Industrial robots and work cells built around them have made great strides in automating foundry operations, but with one big exception: Cleaning castings, which remains the foundry's most labor-intensive operations. In theory, robot-based automation is ideal for sawing off gates and risers, deburring surfaces after those cuts, and removing parting-line flash, especially in aluminum foundries.

However, other than for long runs—thousands or tens of thousands of castings—programming the robot takes too long, often as long as two weeks, during which the entire work cell is out of production. For long runs, a two-week setup time is acceptable. But when runs are dozens or hundreds of castings —increasingly the norm—the entire job may not take two weeks.

The problem in automating the cleaning of castings is the way robots are set up for each new task. In this teach mode, the robot arm is manually moved (or “jogged” with its controller) through each of hundreds of points in 3D space, one point at a time. The robot’s controller incorporates the point data into a task-specific program, a sequence of arcs, distances, vectors, starts, accelerations, decelerations and stops that mimic the way a human would do the job.

Because this must be done inside the work cell, i.e. “online,” it has a big drawback. The robot must be taken out of production during the programming. This is true whether the robot holds the casting or the tool.

To solve this programming problem, ABB Automation Technologies / Robotics in Sweden has developed an off-line programming system called RobotStudio. RobotStudio makes it possible to do all the programming off-line, to simulate workstations and to optimize programs without interrupting the production.

ABB solved the problem with a breakthrough called TeachSaver that automatically generates robot programs from strings of 3D coordinate points. The points are gathered with a Stinger II™ portable coordinate measuring machine (PCMM) from ROMER, a Farmington Hills, Michigan, unit of Hexagon AB. After simulation and optimization in TeachSaver the programs are downloaded to the robot in the work cell when it’s available.

Equipped with a Renishaw touch probe, the Stinger II was developed to gather 3D coordinate points for inspection, dimensional measurement and reverse engineering. With a Stinger II incorporated into each TeachSaver system by ABB, the time to program most casting-cleaning jobs can be reduced to less than four hours.

This means that the time needed to program complex tasks off-line has fallen by as much as 90 percent, and marketing managers at ABB Robotics are delighted with the results. With a Stinger II interfaces to TeachSaver, their customers now have a fast and robust way of gathering reliable 3D point data for off-line programming systems.

When programming and performance tests and debugging are done in an office, away from the work cell, production can continue uninterrupted. TeachSaver programs can be cloned and shared whenever multiple cells are being set up to do identical work; some recalibration will probably be needed.

In turn, that opened a huge number of short- and medium-run foundry production tasks to robotics, from dozens to several hundred castings. Within weeks of the system’s final approval, ABB was quoting dozens of applications in foundries around the world.

For ABB, the biggest opportunity lies with cleaning castings in sand and permanent-mold aluminum foundries. With Stinger II and TeachSaver, these hard, dirty, noisy and occasionally hazardous tasks can now be economically robotized.

The business opportunity of this essentially untapped market is huge by any measure. ABB estimates that 300,000 people work under unpleasant conditions in foundry cleaning departments—200,000 in Europe and 100,000 more in North America.

ABB expects robotization of foundry cleaning operations will eventually pay off big for its foundry customers through:

• Fewer worker’s compensation and disability claims as these dirty and dangerous jobs are automated.

• Less difficulty recruiting good workers; like every other manufacturer, foundries must deal with rising labor costs while customers demand lower prices.

• More consistent part quality.

• Lower scrap and reclaim / rework rates.

About 80% of the work of cleaning castings is still done manually, ABB calculates, despite high levels of automation and robotics nearly everywhere else in foundries. To put this new market in perspective, in its three decades in business, fewer than 11,000 ABB robots have been installed in foundries.

To achieve true off-line programming, RobotStudio utilizes ABB’s VirtualRobot Technology, a copy of the robot controller. The linked technologies have long been instrumental in automating long production runs of castings for the automakers, especially for the heaviest parts such as engine blocks, heads and transmission housings. (ABB also offers its own high-level programming language, RAPID, and an Internet browser-based package called WebWare for local and remote monitoring and control of production.)

These advances have allowed programming to be done in the engineering office for many years while production continues rather than in the work cell. But until TeachSaver and the Stinger II PCMMs were joined, there was no quick and easy way to generate process points and there were no foolproof methods for calibrating the locations and orientations of tools and workpieces.

Teaching a robot to do a single complex task like cleaning castings takes up to 100 hours, during which the robot produces nothing and the work cell’s output is zero. “Until now, that was why only long production runs were economically feasible for robotic cleaning,” said Per Astrom, TeachSaver specialist at ABB Robotics headquarters in Vasteras, Sweden.

“By using TeachSaver, the robot and work cell can be kept in production,” he added, noting that off-line programming, by itself, typically boosts uptime in foundry cleaning cells to 80 percent from 60, a one-third increase in output. Even optimizing programs for cycle time can be done with TeachSaver, along with calculating tradeoffs between robot reach and part weight.

“The economic order quantity (EOQ) of castings for cleaning has now been sharply lowered,” Astrom added. “That accounts for the multiplication of the potential size of the market.

“The biggest part of the savings comes from not having to go back and forth to the fixed-in-place coordinate measuring machine (CMM) to fine-tune the robot's program,” he pointed out. “Tuning just one robot program can take several days,” not counting queue time if the CMM is overloaded. This is why ABB insisted on a PCMM to go with TeachSaver,

The interface between the Stinger II controls and RobotStudio was developed in mid-2003 at ABB’s Northeastern U.S. regional application center in Windsor, Connecticut. The Stinger II interface was developed with the ActiveX applications programming interface (API) in TeachSaver. ActiveX makes it very easy to write code modules for external devices, ABB said, and any necessary interpolation or translation of the 3D coordinate points gathered by the PCMM is done in TeachSaver.

Among the reasons why the Stinger II was chosen:

• Accuracy. “Basically the portable CMM arm needs to be just a little bit more accurate than the robot’s repeatability, which is between 0.1 and 0.2 millimeter,” Astrom said. “The Stinger II gives us that and a lot more.”

• Price. “Other arms that could do the job were as much as three times more expensive than the Stinger II and still were not as accurate,” Astrom said.

• Ease of use. Another firm’s coordinate point data was in a hard-to-handle format, ABB engineers said

Automating those some of those 300,000 jobs is already getting under way. For example:

Cleaning turbine blade and vane castings for a large U.S. jet engine manufacturer, an application developed at ABB’s application center in Windsor. This low-pressure, permanent-mold operation was the initial application of the Stinger II and TeachSaver.

Cleaning die castings, a turnkey-systems approach built by Rimrock Corp., Columbus, Ohio, and its Rimrock Automation Inc. unit in New Berlin, Wisconsin. As ABB’s largest systems integrator in North America, Rimrock focuses exclusively on high-pressure die-casting.

A sand-cast cylinder head was programmed for cleaning with over 1,000 points located and entered in four hours. The job was done by long-time ABB systems integrator Laempe GmbH in Germany.

The technical skill required for TeachSaver users is not trivial, Astrom pointed out. Programmers need to know how to work with robots manually or off-line, including the key task of gathering coordinate points with a PCMM.

Programmers must also have experience with speeds, RPMs and traverse rates appropriate to the sizes and types of robots used in foundries. Skill with foundry cleaning tools is also needed to optimize operations for tool life and cleaning quality. “If cleaning is not done correctly, the tools can be destroyed,” said Astrom. “Part of the casting might even be melted.”