Big Data & Metalcutting
Thomas R. Kurfess knows more than a little about subjects including automotive manufacturing and machining operations. Kurfess is presently professor at the George W. Woodruff School of Mechanical Engineering at the Georgia Institute of Technology. He spent a year as the assistant director for Advanced Manufacturing at the Office of Science and Technology Policy, Executive Office of the President of the United States of America. He’s held the BMW Chair of Manufacturing at Clemson, where he was also the director of the Automotive Engineering Program. And that’s just since 2008. He holds four degrees from MIT, including a Ph.D. in mechanical engineering. He also notes, “I grew up in a machine shop. Back then, it was NC [numerical control].”
So when you want to learn about trends and developments in machining germane to automotive operations, Kurfess is probably the smartest guy in the room.
“You’re always being driven by ‘How can I produce something better? How can I do it at a lower cost, and with shorter lead times?’ You also, on the other side, are getting designs that are (a) more complicated and demanding, so tolerances are tighter and geometries are crazier; you’ve got shapes you would have never thought about making, and (b) the materials are getting nastier. We know the higher temperature you can operate an engine at—a jet engine or an automotive engine—the more efficient you can be,” Kurfess says, adding, “Turns out that the materials that are good at high temperature are harder to machine.”
Be that as it may, there is the need to machine these materials, as well as those that are less “nasty.”
While Kurfess believes that there will continue to be improvements in machinery and tooling—and he says “Machines today are phenomenal”—there is another area where there could be huge advantages to manufacturers.
“The low-hanging fruit,” Kurfess says, “relates to Big Data. There are lots of machine tools out there with lots of sensors—and they’re going to get more sensors on them—generating tons of data.”
And it is the data that can be advantageous.
Kurfess suggests that there are several ways of taking advantage of data, both small and big. For example, he cites the spindle for a bore grinder that runs at high speeds, say 120,000 rpm. The life of spindle bearings is, of course, finite. Say approximately 90 days. While it could be simply just swapping out a spindle every 90 days. But there is a better way, perhaps, which is to mount accelerometers on the spindle to determine its health. Whereas accelerometers were once somewhat costly tech, “Your smart phone has an accelerometer,” he says, adding, with a bit of overstatement, “You could monitor the process with your smart phone.” Kurfess says that getting the information would allow a more timely spindle replacement, one that would help assure that the parts produced are within spec. (“The spindle just doesn’t stop,” he says. The bearings wear, and with the wear can come reduced machining precision, which is certainly not a good thing in bore grinding applications.)
While Kurfess says that improving local operations like that with data is good, bigger opportunities involves something more: “We are now generating huge quantities of data in plants that really aren’t used except for local functions. What if we could take all of those data sets, aggregate them, and make use of them elsewhere?”
This isn’t, he points out, a matter of adding more sensors to machines: “They are already instrumented. We have a good foundation.” Sure, more sensors may be attached and integrated, but Kurfess points out that what’s already there are generating plenty of data.
Kurfess suggests that it is a matter of going beyond utilizing the data from a given plant to optimize the cutting process, the speeds and feeds, to fully aggregating the information from multiple plants, be they operated by OEMs, Tier One suppliers or are the proverbial “Mom and Pop” shop. By taking all of this data, it would allow companies to determine precisely what the best practices are.
One question arises in relation to this, which is simply why someone would allow their machining data to go to a third party. Kurfess admits that there needs to be a business case related to why someone would readily give up this information, but he cites two examples of how this is happening right now.
He says that in many small shops, their photocopying machines are online, such that the copier company knows when the toner cartridge needs to be changed. It is automatically ordered and sent to the shop, which is a convenience for the office manager. “What if it was tracking tools that were being used in machines? You could tell when the tools were wearing out, and ordering replacements when needed.”
“If you provide a service that they use and is of value to them, why would they care if you know what their feeds and speeds are?” The other example is one that is as close as your smart phone. Seemingly everyone has Google Maps on their phones. Kurfess says that most people probably don’t pay much attention to the permissions that are being granted to Google when installing the app. Google is granted access to some of the data about the phone. “It knows you are going 60 mph along a certain road so they are able to put the green line on the map,” he says.
Again, you’re giving up data in order to get use. So he thinks that the aggregation and analysis of machining data on a massive scale would permit significant gains in productivity.
If that’s improving machining opera-tions at the macro level, there are also changes that Kurfess thinks will be occurring at the on-the-ground level, as well. “One of the things we haven’t been particularly good at,” he says, “is using five-axis machines for five-axis machining rather than three-axis machining.” The reason: “Five-axis machining is a hard concept to deal with.” But, he says that CAD/CAM system improvements should make it easier for people to make the most of their equipment.
Then there are the seven- and nine-axis machines, those with secondary spindles. Kurfess says that in many instances, these machines are purchased for special-purpose, high-volume parts production. At Georgia Tech, Kurfess and his colleagues are doing work on high-performance computing that will facilitate using multi-axis machines to run a variety of parts. This can lead to a whole realm of opportunities in terms of plant capacity: “If I have a shop with 15 of these machines, how do I, in real time, move things around?” he says, explaining that it would be possible to quickly react to a machine going down or new, high-demand orders coming in with less turbulence in the scheduling.
And things do change, even in high-production operations. Back when Kurfess was at Clemson, he and some of his colleagues and students regularly worked with BMW personnel at its plant in Spartanburg, SC. “During my interactions with BMW, they were getting changes on a daily basis in terms of production,” he says.
Having a handle on data certainly is advantageous, regardless of what is being produced.
MTConnect: Easier Integration of Information
It probably isn’t going too far out on a limb to say that no machining operation has one kind of machine. Even if every machine tool is precisely the same, chances are there is another type of equipment being used in the facility, such as a coordinate measuring machine.
An existing issue is that there is a lack of easy interoperability and connectivity of these various machines. We think nothing of plugging a new printer into a computer and having it do its job with little fuss (perhaps needing to download a driver). This would certainly not be the case in terms of equipment in a factory.
Enter MTConnect (mtconnect.org), an open, royalty-free standard that’s being embraced by a wide array of equipment manufacturers and other interested parties that is designed to facilitate interoperability between controls, devices and software applications.
What’s notable about MTConnect is that it isn’t based on some specially developed language for
machines, but uses XML (Extensible Markup Language) and the Internet Protocol, so it is leveraging technology that has been developed for the mass market.
Which is to say that MTConnect can help provide information integration throughout one’s own facility in a comparatively simple, straightforward manner . . . as well as the means by which there could be the aggregation of data from a number of facilities that Prof. Thomas Kurfess of Georgia
Tech believes can lead to the optimization of production operations.
As OEMs and suppliers seek lightweight solutions to meet higher fuel economy standards through multi-material structures, conventional welding techniques are beginning to give way to new solid-state joining methods better suited for creating strong bonds between dissimilar metals.
While no single piece of equipment is ideal for everything, those looking for a better way to perform production welding ought to consider these solid-state laser systems for speed, efficiency, and effectiveness.
By John R. (Jay) DoubmanIn concept, not much has changed in double disc grinding since the invention of the machine in the 1890s.