Aerospace manufacturing

First appeared in Manufacturing and Engineering

“If you’re not measuring, you’re guessing”

Speaking of mantras, that’s one of Dr. Scott Smith’s. He’s the group leader for intelligent machine tools at the Oak Ridge National Laboratory (ORNL), Knoxville, Tennessee. ORNL researches ways to improve US manufacturing, and partners with commercial firms to bring these developments to market. While some of these ideas will take years to mature and may require entirely new machines, Smith also seeks ways to help companies today with their existing machines. And while his measuring mantra might appear to apply to QC, he brought it up in reference to improving productivity. Smith said most shops in the US get significantly lower material removal rates than they otherwise could “because they don’t really know what their machines can do. And that’s because you have to consider the tool, the tool holder, the spindle, and the machine tool in combination to determine the performance capabilities.”
Smith observed that “much of our infrastructure for the use of CNC equipment is focused on the geometry. The tool is a cylinder and the workpiece is a prism. And as long as the cylinder intersects the prism in all the right places, then you get the part that you’re meant to create. It’s a purely geometry driven calculation.” But, he explained, machining depends on a number of physical factors, including torque, power, and stiffness. “We write part programs based on geometry, and then we try them out and realize that there are problems that are connected to these other physical phenomena. So, we change the part program, try it again, and after a few cycles we finally get a part program that works. Then everybody says ‘Thank God it worked. That’s the part program. Everybody stop.’ This is particularly problematic for the aerospace industry because the volumes are lower, and you really don’t have time to fiddle around and try to figure out a better process plan. You’re happy enough that you got one that worked.”
The good news is that ORNL partnered with MSC to deliver a solution that “can improve the material removal rate of existing equipment, on average, by a factor of three just by correctly using cutting conditions based on measurements instead of a guess,” said Smith. It’s called MillMax, and simply put, it turns the formerly expensive and difficult tap test into an easy and “inexpensive package you can carry around in your pocket,” explained Smith. The kit includes software that runs on any standard Windows PC, an instrumented hammer, and an accelerometer you affix to the tool. The user taps the end of the tool with the hammer, and the PC shows the vibration characteristics of the tool in a clear graphic that indicates which speeds, at what axial and radial depths of cut, will deliver a high material removal rate without chatter.
If, for example, your part requires 10 different tools, you’d take the measurement on all 10 tools in their holders, in the spindle. Then you’d use the results to program the part, whether on the control or your CAD CAM package. It’s important to treat each setup as unique, Smith explained, because they each behave differently. For example, although Smith acknowledged that today’s NC programming aids allow you to set an approximate tool length and then automatically correct the geometry for different lengths, the length of the tool has a significant effect on the stiffness. “And stiffness is connected to the vibration characteristics.”

“There are many more ways to be wrong than there are to be right”

That’s another Smith quote, which he related to guessing about the axial and radial depths of cut. He explained that although aids like modern CAD CAM and adaptive control are very helpful, they typically try to keep the material removal rate constant. “And that presumes that the material removal rate conditions that you selected are all acceptable. And that’s not necessarily correct. If you’re guessing about the axial and radial depths of cut, then very often you are wrong.”
Scott Lewis, Sandvik Coromant’s aerospace industry specialist for the U.S., ruefully agreed that he often sees programming mistakes in all tiers, including among experienced customers. And Bill Durow, manager of Sandvik Coromant’s global engineering project office for aerospace, space and defense in Mebane, North Carolina, added that some customers choose “standardization over optimization.” In other words, they are reluctant to adjust for different materials or to optimize individual part features. “They’re afraid that if two inserts are very similar, someone might take the wrong insert for a job, put it in the wrong cutter, and have a catastrophe on a component.” “Or the other challenge is, they just don’t have the resource to make the change.” So instead they’ll opt for a “good enough” solution that gets them by.
Lewis said the problem is often as simple as using the wrong grade insert. “Or choosing the right cutting edge for what they’re trying to do with that tool. Or they try to use one insert for multiple operations and techniques. We see that a lot as well. It’s just not the right tool matched up to the operation.” Perhaps more surprising, he observed that it’s very common for aerospace customers to be running nickel alloys with a 5-6% coolant concentration which is more than half from where it should be. “That drastically affects the tool life, or the consistency of tool life. They might be scratching their heads, wondering why, and it’s just a simple adjustment to the coolant concentration. For nickel-based materials, you really should be up in the 10-12% range.” He added that if your machines use individual coolant tanks, checking concentrations is a manual job. But it needs to be done “every day, if not a couple times a day, especially when it’s hot outside. Those concentrations can fluctuate quite a bit.”

New tools, and new techniques for changing them

Both Lewis and Durow reported a higher demand for specials or customized tools in the aerospace industry. On the other hand, Sandvik Coromant has a huge standard portfolio and is regularly adding specialized solutions to that portfolio. One of their new grades is called S205, which Lewis described as a “high-speed carbide grade optimized for nickel-based materials.” The S205 grade uses a second generation Inveio CVD coating technology, which we reported on in May. Sandvik Coromant says it “allows for increased cutting speed by 30–50% versus its predecessor S05F and competitors.” They report speeds of 164-262 SFM for light roughing and 262-425 SFM for finishing operations in heat resistant superalloys.
Lewis also pointed to their SL70 system, “a serration lock 70-millimeter coupling” that makes it possible to attach a variety of insert holders to standard CAPTO tool holders. Sandvik Coromant offers a variety of these insert holders (which they call blades) “with different angles and reach configurations to get into difficult to access features like pocketing and undercuts, or to switch from profiling to turning, to face grooving and parting off,” explained Lewis. “So, we have a greater standard assortment that customers can build on, off the shelf.”
Jeff Wallace, the general manager of national engineering for DMG MORI USA, Hoffman Estates, Illinois, said the aerospace industry has fully embraced automation in the last few years, and his firm has introduced interesting new features to help. At the top of the list is a new technology cycle called “automatic sister tool change and reposition.” (DMG MORI’s Technology Cycles are canned routines that compliment a user’s CAM software, or otherwise make the machine more productive.) Wallace pointed out that many aerospace parts have run times that reach into tens of hours, making them difficult to automate because tools could break down or fail over the course of the job. The new technology doesn’t just stop the job when a tool reaches the end of its life, or exceeds the desired load, or there’s a vibration alarm, etc. That’s the easy part, said Wallace. The technology cycle also gets a sister tool, carefully moves it back to where the cut was interrupted, and then resumes the job.
“Traditionally, when you get an alarm, the machine will stop and sit there waiting for someone to come over and figure out what was wrong with it,” Wallace explained. “With this new system, the machine knows what it did and where it was.” And when it brings the sister tool to resume the cut, it automatically calculates a small offset to ensure that the surface is cleaned up. As Wallace put it, “that’s where some of the magic lies. And this feature can apply to almost any industry that has high value parts, or high run times, that they want to automate. We’re seeing a big push in the automation across these OEMs, and even the second and first tiers.”

Chasing worms around the cube

An increasing demand for part accuracies of 2 tenths or better is adding to the challenge of aerospace manufacturing, reported Cody Berg, national engineering machining manager at Methods Machine Tools, Sudbury, Massachusetts. “If you keep hitting that top-notch tier of suppliers, they’re asking you to go above and beyond,” as he put it. When the parts are titanium, Hastelloy, or Inconel, such requirements often lead to grinding, he added. But they’ve been pushing their Yasda machining centers into such applications with great success. One caution, when you have to rough and finish materials ranging from 58 to 62 Rockwell, you can’t count on the machine to solve every problem. You need a skilled operator, and “a lot of the skilled guys are disappearing in the industry.”
Automatic calibration routines are one technique for maintaining tight tolerances over a run. Berg said Yasda uses an artifact that’s serialized to each machine and a kinematic calibration routine written for that artifact. The artifact has flat and circular ground surfaces that the machine probes during the routine. “It’ll pick up the top, so it knows where Z is, and then pick up the OD, rotate C ten degrees, pick it up and kick B 90 degrees, B 45, and so forth…” The operator can tell the machine to calibrate itself as frequently as desired, even, for example, telling it to calibrate itself twice before the next pallet change. Berg recounted a recent case in which the customer wanted them to hold 2 tenths on thickness, two ID holes, and an OD hole. With only one calibration beforehand, the Yasda cut 480 parts unmanned over two continuous weeks, and made only 3 bad parts, which it labeled as such. In another case, Berg needed to cut small slots over a 12-inch-long part, holding 5 tenths for location. “We were maxing out our stock height and the tool had to be very long. Along with that, the part was hardened to 58 Rockwell, so we were putting a lot of tension on B and C.” Berg noticed the accuracy drifting slowly throughout the week. Not reaching 5 tenths but getting close. So, he decided to simply calibrate the machine every morning and held within 2 tenths, thereafter, as confirmed on a CMM.
Berg also mentioned the importance of building a near perfectly square machine in the first place as a prerequisite for achieving the tightest tolerances, particularly when you consider volumetric effects in larger envelopes. Wallace echoed the point, along with the need for thermal stability. “How do you chase microns around a one-meter cube machine?” he asked. One method is DMG MORI’s own self-calibration routine (3D QuickSet). Another is mechanical. “Our duoBLOCK machines are completely thermally controlled. We’ve got coolant running throughout all the castings and all the components, including the table, the columns, and the beds. We strive for thermal stability. Everyone is chasing the worm.”

Future advances

Smith applauded these efforts, while offering ideas ORNL is working on that promise to make things even better. First, they’ve demonstrated the viability of using a large additive manufacturing (AM) polymer printer to quickly build a machine mold at low cost. (See “Power Parts: Making Parts for Making Electricity” in our December issue.) In addition to enabling machine builders to create customized concrete machine bases at low cost, this approach also permits the inclusion of many more coolant channels throughout the casting, plus vibration and temperature sensors in any desirable location.
Smith offered another method for combating the constant temperature fluctuations that occur within a machine tool: Making adjustments based on markers previously added to the part. Smith explained that if you try to make corrections based on temperature data, you not only need to make a lot of measurements, you also need a thermal model of the machine that accurately predicts the corrections you should make. It short, “you have to know a lot of things in order for that to work.” But if you add markers, or fiducials, to the workpiece, you can measure their position at a known temperature and then “find those same fiducials out on the machine. You can correct the program to allow for the fact that the machine and the part have different dimensions now than they did earlier… And if you use the fiducials, you don’t even know what caused the machine errors. You just know that in the current setup you find those fiducials in these positions, and you know that they were supposed to be in those positions, and you make a correction for the difference between them.”
This last factor means that although the idea was originally intended to correct for temperature fluctuations, it’s also a way to handle the long-range accuracy of the machine. What’s more, it can help do this for very large parts, with further exciting implications. As Smith explained, “typically, if we talk about the accuracy of a machine tool, we say it holds one part in a number of parts. For example, one part in 10 to the fifth. That means that as the part gets bigger, my ability to place the tool accurately goes down. But every time that I can find fiducials, I have the opportunity to reset the start of the position. It acts like a datum. I can reset to zero every time I find a fiducial or a group of fiducials, improving the long-range accuracy of really big machines.”
It gets even better. Smith thinks it’s possible that this concept will allow small machines to make very large parts. And that means many more shops can compete for these jobs, which also lowers the cost for the OEMs. Smith pointed out that many aerospace parts, like wing ribs, “are long in one dimension, but small in the other two. Right now, they are made on machine tools that are bigger than the entire part. But I could make them on a machine that has a throat big enough to pass the part through. I can machine it in segments. To do that, I have to be able to understand where am I on the part, so that I can accurately produce the features that are in the range of the machine at this moment, and then I’ll move. The fiducials allow a way for this to be done. If I can find the fiducials, I know where I am on the part. I can adjust the part program to make the features that are within range of the machine at this moment, and then I can move the part a little farther through.”
One of the machines ORNL built using a 3D printed mold for casting the base is in such a “pass through” configuration. So, look for proof of concept soon!

Sandvik Coromant
Part of global industrial engineering group Sandvik, Sandvik Coromant is at the forefront of manufacturing tools, machining solutions and knowledge that drive industry standards and innovations demanded by the metalworking industry now and into the next industrial era. Educational support, extensive R&D investment and strong customer partnerships ensure the development of machining technologies that change, lead and drive the future of manufacturing. Sandvik Coromant owns over 1700 patents worldwide, employs over 7,700 staff, and is represented in 150 countries.

For more information visit www.sandvik.coromant.com

Share with us on social media: Facebook | YouTube | Twitter | LinkedIn