Investigation of the Impact of Intensive EDM Regimes on Manufacturing Efficiency and Surface Quality of C120 Steel Parts

Abstract

The emergence of new hard and extra-hard materials has led to the development of new technologies capable of processing them, known as unconventional technologies. Electrical discharge machining (EDM) is a very common unconventional technology in the manufacturing industry, used to process special materials. The primary benefit is the ability to machine various complex shapes at a reduced cost. This study addressed the use of intensive machining regimes that would enhance productivity while also maintaining a high quality of the resulting surface. The experimental setup was designed according to a D-optimal response surface method, and the results were statistically processed using ANOVA. The results revealed that it is possible to achieve both high productivity and good surface quality, but it was also found that increasing the processing parameters is feasible only to a certain extent.

1. Introduction

One of the seventeen Sustainable Development Goals is to promote sustainable industrialization and encourage innovation (Goal nine). A competitive and sustainable industry plays an essential role in accelerating economic growth, reducing poverty through productive activities, and achieving all the other goals set out in the 2030 Agenda [1]. From this perspective, in line with the EU’s industrial policy strategy [2], it is necessary to support the strengthening of value chains and the implementation of the most efficient technologies, promote the circular economy and competitiveness, encourage industrial trade, and develop the private sector, agro-industries, and renewable energies.

Industry has an important role to play in what is proving to be both the greatest challenge and the greatest opportunity of our time. All industrial value chains, including energy-intensive sectors, will have a key role to play. They will need to work to reduce their own carbon footprints but also to accelerate the transition by providing clean and affordable technological solutions and developing new business models [3].

Issues associated with industry can be addressed, from a sustainable development perspective, through sustainable production strategies. In practice, sustainable production means “creating goods and services using processes and systems that are non-polluting, conserve energy and natural resources in economically viable ways, are safe and healthy for employees, communities, and consumers, and are socially and creatively appropriate for all stakeholders in the short term, but especially in the long term” [4]. In view of this, advanced manufacturing technologies (unconventional technologies) have emerged, which attempt to solve typical problems where traditional techniques have proven ineffective.

Electrical discharge machining (EDM) belongs to this category, being able to provide industry with cost-effective solutions for manufacturing high-value-added components for key sectors such as tooling and precision machining, mold and die making, aerospace, medical implants and devices, microelectronics manufacturing, etc. EDM using a solid electrode is one of the most effective technologies in the mold and die making industry due to its ability to produce blind cavities with intricate shapes in hard and extra-hard materials. It is also used in the aerospace and automotive industries to manufacture high-precision components from difficult-to-cut materials [5,6].

The working principle is based on the erosive effect of high-frequency electrical discharges (sparks) between the tool (electrode) and the workpiece, which are connected to the poles of a DC source [7,8]. Machining efficiency and efficacy depend on many factors, categorized into electrical variables (e.g., peak current, pulse-on time, pulse-off time, polarity) and non-electrical variables (e.g., properties of electrode and workpiece material, electrode geometry, dielectric type, etc.) [9,10]. For EDM to be a sustainable process, it is crucial to identify an optimal combination of these variables to ensure, concomitantly, high productivity (i.e., high material removal rates, MRR), reduced electrode wear (i.e., tool wear rate, TWR; electrode wear ratio, EWR), and high quality of the machined surface (i.e., low surface roughness; low surface crack density, reduced heat affected zone) [11,12].

Table 1. Machining regimes used for improving productivity of die sinking EDM process and quality of machined surface.

Ref.	Workpiece Material	Electrode Material	Study Conditions	Outcomes
[13]	AISI P20 steel	Copper	Varied parameters: Peak current, Ip (A): 2; 4, 6 Pulse-on time, Ton (μs): 60; 90; 120 Pulse-off time, Toff (μs): 30; 60; 90 Positive polarity Methods: Taguchi and Analysis of Variance (ANOVA)	MRR; EWR; Surface Roughness (SR). Optimum parameters:
				Findings	Ip		Ton		Toff
				MRR	6		120		30
				EWR	2		120		90
				SR	2		60		30
				Factors’ influence: I > Ton > Toff
[14]	90MnCrV8 steel	Copper	Varied parameters: Ip (A): 5; 8; 11 Ton (μs): 50; 100; 200 Toff (μs): 6.4; 13; 25 Methods: Taguchi and ANOVA	MRR; SR; White layer thickness (WLT).
				Findings	Ip		Ton		Toff
				MRR ↑	↑		↓		↓
				SR ↓	↓		↓		↑
				WLTh ↓	↓		↓		↑
				Factors’ influence: I—the most significant parameter with contributions of 49.56%, 69.37%, and 51.83% for MRR, SR, and WLT, respectively
[15]	AISI D2 steel	Copper	Varied parameters: Ip (A): 9; 12; 15 Dielectric type: Distilled water; Kerosene oil; Transformer oil Spark Gap (mm): 2; 4; 6 Electrode polarity: positive; negative Methods: response surface methodology (RSM), grey relational analysis (GRA) and ANOVA	MRR and SR
				Optimum machining parameters
				Ip	Dielectric type		Spark gap		Polarity
				15	kerosene		6		positive
				Optimum outputs
				MRR	17.23 mm3/min
				SR	3.86 µm
				Factors’ influence: Polarity > Spark gap > Peak current > Dielectric type
[16]	AISI 304 stainless steel	Tungsten	Varied parameters: Ip (A): 5; 7; 9 Ton (μs): 50; 150; 200 Gap voltage, V (V): 45; 55; 65 Methods: RSM, Biogeography-Based Optimization (BBO) and Ant Colony Optimization (ACO)	MRR and SR
				Optimum machining parameters
					Ip		Ton		V
				BBO	9		200		45
				ACO	8.97		194.33		44.65
				Maximum MRR
				BBO	9.481 mm3/min
				ACO	9.232 mm3/min
				Optimum machining parameters
					Ip		Ton		V
				BBO	5		50		65
				ACO	5.2		52.8		64.8
				Minimum SR
				BBO	2.582 μm
				ACO	2.966 μm
[17]	X210 steel	Copper	Varied parameters: Ip (A): 15; 20; 25; 30; 35 Ton (μs): 75; 150; 225; 300; 375 Toff (μs): 20; 50; 80; 110; 140 Angle of electrode (deg): 0; 22.5; 45; 67.5; 90 Methods: RSM, ANOVA	MRR, TWR, and WLT
				Optimum machining parameters
				Ip	Ton		Toff		Electrode angle
				30	300		110		67.5
				Maximum MRR: 6.465 mm3/min
				20	300		50		22.5
				Minimum TWR: 0.0112 mm3/min
				Minimum WLT: electrode angles of 45 and 15
[18]	HCHCr die steel	Copper	Varied parameters: Ip (A): 5; 6; 7 Ton (μs): 9; 49; 99 Toff (μs): 2; 6; 9 Dielectric powder: Zr, Ni, Ni + Zr Methods: Taguchi L9 OA and Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS)	MRR and SR
				Optimum machining parameters
				Ip	Ton		Toff		Dielectric powder
				7	9		2		Ni + Zr
				Maximum MRR: 28.608 mm3/min Minimum Ra: 4.845 μm
[19]	AISI D2 steel	Copper, cryogenic and non-cryogenic treated (CTE and N-CTE)	Varied parameters: Ip (A): 7; 11; 15 Ton (μs): 400; 600; 800 Toff (μs): 45; 80; 150 Method: Taguchi L9 orthogonal array (OA)	MRR and EWR
				Optimum machining parameters
				Ip	Ton		Toff
				15	600		45
				Maximum MRR
				CTE	0.42152 mm3/min
				N-CTE	0.33920 mm3/min
				Optimum machining parameters
				Ip	Ton		Toff
				7	400		45
				Minimum EWR
				CTE	0.00076 mm3/min
				N-CTE	0.00078 mm3/min
[20]	AISI H-13 tool steel	Copper	Varied parameters: Ip (A): 1; 3; 5 Ton (μs): 200; 300; 400 Electrode thickness (mm): 5.2; 6.2 Method: Taguchi L18 OA	MRR, over cut (OC)
				Optimum machining parameters
				Ip	Ton		Electrode thickness
				1.4	200		6.2
				Optimum outputs
				MRR	12.254 mm3/min
				OC	0.005 mm
[21]	AISI-D6 steel	Copper	Varied parameters: Ip (A): 8; 10; 12; 14 Ton (μs): 10; 20; 30; 40 V (V): 150; 250 Methods: ANN and adaptive neuro-fuzzy inference system (ANFIS)	MRR, TWR, Ra
				Findings	Ip		Ton		V
				MRR ↑	↑		↑		↓
				TWR ↓	↓		↑		↑
				Ra ↓	↓		↓		↑
				ANFIS is superior to ANN in terms of lower RMSE
[22]	17-7 PH stainless steel	Copper	Varied parameters: Ip (A): 8; 14; 20 Ton (μs): 400; 500; 600 V (V): 40; 50; 60 Inter Electrode Gap, IEG (μm): 100; 150; 250 Method: Taguchi L9 OA	MRR, TWR, Ra, OC, Clearance
				Findings	Ip		V		IEG	Ton
				MRR ↑	20		50		150	400
				TWR ↓	14		50		150	400
				Ra ↓	8		40		150	600
				OC ↓	8		40		150	600
				Clearance ↓	20		60		250	600
[23]	Eglin steel	Tungsten	Varied parameters: Ip (A): 10; 20; 30 Ton (μs): 100; 200; 300 V (V): 40; 50; 60 Method: ANN	MRR
				Optimum machining parameters
				Ip	Ton		V
				30	300		50
				Maximum MRR: 15.68 mm3/min
[24]	304L stainless steel	Copper	Varied parameters: Ip (A): 20; 40 Ton (μs): 50; 250 Toff (μs): 25; 125 Mix of powder (% SiO2): 30; 70 Powder concentration, PC (g/L): 1; 5 Method: RSM	MRR, SR, EWR
				Findings	Ip	Ton	Toff	PC	% SiO2
				MRR ↑	40	250	25	5	30
				SR ↓	20	250	25	1	30
				EWR↓	20	250	125	1	70
[25]	OHNS tool steel	Not specified	Varied parameters: Ip (A): 5; 10; 15 Ton (μs): 50; 75; 100 V (V): 40; 45; 50 Duty factor (Tau): 50; 66.5; 83 Methods: RSM, ANN	MRR
				Optimum machining parameters
				Ip	V		Ton		Tau
				15	40		100		50
				Maximum MRR: 1.272 mm3/min
[26]	SDK11 die steel	Copper	Varied parameters: Ip (A): 1; 2; 3; 4; 5 Ton (μs): 18; 25; 37; 50; 75 Toff (μs): 9; 12; 18; 25; 37 V (V): 30; 40; 50; 60; 70 Method: Taguchi—AHP—Deng’s similarity method	MRR, TWR, SR, HV (hardness of machined surface), WLT
				Optimum machining parameters
				Ip	Ton		Toff		V
				5	25		18		60
				Maximum MRR: 24.68 mm3/min Minimum SR: 1.84 μm Minimum TWR: 0.22 mm3/min Minimum HV: 781.98 Minimum WLT: 8.839 µm
[27]	7Cr13Mo steel	Copper	Varied parameters: Ip (A): 1.5; 4.5; 6; 9 Ton (μs): 15; 60; 90; 120 Method: experimental tests, ultrasonic-assisted and powder-mixed electrical discharge machining (US-PMEDM); no optimization	MRR, SR, micro-hardness Findings: US-PMEDM process leads to better surface integrity; SR increases when Ip and Ton increase; MRR of US-PMEDM increased with 57%; Ip has the highest influence on MRR Micro-hardness was improved by 78% in US-PMEDM
[28]	AISI 304 stainless steel	Copper	Varied parameters: Duty factor (Tau): 2; 4; 6; 8; 10; 12 MoS2 powder size: 40 μm; 90 nm Method: experimental tests, no optimization	Discharge energy, MRR, Ra Findings: Discharge energy increases as duty factor increases; MRR is higher when nano powder mixed dielectric is used for Tau equal to 2; 4; 10 and 12; Micro powder offered better surface finish compared to nano powder, irrespective of the duty factor value
[29]	SKD61 die steel	Copper	Varied parameters: Ip (A): 6; 8; 10 Ton (μs): 100; 150; 200 V (V): 60; 75; 90 Method: Data Envelopment Analysis-based Ranking (DEAR)	MRR, Ra
				Optimum machining parameters
				Ip	Ton		V
				10	100		90
				5% higher MRR with better surface finish
[30]	55NiCrMoV7 tool steel	Graphite (EDM-3 POCO)	Varied parameters: Ip (A): 3; 8.5; 14 Ton (μs): 13; 206; 400 Toff (μs): 9; 80; 150 Method: RSM	MRR, Sa, WLT
					Optimum machining parameters
				Finishing	Ip		Ton		Toff
					3		176		10
					MRR: 1.06 mm3/min; Sa: 1.8 μm; WLT: 6.3 μm
					Optimum machining parameters
				Semi-finishing	Ip		Ton		Toff
					14		52		24
					MRR: 15 mm3/min; Sa: 5.4 μm; WLT: 15.8 μm
					Optimum machining parameters
				Roughing	Ip		Ton		Toff
					14		361		24
					MRR: 28.1 mm3/min; Sa: 12.7 μm; WLT: 30.5 μm
[31]	SKD61 die steel	Copper	Varied parameters: Ip (A): 3; 6; 8 Ton (μs): 12; 25; 50 Toff (μs): 5.5; 12.5; 25 Frequency vibration, F (Hz): 128; 256; 512 Method: Multi-objective optimization based on ratio analysis (MOORA)	MRR, TWR, SR
				Optimum machining parameters
				Ip	Ton		Toff		F
				8	25		5.5		512
				Maximum MRR: 9.564 mm3/min Minimum TWR: 1.944 mm3/min Minimum SR: 3.24 μm
[32]	AISI 304 stainless steel	Tungsten carbide, brass, copper	Varied parameters: Ip (A): 9; 12; 15 V (V): 40; 60; 80 Duty factor (Tau): 0.4; 0.6; 0.8 Method: TOPSIS	SR, WLT, residual stress
				Optimum machining parameters
				Ip	V		Tau		Electrode
				15	80		0.6		WC
				Ip has the highest influence on the outputs

presents machining regimes used by different researchers to improve the performance of the EDM process, expressed in terms of productivity and surface quality.

One should note that choosing the optimal regimes that allow for the simultaneous achievement of high productivity and high quality is not an effortless task. Moreover, there is no universally valid recipe for success, but rather the results strongly depend on each process setup. It should also be noted that the values used for the process parameters by different researchers as presented in the literature are specific to more “conservative” machining regimes (e.g., Table 1). The purpose of the current study was to test process performance under intensive machining regimes, specifically to achieve a reduction in machining time (indicating high productivity) while maintaining high quality in the finished surfaces, which would enhance the sustainability of the process. The results presented in this paper are part of a more extended study that tries to improve the efficiency of the process from a quality and cost point of view, and to make it more environmentally friendly: higher productivity, lower tool consumption, increased part quality, energetic efficiency, greener dielectrics.

2. Materials and Methods

2.1. Experimental Conditions and Processed Material

The experiments consisted of machining blind holes with a depth of 2 mm on a KNUTH FEM 110 CNC EDM machine (KNUTH Werkzeugmaschinen GmbH, Wasbek, Germany), in reversed polarity, using square cross-section electrodes (10 × 10 mm) made of three different materials (pure and alloyed copper). The chemical composition and mechanical properties of the electrodes are presented in

Table 2. Chemical composition of the electrodes’ material.

Electrolytic Copper Electrode (Cu 99.9)
Cu (%)		O (%)				Pb (%)				Bi (%)
99.9		0.04				0.005				0.0005
Tungsten–Copper Electrode (WCu 75/25)
Cu (%)				W (%)				Aditiv (max.%)
25 ± 2				difference				1
Brass Electrode (CuZn39Pb2)
Fe (%)	Ni (%)		Al (%)		Cu (%)		Pb (%)		Sn (%)		Others (%)
max 0.3	max 0.3		max 0.05		59–60		1.6–2.5		max 0.3		total 0.2

and

Table 3. Properties of the electrodes’ material.

Electrode Material		Electrolytic Copper (Cu 99.9)	Tungsten–Copper (WCu 75/25)	Brass (CuZn39Pb2)
	Properties
Ultimate Tensile Strength (MPa)		235–395	585–654	360–440
Modulus of Elasticity (GPa)		115	260	96
Density (g/cm3)		8.9	14.3	8.46
Hardness (HB)		70–120	89–102	86–115
Electrical conductivity (m/Ω·mm2)		100	41–48	24
Thermal conductivity (W/m·K)		388	190	110

as given by the producer (Xometry Europe GmbH, Hohenbrunn, Germany).

The research was organized according to a D-optimal response surface experimental design, and the variation levels of the factors are presented in

Table 4. Process parameters and their levels of variation.

Control Parameters	Levels of Variation (Machine Codification)			Levels of Variation (In Real Values)
	Level 1	Level 2	Level 3	Level 1	Level 2	Level 3
Peak current, Ip, (A)	10	15	20	11.4	57.3	114
Pulse-on time, Ton (μs)	09	29	35	13	380	600
Pulse-off time, Toff (μs)	10	20	30	18	120	420

. The values were chosen to increase the applicability of the process in an industrial environment. The dielectric fluid used was a mineral oil for the Knuth FEM 110 machine, OEST FE FLUID 2460 (Oest Group, Freudenstadt, Germany).

Untreated and heat-treated C120 tool steel samples (

Table 5. Chemical composition of the C120 tool steel.

C	Si	Mn	P	S	Cr	Ni
1.15–1.25	0.1–0.3	0.1–0.4	<0.030	<0.030	-	-

and

Table 6. Properties of the C120 tool steel.

Density (g/cm3)	Electrical Resistivity (μΩ·m)	Coefficient of Linear Expansion (10−6/°C)	Thermal Conductivity (W/m·K)	Specific Heat Capacity (J/Kg·K)	Hardness (HB)
					Untreated (C120)	Heat Treated (C120T)
7.830	1.96	10.30	388	485	273	715

), measuring 60 mm × 40 mm × 10 mm, were used to perform experimental tests due to the wide applicability of material in industry, especially in the field of tool manufacturing.

2.2. Output Parameters

Two output parameters were analyzed to quantify the performance of the electrical discharge machining process under the working conditions mentioned above:

• an economic efficiency parameter—the machining productivity—quantified by the material removal rate and calculated using Equation (1) [33,34]:

$$\mathrm{M}\mathrm{R}\mathrm{R}={\displaystyle \frac{{\mathrm{w}}_{\mathrm{i}}-{\mathrm{w}}_{\mathrm{f}}}{{\mathsf{\rho }}_{\mathrm{m}}\mathrm{t}}}({\mathrm{mm}}^3/\min ),$$

(1)

where w_i, w_f—weight of workpiece before and after machining, in [g], ρ_m—density of processed material, in [g/mm³], and t—machining time, in [min].

• a quality parameter—the surface roughness, Sa—used to quantify the quality of the machined surface.

To calculate the MRR, the weights of the samples before and after each experimental test were measured using an analytical balance (RADWAG, Radom, Poland), with a capacity of 1000 g and a weighing accuracy of 0.001 g. A ZeGage Pro optical profilometer (AMETEK, Weiterstadt, Germany), equipped with a 2.75× ocular, was used to measure the roughness (topography) of the machined surface (Figure 1). The scanned area on each machined sample was 3000 μm × 3000 μm, for which an average of performed measurements was calculated.

3. Results and Discussion

The experimental plan was constructed using five input parameters: three numerical factors and two categorical factors. The numerical factors, Peak current (Ip), Pulse-on time (Ton), and Pulse-off time (Toff), have three levels of variation each (Table 4). The categorical factors are the electrode material, with three levels of variation (Cu 99.9, WCu 75/25, and CuZn39Pb2), and the state of the processed material, C120 steel, with two variation levels (untreated and heat-treated). Their combination resulted in a total of 162 experiments.

3.1. Statistical Analysis of the Results

The analysis of variance was used for the statistical analysis of the results. ANOVA assesses the influence of each input on the responses and generates regression models of the process.

The analysis was conducted for a confidence level of 95%, which means factors are considered significant only if they have a calculated p-value lower than 0.05. However, factors involved in significant interactions may be included in models to preserve the hierarchy, even if the corresponding p-value is above 0.05.

The suitability of the models is evaluated using different residual plots, such as the normal probability plot, indicating minimum errors if the points on the graph are closely following the regression line (Figure 2a,b), and the residual vs. predicted plots, indicating a constant variance if the residuals are randomly scattered around the zero line (Figure 3a,b).

The prediction capability of the models is estimated based on the correlation coefficients R², adjusted R², and predicted R². A difference smaller than 0.2 is recommended between Adj. R² and Pred. R². The calculated values (

Table 7. Fit statistics.

Model	R2	Adj.R2	Pred.R2
MRR	0.9081	0.8978	0.8849
Sa	0.8528	0.8386	0.8198

) highlight a good agreement between the two coefficients, for both analyzed output parameters. The models are capable to cover 89.78% of MRR variation and 83.86% of Sa variation.

3.2. Regression Modeling of Material Removal Rate (MRR)

The ANOVA results for MRR are presented in

Table 8. ANOVA results for material removal rate (MRR).

Source	Sum of Squares	df	Mean Square	F-Value	p-Value	Contribution
Model	335.44	16	20.96	88.89	<0.0001
A-Electrode material	29.47	2	14.73	62.46	<0.0001	7.98%
B-Peak current, Ip	146.19	1	146.19	619.80	<0.0001	39.57%
C-Pulse-on time, Ton	34.48	1	34.48	146.20	<0.0001	9.33%
D-Pulse-off time, Toff	6.31	1	6.31	26.76	<0.0001	1.71%
E-Material state	0.7855	1	0.7855	3.33	0.0701	0.21%
AB	9.53	2	4.77	20.21	<0.0001	2.58%
AC	10.60	2	5.30	22.46	<0.0001	2.87%
AE	2.38	2	1.19	5.05	0.0076	0.65%
CD	16.39	1	16.39	69.50	<0.0001	4.44%
B2	59.15	1	59.15	250.78	<0.0001	16.01%
C2	17.09	1	17.09	72.46	<0.0001	4.63%
D2	1.99	1	1.99	8.42	0.0043	0.54%
Residual	33.96	144	0.2359
Cor Total	369.40	160

. A quadratic model was generated for determining influences. The F-value of the model is 88.89, which means that the model is significant, with a probability of only 0.01% of obtaining such a value because of “noise”. The most influential factors on MRR are B-Peak current (Ip) with a contribution of 39.57%, C-Pulse-on time (Ton) with a contribution of 9.33%, A-Electrode material with 7.98% contribution, and D-Pulse-off time (Toff) with 1.71%. There are significant interactions: CD with a contribution of 4.44%, AC with a contribution of 2.87%, AB with 2.58%, and AE with 0.65%. Quadratic factors also have important influences, especially B² (16.01%) and C² (4.63%). The model also includes an insignificant factor, E, to preserve the hierarchy, as it is involved in a significant interaction (AE).

Figure 4 illustrates the effects of input factors on the material removal rate. The MRR increases with the increases in peak current (Ip) and pulse-on time (Ton). On the other hand, pulse-off time (Toff) has a much lower influence on MRR and leads to a slight decrease. The state of the machined material (C120 vs. C120T) does not significantly influence MRR. The electrode that allows for the highest material removal rate is the pure copper electrode (Cu 99.9).

However, the graphs illustrating the interactions between factors highlight several specific aspects (Figure 5):

• The worst combination in terms of MRR consists of using large pulse-off time (Toff) and small pulse-on time (Ton) values (Figure 5a). This effect is diminished by minimizing Toff to the lowest limit (18 μs) and increasing Ton to around 300 μs; in this case, the highest possible MRR for this combination is obtained (Figure 5b). A significant reduction in MRR occurs when the values of these two parameters rise beyond these thresholds.
• The peak current effect on MRR is much more pronounced when machining with Cu 99.9 and Wcu 75/25 electrodes compared to the brass electrode CuZN39Pb2 when using a peak current of 114 A (Figure 5c). For low values of Ip, MRR values are not differentiated by the type of electrode material. Furthermore, for all variation curves, regardless of the electrode used, a maximum point is observed around 90 A, after which the peak-current effect on MRR is negative (Figure 5d).
• A similar variation can be observed in terms of the influence of pulse-on time (Ton) on MRR. At high Ton values (600 μs), the largest MRR is given by the Cu 99.9 electrode, followed by the Wcu 75/25 electrode. The MRR produced by the CuZn39Pb2 electrode is not influenced by Ton. For the 13 μs Ton, MRR values are almost the same for all electrode materials (Figure 5e). However, from Figure 5f it can be observed that MRR curves exhibit a maximum inflexion point, which for Cu 99.9 and Wcu 75/25 is around the interval 360–390 μs, while for the CuZn39Pb2 electrode the maximum is around 245 μs (Figure 5f). This shows that the increase in pulse-on time above a certain threshold is not necessarily beneficial and only increases the consumption of energy.
• The condition of the material does not influence the MRR, regardless of the electrode used for machining (Figure 5g).

Figure 5. The effect of interactions on MRR: (a) CD; (b) response surface CD; (c) AB; (d) BA; (e) AC; (f) CA; (g) AE.

Figure 5. The effect of interactions on MRR: (a) CD; (b) response surface CD; (c) AB; (d) BA; (e) AC; (f) CA; (g) AE.

In order to improve the prediction model, a Box–Cox transformation has been applied to the response, and a logarithmic function was implemented. The regression model of the material removal rate MRR is expressed through the set of Equations (2)–(7), for each type of electrode and state of the material (untreated C120 and heat-treated C120T):

• For the Cu 99.9 electrode:

$$\begin{array}{l}\ln \left(\mathrm{M}\mathrm{R}{\mathrm{R}}_{\mathrm{C}120}\right)=-0.088381+0.089801\mathrm{I}\mathrm{p}+0.006563\mathrm{T}\mathrm{o}\mathrm{n}-0.007437\mathrm{T}\mathrm{o}\mathrm{f}\mathrm{f}+7.74408\mathrm{E}-6\mathrm{T}\mathrm{o}\mathrm{n}\mathrm{T}\mathrm{o}\mathrm{f}\\ -0.000497\mathrm{I}{\mathrm{p}}^2-8.63081\mathrm{E}-6\mathrm{T}\mathrm{o}{\mathrm{n}}^2+8.00678\mathrm{E}-6\mathrm{T}\mathrm{o}{\mathrm{f}}^2\end{array}$$

(2)

$$\begin{array}{l}\ln \left(\mathrm{M}\mathrm{R}{\mathrm{R}}_{\mathrm{C}120\mathrm{T}}\right)=-0.079190+0.089801\mathrm{I}\mathrm{p}+0.006563\mathrm{T}\mathrm{o}\mathrm{n}-0.007437\mathrm{T}\mathrm{o}\mathrm{f}\mathrm{f}+7.74408\mathrm{E}-6\mathrm{T}\mathrm{o}\mathrm{n}\mathrm{T}\mathrm{o}\mathrm{f}\mathrm{f}\\ -0.000497\mathrm{I}{\mathrm{p}}^2-8.63081\mathrm{E}-6\mathrm{T}\mathrm{o}{\mathrm{n}}^2+8.00678\mathrm{E}-6\mathrm{T}\mathrm{o}\mathrm{f}{\mathrm{f}}^2\end{array}$$

(3)

• For the W/Cu 75/25 electrode:

$$\begin{array}{l}\ln \left(\mathrm{M}\mathrm{R}{\mathrm{R}}_{\mathrm{C}120}\right)=-0.122341+0.089987\mathrm{I}\mathrm{p}+0.006059\mathrm{T}\mathrm{o}\mathrm{n}-0.007437\mathrm{T}\mathrm{o}\mathrm{f}\mathrm{f}+7.74408\mathrm{E}-6\mathrm{T}\mathrm{o}\mathrm{n}\mathrm{T}\mathrm{o}\mathrm{f}\mathrm{f}\\ -0.000497\mathrm{I}{\mathrm{p}}^2-8.63081\mathrm{E}-6\mathrm{T}\mathrm{o}{\mathrm{n}}^2+8.00678\mathrm{E}-6\mathrm{T}\mathrm{o}\mathrm{f}{\mathrm{f}}^2\end{array}$$

(4)

$$\begin{array}{l}\ln \left(\mathrm{M}\mathrm{R}{\mathrm{R}}_{\mathrm{C}120\mathrm{T}}\right)=-0.194056+0.089987\mathrm{I}\mathrm{p}+0.006059\mathrm{T}\mathrm{o}\mathrm{n}-0.007437\mathrm{T}\mathrm{o}\mathrm{f}\mathrm{f}+7.74408\mathrm{E}-6\mathrm{T}\mathrm{o}\mathrm{n}\mathrm{T}\mathrm{o}\mathrm{f}\mathrm{f}\\ -0.000497\mathrm{I}{\mathrm{p}}^2-8.63081\mathrm{E}-6\mathrm{T}\mathrm{o}{\mathrm{n}}^2+8.00678\mathrm{E}-6\mathrm{T}\mathrm{o}\mathrm{f}{\mathrm{f}}^2\end{array}$$

(5)

• For the CuZn39Pb2 electrode:

$$\begin{array}{l}\ln \left(\mathrm{M}\mathrm{R}{\mathrm{R}}_{\mathrm{C}120}\right)=0.235724+0.077633\mathrm{I}\mathrm{p}+0.004102\mathrm{T}\mathrm{o}\mathrm{n}-0.007437\mathrm{T}\mathrm{o}\mathrm{f}\mathrm{f}+7.74408\mathrm{E}-6\mathrm{T}\mathrm{o}\mathrm{n}\mathrm{T}\mathrm{o}\mathrm{f}\mathrm{f}\\ -0.000497\mathrm{I}{\mathrm{p}}^2-8.63081\mathrm{E}-6\mathrm{T}\mathrm{o}{\mathrm{n}}^2+8.00678\mathrm{E}-6\mathrm{T}\mathrm{o}\mathrm{f}{\mathrm{f}}^2\end{array}$$

(6)

$$\begin{array}{l}\ln \left(\mathrm{M}\mathrm{R}{\mathrm{R}}_{\mathrm{C}120\mathrm{T}}\right)=0.714810+0.077633\mathrm{I}\mathrm{p}+0.004102\mathrm{T}\mathrm{o}\mathrm{n}-0.007437\mathrm{T}\mathrm{o}\mathrm{f}\mathrm{f}+7.74408\mathrm{E}-6\mathrm{T}\mathrm{o}\mathrm{n}\mathrm{T}\mathrm{o}\mathrm{f}\mathrm{f}\\ -0.000497\mathrm{I}{\mathrm{p}}^2-8.63081\mathrm{E}-6\mathrm{T}\mathrm{o}{\mathrm{n}}^2+8.00678\mathrm{E}-6\mathrm{T}\mathrm{o}\mathrm{f}{\mathrm{f}}^2\end{array}$$

(7)

3.3. Regression Modeling of Surface Roughness (Sa)

The ANOVA results for Sa are presented in

Table 9. ANOVA results for surface roughness (Sa).

Source	Sum of Squares	df	Mean Square	F-Value	p-Value	Contribution
Model	59.79	14	4.27	60.01	<0.0001
A-Electrode material	0.1437	2	0.0718	1.01	0.3669	0.20%
B-Peak current, Ip	20.12	1	20.12	282.66	<0.0001	28.69%
C-Pulse-on time, Ton	21.33	1	21.33	299.76	<0.0001	30.43%
D-Pulse-off time, Toff	0.5217	1	0.5217	7.33	0.0076	0.74%
E-Material state	0.0093	1	0.0093	0.1303	0.7186	0.01%
AB	3.42	2	1.71	24.04	<0.0001	4.88%
AC	0.4593	2	0.2297	3.23	0.0425	0.66%
BC	1.20	1	1.20	16.91	<0.0001	1.72%
BE	0.1996	1	0.1996	2.80	0.0962	0.28%
B2	9.18	1	9.18	128.95	<0.0001	13.09%
C2	3.59	1	3.59	50.41	<0.0001	5.12%
Residual	10.32	145	0.0712
Cor Total	70.11	159

. A quadratic model was generated for determining influences. The F-value of the model is 60.01, which means that the model is significant, with a probability of only 0.01% of obtaining such a value because of “noise”. The most influential factors on Sa are C-Pulse-on time (Ton) with a contribution of 30.43%, B-Peak current (Ip) with a contribution of 28.69%, and D-Pulse-off time (Toff) with 0.74%. There are significant interactions: AB with a contribution of 4.88%, BC with a contribution of 1.72%, and AC with 0.66%. Quadratic factors also have important influences, especially B² (13.09%) and C² (5.12%).

Figure 6 illustrates the effects of input factors on the surface roughness. The Sa increases with the increases in peak current (Ip) and pulse-on time (Ton). On the other hand, raising the pulse-off time (Toff) leads to a decrease in Sa. The state of the machined material (C120 vs. C120T) does not really induce notable variations in Sa. On average, the electrodes that usually produce lower roughness values are Cu99.9 and WC 75/25.

The graphs illustrating the interactions between factors highlight several specific aspects (Figure 7):

• The interaction electrode material—peak current (AB) reveals that for the larger currents (114 A), the brass electrode CuZn39Pb2 seems more favorable for a better surface quality by producing a lower value of surface roughness (Figure 7a). It can also be observed that the variation curves have maxima at around 100 A for the Cu99.9 and Wcu 75/25 electrodes, and approximately 67 A for the CuZn39Pb2 electrode (Figure 7b).
• The most favorable combination for an improved surface quality is low levels of peak current Ip and pulse-on time Ton (Figure 7c,d).
• The interaction AC reveals that for the lower Ton, the electrode material does not modify surface roughness. However, for larger Ton (600 μs), the surface roughness is significantly higher and increases from the Cu 99.9 electrode to Wcu 75/25 and CuZn39Pb2 (Figure 7e).

Figure 7. The effect of interactions on Sa: (a) AB; (b) BA; (c) BC; (d) CB; (e) AC; (f) CA.

Figure 7. The effect of interactions on Sa: (a) AB; (b) BA; (c) BC; (d) CB; (e) AC; (f) CA.

The regression model showing the dependence of the surface roughness Sa on the factors of influence is expressed through the set of Equations (8)–(13). A logarithm transformation of the results was applied to improve the predictability of the model. There are separate sets of equations for each combination of electrode material and material state (untreated C120 and heat-treated C120T):

• For the Cu 99.9 electrode:

$$\begin{array}{l}\ln \left(\mathrm{S}{\mathrm{a}}_{\mathrm{C}120}\right)=0.965704+0.033596\mathrm{I}\mathrm{p}+0.003029\mathrm{T}\mathrm{o}\mathrm{n}-0.000336\mathrm{T}\mathrm{o}\mathrm{f}\mathrm{f}+8.48406\mathrm{E}-6\mathrm{I}\mathrm{p}\mathrm{T}\mathrm{o}\mathrm{n}\\ -0.000196\mathrm{I}{\mathrm{p}}^2-3.98741\mathrm{E}-6\mathrm{T}\mathrm{o}{\mathrm{n}}^2\end{array}$$

(8)

$$\begin{array}{l}\ln \left(\mathrm{S}{\mathrm{a}}_{\mathrm{C}120\mathrm{T}}\right)=1.05232+0.031914\mathrm{I}\mathrm{p}+0.003029\mathrm{T}\mathrm{o}\mathrm{n}-0.000336\mathrm{T}\mathrm{o}\mathrm{f}\mathrm{f}+8.48406\mathrm{E}-6\mathrm{I}\mathrm{p}\mathrm{T}\mathrm{o}\mathrm{n}\\ -0.000196\mathrm{I}{\mathrm{p}}^2-3.98741\mathrm{E}-6\mathrm{T}\mathrm{o}{\mathrm{n}}^2\end{array}$$

(9)

• For the W/Cu 75/25 electrode:

$$\begin{array}{l}\ln \left(\mathrm{S}{\mathrm{a}}_{\mathrm{C}120}\right)=0.784492+0.034070\mathrm{I}\mathrm{p}+0.003499\mathrm{T}\mathrm{o}\mathrm{n}-0.000336\mathrm{T}\mathrm{o}\mathrm{f}\mathrm{f}+8.48406\mathrm{E}-6\mathrm{I}\mathrm{p}\mathrm{T}\mathrm{o}\mathrm{n}\\ -0.000196\mathrm{I}{\mathrm{p}}^2-3.98741\mathrm{E}-6\mathrm{T}\mathrm{o}{\mathrm{n}}^2\end{array}$$

(10)

$$\begin{array}{l}\ln \left(\mathrm{S}{\mathrm{a}}_{\mathrm{C}120\mathrm{T}}\right)=0.871110+0.032388\mathrm{I}\mathrm{p}+0.003499\mathrm{T}\mathrm{o}\mathrm{n}-0.000336\mathrm{T}\mathrm{o}\mathrm{f}\mathrm{f}+8.48406\mathrm{E}-6\mathrm{I}\mathrm{p}\mathrm{T}\mathrm{o}\mathrm{n}\\ -0.000196\mathrm{I}{\mathrm{p}}^2-3.98741\mathrm{E}-6\mathrm{T}\mathrm{o}{\mathrm{n}}^2\end{array}$$

(11)

• For the CuZn39Pb2 electrode:

$$\begin{array}{l}\ln \left(\mathrm{S}{\mathrm{a}}_{\mathrm{C}120}\right)=1.31043+0.026474\mathrm{I}\mathrm{p}+0.003494\mathrm{T}\mathrm{o}\mathrm{n}-0.000336\mathrm{T}\mathrm{o}\mathrm{f}\mathrm{f}+8.48406\mathrm{E}-6\mathrm{I}\mathrm{p}\mathrm{T}\mathrm{o}\mathrm{n}\\ -0.000196\mathrm{I}{\mathrm{p}}^2-3.98741\mathrm{E}-6\mathrm{T}\mathrm{o}{\mathrm{n}}^2\end{array}$$

(12)

$$\begin{array}{l}\ln \left(\mathrm{S}{\mathrm{a}}_{\mathrm{C}120\mathrm{T}}\right)=1.39705+0.024792\mathrm{I}\mathrm{p}+0.003494\mathrm{T}\mathrm{o}\mathrm{n}-0.000336\mathrm{T}\mathrm{o}\mathrm{f}\mathrm{f}+8.48406\mathrm{E}-6\mathrm{I}\mathrm{p}\mathrm{T}\mathrm{o}\mathrm{n}\\ -0.000196\mathrm{I}{\mathrm{p}}^2-3.98741\mathrm{E}-6\mathrm{T}\mathrm{o}{\mathrm{n}}^2\end{array}$$

(13)

From the presented results, it was observed that for both the material removal rate (MRR) and surface roughness (Sa) there are certain values for which the influence of the factors is reversed.

Thus, passing these thresholds does not further improve the process performance because the maximum point for MRR has been surpassed and surface quality depreciates at accelerated rates. The reduction in MRR over certain limits can be explained by the fact that for the “aggressive” regimes (i.e., larger values of Ip and Ton), there is an increase in melted material but not all of it is removed from the electrode–workpiece interstitial space, and it is partially redeposited on the machined surface. These results are in accordance with other research studies [35,36,37]. Moreover, in the case of the CuZn39Pb2 electrode, there is also a transfer of material from the electrode to the surface of the part, deposited in a visible yellow layer of material (Figure 8), which on one side decreases the effective MRR, but also contributes to an apparent improvement of surface roughness Sa by covering the discharge microcraters. These faux decreases can be seen in Figure 5c for MRR and Figure 7a for Sa.

4. Conclusions

The purpose of the current paper was to investigate the use of intensive EDM regimes to increase the productivity of the process by reducing the machining time, while also maintaining a good quality surface. The results have shown it is possible to achieve such an objective, but it was also found that increasing the processing parameters is feasible only to a certain extent.

For both the MRR and Sa, the most important factors of influence are the peak current (Ip) and the pulse-on time (Ton). Maximum MRR for the Cu 99.9 and WCu 75/25 electrodes was obtained for an Ip of approximately 90 A and a Ton between 360 and 390 μs, while maximum MRR for the CuZn39Pb2 electrode was for an Ip of approximately 78 A and a Ton around 245 μs. Surface roughness (Sa) is highly affected when intensive regimes (high values of Ip and Ton) are used, making it suitable only when productivity is the sole objective of the process (roughing machining). However, an acceptable surface quality may be obtained by reducing the Ton values, which mitigates the influence of peak current on Sa.

The process outcomes can be also controlled by regulating the values of pulse-off time (Toff). Thus, by increasing Toff, the material removal rate reduces slowly, but on the other hand, surface quality is improved.

The recommended electrode materials are Cu 99.9 and WCu 75/25, which yielded better results with more intensive regimes that translate into increased productivity, but also into better surface quality.

The state of the workpiece material (i.e., the thermal treatment) did not influence the results in a notable manner.

Our research will continue in the direction of improving debris flushing when using intensive regimes to prevent their deposition on machined surfaces. We will also test the possibility of eliminating the thresholds encountered for Ip and Ton by replacing the machine’s dielectric with vegetable oils, which, from previous research [34,38], appear to keep their properties under intense regimes, ultimately leading to higher MRRs and better surface quality.