Hybrid Manufacturing: Combining DED Additive and Subtractive Workflows Effectively

JON CALIGURI: All right, we'll go ahead and get started. Thank you, everybody, for joining us today. This is the session on hybrid manufacturing. We'll go over an agenda. And I'll make some introductions, and we'll get straight away.

So there's four learning objectives. This late in AU, probably not your first industry talk, so I'm going to assume you guys know the drill here. Number one-- we're going to understand metal additive, what DED is, and what hybrid manufacturing means. If you've been to the keynote, which I hope you have, you heard a lot about-- well, actually, all three of those. And we want to go over some terminology and just establish a base level of understanding of metal AM.

Then we'll discuss the main DED technology platforms. So once we define what DED is, we'll discuss the different platforms that are available today on the market. We'll discuss the three most common applications, or use cases, for this technology today. And lastly, we will dive into a case study of how our Autodesk advanced manufacturing consulting team applied DED and hybrid technologies with a blisk case study. And at the end, we'll leave some time for questions.

So I'm Jon. I'm from design software. We're an American Autodesk digital manufacturing reseller. We previously were a Delcam reseller, so the power and spec software that you may have just heard on the big stage-- Powermail, FeatureCAM, these things, is really are our background.

JAMES DONNELLY: And I'm James. I work within the advanced consulting department within Autodesk. My experience is subtractive manufacturing, mainly, so adaptive processes, milling of blisks, and such. I'll talk about some of the projects that we do in Advanced Consulting a bit later on, after Jon's gone through the hybrid and his part of the presentation. OK.

JAMES DONNELLY: So straight into objective 1-- so understanding metal AM, or metal additive manufacturing, DED, and hybrid manufacturing. So metal AM, or metal additive manufacturing, is the fastest growing segment of the additive manufacturing market.

So the additive manufacturing market-- really, that term-- well, it's not new. The term is fairly new. It was just called rapid prototyping, or RP, a lot in America for a long time. 3D printing, additive manufacturing-- these all refer to the same processes. 80% growth in metal AM systems sold last year compared to 2016-- massive amount of growth. It is an emerging technology. And hybrid manufacturing really just refers to combining additive and subtractive manufacturing processes together. It can be done in a number of different ways.

So just to establish a little bit of a timeline of when most people would have became aware of metal 3D printing, or metal additive manufacturing-- really, this emerged in the 1990s. So EOS is probably the largest player in powder bed. Metal AM today. 1994 is when they introduced their first prototype machine. Optomec follows in 1998 with a technology called LMD, or laser metal deposition being introduced in 2002 commercially. They can be seen as the first DED.

So historically, these systems, when they hit the market-- and still today-- the vast majority of metal additive manufacturing is based on what we call powder bed systems. So these technologies have been around. Of metal AM, these are the most mature. They also support the most complex geometries.

So when we were looking at those generative design studies, you have the most degrees of freedom in powder bad metal AM. So the least constraints using this technology. So you'll hear things like SLM, SLS, DMLS-- these are all different metal powder bed centering processes for the most part.

So this is very busy slide. I'm not going to read all this. This is really as a reference. But there are seven or eight families of additive manufacturing. So Vat photopolymerization, powder bed fusion, binder jetting, material jetting-- these are in this slide, and the following is in the handout as a reference for you guys-- sheet lamination, material extrusion-- material extrusion is probably what most of us are more familiar with, like MakerBot desktop-style FDM printers.

So lastly, what we're going to talk about today-- so DED, or directed energy deposition, and the concept of hybrid manufacturing. So directed energy deposition refers to an additive manufacturing process by which we combine either a metal powder or wire feedstock with a thermal energy source. So the thermal energy source can vary. The feedstock can vary. And we're going to go over some of those different platforms here in just a little bit.

Objective 2 is just that. We want to understand, really, the four main categories of technology platforms that we would call DED today. So electron beam and wire-- so this uses an electron beam as the source, the energy source, that we were discussing. I don't know if I have to hit to get my-- my gifs aren't animating. Sorry about that.

So either way, electron beam is a unique system. This is where we find the highest deposition rate, so up to 40 pounds an hour of titanium being deposited. It's unique in that it requires a near perfect vacuum to achieve the temperatures to diffuse the feedstock. This technology can take advantage of using multiple materials on the same part. And typically, this technology is found creating large parts, so maybe above eight inches.

Laser and powder, sometimes called laser metal deposition, is on the opposite side of the scale. So we're not depositing near as much material as quickly. It's much finer. Where on an electron beam and wire system you can count on a lot of post-processing machining, the laser powder systems are much closer to creating finished, ready-for-use parts. So they require, in a lot of cases, minimal, post-processing machining.

Where those animating at all, those pictures?

[INTERPOSING VOICES]

JON CALIGURI: I worked very hard on those. I'm not sure what happened. I'm so proud of my gifs.

Laser and wire-- this is another system. So again, the wire-fed system. It's pretty high deposition rates, so one case that we worked with was 23 pounds an hour, which is pretty significant. The wire fence systems have a number of benefits. One, it's basically weld wire. So it's a common material. It's non hazardous. It's easier to store. It's fairly affordable. And the powders can have-- there can be some health concerns, some safety concerns with the powder.

Arc and wire-- this is probably the most familiar from anyone who's ever welded. It's effectively a welding process. These systems are probably the most common. There's a lot of robotic welding wire arc additive systems that we see. And again, you can count on these parts needing at least a moderate amount of post-process machining, because the deposition is not terribly fine or accurate.

So this graph maps out-- and we're not going to spend a ton of time on this, but it maps out the strengths and weaknesses of the different processes. So depending on your part and your business, might make sense for you to-- if you're investigating additive, or metal additive in your business, this can be a guide to which technology might make the most sense for you.

So we'll discuss hybrid manufacturing. Again, like I said, this really just refers to combining additive and subtracting machining processes together. So one way to look at it is that every additive manufactured part is most likely a part of a hybrid process, because additive manufacturing is just not accurate enough for precision mating surfaces and things like this yet. So most metal parts that are 3D-printed are machined, or some post-process subtractive process takes place.

So one way that we can do this is use a traditional; like the EOS machines, powder bed technology; the most mature, that takes advantage of internal micro structures, like we can see in that cylinder headache example, to create a finished metal AM part that we then transfer to a subtractive machining center in this example, and do finished machining to precision faces and maybe cylinder head. So this can create the most complex geometry.

Another method of hybrid manufacturing might be to finish turn. So maybe we turn a large round shape, and then we use additive manufacturing to just add bosses and flanges and things on the side of the part, and then we can just finish mill, or finish machine, those added features.

So for the part that's on the screen, you could only turn to the outermost extent of all of these features, like a Spun profile for my CAD/CAM guys. So that leaves an absolute ton of milling to be done, if you were going to use traditional methods to make that part. So you can quickly see, on some parts, like the one on the screen, this makes a ton of sense.

So the third method features a machine that we work with very closely from our partners at DMS. I have another gif. It's not behaving terribly well. So this is a hybrid manufacturing process that combines additive and subtractive processes on the same platform dynamically. So in this process, we typically deposit one bead of aluminum, or steel, or inconel, and then we machine away the top portion of that.

So that rest time helps to dissipate some heat. And perhaps the material, when we machine away the very top, so that we have a little bit more of a constant known surface that we're welding on for our second and third subsequent layers. So this allows a three-axis machine, with welding and a wire arc additive process and a milling head, to create geometries you could never create in three axis.

So now we'll discuss the most common uses, or use cases, for additive and DED technology today, which would be create a brand new part, add features to an existing part, or repair parts. So this nozzle shape on the screen is an example of creating highly complex geometries, near net shapes, from scratch. So the benefit to using additive manufacturing is, A, you can create geometries that you could not create using traditional methods. That would be impossible to mill out of a billet many of the shapes you can do with this process.

The second is the amount of material. So when we start taking advantage of exotic materials, like inconel and things like this, super alloys, to machine away from a billet, you're going to remove about 80% to 90% of that material on a shape like this. With additive manufacturing, the amount of waste is almost zero. So it's a far more accurate process. And when compared to casting or forging, the environmental impact and the amount of energy used to produce these parts is far, far less using additive manufacturing.

Secondly, taking an existing part and adding a feature to it, like we discussed a little bit ago. This is a little bit more of an advanced application of DED, so it takes a little bit more know-how and engineering to know how to work on whether subtractive or additive processes a near net shape part. So companies that machine castings and forgings and composite parts today are typically where we see these talents and skill sets.

Thirdly, we can extend the life-- the useful life-- of high-value parts by using additive manufacturing-- hybrid manufacturing-- to repair and put these parts back in service. So like what James is going to speak about directly following this-- in-flight turbine blades, blisks, things that are under extreme load and tend to warp, chip, and need repair. Rather than manually repairing those, we're learning that we can adaptively add material, remove it away, and requalify those parts and put them back in service much quicker. This leads to drastically reduced lead times and far less disruption in the supply chain.

So for our fourth and final learning objective, I'm going to pass it over to James to discuss the work that they've done in our Birmingham advanced manufacturing facility at Autodesk.

JAMES DONNELLY: Thanks, Jon. OK. So, I'm just going to go through what we do in the department of advanced consulting, what my role is, the background of the project, and the reason why we did the case study. And then I'm just going to show you some of the results that we got with a short video at the end of the process itself.

So, Autodesk Advance Consulting [INAUDIBLE]. And we're involved in taking the core tech that I think of, like the [? brick bot ?] on the top left-hand corner, and applying it to our customer problems. So we took the AI intelligence that we used on that [? brick bot ?] and then applied it to a blade polishing cell for an aerospace company.

As well as the early stages of generative design-- taking the stuff that we learned from there, and applying it to the engine block which you saw earlier on John's slide, which helps the thermal properties of the part, as well as a study that was an environmental study that was done in the Boston office-- how we can scale all that to a large skyscraper or a factory of the future.

Now, we do that through direct customer engagement and consultancy. So, we team up with a hardware company, and then we'd go and deliver a project to a customer. Or we do collaborative research and innovation projects. So this could be a government funded project, or a research and development project within a company.

So I work within the subtractives and robotics department of advanced consulting. The sort of projects that we take on within there are manufacturing process development, complex geometric shapes like what you see in the picture-- a blisk-- and what I'll talk about today, adaptive manufacture of automated processes, customized software. So we use the technology that Autodesk have, but just re-skin the software to what our customer wants. Manufacturing services-- and, finally, customer-focused research and development.

So with a lot of the projects that we do within our department, they're tied into rules and regulations so that we can't really talk about the tech that we developed during the project. So what we decided to do was come up with the technical demonstrators to show off all of the technology that we developed over the past few years in these projects that we weren't allowed to talk about. And from then onwards, it sort of stemmed and snowballed into a design manufacture and repair of a high-cost aerospace component, adaptively repairing this part, which I'll talk about on the next few slides.

And we just wanted to increase our understanding of DED and that side of things. As well, there was a recent open house event within our advanced manufacturing facility, which is a technology center in Birmingham, and this was one of the demonstrated pieces on the machine.

So typical blade repair is shown there. So the process would be to remove a defect or the defected area from the part. Then, it would go away for a manual welding process, which you see in that picture there, or it would be done on an automated machine. Then, it would be set up again on a CNC machine to have that defected area removed. And, then, final inspection.

Now, the way that differs from the hybrid process that I'm going to talk about today is straight away we can see that two of the three setups have been removed, because everything can be done on the one machine. So you no longer have the risks that are involved in multiple setup processes. There could be an error on the one machine, which is then transferred onto another, and with a high value part like this, setup is very crucial to the process. So saving the time and risk involved in that is one of the benefits of this process, as well as being able to consolidate all the technology into one machine. So you've only got one footprint in your workshop, as opposed to having a welding machine, and a milling machine, and a manual grinding bench, for example.

So I'm going to strip the project down to three key technology areas within this project and talk about those. The first one is the additive repair. So this is what makes the process a hybrid process. We swapped out what we would be doing with the manual welding process with an additive.

So the hardware that we used is a Hamuel Reichenbacher CNC five-axis machine. So this is their blade manufacturing machine. So it's good for rotating parts, single blades, blisks. And what makes it hybrid is the addition of a blown-powder DED head, which, this one in particular is offered by Hybrid Manufacturing Technologies. It's their Ambit head, which is like a retrofit device, so they can implement this into many machines.

It comes in with a single tool change. So it's stored where your milling cut would be, or your probe inside the machine's tool changer. You call the tool out, it clamps itself into the head, and then we perform a docking procedure, where the laser manifold which all the powder comes through, and the laser source itself. And then that's both directed through the copper nozzle you're seeing on the deposition head. It's like it directs the powder into the laser beam, creating the melt pool, enabling us to deposit material similar to what you saw in one of the previous slides.

And, finally, this was programmed using a new bit of functionality added to one of our existing CAM packages, where as well as being able to subtractively machine a part, we can now add material in the same project, which makes it a hybrid software demonstration.

In the next slides, I'm just going to show the results that we got from the builds that we did. There's some process parameter work and tests that you have to do before you can do your first build. But I wasn't going to add this into the presentation. That's in the handout, if you want to know more about what goes into developing the process parameters, and how you get to do you first build.

So we generated a strategy on using the software to be run on the machine. And the thought process would be, try out a first one, see what we got, see how the blade handled the power, see if there was enough material being put in, see if the parameters that we've chosen were correct. And you can see here, that on the front edge of the part, it was blown away, because there's too much heat in the part.

We started at the back edge, where the edge is really good, traversing towards the front edge. And, I think, because of the amount of heat that was being put in the part, it sort of melted the tip of the blade. So then we went back and we thought, well, through watching the machine in process, you could see that there was a lot of heat being put into the part. And, I think, due to the heat build-up, it just melted the material away.

So to test that theory, we switched the direction of travel. So we started at the front this time, and run the same test again, and it was a mirror opposite. So we worked out that when we first entered the blade, the blade was cold. There wasn't as much heat in there, but by the time we got to the very end there was a lot of heat in this part, and it just took away the edge.

So then we went back to the drawing board, and devised new methods of how we could build this part, and what would be the best bet. And we changed two things. The first thing was, we saw that by the time you get to the second layer, the blade's got heat within the part, so you don't need to put as much power in to get it to the required temperature for depositing. So we decided that every other layer, we would lower the heat that we put into the part. So the first layer would be 100%, for example. The second layer after that would be 98%, and so on and so forth as we go up the part.

And we also changed the fact that instead of building in one direction, we now built in two directions, so starting at each edge and meeting in the middle. Because we saw, here in this example, that the edge that you start on was good, but the edge that you finish on was bad. So we thought that if we met in the middle, this would improve the process.

The results were successful. So we had a nice, even build, albeit there were some gaps where material might be missing. But the two edges that we were having problems with before were OK. It was a nice, even level, so you can work with that then. You've got your first build. You can start developing the process further. You could spend a lot of time fine-tuning the parameters and trying to get more material on, but for all we know, that could have been enough material to finish the part.

So the next thing that we did was a blend test. So we decided to machine away the material that we just added, and just see basically what the results were, whether there were any voids in the weld, whether there's any areas that we need to retreat. And, as you can see in the image on the right, there's some sort of gaps where there's missing material, which we thought would be the case. You could see the definitive lines of the layers of the build.

But that's another beauty of this process. If, for example, it was a manual welding procedure, and we found after machining that there was not enough weld to be taken off the machine, sent away to be re-welded, and then that takes time. Whereas, now, with this hybrid process, we can just requalify the surface by removing material, editing our strategy, and then rebuilding, which is what we did.

So this was the final build that we came up with. So we added two additional passes to the build to basically fill in any of the gaps that were left from the subsequent passes. Which, you can see, it's a lot more uniform on the side. There's more material on there, so we were happy with the way that that came out. And we continued the process.

So once we'd done with additive, the next thing I want to talk about is the adaptive process. I don't know if you're familiar with adaptive machining, but you can imagine this part has been through some stresses and strains within the engine. It's now slightly deformed, whether it's been hit with something so that the blade might be in a different place to where you think it is.

Which means that we can no longer run a nominal from the [INAUDIBLE] tool path, because it's slightly away from it. This is also no two parts the same. So one blade's been machined. The next blade was machined. The tool might have slightly worn when it was being produced, so that means that it's slightly bigger. So running a nominal tool path just wouldn't work.

So what we have to do is we have to adapt those tool paths to prevent something like that occurring, where we've run a nominal tool path, the blades move slightly, it causes a gouge. Or, the opposite to that, we leave material on the blade. So what we do is we adapt the geometry-- the G-code-- to this specific blade, and then rerun it.

So there's several ways of doing this, but the one that we chose on this method was to use a 3D probe, which you're seeing simulated there, to capture the data around the blade and around the welded area. And then create an area map of where the blade is, what's it deformed, how much is it deformed. We then take the measurement data that we receive from the probe, input it into a modeling software, and then we're able to create new geometry specific and unique to that blade using various morphing techniques.

An example-- so once we've created the blend surface, we scan the part using the GOM scanner to give us a 3D image of this part. And then we overlaid the surface that was created using the probing results to give us an idea of how good the blend would be, or in a virtual world, how good the blend would be.

I don't know if you can see there, but the green's around about 10 microns, so just under half of that. And the blue areas-- I think it was 50 microns. So where the probing data is, the blend is extremely good. And then as we go up the blade further away, we expect to see some deviations, but that's the nature of this sort of work.

Once we have the surface generation, we then import it into our CAM software, and regenerate new adaptive G-code for each specific blade, enabling us to finish machining them successfully. One thing I will say about this, is that the previous image of the blend, you can see some areas of the weld where it's not cleaned up, or it's not put enough material on. Whereas this, when we revisit it and apply that new approach of building, you can see that all of the areas have been covered, and it's a nice, uniform-looking surface. The only way you'd tell that was repaired was by the blend line, which is due to manufacturing tolerances.

So the third and final aspect of this project, which I think is key, is the fact it's fully automated. So like a lot of the projects that we do within our advanced consulting, the customer doesn't want any interaction, or as little interaction as possible by the operator. They want to press a button or scan a barcode, and the machine jumps into life, signaling certain events and automating the process fully.

So we use a bit of software that communicates with the machine and the PC. So it's like a handshake scenario. So when the machine needs data, it'll request it from the PC. The PC will wait, sit in a loop until that's finished, so on and so forth. And we are able to automate different software tasks and file transfers using this software, making it the safest possible way.

There's a lot of aspects within this process. There's new G-codes we run, there's surfaces to be made, there's measurement data to be transferred. It's very easy for you to put the wrong file in the wrong place, or run the wrong G-code, which in turn will possibly scrap your part.

So, finally, this is just an overview of the part of the process itself, from the cropping of the tips, or the removal of any defected area. Then we build using a sort of nominal preform, plus one millimeter. So we built the geometry on the top of the blade using the additive head. Then we go in, and we inspect the geometry to create measurement data on our unique surfaces. Then we bring that into our CAD software, and develop those surfaces. Finally, we take that surface and import it into the CAM, and we create our unique G-code, which is then run on the machine.

And here's a short video of the process, if it works. Not yet. OK. So here we're cropping the material away. And this is the first part of the process-- can't really see much because there's a lot of [INAUDIBLE] in there.

So one thing I will add-- because these blades have changed slightly, what's really the nominal tool path for the cladding might not have worked. We found that, because the blade is in the wrong place, sometimes it would favor one side over the other. So the first bit of probing there was creating a best fit. So that we'd probe the blade, find out where it is, and put a best fit procedure so that we would make sure that we were running down the center of the blade with our cladding, which you can see there every time.

And then, once that was finished, we would reinspect the geometry, create the new surfaces, and then finish the part.

So that's the end of the presentation. Any questions? Thanks for listening.

JON CALIGURI: We threw an awful lot at you guys.

[APPLAUSE]

So we wanted to make sure we left some time for questions. I know that for most people, this is still sort of out there a little bit. Anybody in this room doing metal additive manufacturing right now? I know you guys. I know you.

So the reason why I wanted to do this presentation at AU was a couple of reasons. But, in my job, we support manufacturing companies all over the US, Canada, and Mexico. And this question has been coming up on a constant basis for the last maybe 18 months. We talk a lot about it on the main stage. There's a lot of very nice things out on the exhibition floor for this AU. And this is obviously a hot topic.

But I think that some of it can-- it helps to dive in a little bit deeper on some of this stuff, and just understand, really, who are the players? Where does this technology make sense? And help show some cases, like what James just talked about-- not that everyone has to do blisks to apply this-- but see some real-world applications for this technology. This is becoming a reality for a lot of our customers.

So if there's a way that we can help you guys implement this technology, or just investigate it inside of your companies, James's group is a fantastic resource. We're always more than happy to help design software, and this side of the room primarily is made up of maybe the best expertise in the industry around this kind of technology.

So if we do have five minutes or so at the end of this, feel free to at least make some connections, and there's a ton of knowledge in this room that helps us greatly, and it's all available to you guys. And that's really the beauty of AU. So I encourage you to take advantage of it, or if you'd rather ask questions offline, I'll hang out.

AUDIENCE: I just have a quick question-- just an adaptation for mold repair [INAUDIBLE]

JON CALIGURI: Yeah, so adaptive mold repair is one of the applications that we've identified. Blade repair, mold repair-- there's more to it, obviously-- and you know this. But there's more to it than just, oh, there's a bit of wear on this part of the shut-off surfaces of a mold. If we just use what would be like a surface coating operation to clad some more material on, we could just remove it, and the mold's ready to go. There's more processes involved in that. There's heat treat. There's all sorts of things.

But, yes, I do believe that a lot of this technology makes a ton of sense for that. But it is fairly early days. So even on the blade repair example, you can see the iterations, and that's why I think that was great to see. This is not putting a new Haas mill on your floor, and getting new parts 45 minutes after the machine is hooked up.

There is a lot of process development that takes place. And I think there's a tremendous opportunity for CNC programmers or manufacturing engineers to take the leap to learn this technology, because these early adopters are in desperate, desperate need of highly qualified operators to help solve these problems. So Autodesk Advanced Consulting and the dealer channel. That's a lot of what we do is support our customers-- you folks-- in trying to figure out how to implement these technologies in your business, as well as machine tool partners.

We're actually lucky. One of the machines I featured-- we have some senior leadership from DMS in the back of the room. We work with them really closely to develop these processes before they even hit the customer floor, as well. So we'll take tools like the PowerInspect, PowerMILL, Fusion-- we'll take those tools and we will customize them for your company, or for the machine tool OEMs, so that maybe we can help you go from 10 iterations before a good part to maybe three or four. But there's no magic button for this stuff yet.