Robotic Grinding in a Steel Job Shop

Nic Tarzwell

Hiring has become a huge challenge to attract new workers and retain them, and it’s a topic of much discussion. Eagle Alloy is also impacted by its location in Muskegon, Michigan, which is a heavy industrial area with a lot manufacturing companies that pay very well; the competition in this area for these workers is even more challenging. Anticipating the hiring challenge would continue, Eagle Alloy began its robot journey in 2016 to move ahead of the curve when it comes to extending the workforce.

Eagle Alloy is part of Eagle Group. Its sister companies include Eagle Precision—a lost wax investment facility, which is physically attached to Eagle Alloy; Eagle CNC Technologies—the sister machine shop approximately 20 minutes north of Eagle Alloy; and Eagle Aluminum—a smaller green sand molding facility located in downtown Muskegon.

Eagle Alloy is a job shop steel foundry that services about 20 different industries. The average casting it produces is comparable to a 20-lb. valve running 1,000 pieces a year. The metalcasting facility uses lean practices to target six- to eight-week deliveries, and prior to robotic grinding, its manhours-per-net-ton was between 55 and 60.

When Eagle Alloy first began exploring robotic solutions on the market, the foundry discovered that most automation companies focused on high volume. The available options had very high initial cost yet were slower than manual production—Eagle Alloy’s part tolerances were not ideal for robotic programming, and surface finish and flaws would affect part locating.

Ultimately, the company decided to create its own robotic grinding cell that would fit its facility.

The first initial concept was to incorporate two robots, with one performing the plasma cutting and heavy grinding, and a second, smaller robot addressing the parting line, core scar, and poppers. One conveyor removed the cut-off gating, and two shuttles brought the parts into the cell.

The thought was that this robot cell potentially could turn a manual, three-person cell into a one- or two-person cell with a robot that matched or exceeded the output of the three-person cell.

The cell was first built in a spare warehouse for testing out the concept. The robot used was an ABB M97 purchased fully refurbished for $15,000 in 2016. The fencing around the cell was about $5,000 and the tables and fixturing cost another $5,000. The plasma cutter was a used unit that Eagle Alloy already had. Initial trials were conducted by the foundry’s engineering department and an outside programmer who was local but not familiar with the foundry industry. 

Once the engineers were confident in the working robotic cell design in the spare warehouse, they disassembled and easily re-assembled the cell for production trials to begin in the production environment.

As more trials were conducted, Eagle Alloy began to notice some deflection caused by excess parting lines being too much for the robot and pencil grinder to address. As further trials pushed this robot for more speed and output, the foundry ran into this issue more and more.

At this time, Eagle Alloy had also realized its robot cell wishes had exceeded the outsourced robot programmers’ ability. So, a new robot programmer was brought on board. The foundry also switched out the belt grinder for a standard 30-in. snag grinder. Due to the high cost of the belts and how often they wore out— which would shut down the cell for a changeover—Eagle Alloy determined the cost per square inch of a snag grinder and uptime to be much more desired.

At this stage of the first robotic cell, Eagle Alloy needed to find a way to take dimensional measurements of the snag wheel diameter during full-speed operation. The new programmer retrofitted a reflective laser to dimensionally inspect the ever-changing outside diameter of the snag wheel. The current dimension would automatically inform the program to adjust the starting point accordingly to ensure each casting was ground correctly. Trials were performed to determine if the wheel needed to be re-inspected after grinding each part or if the inspections could be spaced out every three or five parts with an auto offset calculation to cover the gap between inspections.

Eagle Alloy’s first robotic cell concept was not perfect, of course. The foundry had trouble with inconsistent plasma cutting and could not get a consistent cut start and stop point so the gating would fall where it was supposed to onto the conveyor. This was most often affected by varying degrees of parting line fin around the gate contacts.

At this point, Eagle Alloy was ready to add a second robot cell—this time with a revamped concept that had a smaller footprint and used just one robot. In this cell layout, the engineers decided to put the torch back in the main operator’s hand. The parts would still enter on the two shuttles and exit on the one down chute, and the burden of the grinding would be on the snag wheel.

After finding success with the first part in the version 2.0 cell, Eagle Alloy began programming additional part numbers. The robotic cell was proving to hit the goal of meeting or exceeding the manual operation’s productivity. For example, a three-man cell conducting torch, fin, and snag grinding on a 4-lb. carbon steel was able to complete 65 pieces per hour.

The single-person robot cell completes 86 per hour. A 20-lb. carbon steel part requiring torch, arc, and pencil grinding was finished at a rate of 45 per hour in a three-man cell versus 41 per hour at first then 56 per hour after some improvements. A 6-lb. carbon and stainless steel casting had a production rate of 47 per hour by the three-man cell and 63 per hour by the robotic cell.

Consumables in the manual cell, which include torch, arc, and pencil grinding, cost less than the consumables in the torch/robotic snag grinding cell, but they are offset by the larger labor savings. All seven initial jobs ran at least 20% faster in the version 2.0 robotic cell versus the manual cell. In addition, grinding consumables (wheels and dressers) are easily calculated and consistent.

Now, customers whose parts were in those initial trials are requesting robotic grinding as standard due to improved consistency and final surface finish.

As successful as version 2.0 was, Eagle Alloy still saw room for improvement. (1) They wanted to address operator-to-operator contact height variations from inconsistent torch processing. The inconsistency led to more rework, and there was no input to tell the robot whether the contact height was too high. (2) The robot cells were built on a 1-ft. raised platform to avoid vibrations, but it created a trip hazard and an unnecessary step.

When it was time to add a third robotic cell, Eagle Alloy built it flat on the ground without the raised platform and added dimensional measurements of the initial contact height prior to grinding and the final contact height after grinding.

The initial dimensional check takes about two seconds, and the snag wheel inspection takes about four seconds. The initial inspection tells the robot the starting height, in the event a new in-training employee sends a very high contact inside the cell. The final inspection checks to see if the part meets the acceptable tolerance. Based on the measurement, if the parts pass, they are dropped on the down chute. If one of the parts or both parts fail, they go into a snag grinding rework cycle until the dimensional height of the contact is achieved.

By 2019, Eagle Alloy felt comfortable enough with the robots to try to do its programming in-house. The outside programmers had little understanding of foundry needs and often took longer than desired to program a new part.

One shop floor employee and one engineering intern with an interest in robotics were selected to train for this programming role. Eagle Alloy took advantage of ABB University’s training classes and partnered with a local community college for specific training classes. It also used a training grant from the state of Michigan.

The in-house programming was a gamechanger for Eagle Alloy. Where it would take an outsourced programmer two weeks to develop a new part program, the foundry is now only two days away from grinding a new part number.

ABB also has Robot Studio programming that can be done offline to avoid stopping work in the cell. This has allowed the foundry programmer to quickly construct robotic programming code using CAD built-in Solid Edge and create it for multiple robotic cells.

When the third robotic cell concept was installed at Eagle Alloy, the foundry also upgraded the other robots, so all three cells are now using the ABB M2004 IRC5 platform. This enables the robots to handle the new vision system with ease. Each upgraded robot cost less than $20,000 fully refurbished with warranty.

Currently, Eagle Alloy has more than 32 part numbers programmed on its first three robotic cells. The majority of these parts are taught on multiple robot cells to avoid delays. The parts are chosen first based on volumes, and second based on issues surrounding the part—the pieces the operators hate to manually grind. Setup times vary from 10–25 minutes with one operator.

Eagle Alloy estimates the robotic cells have reduced the manhours per net ton from 55-60 to around 40-48. The labor savings amounts to about 15,000 hours per year, or approximately seven full-time employees who can now fill labor gaps elsewhere in the facility.

A fourth robotic cell was just completed in July. While the first three cells were set up with standard ventilation for work on carbon steel only, the newest cell has a hex booth for processing both stainless and carbon steel alloys.

The average weight processed per hour on the robot cells is about 35% higher than the manual cells.

The robots can accommodate high and low gate contacts, which has been beneficial when training new employees. And, robotic grinding removes the operator away from the large snag wheel—avoiding potential slips or nick injuries while grinding.

Based on the success of the robotic cells, Eagle Alloy continues to work to improve robotic cycles to remove wasted motion, improve dimensional control of the parts existing in the cells, and revamp fixtures to move from one-on to two-on for grinding multiple castings. This metalcaster is constantly increasing its list of programmed part numbers across all four cells now and is building in redundant programming so the same part can run across multiple cells.

A fifth robotic cell is already being visualized. This cell would handle larger parts that enter on a table and then a robot would bring a grinder to the part. Eagle Alloy is also exploring adding A/C or cooling in the summer (right now they have heated air in the winter).

While automation may at first bring to mind high-volume production, at Eagle Alloy, robots have proven to be just as useful in a job shop—improving the safety of employees while extending the existing workforce to increase productivity.