Sea Control - Sea Control 320 - Better, Faster, Cheaper Ship Maintenance - Now! with Artem Sherbinin

Episode Date: February 20, 2022

Links1. “Better, Faster, Cheaper Ship Maintenance – Now!” by LTJG Artem Sherbinin, USN; LT Thomas Wester, USN; and CPT Richard Kuzma, USA, USNI Proceedings, January 2022.2. “A Fleet without a Rudder” by LTJG Artem Sherbinin, USNI Proceedings, October 2020.

Hey, folks, Sherrod here. Andrea Howard is back as host this week, and she's joined by Artem Shcherbinin to discuss his proceedings article on Better, Faster, Cheaper Ship Maintenance. This episode was edited and produced by Joshua Gruber. We're still looking for audio editors to add to our team, and if you're interested, please drop us a line at ccontrol at simsec.org with your resume. I want to highlight a couple of developments over at the website. In the last two months, we've launched the Flotilla, SimSec's warfighting society focused on tactics. and just this week we launched our naval wargaming discord you can find more information over at simsec.org finally i want to take the opportunity to recommend our partners in the simsec podcast

network the bilge pumps you can find alex jamie drack and a pot of iron brew bottles wherever you download your podcasts on that note i'll turn it over to kimber's men you're listening to sea control hosted by the center for international Greetings Sea Warriors! Let's dive, dive back into sea control. I'm Andrea Howard and today we're discussing an article titled Better, Faster, Cheaper Ship Maintenance Now. It's featured in the January 2022 edition of the U.S. Naval Institute's preceding magazine and penned by three junior officers, Lieutenant Junior Grade Artem Shcherbinin and Lieutenant Tom Wester of the U.S. Navy, as well as Captain Richard Kuzma of the U.S. Army. Due to operational

constraints, I am joined today with a sole representative from this rock star group of authors, Lieutenant Junior Grade Sherbinin. Artem is a Surface Warfare Officer Station in San Diego, currently leading the team writing the Navy's Artificial Intelligence Strategy. He previously established the Hopper Group, which is the Surface Navy's AI Development and Governance Team. As a reminder, all opinions expressed are personal opinions and are not representative of any institutions with which we are otherwise associated. Thank you so much for joining us today. Yeah, thanks for having me, Andrew. Yeah, so clearly you're involved in some pretty light stuff with the Navy. And of course I'm

joking. So your pieces have been featured and received some of the highest hits across the U.S. Naval Institute and in the other forums where you write. Let's start there by talking about the motivation behind this particular article. If we're looking at my home community to submarines, you cite the government accountability office and noting that attack submarines awaiting maintenance between 2008 and 2018 sat idle for 10,363 days. How bad are these maintenance delays across the entire Navy? Yeah, so great question. Across the Navy right now, things are getting better. But let me just tell you where we were just two years ago. Back in 2019, it was estimated also by the GAO and Congress that only about 60% of ships, or sorry,

60% of ships failed to complete maintenance on time. That's a pretty staggering, and when I say ships, I mean submarines and surface ships. That's a pretty staggering number to think about. Now, during a window, it was clear to the new CNO that this was a problem. So during the window of time between 2015 and 2019, there was a huge capital investment into ship maintenance. It was recognized that this was problematic. I think the capital investment was about, it was almost $3 billion. in that between those four fiscal years. Despite that, there were over 7,400 days worth of

maintenance delays. So in a four-year period, you experienced approximately 20 years worth of delays, which is pretty mind-blowing to think about. There's a number of factors that contribute to this, whether it's inexperience by shipyards, a loss of talent in the shipyards, bad processes, poor management. I mean, there was a whole lot of things that went into this. So the Vice Chief of Naval Operations brought on board the Center for Naval Analysis to look at these trends and instituted a program called P2P, Performance to Plan. That program is now in place and it seeks to use data and data analytics to make data-driven decisions regarding the ship maintenance process. So that's where we were. And now with P2P, I'd estimate right now there's about

a 25 to 30 percent improvement from that previous 60 percent of ships failing to complete maintenance on time. So we're nowhere near where we need to be, but we are driving to get better. But I want to accelerate, you know, and I and so many folks, especially the type commanders like Vice Admiral Kitchener, are looking to accelerate that. And we see emerging technologies as the answer to doing to doing that and getting that 60% number down. It's certainly encouraging to hear during my time on USS Ohio, really two thirds of my JO tour was consumed by being stuck in different shipyard environments, whether in Puget Sound, Pearl Harbor, or in Guam. And in all three of those places, the themes of what was happening to the Navy at a larger scale were certainly prevalent. So

In turn, that causes an over-reliance on operational vehicles, unexpected delays in overstretched shipyards, and people that are on these crews in their shipyards, which then overall just deteriorates the warfighting readiness of the Navy. Thankfully, you guys are offering solutions, though, into how we can expedite these shipyard periods and therefore open up more space in shipyards for ships that have the highest need of maintenance. And one of the concepts that you discuss is predictive maintenance. What is this and how does it differ from traditional maintenance techniques such as preventive maintenance? Yeah, so there's actually two things that we're talking about here. It's getting ships out of the yards faster, right? And that's one component of this, which is the primary responsibility of the type commanders, right? They need to generate forces for the operational commands to use. But there's also a second component of maintenance. This is really what's near and dear to junior officers like myself and you and all of our friends. And that's the daily process of doing maintenance. And that's where predictive maintenance comes into play. So we use the term predictive maintenance, the technical term used within the Navy, which we chose to avoid in our article because it's not very common in industry, is conditions-based maintenance. And what conditions-based maintenance is, is exactly what it sounds like.

When a certain piece of equipment meets certain parameters, then you conduct maintenance on it. So let's say you're an engineering officer and you're working on a fuel oil filter. Currently, sailors have to perform that check on a certain, you know, on that fuel oil filter on a certain time-based period to see. Let's just say once a year, right? But what if when they do that, that maintenance on that filter, the ship has only been to sea for 100 hours in that year because of various technical problems. And so they do this maintenance on this filter, and they take out the old filter, and they put in a new one, and they throw away the old filter, right? That was man hours, that filter might have not needed to be replaced. And, and you're getting the trend. Predictive maintenance or conditions-based maintenance happens when sensors tell you that a part is nearing failure and needs to be replaced, or when it's met certain conditions like X number of hours of use, and then you replace that filter. This saves time, and it also saves a significant amount of money. And that's that second maintenance piece.

So that's not necessarily happening in the shipyards. That's everyday maintenance that sailors are doing on deck plates around the world today. Part of the issue as well, if you look at the larger scope of wear and tear on these boats, is from the December 2020 30-year shipbuilding plan for the Navy, there's a goal to decommission 40-plus different naval vessels between fiscal year 2022 and 2028. With that, in the programs that you're talking about, if you were able to extend the viability of different components on board, you're offering solutions to then extend the lifetimes of these ships. Can you talk to us about the Navy's integrated condition assessment system, ICAST, which is one of these methods that can help elongate the livelihood of different pieces of equipment on board? Yeah, so we'll talk a little bit about ICAST, but I also want to share something that hasn't hit too many ships yet, but will be on board your next ship and submarine very soon called ERM, which is a successor to ICAST. But what these systems are in general, ignore just the acronym of ICAST or ERM, it doesn't matter. These are automatic logging systems where they take the sensor data that's coming off of these highly censored new ships.

So consider the Elm 2500 gas turbine engine. I was an engineering officer on one of my tours, and the Elm 2500 gas turbine engine is extremely well censored. There are dozens of sensors across the entire powertrain of modern destroyers and cruisers. Those systems log, like ICAS or EM, log the data that's being recorded by these sensors, and they assess in the back end what's happening with that component. So we can actually use that data to train machine learning algorithms. And really all that means is we take data and we conduct a process called supervised learning where we're teaching a computer what to look out for. We're telling the computer the answer of, hey, this is what happens when a part fails. And then we're having the computer look out for that trend that leads up to that part failure. Thanks to these logging systems like ICAST and ERM, we're actually able to do a lot with that data.

So instead of that data just going to waste or being manually logged in a logbook for sailors to have to do that trend analysis on their own, there's a machine learning algorithm doing it. So we can actually predict, for instance, when an LM2500 will fail before it does. So let's say that the analytics that you're talking about actually do come to fruition, you know, in the hopes of assisting engineering officers, engineering officer watches, sailors on these operational platforms. What is the best way to then pass these deliverables to the operator from this computer that is starting to learn this information and look for those patterns? That's a really great question, because I think a lot of times when we talk about technical solutions for the Navy or technical solutions in the Department of Defense, we want to say, hey, let's slap some AI onto this or let's do big data analytics on this. And that's all well and good. But unless you can build a really great user experience or user interface, so UX, UI, as we say in the software world, it doesn't matter. Because fundamentally, what we're looking to change is how the sailor does work, not what work they're doing. They're still doing maintenance, but we want to change the

process by which they're doing it, which in turn will save us money. So there's a number of different ways to go about this. For one, one option that we've considered is using augmented reality. This is a very mature technology that already exists on your smartphone. For those of you listening, if you have an iPhone, you can actually measure a wall or a painting in your house or in Ikea, you know, a couch to see if it'll fit using your phone. And it built an augmented reality. What that technology does is it imprints a virtual image on a live image that you're looking at. So let's say a sailor is working with a piece of equipment. They're not going to be looking at a printout in the future of what ICAS or ERM told them is wrong with that piece of

equipment, they're going to be doing something either on their phone or with a tablet or wearing some kind of augmented reality glasses, which are showing them in real time, a digital imprint on a physical image that they're staring at of what's wrong and how to fix that piece of equipment. And by the way, for those of you listening to think that this is far-fetched and something you'd see in Star Trek, this already exists at companies like Boeing or UPS. And this is a very mature technology. Taking it a step further, you know, sailors that are on watch are going to be able to utilize this technology to diagnose what's happening with the equipments. How will this actually transform the training environment for them too? We already use virtual reality trainers

and in our, you know, normal officer and sailor training pipelines, and augmented reality is really just the next step for that. So I see in the future a world in which sailors are able to instead of necessarily having to have a chief stand over them and provide them training, they're going to be able to go conduct that maintenance and get the help of this virtual assistant and these virtual images appearing across their face or on their phone. And it's going to be telling them step-by-step of how to perform this maintenance. And again, these are things that are already happening in the private sector. So zooming out a bit too, what does the ideal outcome and look like in terms of communication across different ships in the Navy

and how it informs the decision-making process for captains and commodores? So at a macro level, at the end of the day, the next iteration of naval warfare, but arguably all of warfare, the character of war is changing thanks to emerging technologies. And the next iteration of that

is this world in which you have interconnected systems, right? Part of that process underway in the Department of Defense right now is something called CHAD-C2, an effort to network every sensor in the entire U.S. military across the joint force. In the Navy, we have something called Project Overmatch. That's our enclave of JADC2. Part of that effort is to get all of this sensor data, this raw data coming off of your engineering equipment, your radars, everything off of a ship and into an environment somewhere, whether it's

on-premises or up in the cloud in which all this data can be analyzed, right? And it's not going to be a person analyzing this. In fact, my team that I previously worked on, the surface analytics group, we estimated that a single surface ship produces 118 terabytes of data per day, right? That is more than any platform in the US military, single surface ship. So no human can look at 118 terabytes of data per day. For context, that's about 200,000 CD ROMs. So if you can picture that in your office right now, that's how much data we're talking about. What's going to happen is this data is going to go off of ships into the cloud, thanks to the amazing work that Overmatch team is doing. And that data is going to be run through machine learning algorithms. So we're not

just getting data off of your LM2500 and my LM2500, and then someone is separately processing all that and looking at it, which is happening today. We're going to be getting data from every LM-2500 and every fire pump and every sensor all at once. And with that kind of data, we're going to be able to predict part failures as well as do a lot of other really cool warfighting stuff. Yeah, I think one of the neat concepts is that a captain could see what the trend analysis is for particular bits of equipment on similar vessels of the same class. And then you also will have commodores who are then empowered to look at the performance levels of the different pieces of equipment across all the boats under their cognizant. That's really special and totally

innovative. Do you see any other major impacts that this might have in regards to that chain of command informing? Yeah, so we're actually seeing a lot of those impacts today already. The surface force, I can't speak to the other TICOMs, but I do know that there are really great efforts underway at all of the type commands, whether it's aviation, subs, or surface. But in the surface Navy, our leadership has really embraced data-driven decision-making. Vice Admiral Kitchener now gets briefed on the status of every single ship in the fleet simultaneously using various analytics tools that our team has developed for him and his staff. Right. So we're already seeing the ability of the type commander to now directly support ships that are at risk or in trouble and need to get a little extra assistance on the maintenance side or on the personnel manning side. So in the long run, placing more data at the fingertips of senior leadership is only going to drive to get more ships out to sea ready to execute tasking. I really could not agree more. We wouldn't be doing our due diligence, though, if we didn't compare ourselves to the Air Force and the Army and some of our sister services.

So how have these other branches capitalized on commercial technologies? And do you believe that the Navy has the same capacity to do so? I think absolutely the Navy can do the same. Across the board right now in the Department of Defense, you're seeing a really huge revolution in the IT space. Whether that's the adaptation of DevOps, software practices, agile acquisition practices, and agile software development, or obviously the adaptation of artificial intelligence, which is very near and dear to my heart. Across the board, one of the key drivers of this movement hasn't necessarily been internal development of these technologies, but the purchase of commercially available solutions. So we call commercial off-the-shelf technology. In other words, let's say you've got a radar that is on a merchant ship and it's a really good radar. We just buy that and install it on our surface ships or submarines. We're doing the same thing with software today. And there's some amazing programs in the Air Force, for instance. There's a program called AFOREX, where they issue out CBER contracts. These are small business contracts. At various levels, they fall below usually a million dollars, I think, on average, or $800,000 contracts. These are really small contracts that a major company like Lockheed Martin wouldn't be competing for. But we're getting this amazing technology through these small startups. The Air Force has shown that it's doable.

I think the Navy can do the same thing. All of this sounds so fantastic. But ultimately, if you have data that is going to inform you on when pieces of equipment are going to fail, that's only good enough if you actually have replacement parts for that equipment. So many of us who have served on older platforms understand the woes that it takes to kind of seal and cop these other pieces of equipment from boats that are less operational and scheduled to be in the yards for a bit longer. Your article also addresses a solution for that. So what roles do 3D printing and additive manufacturing play in bolstering the supply chain in both peacetime and in wartime? And how does that fit into the larger scope of what you're talking about in terms of the future of maintenance and operations? Yeah, so in, you know, before I did data science and machine learning, I was a student of strategy and have a background in national security policy.

And there's this old adage that in war, amateurs study tactics, professionals study logistics, right? So you can't fight a war without the logistics train to back it up. Additive manufacturing brings the logistics system to you, right? It brings the Navy supply system to your ship or submarine or aviation squadron. Additive manufacturing is not creating, you know, play things out of plastic, right? Oh, you can go out to a store, to a Target today and buy a 3D printer. Don't quote mail that. Maybe it's not Target, but maybe Best Buy.

But you can buy a commercial 3D printer and make some really cool little widgets. But guess what? You can also make a gun that shoots real bullets. you can make a on a surface ship a strainer for thousands of psi of water to go through it right so these aren't just uh little things that you tinker with this is an existing capability that we have today uh and at the very least we can use this capability if it's properly you know given to our sailors and they're provided training on how to use it which some already are and i'd like to shout out uh admiral small over at navsea who's a big champion of this technology and getting it

to our surface ships we can produce parts and in the short term at least it's going to be a stop gap in case during wartime you know supply chains are cut off and things like that but in the long term it can really help augment the situation where we have a lot of aging platforms and not enough replacement parts i've served on a really old ship i've served on you know the oldest cruiser in the fleet uh that that ship requires parts that are sometimes not manufactured and we would have to call a manufacturer to custom make that part. As the law of supply and demand goes, if demand is very low, supply is also very low, and costs go up, right? Now imagine if instead of calling somebody and paying $100,000 for a part, because you only manufacture a small number of

them, you've got a third-class petty officer who can make that part on board using a 3D printer. Well, we are certainly kin in that regards. I served on the oldest operating submarine in the fleet on the USS Ohio. And so I can certainly share your limitations about the material lows on the oldest cruiser in the fleet. One of the applications that I foresee for this, at least on the submarine side, is in the development and delivery of the Columbia-class submarines, which is arguably one of the most important missions of the U.S. Navy because of the role that those submarines play in the deterrence piece of our strategic triad. additive manufacturing helps mitigate risk of not delivering parts on time, which is one of

the greatest areas of risk right now for that program. What other programs do you foresee that this is going to assist? I'd love to hear your brainstorming on this. That's a, that's a good question. Put me on the spot there. If I have to think off the top of my head, it's really going to be, we're not going to be manufacturing huge parts. Additive manufacturing is capable of doing that but the real value in this is manufacturing your individual like nuts and bolts and uh you know small pieces of engineering equipment like sections of piping and things that you have to do in a yard period that take extensive amounts of time and can't normally be done by ship's force so anything that your sailors currently can't do and have to outsource to the shore side

infrastructure which by the way will come under direct threat uh in the event of a conflict even mainland infrastructure. In other words, things in the continental United States aren't necessarily going to be safe. That's a very different way of fighting a war, by the way, than we're used to as a country. The ability to manufacture those small pieces that currently can't be done by ship's force, but play a huge role in the overall, you know, functionality of a ship. That's where additive manufacturing has the biggest return on investment. And really another great platform that could become the experts of additive manufacturing are tenders, the submarine tenders and all the kind of various war games that have been played

over the last few years out in the Pacific theater, all highlight the importance of these tenders being able to deliver parts and deliver support in the outbreak of conflict. And for them to be the subject matter experts on additive manufacturing would be an absolute game changer. Yeah, as a complete kind of very quick sidebar, in the opening days of World War II, I think very few people realize just how important those tenders are. There are two submarine tenders, one in Guam and one in Diego Garcia. At the opening stages of World War II in the Pacific, when the Japanese attacked the U.S. Asiatic Fleet in the Philippines, the Asiatic Fleet had a number of submarine tenders. Those submarine tenders kept the entire force of submarines and destroyers and the three cruisers that the Asiatic fleet had afloat for just a few extra months, buying the U.S. Navy significant amount of time to rebuild after Pearl Harbor. Those tenders played a crucial role in the opening days of the conflict and allowed submarines to operate with near impunity inside the first island chain against Japanese shipping. their ability to manufacture parts at that time kept the US Navy alive. So it can't be

understated just how important that supply chain is. Certainly. And look, I'm a believer of everything that you've stated thus far in this podcast. Again, like I said, somebody who's operated on older platforms, the ideas that you guys have put out here and the initiatives that are actually taking place to transform them into reality are invigorating and make me believe that You don't have cases where boats like the USS Boise end up for five years in shipyard with no patrols and, you know, to date, over $355 million worth of repairs. You are offering solutions that are going to expedite and more efficiently and at lower cost deliver these platforms back out to sea where they belong, which then in turn also makes sailors and officers happier and more tactically proficient operators. However, I can foresee that there's going to be some pushback from some of the old guard, from some people in the chain of command in regard to how this technology impacts decisions

that are made. Like just for one example, if you have augmented reality glasses that are going to empower operators to identify equipment that's going to fail, are they then able to perform maintenance on board? What kind of level of permission do they need, even if they know that the problem is imminent and the failure is therefore imminent? So, yeah, what do you foresee in terms of that example, but then also more broadly in terms of pushback from the chain of command? I'll start with saying that conditions-based maintenance, which is really the crux of what this article talks about, the other capabilities like additive manufacturing and augmented reality are really enablers to conditions-based maintenance.

It has broad support across the Department of the Navy right now. The Vice Chief of Naval Operations is a huge supporter. And you've got entire program offices and NAVSEA who are very eager to tackle this problem. So in that regard, I think we've really moved past the old guard problem. However, it's well known that the Department of Defense, and I'm going to quote the former CEO of Google here, Mr. Schmidt, that the Department of Defense doesn't have a technology development problem. It has a tech adaptation problem. So exactly the kind of problem you highlighted, like why would a captain trust this new technology? Or better yet, why would senior leadership pay for the storage of 118 terabytes of data a day?

Currently, that 118 terabytes is overwritten constantly. You know, a resource sponsor such as Opnev in 96 and 97 and 98, they've got to worry about fielding the Columbia class, for instance, for Opnev in 97. And then you imagine walking into their office and saying, hey, I need you to pay the cloud storage costs for 118 terabytes of data a day off of every single ship. They're going to say, well, what's that buying me? Right. Well, the response is it's buying you a huge return on investment in the long run. You really have to believe in that and to believe in that. And also for you as an individual captain at the deck plate level, not sitting in the Pentagon's resource sponsor, but as a captain on the deck plates, to believe that this system and trust the system, you're going to need to perform a lot of training and education. And I'm happy to say that this year, we've taken a significant step forward in that direction.

Certainly in the surface warfare enterprise, Admiral Kitchener has done a lot of AI education and data literacy for himself and his team. We have a program in which we're looking to send every flag officer and senior executive service member through an AI and data course over the course of a couple of days at MIT. Obviously, these are all voluntary programs, but there's a huge drive to get data literate. And then we're going to trickle that down to the deck plates. Eventually, the surface worker officer school, as just one example, or different A schools that sailors go to are going to have to implement, and they already are thinking about this,

are going to have to implement data literacy and AI education. The U.S. Naval Academy is starting a data science major because they recognize that just like 200 years ago, it was essential, sorry, it's about 150 years ago. It was essential for our new young officers to learn about steam propulsion and mechanization. And they studied mechanical engineering and electrical engineering. In the future, our officers will have to know about data and the importance of clean data and how that's going to drive senior leaders' decisions. And that's all happening today. We are, 2022 will be a big inflection point for the Navy in this regard, I think. That's entirely great news.

You became a champion for JOs across the Navy with your article, A Fleet Without a Rudder, that was published in Proceedings as well. How does this article, Better, Faster, Cheaper Ship Maintenance Now, feed into those themes as well as these initiatives that you were talking about in a brighter future for 2022? to. So a fleet without a rudder can be, is an article about, you know, I'll preface by saying that I think any officer or sailor who had served between about 2018 to today could have written a fleet without a rudder. I just happen to have been the guy that did it. And I really appreciate all the sailors and fellow junior officers and senior officers who landed mentorship and support in that process. But what that article was about is the personal, meaning the manpower. So individual sailors, the personal and technical and ultimately operational toll on the U.S. Navy over the last

20 years, and especially the last couple of years. That article was a great experience because a lot of people stood up after the fact and said, hey, we really do have a supply and demand problem where we don't have enough ships to fulfill our global commitments. And the demand from combatant commands is ever increasing, whereas the supply from administrative commands is falling behind. I believe that the root cause of it falling behind, other than a lack of ships, is not a lack of investment in maintenance, but a lack of the maintenance process delivering results. And so this article hopes to address just one way in which we can get more ships out to sea better, faster, and cheaper.

You are certainly a role model for not just junior officers, but officers at all levels in the Navy and sailors as well, Marines, to engage in these issues, think about these issues. And so what advice do you have for people that are trying to follow in your footsteps to write like you and to pen articles like this one with the junior officers that you took this endeavor with? And where can we find you online to keep following your thought processes and ultimately your presence? Yeah, so the best piece of advice I give to any sailor, marine, airman, soldier, and officer out there that has a good idea is think through it and then get it on paper and put it out there for other people to critique. But also, after you do that, especially if you write a technical article or you want to change something or let's say you don't like NAVFIT 98, the performance evaluation software that we use. go out and make a new one. And I can, I can tell you right now that that's what my amazing team is doing this year is we've seen a bunch of things that we're unhappy about that we see at the deck plates. And we see a system where program offices fail to deliver capabilities at speed. And that's

not a knock on the, on the incredible men and women and government civilians and officers who work in those offices. That's a, that's actually a feature of the system. It's meant to be risk reverse. It's meant to be slow. But in the age of great power competition, we don't have that luxury. So go write something down and then go build it. Because I promise you that that's what my team's going to do this year. And you're going to see some amazing software coming out to the fleet and in your pocket on your iPhone pretty soon. It is great news to hear that this might be the last time that I have to engage in writing a fit rep on NavFit 98. I don't think I could end on a higher high note than that. So thank you for joining us today. That's unfortunately all that

we have time for. My guest, Lieutenant General Grade Artem Shcherbinin, has been fantastic, and we look forward to seeing you next time. Thank you. Thank you.