f flu Copyright 1996 by ASME

Transcription

1 Analysis of Drill Failure Modes by Multi-sensors on a Robotic End Effector OUTER CORNER WEAR FLANK WEAR Shane Y. Hong, 1 J. Ni, 2 and S. M. Wu 3 The characteristics of nine common drill failure modes have been analyzed. To facilitate detecting and diagnosing these failure modes, a multi-sensor approach is developed. Based on the dynamic characteristics of three different measurements including thrust, torque, and lateral vibration, necessary signatures are extracted for the nine failure modes. Experimental tests were conducted to confirm the characteristics of these nine drill failure modes. 1 Introduction A drill can fail in different modes. These modes include chisel edge wear, flank wear, outer corner wear, margin wear, crater wear, breakage, chipping at lip, lip height differences, and drill asymmetry. Each mode produces a different signature. Past drill wear monitoring approaches were based on signature analysis of thrust force, torque, power consumption, temperature, vibration, acoustical signal, and radio active materials [1-8 ]. The bulk of experimental methods studying tool wear and tool life are derived from direct observation of the wear zone. Generally, selected dimensions of the flank wear area are measured and used to estimate actual tool wear. But, flank wear is just one of many drill failure types. Other types of drill failures may be more critical in certain applications. For instance, a drill with severe flank wear may still work well if the ratio of feedrate to speed is modified. On the other hand, a newly ground drill with sharp cutting edges may cause excessive vibration and produce poor quality holes due to asymmetry or height differences between drill lips. From a practical standpoint, drill life should be defined by performance, not by a specified flank wear limit. A drill should not be used if one of the following conditions exists: (a) excessive power consumption; (b) excessive force and torque; and (c) poor work quality. This paper examines nine different types of drill failures, as shown in Fig. 1, and analyzes their failure characteristics. A multi-sensor approach facilitates the detection and recognition of these different failure modes. Experimental test results are presented to confirm the described characteristics. 2 Effect of Drill Failure on Thrust, Torque, and Vibration Three physical parameters: thrust, torque, and lateral vibration, were chosen to analyze drill failures because of their direct relationship to drill performance. Thrust force determines the machine rigidity and strength requirements. Torque and speed set the power requirement. The lateral vibration has a direct effect on the hole location, dimension, geometry, and surface finish. All other phenomena can be derived from these three parameters. For example, any temperature rise in the cutting 1 Associate Professor, Department of Mechanical and Materials Engineering, Wright State University, Dayton, OH Associate Professor, Department of Mechanical Engineering and Applied Mechanics, The University of Michigan, Ann Arbor, MI Professor, (deceased), Department of Mechanical Engineering and Applied Mechanics, The University of Michigan, Ann Arbor, MI Contributed by the Manufacturing Engineering Division for publication in the JOURNAL OF MANUFACTURING SCIENCE AND ENGINEERING. Manuscript received May 1992; revised Aug Technical Editor: S. K. Kapoor. MARGIN WEAR CRATER WEAR BREAKAGE f flu CHISEL EDGE WEAR ^ ft\ DIFFERENCE IN UP, HEIGHT ASYMMETRY CHIPPING AT UP Fig. 1 Different types of drill failure region is due to excessive power consumption i.e. excessive torque. 2.1 Thrust and Torque. Research and analysis indicate that drill torque and thrust are functions of drill diameter, drill chisel edge length, feed per revolution, and work material [7, 9 ]. The thrust force will increase as the contact area increases, which may be a result of tool wear. The torque will also increase due to frictional force increases caused by flank wear. However, the primary factor affecting torque force is rake angle. Rake angle changes the shear angle, and any breakage or built-up edge results in a drastically increased torque. The crater wear on a carbide drill also has a direct effect on the torque, but less so on the thrust. Note that torque is the sum of the cutting forces multiplied by the distance from the drilling center. Increases of cutting force at the outer radius will increase torque more than increases at the inner radius. However, thrust is not a function of the radius. This difference in radius-dependence can help identify the location of the wear or the type of defect. The thrust to torque ratio can also be used to recognize different drill failures. 2.2 Vibration. Three kinds of vibration are possible in a drilling process: torsional, longitudinal, and transverse. Since torsional and longitudinal vibrations do not significantly affect the hole surface quality, this analysis focuses on the transverse vibration, especially the forced vibration case. Cutting forces can be decomposed into three components, P x, P y and P z. The P y 's are the indenting forces in feed motion and contribute to the thrust. The P z 's are a couple, which needs a torque to overcome it. In the ideal case, the forces of P x acting on the drill are of equal magnitude but opposite directions along 672 / Vol. 118, NOVEMBER 1996 Transactions of the ASME Copyright 1996 by ASME

2 Table 1 Drill samples used in the test Drill Major Defects Description No. 0 New drill A new drill without any defects 1 Chisel edge wear Pure chisel edge wear only; cutting lips and margin are new 2 Cutting lips and margin Chisel edge in good condition; outer corner wear not so clear; cutting lips with flank wear; margin wear 0.25 mm from the corner 3 Lip difference One cutting lip with flank wear; the other lip and chisel in good condition; margin and comer in good condition 4 Worn all over Chisel edge wear; flank wear on both cutting lips; a litde corner wear; margin in good condition 5 Outer corner and margin wear Chisel and cutting lips in good condition; outer corner wear and margin wear 6 Breakage Small pits of breakage in the cutting lips 7 Chipping at one lip Everything in good condition except a small chipping at the center of one lip the same line, thus canceling each other. But when the drill is asymmetrical, because of poor drill point grinding or drill wear/ breakage, the two P x forces tend to be unbalanced. The unbalanced force causes a sinusoidal vibration with a frequency identical to the spindle speed. The deflection of the drill or lateral displacement caused by this vibration tends to overcut on one side and undercut the other. Depending on the vibration pattern, the drilled holes may be oversized, have irregular geometry and shape, or have poor surface finish. 3 Experimental Results A computer controlled drilling machine, Robotic Drilling Unit (RDU), was used for the experiment [10]. The spindle of the machine was driven by two air motors. During the experiment, a feedrate of mm per minute (1.2 inch/minute) was maintained. The air supply was kept at 465 KN/m 2 (67.5 psi), resulting in a free running speed of the spindle around 2100 rpm. The material used in the experiment was 2024T851 aluminum plate, 12.7 mm (0.5 inches) thick. Spiral point drill bits with diameters of mm ( inch), made of high speed steel, were used in all of the tests. Different drill samples having defects listed in Table 1 were used. In the experiment, each drill sample was used to drill 12 to 16 holes. The output curves were fairly similar for the same drills, and easily distinguished from those of different drills. The results of the experimental tests are discussed as follows. 3.1 Thrust Test. When drilling aluminum plate, the thrust history has the following characteristics: The thrust curve of a new drill is flat and low (Fig. 2(a)). The thrust curve of drills with only chisel edge wear, Fig. 2(b), is slightly higher than that of a new drill. The thrust for drills with flank wear on the two cutting lips, Fig. 2(c), is higher than in the former two cases. The thrust history, Fig. 2(d), of drills with a new chisel edge, one new cutting lip, and one worn cutting lip, fluctuates which could be explained as the variation in the depth of cut of the two lips. The thrust for drills worn all over (including chisel edge, cutting lips, and outer corner) is extremely high compared with the former samples, Fig. 2(e). The thrust for drills worn at the margin corner, one side a little more seriously than the other, Fig. 2(f), has an average value close to the one shown in Fig. 2(c). The high thrust is partially due to speed reductions and depth of cut increases per revolution. Some fluctuation is also observed. The thrust of drills with several small pits of breakage, Fig. 2(g), is high and fluctuates greatly. The thrust of a new drill with one small chip at the center of a cutting lip has sharp pulses appear on the curve, Fig. 2(h). 3.2 Torque. Figure 3 shows the smoothed curve of the torque (speed reduction) produced by different drill samples. Drill No. 0 (new) had the least speed reduction. Drill No. 1 (chisel edge wear only) had just a little more speed reduction than the new one. Drill No. 2 (flank wear) had even more speed reduction than the previous one. Drill No. 4 had both chisel edge wear and cutting lip wear, and the speed reduction can be seen to be almost the sum of the reductions due to chisel and flank wear when compared with drills No. 1 and 2. It was observed that drill No. 5, with outer corner and margin wear had a very high torque. Drill No. 6, which has pits of breakage, has the most speed reduction in the drilling process. 3.3 Lateral Vibration Test. Figure 4 shows the vibration amplitude drilling history of different drill samples. The vibration amplitude is greatest when the drill touches the workpiece and is "walking around." As the drill enters the hole, due to the centering capability of the drill point as well as the constraint of the drilled hole to the drill bit, the vibration amplitude is reduced. To analyze the drill condition, the curve was divided into three portions: Vibration at Beginning (VB): The centering ability of the drill bit is the major concern. Vibration at Middle (VM): The cutting force balance and constraint of vibration are the point of interest. Vibration at Retracting (VR): The rotating drill is retracting, only marginal contact is possible with the hole surface. The centering capability of the drill by the chisel edge and cutting lips no longer exists. The results show that only the drills with chisel edge wear or breakage cause increased vibration in the VB region. It is also seen that drill No. 3, which shows a great thrust variation, does not have a significant increase in lateral vibration. In the VM portion, drills No. 2 and 6 exhibited more vibration than that of a new drill. Drills No. 1,5, and 7 exhibited small vibration increases compared with a new drill. It is observed that drill No. 3, having lip height difference and large thrust fluctuation, does not show much lateral vibration. Drill No. 4 is worn all over, but does not show excessive vibration either. In the VR portion, only drills No. 2, 5, and 6 had significant vibration. It is found that drills No. 2 and 5 are the only two drills which have margin wear. It is interesting to note that the surface roughness produced by those drills coincides with the vibrational pattern in the VR portion. This is because only the margin contacts the finished hole surface during the final stage of the drilling process. The margin wear or excessive vibration from breakage of a drill bit has a significant effect on the hole surface. 4 Discussion of the Experimental Results Some general observations can be drawn from the experimental results: (a) (b) (c) Breakage and margin wear can cause more lateral vibration when the spindle is retracting. Flank or chisel edge wear has little influence on the torque, mainly due to an increase in frictional forces. However, drill failures that tend to increase normal rake angle (such as tool breakage, corner wear, or crater wear), have more significant influence on the torque. Hole quality is highly influenced by the margin wear or excessive imbalance of the cut. It is less affected by pure chisel edge or flank wear. The amplitude of lateral Journal of Manufacturing Science and Engineering NOVEMBER 1996, Vol. 118/673

3 Fig. 2 The thrust force curve of different drill samples: (a) drill #0, new (b) drill #1, chisel edge wear (c) drill #2, flank wear (d) drill #3, lip height difference (e) drill #4, worn all over (f) drill #5, margin wear (g) drill #6, breakage (h) drill #7, chipping at lip vibration is an indicator of hole quality, especially the lateral vibration occurring during spindle retraction. As expected from the analysis in previous sections, different drill defects yielded different combinations of sensor outputs. (1) Chisel Edge Wear: Compared to a new, flawless drill, a drill with chisel edge wear increases the average thrust and t- -+- NO. O NEW DRILL N0.4 WORN ALL OVER NO. I CHISEL WEAR N0.5 CORNER WEAR N0.2 CUTTING LIPS WEAR N0.6 BREAKAGE NO 3 ONE SIDE WORN ONLY TIME (Sec.) N0.7 CHIPPING Fig. 3 Smoothed torque (speed reduction) curves during the drilling by different drill samples the lateral vibration at the entrance. The ratio of thrust to torque for a drill with chisel edge wear tends to be larger than that of a new drill. Unless mixed with other drill defects, chisel edge wear alone does not cause fluctuation. Chisel edge wear will increase torque, but not significantly. (2) Flank Wear: A drill with flank wear tends to dramatically increase thrust, and moderately increases torque. Its ratio of thrust to torque is larger than that of a new drill. Unless there is uneven wear at the two flank edges, flank wear generally does not cause thrust variation and lateral vibration. (3) Crater Wear: Although crater wear was not tested in the experiment due to lack of available samples, it is expected that crater wear will increase torque, but not thrust. The thrust/ torque ratio should be smaller than that of a new drill. Vibration in thrust direction or lateral direction should be negligible. (4) Corner Wear: A drill with outer corner wear tends to increase both torque and thrust. It increases torque more than thrust as it occurs at the outer radius, making the thrust to torque ratio smaller than that of a new drill. Because it is not easy to wear two corners equally, the nature of slight asymmetry will make some thrust fluctuation and lateral vibration at the final stage. (5) Margin Wear: Margin wear does not have much effect on torque and thrust. The margin edge of a drill contacts the drilled hole surface, therefore, its wear has a large impact on the hole quality. However, the sensing of margin wear is rather difficult. The only clear sign of margin wear is revealed by lateral vibration at the final stage of the drilling process. 674 / Vol. 118, NOVEMBER 1996 Transactions of the ASME

4 (b) Drill fl, Chiiel Edge Wear TIME (Sac) TIME T",i.i i ii.tttrttni (Sacl -I 1 1 l 1 1- (c) Drill 12, Flank Wear -I 1 I (d) Drill #3. Up Height Difference so TIME (sac) B TIME (Sac) I I v \ 1- (c) Drill #4. Worn all Over 1 i 1 1 (0 Drill K, Margin Wear ~T~t 10 B TIME (SIC) li lliitl E 40 -I H 1- (g) Drill #6, Breakage -f- (h) Drill n. Chipping al Up Fig. 4 The vibration amplitude history of different drills (6) Chipping at Lips: Chipping at the cutting edge will increase torque and thrust to some degree, depending on the chip shape, size and location. The increased thrust fluctuation results from an unbalanced cut. The thrust to torque ratio decreases as torque increases more than thrust due to the geometry change at the cutting edge. (7) Drill Asymmetry: Drill asymmetry may be caused by unbalanced wear, misaligned grinding of a new drill, or bent axis of the drill. The thrust and the torque may not change significantly, but there is great vibration in the axial direction (thrust variation) and in the cross direction (lateral vibration) at the beginning, middle, or final stage of the drilling process. (8) Breakage: Drill breakage changes the drill cutting geometry and causes disruptive sensor signals, including a significant increase of the cutting torque, lateral vibration at any stage, thrust and thrust average fluctuations. (9) Lip Height Difference: A drill with height differences between lips is caused by poor grinding. Thrust fluctuation is the strongest indication of height differences between lips. This drill defect may not affect the thrust average, torque, or lateral vibration if it is not associated with other drill defects. To facilitate the implementation of diagnostic functions, the results and discussion from previous sections can be mathematically represented using set theory. The set theory representation is given in the reference [11] for establishing the criteria of detecting and diagnosing drill failures. The set theory representation will link the sensor outputs to the drill conditions. 5 Conclusion The following conclusions can be drawn: (a) The thrust force signal can be decomposed into: thrust average and thrust fluctuation. The former increases as flank wear increases, while the latter increases when cutting lip height is imbalanced. Therefore, the uniform (b) (c) (d) wear of a drill will increase the average thrust, but not the thrust fluctuation; and a drill with lip height differencse will increase the thrust variation, but not the average thrust. The torque increases more significantly due to drill cutting edge geometry changes, rather than from pure wear. The change of thrust to torque ratio is a valuable criterion for separating types of drill failures into two groups depending on the drill geometry change and the location of defects. The lateral vibration at the beginning of drilling indicates the drill point centering capability, reveals chisel edge wear, and drill asymmetry; while the lateral vibration at the time of retracting is a good indication of hole quality and is influenced by margin wear, breakage, or asymmetry. References 1 Amini, E., Kwaitkowski, A. W., and Winterton, R. H. S., 1977, "Twist Drill Wear Measurement Using Surface Activation," Proc. of the Int. Mach. Tool Des. and Res. Conf. 18th, London, England. 2 Beer, L. D., 1979, "Power Consumption A Measure of Tool Performance," Technical Paper MR79.398, SME, Vol. 103, pp Boothroyd, G., Eagle, J. M., and Chisholm, A. W. J 1976, "Effect of Flank Wear on the Temperature Generated Metal Cutting," Proceeding of the 8th MTDR Conf., pp Braun, S., Lenz, E., and Wu, C. L., 1982, "Signature Analysis Applied to Drilling," ASME paper 81-DET-9 for meeting September 20-30, 1981, p Cook, N. H., and Subramanian, K., 1978, "Micro-isotope Tool Wear Sensor," Annals of the CIRP, Vol. 21, No Jetley, S. K., 1984, "A New Radiometric Method of Measuring Drill Wear," SME Manufacturing Engineering Transactions, 12th NAMRC, Houghton, MI, May 30-June 1. 7 Shaw, M. C, and Oxford, Jr., C. J., 1957, "On the Drilling of Metals, II The Torque and Thrust in Drilling," ASME JOURNAL OF ENGINEERING FOR INDUSTRY, Vol. 79, pp Thangarai, A., and Wright, P. K., 1988, ' 'Computer-assisted Prediction of Drill Failure Using In-Process Measurements of Thrust Force," ASME JOURNAL OF ENGINEERING FOR INDUSTRY, Vol. 110, pp Journal of Manufacturing Science and Engineering NOVEMBER 1996, Vol. 118/675

5 9 Williams, R. A., 1974, "A Study of the Drilling Process," ASME JOURNAL OF ENGINEERING FOR INDUSTRY, November. 10 Horng, S. Y Wu, S. M., and Voss, T., 1982, "An End-Effector for Robotic Drilling," Proceedings of the 14th National SAMPE Technical Conference, Atlanta, GA, October Horng, S. Y., "Development of an End Effector for Robotic Drilling with On-line Sensing and Diagnosis," Ph.D. dissertation, University of Wisconsin- Madison, Table 1 Order of magnitude analysis Inertia = 10 s Pa Centrifugal = 105 Pa Surface Tension = 10 s Pa Pressure Boundary KIO 5 Pa generalized Eularian-Lagrangian finite element formulations for incompressible viscous flow are as follows Free Surface Modeling of a Glass Tube Sealing Process Mi] dv= 0 H. P. Wang 1 and Y. F. Zhang 2 Introduction The tube sealing process is implemented by heating the surface of a partially evacuated glass tube by means of a gas torch. Once the glass temperature reaches the softening point, the tube shrinks because of pressure differential to form a seal. An order of magnitude analysis first shows that both the centrifugal force, due to tube rotation, and the surface tension are negligible compared with the pressure force on the boundary. One intrinsic characteristic associated with the mechanics of the sealing is that the physical boundary of the glass tube deforms during processing. In addition, the glass viscosity is also highly temperature-sensitive; therefore, a numerical approach is necessary for obtaining quantitative process information. The process can be described as an incompressible viscous flow having a distributed heat flux on the boundary. The Galerkin finite element formulation based on the updated Lagrangian technique [1] was used to model this complex problem. The sensitivity analyses and the comparison between the results from modeling and the experiment are also demonstrated. Order of Magnitude Analysis The geometry, the processing conditions, and the glass material properties [2] for a sample sealing process are specified as: Tube Dimension 0.008m(ID) and 0.01 (OD); Pressure Conditions 10 s Pa (ambient) and Pa (internal); Tube Rotation fi 105 rad/s (1000 rpm); Maximum Processing Temperature 1300 C; Estimated Radial Velocity U r» to 0.01 m/s; Density p 2240 Kg/m 3 ; Viscosity p, «200 Pa at 1300 C. From these data, each force is estimated in Table 1. Generalized Eulerian-Lagrangian Finite Element Formulation The glass material is considered here as Newtonian incompressible viscous fluid having a temperature-sensitive viscosity. Following the previous finite element formulations for phase change problems [3, 4], a general Eularian-Lagrangian finite element formulation for Navier-Stokes equations has been derived by using Galerkin's weighted residual method [5]. The 1,2 General Electric R&D, Schenectady, NY. Contributed by the Manufacturing Engineering Division for publication in the JOURNAL OF MANUFACTURING SCIENCE AND ENGINEERING. Manuscript received March 1994; revised Aug Technical Editor: S. K. Kapoor. JV c,[f + <4- u;,^v - \ pg&dv - I ^TijitjdA = 0!<%«-! qrijda = 0 When the element velocity U' = 0, these equations are Eularian, and when U' Uj, these equations are Lagrangian. For modeling the sealing process, the generalized Eulerian-Lagrangian formulation was used and the inertia terms in the momentum equations were neglected. Finite Element Model and Results A Gaussian distribution of surface heat flux is assumed in the model for simulating the heat input of a gas torch. The total input power is about 80 W with a span of 0.004m in the axial direction. The radiation boundary condition is on the rest of the outer surface. Because of the symmetry, the boundary at the inner surface is assumed to be insulated. Both end conditions are specified as the initial temperature (which is the room temperature) because the temperature fronts have never reached the ends in such a short period of time. For the boundary condition of the momentum equation, the ambient pressure and the internal pressure are specified on the outer and inner surfaces, respectively; both end conditions are assumed to be rigid, i.e., zero velocities. Sensitivity Analyses The sensitivity analyses for the sealing process were implemented by varying these five parameters: power level, power variation, positioning, pressure differential, and preheating in the finite element model. Table 2 summarizes the results of this Table 2 Summary of sensitivity analyses Processing condition Geometry variation a. 0.1 s of Timing Control (no preheating) 90 W, 1.7 s (standard) 0.8 mm 60 W, 3.3 s 0.21mm b. 1% Power variation 90 W, 1.7 s 0.14 mm 60 W, 3.3 s 0.12 mm c. 0.2 mm Position change 90 W, 1.7 s 0.20 mm d. 76 Torr Pressure Change 90 W, 1.7 s 0.10 mm e. 0.1 s of Timing Control (with preheating) 60 W, 1.7 s 0.31mm 40 W, 4.0 s 0.06 mm 676 / Vol. 118, NOVEMBER 1996 Transactions of the ASME