Reinventing the wheel - Chrome plater develops automated wheel polishing process
Chrome plater develops automated wheel polishing process
Kuntz Electroplating, a Kitchener, Ontario-based independent OEM chrome-plating supplier of automotive wheels and bumpers, the management watched as hundreds of its workers manually polished its wheels. Because this made it difficult for the company to find, train, and retain the workers it needed, it decided to develop its own automated process for wheel finishing.
The subsequent chrome plating operation requires that the finish on the wheels be of exceptionally high quality, because even the slightest imperfections stand out once the plating is complete.
Next imagine how difficult it is for your company to find, train, and retain hundreds of employees to perform manual finishing. Manual polishing is tedious and monotonous—the average new worker lasts less than a month. Polishing accounts for more than half the entire company's repetitive-strain injuries.
Because of these difficulties, the company decides to try to automate the process. After surveying available technology, the company decides to reinvent the wheel—or at least how it's polished. The company opts to develop its own automated robotic process for wheel finishing.
Collecting the Tools
Kuntz Electroplating Inc., Kitchener, Ontario, found itself in exactly this situation. Kuntz is an independent OEM chrome plating supplier that produces millions of plated automotive wheels and bumpers annually. More than 50 years old, the family-owned company employs more than 1,000 people at peak production and is an approved supplier to such customers as the Big Three and Harley-Davidson. The company had designed and implemented fixed automation for automotive wheel finishing, but fixed automation wouldn't work for the newer wheel styles with deeply contoured spokes. A new approach was needed.
Kuntz contacted FANUC Robotics North America, Rochester Hills, Mich. (www.fanucrobotics.com), looking for an off-the-shelf system. Laxmi Musunur, manager of FANUC's material removal group, had worked with robotic finishing and deburring systems but had never attempted to polish aluminum wheels with such complex geometry. FANUC provided technical support in several areas, including robot selection.
PushCorp Inc., Dallas (www.pushcorp.com), provided the end-of-arm tooling. Its active force control devices allowed the sandpaper cartridges or other polishing media to be pressed against the workpiece with a controlled amount of force that was maintained as the media wore down, even if the casting had dimensional irregularities. It also can compensate for the tool's weight in any direction or position.
Teaching a Robot How to Be Human
Initial test results demonstrated that the compliant tools worked and that the desired finish could be achieved, at least on small test areas. Kuntz installed 20 M-710i robots in 10 cells. The cell design featured two inverted robots mounted on a steel superstructure. The superstructure was situated above a large turntable capable of fixturing 10 wheels.
The first of these cells was commissioned in mid-1999, with the last cell being completed about one year later.
Robot paths for the first three wheel styles were programmed by manual lead-by-teach. In other words, the robot path was created by manually moving the robot in small increments along the desired path using a teach pendant and recording the positions and control parameters at each point. Then the robot program replayed the series of points in order. Each wheel path program took eight to 10 weeks or more to create and perfect.
Another challenge was the learning curve necessary to understand the fine points of the human operator's techniques so they could be reproduced. The goal was for the robots to achieve the same smooth, blended finish as on a manually polished wheel.
Kuntz needed to investigate and develop faster ways of generating robot programs. The company purchased an RODYM-6D Robot Measurement System from Krypton Electronic Engineering, Belgium (www.krypton.be). The system was used to capture and digitize human motion and turn it into a robot program, calibrate the robots, and transfer programs from one cell to another without touching up the programs.
At the same time Kuntz contracted Al Knasinski at Krypton to provide training on how to use the measurement system and develop alternative programming methods.
This combination of equipment and skills eventually led to the development of an offline programming system using a 3-D CAD model of the wheel.
This software was designed to control better such parameters as force head pressure, spindle speed, and travel speed to produce smoother, more consistent paths in less time.
Work began by employing the robot measurement and programming system to develop robot programs for the new model-year wheels. Unfortunately, this effort brought to light a shortcoming in the cell design that wasn't apparent with the original 1998 model-year wheels.
When the project began, the 1998 model-year wheels were in production, so they became the focus of the company's initial efforts. The same style wheels carried over into the 1999 and 2000 model years. But in 2000 the company began developing robot polishing paths for the 2001 model-year wheels.
A fundamental problem with the cell layout then emerged: The 2001 wheels had a significantly different spoke shape, which required the sandpaper cartridge to reach positions at a shallower angle relative to the wheel face than what was needed to polish the earlier wheels. For the new wheels, the robot reach was insufficient in the existing configuration.
After repositioning the wheels, the company discovered other concerns. The new robot path programs required faster motions with abrupt changes in direction, which caused flexing in the robot support superstructure. This led to surface defects in the polishing when one robot's motion would cause the other to vibrate. Also, the new wheel styles required larger robot motions.
These larger motions resulted in singularities in the program. A singularity occurs when a typical six-axis robot reaches certain positions, and two or more joints become aligned so that the robot controller can move either joint to reach the next point on the robot path. Consequently, it can't decide which one to move, and the resulting error causes the robot program to stop or exhibit erratic behavior—it's like trying to get a calculator to divide by zero. After trying to overcome all of these difficulties, the company ultimately abandoned the original cell design and started over.
Reconfiguring the Cell Design
Beginning work in its development lab and then in a two-robot pilot cell, the company rearranged the cells into a new design configuration referred to internally as the Second Generation.
In the new layout, the robots were mounted conventionally: upright on a pedestal. In front of each robot stood two programmable wheel mounting platters, the robot's seventh and eighth auxiliary axes. FANUC helped the company set up an auxiliary axis system to meet this application's requirements, in which the auxiliary axis was the first joint out of another robot.
The electric spindles that powered the polishing media also caused problems for the original robot cells. The original design, even after a problem-free second attempt, was too large and had too little torque to handle the larger polishing media targeted for the Second Generation cells.
In response, PushCorp developed a 15,000-RPM spindle that had enough torque to handle larger polishing media up to 4 inches in diameter, which enabled a full range of polishing media to be used on the robot in addition to the sandpaper cartridges.
As the robot polished the wheel, the wheel was rotated about its own axis, allowing full coordinated motion with the robot. Because the wheel could be repositioned, the robot needed to work only within the dexterous part of its work envelope. The ability to support coordinated motion also reduced the need for the robot to make large motions with sharp directional changes, which reduced wear and tear and cycle time.
While the robot polished a wheel on one auxiliary rotary axis, the operator unloaded and loaded the other unit, permitting continuous operation. The two auxiliary axis units were separated by safety fencing, each with pressure safety mats and a complete set of operator controls.
A system of safety interlocks on the auxiliary rotary axis units and on the robot ensured that the operator could load one unit safely while the robot worked on the other. This was not a trivial consideration, because the operator was loading a "live" unit to which the robot controller still was connected throughout the duration of the operator's exposure.
While trying to get the original cells to perform properly and developing the Second Generation concept, the company also created an offline programming system. The basic concept of offline programming was not new, but it had challenges—specifically, achieving a simulation-created robot path that could be downloaded to a robot without program touchup. Because the wheel polishing programs were complex, this task was challenging, requiring new pieces of software to be developed.
To polish automotive wheels, it's necessary to coordinate numerous complex and interrelated factors:
- The working axis of the compliant tool always must be perpendicular to the surface at the point of contact so that the tool pressure is applied in the right direction toward the surface at all times during the program.
- The motor travel speeds, compliant tool forces, and motor RPM are in a constantly varying balancing act, depending on the local surface geometry. If one factor changes, the others must adjust smoothly and automatically.
- Acceleration and deceleration must be controlled carefully at all times.
- Initial contact and lift-off points must be feathered to prevent polishing discontinuities.
- Wear patterns on the sandpaper cartridges must be controlled for a long, even polishing life.
- The sandpaper cartridge lead- and toe-in angles must be adjusted continually for smooth blending of polishing marks.
Inconsistency of the lug nut hole locations relative to the wheel spokes was another consideration. The wheels were positioned by these holes; but, according to the manufacturer's specifications, the placement of these holes could vary rotationally by ±1 degree. On an 18-in.-diameter rim, this led to an error of almost 1/4 in. near the outer edge of the wheel. This caused problems with robotic polishing. Kuntz subsequently developed a proprietary method for automatically detecting and compensating for this error that didn't require the expense of a vision system.
Also, the offline programming system had to create different versions of any one wheel polishing path quickly. This was necessary to cope with the varying surface quality of the incoming raw cast aluminum wheels. When the surface conditions deteriorated, the robot operator needed to be able to add polishing paths. This meant that the programs had to be able to adapt to changing requirements without having to create or load new programs.
While the company was dealing with all of these issues, it simultaneously formed a new division, Kuntz Logic Systems Inc. (KLS), to develop and market the new offline programming software technology and adapt it to other automotive applications, such as welding, painting, and material handling.
The Future of Automated Polishing
By mid-2001 the company had built a prototype Second Generation pilot cell comprising two robots and four auxiliary axis units. The company had been using its robot lab development cell to test and develop a way to incorporate other types of polishing media in addition to the sandpaper cartridge, such as small flap wheels and brushes.
The first wheel to be programmed for robotic polishing using all these types of media was a wheel for a major automaker's SUV, for which a large, complex surface area had to be polished. According to KLS, the new system produced consistent work and exhibited none of the problems that plagued the first-generation design.
The company also said quality of the output is indistinguishable from that of the best human polisher, but at only one-third to one-half the time in most cases.
In addition, the automatically generated paths are superior in smoothness of motion and overall improvement of wheel quality, said Peter Forth, director of technology and innovation at Kuntz Electroplating.
"Robotic polishing has allowed us to achieve a far more consistent surface quality finish, leading to improved visual electroplated appearance on many of our more complex-shaped aluminum wheels," he said.
A Kuntz programmer currently can develop a polishing path for a new wheel in one to two weeks, including testing and optimization, using all of the tools developed so far. This compares to the eight to 10 weeks or more that it took to generate paths by the lead-to-teach method in the original polishing cells.
However, the company expects that this time can be reduced. In 2001 it began a two-year research and development project at the University of Waterloo, which tests and develops additional refinements to this process. The project includes developing optimum performance parameters for polishing media and new methods to optimize the polishing programs automatically by considering the dynamic behavior of the robot.
After completing an extended trial on the pilot cell, the company now is reinstalling the remaining robots in the new configuration, which now will include fully automated loading and unloading of the cast aluminum wheels. One operator will be able to operate up to 16 robotic polishing cells.
Practical Welding Today
Practical Welding Today was created to fill a void in the industry for hands-on information, real-world applications, and down-to-earth advice for welders. No other welding magazine fills the need for this kind of practical information.