Experiment Assessment and Adam Gene Algorithm To Optimize The Process Parameter of Wire Spark Erosion Material Removal While Machining Nickel Alloy
Experiment Assessment and Adam Gene Algorithm To Optimize The Process Parameter of Wire Spark Erosion Material Removal While Machining Nickel Alloy
Experiment Assessment and Adam Gene Algorithm To Optimize The Process Parameter of Wire Spark Erosion Material Removal While Machining Nickel Alloy
https://doi.org/10.1007/s41939-023-00343-w
ORIGINAL PAPER
Received: 11 October 2023 / Accepted: 10 December 2023 / Published online: 11 January 2024
© The Author(s), under exclusive licence to Springer Nature Switzerland AG 2024
Abstract
The nickel-based superalloy, CM247 is machined using wire electro spark material removal process with three different levels
of process parameters. The material removal rate and the surface finish are evaluated individually to assess the influence
of process parameters. Maximum material removal is noticed with increase in pulse on duration and superior surface finish
with average pulse on time. At 30 V and 120 µs in pulse on time, maximum material removal is noticed and better surface
roughness as 2.211 µm. The machinability of the material depends on the duration of electro spark intensity generated during
machining. Pulse on time has a maximum influence (53%) on surface roughness followed by the applied voltage (30%). In
material removal, the applied voltage (87%) is a predominant factor. A modified genetic algorithm, mimic of iconic pair Adam
and Eve is simulated to study the varying populations. From the optimization technique, a globally best process parameter
is identified for the electro spark material removal of CM247 alloy. The optimal process condition is 18 V and 126 µs for
applied voltage and pulse on time.
123
2394 Multiscale and Multidisciplinary Modeling, Experiments and Design (2024) 7:2393–2400
hard and dense metallic materials is the electro spark mate- 2 Experimental procedure
rial removal technique (Adam Khan and Gupta 2020). It is
challenging to make predictions from the trial on process The CM247 nickel-based superalloy is the work material
parameter selection for wire electro spark material removal. used to machine using wire electro spark material removal
The need for search algorithms to forecast the ideal process process. The material composition identified through the
parameters has increased. optical emission spectroscopy for the work material CM247
Different types of statistical and heuristic techniques are is as follows: Ni—62.2%, W—9.3%, Co—9.1%, Cr—8.2%,
available to predict the machining process parameter. The Al—5.5%, Ta—3.0%, Hf—1.3%, Ti—0.7%, Mo—0.6%,
constraints and process requirements are taken into con- and C—0.07%. The material is in the form of rod with a
sideration when choosing these optimization strategies. In diameter of 10 mm and 150 mm length. The machining
general, statistical analysis tools are based on the process con- studies are performed with Electronica wire cut discharge
ditions [12.13] which include the Taguchi methodology, Gray machine using 0.3 mm copper wire with dielectric fluid.
Relational Analysis (GRA), TOPSIS, MOORA, Hook & The work material CM247 alloy is high in hardness (379
Jeeves, etc. These methods should result in a large number of HV) and density (8.62 g/cc) to face engineering applications.
test experiments and trail experiments. In contrast, artificial This material has been developed to replace the existence of
neural network (ANN) approaches, particle swarm optimiza- Inconel718 and to compete the shortfall in high-temperature
tion (PSO), genetic algorithms (GA), simulated annealing applications. Especially, CM247 is designed and developed
(SA), and genetic algorithms (GA) are utilized as search algo- for high-temperature applications such as gas turbine, ther-
rithms. Literatures and research monographs are enormous to mal combustors, and other high-temperature applications.
discuss more on process optimization process (Chakraborty The experimental investigation on CM247 has not been
and Chakraborty 2022). Genetic algorithm has been highly initiated in the current scenario and it cannot be compared
recommended for machining process parameter optimiza- with commercial nickel alloy as the properties are bet-
tion with constraints or unconstraint’s in objective functions. ter. Hence, a basic study has been performed before the
As the search space are widely used to find the globally actual experimentation. A set of process parameters has been
best parameter from the proposed process conditions. Adam identified from the literature and it has worked for some
et al. made an attempt to modify the conventional genetic trail logics. A thorough investigation and process evalua-
algorithm with the base of Noah’s Arc concept in parent chro- tion are made with the expertise to reach the outcome. As
mosomes swapping techniques (Adam Khan et al. 2021). It is an outcome the input process parameters are framed for the
a technique with varying in population size mimic to nature proposed research investigation. Application voltage (10, 20,
of a human life. From the existing research, it can be under- and 30 V), pulse on time (110, 115, and 120 s), and pulse
stood that the machining of hard material and its process of time (50, 55, and 60 s) are the process parameters. The
evaluation are very much commercialized (Alfattani et al. set of experiments planned for the spark erosion is as given
2023; Selvarajan et al. 2023a, b; Selvarajan and Venkatara- in the Table 1. Throughout the cutting operation, potential
manan 2023). However, the demand for advanced and novel (230 V) and current (12 A) are kept constant and ensured as
technique is still open to researchers. the energy input for the EDM machine. The hard brass wire
From the literature, it has to be confirmed that the inves- with a combination of copper, zinc, and molybdenum mate-
tigation on CM247 nickel-based superalloy has not been rial in the form of wire with a uniform diameter of 0.3 mm is
initiated. These materials are very rare and the novelty of used as an electrode / tool for spark erosion machining pro-
the material has to been highlighted as a research gap in cess. The quality of the electrode refers to a good conductor
this paper. Simultaneously, the machining studies and wear and high in tension to substantiate the wire feed. The cut-
analysis are the new features added to this manuscript. The ting wire travels at speeds between 3 and 1 mm/min, while
nickel-based superalloy is machined using an electro spark maintaining a constant tension of 7 kg/mm throughout the
material removal process with different process parameters. entire procedure. The machined surface is used to measure
Additionally, it is suggested to statistically examine the pro- the roughness range using roughness tester (Model: Mitu-
cess parameters and use the Adam–Eve genetic algorithm toyo Surftest SJ410 with an accuracy of 0.01 µm) with a
to optimize the process condition. To assess the process cut-off length of 8 mm and the material removal rate are
condition, surface roughness and material removal rate are calculated. The machined surface is observed through scan-
recommended. ning electron microscope (Model: ZEISS EVO 18) to study
the wear morphology. The designed experiments are statisti-
cally evaluated through a MINITAB software to identify the
contribution of input process parameters. A linear regression
technique is used to identify the goodness of fitness and Tay-
lors equation. To predict the optimal process parameter, a
123
Multiscale and Multidisciplinary Modeling, Experiments and Design (2024) 7:2393–2400 2395
Table 1 Experimental plan for the proposed spark erosion method The process depends entirely on the pulse on and / or
off for electro spark generation. In this research, the mate-
Exp No. Voltage Ton Toff
rial removal was found to be maximized in a steady state
1 10 110 50 with increase in pulse on time. This indicates that the effi-
2 10 115 55
ciency of the plasma spark has highly supported for the
material removal in large scale. However, this will be inter-
3 10 120 60
rupted against the pulse off time as designed for machining
4 20 110 50
experimentation. The increase in pulse of time will pull the
5 20 115 55
material fusion and evaporation during machining process;
6 20 120 60 subsequently, the material removal has found fluctuating and
7 30 110 55 has reduced to some extent. As a result, the distraction in
8 30 115 60 material removal over the machined surface may lead to some
9 30 120 50 defects in terms of metallurgical aspects. Due to thermo-
10 10 110 60 metallurgical processes, it has a direct impact on surface
11 10 115 50 roughness with craters and valleys as well. The surface rough-
12 10 120 55 ness of the machined samples were measured with respect
13 20 110 55 to applied voltage and pulse duration. Figure 2 shows the
14 20 115 60 average surface roughness of the sample, measured for three
15 20 120 50
samples against their experimental design. The spark pulse
off time on machining has high impact over the surface rough-
16 30 110 60
ness. For minimum pulse off time (at 50 µs), the surface
17 30 115 50
roughness is less and it is in the range of 2.2–2.45 µm. Dur-
18 30 120 55
ing the same machining process, with increase in pulse off
time, the distraction or shortened pulse duration has highly
influenced the machined surface to cause rough layers. Sim-
modified genetic algorithm is used to find the globally best
ilarly, the surface roughness has found to be less for average
process parameter from the designed process parameters.
voltage (20 V) within the range of 2.2–2.4 µm. Wide vari-
ation in surface roughness has been recorded for different
pulse on time. The significant reason for changes in surface
3 Machinability of superalloy roughness is due to pulse duration and wire feed rate.
The electron image of the machined surface is shown in
From the experimental investigation, the material removal
Fig. 3 for two different voltages. While increasing the pulse
rate is calculated for each machined specimen with a repeata-
on time, the intensity of the spark was found to be increased
bility of three samples using the standard mathematical
to cover machining for a maximum surface area. Due to the
relation as given in Eq. 1.
break in electro spark in terms of pulse of time, the efficiency
Ax W of cutting is forced to stop. As a result, surface is saturated
MRR (mm3 /min), (1) with partial melted metal splats which lead to production of
M
maximum recast layer. Since, the machined surface with high
where MRR is the material removal rate, A is the machined pulse on time has maximum wear scar, craters, and recast
surface area (mm2 ), W is the width of the cut (mm), and M layers in the surfaces. With respect to the pulse off time
is the machining time (in min). Calculated values are given for same voltage, the surface morphology has found to be
in Fig. 1; with respect to the pulse duration. varied. Increase in pulse off time reveals with minimum sur-
The material removal tends to follow increasing trend face craters and recast layers. However, for minimum applied
with maximum pulse on time and applied voltage. It directly voltage (10 V) at same pulse off time, the inefficiency of the
relates with the increase in voltage and prolong pulse dura- plasma spark has led to rough and aggressive surface with
tion. The efficiency of electro spark has supported to remove recast layers and craters. The micro structure developed on
maximum material. The duration of the electro spark inten- the machined surface are in the form of spheroids. The micro
sity is short for minimum pulse on, removing the material structural change is observed as a result of partially melted
with a smaller volume. It has been understood that the layer bulk which is caused by the sudden solidification of
material removal has gradually increased to a maximum of fused metal. While machining the alloy, the spark developed
1.169 mm3 /min with maximum pulse on time 120 µs at 30v. between the electrodes (work piece and wire) diffuse the
Similarly, the interruption in pulse off time has also influ- metal followed by the flushing of fused metal using elec-
enced the volume of material removal. trolyte. The spark intensity decides the rate of metal fusion
123
2396 Multiscale and Multidisciplinary Modeling, Experiments and Design (2024) 7:2393–2400
Fig. 1 Material removed calculated on electro spark machining with respect to pulse duration and applied voltage
3.000 3.000
10v 20v 30v
10v 20v 30v
2.800 2.800
2.600 2.600
2.400 2.400
2.200 2.200
2.000 2.000
50 55 60 110 115 120
Pulse Off Time (µs) Pulse On Time (µs)
Pulse off time Vs Voltage Pulse on time Vs Voltage
Fig. 2 Surface roughness measured on electro spark material removal with respect to pulse duration and applied voltage
and surface quality. Distraction in pulse duration as observed Literature proved the justification for material removal con-
in the surface roughness decides the quality of the machined dition with spark intensity has derived the track for any form
samples. of metals (Adam Khan et al. 2020). An empirical equation
for the wire spark material removal process is developed
using Gaussian elimination method considering the predom-
inant process parameters. It is used to identify the response
4 Statistical analysis on process parameters [Z C.X α .Y β ] in terms of influencing parameters. The
equations developed for the response on surface roughness
The experimental data obtained from the investigations are (Ra) and material removal rate (MRR) are given in the Eqs. 2
statistically evaluated through analysis of variance (ANOVA) and 3.
using Minitab software. The analysis is considered for min-
imum surface roughness and the reduced materials removal. Ra 7.501V −0.021 Ton −0.225 , (2)
The response on surface roughness and material removal
rate are taken into consideration together with three input M R R 0.017V 0.193 Ton 0.61 . (3)
process parameters such as applied voltage, pulse on time,
and pulse off time. From the analysis, the contribution of The outcome of the equation is evaluated and the vari-
input process parameters is statistically evaluated. Figure 4 ations in the results are established through the statistical
illustrates the contribution of input process parameters. The analysis. It has been confirmed that the fitness of data is
pulse on duration has highly influenced (53%) toward the in line with the experimental range with the confident level
surface roughness followed by the applied voltage (30%). of adjacent square (R2 ) as 96.54% (for Ra) and 94.1% (for
Subsequently, for material removal rate the applied voltage MRR), respectively. In addition, the Minitab software is used
is the predominant factor (87%) and the pulse on dura- to study the interaction of input process parameter in rela-
tion (2%), respectively. It has been evidently proved that tion to the responses on machining process. Interaction plot
the spark voltage has a significant impact on the fusing of for the surface roughness in given in the Fig. 5a–c. The
materials, and that electrolyte flushing removes the fuses. interrelation between the input process parameter for surface
123
Multiscale and Multidisciplinary Modeling, Experiments and Design (2024) 7:2393–2400 2397
Fig. 3 Electron image of machined surface at different voltage at a constant pulse duration
100 Fig. 5b. Similar results were revealed for material removal
90 Surface Roughness rate in connection to input process parameters. For material
Contribution percentage (%)
123
2398 Multiscale and Multidisciplinary Modeling, Experiments and Design (2024) 7:2393–2400
30 30
Ra MRR
< 2.3 < 0.8
2.3 – 2.4 0.8– 1.0
2.4 – 2.5 1.0– 1.2
2.5 – 2.6 1.2– 1.4
25 > 2.6 1.4– 1.6
25
> 1.6
Volt
20
Volt
20
15 15
10
110.0 111.5 113.0 114.5 116.0 117.5 119.0 10
110.0 111.5 113.0 114.5 116.0 117.5 119.0
Ton
Ton
20
Volt
20
15
15
10
50 52 54 56 58 60 10
Toff 50 52 54 56 58 60
Toff
(b) Volt Vs Toff
(b) Volt Vs Toff
Ra
< 2.3
119.0 MRR
2.3 – 2.4
< 0.8
2.4 – 2.5 119.0 0.8– 1.0
2.5 – 2.6
1.0– 1.2
117.5 > 2.6
1.2– 1.4
117.5 1.4– 1.6
> 1.6
116.0
116.0
Ton
114.5
Ton
114.5
113.0
113.0
111.5
111.5
110.0
50 52 54 56 58 60
Toff 110.0
50 52 54 56 58 60
bio-mimic algorithms, the chromosomes / populations are this algorithm, the chromosomes generated or born during
handled in different form to increase the efficiency of the crossover are limited to sustain (to live) for three generations
algorithm. One of the concepts in handling the chromosomes only. The chromosomes designed in the Adam gene algo-
/ population is variation in size (number of chromosomes / rithm is given in the form of population separation (Fig. 7).
populations) defined in the multi-population in genetic algo- The Adam gene algorithm is similar to Genetic Algorithm
rithm. The development of the population variation in genetic with varying population size (GAVsPS), with a control in
algorithm has reached to expand as Adam–Eve Gene Algo- population through the death factor. It will control the pop-
rithm (Alfattani et al. 2023; Selvarajan et al. 2023a). The ulation by removal of chromosomes after three consecutive
factors in Adam gene algorithm is slightly modified from generations. The objective function used for this algorithm
conventional genetic algorithm. Some of the highlights are is minimization of the function given in Eq. 5.
as crossover (reproduction) from single pair randomly gen-
erated chromosomes, mutation in chromosomes and death
of chromosomes after a defined number of generations. In f (x) 0.5Ra + 0.5M R R, (4)
123
Multiscale and Multidisciplinary Modeling, Experiments and Design (2024) 7:2393–2400 2399
123
2400 Multiscale and Multidisciplinary Modeling, Experiments and Design (2024) 7:2393–2400
1.500 Adam Khan M, Gupta K (2020) On heat treatment and surface char-
acterization of spark eroded nickel-based superalloy developed by
1.480 additive manufacturing. Trans Indian Inst Metals 73:429–439
Alfattani R, Yunus M, Selvarajan L, Venkataramanan K (2023) Spark
Fitness value f(x)
123