MRR and REW it has been found that pulse durations of 250 /.B and 467 ps respectively and cobalt binder percentages of 20% and 6% respectively are best suited. A carbide with a high cobalt content minimizes the resolidified layer thickness at lower pulse durations and yields a better surface finish. The REW in the case of GT-40 is usually higher than that for GT-10 or GT-20.

References 1 N. K. Poteev, G. V. Merkulov and I. N. Savelev, Manufacture of sintered carbide parts for press tools by electro-spark method, Electra-spark Machining of Metals, Vol. 3, Consultants Bureau, New York, 1965, p. 154. 2 A. I. Kruglov, I. P. Korobova and V. N. Pukin, A Thyratron pulse generator for electro-spark machining of sintered carbides, Electra-spark Machining of Metals, Vol. 3, Consultants Bureau, New York, 1965, p. 122.

88 3 S. S. Chetverikov and N. K. Poteev, Electra-spark machining of the cutting elements of sintered carbide blanking and piercing dies, Electra-spark Machining of Metals, Vol. 2, Consultants Bureau, New York, 1964, p. 85. 4 E. Lenz, E. Katz, W. Konig and R. Wertheim, Cracking behaviour of sintered carbides during EDM, Ann. CZRP, 24 (1) (1975) 109. 5 S. H. Watson and H. E. Freer, A comparative study of the electro-chemical and electro-discharge machining of a tungsten carbide-25% cobalt alloy, Znt. Symp. on Electromachining, Cracow, Poland, 1980, pp. 34 39. 6 Y. Kimoto, H. Maehata and K. Tamiya, Electrical discharge machining of cemented carbide alloys, BUZZ.Jpn. Sot. Precis. Eng., 12 (2) (1976) 89. 7 S. M. Pandit and K. P. Rajurkar, Analysis of electric discharge carbides,Ann. CZRP, 30 (1) (1981) 111.


of cemented

8 C. J. Heuvelman, Summary report on the CIRP co-operative research on spark machining of cemented carbides, Rep. presented to CIRP, 1980. 9 S. T. Jilani and P. C. Pandey, Analysis and modelling of EDM parameters, Precis.

Erg., 4 (2) (1982)



S. T. Jilani, Some investigations into the process of metal removal by EDM, Ph.D. Thesis, University of Roorkee, 1983. 11 J. R. Crookall, A basic analysis of pulse trains in electro-discharge machining, Int. J. Mach. Tool Des. Res., 13 (1975) 199. 12 J. M. Galimberti, Applications and limitations of carbide tools, American Society of Tool and Manufacturing Engineers, 1964, Paper 632. 13 C. S. Beightler, D. T. Phillips and D. J. Wilde, Foundations of Optimization. PrenticeHall, India, 2nd edn., 1982.

Appendix a

a’ c

z I

h lf k

m Q Ii t T T,

uf X,Y,Z P

A: Nomenclature

thermal diffusivity of the electrode material adapted thermal diffusivity specific heat of the electrode material adapted specific heat crater diameter crater depth mean pulse current thermal conductivity latent heat of fusion energy density radius radius of heat source time temperature in excess of ambient melting temperature mean voltage during sparking reference frame coordinates specific mass