USING A CNC MILLING MACHINE FOR INCREMENTAL FORMING

Transcription

1 Proceedings in Manufacturing Systems, Volume 9, Issue 2, 2014, ISSN USING A CNC MILLING MACHINE FOR INCREMENTAL FORMING Melania TERA 1,*, Radu BREAZ 2, Octavian BOLOGA 3, Gabriel RACZ 4 1) Assist. Prof., PhD, Industrial Machines and Equiments Deartment, Faculty of Engineering, "Lucian Blaga" University of Sibiu, Romania 2) Prof., PhD, Industrial Machines and Equiments Deartment, Faculty of Engineering, "Lucian Blaga" University of Sibiu, Romania 3) Prof., PhD, Industrial Machines and Equiments Deartment, Faculty of Engineering, "Lucian Blaga" University of Sibiu, Romania 4) Prof., PhD, Industrial Machines and Equiments Deartment, Faculty of Engineering, "Lucian Blaga" University of Sibiu, Romania Abstract: The Asymmetric single oint incremental forming (ASPIF) is a manufacturing technology with a high degree of novelty which has been widely recognized as a solution with great otential in manufacturing small batches or even single sheet metal arts. Machines for this secifically rocess are hardly are almost inexistent for that reason some solutions would be using CNC machines or industrial robots. The technological equiment used for the incremental forming is the CNC cutting milling machine-tools. The control system of the CNC equiment is tuned according the accuracy domain of the milling rocess, which lies within micrometers range. The control arameters are chosen in order to minimize the ositioning errors and to deal with redictable and constant forces. This research resents an aroach of re-tuning the control system of the CNC equiment for the requirements of the ASPIF rocess, a smaller accuracy, lying within millimeters range and unredictable forces. In the last stage of the aer there was realized a comarative study of the arts' roughness function of the emloyed incremental forming rocess. Key words: CNC milling machine, control arameters, incremental forming, roughness, simulation. 1. INTRODUCTION 1 New manufacturing technologies are develoed fast in order to kee u with the market demands, both in the field of metal cutting and metal forming. Asymmetric single oint incremental forming (ASPIF) has been widely recognized as a solution with great otential in manufacturing small batches or even single sheet metal arts [1, 2]. A brief descrition of the ASPIF rocess rincile is resented in Fig. 1. The blank (2) is fixed by mean of the blank holder (3). In order to realize the shae of the sheet metal art, one of the active elements, usually the unch (1) has an axial feed movement on vertical direction, continuous or in stes s (incremental), while the other element, the active late (4) carries out a lane horizontal movement. However, in site of its great otential, the industry is still reluctant to aly the ASPIF rocess on a large scale [3, 4, and 5]. One of the reasons for that is the lack of a dedicated machine. The ASPIF rocess is now unfolded on CNC milling machines, robots and Stewart latforms and hexaods [1]. There are also in use some urose built machines, but most of them are only rototyes [6, 7]. Common CNC milling machines are the most used technological equiment for manufacturing arts by ASPIF. However, there are some drawbacks related with their use. First of all, the technological forces are bigger * Corresonding author: Faculty of Engineering, 4, Emil Cioran Street, Sibiu, , România, Tel.: ext.453 Fax: addresses: melania.tera@ulbsibiu.ro (M. Tera); Fig. 1. The ASPIF rocess. than in normal cutting (milling) rocesses and moreover, their values cannot be estimated. There are still no reliable relationshis for calculating these forces, and the users are concerned about their effect uon the CNC machines. There are some secial designed sensors able to measure the forces in the ASPIF rocess [7, 8] but measurement has to be done on-line and no estimation can be made before manufacturing the art. Usually, the CNC controllers have failsafe systems which rotect the machine if the technological forces or toques exceed the maximum allowable values. However, using the machine at its limits is not recommendable, because such regimes could affect, in time, the machine s erformance. The control system of the CNC equiment is tuned according the accuracy domain of the milling rocess, which lies within micrometers range. The control arameters are chosen in order to minimize the ositioning errors and to deal with redictable and constant forces. The

2 100 M. Tera, R. Breaz, O. Bologa and G. Racz / Proceedings in Manufacturing Systems, Vol. 9, Iss. 2, 2014 / accuracy domain of the ASPIF rocess lies within millimeters range, while the technological forces, as stated before, are quite unredictable. Also, the main objective while rocessing a art by ASPIF on a CNC milling machine is to obtain an accetable roughness. The elastic sringback of the art has a great influence uon the shae and dimensional accuracy, so it is not ractically ossible to obtain accuracies as higher as the ones obtained in the milling rocess. 2. TECHNOLOGICAL FORCES AND TORQUES The feed drives of CNC machine tools control the ositions and velocities of machine tool slides in accordance with commands generated by the CNC equiment. Basically, the feed drive is comosed of a rotary servomotor as actuation device and a recirculating ball screw as the solution for converting the rotary motion of the servomotor into linear slide motion, its bearing taking u all axial forces of the slide. The servomotor and ball screw drive are usually directly couled. The total static torque for a horizontal feed drive within a CNC milling machine may be calculated with the following relationshi: [ Ft + m( a + µ g)] M st = iη tot r, (1) where: M st total static torque Nm; F t technological force, N; m total mass (machine slide and workiece), kg; a linear acceleration, m/s 2 ; µ friction coefficient; g gravity acceleration, m/s 2 ; lead ball-screw ste mm; i transfer ratio between rotational motor and machine slide; η tot total efficiency of the mechanical transmission between rotational motor and machine slide and M r re-loading torque of the nut of the main ballscrew system. Also, the total dynamic torque may be calculated as: M = ε, (2) dt J t where: M dt total dynamic torque, Nm; J t equivalent moment of inertia of the feed drive, reduced at the rotational motor sindle, kgm 2 ; ε M angular acceleration of the rotational motor. For a vertical feed drive, one has to take into consideration the following secific characteristics: weight of the vertical slide has to be introduced. Normally, at the heavy duty machine tools the weight force is comensated by means of a counterweight which balances the vertical slide mass. However, the small machine tools, as the one which was used for this research are not equied with the counterweight system; effect of the friction force may be neglected, because its value is much smaller comared with the weight of the system; rotational motor of the vertical feed drive is equied with a brake, which has its own moment of inertia, which has to be considered when calculating the equivalent moment of inertia of the feed drive, reduced at the rotational motor sindle. M Consequently, for a vertical feed drive, relation (1) has to be modified, as: [ Ft + m( a ± g)] M st = iη tot r, (3) where the sign + stand for u movements and for down movements. As exerimental layout, a Haas MiniMill vertical machining center was used (Fig. 3). Some characteristics of the machine are resented below: the ste of the lead ball-screw on X, Y, Z axes, = 6 mm itch screw (25.4 / 6 mm), maximum feed force on X, Y, Z axes, F maxx = F maxy = F maxz = N (when these values are reached, all movements sto on every axis). Consequently, these are maximum forces which may aear during the machining rocess. Greater values of technological forces will lead to a rocess halt. The servomotors on each feed drive are Yaskawa SGMGV 09ADA6C ermanent magnet synchronous motors, with the following characteristics: nominal ower P n = 850 W; nominal torque M n = 5.39 Nm; maximum torque M max = 13.8 Nm; angular acceleration for X, Y axes motors ε M = rad/s 2 ; angular acceleration for the Z axis motor (equied with brake) ε Mb = rad/s 2 ; moment of inertia J M = kgm 2 ; moment of inertia for the motor equied with brake J M + J b = kgm 2. Taking into consideration the fact that the rotational motors are directly couled with the lead screws (i = 0), the following relationshis regarding the static torques on each axis may be written: [ FtX + mx ( a + µ g)] M stx = [ FtY + my ( a + µ g)] M sty = [ FtZ + mz ( a g)] M stz = s s s rx ry rz, (4), (5), (6) where: η s efficiency of the lead ball-screw transmission which is considered 0.9. Exressing the left art of each of (4) (6) equation by means of the maximum feed force on X, Y, Z axes, the following relationshis may be written: Fmax X M = = 6 st max X = 6.66 Νm, (7) 2 π 0.9 sb Fmax Y M = = 6 st max Y = 6.66 Νm, (8) 2 π 0.9 sb Fmax Z M = = 6 st max Z = 6.66 Νm, (9) 2 π 0.9 sb

3 M. Tera, R. Breaz, O. Bologa and G. Racz / Proceedings in Manufacturing Systems, Vol. 9, Iss. 2, 2014 / N im K e = ulses/rad. (10) According to the machine documentation, K e = 8192/ ulses/rad. a b Fig. 4. Simulation diagram. Fig. 3. Exerimental layout: a Haas Mini Mill CNC milling machine; b examle of rocessed art. where M stmaxx,y,z is the maximum values of the static torques. It is noticeable the fact that these values are inbetween the values of the nominal torque (M n = 5.39 Nm) and maximum torque (M max = 13.8 Nm) of the motors. 3. SIMULATION In order to simulate the behavior of a CNC drive a simulation diagram under Matlab and Simulink was build (Fig. 4). The diagram simulates the behavior of a osition control system using a ermanent magnet synchronous motor as actuating device. The subsystem which simulates the ermanent magnet synchronous motor was adated from the existing diagrams within Matlab and modified according to the Yaskawa SGMGV 09ADA6C motors characteristics. The osition loo which takes into consideration the mechanical transmission and the incremental encoder was introduced by the authors. The simulation diagram was also comleted with a kinematic generator, which issues the kinematic inuts (sace, velocity and acceleration) using a traezoidal velocity rofile. The kinematic inuts for a linear dislacement are resented in Fig. 5. The encoder is characterized by the encoder gain K e, which aears in Fig. 4, defined as the number of ulses emitted for one rotation of the lead screw: Fig. 5. Kinematic inuts for a linear dislacement for a traezoidal velocity rofile.

4 102 M. Tera, R. Breaz, O. Bologa and G. Racz / Proceedings in Manufacturing Systems, Vol. 9, Iss. 2, 2014 / Another inut from Fig. 4 is the mechanical transmission constant, K tr, defined as: = K tr i m/rad. (11) where: ste of the leadscrew m; i gear reducer ratio. For this articular machine tool, = m and i = 1, so K tr = / m/rad. The main controllers of the feed drive are the osition and the velocity PID controllers. The tuning arameters of these controllers are resented in Table 1. Each arameter form Table 1 is also available at the oerator anel of the CNC milling machine and can be changed by the user. The initial set of values (factory settings) for the control arameters are: K PP = 40, K IP = 64, K DP = 500 and K PV = 60, K IV = , K DV = 0.1. The initial set of control arameters were chosen by the machine manufacturer taken into consideration the main objective of milling rocess: art dimensional accuracy within micrometers range, which may be also exressed as small ositioning errors, also in the micrometers range. The tuning strategy roosed in this aer is oriented to obtain a better quality of the art surface, instead of a small ositioning error. In terms of control, it means that the velocity outut should oscillate much less than in the initial situation. The simulation rocess was unfolded for a linear dislacement x = 2 mm, with a feedrate F = 240 mm/min, a linear acceleration a = 2.68 m/s 2 and an angular acceleration ε = rad/s 2. The numerical character of the feed drive system was also taken into consideration, by introducing a samling time of T =10-6 s. The PID controllers were introduced as discrete ones, according to the diagram resented in Fig. 6 for the osition controller (the same structure was used for the velocity controller). PID Controller Control arameters Proortional gain Gains Integral gain Table 1 Derivative gain Position K PP K IP K DP Velocity K PV K IV K DV Fig. 6. Discrete osition controller. Fig. 7. Simulated technological torque. The simulated technological torque was introduced according to the diagram resented in Fig. 7: at the beginning of the movement cycle the torque is equal to the nominal value M n = 5.39 Nm and after 0.2 the value increases to M stmax = 6.66 Nm. As resented above, any value greater that 6.66 Nm will lead to a sto in the movement of the feed drive. Two sets of simulations were unfolded, the first one with the initial control arameters resented above and the second one with the following arameters: K PP = 4, K IP = 64, K DP = 500 and K PV = 15, K IV = , K DV = 0.1. The first set of control arameters characterize a system tuned for milling while the second set characterize a system tuned ASPIF. As it can be observed, changes occurred only at K PP, K PV and K IV values. The tuning rocess was based artially on a trial and error rocess, artially on some analytic relationshis recommended by the machine manufacturer For the velocity controller, the integral time T IV was introduced, together with the following relationshis: T IV =, (12) K PV K PV K IV =. (13) T The value K PV = 15 was recommended by the machine manufacturer for low-rigidity systems and K IV was calculated according to equations (12) and (13). In Fig. 8 the velocity ste resonses for the two systems are resented. It is here noticeable the fact that the system tuned for milling is faster than the one tuned for ASPIF. However, neither overshoot, nor steady state errors are resent for both systems. Fig. 9 shows the simulated ositioning errors for both systems. The error for the system tuned for milling is four times smaller than the error for the system tuned for ASPIF. The velocity oututs are resented in Fig. 10. In order to show more accurate results, only a detail form the oututs are shown. One can notice that the velocity outut oscillations are significantly greater fir the system tuned for milling than for the system tuned for ASPIF. IV

5 M. Tera, R. Breaz, O. Bologa and G. Racz / Proceedings in Manufacturing Systems, Vol. 9, Iss. 2, 2014 / Fig. 11. Measured roughness for the system tuned for ASPIF Fig. 8. Simulated ste resonses. Tuning for milling and ASPIF Table 2 Tuning R a µm Tuned for milling Tuned for ASPIF Roughness R a for different rocessing methods The rocessing methods Accuracy classes R a µm Rough turning Finish turning Rough milling Finish milling Grinding Suerfinishing Table 3 Fig. 9. Simulated ositioning errors. Fig. 10. Simulated velocity oututs (detail). 4. EXPERIMENTS In order to validate the roosed control strategy a number of simle arts were rocessed by means of ASPIF rocess on the Haas MiniMill vertical CNC machining center. The same set of control arameters as the ones used for simulation were used during the machining rocess. The arameters were altered from the CNC oerator anel. The arts were generated as hemisheres and simle circular trajectories in the horizontal XY lane, equally saced between each other on the Z axis were used to obtain the shae of the arts. Further exerimental research studied the surface quality of machined arts, with different variants of the rocess of incremental forming. The roughness of the arts was measured and one result is resented in Fig. 11. For examle, the R a value for this case was between µm and µm as shown in Table 2, where the most reresentative of these are resented. Surface roughness analyze is made using quality assessment criteria R a. In order to make a comarison of the surface roughness of arts made by incremental forming with cutting rocess in Table 3 was resented standard roughness classes R a for several machining rocesses. Among the criteria showed the highest accuracy is assured by roughness R a, its evolution is shown in Fig. 12, where are resented seven different tyes of arts obtained through incremental forming. It is noted that the tye ieces hemishere arithmetic mean deviation R a irregularities values have values between µm and µm are equivalent with flat finish milling rocess. For arts of tye roughness hemishere best variant obtained rocessing method which incororates the comlex contour siral buoyancy comleted circle. Comlex arts have values aroximately 70% lower roughness R a, and the obtained values fall within rocessing by grinding. The lowest roughness value of micrometres was obtained for comlex song erformed by otimized roughing and finishing.

6 104 M. Tera, R. Breaz, O. Bologa and G. Racz / Proceedings in Manufacturing Systems, Vol. 9, Iss. 2, 2014 / Fig. 12. Arithmetic average deviation of irregularities R a µm. 5. CONCLUSIONS The ASPIF rocess is still a new and romising manufacturing technology, yet it lacks of dedicated technological equiments. CNC milling machine are used for this urose but the control arameters are chosen for a very different rocess, milling. The work resented in this aer roosed some simulation tools, which were used for testing a new control strategy for the CNC feed drives used for the ASPIF rocess. Instead of focusing on the ositioning accuracy, the CNC controller was tuned in order to reduce the oscillations of the velocity outut, which consequently lead to a better surface quality of the rocessed arts. The simulated control strategy was validated by exeriments. The arts rocessed with the CNC equiment tuned secifically for the ASPIF rocess had shown a significantly smaller roughness as the arts rocessed with the initial tuning, secific for the milling rocess. In the last stage of the aer there was realized a comarative study of the arts' roughness function of the emloyed incremental forming rocess. ACKNOWLEDGMENTS: This work was suorted by the strategic grant POSDRU/159/1.5/S/133255, Project ID (2014), co-financed by the Euroean Social Fund within the Sectorial Oerational Program Human Resources Develoment REFERENCES [1] J. Jeswiet, F. Micari, G. Hirt, et al., Asymmetric single oint incremental Forming of Sheet Metal, Annals of CIRP, 54(1), 2005, [2] V. Oleksik, A. Pascu, C. Deac, R. Fleaca, M. Roman, Numerical simulation of the incremental forming rocess for knee imlants, Proc. COMPLAS X, X International Conference on Comutational Plasticity, Fundamental and Alications, eds. E. Oriate and D.R.J. Owen, Barcelona, [3] V. Oleksik, O. Bologa, R. Breaz, G. Racz, A. Găvruş, Numerical simulations and exerimental researches regarding the recision of the incremental forming rocess, The 12th Conference on Metal Forming, Krakow Poland, Setember 21 24, 2008, ublished on Steel Research International, No. 79 (2008), Secial Edition, Vol. 1, Publishing Comany: Verlag Stahleisen Gmbh, , [4] V. Oleksik, A. Pascu, D. Mara, O. Bologa, G., Racz, R. Breaz, Influence of geometric arameters on strain and thickness reduction in incremental forming rocess, The 13th Conference on Metal Forming, Toyohashi Jaan, Setember 19 22, 2010, ublished on Steel Research International, Vol. 81 (2010), Secial Edition, nr. 9, Publishing Comany: Verlag Stahleisen Gmbh, [5] M. Rauch, et al., Tool ath rogramming otimization for incremental sheet forming alications, Comuter-Aided Design, 41, 2009, [6] J. M. Allwood, N. E. Houghton, K. P. Jackson, The design of an Incremental Forming machine, 11th Conference on Sheet Metal, Erlangen, , [7] J. Jeswiet, J. Duflou, A. Szekeres, P. Levebre, Custom Manufacture of a Solar Cooker a case study, Journal Advanced Materials Research, Vols. 6 8, May 2005, [8] J. Jeswiet, J. Duflou, A. Szekeres, Forces in Single Point and Two oint Incremental Forming, Journal Advanced Materials Research, Vols. 6 8, May 2005, [9] J. Duflou, A. Szekeres, P. VanHerck, Force Measurements for Single Point Incremental Forming: and exerimental study, Journal Advanced Materials Research, Vols. 6 8, May 2005,