The purpose of this research is to obtain the optimum cutting parameters to achieve the dimensional accuracy of Nimonic alloy miniature gear manufactured using Wire EDM. The cutting parameters investigated in this study are current, pulse on time (PON), pulse off time (POFF), wire tension (WT) and dielectric fluids. Ethylene glycol, nanopowder of alumina and oxygen are mixed to demineralized water to prepare novel dielectric fluids. Deviation in inner diameter, deviation in outer diameter, deviation in land and deviation in tooth width are considered to check the dimensional accuracy. Taguchi L16 is employed for experimental design and multiple response optimization is performed using Entropy TOPSIS and Pareto ANOVA. Results indicate that pulse on time is the most notable parameter for good dimensional accuracy followed by dielectric fluid, current, pulse off time and wire tension. Ethylene glycol mixed demineralized water is preferred for low dimensional deviation. The optimum WEDM parameters are pulse on time at 20
Among various micro parts, micro gear is the extensively used components in MEMS and NEMS technology, watches, turbines, pumps, harmonic drives, dental and medical devices, micro-motors, precision measuring instruments, and electronic home appliances, etc. [
Mainly, miniature products including mini gears are made from a variety of materials like aluminium, bronze, brass, nickel, steels, and titanium based alloys, etc. [
Dimensional accuracy (DA) plays crucial role in deciding the quality of die manufactured using Wire EDM [
It is important to note that more than one output responses are always involved when defining the quality of surfaces machined by WEDM. At times the effects of a single parameter on different output response are often contracting. This makes the multiple response optimization important in WEDM research. In recent researches, fuzzy logic multiple response decision making technique has become popular to optimize various manufacturing processes. Fuzzy-TOPSIS was applied to get the optimum responses like machining time, electrode wire wear rate, and dielectric fluid consumption during WEDM machining of steel [
Dielectric fluid helps to remove material and debris from surface after machining. It acts as an insulating medium between the wire and workpiece and generates required energy. Mostly traditional dielectric fluid such as Deionized water, kerosene, white spirit has been used as a dielectric but researchers have identified new additive mixed dielectric fluids as well. Mixture of Aluminum and SiC in Kerosene improves the surface finishing by enlarging the space between wire electrode and workpiece which scattered the spark energy. Researchers have demonstrated that the mixture of aluminium powder with kerosene is better than mixture of silicon carbide with kerosene [
In the maiden present work hybridization of dielectric by mixing additive such as metals and ceramic powder, gas and liquid form in basic dielectric “DM water” have not been investigated. The hybridization by mixing additives in all three physical phases solid, liquid, gas (such as alumina ceramic powder, ethylene glycol and oxygen) in the basic dielectric DM water has been carried out. A hybridization of such as kind is also one of the novelty characteristics of this work.
The purpose of the study is to find out the optimum cutting parameters to achieve improved DA. Deviation in inner diameter (DID), Deviation in outer diameter (DOD), Deviation in land (DLAND), and Deviation in tooth width (DTW) are the parameters to investigate the DA. The scheme for the final experiments was chosen and the final experiments were performed to analyze the influence of process parameters and different dielectric fluids on the DA. Entropy TOPSIS and Pareto ANOVA were applied for multiple optimization of cutting parameters.
Nimonic alloy is used to manufacture miniature gear of inner diameter of 5 mm, outer diameter of 7 mm and land of 0.5 mm. At present, Nimonic alloy is extensively used in many fields such as aerospace, transportation, and medical, etc. Main reason for extensive use of Nimonic alloys is its high temperature properties like wear resistance, strength and hardness, however it is difficult to cut. Further, little literature is available machining on Nimonic alloy for miniature gear. The chemical constituents of Nimonic alloy is shown in
Constituents | Al | C | Fe | Ni | Cu | V | Mn | Ti |
---|---|---|---|---|---|---|---|---|
Concentration | 0.004 | < 0.03 | 0.042 | 0.006 | 0.009 | 0.022 | 0.005 | Remaining |
Wire EDM machine DK7712 CNC of Steer Corporation was used in experimentation. Schematic diagram of Wire EDM is represented in
Model | DK7712 |
---|---|
Maximum cutting speed | 120 mm/min |
Maximum workpiece height | 2 mm |
Wire electrode material and diameter | Molybdenum, |
Machine dimension | |
Process usage | Metal-cutting CNC machine tools |
Movement method | Linear control |
Control method | Open-loop control |
Machine weight | 600 kg |
The dielectric fluid is one of the most important factors in WEDM which affects all the response factors. Usually WEDM employs demineralized water (DM) as dielectric. In the present work the dielectric fluid was used as a base-dielectric. Constituents such as, Ethylene Glycol (EG, an organic liquid) Al2O3 (alumina nanopowder) and oxygen gas were blended in the base-dielectric and four different blends of dielectric fluids were prepared as per the details given in
Symbols | Process parameters | Level 1 | Level 2 | Level 3 | Level 4 | |
---|---|---|---|---|---|---|
A | Current (A) | 1 | 2 | 3 | 4 | |
B | PON time ( |
10 | 15 | 20 | 25 | |
C | POFF time ( |
1 | 2 | 3 | 4 | |
D | Wire tension (N) | 6 | 10 | 14 | 18 | |
E | Dielectric fluid | DMwater |
DM |
All the hybrid dielectric fluids hold different dielectric strength and play significant role in spark gap and discharge energy. The dielectric strengths of DM water, oxygen, alumina nano powder and ethylene glycol are 70 kV/mm, 0.92 kV/mm, 15 kV/mm and 35 kV/mm respectively. Dielectric fluid which has high strength delivers high spark energy and the one with low strength deliver slow spark energy. After machining dimensions of the machined miniature gear were measured using optical projectors (diameter 300 mm) of Banbros, JT series as shown in
The WEDM process works melting evaporation and erosion by electro-thermal energy which removes the material by distinct sparks between the workpiece and electrode wire. The sparking takes place with a present periodicity controlled by spark on and spark off time. During spark on the dielectric breaks down and releases the energy resulting in material removal. During spark off the dielectric fluid flushes out the debris from the gap and also cools down the workpiece. Dielectric is an insulating medium which avoids electrolysis effect on the wire electrode during machining. For every pulse, discharge takes place at single point where wire electrode material releases the energy which melts and vaporizes the material. Thus, small crater forms on electrode and workpiece surface [
There are various methods to access the weight of decision making processes like eigenvectors method, entropy method, weighted least square method, and linear programming method for multiple dimensional analyses of preferences (LINMAP) [
The basic approach of TOPSIS methodology is the perfect alternative having minimum distance from positive ideal solutions (PIS) and maximum distance from negative ideal solutions (NIS) [
Entropy TOPSIS method is used in this work to obtain the optimum cutting parameters for improved DA. Since, DA is defined in terms of DID, DOD, DLAND and DTW. Entropy TOPSIS method transforms multiple DA measures viz. DID, DOD, DLAND and DTW into a single parameter Pi. Entropy method is used to determine the weights of the responses. Whereas, TOPSIS method is employed to compute the single replicated parameter Pi for DA. The computational steps involved in Entropy TOPSIS method are as follows:
Formulate a decision matrix by arranging the alternatives in rows and attributes in columns as shown in Normalize the decision matrix using Compute entropy of the attributes using where, Compute weight of the attributes using where, Calculate weighted normalized decision matrix as per where, Determine Positive and negative ideal solutions by using Calculate separation measures by using Calculate relative closeness from the ideal solutions as per Hence, Ranking the order on the basis of descending order of
Pareto ANOVA is mostly used to examine the data for process optimization [
The experiments were accomplished using L16 orthogonal array (OA) of the Taguchi’s design of experiment. Three replicates of each experiment were performed and average values of DID, DOD, DLAND, and DTW were recorded. The experimental data for deviation in inner diameter, outer diameter, and tooth width are shown in
Expt No. | A | B | C | D | E | DID (mm) | DLAND (mm) | DOD (mm) | DTW (mm) |
---|---|---|---|---|---|---|---|---|---|
1 | 1 | 1 | 1 | 1 | 1 | 0.223 | 0.097 | 0.228 | 0.005 |
2 | 1 | 2 | 2 | 2 | 2 | 0.2 | 0.077 | 0.241 | 0.041 |
3 | 1 | 3 | 3 | 3 | 3 | 0.182 | 0.062 | 0.198 | 0.016 |
4 | 1 | 4 | 4 | 4 | 4 | 0.202 | 0.060 | 0.231 | 0.029 |
5 | 2 | 1 | 2 | 3 | 4 | 0.465 | 0.004 | 0.591 | 0.126 |
6 | 2 | 2 | 1 | 4 | 3 | 0.212 | 0.063 | 0.262 | 0.05 |
7 | 2 | 3 | 4 | 1 | 2 | 0.228 | 0.064 | 0.238 | 0.01 |
8 | 2 | 4 | 3 | 2 | 1 | 0.178 | 0.027 | 0.243 | 0.065 |
9 | 3 | 1 | 3 | 4 | 2 | 0.228 | 0.062 | 0.259 | 0.031 |
10 | 3 | 2 | 4 | 3 | 1 | 0.157 | 0.013 | 0.205 | 0.048 |
11 | 3 | 3 | 1 | 2 | 4 | 0.185 | 0.056 | 0.244 | 0.059 |
12 | 3 | 4 | 2 | 1 | 3 | 0.158 | 0.011 | 0.209 | 0.051 |
13 | 4 | 1 | 4 | 2 | 3 | 0.2 | 0.079 | 0.245 | 0.045 |
14 | 4 | 2 | 3 | 1 | 4 | 0.178 | 0.100 | 0.201 | 0.023 |
15 | 4 | 3 | 2 | 4 | 1 | 0.141 | 0.024 | 0.187 | 0.046 |
16 | 4 | 4 | 1 | 3 | 2 | 0.199 | 0.071 | 0.214 | 0.015 |
Deviations in the inner diameter, land, outer diameter and tooth width are the factors that were taken into account to define the dimensional accuracy of the miniature gear. A single multi-response comprising of all key elements of miniature-gear geometry are investigated in this maiden work. Hence, Entropy-TOPSIS multiple response characteristics was applied to assess the deviation of gear and Pareto ANOVA analysis was conducted to obtain the optimum combination of parameters shown from
Expt No. | DID | DLAND | DOD | DTW |
---|---|---|---|---|
1 | 0.2532 | 0.3936 | 0.2145 | 0.0251 |
2 | 0.2271 | 0.3125 | 0.2267 | 0.2057 |
3 | 0.2067 | 0.2516 | 0.1863 | 0.0803 |
4 | 0.2294 | 0.2435 | 0.2173 | 0.1455 |
5 | 0.5280 | 0.0162 | 0.5560 | 0.6320 |
6 | 0.2407 | 0.2557 | 0.2465 | 0.2508 |
7 | 0.2589 | 0.2597 | 0.2239 | 0.0502 |
8 | 0.2021 | 0.1096 | 0.2286 | 0.3260 |
9 | 0.2589 | 0.2516 | 0.2437 | 0.1555 |
10 | 0.1783 | 0.0528 | 0.1929 | 0.2408 |
11 | 0.2101 | 0.2273 | 0.2296 | 0.2959 |
12 | 0.1794 | 0.0446 | 0.1966 | 0.2558 |
13 | 0.2271 | 0.3206 | 0.2305 | 0.2257 |
14 | 0.2021 | 0.4058 | 0.1891 | 0.1154 |
15 | 0.1601 | 0.0974 | 0.1759 | 0.2307 |
16 | 0.2260 | 0.2881 | 0.2013 | 0.0752 |
Terms | DID | DLAND | DOD | DTW |
---|---|---|---|---|
Ej | 0.1704 | 0.1625 | 0.1701 | 0.1600 |
Dj | 0.8296 | 0.8375 | 0.8299 | 0.8400 |
Wj | 0.2486 | 0.2510 | 0.2487 | 0.2517 |
Response | V+ | V − |
---|---|---|
DID | 0.0398 | 0.9418 |
DLAND | 0.0041 | 0.1019 |
DOD | 0.0438 | 0.1383 |
DTW | 0.0063 | 0.1591 |
Expt No. | Pi | Rank |
---|---|---|
1 | 0.6570 | 10 |
2 | 0.6353 | 12 |
3 | 0.7532 | 4 |
4 | 0.7167 | 7 |
5 | 0.3266 | 16 |
6 | 0.6288 | 13 |
7 | 0.7331 | 5 |
8 | 0.6596 | 9 |
9 | 0.6875 | 8 |
10 | 0.7670 | 1 |
11 | 0.6270 | 14 |
12 | 0.7541 | 3 |
13 | 0.6175 | 15 |
14 | 0.6385 | 11 |
15 | 0.7669 | 2 |
16 | 0.7201 | 6 |
Expt No. | A | B | C | D | E | Pi | S/N |
---|---|---|---|---|---|---|---|
1 | 1 | 1 | 1 | 1 | 1 | 0.6570 | −3.6485 |
2 | 1 | 2 | 2 | 2 | 2 | 0.6353 | −3.9406 |
3 | 1 | 3 | 3 | 3 | 3 | 0.7532 | −2.4614 |
4 | 1 | 4 | 4 | 4 | 4 | 0.7167 | −2.8936 |
5 | 2 | 1 | 2 | 3 | 4 | 0.3266 | −9.7193 |
6 | 2 | 2 | 1 | 4 | 3 | 0.6288 | −4.0304 |
7 | 2 | 3 | 4 | 1 | 2 | 0.7331 | −2.6971 |
8 | 2 | 4 | 3 | 2 | 1 | 0.6596 | −3.6150 |
9 | 3 | 1 | 3 | 4 | 2 | 0.6875 | −3.2548 |
10 | 3 | 2 | 4 | 3 | 1 | 0.7670 | −2.3040 |
11 | 3 | 3 | 1 | 2 | 4 | 0.6270 | −4.0544 |
12 | 3 | 4 | 2 | 1 | 3 | 0.7541 | −2.4509 |
13 | 4 | 1 | 4 | 2 | 3 | 0.6175 | −4.1867 |
14 | 4 | 2 | 3 | 1 | 4 | 0.6385 | −3.8972 |
15 | 4 | 3 | 2 | 4 | 1 | 0.7669 | −2.3053 |
16 | 4 | 4 | 1 | 3 | 2 | 0.7201 | −2.8516 |
Factors | Level 1 | Level 2 | Level 3 | Level 4 | Rank | |
---|---|---|---|---|---|---|
A | −3.236 | −5.015 | −3.016 | −3.31 | 1.999 | 3 |
B | −5.202 | −3.543 | −2.88 | −2.953 | 2.323 | 1 |
C | −3.646 | −4.604 | −3.307 | −3.02 | 1.584 | 4 |
D | −3.173 | −3.949 | −4.334 | −3.121 | 1.213 | 5 |
E | −2.968 | −3.186 | −3.282 | −5.141 | 2.173 | 2 |
Terms | A | B | C | D | E |
---|---|---|---|---|---|
1 | −12.944 | −20.809 | −14.585 | −12.694 | −11.873 |
2 | −20.062 | −14.172 | −18.416 | −15.797 | −12.744 |
3 | −12.064 | −11.518 | −13.228 | −17.336 | −13.129 |
4 | −13.241 | −11.811 | −12.081 | −12.484 | −20.565 |
Sum of sq of difference | 163.391 | 224.048 | 91.143 | 68.115 | 194.476 |
Contribution ratio (%) | 22.045 | 30.229 | 12.297 | 9.190 | 26.239 |
Max | −12.064 | −11.518 | −12.081 | −12.484 | −11.873 |
B | E | A | C | D | |
Cumulative contribution (%) | 30.229 | 56.468 | 78.513 | 90.810 | 100 |
Normalized decision matrix and
Weighted dimensionless matrix was obtained by the entropy methodology. Response variable weight for multiple responses is shown in
Then positive and negative ideal solutions were identified. Values of the distances from PIS and NIS have been obtained as shown in
It can be seen from
It may be noted that relationship formula of S/N ratio calculates its values which makes it independent of the characteristic type, i.e., whether it is lower-the-better or higher-the-better or nominal-the-better, maximum value of S/N ratio becomes the representative of desired performance. Consequently, it is evident from the results shown in
The dielectric fluid is the battery of electric power which is discharged as pulse energy during spark. In case of hybrid the nature of constituents such as oxygen and Al2O3 nanoparticles in the present case will alter the manner, this energy is dissipated. The presence of oxygen involves surface oxidation and adjustment of thermal expansions of constituents. The presence of oxygen, thus, creates the violent dissipation of energy and may increase the dimensional deviations. Similarly, the nanoparticles being solid insulating particles alter the average spark gap during travel of spark plasma between the electric poles. In the present case the pure DM water could be able to create a moderate spark and limit the deviation and enhanced the DA.
The current creates an effect of third largest contribution. The effect that a parameter creates on a response mainly depends on: (a) The range of parameters values and (b) The types of selected response. Further, the current is a factor which is responsible for the intensity of energy which is discharged in the spark gap. Under the range of parameters selected in this study the current at level 3, i.e., 3 A created optimum deviation which gave the best over accuracy of the multi-performance characteristics.
The fourth most significant parameter for dimensional deviation turns out to be POFF time. The DA is poor at low POFF time. As the POFF time increases, the dimensional deviation is lower because the pause of spark dictated by higher POFF the available time for the flushing-out of the debris is sufficient and the spark gap is properly cleared. The trapped debris being bad conductor also makes the spark erratic and leading to reduced accuracy. Thus, POFF time at level 4 gave best dimensional accuracy.
Among all the parameters chosen for the investigation in this work, WT is the only electrically neutral parameter. But is produces an indirect effect on the quality of machined surface. WT should be at optimum value to achieve better accuracy. Forces due to sparking act in the transverse direction on wire electrode, and this affects the dimensional inaccuracy in the manufacturing parts. The wire experiences transverse force every times the spark occurs. Under transverse forces, the wire vibrates and the frequency of vibration is proportional to the square root of wire tension. Here lies the main reason, that the wire vibrates following a wave-form, and at low tension the amplitude of vibrations is large. Large amplitude results in greater change in spark gap. Thus, every wire tension value results in a path-change of the spark front. This issue becomes compounded by all factors which influence the transverse force on the wire (e.g., peak current, type of dielectric, PON and POFF time, etc.). The wire tension at low level results in amplitude of vibration of wire to increase and significantly reduces the accuracy. As WT increases, deflection in wire reduce and frequency increases which consequently makes the path of spark front stabilized and results in improvement the DA.
The correctness of the obtained results is ensured by comparing the experimental value of the Pi with that of its corresponding predicted value at optimum combination. In order to estimate the predicted value of the Pi,
where, (
To calculate the experimental value of Pi, it was required to perform the experiment for these settings. However, Pi value cannot be determined by performing only one experiment. Therefore, to determine the experimental values of Pi at optimum combination, regression analysis was performed [
Further, a confirmatory experiment was performed by setting input parameters at the optimum combination and the values of the DA were obtained which were as follows:
Multiple response optimization for Wire EDM Nimonic alloy miniature gear was performed in this research. Taguchi L16 orthogonal array was applied for design of experiments and Entropy TOPSIS was performed for multiple optimization of parameters. Pareto plot was employed to indicate the cumulative percentage contribution of the parameters to improve dimensional accuracy. The cutting parameters included in this research were current, pulse on time, pulse off time, wire tension and dielectric fluids while deviation in inner diameter, deviation in outer diameter, deviation in land and deviation in tooth width were main responses. Results indicated that pulse on time was the highly significant parameter for dimensional accuracy followed by dielectric fluid, current, pulse off time and wire tension. The PON time supplies time for the steady discharge of pulse energy and is the most significant parameter. The PON time at level 3 provided sufficient pulse energy discharge time to develop optimum accuracy. The dielectric fluid was next higher parameter to affect dimensional accuracy significantly. The pure DM water created conditions favorable for the optimum accuracy with the selected range all the parameters. The current which is responsible for the intensity of energy discharge in the gap was the third most significant parameter and the current at level 3 (3 A) created condition for minimum deviation and highest accuracy. The pulse OFF time provides essential pause in the discharge and makes way for the cooling and flushing-out of debris and increase the accuracy. The POFF time at level 4 produced conditions for the best dimensional accuracy. The only non-electric parameters, i.e., wire tension affected the machining performance mainly because of the deflections caused by the transverse forces generated due to spark. At lower tensions the magnitude of deflections are large and cause greater variation in the dimensions of machined part. The wire tension at level 4 created minimum deviation and best accuracy. Based on the study, the optimum machining parameters are pulse on time at 20