SciELO - Scientific Electronic Library Online

 
vol.34 issue1Laser beam welding tempered 300M ultrahigh mechanical strength steelModeling and optimization of cylindrical grinding of Al/SiC composites using genetic algorithms author indexsubject indexarticles search
Home Pagealphabetic serial listing  

Services on Demand

Article

Indicators

Related links

  • Have no similar articlesSimilars in SciELO

Share


Journal of the Brazilian Society of Mechanical Sciences and Engineering

Print version ISSN 1678-5878

J. Braz. Soc. Mech. Sci. & Eng. vol.34 no.1 Rio de Janeiro Jan./Mar. 2012

http://dx.doi.org/10.1590/S1678-58782012000100004 

TECHNICAL PAPERS
MANUFACTURING PROCESS

 

Determination of the relative position between grinding wheel and a cylindrical workpiece on a 7 axis grinding machine by acoustic emission

 

 

Prof. Dr.-Ing. Walter Lindolfo WeingaertnerI; M.Eng. Adriano BoaronII

IUniversidade Federal de Santa Catarina, Department of Mechanical Engineering. 88040-900 Florianópolis, SC, Brazil. wlw@emc.ufsc.br
IIUniversidade Federal de Santa Catarina, Department of Mechanical Engineering. 88040-900 Florianópolis, SC, Brazil. aboaron2002@yahoo.com.br

 

 


ABSTRACT

The contact between grinding wheel and workpiece in the grinding process is recognized by acoustic emission (AE). Two acoustic emission monitoring systems (MS) were integrated into a 3 axis CNC grinding machine. A laptop allows the signal acquisition and visualization. The acquired AERMS signals from the contact between tool and workpiece are analyzed permitting to establish the most suitable AE monitoring system to recognize the contact in a particular grinding machine. In a second experimental setup the selected MS was installed on a 7 axis tool grinding machine at an industrial partner. At this partner, the relative position between grinding wheel and workpiece was previously determined manually. This procedure has a direct influence on the results depending on the technical skills of the operator. The automation of this activity supported by acoustic emission has led to satisfactory results regarding the relative position between grinding wheel and workpiece and contributed to the setup time reduction.

Keywords: monitoring systems, acoustic emission, external cylindrical grinding process


 

 

Introduction

In the manufacturing industry, the grinding process is one of the most common processes required when great quality and close tolerances are desired. A frequent way to improve such needs consists in using monitoring systems (MS) based on the acoustic emission (AE). Besides the possibility to control the process characteristics, these systems also allow an accurate detection of the contact between grinding wheel and workpiece. The position associated with the contact usually represents a reference, serving as a starting point to the following grinding operations.

In the first stage of the present work, the contact recognition between grinding wheel and workpiece is evaluated by two AE MS integrated into the CNC of the machine tool. Each MS works separately with specific AE transducers. The observation of the AERMS signals is done in real time on the screen of a laptop. The acquired contact signals are recorded and sampled aiming at additional analysis. The contact between grinding wheel and workpiece generates a grinding mark on the workpiece's surface. After the contact experiments the depths of the marks were measured in order to use these values as input data in a Factorial Analysis. The Factorial Analysis leads to the determination of an optimized condition to the contact recognition with minimal metal removal.

In a second stage the most suitable MS was installed in a tool grinding machine at an automotive part deliverer. During the machine setup, an activity that demands considerably amount of time consists in determining the reference position of the grinding wheel in relation to the workpiece. Due to the design characteristics of the machine tool and the lack of instrumented support, the machine operator needs to use try-error manual procedures. These procedures often result in errors and exert a direct influence on the machined geometry and do not permit to achieve the tied required tolerances. The present work suggests an instrumented procedure, based on AE, to automatically find the reference position between grinding wheel and workpiece.

 

Nomenclature

AE

= acoustic emission

MS-A = monitoring system A
MS-B = monitoring system B
ZEROYAUTO = procedure to automatically centralize the grinding wheel in relation to the workpiece's axis, using an AE monitoring system
ae = increment on the workpiece surface (infeed direction), mm
ae,m = depth of the measured mark, µm
ae,SIGNAL = depth of the mark evaluated by analyzing the AERMS signal, µm
tA = approaching time in the AERMS signal, ms
tR = rising time in the AERMS signal, ms
vfr = infeed velocity (radial direction), mm/min
vs = cutting speed, m/s
vw = workpiece speed, m/s
DMIN = inner diameter of the grooved profile, mm
DMÁX = external diameter of the grooved profile, mm
YA = mean value of the measured marks obtained
with MS-A, µm
YB = mean value of the measured marks obtained with MS-B, µm
α = significance level during statistical analysis (0.05)
v = degrees of freedom during statistical analysis
λ = relative angular position between grinding
wheel and the axis of the workpiece, degrees

 

Acoustic Emission on Grinding

Acoustic emission (AE) is defined as the transient elastic wave generated by the rapid release of energy from a localized source or sources within a material when subjected to a state of stress. This energy release is associated with the abrupt redistribution of internal stresses, and as a result, a stress wave is propagated through the material (Ravindra et al., 1997).

The grinding process is characterized by the randomic contact of a large amount of cutting edges on the surface of the workpiece. All the individual contacts caused by the grits can be considered as a source of pulse deformation or stress on the workpiece. Figure 1 exemplifies the major AE sources that can be found in the grinding process (Karpuschewski, 2001).

 

 

Acoustic emission signals on grinding

The raw AE signal (AERAW) is composed of different high frequencies on different energy levels and is difficult to interpret. One of the most employed techniques to extract useful information from AERAW signals consists in using the root mean square value (RMS) of the AERAW signals (Hwang et al., 2000). The AERMS represents a physical dimension of the AERAW signal intensity and depends directly from the amount and dispersion of stress waves on the material (Meyen, 1991). According to Hwang et al. (2000), the AERMS signal is defined as:

where:

V = raw acoustic emission signal (AERAW), and
ΔT = integration time constant.

The AERMS (rectified value of AERAW signal) has been successfully used to monitor several grinding situations. However, the spectrum analysis can complement the interpretation in situations where the RMS technique cannot allow satisfactory results (Gomes, 2001).

Acoustic emission signals during the contact between grinding wheel and workpiece

The contact recognition between grinding wheel and the workpiece depends on the transducer, the amplifier and the signal conditioning. This leads to a time delay and the first physical contact of grits and workpiece may happen before any appreciable change in the signal, especially when employing AERMS signals. The contact is usually judged according to a significant change of the amplitude of the AERMS signal, or AERAW signal. Therefore, understanding the instantaneous features regarding wheel/workpiece interaction may help to define "contact" for performing efficient use of the AERMS signal (Dornfeld et al., 1995; Leme, 1999; Dornfeld and Oliveira, 2001). Theoretically, the cutting grit generates a burst type of AE signal when it cuts through the workpiece. When numerous grits cut through the workpiece in such a way the interval of two consecutive cuts (which are not necessarily in the same place) is much shorter than the decay time of each burst signal, then a continuous type of AE signal is formed (Webster et al., 1996). The continuous AE signals generated when many grains randomly touch the surface of the workpiece can be represented by diverse parameters, Fig. 2 (Asher, 1997).

 

 

Experimental Setup for Contact Recognition Experiments

The experimental setup used for the contact recognition experiments performed in a cylindrical CNC grinding machine (Zselics Pratika Flexa 600-L) is schematically represented in Fig. 3.

 

 

Two AE monitoring systems were used separately. The AE signals related to the event of contact were recognized by employing piezoelectric AE transducers with direct transmission. The AERAW signals from the transducers are transmitted to the MS through appropriate cables. When MS-A (Dittel, 2007) was used, the AERMS signals were sent directly to a laptop by a RS-232 interface and visualized on the laptop's screen after treated by a specific software (Dittel, 2007). When MS-B was used (Sensis, 2002), the AERMS signals assigned to its analog output were sent to a multi-analyzer system (Oros, 2006) and to a laptop and the results are presented on the screen. All data were stored for a further analysis.

Both MS carry out the signals treatment in order to convert the AERAW signal into AERMS signal. The signal conditioning chain for the MS-A includes many stages: amplification, band-pass filtering, rectifying, and low-pass filtering at the end. MS-A uses a specific software to digitalize the AERAW signal up to 1000 Samples/s, based on the highest cut-off frequency in the conditioning chain, and then avoiding aliasing errors. MS-B has a similar signal conditioning chain permitting to sample the AERAW signals at 2048 Samples/s situations where the RMS technique cannot allow satisfactory with the aid of a particular analyzer (Oros, 2006), and then, avoidingresults (Gomes, 2001). aliasing errors.

Both MS were connected to the CNC of the grinding machine by means of a DB-15 connector installed into the CNC of the machine. As the AERMS signal from the contact exceeds a static threshold (previously adjusted by the user), a electric signal is delivered to a specific input in the CNC, which acts on the stoppage and reverses the grinding wheel's infeed motion (Boaron, 2009).

 

Experimental Procedure for Contact Recognition Experiments

The experiments were carried out in plunge grinding of ABNT 1040 steel specimens with a CBN grinding wheel (406 mm diameter). During the contact recognition experiments the specimen was kept static (vw = 0 m/s) without cutting fluid. The cutting speed of the grinding wheel was maintained constant at vs = 22.5 m/s.

Before starting the experiments, the grinding wheel was dressed, the AE transducers installed, and both MS were adjusted. The adjustments of the MS were firstly realized by setting the available filters in order to avoid the background noise influence in the AE contact recognition signals. The parameters related to the static threshold, gain, and RMS time constant play an important role in the contact recognition procedure. The parameters selection has been based on the binary technique (Dornfeld and Oliveira, 2001), i.e, threshold and RMS time constant should be as small as possible, and gain and noise reduction parameters, as high as possible.

Due to the fact that the amount of material removed on each contact experiment is very small and the specific removal rate Q'w is also very small, the wear of the CBN grinding wheel can be disconsidered. The specimen was fixed between the tailstock and the headstock and positioned orthogonally to the infeed direction, as illustrated in Fig. 4(a).

 

 

At the beginning of each experiment the grinding wheel is positioned 250 mm from the specimen (X+ = 250). Figure 4(b) shows the stages of movement described by the grinding wheel during the experiments. The grinding wheel plunges with the infeed velocity vfr1 = 600 mm/min to a position at 0.5 mm away from the specimen (point 1 in Fig. 4(b)). Thereafter, the infeed velocity is dropped to the infeed velocity vfr2, until the contact is recognized by the AE monitoring system (point 2 in Fig. 4(b)). At this point the infeed motion is automatically stopped and reversed by the CNC of the grinding machine. The relative displacement described by the grinding wheel from point 2 to point 3 defines the depth of the mark ae on the surface of the specimen. This depth is due to the time delay in processing the signals and informing the CNC, as well as the time delay of the CNC to stop and reverse the infeed movement of the grinding wheel.

The contact between grinding wheel and the specimen is featured by a physical mark on the specimen's surface, which results from the material removal from the specimen during the time between the first contact of a grit and the workpiece until the complete stoppage of the infeed motion and relaxation of all elastic deformations of the system after the reversion of the infeed motion. The level of the AE signals depends on factors like the infeed velocity of the grinding wheel vfr, the integration time constant ΔT, the transducers (magnetic or threaded fixture) and the AE monitoring system. These factors have been varied to produce contact recognition marks on the specimen. The two AE monitoring systems have been used separately.

 

Structure of the Contact Recognition Experiments

The experiments were conducted based on a Factorial Analysis involving the 3 major factors which influence on the first contact AERMS signals. Among these factors were considered the infeed velocity vfr2, the integration time constant ΔT, and the type of the employed transducers in the experiments. These 3 factors were varied in 2 levels (high ↑, and low ↓) whose magnitudes were previously defined, Table 1.

 

 

Line 1A illustrates the experimental situation in which the factors "Integration time constant" (ΔT), "transducer", and "infeed" are set at the lower levels (10 ms, magnetic sensor, 3 mm/min, respectively). On the used abbreviation, the number 1 means the first combination between factors and levels, whereas letter "A" means the MS-A was used (instead of MS-B). Each experimental situation was repeated 6 times, leading to a total of 48 experiments for each MS. When using the MS-B the same methodology was implemented. The only difference consisted in the higher value (↑) for the factor "Integration time constant" (ΔT), which assumed the value 400 ms.

The depths of the marks generated during the contact, measured in a precision metrology device (Mahr MMQ40 Formtester), were used as input data on a Factorial Analysis which permitted an optimized use of both AE monitoring systems in recognizing the first contact and to verify the effectiveness of the AE monitoring systems.

 

Contact Recognition Results

Through the realization of the Factorial Analysis, an optimizing condition for the contact recognition was achieved for each MS. This condition takes into account all the combinations between the 3 factors involved and their respective levels of variation. The input values for this analysis were the values of the depths of the measured marks (ae,m). The optimized condition has been characterized by the specific combination of factors and levels that present the mark with the smallest value of depth. Figure 5 shows the analyzed results for MS-A and MS-B. YA means the average value of the depths of the marks when using MS-A, whereas YB is the average value of the depths of the marks when using MS-B. For both MS the optimized condition is represented by a small vfr2, the transducer with the magnetic base, and low integration time. The constant values that appear at the beginning of both equations represent the mean values of ae,m along the 48 runs for each MS. Additionally, the coefficients refer to the statistical effect of the analyzed factors on the mean values of ae,m.

 

 

Based on these results, the values of the infeed velocity vfr2 were gradually reduced for each MS in order to carry out a comparative study regarding the efficiency in recognizing the first contact. The contact signals were recorded and post-analyzed and showed to decrease when reducing the infeed velocities, as it is known from the literature (König et al., 1994; Dornfeld et al., 1995; Klocke, 2009). The obtained marks in this experiment were also measured by the same way as done before, Table 2.

 

 

Despite the lower values observed in the majority of the situations when using MS-B (except for vfr2 = 0.1 mm/min), it was not possible to affirm that this system would have a better efficiency than MS-A only by a simple comparison of these values. Therefore, an additional study was realized to compare both MS. This study was based on a Statistical Hypothesis Testing which considered the difference in the means of the obtained depths by using the optimized situation determined earlier (See Eq. (3) and Eq. (4)). This experiment starts with two initial hypotheses (H0 and H1). The hypothesis H0 considers that thedifference in the means is zero,(that is,H0:µAB=0) and the hypothesis H1 considers that the difference in means verified during the experiments should represent a better efficiency by the MS-B (that is, H1: µAB > 0) conducting to small values of the marks on the specimen after the contact recognition. During the evaluations, a significance level of α = 0.05 was used. Figure 6 shows the major statistical parameters which have been determined to achieve the conclusion about the available efficiency for both MS.

 

 

According to Montgomery (2001), as T0 > t0,05;12 then the hypothesis H0 (H0: µAB = 0) must be rejected and the hypothesis H1 can be accepted. Based on these results, it is possible to conclude that the observed difference on the mean values ( and ) is representative in terms of a statistical sense. Then it can be affirmed that the MS-B has presented a better efficiency in recognizing the first contact when using the optimized condition predicted by the model.

 

Analysis of the AERMS Signals from the Event of Contact

After determining the optimized conditions for both MS the experiments were conducted in order to predict the depths of the marks by employing the AERMS signal as a reference. For each MS, six repetitions were executed and their AERMS signals have been recorded. The contact recognition signals were described and analyzed based on the parameters shown previously (see Fig. 2). By assuming that: a) the stoppage of the grinding wheel motion is associated with the higher value of the AERMS signal; b) during the time characterized by the parameter tA (approaching time) the infeed velocity vfr2 is equal to the programmed velocity; c) the infeed velocity vfr2 is uniformly decelerated during the time represented by the parameter tR, which is associated to the first overstepping of the static threshold; d) the displacement of the grinding wheel along the decelerated movement numerically corresponds to the area in the graphic vfr2 x t (uniformly decelerated motion). Regarding the AERMS signals connected to the experiment 1A1, the evaluation of the depth of the mark corresponds to the time in which the AERMS signal is recognized. The depth of the mark was evaluated by the following manner:

Approaching time, tA: tA,1A1 = 4 ms
Rising time, tR: tR,1A1 = 228 ms
Infeed velocity, vfr2: vfr2 = 3 mm/min = 50 µm/s

then

(ae,SIGNAL)1A1 = 5.95 µm

The same procedure has been used to the evaluations involving the other signals (1A2 to 1A6 and 1B1 to 1B6). Figure 7 displays the depth of the marks obtained after measuring (ae,m) as well as the depths evaluated through the analysis of the AERMS signal, (ae,SIGNAL). Observing the obtained values it is possible to conclude that the values related to ae,SIGNAL were considerably higher than those obtained by ae,m, for every experimental conditions for both MS. This information makes sense as the AE contact signal in this process happens much earlier than any notable material removal. At the beginning of the contact, the elastic strains related to the system (grinding wheel, workpiece, machine tool) increased until plastic strains initiate to dominate. The initial plastic strains are not enough to cause any material removal. As the plastic strain reaches a specific level, the removal process begins to occur, being characterized by chip formation. After the infeed stoppage, the contact between grinding wheel and workpiece still occurs until the moment in which every elastic strains are attenuated (König, 1989; Dornfeld and Lee, 2008; Klocke, 2009; Boaron, 2009).

 

 

Experimental Setup for the Relative Position Experiments

Despite the better contact recognition accuracy presented by MS-B (Fig. 7), the MS-A was used in an industrial application due to its flexibility and easy-to-use characteristics. In addition, cost factors have also been decisive as MS-B needs an auxiliary analyzer in order to digitalize the AERAW signals and to allow a signal analysis. The designed setup for experiments was implemented in a cylindrical CNC tool grinding machine for broaches (Stauffer/Zen), Fig. 8.

 

 

The axis x, y, z and a (rotation of the workpiece) are CNC controlled. The rotation of the grinding wheel (b) and the additional rotational axis b1 and c (rotation and tilting of the wheelhead) are manually operated with indication of the angular position on the CNC's screen. The grinding speed and feed rates are controlled by the CNC program. The grinding wheel has a diameter of 100 mm. The maximal workpiece length is about 1000 mm. As a manner to allow the implementation of the automatic contact recognition between grinding wheel and workpiece, the MS-A is integrated into the CNC.

The AE transducer was screwed on the tailstock. This position showed the lowest interference from the moving components on the machine and a good signal from the process. The outputs from MS-A were delivered to the CNC by means of pin-6 (connector DB-25) of the MS-A. This pin is associated with the digital output from the MS-A and delivers a voltage signal to the CNC input every time the AERMS signals (from the contact) exceed the static threshold previously adjusted by the user. The AERMS signals were directly sent to a laptop with a specific MS-A software through a RS-232 interface and could be visualized on the laptop's screen. This software digitalizes the AERAW signal using a sampling rate up to 1000 Samples/s avoiding aliasing problems. In parallel, the coordinates associated with the spark in and spark out signals are stored in the CNC.

 

Experimental Procedure to Determine the Relative Position

The main goal of the experiments consisted in implementing a procedure to automatically determine the centralized cross position of the grinding wheel in relation to the specimen. All the experiments have been executed without cutting fluid. The centralized cross position between grinding wheel and specimen is associated with the Y-axis of the grinding machine so that the procedure receives the name ZEROYAUTO. The term "ZERO" refers to the centralized position, whereas the term "AUTO" means the automatic use of the AERMS signals from the spark in and spark out events. During the experiments to verify the ZEROYAUTO procedure the specimen was kept static (vw = 0 m/s). The contact between grinding wheel and the specimen behaves as described before. It is always desirable to achieve the smallest mark as possible to minimize the influence on the dimensional tolerances of the workpiece.

Even considering that the metal removal during the centering experiments is extremely small, before starting each group of experiments to verify the ZEROYAUTO procedure, the grinding wheel was dressed with a symmetric profile and the MS-A parameters were adjusted based on the binary technique (Oliveira and Dornfeld, 2001). At the beginning of each experiment the grinding wheel was positioned in a secure distance above the surface of the specimen (Z+= 2 mm), close to the centralized position of the grinding wheel in respect to the specimen's axis, (Fig. 9, position "b" at left). The grinding wheel is then ordered to move toward the workpiece (vfrz = 10 mm/min) on the Z-axis until the contact with the specimen is recognized by the MS-A and the infeed motion is stopped (Fig. 9, position "a" at left). The recognized contact position is stored in the CNC for further use. The grinding wheel returns to the safe position "b" and moves along the Y-axis for about 10 mm and more 10 mm on the X-axis (Fig. 9, position "c" and "d" respectively). The grinding wheel is ordered to move in Z-axis down to the "z" coordinate and incremented 0.01 mm (ae1) in relation to the reference position recognized earlier on point "b", then reaching point "e". The grinding wheel moves along the Y-axis crossing the workpiece completely until point "f". During this trajectory the grinding wheel touches the workpiece. This contact is recognized by the MS-A and is represented by the smaller mark on the specimen's surface. The AE signals in this first interaction have shown to be not adequate for a centering procedure. The position of the grinding wheel is incremented for up to 0.01 mm (ae2) along the Zaxis, position "g", and then returned to the position "h" on the back side of the specimen. During this movement the coordinates associated with the spark in (Y1) and the spark out (Y2) positions are stored into the CNC and serve as reference positions for centering the grinding wheel in respect to the specimen. The grinding wheel is lifted to position "i", moved to "j" and "k" centered over the workpiece axis and plunged into the specimen until the contact is recognized, Fig. 9 (position "l"). The coordinates of this contact position are also stored (Boaron, 2009).

 

 

Structure of the Relative Position Experiments

The structure of the experiments is divided into two stages (stage-a and stage-b). The stage-a experiments aimed to determine the appropriate conditions in recognizing automatically the centralized position between grinding wheel and specimen by analyzing the major influencing factors on the AERMS signals and thence on ZEROYAUTO procedure. During this stage, the values of the centralized position by using the ZEROYAUTO procedure were compared to the mean value achieved when using the manual procedure. Among the factors that significantly influence the ZEROYAUTO procedure are the cutting speed vs, the depth of cut ae2, the traverse infeed along the Y-axis vfry, and the value of the integration constant time ΔT (selected on the MS-A). Moreover, the relative angular position between grinding wheel and specimen λ has also shown an evident influence on the AERMS signals and, consequently, on the centering values by using the ZEROYAUTO procedure. This angle was chosen equal to λ = 18º and λ = -60º, representing the angular limits of the helical angles of the broaching tools to be manufactured. Figure 10 shows a schematic top view of the tool grinding machine's working chamber and the angular positions used during the experiments.

 

 

The 4 mentioned factors were varied each at 2 levels following the scope of a Factorial Analysis. The combination of the 4 factors and their respective levels of variation have led to a total of 16 experiments, Table 3.

 

 

For each of the 16 experiments, 3 repetitions (R1, R2, R3) have been done in order to achieve a representative mean value and a standard deviation of the centered position, using the ZEROYAUTO procedure. The mean value for each experiment was compared to the mean value of the manual procedure.

The levels of variation have been determined by observing the boundary limits to be used without damaging the machine. These values also corresponded to those normally used during the daily jobs on the machine. The level of variation connected to the factor ΔT was selected in such a way that the ZEROYAUTO procedure could be implemented. The analysis of the 4 factors has been carried out for the critical angular positioning λ = -60º. The results for this position were compared to the results for λ = 18º. The best results obtained automatically were close to the mean value obtained manually. Table 3 detaches the combination of factors presenting the best achieved results. After conducting the 48 experiments regarding the stage-a experiments, it was possible to verify the experimental conditions which conducted to the nearest mean values between both procedures (ZEROYAUTO and the manual procedure). Among all the 16 experimental conditions just 4 have shown to be useful. The best results were achieved when using the experimental conditions "ab", "abc", "abd", and "abcd" which are detached in Table 3. By employing the condition "ab" the angular position of the grinding wheel was then modified to λ = 18º and the ZEROYAUTO procedure was verified again. Table 4 shows the obtained results after 6 repetitions using this condition:

 

 

Despite the difference in about 0.15 mm observed in the mean value founded with ZEROYAUTO procedure in comparison to the mean value obtained through the manual procedure (ZEROYMANUAL), the results have shown this experimental condition could be considered as possible to be used on both angular positions of the grinding wheel (λ = 18º and λ = -60º) leading to close values between both procedures. Meanwhile, to prove the real efficiency in finding the centralized position with ZEROYAUTO procedure, it was necessary to analyze the results obtained along the second phase of the experiments, stage-b.

The stage-b experiments consisted in comparing the efficiency of the ZEROYAUTO procedure and the manual procedure in achieving a centralized symmetric groove on a specimen for the angular position of λ = 18º. The comparison was made by measuring the ground groove on the specimen. The groove was measured on a coordinate measuring machine and referenced to the axis of the workpiece and the reference profile. In this procedure the reference profile is independent of the judgment of the grinding machine operator. The machined groove profile was scanned and exported to specific software allowing the visualization of the actual and the designed profile. The software also permitted to determine the distances between the measured and designed profile on desired positions.

 

Results and Analysis for the Relative Position Experiments

Along the stage-a experiments, all the AERMS signals originated during the interaction between grinding wheel and workpiece could be recorded and analyzed. Figure 11 shows a representative AERMS signal captured along the second traverse movement of the grinding wheel in relation to the specimen (displacement g→h).

 

 

It is possible to observe that the entrance slope is higher than the outgoing slope. This behavior is highlighted through the auxiliary dashed lines in the figure. The difference in the entrance and outgoing slopes is due to the metal removal at the beginning and at the end of the contact between the grinding wheel and the specimen. During the traverse movement of the grinding wheel in respect to the specimen, the metal removal starts when the first corner of the grinding wheel gets in contact with the specimen and stops when the second corner of the grinding wheel loses contact with the specimen. During the time the grinding wheel is in contact with the specimen a nearly continuous AERMS signal is generated.

The obtained results during the stage-b experiments have been achieved with an angular positioning of λ = 18º and a grinding wheel presenting a symmetric involute profile. For a first approach, the grinding wheel was manually centered. The 4 best experimental conditions that have been encountered previously were verified with ZEROYAUTO procedure, in order to guarantee a reliable result. The combination "ab" (see Table 3) has led to the closest mean value to those obtained manually (54.84 with ZEROYAUTO procedure, against 54.83 obtained with manual procedure). By using these mean values of the centralized position, a groove was machined on the specimen for each centralized position. Thereafter, the position of the grooved profile was measured in relation to the axis of the specimen. To compare the efficiency of both procedures in reaching a centralized position between grinding wheel and specimen, it was necessary to evaluate the deviations from the designed profile. Figure 12 illustrates an example of the measured profile (continuous lines) and the overlapping of the designed profile (dashed lines).

 

 

The scanned profile (measured profile) shows to be extremely out of the desired tolerances. As the main goal of this study was related to the determination of a centralized position between the grinding wheel and specimen, the correction of the dressed profile has to be done in a second step, out of the scope of this research. It is shown that the achieved centralization by using both procedures presents a good result in terms of the relative position to the designed profile. Both procedures lead to machined grooves whose profiles appear to be adequately centralized in reference to the designed profile.

The symbol Δ corresponds to the gap (error) between the measured and the designed profile. The letters "R" and "L" represent the sides of the groove in which the measurement of the deviations was carried out ("right" and "left" side, respectively).

Figure 13 compares the measured values at the upper (U), middle (M) and bottom (B) positions of the ground and designed profiles by using both centralizing procedures. The deviation of the centralized position is also shown. The software that superposes the measured and the designed profiles considers the best fit at the middle position M. If there is a deviation at the top, it indicates that the centralization of the grinding wheel is not adequate. The centralization that has been done by the operator (manual procedure) shows a deviation of 0.01 mm at the upper position, which is close to the required tolerances limits and far worse than that achieved by ZEROYAUTO procedure.

 

 

Conclusions

Based on the results that have been presented, it was possible to verify that both AE monitoring systems are feasible to detect the first contact between grinding wheel and specimen. The obtained results from the Factorial Analysis show the average values of the marks when using the MS-A and MS-B were = 6.7 µm and = 8.6 µm, respectively (Eq. (2) and Eq. (3)). The final results achieved from the Statistical Hypothesis Testing have exhibited a better efficiency by MS-B in recognizing the contact on the 3 axis CNC grinding machine employed. This behavior was also verified when comparing the average values of the measured marks (ae,m) and the averages values of the marks obtained by analyzing the AERMS signals (ae,SIGNAL) from the contact events.

The proposed procedure (ZEROYAUTO) is feasible to be implemented in a practical sense, especially when analyzing the angular position of λ = 18º, which represents the smallest helical angle used for machining the broaching tools. As a first advantage, the ZEROYAUTO procedure has led to an insignificant deviation in relation to the designed profile (0.003 mm at the top measuring section) while the use of the manual procedure conducted to a higher deviation (0.01 mm) at the same measuring section. The second advantage that was noted in using the ZEROYAUTO procedure consists in the centering time of 30 s required to determine the centralized position between grinding wheel and the specimen. The production is habilitated to start after finding the centralized position. The manual procedure takes an average time of about 5 min to find a centralized position. For this procedure it was always necessary to control the position of the first ground grove in the metrology laboratory, demanding time (up to several hours) to start production.

 

Acknowledgements

We acknowledge CNPq/CAPES/FINEP/IEL and company ZEN S.A. for their support.

 

References

Asher, R.C., 1997, "Ultrasonic Sensors for Chemical and Process Plant", Institute of physic publishing Bristol and Philadelphia.         [ Links ]

Boaron, A., 2009, "Determinação do Posicionamento Relativo entre Rebolo e Peça com o Auxílio da Emissão Acústica" Dissertação de Mestrado, Laboratório de Mecânica de Precisão, UFSC, Santa Catarina, Brasil.         [ Links ]

Dornfeld, D.A. et al., 1995, "Investigation of Acoustic Emission for Use as a Wheel-to-Workpiece Proximity Sensor in a Fixed-Abrasive Grinding", American Society for Precision Engineering, Austin, TX.         [ Links ]

Dornfeld, D.A., Oliveira, JJF., 2001, "Application of AE Contact Sensing in Reliable Grinding Monitoring" CIRP Annals - Manufacturing Technology.         [ Links ]

Gomes, J.J.F., 2001, "Identificação em Processo de Mecanismos de Desgastes em Rebolos", Tese (Doutorado em Engenharia Mecânica), Escola de Engenharia de São Carlos, São Carlos.         [ Links ]

Hwang, T.W., Whitenton, E.P., Hsu, N.N., Blessing, G.V., and Evans, C.J., 2000, "Acoustic Emission Monitoring of High Speed Griding of Silicon Nitride", National Institute of Standards and Technology, Manufacturing Engineering Laboratory, Gaitherburg, USA.         [ Links ]

Karpuschewski, B., 2001, "Sensoren zur Prozessüberwachung beim Spanen" , Habilitationsschrit, Universidade de Hannover.         [ Links ]

Klocke, F. 2009, "Manufacturing Process 2, Grinding, Honing, Lapping", RWTH Aachen University, Germany.         [ Links ]

König, W., Altintas, Y., Memis, F., 1994, "Direct Adaptive Control of Plunge Grinding Process Using Acoustic Emission Sensor", Institute of Machine Tools and Production Engineering (WZL), Aachen, Germany.         [ Links ]

König, W., 1989, "Fertigunsverfahren Band 2, Schleifen, Honen, Läppen", Düsseldorf, Germany.         [ Links ]

Leme, P.L.S., 1999, "Dressamento por Toque de Rebolos de CBN", Dissertação (Mestrado em Engenharia Mecânica), Florianópolis-SC, Brazil        [ Links ]

Meyen, H.P., 1991, "Acoustic Emission - Mikroseismik im Schleifprozess", RWTH Aachen,         [ Links ]

Montgomery, C.D., 2001, "Design and Analysis of Experiments", Arizona Estate University. Fifth edition.         [ Links ]

Oros, Oros 3-Series/NVGate Reference Manual. Oros Gmbh, 2006.         [ Links ]

Ravindra, H.V., Srinivasa, Y.G., and Krishnamurthy, R., 1997, "Acoustic Emission for Tool Condition Monitoring in Metal Cutting". Department of Mechanical Engineering, Indian Institute of Technology, June, India.         [ Links ]

Sensis, 2002, "Monitor de processos MSM 13 entradas, 2 níveis de comparação", São Carlos.         [ Links ]

WALTER DITTEL Gmbh. AE 6000 Process monitoring, 2007.         [ Links ]

Webster, J., Dong W.P., and Lindsay R., 1996, "Raw Acoustic Emission Signal Analysis of Grinding Process", Center of research and development, University of Connecticut, USA.         [ Links ]

 

 

Paper received 9 February 2011.
Paper accepted 4 April 2011.

 

 

Technical Editor: Anselmo Diniz