Robotic Additive Manufacturing of Large Steam Turbine Blades

Ed Brown

Tech Briefs: What got you started on this project?

Michael Kirka: The project was originally birthed out of our industrial partner, Siemens Corporation (Now Siemens Technology), because of a need to repair large-scale steam turbine blades and other components that even pre-COVID had long lead times. They asked us to partner with them as part of a larger project. Our portion was focused on the large-scale additive manufacturing.

Tech Briefs: Is that something you were doing before this particular project?

Kirka: Yes, the Manufacturing Demonstration Facility at Oak Ridge has been working in large-scale manufacturing, going back to about 2015. We have been working with the welding machine manufacturer Lincoln Electric on developing the process and industrializing it.

Tech Briefs: Could you describe your process?

Kirka: It’s large-scale manufacturing via arc welding. We use a standard electric arc welder on a robotic arm that can move in three-dimensional space. The welder feeds a wire, melts it, and the robot moves it around to deposit the molten metal in a controlled manner.

The concept was extended from large-scale polymer printing. Oak Ridge had developed large-scale polymer extrusions going back to around 2013, and this was a similar idea. But, of course, metals have a lot higher melting temperature.

Although electric arc welding is an old technology, to apply it to this application, we had to do a lot of research into how to control the robotic arm — you have to deal with the material science as well as the robotics.

Tech Briefs: It seems to me that this is very different from standard 3D printing. It must be challenging to precisely control the robotic arm.

Kirka: Yes, with the most common printing techniques, you are looking at printing parts that are 20 pounds up to several thousand pounds. Standard metal printing does five- to maybe 50- or 100-pound parts, using powder bed techniques. We’ve been looking at how to scale up to large objects — the steam turbine blade we printed, for example, is about 3 feet tall. Generally, you can’t print something like that in a powder bed system. Although there are some systems that can print something the size of a large turbine blade, there are few that have the required dimensional capability and efficient build rates.

Tech Briefs: Is there an upper limit to your process in size and weight?

Kirka: In theory, you can scale the large metal process indefinitely from a robotic standpoint, but there probably is a theoretical limit from a thermal mass standpoint. If you attempted to print a 10,000-pound part, as opposed to a 1,000-pound part, the amount of heat retained during the printing and the stresses that accumulate will be very different as you scale.

ORNL researchers here have been working on scaling that process to go from deposition of material at 20 to 25 pounds an hour to 100 pounds per hour by introducing multiple robots — that system is called Medusa.

Tech Briefs: What rate does your system print at?

Kirka: Probably around 10 to 20 pounds per hour. There's a little bit of variability because it's also dependent on the cross section that you're filling in with material. So, if you were just printing straight back and forth, it's a different deposition rate than for trying to fill in, say, a turbine blade pattern that's constantly changing, going from thick to thin.

Tech Briefs: I read that you use scanning to check the piece as you're going along.

Kirka: We've used different types of scanning: blue light scanning to get an idea of the geometry as it was growing, and after the print is completed, we use computed tomography (CT) to look internally into the structure for defects.

You can use the machine vision and other techniques to pick up surface qualities that are unique to a location and watch to see if they're moving as you're printing to get an idea of distortions.

But also on the robot, when we go to start a new layer, there's a sensor that will touch the very top of the part to locate it, because when you print, you're at very high temperatures, close to the melting point, as you're depositing material. But then it cools down and shrinks. So, for every layer you have to find the top of the part because it's not where it was when you stopped the process, and then relocate the robot accordingly.

Tech Briefs: What about the problem of having contoured surfaces so there are no perpendicular edges to use as locating features?

Kirka: In general, that’s a challenge with any part that has something like a turbine-blade-type geometry — they all have a taper or curvature. So, we call what we print, the preform. We then have to identify where the part is within that preform to be able to machine it. Even if a turbine blade is made by casting or forging, every surface has to be machined because it's a precision object.

Tech Briefs: Do you do other non-destructive testing in addition to the CT scan to qualify the structure?

Kirka: We also do blue light scanning. You put fiducial markings on the object and the blue light scanner looks for those markings and reconstructs an exact image of the 3D solid from it. You could consider it almost like a CT scan, but you're not peering inside the material, you're only looking at the surface.

Tech Briefs: Have you made actual samples of turbine blades?

Kirka: Yes, we've printed a number of blades, and also taken mechanical samples out of the material to validate the results, to see how our material performs at higher temperatures and in tests for tensile strength, fatigue, and creep.

Tech Briefs: What are your next steps?

Kirka: This project was originally looking at ways to repair blades. Although we're looking for funding for new projects in that area, we’re also looking at printing larger components of steam turbines. We’d like to apply our technology to other areas of industrial needs in the energy sector that have long lead times associated with them — six months to two years. If you can print a rotating component and scale it to even larger stationary components, it gives you more faith in these new processing techniques.

Tech Briefs: Can you give me a couple of examples of other possible applications?

Kirka: One application is for what are called ring segments in steam and gas turbines. They are giant forgings that go around the turbine blades, circular rings that weigh a couple of hundred pounds.

Other things that are starting to be looked at are replacements for cast and forged components in the nuclear industry. There are, for example, giant castings of valve bodies that probably weigh 2,000 pounds and take a year or two to obtain. So, perhaps we can start replacing those with something that might start out being made by wire arc additive manufacturing and then adding to it. Or perhaps we can start with a smaller casting and add to that.

Tech Briefs: Are these turbine blades ready to be put into actual commercial use?

Kirka: Some of the turbine blades — there have been different sizes manufactured in this project — have been tested at this point by Siemens, starting with the smaller ones. As far as full commercial industrialization, there's a bit more effort to get there on the certification and qualification side — this was just a demonstration project. The industry is focusing now on how to certify and qualify these processes so that you don't have a turbine that goes down and because you can't get components for two years, it's not generating power and it's losing money for the owners.

Our role in this is to demonstrate how you can manufacture the blades with wire arc additive manufacturing. It’s always most difficult to figure out the processing and all the steps necessary for the prototypes, the demonstration pieces. Having done that, we hand it off to our industrial partners.

---