Podcast: Improving the reliability of the PATRIOT missile defense system with vibration analysis

Dave Aebischer is a senior data analyst for Cintel. Throughout his career, Dave has focused on improving Army weapon system sustainment outcomes and developing new diagnostic technologies for combat equipment applications. He has introduced innovative technologies and techniques for meeting the challenges of sustaining complex weapon systems in combat environments.

Dave recently spoke with Thomas Wilk, editor in chief of Plant Services, about a project he and his team worked on to support the PATRIOT missile defense system program. In part 1 of their conversation, Dave details how he combined vibration analysis and 25 years of experience on power generators to successfully identify and correct an issue with new generator assets.

Below is the transcript of the podcast:

PS: Maybe we can start with you providing a little background on the projects, maybe starting with your role and primary research interests.

DA: Thank you, Tom. We would have to start, as you mentioned, with the presentation that I gave at Leading Reliability. It was about a project that were working for the PATRIOT missile defense system, and it's generally a well known system, one of the Army’s Big Five, so to speak.

People may not know the PATRIOT missile defense system is made-up of several subsystems, the most recognizable of which is the launcher station. When you see a movie clip or a television clip about the PATRIOT missile system, you'll see this large trailer with the launcher, with the missile tubes, and you'll see it firing. Well, right on the front of that trailer sits our tactical generator, and that generator is responsible for providing power to that sole source, to that launcher station.

Around 2019, we had had one family of generators or one type of generator that had been fitted on there for almost 30 years, since the early ’90s, when we put the tactical quiet generators (TQG), and that's a name for a family that the DoD puts out for all of its tactical power systems. Those were in need of refresh for 30 years, so the new family is called the AMMPS and that stands for Advanced Medium Mobile Power Sources. They developed those in the same sizes and variants as the TQG.

So in 2019, and we we’re beginning to field AMMPS and the first fielding was not for U.S. units. It was for a foreign military sales (FMS) unit we had integrated for the PATRIOT. We had to get all these generators ready and we fielded them to an to an FMS partner. This was catastrophic – we had immediate failures of these generators, right off the bat, first time out-of-the-box. The failures were characterized by, if you can visualize the back end of the generator with the main alternator, the welds that hold that generator together – the outer frame of the generator – were all cracking and fracturing. The welds would fracture and that would cause contact between the generator rotor and stator, and then catastrophic failure.

This is what we were presented right at the beginning. My role is for Cintel was and is PATRIOT power support, so all the power systems that power the different weapon systems within PATRIOT, the launcher station included. So this fell exactly in my wheelhouse, and my research interests were always analytics from a very long Army career of sustainment and maintenance. I spent my whole career working on maintenance, and evolved into AI for maintenance in different aspects. We had included vibration analysis as one of our tools as well. So that's a little background there.

PS: This is fascinating because it reminds me of a different podcast episode that we did looking at EV vehicle charging stations, and the reliability of those charging stations. And it turns out that the issue there wasn't with the actual mechanics of the charger; it was with the cellular connection that was being used to bill customers for the power they were using it. It sounds similar to what you're describing, in that this an asset which is used out in the field, so it's got to withstand a certain weathering and use, but also if I understand you correctly, the failure mode you’re identifying wasn't even the electrical mechanics of the generator. Like you said, it was cracks in the weld.

DA: That's correct, it was structural. The way that the generator is put together, that structure where you want the rotor and stator are built by the DoD to very exacting power quality standards, so that the air gap between the rotor and the stator, the dimensions, are very close because it has to meet a very broad variety of load profiles, and power quality is very strict. Any change or anything that disturbs that air gap is not good, so when we lost the structural integrity of the stator, then the failures were pretty quick.

PS: If I heard you right, you mentioned that this was a transitional piece of equipment too. There was a previous generator that had occupied this space for the previous 30 years, and so this new version comes along developed for the same form factor. Was this the first time it was being tested out in the field, or one of the first?

DA: One of the first. We had done a lot of testing before we ever came to the fielding for the FMS partner. One of key things: PATRIOT takes a stock 15 kilowatt, 400 Hertz generator from the DoD, and then we adapt it to PATRIOT. That's a process where we have to make sure that when the launcher is in an engagement, it has to be that there are no personnel around it. We have to be able to monitor different parameters from remote. We need to be able to put it in what we call “battleshort.”

Battleshort means that the generator overrides all of its safety, so if it's in an engagement, it gives this generator every chance to deliver power through the engagement, even if it's failing. So it will override key (elements) like oil pressure, it'll just keep running through that. So it has to have a means to remotely put it in battleshort, and the PATRIOT Applications Kit (PAK) is what we call it. It includes several different components, and the core of it is what we call a remote functions assembly, which basically extends some of the controls that you have at the control panel back into the engagement control station (ECS) where there are personnel. So the personnel in the ECS can by remote put the machine in battleshort when they get a track and they have to go into engagement mode. So when they're in that scenario, everything has to be 100%, you have to have power through that. The whole system is designed to be resilient to anything that might occur in that space of time, anywhere from 5 minutes to several hours, where it has to be delivering power.

To extend that, we were required to do the testing for all that PATRIOT Applications Kit or PAK. It had to go through PATRIOT reliability standards that includes a rail impact test, road tests, all kinds of power quality tests for the PAK. For the most part, those tests were successful, so we were not expecting this particular problem to occur when we made our first fielding.

We had specked this out to be essentially a drop-in replacement for the TQG. We had done things very similarly with the PAK with our approach, with our testing standards. We almost just drew from those previous test protocols that we've done with the TQG several years ago and put together a new PAK, obviously that was tuned to the new parameters, the new control panel for the AMMPS and different components.

PS: Let's talk about the right mix of tools that you and the team used for both the RCA project and beyond. You chose a very specific set of non-destructive testing tools – vibration analysis, motion amplification. Could you talk about how that mix of tools was identified?

DA: I sure can, so a little bit background to get to that question, is that with the initial analysis of the weld fractures, the OEM did identify some issues with weld quality on that. It's a tough place to do a weld. Some of the pictures that I presented during the presentation showed the angles and the structure of the stator. It's a tricky place to weld and there were some problems with weld penetration that the OEM addressed.

The key thing here – I was very unsatisfied with the weld quality being a root cause. I definitely understood and I was on hand where we saw some of the dimensional problems with the weld, and then the OEM did a really good job of some destructive testing on the weld to identify where they might have had some issues at some voids. But again, I was not satisfied with that for good reason. The tactical quiet generators, and I'm going to date myself here, but when we fielded those we had a similar problem that did not involve weld fractures, but it involved what we call a cleaving of the rotor – a very difficult, very odd type of failure. But if you can imagine a piece of steel rotor which is maybe 3 inches in diameter, and if you just wrung it, that's what it looked like, like the shaft would get twisted.

And so if you go back maybe 25 years, we addressed this at the beginning of the TQG fielding for PATRIOT, where we had these catastrophic failures with the rotor. Now this was my first introduction to vibration analysis. Before that I knew nothing about it and at the time we guessed that we might need to do that kind of analysis to figure out what was going on. What we found was a structural resonance right at turning speed. This will make more sense in a couple of seconds when I get to the to the next part, but turning speed on these sets is 2000 rpm instead of 1800 rpm that are on the 60 Hertz sets. It happened that on the TQG, this resonance occurred right at turning speed, right about 33 Hertz.

And you could hear it! You almost didn't need a vibration analysis tool, you could hear it – when this machine moved in and out of that resonance, there was a grinding sound. So we treated that at the time, again I was a novice vibration analysis person at the time, but we treated that, based on a lot of good advice I received and research that we did, with a stiffening bracket.

We wanted to move that natural frequency up and out of the band around our turning speed, so we did a treatment of stiffening, which involved putting in a middle row of brackets. The generator is supported at four points – two under the engine, two under the generator. And we added a middle set of brackets right at the antinode between the engine brackets and stator brackets, with isolators underneath them, and then we also added some damping in between the flexplate of the rotor and the flywheel. This was very effective. We installed that on every single tactical quiet generator on every PATRIOT launcher in the fleet, and from that day forward for the next 25 years, we did not have another rotor cleaving issue after we'd had 60-70% failures for a good bit of time when it first occurred, so we absolutely cleared that problem up.

All that to say is when we got to when we got to this AMMPS problem, there were only a few people that were around when that all occurred, me being one of them. So I was very skeptical and I'm wondering, OK, this machine is very similar. It also runs at 2000 rpm, it's geometrically very similar, and in talking with some of the OEMs they said that they did see in their analysis some first bending mode anomaly, so that was enough for me.

At that point I said I think that we need to look at this a little bit broader. I'm not disagreeing with the weld quality issues. I think there's a combination that we need to do here of the root cause analysis in conjunction with redoing the welds, and that's what brought me to this set of tools. There were lots more tools in 2019, available to us to do this than there were back when I was first doing it. So this process was partly me getting up to speed on what kind of tools could be used, and how to use them.

So I got a bunch of help there. I'm getting myself back spun up on vibration analysis, details about how to perform an experimental modal analysis (EMA), how to perform operational deflection shape (ODS) analysis. But once I got to that point where I was capable of doing that, then those tools were absolutely ideal, because were able to characterize the failure. We were able to repeat the tests under different conditions on different sets on the trailer, off the trailer, with our PAK installed, without the PAK installed. We were able to prove that were not introducing any problem. What were doing with by putting the PAK on or by installing it on the tongue of this trailer. It was a generator problem. That was probably the most important step there is, to prove that weren't introducing the problem from an external source.

And then once we were able to do the tool, and were able to capture all this data in different conditions, just the precision and the explainability of the data, I think that was key. The ODS animates the machine at different frequencies, so you could just absolutely see it. And what we found during this test was a very strong structural resonance, major amplification around turning speed about 3x turning speed, right in the pocket, it couldn't have been any clearer once we got all this data.

The explainability aspect was key there. I think even more important, it guided our treatment so we could then go, “OK now we had some experience with stiffening on the TQG, we we’re sort of able to do that with the AMMPS as well.” We had the places we could attach it to and try it, so we prototyped a stiffening bracket set which on these were different, they had a little bit different geometry to get it around different components. We did that all in-house and built a bracket, and then were able to repeat all those tests, the EMA and ODS, and see what effect the stiffening had. We did that treatment independently, I called it Treatment One.

And then we were able to see that immediately after we put in the stiffening bracket, we had moved that major amplification out of the ±10% band that we were concerned with around turning speed. So we got a great result from that. But also looking at that – this the beauty of the precision and the explainability of those tools – is that you can see on your diagram where your amplification points are. It was clear to us that we'd cleared that little path around 33 Hertz nicely, but if we did any further stiffening, we would start to introduce more into the window.

PS: Interesting.

DA: We had to call it good on stiffening because we would have introduced more problems if we had kept going. So there we had what I would say is a good result. The amplification was much lower, but we had more to do. So this goes back to the testing that I alluded to earlier, during the testing they had changed the isolators, they had a problem during testing with the isolators that sit under each one of those brackets, under the two engine brackets and the two stator brackets, and they had changed it to a stiffer durometer, a stiffer isolator in order to let it pass PATRIOT reliability standards.

That was an area right there that the stiffening was effective in making the set pass the reliability standards, but there was, if you will, some elasticity in there where we could we could soften the durometer again, get some damping effect, and still not lose the structural reliability of the mount. So went back at that point and reduced the durometer rating of each one of those isolators, and then we again went back to the EMA/ODS iterations and tried and tested where we could at different levels of durometer ratings. We were able to just kind of fine tune this whole process into a wheelhouse; in the presentation I showed ODS animation where we went from no treatment to Treatment One where we got a good result. You could see that the machine was not twisting, the bending mode was reduced all the way to Treatment 1 + 2 where we had the stiffening bracket plus this damping effect of reducing the durometer rating on the isolators, and it the result was just perfect. Honestly, it was just right in the wheelhouse where we had this nice clear 10± band, and then we reduced the Q factor around the point nicely. At that point we were pleased, and we then went into prototype development of those of those parts to add them to our PAK.

PS: Well, how great is it that the previous project gave you insight into some solutions that might be an option for this one. I want to say it was Mark Twain that says “history doesn't repeat itself, but it often rhymes.”

DA: Yeah, and a benefit of old age is that I remembered back to do that stuff, because there were only one or two people in the room that could remember that far back to have the history. But that was, yes, very informative and too coincidental that we'd had this “problem” with the TQG, and I wouldn't say it was a problem but it’s something that goes along with operating this machine at 400 Hertz where you've got that additional speed and you got additional mass that it's different from the entire rest of the fleet that runs at 1800 rpm in a different mass. That introduces a whole different set of problems that we dealt with.

About the Podcast
Great Question: A Manufacturing Podcast offers news and information for the people who make, store and move things and those who manage and maintain the facilities where that work gets done. Manufacturers from chemical producers to automakers to machine shops can listen for critical insights into the technologies, economic conditions and best practices that can influence how to best run facilities to reach operational excellence.