Vibration Analysis in Robotic Ultrasonic Welding For Battery Assembly

Transcription

1 8th IEEE International Conference on Automation Science and Engineering August 20-24, 2012, Seoul, Korea Vibration Analysis in Robotic Ultrasonic Welding For Battery Assembly Sang Choi, Thomas Fuhlbrigge, and Srinivas Nidamarthi, Member, IEEE Abstract This paper presents the vibration analysis of a robotic system, which performs ultrasonic welding of metal tabs on rechargeable batteries commonly used for hybrid or electric vehicles. A robotic battery-welding cell was set up, battery tab welding was performed and the vibration data was collected by using accelerometers and LVDT sensors. The data was analyzed and we learned if/how the vibration affected the robot arm and the battery object under various welding configurations. I. INTRODUCTION As there has been an increasing concern regarding global warming and the increasing gas price, it becomes more and more obvious that Electric Vehicle EVs (Electric Vehicles) and HEVs (Hybrid Electric Vehicles) will be dominant over the vehicles of conventional petroleum-based fuels in next few decades. The vehicle shipments in 2012 will rise to 2.7 million, 250,000 and 40,000 units of HEV, PHEV (Plug-in Hybrid Electric Vehicles) and EV respectively. It is forecast that the battery market size in 2012 reaches $ 3.8 billion. (Lux Research Inc, 2008) [1][2]. Battery assembly process emerges as one of the most critical and challenging steps of EV/HEV manufacturing. The critical motivation of automation is lower production cost, quality improvement and consistency. There are three distinctive levels of assembly; cell, module and pack. Cell level manufacturing involves raw material mixing, coating, winding, sandwiching, tab welding and sealing [3]. It is followed by module level processes including heat-sink assembly, layering/stacking, compressing and welding or joining. Pack assembly requires processes like pick-n-placing, liquid dispensing, harness wiring and probing. II. ULTRASONIC WELDING OF METAL AND ISSUES One of the most critical steps in battery assembly process is tab-welding/joining. Ultrasonic welding joins metal parts by applying the energy of high frequency vibrations onto the interface area between the parts to be welded. Electrical energy is transformed into high-frequency mechanical vibration, which is then transferred to a welding tip through an acoustically tuned horn as shown in Figure 1. The parts are scrubbed together under pressure at 20,000 or 40,000 cycles per second. This high frequency vibration, applied under force, disperses surface films and oxides, creating a clean, controlled, diffusion weld. As the atoms are combined This work was supported by Corporate Research Center, ABB Inc. The first, second and third author of this paper are with the ABB, Inc. U.S. Corporate Research Center - Mechatronics and Robotic Automation, 5 Waterside Crossing, Windsor, CT 06095: sang.choi@us.abb.com, Thomas.a.fuhlbrigge@us.abb.com. The third author is with Discrete Automation and Motion at ABB Inc. The Division of Robotics, Auburn Hills, MI, USA: srinivas.nidamarthi@us.abb.com between the parts to be welded, a true, metallurgical bond is produced [8]. Figure 1. The operational concept of ultrasonic metal welding [8] The ultrasonic welding is a cold process, and therefore can replace the commonly used laser welding, where the heat generated from the laser power causes problem especially in battery manufacturing process. However, there are issues in the ultrasonic metal welding due to the nature of the ultrasonic welding that requires very high-frequency vibration. It is a good assumption that the vibratory displacement x(t) at the welder tip as a simple sinusoidal wave, which, when double-integrated, gives us the acceleration at the horn tip. a (t) = A (2 ) 2 f cos (2 f t) (1) A is the amplitude of the tip motion, and f is the welding frequency. In our case, the welding frequency of 20 khz and the welding amplitude of about 45 µm are used. This translates into the maximum acceleration at the tip to be about 70,000g. Such an extreme g-level is equivalent to that of a hand-gun bullet [4]. The energy generated by such a high-g level vibration is expected to be utilized for welding, but the energy will be inevitably dispersed and transmitted into the tool, robot, workpiece, fixtures and surrounding structures. This leads to issues as follows: Vibration seen by robot Vibration seen by workpiece Even though, it is widely accepted that the effect of welding vibration on robot arm will be negligible, we will quantify and validate this effect by collecting and analyzing the vibration data from welding in section IV. Bigger concern to battery manufacturers of electric vehicle is welding vibratory energy transmitted "into" the workpiece. The ultrasonic vibration caused by metal welding delivers transverse vibrations that are parallel to the weld surfaces, whereas plastics welding delivers longitudinal vibrations that are normal (i.e., at right angles) to the weld surfaces [6][7]. For this reason, high-energy vibration in /12/$ IEEE 550

2 metal welding tends to easily transmitted into the battery object. Therefore, the vibration from metal welding is very likely to cause serious damage on the internal structure of battery object. Welding tool may need to have a feature that mitigates the vibration undesirably transmitted to the battery object. We will investigate these issues in section V. III. ROBOTIC WELDING SYSTEM SETUP A robotic battery tab-welding system was built for investigating the vibration effect on the robot and battery cell during the welding. We used a metal ultrasonic welder of 3.3 kw, 220 VAC, and 20 Amp [5]. The welding head shown in Figure 2. weighs about 50 kg, and therefore a high payload (60 kg) industrial arm was used (see Figure 3. maximum reach of 2.55 m, and position repeatability of mm) [9]. Figure 4. Welding tip (anvil and horn) and welding mark of 5 x 25 mm IV. VIBRATION SEEN BY ROBOT We first investigated how much the welding vibration affect the robot manipulator by using accelerometers attached to the various locations on the robot. Depending on the signal range, two different accelerometers of different sensitivities were used. The locations of the sensors are labeled in Figure 5. Figure 5. Robotic welding systems setup robot, welder and fixture table Figure 2. The welding head: Amtech Inc. 33.kW, 220VAC, 20 Amps (Left) and Battery mock-up cell used for welding experiment (Right) For the welding, we used a pair Al-Cu tabs for anode and cathode, whose specification is listed in TABLE II. TABLE II. CONFIGURATION OF TAB USED FOR EXPERIMENT Metal Type Welding side Width Thickness Al Alloy µm Horn side 40 mm Meets ASTM B209 (0.006 ) Cu Alloy µm Anvil side 40 mm Meets ASTM B152 (0.006 ) Figure 3. The 6-axis robot for welding experiment IRB 6400 RF [9] TABLE I. WELDING PARAMETERS TUNED FOR AL-CU PAIR Material Al = 150 µm : horn side Cu = 150 µm : anvil side Amplitude 45 µm Welding Pressure 45 psi Welding Energy 800 Joule Welding Time 0.5 sec 5 sec (min-max) Welding Freq. 20 khz The top plot in Figure 6. shows our initial vibration data collected at the lateral side of the welding tool adaptor (dotted circle in Figure 3. ). Welding energy of 800J and pneumatic pressure of 45 psi was used as shown in TABLE I. The middle plot is the zoomed plot of the top plot 0.1-sec period from 0.64 to 0.74 sec, which is again zoomed for 1- msec period starting from shown in the bottom plot. Battery cell samples were built in-house as shown in Figure 2. A pair of 3mm-thick, 120mm 120mm- square plastic plate were sandwiched each other as a cell housing, which was then enclosed by a pair of thin, 150mm 150mm - square aluminum foil. This structure is typical to a prismatic battery cell used in the current electric vehicle industry [10]. 551

3 TABLE III. VIBRATION MITIGATION COMPARISON BETWEEN RUBBER AND POLYURETHANE INSERTS Displacements (% decreased to) Inserts On tool adaptor On link 6 On link 5 None µm µm µm Rubber µm (11.5%) µm (3.5%) µm (11.1%) P.Urethane µm (10.7%) µm (2.6%) µm (11.1%) Figure 6. Vibration data from accelerometer mounted on the robot: Fullrange plot (top) followed by zoom-in plots below. (0.1 and sec) Power spectrum analysis (Figure 7. shows that the strongest signal is located around 20 khz, which is the ultrasonic welding frequency. We use this mean-squared power spectrum to verify the most dominant frequency component occurs at 20 khz with a harmonic around 40 khz. Figure 9. Peak displacement (µm) in three cases (plot from TABLE III. ) Figure 7. Frequency plot of the top plot in Figure 6. A. Vibration Mitigation Compliant Material Even if the vibration on the robot arms are small (bluecolored line in Figure 9., in order to avoid any possible damage to the robot, we attempted to further minimize the vibration on the upper links (tool adaptor, link 6 or link 5) by introducing compliant materials between the tool adaptor and the welder body, as shown in Figure 8. V. VIBRATION SEEN BY WORKPIECE In this section, we will investigate the problem of welding vibration transmitted to the battery object by conducting a series of welding experiments. Then, the vibration isolation or mitigation method will be proposed and verified by experiment. We used LVDT for measuring the vibration on the tab and battery cell pouch as shown in Figure 10. Figure 10. LVDT probe used for measuring vibration of battery tabs: Measurement range: 10 mm, resolution: 0.5 µm, accuracy: ± 1 µm Figure 8. Tool adaptor and compliant materials for vibration reduction TABLE III. shows the peak vibratory displacements in three distinctive cases: without any insert, with rubber insert and with polyurethane insert. TABLE III. and Figure 9. show that the vibration seen and measured by the robot has been significantly mitigated with the compliant material inserts. Figure 11. Two measurement points (or levels) defined to measure the vibration delivered to battery object (in this case a cell pouch). Each point is probed by LVDT. 552

4 A. Vibration under Different Tab Configurations In the practical application, the battery tabs are configured into different combinations of tab material or various numbers of tabs. Multiple battery cell objects are grouped and to be joined together by external connectors (or terminals) and advanced to the next level of assembly. (i.e., cell to cartridge/module or module to pack). Factory-tuned optimal energy (800 J in our case), was held fixed through this test, and one aluminum and one copper tab (both 150 µm-thick) are paired or tripled together making six different configurations as listed in TABLE IV. Two LDVT probes are set up and point at the two different vertical levels and data was simultaneously collected during the welding. At each configuration, the probe data pair (at level 1 and 2) was collected multiple times, from which the two extreme outliers have been eliminated and the rest were averaged. TABLE IV. Energy 800 J PEAK-TO-PEAK DISPLACEMENT UNDER SIX DIFFERENT TAB COMBINATIONS. Tab Configuration Peak-to-peak Amplitude (mm) (Horn Anvil) Level 1 Level 2 Al-Cu Cu-Al Al-Al Cu-Cu Al-Al-Cu Cu-Cu-Al Figure 12. shows that, except for Al-Cu pair, most of the peak-to-peak displacement lies between 0.5 mm ~ 0.7 mm and 0.2 mm ~ 0.45mm at level 1 and level 2, respectively. This variation is well within the nominal data variation among the multitude of the data collected exactly under the same condition. Therefore, we conclude that the tab configuration does not affect the change in vibration transmitted to the battery object. Figure 12. Plots based on the measurement data in TABLE IV. ; Peak-topeak displacement (mm) under six different tab combinations, measured at the two different locations (level 1 and 2) on the battery mockup B. Vibration at Different Welding Energies The welding energy is one of the most important welding parameters, which dominantly affect the final welding quality. This time, the battery tab combination is held fixed to Al-Cu (aluminum on the horn side and copper on the anvil side), and four different level of energies were set to 400, 600, 800, 1000 and 1200 J. Similar to the previous section, two LVDT probes are located on the two levels to simultaneously collect data in TABLE V. TABLE V. Tab Configurations (Horn - Anvil) Al-Cu PEAK-TO-PEAK DISPLACEMENT UNDER FOUR DIFFERENT WELDING ENERGY LEVELS Energy Peak-to-peak Amplitude (mm) Level 1 Level J J J J The result shows that the four averaged displacement values ranges from 0.28mm~0.55mm and 0.6~1mm at level 1 and level 2, respectively, which are within the variations of the five measurement at each energy setup. This result shows that the welding energy change does not significantly affect the peak-to-peak vibration displacement. In general, the increase in the welding energy lead to the increase in the welding time, with the peak acceleration level kept approximately the same. The welding control system is required to accordingly compensate for normal variations in the surface conditions of the metals by delivering the specified energy value, which is done by allowing the welding time to adjust to suit the condition of the materials and deliver the desired energy [8]. C. Vibration Mitigation As one of the vibration mitigation strategies, tab-bending is commonly adopted technique in the current battery manufacturing industry. In most cases of welding on a pair of flat tabs (e.g. Al-Cu pair), a wave or ripples permanently deforms tab surface when the welding is finished. The forming of the ripple is due to the excitation of a resonance of the tab causing an instant plastic deformation of the tab surface. Therefore, it has been a common practice that some geometrical features are added in order to stop or keep the wave from propagating beyond the region where the feature is located. As another approach to prevent the undesirable wave or ripples from forming over the surface, the tabs are clamped by some mechanism in order to block the wave propagation. We set up a tab-clamping mechanism and the measured peak-to-peak vibration displacements at two different levels (level 1 and 2) similar to the setup in Figure 11. TABLE VI. shows the resulting data under different clamping conditions. The corresponding plot in Figure 13. clearly shows that the clamping mechanism somewhat reduces the amount of vibration at the tab level (level 1). This can be explained by the fact that the tab, when clamped, becomes structurally stiffer and thus displaces less in response to the welding vibration. The plot also tells us that the amount of air pressure does not noticeably affect much such vibration reduction at the tab level. On the other hand, the air pressure does affect the amount of reduction in displacement at the pouch level (level 2). This intuitively makes sense in that the increased air pressure allows firmer grip on the tab to block the wave more efficiently. However, we would not 553

5 increase the air pressure to gain such a small amount of reduction at the cost of the possibility of damaging the soft surface of the battery tabs or terminals. TABLE VI. PEAK-TO-PEAK DISPLACEMENT DATA COLLECTED WITH TAB-CLAMPING CONDITIONS Tab Configuration Al - Cu (Horn side - Anvil side) Clamping Material Buna-N Rubber, Durometer = 50A Clamping Area (m 2 ) Air Cylinder 3/8" Bore, 1/2" Stroke, 100 psi. max Cylinder Area (m 2 ) 7.13E-05 Welding Energy 800 J Air Supply Clamping Clamping Peak-to-peak Force Pressure displacement (mm) psi N psi Level 1 Level pressure beyond this range is not recommended due to possible damage to the tab. REFERENCES [1] Annual Merit Review - Energy Storage R&D Overview, by David Howell, Manager, Energy Storage R&D Acting Team Lead, Hybrid Electric Systems: Office of Vehicle Technologies, May 19, /energy_storage/es_0_howell.pdf [2] Electric Vehicle Battery Systems: Sandreep Dhameja, Newnes, Oct. 9, 2001, ISBN [3] Batteries: University of Cambridge: Teaching and Learning Package: [4] G-Force: DOE Office of Science: [5] Emerson Industrial Automation: [6] Metal vs. Plastic Ultrasonic Welding by Austin Weber, Assembly Magazine July 2002 [7] Robotic Ultrasonic Welding Six-axis robots offer flexibility for welding plastic and metal partdd, by Austin Weber. Assembly Magazine, [8] L 20 - The Ultrasonic Welder Product Manual, AmTech Inc. [9] IRB 6400RF: ABB Robotics BB3A.aspx [10] Battery Choices and Potential Requirements for Plug-In Hybrids - Plug-In Hybrid Electric Truck Workshop, Los Angeles, CA Feb 2007; Ahmad Pesaran, Ph.D. National Renewable Energy Laboratory [11] ABB Battery Manufacturing Workshop June 15, 16 & 17th, Auburn Hills, MI [12] [Battery Manufacturing and Joining Technology Symposium, September 8, 2010, Troy, MI Figure 13. Peak-to-peak displacement with tab-clamping method VI. SUMMARY AND CONCLUSION The vibration of robotic ultrasonic metal welding was investigated. A robotic battery welding cell was set up and battery tab welding was performed and the vibration data was collected by using accelerometers and LVDT sensors. The data was analyzed and we learned if/how the vibration affected the robot manipulator and battery objects. From the investigations, the following conclusions were drawn: The vibration introduced by the ultrasonic metal welding process is considered non-critical for the lifetime of the robot. The residual vibration on link 6 and 5 can be further reduced by inserting compliant material between the robot and the welder with a minimal design change of the robottool interface. The welding energy or tab configuration does not affect the vibration propagated to the batteries. In addition to the commonly used vibration reduction by tab-bending, tab-clamping with the pressure of 15~35 psi significantly reduces the vibration of the battery. The 554