Dimensional and Surface Quality Evaluation of Inconel 718 Alloy After Grinding with Environmentally Friendly Cooling-Lubrication Technique and Graphene Enriched Cutting Fluid

Abstract

Properly refrigerating hard-to-cut alloys during grinding is key to achieve high quality, strict tolerances, and good surface finishing. Nonetheless, literature about the influence of cooling-lubrication conditions (CLCs) on dimensional accuracy of ground components is still scarce. Thus, this work aims to evaluate surface quality, grinding power, and dimensional accuracy of Inconel 718 workpieces after grinding with silicon carbide grinding wheel at different grinding conditions. Four different CLCs were tested: flood, minimum quantity of lubrication (MQL) without graphene, and with multilayer graphene (MG) at two distinct concentrations: 0.05 and 0.10 wt.%. Different radial depths of cut values were also tested. The results showed that the material’s removed height increased with radial depth of cut, leading to coarse tolerance (IT) grades. Machining with the MQL WG resulted in higher dimensional precision with an IT grade varying between IT6 and IT7, followed by MQL MG 0.10% (IT7), MQL MG 0.05% (IT7-IT8), and flood (IT8). The lower tolerances achieved with MG were attributed to the lowering in the friction coefficient of the workpiece material sliding through the abrasive grits with no material removal (micro-plowing mechanism), thereby reducing grinding power and the removed height in comparison to the other CLC tested.

1. Introduction

The search for the production of parts with high dimensional accuracy by using machining as first option [1] or its use in post-processing [2] of several types of components has been increasing abruptly in recent years. With industrial development, with production of novel materials, and demand for miniaturized parts, the components’ requirements became stricter leading to the constant need to improve accuracy. Recent work carried out by Alfattani et al. [3] evaluated electro-discharge machining (EDM) of MoSi₂–SiC ceramic composites aiming to achieve higher dimensional and geometrical accuracy in the EDM process for steel. Prakash et al. [4] applied a genetic algorithm to achieve higher accuracy of cavities. Muthuramalingam et al. [5] studied the quality of Ti-6Al-4V parts manufactured by the laser beam process, focusing on the accuracy of holes.

However, to achieve strict geometrical and dimensional tolerances, special processes may be required, and grinding is usually employed in different fields as a first alternative among several machining processes: some applications are in grinding of diamond cutting tools [6], cylindrical microlens array [7], monocrystal sapphire for optics and electronics [8], polycrystalline yttrium aluminum garnet for optoelectronics [9], and the aerospace ceramic RB-SiC [10].

Especially in the aerospace industry, the production of blades of compressors and turbines having complex geometry deserves attention because the of titanium and nickel alloy materials that are used to produce them, since they have peculiar properties that impair their machinability [11], in addition to the surface and sub-surface integrities of components that must be preserved in combination with low dimensional tolerances [12].

This specific part has been studied both as the complete blisk (disc with blades) or interchangeable blades [13], with dimensional tolerances reported as 10 µm for parts varying between 10 mm to 30 mm [14] or −10 µm to +30 µm [15] according to the literature. Specifically, the nickel-based Inconel 718 alloy is able to operate under severe conditions of stress, high temperature (around 500 °C), and corrosion environments [16]. However, this is a poor machining and grinding material [17]. In a study comparing the grindability of Inconel 718 with gray cast iron, the authors reported that Ra roughness values were up to 250% higher for the Inconel 718 in comparison with those recorded when grinding gray cast iron, thereby reinforcing its poor grindability [18].

As previously mentioned, different methods can be applied to increase precision of a given manufactured component. Regarding the grinding process, dimensional tolerances can be associated with the real or true depth of cut (a_r), which differs from the set depth of cut (a_e) due to system deflection (δ), wheel wear (a_sw), and thermal expansion of the workpiece during grinding (a_t) [19], as shown in Equation (1).

$$a_r=a_e-\delta -a_{sw}+a_t\left[\mu \mathrm{m}\right]$$

(1)

The workpiece elastic deformation (ε) also influences by reducing a_r, and it can be calculated in terms of the grinding wheel width (l_r), the external diameter of the grinding wheel (d_s), normal force (f), and the workpiece’s elastic modulus (E) as shown in Equation (2) [20,21].

$$\epsilon ={\displaystyle \frac{f}{l_r{\left(a_e{d}_s\right)}^{\raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$2$}\right.}E}}$$

(2)

Besides the grinding machine stiffness and its own precision capacity and uncertainty in terms of depth of cut adjustment, as well as spark-out strategy, the cutting fluid delivery method, when properly selected and adjusted, can provide both lubrication and cooling that positively contributes to reducing the disparity between the real and set depth of cut. This can be achieved by either reducing the grinding wheel wear and thermal expansion due to better temperature control, or by decreasing cutting efforts due to friction coefficient reduction, thereby attenuating system deflection. As a result of that, geometric and dimension tolerances can be reduced, with potential for decreasing the necessity for further finishing operations.

With respect to research on environmentally friendly techniques for temperature control in grinding, Esmaeili et al. [22] tested different cooling-lubrication conditions during grinding of Inconel 718 including dry, flood, Minium Quantity of Lubricant (MQL), and MQL with graphene. Although no analysis of dimensional or geometric tolerance was carried out, the authors observed that using the MQL technique reduced grinding forces (normal and tangential components) and friction coefficient in comparison to dry and flood environments, and the addition of graphene to the MQL system further improved grinding efficiency. The authors pointed out that the use of nanoparticles as additives for fluids applied via MQL technique has been a trend in the manufacturing sector to enhance lubrication and heat exchange properties of cutting fluids.

Virdi et al. [23] also carried out a study on grinding of Inconel 718 with nanofluids delivered via the MQL technique and observed that this cooling-lubrication strategy was able to improve finishing of the ground surface (low values of the Ra parameter) by up to 59% and 42% in comparison to flood cooling and MQL (only oil), respectively. The best result in terms of reducing Ra was found when machining with nanofluids having sunflower as base oil and 0.5 wt.% of Al₂O₃ nanoparticles. The authors also found that, compared to MQL (only oil), using nanofluids can reduce grinding forces and coefficient of friction, as well as significantly decrease grinding wheel wear (increase in G-ratio).

Most research regarding the grinding of Inconel 718 using nanofluids investigates surface finish and other output parameters like those evaluated in the several works previously mentioned such as grinding forces, coefficient of friction, and surface morphology. However, the dimensional accuracy of the parts is usually omitted in machining research.

The International Tolerance (IT) grade system, as defined by the ISO 286 standard [24], refers to the allowable variation in the size of a part, ensuring functional fit and interchangeability between components. The grades range from IT01, which represents extremely high precision, to IT16, which allows for more open tolerances. The selection of an appropriate IT grade is critical during the design phase, as it directly influences the functionality, assembly, and production cost of mechanical components. For instance, high-precision applications such as bearings or aerospace parts may require IT grades between IT01 and IT5. Specifically, turbine blades, a part commonly made with Inconel 718, usually require tolerances varying from IT4 to IT6 [13,14,25].

The literature regarding a more detailed analysis of dimensional tolerances in grinding of Inconel 718 is still scarce, especially considering graphene enriched cutting fluids and their relation to real depth of cut, chip formation mechanism, surface morphology, and electric power during grinding. To fill this gap, this work sought to evaluate the dimensional tolerance and surface quality of Inconel 718 after grinding with different operational conditions, including different cooling-lubrication methods such as flood and MQL with graphene enriched fluids. Different cutting conditions in terms of radial depth of cut values were also tested. The dimensional deviation of the workpiece was evaluated in terms of the removed height (real depth of cut) and comparison with set depth of cut values was performed. Surface roughness (Ra parameter) and surface texture of the machined surfaces were used for surface quality analysis and the electric power during grinding was also monitored.

2. Materials and Methods

In this section, the methodology of grinding experiments including materials, equipment, grinding conditions, and the methods for output variables evaluation are presented in detail.

2.1. Material Characterization

The workpiece selected is the Inconel 718 alloy, which was produced by casting and subjected to age hardening. Its main mechanical properties and chemical composition are presented in

Table 1. Mechanical properties of Inconel 718 [26].

Tensile Strength (σt—MPa)	Yield Strength (σy—MPa)	Young’s Modulus (E—GPa)	Hardness (HRC)	Density at Room Temperature (ρ—g/cm3)	Melting Range (°C)	Thermal Conductivity (λ—W/mK)
1275	1034	200	40	8.22	1260–1336	11.4

and

Table 2. Chemical composition of Inconel 718 [26].

Element	C	Al	Ti	Cr	Fe	Ni	Nb	Mo
Percentage	0.04	0.50	0.90	19.00	18.50	50.66	5.10	5.30

, respectively. The dimensions of the workpiece were 15 mm × 20 mm × 15 mm.

2.2. Grinding Parameters

Grinding tests were carried out in a P-36 peripheral surface grinding machine (manufactured by MELLO, São Paulo, Brazil), with spindle speed of 2400 rpm and power of 2.25 kW. This grinding machine has a vertical movement resolution of the grinding wheel of 0.005 mm (5 µm). The workpiece sample was held in a precision vise mounted on the grinder table, positioned between two fixture devices of gray cast iron to improve the clamp’s rigidity. A Saint Gobain vitrified bond silicon carbide (SiC) grinding wheel with designation of 39C60KV and dimensions of 305 mm (external diameter), 25 mm (width), and 76 mm (internal diameter) was employed, the cutting parameters were selected based on previous work in the aiming to access if the geometrical tolerances would also been influenced by the CLC and cutting conditions [27,28].

To ensure that no clogging or grinding wheel wear would interfere in the results, the grinding wheel was dressed using a single point diamond dresser, with an overlap ratio of 3. It should be mentioned that for all the experiments a spark-out cycle was performed, with the same cutting parameters of the grinding pass, i.e., the grinding was performed in the discordant cutting direction while the spark-out process was performed in the concordant cutting direction.

The grinding process is represented in the diagram of Figure 1. The parameters selected for this work are detailed in

Table 3. Machining parameters.

Parameter	Cutting Speed (vs)	Workspeed (vw)	Radial Depth of Cut (ae)
Value	38 m/s	10 m/min	20 µm; 40 µm

, which were previously selected based on pilot tests. The selection of the radial depth of cut as input variable is due to its significant influence on the removed height from the machined surface, once it is necessary a minimum abrasive grain penetration into the workpiece for chip formation (material removal) [29].

Differences between the radial depth of cut selected and the one adjusted in the grinder affect the final dimension and surface quality of the workpiece, and it depends on the vertical movement resolution/adjustment of the grinding wheel, as illustrated on Figure 2. To check this variation, positioning errors in the vertical movement were measured with the aid of an analogic precision dial gauge, and the higher positioning error in the range of radial depth of cut employed in this work, 20 µm and 40 µm, was 1 µm, similar results on the machine positioning precision were observed by De Oliveira et al. [18] in research carried out on grinding F250 gray cast iron and Inconel 718 using the same grinding machine employed in this work.

Cooling-Lubrication Condition (CLC)

Four different CLCs were tested, with and without multilayer graphene (MG), which were evaluated with the concentrations based on different studies [30,31], as follows:

• Flood: semi-synthetic emulsifiable oil, Vasco 7000 from Blaser Swisslube (Blaser Swisslube, Hasle-Rüegsau, Switzerland), diluted in water in a proportion of 1:9, which was monitored with the aid of a refractometer (N1, ATAGO). The coolant was delivered with a pressure near to the atmospheric one and with a flow rate of 522 L/h (522,000 mL/h).
• MQL WG: WG stands for without multilayer graphene platelets, i.e., only oil. In this condition the oil (Vasco 7000) was delivered without water and solid particles. The flow rate and compressed-air pressure for the MQL technique were 240 mL/h and 0.5 MPa, respectively, for all the tests using the MQL technique;
• MQL MG 0.05%: this condition refers to Vasco 7000 oil with multilayer graphene (MG) platelets added as 0.05% by weight, applied via the MQL technique;
• MQL MG 0.10%: this refers to Vasco 7000 oil with multilayer graphene platelets added as 0.10% by weight, also applied via the MQL technique.

The process of production of multilayer graphene platelets is detailed in the work carried out by Augusto et al. [32]. However, it is worth mentioning that the flake thickness is in the range of 1–10 nm, resulting in 6 to 14 graphene layers in the flake. Also, the methodology results in very low structural defects, low impurities concentration, and high electrical conductivity therefore promoting high thermal conductivity, which improves heat transference [32]. An ultrasonic bait (40 Hz) was used to disperse the particles in the cutting fluid. The process was maintained for two hours, achieving both homogeneity and stability.

The morphology of a representative graphene flake was performed by atomic force microscopy in dynamic mode using a Shimadzu SPM9700 device. In Figure 3a is shown the 3D topography of a representative MG flake herein used, Figure 3b illustrates its 2D topography as well the lines used to indicate the profiles for thickness estimation whose values are presented in Figure 3c. The image treatments and profile analysis were obtained by using Gwyddion software (version 3.2). In this representative flake it is possible to infer that the particle has a thickness of 6 to 8 nm along its horizontal profile and about 8 to 10 nm in the other one.

The different cutting fluids tested in this work were characterized in terms of dynamic viscosity with the aid of an Anton Paar SVM 3000 viscometer (Anton Paar, Styria, Austria) with a measurement range of 0.2 mm²/s to 30,000 mm²/s and a precision of 0.0001 g/cm³ with a repeatability of 0.01%, and thermal conductivity with a Linseis THB-1 thermal conductivity analyzer (Linseis, Bavaria, Germany), possessing an accuracy for thermal diffusivity around ±2.4% and specific heat around ±5%. The measurement results are shown in

Table 4. Dynamic viscosity and thermal conductivity at 25 °C and 50 °C for the cutting fluids employed in this work.

	Dynamic Viscosity (m·Pa·s)		Thermal Conductivity (W·m−1K−1) × 10−001
	At 25 °C	At 50 °C	At 25 °C	At 50 °C
Flood (oil + water)	1.38 ± 0.01	0.86 ± 0.00	5.41 ± 0.08	5.00 ± 0.22
MQL WG	145.2 ± 0.41	44.82 ± 0.03	2.65 ± 0.04	2.64 ± 0.05
MQL MG 0.05%	157.18 ± 2.05	48.53 ± 0.54	2.83 ± 0.13	2.85 ± 0.21
MQL MG 0.10%	159.42 ± 0.74	48.97 ± 0.08	2.81 ± 0.31	1.65 ± 0.21

, in which the mean value of three measurements is presented with the standard deviation (68.27%). It is possible to observe that the lower graphene concentration led to a better thermal conductivity which can be associated with better distribution of the flakes allowing the thermal exchange or with errors in measurements, which may also result from variable heating power and predominantly from natural convection [33].

It also should be noted that in previous investigations conducted by researchers at the Laboratory of Teaching and Research in Machining at the Federal University of Uberlândia, it was observed that with a graphene content above 0.10 wt%, agglomeration occurred in the nozzle, obstructing the cutting fluid output.

In this study, the volumetric efficiency of the coolant was not directly measured, as it fell outside the primary scope of the investigation, which focused on the influence of process parameters on grinding performance under MQL conditions. The atomized nature of the lubricant in MQL, delivered as a fine aerosol spray, introduces significant challenges in accurately quantifying the volume of fluid that effectively reaches the grinding zone. Factors such as dispersion, partial evaporation, and the absence of a standardized measurement methodology under dynamic conditions further complicate this assessment.

2.3. Dimensional Measurements

To measure the removed height from the workpiece material when grinding using each CLC in combination with depth of cut, the initial height and final height of the workpiece were measured as a linear dimension and their difference indicates the removed height for each trial. In Figure 4 these elements are schematically illustrated.

As mentioned in Section 1, it is possible to calculate the theoretical variation occurring due to elastic deformation as presented in Equation (2). The consistency of Equation (2) can be verified by using the experimental force results in grinding Inconel 718 found by Tso [34]; combining those values with Inconel 718 properties, it is possible to obtain the deformation ($\epsilon )$ for each cutting condition, resulting in $\epsilon$ equal to 6.2 μm and 8.7 μm, for 20 μm and 40 μm radial depth of cut values, respectively. Considering this elastic deformation and the theory developed by Malkin and Guo [21], it is possible to notice that when the depth of cut is set as 20 μm, for instance, the predicted material removal, in linear dimension, is about 13.8 μm, or even less when considering other influence factors pointed out by Marinescu et al. [35] as shown in Equation (1). Similarly, for 40 μm, the predicted height removed is 31.3 μm.

During some pilot tests, a second source of variation was identified, the deformation caused by the precision vise itself. As previously mentioned, Inconel 718 is a nonmagnetic material, so the use of a precision vise for workpiece fixture is indispensable for machining. Thus, to attenuate the influence of the vise’s deformation in the measurement results, 11 steps were performed for each grinding test and replica as follows:

(1)

Fixation of workpiece in the precision vise;

(2)

Preparation of the workpiece surface by grinding with the same cutting parameters detailed in Table 4, with the exception of radial depth of cut, which was equal to 10 µm, i.e., soft grinding conditions;

(3)

The set (vise + workpiece), gauge blocks, and the analog outside micrometer were kept on a granite table in a temperature-controlled room for 12 h to ensure the measurement temperature of 20 °C;

(4)

A gauge block was used to avoid errors due to the vise geometry;

(5)

Measurements were performed following all the statistical recommendations;

(6)

The set was put in the same room as the grinder machine for 12 h before the grinding trial to achieve thermal balance;

(7)

The grinding operation was performed;

(8)

The set was cleaned;

(9)

Step 3 was repeated to ensure the proper measurement temperature of 20 °C;

(10)

Measurements of the height of workpieces included ten measurements for each workpiece;

(11)

Repetition of tests for the different cutting parameters and colling-lubrication conditions and replica.

The selected measurement system was an analog outside micrometer, model M120-50 (serial nº 27204214) with a 0.001 mm graduation and a measuring range from 25 mm to 50 mm, and an accuracy of 0.0025 mm, according to the certificate of inspection (nº 13A09E2C). It is important to remark that the temperature was kept constant at 20 °C during all the measurements This was verified by digital thermo-hygrometer, with calibration certificate number R4996/13, a resolution of 0.1 °C, and a nominal range of -20 to 60 °C. A measurement expanded uncertainty analysis was developed, according to the methodology proposed by GUM (Guide to expression of Uncertainty in Measurement) namely [36]:

• Definition of the measurand or output variable;
• Identification of input variables that may affect the measurement of the output variable;
• Mathematical model of the measurand as a function of all input variables;
• Calculation of the standard uncertainty associated with each input variable (u);
• Calculation of the combined standard uncertainty regarding the output variable (uc);
• Calculation of effective degrees of freedom;
• Assessment of the expanded uncertainty regarding roughness (U);
• Mathematical expression of the measurement result.

The certificate of inspection does not specify the measurement uncertainty associated with the spindle flatness deviation, the stop flatness deviation, or the parallelism deviation. In this case, the mathematical model that correlates the input variables that influence the measurement with the output variable (dimensional deviation—D) is the variability of the readings ($\overline{\mathrm{x}}$), finite graduation of the outside micrometer (∆Rm), and uncertainty associated with outside micrometer calibration (∆Cm). The respective mathematical model is described according to Equation (3):

$$\mathrm{D}=\overline{\mathrm{x}}+∆\mathrm{R}\mathrm{m}+∆\mathrm{C}\mathrm{m}$$

(3)

Output variables information’s for standard uncertainty calculation are shown in

Table 5. Variable information for the uncertainty calculation.

Quantity	Estimation	P. D	T. A	D. F	C. S	Standard Uncertainty
$\overline{\mathrm{x}}$	$\overline{\mathrm{x}}$	t	A	n − 1	1	$\mathrm{u}\left(\overline{\mathrm{x}}\right)={\displaystyle \frac{\mathrm{s}}{\surd \mathrm{n}}}$
∆Rm	Rm	R	B	∞	1	$\mathrm{u}\left(∆\mathrm{R}\mathrm{m}\right)={\displaystyle \frac{\mathrm{R}\mathrm{m}}{\surd 3}}$
ΔCm	0.0015 mm	t	B	∞	1	$\mathrm{u}\left(∆\mathrm{C}\mathrm{m}\right)={\displaystyle \frac{\mathrm{U}(∆\mathrm{C}\mathrm{m})}{\mathrm{k}}}$

, as well as the type of uncertainty assessment (T.A), the probability distribution (P.D), the degrees of freedom (D. F) as well as the coefficients of sensitivity (C.S) and calculations for standard uncertainty. Also, as detailed in Table 5, “s” is the sample standard deviation and “Rm” is the outside micrometer graduation.

After calculating the standard uncertainties, it is necessary to calculate the combined standard uncertainty associated with the dimensional deviation, according to the uncertainty propagation law, Equation (4):

$${\mathrm{uc}}^2\left(\mathrm{D}\right)={\mathrm{u}}^2\left(\overline{\mathrm{x}}\right)+{\mathrm{u}}^2(\Delta \mathrm{R}\mathrm{m})+{\mathrm{u}}^2(\Delta \mathrm{C}\mathrm{m})$$

(4)

After calculating the combined standard uncertainty, it is necessary to calculate the number of effective degrees of freedom, Equation (5), so that the coverage factor k can be identified. It is worth mentioning that k is a tabulated constant value, which relates the number of degrees of freedom (v_eff) with the coverage probability (95%) [36]. It is used to find the expanded uncertainty, when multiplied by the combined uncertainty, so that the value of the expanded uncertainty is obtained, according to Equation (6).

Finally, percentual contributions of input variables can be achieved by Equation (7).

$${\mathsf{\vartheta }}_{\mathrm{e}\mathrm{f}\mathrm{f}}={\displaystyle \frac{{\mathrm{u}}_{\mathrm{c}}^4\left(\mathrm{y}\right)}{\sum _{\mathrm{i}=0}^{\mathrm{N}}\frac{{\mathrm{u}}^4\left({\mathrm{y}}_{\mathrm{i}}\right)}{{\mathrm{v}}_{\mathrm{i}}}}}={\displaystyle \frac{{\mathrm{u}}_{\mathrm{c}}^4\left(\mathrm{y}\right)}{\sum _{\mathrm{i}=0}^{\mathrm{N}}\frac{{[\mathrm{u}\left({\mathrm{x}}_{\mathrm{i}}\right)·{\mathrm{c}}_{\mathrm{i}}]}^4}{{\mathrm{v}}_{\mathrm{i}}}}}$$

(5)

U = k · u_c

(6)

$$\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\left(\mathrm{\%}\right)={\displaystyle \frac{\left[\mathrm{u}\right({\mathrm{x}}_{\mathrm{i}})·{\mathrm{C}}_{\mathrm{i}}]}{{[{\mathrm{u}}_{\mathrm{c}}·\left(\mathrm{y}\right)]}^2}}$$

(7)

The reduction in the workpiece height was obtained by subtracting the mean value (of ten measurements) after grinding from the mean value of the ten measurements performed prior to the grinding process. Worth mentioning that the selection of ten measurements for each condition was also made by preliminary trials. To verify the International Tolerance (IT) grade, the reference values from

Table 6. International Tolerance (IT) grade, adapted from [24].

Nominal Dimension (mm)		International Tolerance (IT) Grade
		IT1	IT2	IT3	IT4	IT5	IT6	IT7	IT8	IT9
Above	Up to (Including)	Tolerance (µm)
6	10	1.0	1.5	2.5	4.0	6.0	9.0	15.0	22.0	36.0
10	18	1.2	2.0	3.0	5.0	8.0	11.0	18.0	27.0	43.0
18	30	1.5	2.5	4.0	6.0	9.0	13.0	21.0	33.0	52.0
30	50	1.5	2.5	4.0	7.0	11.0	16.0	25.0	39.0	62.0

were used to establish comparisons with the those obtained by calculation of the mean values.

2.4. Surface Roughness and Surface Morphology

The roughness measurement was carried out using a MITUTOYO surface roughness tester model SJ201 P/M, with a resolution of 0.01 μm and a measuring range of ± 5 µm, following the compliance standards: ISO (International Organization for Standardization), ANSI (American National Standards Institute), JIS (Japanese Industrial Standard), and VDA (Verband der Automobilindustrie—German Automotive Industry Association). A diamond stylus tip with 5 μm radius was used. A sampling length of 0.8 mm and a gaussian filter were adopted, according to the ISO 12085 standard [37]. Five sampling lengths were considered, resulting in a 4 mm evaluation length. For each workpiece, nine measurement cycles were performed. Measurements were equally distributed on the machined surface, and they were carried out perpendicularly to the grinding direction, i.e., perpendicular to the feed direction. The Ra parameter was selected for analysis.

All measurements were carried out at a controlled room temperature of 20.0 ± 1.0 °C. The instruments and workpiece were maintained at this temperature for approximately 12 h prior to the measurements under a granite table. The roughness tester was turned on 3 h before starting the measurements. The temperature was monitored by a digital thermo-hygrometer with a resolution of 0.1 °C and a nominal range from −20.0 to 60.0 °C. According to the calibration certificate, this thermo-hygrometer has an expanded uncertainty of 0.3 °C for a coverage factor of 2.00%.

The surface morphology of ground surfaces was evaluated with the aid of a HITACHI TM 3000 (Hitachi, Tokyo, Japan) scanning electron microscope (SEM) using 1k magnification and an accelerating voltage of 10 kV. The SEM images were also used to elaborate 3D schemes in the open-source software Octave (version 8.3) to facilitate comparative analysis.

2.5. Instant Grinding Power Measurements

The electric power gathered during grinding was calculated throughout the instantaneous signals of electric current and voltage of the electric motor of the machine tool. Power consumption was determined from voltage and current values, corresponding to the energy required to drive the spindle motor of the grinding wheel. The motor is a three-phase induction type with a rated power of 2.5 kW. Two Hall-effect sensors were installed on each phase for monitoring. Current was measured using an HAS 50-600S sensor (LEM International SA, Meyrin, Switzerland), which detects the magnetic field and converts it into a voltage signal. Voltage was measured with an LV-20P sensor, connected in parallel to the potential difference of interest.

A Ni 6001 Board, National Instruments, was used for the analogic/digital conversion; with 9.1 mV typical accuracy, the acquisition rate selected was 20 kS/s. Finally, to amplify the signal and reduce the noise an electrical circuit was used. After the trial, the collected signal was treated and analyzed. The mean value of the electric power during cutting (material removal) was selected to compare the different grinding conditions tested.

2.6. Design of Experiment (DOE) and Statistical Analysis

A full factorial design was chosen as shown in

Table 7. Design of experiment (DOE) employed for grinding tests.

Test	Cooling-Lubrication Condition (CLC)	Radial Depth of Cut (ae) [µm]
1	Flood	20
2		40
3	MQL WG (only oil)	20
4		40
5	MQL MG 0.05 wt. %	20
6		40
7	MQL MG 0.10 wt. %	20
8		40

. It contains two factors: cooling-lubrication condition (CLC) and radial depth of cut (a_e), the former with four levels and the latter, two. Each test was replicated once. As previously mentioned, ten dimensional measurements were conducted for each test, thereby resulting in a total of twenty measurements for each test as detailed in Table 7.

An analysis of variance (ANOVA) was performed to better analyze the effect of each factor, including their interaction. A confidence interval of 95% was considered in the analysis and therefore a p-value less than 0.05 indicates a statistically significant effect. The ANOVA was performed for the results of surface roughness (Ra) and electric power, which, different from removed height results, presented normally distributed data as shown in Figure 4 (p value higher than 0.05 for the Anderson–Darling test).

3. Results and Discussion

3.1. Removed Height and Dimensional Tolerance After Grinding

The removed height values as a function of cooling-lubrication condition (CLC) are shown in

Table 8. Removed height, lowest and highest deviations, and international tolerance grade calculated for Inconel 718 alloy height after grinding with each cooling-lubrication condition tested in this work. a_e = 20 µm.

		Flood	MQL WG	MQL MG 0.05%	MQL MG 0.10%
Removed height (mean value ± expanded uncertainty) [µm]		10 ± 2	11 ± 2	7 ± 2	5 ± 2
Difference between removed height and set radial depth of cut (20 µm)	Highest deviation [µm]	12	11	15	17
	Lowest deviation [µm]	8	7	11	13
International tolerance grade for nominal dimension up to 30 mm		IT5–IT6	IT5–IT6	IT6–IT7	IT6–IT7

and

Table 9. Removed height, lowest and highest deviations, and international tolerance grade calculated for Inconel 718 alloy height after grinding with each cooling-lubrication condition tested in this work. a_e = 40 µm.

		Flood	MQL WG	MQL MG 0.05%	MQL MG 0.10%
Removed height (mean value ± expanded uncertainty) [µm]		17 ± 3	27 ± 1	9± 2	23 ± 2
Difference between removed height and set radial depth of cut (40 µm)	Highest deviation [µm]	26	14	29	15
	Lowest deviation [µm]	20	12	33	19
International tolerance grade for nominal dimension up to 30 mm		IT8	IT6–IT7	IT7–IT8	IT7

for a_e = 20 µm and a_e = 40 µm, respectively. The results include mean and expanded uncertainty, as well as the International Tolerance (IT) grade in accordance with Table 6 [24], determined as the difference between the measured removed height and the set radial depth of cut, considering the nominal dimension of the workpiece (total height up to 18 mm).

The tolerance grade presented on the last line of both Table 8 and Table 9, was calculated considering the workpiece dimension, in height, which was 20 mm, thus being categorized in the range “above 18 mm and up to 30 mm” from Table 6. As the removal was small (20 µm and 40 µm) when the grinding was completed, the workpiece remains in the same range for IT evaluation. To exemplify this, for such a range to be considered IT4 grade the maximum variation should be ±6.0 µm, while for IT5, IT6, and IT7 the deviations should be ±9.0 µm, ±13.0 µm, and ±21.0 µm, respectively.

The results are also graphically presented in Figure 5 and Figure 6, for radial depths of cut of 20 µm and 40 µm, respectively. The expanded uncertainty (95%) is shown as error bars. It is worth mentioning that the dashed line represents the radial depth of cut (a_e), and the removed height denotes the linear dimension reduction in the workpiece’s height after grinding, which is the real depth of cut (a_r), as illustrated in Figure 2.

It can be seen from Table 8 and Figure 5 that the highest values of removed height were recorded after grinding with flood and MQL WG, which represents an achievement considering the predicted material removal in linear dimension for a_e = 20 µm was about 13.8 µm, as previously mentioned in Section 2.3. Considering the workpiece height and the linear deviation from the predicted height after grinding, the dimensional tolerances when applying these cooling-lubrication conditions fall into the IT grade in the range between IT5–IT6.

According to Marinescu et al. [38], system deflection (δ) and grinding wheel wear (a_sw) are the major factors that can contribute to reducing the actual depth of cut (removed height). Thermal expansion (a_t), on the other hand, presents the opposite effect, thereby contributing to increase the removed height. In this context, worth mentioning that, despite the better cooling capacity of the flood technique in comparison to MQL [38,39], both CLCs resulted in similar IT grades for a_e = 20 µm, while MQL MG 0.05% and MQL MG 0.10% (IT6–IT7) promoted lower material removal (lower removed height), which is usually associated with system deflection and/or wheel wear.

The use of nanoparticles dispersed in the cutting fluid applied via the MQL technique usually contributes to reducing grinding forces as reported by many authors [40,41,42,43], especially considering low concentrations of solid particles which prevent agglomeration issues. With respect to grinding wheel wear, Kalita et al. [44] observed a 55% reduction in wheel wear rate when grinding with nanoparticles (molybdenum disulfide—MoS₂) dispersed in cutting fluid and delivered via the MQL technique. In this context, it is reasonable to infer that the use of nanofluids would, in principle, reduce cutting efforts, which tends to reduce system deflection and wheel wear, thereby increasing the removed height. Thus, the results found in this work indicate that another factor, besides system deflection, wheel wear, and thermal expansion, plays a significant role in the removed height in the grinding process, which may be associated with chip formation.

The chip formation mechanism in grinding of ductile materials can be divided into three stages: rubbing, plowing, and cutting [45]. In the first stage, the abrasive grit penetration in the workpiece is low and only elastic deformation takes place, with no material removal. The friction then increases with grit penetration and the material is submitted to plastic deformation by sliding through forwards and sideward of abrasive grits, which is a type of abrasive wear mechanism known as micro-plowing.

When abrasive grit penetration reaches a critical value, chip formation due to material shearing occurs (micro-cutting). In practice, micro-plowing and micro-cutting occur simultaneously during chip formation, and the predominance of one over the other depends on tribological conditions at the contact zone such as type of abrasive grit and workpiece, cutting and cooling-lubrication conditions, and cutting-edge geometry [31].

With respect to the cooling-lubrication condition, the increase in lubrication can reduce the friction of workpiece sliding through the abrasive grits with no material removal, consequently favoring micro-plowing over micro-cutting, which might have caused the least IT grade for the sample machined under the CLC with MG.

On the other hand, the improved lubrication can also reduce the friction between the chip being formed and the rake face of abrasive grits, which in turn favors micro-cutting over micro-plowing [43]. In this context, the results in Figure 5 suggest that the higher lubrication capacity of the cutting fluid enhanced with multilayer graphene (higher viscosity than coolant without graphene—MQL WG, as shown in Table 4) contributed to reduce the friction of material sliding through abrasive grits with no material removal, thereby reducing the removed height in comparison to the other cooling-lubrication conditions.

When grinding Inconel 718 alloy with a_e = 40 µm, different trends can be noticed as shown in Table 9 and Figure 6. When grinding with the severest cutting condition, the traditional MQL technique (MQL WG) promoted the highest removed height, thereby resulting in an IT grade in the IT6–IT7 range. This range, however, was the same as that obtained for workpiece ground under MQL MG 0.10%. Machining with the flood coolant delivery technique, on the other hand, resulted in lower removed height, when compared to MQL WG and MQL 0.10%, thereby indicating that thermal expansion played an important role. Higher values of a_e increase heat generation at the contact zone, thereby developing higher temperatures [46], especially in the grinding of Inconel 718 with a SiC abrasive wheel due to the low thermal conductivity of the workpiece material and the high friability and micro-fracture and wear flat tendency of the abrasive grits, which contributes to increasing heat generation and temperature during grinding [47].

In this context, the higher temperature at the cutting region increases workpiece thermal expansion (a_t), consequently increasing the real depth of cut (a_r) and the removed height as well [Equation (1)]. The enhanced cooling capacity of flood in comparison to the MQL technique stands out in this case, attenuating thermal expansion and reducing the removed height, thereby resulting in higher dimensional deviations (higher IT grade).

Regarding the use of graphene dispersed in the cutting fluid when grinding with a_e = 40 µm, lower removed height was generated when grinding with this CLC in comparison to the MQL WG condition as shown in Table 9 and Figure 6, thereby indicating that the presence of graphene contributed to a better temperature control, irrespective of the nanoparticle concentration tested. This can be attributed to the increase in cooling capacity which is directly proportional to the thermal conductivity of the nanofluids [48].

Low grinding temperatures reduce workpiece thermal expansion and the removed height, similar to what was observed when grinding with the flood CLC technique. Also, it is possible to notice from Figure 5 and Figure 6 that grinding using the flood technique resulted in a slighted increase on IT tolerance (IT6–IT7) grades than MQL MG 0.05% (IT7). This effect can be associated with the different cooling rates in the workpiece center and the extremities of the workpiece, thereby increasing the difference between the removed height at such locations. On the other hand, when grinding with the MQL technique, which presents a lower cooling capacity, the lower and more homogeneous cooling rates led to smaller error bars.

Machining with higher radial depth of cut values tends to favor the micro-cutting wear mechanism over micro-plowing due to the increase in abrasive grit penetration in the workpiece [49]. Thus, the strong influence of a_e on removed height was predicted, as well as the increase in removed height with a_e. However, it is noteworthy that, although the removed height increases with depth of cut, the dimensional tolerance is adversely affected as can be noticed by comparing the values in Figure 5 and Figure 6, which show the importance of reducing material removal rate, in this case the radial depth of cut, to achieve lower dimensional deviations (lower IT grade).

3.2. Surface Finish and Surface Morphology

The main effects and interaction graphs from the ANOVA results for surface finish (the Ra parameter) are shown in Figure 7 with their respective p-values. It is notable that CLC and its interaction with depth of cut (a_e) presented a significant effect on surface roughness results (p-value < 0.05). The radial depth of cut, on the other hand, did not significantly affect roughness (Ra) for the conditions tested in this work, although there is a tendency of increasing Ra with a_e as is usually reported in the literature. This is because for smaller radial depths of cut, a smaller height of each abrasive grain will interact with the workpiece surface, resulting in shallower abrasive marks, thus reducing the overall surface roughness.

In Figure 7a it is also possible to notice that the CLC that resulted in the lowest value was MQL MG 0.05%, followed by MQL MG 0.10% and flood environments. It is important to highlight the lower efficiency of the MQL WG in comparison with flood, which are the most common coolant delivery techniques. This difference has been attenuated with the presence of the MG particles, indicating that they are a possible method to approximate the MQL to flood, which is widely pursued in the machining literature [50].

From Figure 7b, with respect to the effect of CLC, the grinding with multilayer graphene (MG) significantly reduced Ra roughness for the conditions used in this work. However, worth mentioning that this reduction was dependent on MG concentration, especially when machining under the severest cutting condition (a_e = 40 µm), in which grinding at the lowest MG concentration outperformed the other CLCs tested.

Cao et al. [51] evaluated the grinding of Inconel 718 with the aid of an ultrasonic system to perform intermittent cutting using a water-miscible cutting fluid (Brix 5%) and a radial depth of cut (a_e) of 20 µm and analyzed the surface finish. The authors obtained Ra values of approximately 0.35 µm, which is higher in comparison to the results of Ra found in this present work considering the flood condition and a_e = 20 µm. The difference in the results may be attributed to the presence of deeper marks at the ground surface observed by Cao et al. [51], which the authors attributed to the kinematics of the vibration-assisted process. This indicates that depending on the application, ultrasonic-assisted grinding might be overcome by simply applying the proper cooling-lubrication condition.

The SEM images of ground surfaces are shown in Figure 8 with their respective 3D scheme. It is notable from this figure that all machined surfaces presented a typical ground surface morphology, showing well-defined grooves and some regions with evidence that intense plastic deformation took place, especially when grinding with the combination of the lowest a_e and multilayer graphene [Figure 8c,d]. This result indicates that micro-plowing mechanism was predominant during chip formation when grinding using these CLCs, which corroborates the previous discussions.

The predominance of the micro-plowing mechanism over micro-cutting is associated with the abrasive grit penetration degree and the interfacial shear strength between the grit and the surface. The latter is in turn related to the lubrication between the grit and the workpiece, in which high lubrication contributes to plastic deformation only, therefore involving the micro-plowing mechanism [52]. In this context, the greater predominance of micro-plowing when using MG can be attributed to better lubrication at the interfacial shear strength between the grit and the surface, thereby facilitating the displacement of the workpiece sideways (material side flow) of the abrasive grits, resulting then in plastic deformation with no material removal.

For the more severe cutting condition (a_e = 40 µm), this issue is strongly minimized due to the increase in the abrasive grit penetration degree, with the exception of the grinding condition with the traditional MQL technique (MQL WG) as shown in Figure 8f. This suggests that the material presented more plastic behavior for this specific grinding condition, thereby increasing the predominance of the micro-plowing mechanism in comparison to the surface ground with lower a_e [Figure 8b]. This may be associated with higher temperatures due to the high depth of cut values and insufficient cooling capacity provided by the traditional MQL technique without multilayer graphene (MG), only oil, which supports the assumption that the higher removed height for this condition is associated with thermal expansion.

3.3. Electric Power

The main effects and interaction graphs of ANOVA for electric power results are shown in Figure 9. For a_e = 20 µm, the lowest value was observed for the MQL 0.10% condition (0.82 kW) and the highest value was observed for the MQL WG (1.24 kW). With a_e = 40 µm, the lowest and highest values were obtained machining with MQL MG 0.05% (1.51 kW) and MQL WG (1.91 kW), respectively. It can be noticed that the electric power during grinding significantly increased with the radial depth of cut (ae), irrespective of the cooling-lubrication condition (CLC). This is attributed to the higher chip thickness when using higher values of a_e, which increases the amount of material being removed from the workpiece, thereby increasing grinding forces and electric power as well. The increase in grinding forces with a_e in grinding of Inconel 718 is predicted once it improves material removal rates, and it was also observed by Yao et al. [53] and Ruzzi et al. [49].

With respect to the influence of CLC, it can be seen from Figure 9 that, although not statistically significant (p-value > 0.05), the use of the MQL technique with multilayer graphene (MG) tends to decrease the electric power during grinding in comparison to flood and traditional MQL technique (without nanoparticles). The interaction between CLC and a_e also shows no statistical significance for electric power results. However, it is important to notice that both main effects and interaction graphs for electric power results (Figure 9) presented similar trends as those observed for removed height results. This indicates that these output variables can be well correlated as shown in Figure 10, in which the lower the electric power, the lower the removed height.

Figure 11 contains the mechanisms diagram to illustrate the impact of the cutting on the grinding power. It should be highlighted that it is expected that a lower power is required for lower depths of cut, when machining with similar cutting parameters and CLC conditions, since the material removal rate will be lower, thus less rubbing, plowing, and shearing mechanisms will occur, resulting in lower required power.

Onishi et al. [54] made a similar evaluation for cylindrical grinding of S45C steel and verified a strong correlation between the machine grinding power and the volume of removed material (grinding stock). They reported that it was possible to estimate the final dimension of the workpiece by monitoring the power consumption of the wheel motor.

In this sense, the association of the reduction in both power and removed height on the Inconel 718 sample, especially when machining with a_e = 20 µm; it can be inferred that the presence of MG at the contact zone indeed improved lubrication with respect to friction with no material removal (micro-plowing) when grinding with a SiC grinding wheel under the conditions investigated. A reduction in friction force strongly reduces the total grinding force, since it is its main component [55], thereby reducing electric power as observed in the results found in this work.

Similarly, Kishore et al. [56] applied different CLCs for the grinding of Inconel 625, applying MoS₂, multi-wall carbon nanotubes (MWCNTs), and a fluid in combination with 0.01, 0.05, and 0.25 wt.% concentrations of nanoparticles. In terms of cutting forces, the authors observed lower values when grinding using nanofluids with concentration of 0.05 wt.%, regardless of nanoparticle type. However, it is worth mentioning that lower cutting forces were associated with higher Ra values for the conditions tested by the authors.

4. Conclusions

After the experimental trials and the statistical analyses, the following conclusions can be drawn:

• Considering the lowest depth of cut (a_e = 20 µm), the dimensional accuracy in terms of International Tolerance (IT) grade was in the range from IT5–IT6 for flood and MQL WG (only oil), and IT6–IT7 for MQL MG 0.05% and MQL MG 0.10%;
• The improved lubricating properties of the MQL MG 0.10% contributed to reduce the friction of material sliding through abrasive grits with no material removal, thereby reducing the removed height in comparison to the other cooling-lubrication conditions;
• Grinding with radial depth of cut (a_e) of 40 µm increased the removed material in comparison to the lower a_e (20 µm). In terms of the cooling-lubrication conditions tested, the MQL WG resulted in the lowest International Tolerance (IT) grade (IT6–IT7), followed by MQL MG 0.10% (IT7), MQL MG 0.05% (IT7–IT8), and flood (IT8);
• Overall, regardless of the a_e, the MQL without graphene (MQL WG) was the condition in which the removed height was closest to the set depth of cut;
• With respect to the multilayer graphene (MG), its presence in the cutting fluid contributed to reducing the removed height due to an enhanced micro-plowing mechanism, thereby indicating an improvement in lubrication capacity related to plastic deformation with no material removal. Additionally, the reduction in the removed height when grinding with MG suggests lower thermal expansion of the workpiece during grinding, which may be associated with better temperature control;
• The addition of multilayer graphene (MG) significantly reduced Ra roughness, and the condition MQL MG 0.05% outperformed others when grinding with the severest cutting condition (a_e = 40 µm);
• Evidence of intense plastic deformation occurred when grinding with MG and a_e = 20 µm, indicating that micro-plowing was the predominant mechanism during chip formation. This characteristic was also observed when grinding at the highest depth of cut value combined with the MQL WG condition, possibly associated with insufficient cooling capacity of the MQL technique without MG;
• The electric power increased with the radial depth of cut due to the higher chip thickness, which led to more material being removed from the workpiece, which increases grinding forces and electric power as well. The highest measured values were obtained for the MQL WG condition being 1.24 kW and 1.91 kW for a_e equal to 20 µm and 40 µm, respectively. The lowest values for a_e = 20 µm were 0.82 kW measured for the MQL 0.10% condition and for a_e = 40 µm it was 1.51 kW as a result of the MQL 0.05% condition;
• With respect to the influence of CLC on the electric power, the lower values were obtained for the conditions with MG, evidencing that the presence of MG at the contact zone in fact improved lubrication, even resulting in no material removal (micro-plowing) when grinding with a_e = 20 µm, and that a reduction in friction force strongly reduces the total grinding force, and therefore reduces the grinding power;
• The use of the MQL technique with graphene enriched cutting fluids can improve Inconel 718’s grindability in terms of surface finish and electric power, although slightly increasing dimensional deviation in comparison to traditional MQL (only oil). Thus, based on the findings of this work, these cooling-lubrication conditions can replace the flood technique for a more environmentally sustainable grinding process.