Novel application of multitubular nozzle in throughfeed centerless grinding process of bearing steel SAE 52100

Through-feed centerless grinding is a high-productivity machining process widely used for mass production of cylindrical parts and rotationally symmetrical parts in automotive and bearing industries. Grinding process is strictly related to large amount of heat generated in the cutting zone. This process characteristic makes pivotal and indispensable the effects of lubrication and cooling provided by metal working uid (MWF). In this regard, this work aims to contribute to the study of MWF application in grinding process, analyzing the effects of the optimized technique developed for an eco-friendlier use of MWF by the application of a novel multitubular nozzle. The results obtained for the novel designed multitubular nozzle with compressed air outperformed the conventional nozzle system with lower oil volume employed.

methods. According to Maruda et al. [24], dry machining is not recommended for surfaces with high accuracy requirement, because higher amount of heat is transferred to the workpiece as a result of the friction between tool/chip interface, affecting negatively the tool life. In grinding, the total elimination of MWF results in higher temperatures during the grinding process, which may affect the surface integrity and geometric precision of the ground part, in addition to an increase in wheel wear and clogging phenomenon as reported by Oliveira et al [25]. This technique is generally applied to processes with de ned cutting geometry, such as turning and milling.
A promising approach to overcome the limitations of dry machining is through the MQL technique. This technology combines the functionality of refrigeration with very low uid consumption, with ow rates ranging from 10 mL/h to 100 mL/h and application pressures ranging from 0.39 MPa to 0.59 MPa, eliminating the need for disposal and minimizing possible impacts to the environment as inferred by Klocke et al. [26], since all the uid amount is consumed directly during process. However, this technique still needs to be improved in order to be used in grinding processes, especially in centerless grinding processes using conventional grinding wheels. For high rates of MRR, the ow rate of lubricant is ine cient to remove the chip from the cutting zone, resulting in wheel clogging. Thus, the chip lodged in wheel pores reduces the grinding wheel active surface (GWAS) and increases load on tool, decreasing the lubrication capacity of the cutting zone, hindering the surface nal quality and greater diametric wheel wear. According to Rodriguez et al. [27], the MQL technique with water addition and wheel cleaning jet indicates an environmental-friendly potentiality to reduce risks to environment and carbon footprint. On the other hand, Lopes et al. [28] infer higher cutting temperature are generated in pure MQL grinding, and consequently accentuating the clogging occurrence on GWAS.
De Moraes et al. [29] studied application of the MQL with water for AISI 52100 steel grinding and reported that this lubri-cooling technique reduced surface roughness, form deviation error and wheel wear due to its higher cooling capacity, presenting results close to conventional MWF application ( ood). On the other hand, the authors infer that the addition of water in MQL grinding decreases the lubrication effect in cutting zone, consequently increasing the cutting power. However, the addition of water in MQL grinding shows to be a viable alternative to improve the wider employability of this lubri-cooling technique. Garcia et al. [30] tested the addition of 1:5 oil-water ratio in the MQL grinding with wheel cleaning jet of AISI 52100 and concluded that this technique outperformed the other dilution proportions (1:0, 1:1 and 1:3) and produced surface nishing quality results 22% higher than the ood technique.
Studies carried out by Tawakoli et al. [14], Webster and Grün [17] and Monici et al. [31], indicate that the optimized application and new nozzle designs can be the most e cient means to make the application more e cient possible, allowing the reduction of the consumption of MWFs in the grinding processes.
Aiming to guarantee a better quality of the ground parts and the reduction of manufacturing costs and risks to human health related to the use of MWFs during the through-feed centerless grinding process, this current work propose the development of an optimized method of lubri-cooling in order to make it possible a more environmentally friendly application of MWFs, thus guaranteeing the integrity of the ground surface part, while reducing manufacturing costs and lowering environmental impacts. In a groundbreaking way, this study compared the effects caused on ground samples by the conventional application of cutting uid and the optimized lubri-cooling method that employed a new design of multitubular nozzle with uid emulsion and emulsion with compressed air simultaneously.

Experimental Procedure
Tests were carried out on a centerless grinder, manufactured by Herminghausen, model SR4-25. A silicon carbide grinding wheel (model C120 TB24, outer diameter of 500 mm and 250 mm of width) was used in tests. The Al2O3 regulating wheel with rubber bond (model ARR120, outer diameter of 300 mm and 250 mm of width). Both wheels were supplied by SIVAT abrasives. For the specimens, geometries were selected with 13 mm in diameter and 18 mm in length, consisting of SAE 52100 steel (quenched and tempered), with an average hardness of 62 HRC and used for rollers bearings. Details related to wheels and specimens are shown in Fig. 1. In order to guarantee an ideal condition for the evaluation of the specimens, the wheels were dressed after each test.
The cutting wheel was dressed with a conglomerate dresser (10 mm wide, 4 mm heigh and 10 mm thick), while for the regulating wheel a single point dresser was applied. The processes were done according to the manufacturer's recommendations, due to the nature of each wheel.
This study evaluated different lubri-cooling techniques: conventional (CN), multitubular with emulsion (ME) and multitubular with emulsion and compressed air (ME + CA). The emulsion cutting uid was employed at 4 different ow rates using conventional nozzle and multitubular nozzle. For all tests, the cutting uid used was Syntilo 290 BR, supplied by Castrol, 4% of oil concentration. The conventional application system is shown in Fig. 2 (a) Fig. 2 (b) shows the position of the conventional nozzle in front of the wheels and the workpiece. The conventional nozzle delivers the uid abundantly, with great dispersion and with a non-uniform pro le in the lubrication area. The ow rates for conventional application were 10; 20; 30 and 40 L / min with a pressure of 0.5 MPa.
The system is comprised of the multitubular nozzle, machine tool, grinding wheels and workpiece is shown in Fig. 3a. The nozzle geometry was based on the one proposed by Webster [17], indicating that application of this nozzle generates a reduction in the cutting uid turbulence at the nozzle outlet, providing a jet of directional uid with low dispersion. In order to produce a good uid delivery, the nozzle was positioned at a distance of 80 mm from the cutting region. The positioning of the nozzle at the cutting interface is shown in Fig. 3b. The ow rate of compressed air during the tests using the multitubular nozzle was 170 L / min, with a pressure of 0.6 MPa, generating a uid speed of 45 m / s (similar to the peripheral speed of the wheel, i.e., cutting speed). Table 1 shows the process parameters and the lubri-cooling methods employed in the tests. In order to check and compare all the grinding conditions produced, surface roughness, roundness deviation and residual stress were analyzed. The integrity of the ground surface was also checked by magnetic particles inspection and super cial burning was detected by the chemical attack method and micrograph. For the GWAS (grinding wheel active surface), the topography of the cutting surface was evaluated.
The mean surface roughness (Ra) was measured the rougher equipment model Perthometer M2, manufactured by Mahr Ltda. A 0.8 mm cut-off value was applied, using a Gaussian lter for roughness measurements. The measurements of the roundness deviations were performed using an equipment model MWA 100 B, manufactured by SKF. All results shown are averages of 5 measurements performed.
One of the parameters that can indicate the effectiveness of the grinding process is the residual stress.
Due to the heat generated by the friction in the cutting zone, along with the mechanical deformations resulting from the chip removal process, the ground surfaces show surface changes, whether they be tensile or compression as reported by Macherauch [32]. According to Field et al. [33], excessive tensile in these components can compromise the mechanical reliability of the part. The Barkhausen Noise method was employed to measure the residual voltage, performed by a Rollscan 300 CPU s / n 1775 device, adjusted with a magnetization frequency of 200 Hz, with a magnetization voltage of 15 V, together with a 70-200 kHz and an S6362 sensor. The pieces were xed on a bracket that allowed them to rotate when measuring their axial direction. The data were collected using ViewScan software.
Finally, the GWAS was observed after performing each test by a portable digital microscope under magni cation of 50X. This process observation aimed to investigate GWAS condition in terms of clogging occurrence.

Results And Discussions
This section discusses the results obtained for the different MWFs application proposed (CN, ME and ME + CA) in through-feed centerless grinding process of bearing steel SAE 52100.

Analysis of surface roughness and GWAS condition
The Fig. 4 presents the mean surface roughness (Ra) values for different lubri-cooling techniques (CNconventional nozzle, ME -multitubular with emulsion and ME + CA -multitubular with emulsion and compressed air) and different ow rates employed.
By Fig. 4 and irrespective of nozzle used, it can be seen that the reduction of surface roughness values is related to the increase of ow rate. As reported by Ramesh et al. [34] and Bianchi [10], a higher ow rate results in higher coolant velocity, improving the lubrication effect and contributing to a better grinding heat dissipation characteristic. In general, the results obtained for ME and ME + CA nozzles were lower than those obtained for CN.
Regarding to application of ME + CA technique, it was noticed that this technique resulted in lower values of surface roughness and variability in comparison to ME and CN. For tests with ME + CA nozzles at the ow rate of 10 L/min, the surface roughness obtained was 73 and 28% lower than, respectively, CN and ME nozzles.
For tests with ME + CA nozzles at the ow rate of 20 L/min, the surface roughness obtained was 24 and 18% lower than, respectively, CN and ME nozzles. For tests with ME + CA nozzles at the ow rate of 30 L/min, the surface roughness obtained was 30% lower than CN nozzle and no mathematical difference in comparison to ME nozzle. For tests with ME + CA nozzles at the ow rate of 40 L/min, the surface roughness obtained was 17 and 9% lower than, respectively, CN and ME nozzles.
The lowest surface roughness values obtained for ME + CA can be justi ed by the fact that compressed air increases the velocity of cutting uid in the nozzle end, producing a higher force in the cutting zone.
Higher is the force of uid jet, higher is the pressure of cutting uid in the cutting zone; promoting a better cleaning of chips generated in the cutting zone during the grinding process and consequently increasing the lubricant effect of lubri-cooling method. According to Bianchi et al [10], the use of wheel cleaning jet in MQL with different ow rates (30, 60 and 120 mL/h) signi cantly lowered the surface roughness Ra for AISI 4340 steel grinding process, producing results similar to ood MWF method.
The best surface roughness values were obtained for the highest ow rates of the cutting uid, that is, 30 L / min and 40 L / min using the ME + CA nozzle, thus con rming a better grinding condition. Unlike the CN, the ME + CA nozzle promotes a directional jet, with low dispersion at the nozzle outlet and with a speed close to the peripheral speed of the grinding wheel, due to the use of compressed air as an auxiliary uid. With this nozzle, the uid penetrates easily in the aerodynamic barrier around the wheel and guarantees a greater amount of uid in the cutting zone, thus ensuring a better lubricating and cooling effect. The greater dilution of the cutting uid provides a faster removal of the chips, preventing them from affecting the quality of the piece. Bianchi et al. [10] state that the occurrence of clogging phenomenon on GWAS is hindered by pure oil application because of higher uid viscosity results in solid suspension, and consequently grout formation which lodges in the wheel pores.
Yoshimura et al. [35] infers that the addition of water in MQL grinding generates lower plastic deformation caused by higher coolant effect by water addition, i.e., higher heat dissipation in the cutting zone by phase transition (from liquid water to vapor). According to Sato et al. [36], this reduces the wheel clogging phenomenon and keeps the wheel grit sharp [37].
According to Ramesh [34], the increase of uid speed promotes a lubricating effect with better heat dissipation effect in the grinding and, thus, lower surface roughness values. According to Daniel et al.
[38], the MWF heat transfer increases with the uid ow velocity. The Turkey method was employed in order to compare the surface roughness values for ME + CA technique in different ow rate, considering a signi cance level of 5%. Regarding to ow rate values, the differences in surface roughness were small and were signi cant between 40 L / min − 10 L / min and 40 L / min − 20 L / min, as shown in Fig. 5.
The high values of surface roughness Ra for the tests using the CN and ME application, for ow rates of 10 and 20 L / min, as shown in Fig. 4, are related to the low e ciency in heat dissipation and in removing the chip from the cutting zone, which may cause the clogging phenomenon on GWAS. Figures 6 (a) and 6 (b) depict evidence of the occurrence of this phenomenon. For the other conditions, no wheel clogging was observed. According to Malkin [39], the presence of chips in the cutting region hinders the cutting performance, increases efforts during material removal and, consequently, increases the surface roughness value. According to Walker et al. [40], the eco-e cient application of a cleaning jet by compressed air diminishes the clogging phenomenon, further decreasing the volume of grout formation on GWAS and, consequently, minimizing the effects of the rubbing and ploughing effects on the ground surface. Thus, Rodriguez et al. [41] state that the surface roughness results can be reduced by 45%.
Additionally, the authors infer that excessive temperatures in the contact zone result in reduction of yield strength and increasing of material strain from the ground material, increasing the lodging and adherence of chips in the grinding wheel pores.
Comparing the surface conditions of the grinding wheel using the CN (Fig. 6a) and those obtained for ME (Fig. 6b) at 10 L / min, it is possible to observe in both cases a probable clogging of the GWAS. For the condition employed in the current tests, the surface roughness values generated for CN and ME application did not show mathematically signi cant differences.
The ME + CA application presented a more e cient condition than the other two lubri-cooling methods, resulting in lower surface roughness values. This can be explained by the high outlet speed of the cutting uid jet using this technique. According to Webster and Grün [17], the uid outlet speed associated with the ow rate and the type of nozzle have a signi cant in uence on the coolant effect, as it increases the cleaning e ciency in cutting zone. Moreover, it facilitates the removal of material and results in better surface quality. Thus, the process performance was improved by the application of the cutting uid associated with the high speed provided by the auxiliary uid using ME + CA technique and, thus, greater pressure force in the cutting zone. For all conditions tested for ME + CA, no evidence of clogging on GWAS was observed. Figure 6 (c) shows the condition of GWAS for the lowest ow, i.e., 10 L / min. As can be seen in Fig. 6, higher wheel wear could be seen for ME application. According to Rodriguez et al. [41], the abrasive tool wear occurs by the following mechanisms: grain detachment from bond, grain fracture and wear. As reported by Martini et al. [42], excessive temperatures promote the reduction of bond material strength, what increases tool wear. As observed for ME + CA application, the cutting force is   In general, it can be seen in Fig. 8 that the Ra surface roughness values for all conditions tested were less than the maximum tolerance speci ed for the grinding process of cylindrical rollers for SAE 52100 steel bearings, i.e., 0.15 µm. According to Malkin and Guo [44], the allowable range of Ra roughness values for the grinding process is between 0.2 to 1.6 µm. Thus, all values obtained using the CN method and the optimized methods (ME and ME + CA) are within the tolerance range for the grinding process.
Analyzing Fig. 8, it can be observed that lower values of surface roughness at 40 L / min when the ME + CA method is employed are related to the higher speeds of the cutting uid and the directional output of the cutting uid promoted by multitubular nozzle. Despite presenting lower values of surface roughness when compared to ME at 40 L / min, the multiple comparison test, as shown in Fig. 9, does not present mathematically signi cant differences between both techniques at 40 L / min. The higher speeds of the uid at the outlet of the nozzle promote the reduction of friction, due to greater presence of lubricant in the cutting region.

Roundness Deviation
The values of the roundness deviation obtained for the tested lubricooling conditions and material removal of 0.10 mm are shown in Fig. 10. Comparing the results of the roundness from Fig. 10, it is observed that the increase in ow rate resulted in lower roundness deviation, in particular, conditions tested with the CN and ME. In general, the values of the roundness deviation for ME + CA application were lower in comparison to CN application.
The reduction of roundness deviation is due to the fact that the higher ow rate promotes a better cooling effect in the cutting zone. According to Demeter and Hockenberger [45] and Malkin [39], greater is the di culty of the uid penetration in the cutting zone, greater is the amount of heat distributed to workpiece and consequently producing thermal expansion and causing an increase in the form deviation, i.e., mainly amplifying the roundness deviation. The best result of the roundness deviation can be seen in Fig. 10 for ME and ME + CA technique at 40 L / min.
The results of the mean values and variability of the roundness deviations (for total removal of 0.03 mm) are shown in Fig. 11.
By Fig. 11, it can be seen that higher values of the roundness deviations were recorded for the tests ME + CA at 20, 30 and 40 L / min. For the ow rate of 10 L / min, the value of the roundness deviation was lower in relation to the tests with CN and ME application.
For all tested conditions, the mean values of the roundness deviations were kept below the maximum allowed tolerance (1µm) for the ground rolling rollers. The roundness deviation during the centerless grinding process is mainly in uenced by the geometric variables (inclination angle of regulating wheel, rest blade) and dynamics of the grinding process (natural frequency of machine, wheel rotation, among others). In addition, it is in uenced by the magnitude of wheel wear, the MRR (material removal rate) and speci c energy.
In addition, the roundness pro le of the machined samples was evaluated and concluded that the deviation was not in uenced by the dynamic variables of grinding process. The roundness deviations for all conditions tested did not show deviations in the form of periodic undulation, therefore, not due to the in uence of dynamic factors in grinding process.

Residual Stress
The results of the residual stress measurements in "mp" (elastic magnetic parameter) using the Barkhausen noise method for all tests with CN, ME and ME + CA techniques with material removal of 0.03 mm, as shown in Fig. 20.
The mean values of the residual stress results for the lubri-cooling conditions with CN, ME and ME + CA, as shown in Fig. 12.
Barkhausen residual stress analysis is a non-destructive quantitative control method. Workpieces with high values of residual compressive stresses indicate low Barkhausen noise intensity (mp). In order to de ne the limits between the residual compressive and tensile stresses, an investigation of the ground surface using the X-ray diffraction technique was performed. Through the stress values (MPa) by the Xray diffraction and Barkhausen noise (mp) technique it is possible to de ne a factor to convert the values in "mp" to "MPa".
As shown in Fig. 20, it can be seen that the lubri-cooling method using the ME + CA presented, in general, residual stresses lower than those generated through the methods using the CN and ME + CA techniques.
Thus, it can be observed that workpieces ground with ME + CA technique suffered fewer thermal effects.
Comparing the residual stress values (Fig. 12) with the results of the metallographic analysis, a maximum residual stress value can be admitted for the recti cation of SAE 52100 steel of 80 mp, due the fact that the conditions using the ME with a ow rate of 40 L / min and for the ME + CA with a ow rate of 30 and 40 L / min, did not produce changes in the microstructure, i.e., the ground workpieces without grinding burns.

Conclusion
Based on the results and discussions previously presented, the following sentences can be concluded: 1. In general, the conditions obtained for the ME and ME+CA application proved to be more e cient than those obtained of CN application regarding to all parameters analyzed, i.e., surface roughness and integrity, roundness deviation and thermal damage.
2. Regarding to surface roughness results, it was noticed that the application of ME+CA technique resulted in lower values of surface roughness and variability in comparison to ME and CN. For tests with ME+CA nozzles at the ow rate of 10 L/min and depth of cut of 0.1 mm, the surface roughness obtained was 73 and 28% lower than, respectively, CN and ME nozzles. For tests with ME+CA nozzles at the ow rate of 20 L/min, the surface roughness obtained was 24 and 18% lower than, respectively, CN and ME nozzles. For tests with ME+CA nozzles at the ow rate of 30 L/min, the surface roughness obtained was 30% lower than CN nozzle and no mathematical difference in comparison to ME nozzle. For tests with ME+CA nozzles at the ow rate of 40 L/min, the surface roughness obtained was 17 and 9% lower than, respectively, CN and ME nozzles.
3. Both ow rate and the type of nozzle signi cantly in uence the values of surface roughness, roundness deviation, residual stress and depth of the thermally affected layer during the grinding process.
4. Comparing the results of the roundness for depth of cut of 0.10 mm, it is observed that the increase in ow rate resulted in lower roundness deviation, in particular, conditions tested with the CN and ME. In general, the values of the roundness deviation for ME+CA application were lower in comparison to CN application. For the application of ME+CA technique, the mean values of the roundness deviations obtained, in all conditions tested, were below the maximum allowable tolerance for through-feed centerless grinding process of rolling rollers, that is, 1 μm. 5. The surface roughness values for the tests using the ME+CA were satisfactory for all conditions tested, with an average roughness value (Ra) below 0.15µm. The application of compressed air proved to be an e cient way to improve the cooling condition during the grinding process, since during the tests under ME+CA technique, no clogging occurrence on GWAS was recorded.
6. The analysis of results indicates that the new nozzle concept developed using either ME or ME+CA technique proved to be a viable alternative to optimize the grinding conditions and to replace the conventional nozzle commonly used in through-feed centerless grinding process. By the results obtained in this work, it can inferred that the use of the new nozzle concept is a viable way to reduce the MWF consumption during the centerless grinding process, guaranteeing the integrity of the ground part.

Declarations ETHICAL APROVAL
The authors declare that this manuscript was not submitted to more than one journal for simultaneous consideration. Also, the submitted work is original and not have been published elsewhere in any form or language CONSENT TO PARTICIPATE AND PUBLISH The authors declare that they participated in this paper willingly and the authors declare to consent to the publication of this paper.    The system is comprised of the multitubular nozzle, machine tool, grinding wheels and workpiece is shown in Figure 3a. The nozzle geometry was based on the one proposed by Webster [17], indicating that application of this nozzle generates a reduction in the cutting uid turbulence at the nozzle outlet, providing a jet of directional uid with low dispersion. In order to produce a good uid delivery, the nozzle was positioned at a distance of 80 mm from the cutting region. The positioning of the nozzle at the cutting interface is shown in Figure 3b. The ow rate of compressed air during the tests using the multitubular nozzle was 170 L / min, with a pressure of 0.6 MPa, generating a uid speed of 45 m / s (similar to the peripheral speed of the wheel, i.e., cutting speed). Multiple comparison test for ME+CA technique.

Figure 6
Wheel topography for (a) CN, (b) ME and (c) ME+CA at 10 L/min. GWAS condition for different lubri-cooling techniques tested.

Figure 8
Mean surface roughness (Ra) values for different lubri-cooling techniques (depth of cut of 0.03 mm).

Figure 9
Multiple comparison test for different lubri-cooling test at 40 L/min and depth of cut of 0.03 mm.

Figure 10
Page 23/24 Roundness deviation values for different lubri-cooling techniques (depth of cut of 0.10 mm).

Figure 11
Roundness deviation values for different lubri-cooling techniques (depth of cut of 0.03 mm). Figure 12