The Analysis of the In uence of Technological Parameters on the Grinding Temperature in the Single-Pass Grinding Process of Solid Carbide End Mill Flutes

A method of the measurement of temperature and grinding force components during fl ute grinding of solid carbide end mills was presented. The tests were performed using two diamond grinding wheels: with a resin bond (S1) and with a metal bond (S2). A diamond grinding wheel of 1A1 type and a rectangular cross-section was used. An ultrafi ne grained carbide with a grain size of 0.2 to 0.5 μm, characterized by high hardness (1730 HV30) and very high tensile strength (4600 MPa) was used. The fl utes were grinded in one pass, thus replacing rouging and fi nishing. The temperature of grinding was recorded using thermocouple type K, NiCr-Ni. The measurement of forces was conducted using a rotary piezoelectric dynamometer. Surface topography was recorded with the Alicona Infi niteFocus G4 microscope. The experiment was planned in Design-Expert 13 software. Mathematical models were developed, describing the relationships between the grinding speed and the feed rate as a function of the grinding temperature, grinding force and roughness parameters of the fl ute. For both grinding wheels, the Fn grinding force was recorded in the range of 28–110 N, the temperature in the range of 32–200 °C and surface roughness Sa in the range of 0.31–0.76 μm. The best grinding result, in terms of low grinding forces and temperatures was achieved for the grinding wheel with the metal bond (S2). For the resin grinding wheel (S1), for the selected technological parameters, grinding burn occurred.


INTRODUCTION
The main issue of the currently manufactured solid carbide end mills is the production time and the complexity of the tool design as discussed by Malkin [1], Hubert [2] and also Fujara [3]. The basic operation in the manufacturing of carbide tools is the fl ute grinding process, presented by Christoph [4] as well as Guachao [5]. It takes up to 70% of the production time, which was proved in works presented by Burek [6,7]. Machining of the fl ute takes place in many stages, and the roughing and fi nishing require from a few to a dozen grinding wheel passes, which significantly extends the grinding time as discussed by Yang [8]. The method of deep grinding, which is constantly the subject of research, consists of removing the entire machining allowance in a single pass of a grinding wheel, thus replacing the multi-stage roughing and fi nishing presented by Uhlmann [9] and also Badger [10]. Therefore, the optimization of technological parameters (v s , v f ) is of great importance which was proved in works presented by Bazan [11,12].
Among the many grinding parameters, in the monitoring systems used so far, the most common monitoring parameter has been the grinding force, because its variations very well correspond with the characteristics of the process which was also discussed in the work by Ren [13]. Grinding temperature is another parameter allowing to monitor the machining as discussed by Shatov [14] and Sui [15]. High temperatures, especially in single-pass grinding can lead to damage of the grinded end The Analysis of the In uence of Technological Parameters on the Grinding Temperature in the Single-Pass Grinding Process of Solid Carbide End Mill Flutes Marcin Sałata 1 mills as well as diamond grinding wheels, which was proved in works presented by Ren [16] and Yang [17]. In addition, due to thermal expansion, the dimension of the workpiece may change, which may result in shape and dimensional errors presented by Bazan [18]. Thus, thermal analysis of the grinding process is so important, which was proved in works presented by Grochalski [19].
Ren et al. [20] describes a method (an algorithm) of five-axis flute grinding using standard 1V1/1A1 grinding wheels. Adopting the algorithm allows for precise calculation of the grinding wheel's position in terms of the three most important parameters defining the flute (rake angle, core radius and flute width). The experimental tests described in the following paper adopted a standard algorithm of calculating the flute geometry available in the CNC machine tool control. Abdullah et al. [21] in their paper describe the diamond grinding wheel wear process during single-pass deep grinding, also called creep-feed grinding. Analyzing the results of their work, one can observe that technological parameters influence wear of the diamond grinding wheels. In the following paper grinding wheel wear was considered negligible, due to the short length of the grinded flute. Habrat et al. [22] in their work describe the influence of diamond wheel's bond on the grinding force components during carbide grinding. The obtained results indicate that the grinding speed, feed rate and grinding depth influence tangential and normal grinding force components. Grochalski et al. [23] in their work present two methods of temperature measurement: with the use of a thermal imaging camera and with a thermocouple. For the most resistant to measuring problems method authors indicate measurement with a thermocouple. Such a method was implemented in the following paper.
Regardless of the direction of the studies, researches emphasize the fact that machining of carbide, which is one of the difficult-to-cut materials, remains a constant challenge in terms of machining efficiency. It is especially important when considering new carbide types, such as ultrafine grade carbide or new diamond wheels with hybrid bond. It should also be noted that in solid end mills, flute is a critical part that determines parameters such as rake angle, core radius and flute width. These three parameters significantly influence grinding force, temperature, chip removal efficiency and tool life. Thus, ensuring their accuracy during machining and their continuous monitoring is of utmost importance.
The paper presents the attempt of using the K type thermocouple (NiCr-Ni) to measure the temperature of solid carbide end mill flutes during single-pass grinding with diamond grinding wheels.

EXPERIMENTAL RESEARCHES
Equipped with ISOG CNC FORTIS grinding machine test stand, researches of grinding process of flute in solid carbide end mill were performed (Fig. 1). The machine is a 5-axis grinding centre with three linear axes (X, Y, Z) and two rotary axes (A, C), spindle power 6 kW, spindle speed from 1000 to 8000 rpm. During the measurement of grinding forces KISTLER rotary dynamometer type 9123 was used. It is multi-channel, piezoelectric, adapted to measure three grinding forces components as well as torque (M c ) in two measurement ranges. To make the measurement even more accurate and to minimalize measuring noise, the accuracy of measurement increased through measuring in second range which amounts to F x = 0-450 N, F y = 0-450 N, F z = 0-1800 N.
In the measuring scheme measured values of cutting force components are converted in dynamometer's 12-bit A/C converter into digital value. Next they are transmitted wirelessly to nonrotating parts of the dynamometer. Then the signal is sent to amplifier, type 5223B. In the amplifier the signal is again changed into analog. Measurement frequency bandwidth of the device is in the range from 0 to 1 kHz. These signals are then transmitted to the National Instruments A/C NI USB-6009 converter (14 bit ,+/-10 V) working under control of the LabVIEW SignalExpress software, which registers test signals. It was registered with 5 kHz rate and 250 samples, which enables to achieve proper resolution allowing to measure grinding forces components [24].
Processing of the measurement data together with the calculation of 3D parameters of the geometric structure of the surface was carried out in SPIP 6.4.2 software. Data processing included removal of unusually high peaks and deep cavities resulting from coarse measurement errors and filtering of the shape profile with a polynomial of the first degree. 2D and 3D images of the topography maps of the original profile of the measured surfaces were produced [25]. 3D parameters such as Sa, Sz were calculated. An x50 objective was used for the study. More measurement conditions are shown in Table 1.
The Alicona InfiniteFocus microscope uses a measurement method called Focus Variation (FV). FV is a surface-based method in which multiple points are measured during a single scan. FV microscopes use the image sharpness information of the measured surface to determine its height as a function of the z position (x,y), which is made possible by combining the small depth of field of the optical system with vertical scanning. The operating principle is that the optical system is moved vertically along the optical axis to the extent that all points of the surface to be measured are in focus. At certain  positions within the given range, white light is emitted, which passes through the optical system and is focused on the measured surface. All rays reflected from the surface that returns to the objective lens end up on the detector. Due to the small depth of field of the optical system, only a small area of the measured surface is in focus for each position, allowing the vertical position of the optical system to be associated with high information from the measured surface. For each measured point, the actual color information is obtained, allowing photos of the surfaces to be taken.

MEASUREMENT OF GRINDING TEMPERATURE
In the article the method of temperature measurement during flute grinding was proposed. The method is easy to apply, fast as well as it offers the ability to measure and acquire the results online without stopping the machining. It is a contact method which uses a calibrated thermocouple. Its usage is connected with many advantages comparing to different kinds of temperature sensors. For example: it doesn't require external power supply, small size, low heat capacity, low response time, simple design as well as high reliability. Thermocouple is an element of electric circuit consisting of two different metal made wires. The place of their connection is called a measuring joint, remaining ends are called "cold ends". In that thermocouple electromotive force occurs in case when measuring joint and cold ends are kept in different temperatures (Fig. 3). The thermocouple uses the Seebeck effect, which was discovered in 1821 by German physicist Thomas Johann Seebeck. The reference point of a thermocouple is located in a converter. Temperature of surrounding indicated there is registered in constant way and included in the calculations.
The temperature of grinding was recorded using thermocouple type K, NiCr-Ni class I coupled with National Instruments A/C NI 9211 converter working under supervision of LabVIEW SignalExpress software which registers test signal on the computer.
One of disadvantages of using a thermocouple is that output voltage equals a few microvolts per degree Celsius. That fact forces to apply A/C card with much higher measure resolution than the one that has been used to measure grinding force component. The input range affects the resolution of the USB-6009 device (for measurement of cutting forces) for an AI channel. For example, a 14-bit ADC converts analog inputs into one of 16 384 (= 2 14 ) codes-that is, one of 65 536 possible digital values [26]. So, for an input range of -10 V to 10 V, the voltage of each code of a 14-bit ADC is: Due to too low card resolution, Ni 9211, (24 bit +/-80 mV) measure card was used. So, for an input range of -80 mV to 80 mV, the voltage of each code of a 24-bit ADC is: Card resolution of approximately 0.009 μV was obtained, which is enough to measure the temperature where the thermocouple sensitivity equals 40 μV/°C [27].
Inserting measuring probes as close as possible to the machining zone was a problem during contact measurement. For this purpose a hole was made in the blank's axis and consequently the thermocouple was put in there. The hole was made in the process of EDM (electrical discharge machining) on the AccuteX AH-35ZA device using copper electrode of a 1.4 mm diameter (Fig. 4).
For the purpose of ensuring good thermal conductivity -space between the material and the thermocouple was filled with some thermal compound [19]. This procedure prevented the coolant getting into the eroded hole. In order to ensure repeatability of experimental tests, the eroded holes were subjected to the measurements of the diameter, out-of-roundness and concentricity. The eroded holes with the highest deviations were rejected. The hole and its setting is shown in Figure 5. In Table 2 the accuracy of eroded hole is shown.

MATERIALS AND METHODS
For single -pass grinding process of fl ute two diamond wheels were used: • S1: 1A1 D100 U10 X10 H20 D64 C115resin bond The diff erence between those diamond wheels is their bond. These are: • 1A1 wheels with a rectangular cross -section, • nominal diameter of 100 mm, 10 mm width, • with an average size of a diamond grain of 64 μm in diameter, • with C115 grain concentration which is 28.75% of the diamond volume in the entire abrasive layer.
Carbide type TSF44, 10.2 mm in diameter was grinded during the tests. Its features are shown in Table 3. This type of carbide is classifi ed as ultrafi ne grade and is characterized by the highest transverse rupture strength available on the market (4600 MPa), with maintaining high hardness (1730 HV30). Ultrafi ne carbide grade for HSC   Flood cooling was applied with the usage of "Sin-toGrind" oil. For simplicity straight flutes were cut as shown in Figure 6.
The experiment (DoE) was planned in Design-Expert 13 software. One of the methods used in designing of experiments is the surface response method (RSM) -multicomponent research designs are based on it [28]. For the researches central composite face centered design (CCF) was applied for two dependent variables assuming levels of values (3 2 ) where α = 1. Graphic interpretation of this plan was shown in Figure 7 [29,30].  This experiment is characterized by the fact, that each factor has only three levels. It was used due to the fact that region of interest and region of operability are almost the same. It is characterized by good design properties for designs up to 5 factors, little collinearity, cuboidal rather than rotatable, insensitive to outliers and missing data. It is not recommended for six or more factors due to high collinearity in squared terms [31]. A modified polynomial models was selected for fitting, containing only statistically significant elements. The significance of the influence of individual input parameters and their interactions was determined based on the ANOVA analysis of variance. Statistical significance of the effect was demonstrated for an assumed confidence level of 95%.
Models obtained by this method can be the basis for optimizing process inputs due to the adopted objective function [32]. The experiment was carried out according to appropriate sets of process input parameters values generated using the Design-Expert software (Table 4).

RESULTS AND ANALYSIS
According to the assumed research parameters, the performance indicators were calculated as follows: • grinding efficiency: where: A -is the cross-sectional area of the flute, mm 2 , v f -is the feed rate, mm/min.
where: b s = a p -is the active grinding wheel width, mm.
After the tests had been carried out, the values of F n grinding force, temperature T and surface roughness Sa (arithmetical mean height of a surface), Sz (the greatest height of the surface) were analyzed. For the fitting, a modified square model was selected. It contained only statistically significant elements (alpha < 0.1). The goodness of fit for the obtained models was determined based on: • standard deviation of the residua component s, • mean value y, • coefficient of variation C.V., % • determination coefficients R 2 , • adjusted determination coefficients R 2 , • predicted determination coefficients R 2 .

Results and analysis for S1 -resin bond
Dependency between technological parameters for wheel with resin bond S1 as a function of temperature T and force F n was shown in Figure 8.
Equations of mathematical models describing the changes in temperature T are presented below (Exp. 5) as well as the component Fn of the grinding force (Exp. 6) as a function of technological parameters. The statistical parameters for the model (5), (6) are presented in Table 5.
The F n normal component of the grinding force was analyzed and it was noticed that depending on the experiment, the force F n varies from 37 N to 110 N and the range of the value for all tests is as high as 73 N. The highest value of F n was registered for the highest feed rate v f . The cutting speed v s increased with decreasing of the force Fn. From the obtained mathematical model -feed rate v f and cutting speed v s have effect on the increase of F n force. According to the conducted analysis the feed rate vf exhibits the greatest influence. Habrat et al. [22] registered the influence of technological parameters on the grinding force Fn as well. The recorded force was in the range of 73-245 N.
While analyzing the results it was noted that the T grinding temperature varies from 200 to 68 °C depending on the experiment, and the range of the values is equal to 132 °C. The highest T temperature was registered for number 3 and 7 grinding sample where grinding burns occurred. According to the conducted analysis that only feed rate vf influences the rise of grinding temperature. Yang et al. [17] recorded temperature during grinding using a thermocouple as well. Grochalski et al. [23] recorded temperature using a thermocouple, but in the turning process.
Dependency between technological parameters for wheel with resin bond S1 as a function of surface roughness Sa, Sz is shown in Figure 9.
The dependencies between roughness parameters Sa and Sz, and the technological parameters for the wheel with resin bond S1 are presented below: The statistical parameters for the model (7), (8) are presented in Table 6.
Analyzing the parameters of flute surface roughness Sa, it was discovered that depending on the experiment, it varied between 0.309 to 0.49 μm, and the range of the values for all the   samples was as low as 0.191 μm. According to the conducted analysis feed rate v f and cutting speed v s influence the surface roughness parameters. It was determined that technological parameters influence in statistically significant way the roughness parameters Sa and Sz, although the influence is low judging from the value's range.

Results and analysis for S2 -metal bond
The dependency between technological parameters for wheel with metal bond S2 as a function of temperature T and force F n is shown in Figure 10.
The equations of mathematical models describing the changes of temperature T are presented below (Exp. 9) as well as the component F n of the grinding force (Exp. 10) as a function of technological parameters.
The statistical parameters for the model (9), (10) are presented in Table 7.
The F n normal component of the grinding force was analyzed and it was noticed that depending on the experiment, the force varies from 28 to 90 N, and the range of the values for all tests was equal to 62 N. The highest force Fn was registered for  the highest feed rate v f and with increasing cutting speed v s , a decrease of the force F n was observed. According to the conducted analysis, feed rate v f and cutting speed v s influence the force F n , but feed rate v f shows the most significant influence. Habrat et al. [22] registered the influence of technological parameters on the grinding force Fn as well. The recorded force was in the range of 73-245 N. While analyzing the results, it was noted that the grinding temperature T varied from 105 to 32 °C depending on the experiment and the range of the values was equal to 73 °C. The highest T temperature was registered for grinding sample number 1 and 7, where the highest value of speed rate v f occurred. According to the conducted analysis, feed rate v f and cutting speed v s influence the rise of temperature T, but feed rate v f shows the most significant influence. Yang et al. [17] recorded temperature during grinding using a thermocouple as well. Grochalski et al. [23] recorded temperature using a thermocouple, but in the turning process.
On the basis of the results it was not possible to develop mathematical models of Sa and Sz as a function of technological parameters. It follows from the fact that the range of the values for individual tests is only 0.140 μm. Analyzing the parameter values of flute surface roughness Sa, it was discovered that depending on the experiment, it changes between 0.624 to 0.764 μm. Abdullah et al. [21] registered slight influence of feed rate v f and grinding speed v s on the surface roughness as well. The recorded surface roughness was in the range of 0.25-0.4 μm.

Results and analysis for booth wheel (S1, S2) -summary
While examining the grinding temperature T and the two wheels with two different bonds (S1, S2), it can be noticed that wheel with metal bond (S2) generates lower grinding temperature T in comparison to wheel with resin bond (S1), which is shown in Figure 11. Average value of temperature for all the grinding passes (DoE) equals to 132 °C for the S1 wheel and to 63 °C for the S2 wheel. Usage of the wheel S2 lowers average value of flute temperature in single pass by 53%.
Analyzing the grinding force F n and the two wheels with two different bonds (S1, S2), it can be noticed that wheel with metal bond (S2) generates lower grinding force F n in comparison with the wheel with resin bond (S1), which is shown in Figure 12. Average value of the force F n for all the passes (DoE) for the wheel S1 equals to 66 N, and for the wheel to S2-52 N. Usage of the wheel S2 lowers average value of flute temperature in single pass grinding by 21%.
Analyzing roughness parameters Sa and Sz and the two wheels with two different bonds (S1, S2), it can be noticed that the wheel with resin bond (S1) generated lower values of roughness parameters in comparison with the wheel with metal bond (S2) - Figure 13, 14. The average value of the Sa roughness parameter for all grinding passes (DoE) for the S1 grinding wheel was equal to 0.401 μm and for the grinding wheel S2 to 0.688 μm. The use of the S1 grinding wheel reduced the value of the Sa roughness of the flute on average by 41%. Fig. 11. The results of the temperature T measurement for both wheels   The average value of the Sz roughness parameter for all the passes (DoE) for the S1 wheel was equal to 3.20 μm and for the S2 grinding wheel to 4.42 μm. The use of the S1 grinding wheel reduced the value of the Sz roughness of the flute on average by 27%.

CONCLUSIONS
Thanks to the designed measuring stand it was possible to register not only grinding force F n but also grinding temperature T which is a significant aspect. Measurement of grinding force normal component F n is also crucial, as value of the force influences the elastic displacements of the tool. High values of the elastic displacements lead to dimensional errors of grinded end mills.
As the experiments has shown, the measurement of temperature in the grinding process indicates the occurrence of grinding burns. It is a key information concerning correct single pass flute grinding process of a TSF44 carbide.
Grinding burns were observed during the experiment, although they were classified as "light ones"-which enabled further experiments. Outside of the assumed design of experiment, for parameters v s = 25 m/s, v f =30 mm/min, a e = 3.5 mm a p = 4.77 mm "strong" grinding burns were observed, which did not allow to continue further experiments. Measured temperature value was equal to 485 °C.
The grinding efficiency Q w in carried out researches varied from 100 to 301 mm 3 /mm. For the S1 wheel, for grinding pass number 3 and 7 (in which the grinding efficiency was the highest Q w = 301) with the feed rate v f = 30 mm/min, grinding burns were observed. For grinding pass number 1, where feed rate v f was also 30 mm/min the grinding burn was not observed, what results from using the smallest grinding speed v f = 15 m/s. It should be mentioned that the flute in solid carbide end mills is responsible for the evacuation of the chip from the machining zone. Therefore, the roughness of the flute is the key to proper chip evacuation. In addition to the roughness parameters, an important factor in assessing the quality of the flute surface is the presence of defects in the form of scratches or inclusions. No surface heterogeneity or irregular inclusions were observed in any grinding test.
The developed mathematical models for different technological parameters (v f , v s ) as a function of T, F n , Sa, Sz can help to optimize the grinding process. The correlation between the temperature T and the grinding parameters (v f , v s ) allows to determine the non-recommended parameters, which allows to avoid grinding burns, thus extending the grinding wheel time. The correlation between the grinding force F n and the grinding parameters (v f , v s ) allows to reject grinding parameters, for which the dimensional errors of grinded end mills would be the highest (F n = max).