Robbie Stewart speaks with Mark Hinaman about his time at GE, studying at MIT, and his nuclear startup, Boston Atomics.
[00:00:00] Robbie Stewart: You know the one piece on customers that you mentioned at this ARPA-e workshop last week, all of the, not all, but a number of the petrochemical companies and our, oil and gas companies were like, we would be excited to deploy nuclear reactor at one of our facilities, but we want to go and touch it. We need to go and see it built.
We need to see it operating. I wanna send my engineers to go walk around it. And the question I think everybody had in the room, and this was asked by multiple folks. Is, how do you get the funding? How do you get the capital to go build those demonstrations so that you can then walk a customer around it?
[00:00:32] Intro: Just because the facts are A, if the narrative is B and everyone believes the narrative, then B is what matters. But it’s our job in our industry to speak up proudly, soberly. And to engage people that are in energy poverty, they need us. America cannot meet this threat alone. If there is a single country.
Of course the world cannot meet it without America. That is willing to. We’re gonna need you. The next generation to finish the need scientists to design new fuels. And focus on net public benefit. We need engineers to invent new technologies. Over absurd levels of radiation. Entrepreneurs to sell those technologies.
And we will march towards this. We need workers to operate a. Assembly lines that hum with high tech, zero carbon components. We have unlimited prosperity for all of you. We need diplomats, businessmen and women, and Peace Corps volunteers to help developing nations skip past the dirty phase of development and transition to sustainable sources of energy.
In other words, we need you.
[00:01:41] Mark Hinaman: Okay. Welcome to another episode of the Fire Division podcast, where we talk about energy dense fuels and how they can better human lives. Today we’re joined by Robbie Stewart, co-founder of Boston Atomics. Robbie, how you doing?
[00:01:56] Robbie Stewart: Doing great. Excited to be here. Thanks for having me. Excited about what you’re doing with this podcast and the Fire two vision mission overall, so awesome.
Excited about our conversation.
[00:02:06] Mark Hinaman: Yeah, I’m stoked to chat with you. So Robbie, for our audience, why don’t you give a 30 to 62nd brief intro about yourself and we’ll dive into your background and, yeah.
[00:02:14] Robbie Stewart: Yeah. So I am, I think there’s a common theme of mechanical engineer by training, undergrad and masters.
And I’m now the co-founder and CEO of Boston Atomics, a nuclear reactor developer. So we got our start in the middle of Covid, actually, and we’ll get into more about Boston Atomics. As an entity later, but more in my background is, like I said, mechanical engineer. I went to work for GE right outta my master’s degree, and worked for their global research center.
So they have a, a research center that supports all of, at that point, the individual GE businesses, which are now separate. Um, and I worked there for a number of years doing research projects for the commercial aviation business, for the most part, wanted to get into decarbonization in a high impact way.
Made my way to nuclear, to m i t. Just finished my PhD last May, so a year ago in nuclear science and engineering. Um, and along the way started Boston Atomics with my co-founder Enrique Be Lopez.
[00:03:12] Mark Hinaman: Nice. Yeah. US mechanical engineering students here, at least the mechanical engineering educated have to represent right?
[00:03:19] Robbie Stewart: We do, we do. Well, you know, we’re the most versatile, so we have to, we get our fingers and everything. Yeah. Yeah.
[00:03:26] Mark Hinaman: Um, what was it like being a ge?
[00:03:29] Robbie Stewart: I love ge. I learned a ton there. The people I worked with were phenomenal, so my first mentor was a guy, his name’s David Hemer, and he still is, a mentor and he was phenomenally smart and generous with his time, and I learned a lot about.
The business. I learned a lot about research and development and implementing new technology into a complicated and highly system integrated machine, commercial jet engines. Yeah. And some land based gas turbines and the complexities of operating equipment all around the world in different environments and how you have to deal with different conditions and what that affects for the life and maintenance and performance of machines all over the place.
Um, my first job there was a lot of fun, so I was Basically in the lab every day. I came in right as they were shaking down an experimental rig and to test the heat transfer performance of a turine cooling system and certain let’s say hot and harsh environments. And it was a lot of fun.
I mean, it was every day. This didn’t work. Sit, let’s go back to the chalkboard. Redraw up. We gotta change the rig. It’s gotta look like this. Okay, this piece is melting. We’ve gotta reform it to look like this. Um, it’s like gotta
[00:04:45] Mark Hinaman: iterate to figure out what works and what doesn’t. Right?
[00:04:47] Robbie Stewart: Exactly. Very hands-on experimental, high pressure, high temperature testing stuff.
It was a lot of fun. So I worked with a great team that those first few years. My ended the GE in their probabilistic group. So we did, this is the front end of digital twin and industrial. So this is like 2015 timeframe. Building condition based maintenance models. So that’s how most folks view them, and predicted black models.
So could you look at the trajectory of how an aircraft operated for all like 3000 to 10,000 flights and estimate. Things from crack growth to pitting to, whatever the failure mechanism is. Um, maybe even update your model as you do inspections because the FMA requires certain inspections.
So this is like the forefront of the digital twin work. And I got to work with a global team. At that point, it was folks in Cincinnati, India. China, Japan. So, it was a lot of fun. I really enjoyed my time at ge.
[00:05:42] Mark Hinaman: That’s great. Yeah. During the design process, how many people would typically be on a team working on a certain project?
Like jet engines are awesome machines, but they’re pretty complex.
[00:05:54] Robbie Stewart: Yeah, so I’ll be truly honest with you, I have no clue how big the overall, so like overall team was, so when I was there, We were in the middle of shaking down the Leap X engine, which is what supported the new 7 37 max. And then the, new Airbus A three 20, A 3 21 Neos.
And so they were just, it was like first engine to test, second engine to test, and, G Aviation, I think at that point had like 50,000 employees. And they’re all over the place, right? They’re everything from maintenance and op to fleet support, to design, to experiments, to not even commercial aviation.
There’s a lot of military work as well inside the company, so, I have no sense of actually how many engineers we’re working on like the design of, about the
[00:06:37] Mark Hinaman: manufacturing center. Like, I mean, you got machine shops all over the place and p
[00:06:41] Robbie Stewart: Yeah, the research center. The research center was smaller, so we were like 2000 employees, maybe a little bit bigger, 1500 PhDs.
So we had a thousand PhDs. A handful of master students like myself that squeaked our way in there. And like the team for designing our experimental rigs was small. I mean there was, there’d be like three of us. It would be like me, one other research engineer and then a technician. And we’re all sitting there
[00:07:05] Mark Hinaman: of like rapid prototyping, hands on.
Exactly. Build stuff to test our
[00:07:10] Robbie Stewart: thesis and Exactly, and, you’re like the support network there is, while there’s only three of us staring at the chalkboard trying to figure out what we’re gonna do the next day, it’s the ability to walk up the hill and go to the machine shop and say, Hey, can you water jet?
This for me real quick? Yeah. Or can you, at which point the
[00:07:24] Mark Hinaman: machine is roll their eyes and be like, there’s no way we’re gonna build it correctly. They like that, you know, have you thought about doing it this way? And,
[00:07:30] Robbie Stewart: yeah. Or they’re like, Hey, I’ve got a queue and I.
The power of relationships are we benefited greatly from a technician that was really good at working his way up the queue. Yeah.
[00:07:45] Mark Hinaman: Great at buying
[00:07:45] Robbie Stewart: people beer, right? So yeah, dunking Donuts was the Thepath good too. Nice.
[00:07:51] Mark Hinaman: Okay. Well, did you have a favorite project that you worked on or was it just kind of one project the whole time?
[00:07:56] Robbie Stewart: I mean, I have a favorite project that first those, that first two years. Where I was in there and we were in the lab every day. After I did a lot of that, um, high temperature, high pressure testing, I shifted to a a rule where we were actually 3D printing test rigs. So we tried to take kind of like off the, shelf 3D printers at that point in time, whether they’re made by like form labs or maker bot and gas turbine, um, nozzles and.
Turbine blades have pretty complicated cooling geometries, so it’s actually a really nice feature to be able to say, instead of machining this geometry at the experimental scale, can I just go 3D print it? And then hook it up to like a low pressure scale, low temperature test rig
[00:08:40] Mark Hinaman: was the tolerance of the 3D printers good enough?
In my mind, interacted with a really janky, you know, in your garage, DIY 3D printers, right?
[00:08:48] Robbie Stewart: Like, yeah. So, the filament printers have a lot of uncertainty, but by the time you, and so you have to go in and sometimes we like micro CT them to figure out what they’re drawing.
Sometimes you just get out the, the caliper and measure exactly what you’re talking about. Seeing where pins are, I mean, the CFD folks really when they wanna do so computational fluid dynamics, when they want to validate against your experiment, they really want you to, know where everything is because the smallest errors and things can, can shift performance and to validate their models, they want really high, high tolerances.
But, you know, casting is also a highly uncertain process. That’s true. So the tolerances on turbine blades and nozzles. Um, are not huge. They’re, you know, on the orders of thousands of an inch, fractions of thousands of an inch. But when you scale up a test rate, that’s gonna be, you know, 15 times that size.
Then the tolerances on my 3D printer are pretty manageable.
[00:09:40] Mark Hinaman: Got it. All right. We can stop nerding out about the meki stuff, 3D
[00:09:44] Robbie Stewart: printings.
[00:09:46] Mark Hinaman: Um, okay. So, what prompted your decision to go to mit? I mean, you kept, ge you know, what, what were you thinking?
[00:09:53] Robbie Stewart: Yeah, so there was like practical, familial reasons, and then there was a broader interest in decarbonization.
So at the time we were living in upstate New York, my wife and I, where she had a research center, she wanted to go to law school. So she was like, I’m gonna head to Boston and I’m gonna go to law school. And I’m like, okay, well, I worked it out so that I could work remotely. And I kind of commuted back and forth every now and then to make that work.
And after a few years of that, I was like, I need to do something in Boston. I really want to have a bigger impact in the, climate space. Um, actually a few conversations with folks inside ge. One particular chief scientist was like, if we are to decarbonize, nuclear’s gonna play a huge role there.
And that was early on in my time there. And I that kind of just like sat in the back of my mind and as I was thinking about, okay, what’s next? I’ve gotta stop commute, you know, driving three hours to work. Um, I need to,
[00:10:45] Mark Hinaman: this is miserable. I ran out a podcast and audiobooks, right?
[00:10:48] Robbie Stewart: Yeah. I. I listened to, you know, I, I just, just dove in.
I was like, okay, well, what does it look like to expand nuclear? What, what is the technology? If we were really able to decarbonize, like what does that look like from a resource standpoint on wind, solar, batteries, hydro? And it was like, okay, actually there is a huge opportunity here for nuclear.
If, yeah, we can overcome a huge long list of barriers. There’s a pretty huge potential market opportunity for nuclear if to decarbonize our economy. So, that led to the motivation I was already living in Boston, so it was relatively easy. I applied to one school for grad school. Um, I mean, it’s a good school.
Yeah. Um, I kind of told myself like, I already live here, so if it takes me a couple years to get in and I’ll just keep applying it until it works out. Um, lucky enough to get in on the first shot and started in the fall of 2018. Okay. And spent my first year at m MIT trying to figure out what I wanted to do.
Cause at that point I was still like, I wanna be in the nuclear space, but I have no clue what it looks like to have impacted here. Where are the problems?
[00:11:52] Mark Hinaman: I mean, being a PhD student, that’s like a pretty good opportunity. You’re basically, you know, you’re, it is, your advisor typically tells you like, go, go read the literature.
What does the literature say? Right? So you just gotta go read and learn kind of independently for,
[00:12:05] Robbie Stewart: but, but this is where, is really interesting because a lot of research. Um, at the academic and even national lab level, the timeline for impact is really long. Yeah, right. If you are studying salt chemistries or you are looking at developing new materials, so you wanna develop a new material that’s gonna, you know, allow reactors to operate for longer or higher temperature, it’s gonna have some cost or constructibility benefit.
You know, ultimately that material, you’ve gotta figure out what it is. And then ultimately you’ve gotta code, qualify it through ASME to get it through to the nrc. And that’s a, you know, that’s a 15 year plus or minus, usually plus five years timeline. And so research at some of these institutions, while it’s high impact, the timeline for impact is long.
And I was like, and I was looking at that and I was like, okay, I’m not sure that that is, Where I feel motivated to spend my time. I’m really glad there are people doing that because we need people thinking decades into the future. But it’s not, it’s not what personally motivated me to get into, to meet with them.
Yeah. So I was trying to figure out how do I find an academic research project that fits into more of an applied impact point of view, which are not always cohesive. Right? Like the, um, the, the interesting research questions that get you published in, um, You know, the, the good journals are usually sort of niche science questions where their impact on technology is in the distant future.
And so interesting to about
[00:13:33] Mark Hinaman: a dozen people right now. And they’ll be real interesting to, you know, thousands of people and 20 years from now, just like you said,
[00:13:39] Robbie Stewart: but Exactly, exactly. So, I just started, you know, asking questions and it doesn’t take long. You know, m i t had just finished a study, which I recommend to anybody interested in getting into nuclear called the Future of Nuclear in a Carbon constrained World.
And so maybe you’ve heard you’ve seen the study.
[00:13:55] Mark Hinaman: Yeah. 2018, right? Yeah. That’s a
[00:13:56] Robbie Stewart: pretty good report. I’ve read that one. Yeah, so they finished that the year I started, and the report has like a pretty obvious conclusion, which is there’s a big opportunity for nuclear, but if we can’t constrain cost and construction, then then, There’s actually not a big opportunity for, there’s no future, pretty much.
Um, and I appreciate that about the executive director of YA is, you know, he was very clear about that and every time he speaks about it, he makes that point pretty clearly, that the cost is the primary. There are other constraints in nuclear, there are regulatory burdens, there are customer questions customer takeoff questions.
There’s financing questions, but Nuclear continues to take 14, 15 years and cost 20 billion. There’s those other questions are in the noise.
[00:14:44] Mark Hinaman: I mean, it’s pretty practical when you think about it, right? Like if anyone has their capital, they’re not, if you have a dollar, right?
If you give me a dollar, your one question’s gonna be, when can I get my dollar back and how many more can I get?
[00:14:58] Robbie Stewart: Yeah. Like it’s not, and if someone says, I think six years, but maybe 16. You’re gonna be like, I’m gonna find another place for my dollar. Yeah, exactly. Yeah. Um, and so the focus on my research then, my PhD research was, what is the cost and construction timeline for advanced nuclear?
So what is it gonna, you know, there’s a lot of small modular technology out there being developed. The BWS 300 was I think, sort of the nameplate at the time in the lightwater reactor space. But more recently now, I mean the last month or so, you got the AP 300 from Westinghouse. And so the question was does small modular technology reduce cost overrun risk and does it reduce, construction, delay risk and advance construction schedules?
So that, that’s where I got to spend my, four years at mit. So I had a lot of fun working with Professor KK about on that question and. I think you still got, what was the answer? Well, I wish it was like easy to distill down, right? In a simple question. I would say, there’s a number of takeaways.
The one I like to give, is that if you look at the A w R, so this is the advanced boiling water reactor from G Hitashi first built in the 1990s by Teco and GHI Tashi in Japan. And the first ones is the first of a kind reactor. Took 36, 38 months to build I, which is an absolutely phenomenal performance.
Right. If we were building reactors in 36 month timelines, they would growth of the nuclear industry, oil, gas industry wouldn’t exist. Yeah. No. It would exist. They would just be using nuclear to power their petro can facilities and refinery, you know, like that’s true. We still need a lot of oil and gas.
Let’s be clear. We need a lot of fertilizer. We need a lot of you know, chemicals.
[00:16:43] Mark Hinaman: I’ll say the synthetic fuel industry would be much more prolific.
[00:16:46] Robbie Stewart: Yeah. That way. And then you compare that to the first deployment of the AP 1000 at vog, and it took, you know, 16 or 14 years I think is what it’s gonna final finalize out at.
And you’re like, okay, well what, what is the difference here? And if you do just a pure cost estimate. So this is like what I spent my first year doing. The A W R is a more expensive reactor than the AB 1000 concrete per megawatt. Safety grade steel like embedment, rebar, that, all that kind of stuff, more per megawatt than the AP 1000, but clearly dramatically easier to build than the AP 1000 is.
And so it sort of brought up, this question of constructability, which is, I did a lot of cost estimates. That was basically quantifying, how big are your heat exchangers, how big are your reactor vessels? You know, how thick are your concrete walls? But it takes an epc. To an engineering procure construction firm to come in and say, you have a lot of concrete, but it’s actually in places where it’s really easy to put concrete and you have a lot of nqa one welds, nuclear quality assurance plus one welds, but they’re in places that are really easy to do those welds.
Yeah. Um, as opposed to 45 meters off the ground, where it’s really complicated to do that well. So not only constrained by the supply chain of welders who can do that, but now you’ve gotta have them doing it. Um, Far off the ground. So the, the
[00:18:09] Mark Hinaman: or, yeah. Suspended. Can’t do it when it’s windy, right? Yeah. Like a crane.
[00:18:14] Robbie Stewart: So there it came to this question of like, actual cost as risk estimates are not something that there, there’s something we could do inside of an academic institution, but to really understand and, to design the future nuclear reactors that are gonna help this industry scale. It’s gonna take a civil construction focused mindset on the design side.
And some of that led to, a lot of that led to Boston Atomics.
[00:18:43] Mark Hinaman: Okay. Well, that’s a, that’s a pretty good segue. So what is Boston Atomics?
[00:18:49] Robbie Stewart: Yeah, so Boston Atomics, we’re developing a high temperature gas reactor. So this is, as far as reactor architectures or reactor archetypes. So be similar to what you see from.
X energy or U S N C or radiant on the micro scale. So this is, helium cooled or gas cooled reactor, or operating at high temperatures, coral temperatures in the six or 700 degrees Celsius range. Fueled by trio, which is the you know, the famous line is the DOE says it’s the most robust fuel on earth.
So this is basically, instead of. Fuel pellets that are in cylinders and fuel pellets are microsphere is covered by multiple layers of protection, which provides barriers to, to release. Um, and so that’s our reactor architecture type, and, that’s where we start from. But from that point forward, we make a dramatic shift.
So we sit, we basically say if you were to go to an EPC or a civil construction firm of any type and say, Can you build me a building that has X volume and is the easiest and fastest to build? You can imagine, and I don’t care what the building looks like, they’re gonna build you something that looks like a Walmart, right?
It’s gonna be low and height. It’s gonna be sheet metal. Sheet metal, steel, steel framed. Um, now unfortunately, there’s a lot of things about that we can’t do a nuclear right. We need a lot of concrete for radiation shielding to shield workers from moving on the plant. We need our walls to take different types of loads for earthquake response.
So we can’t, you know, it’s, it’s dramatically different. But the primary thing there is that they wouldn’t go much taller probably than three stories, right? They’re not gonna build you a skyscraper. If they can avoid it. They’re gonna go somewhere where they can, where land is cheap and they can build something long and flat.
And so we looked at the traditional HG G R architecture and whether it’s. The large sort of 600 megawatt thermal class hg or even the smaller ones in the, you know, hundred to 200 megawatt thermal classes, you’re still talking about structures that are between probably 35 and 80 meters tall.
[00:20:52] Mark Hinaman: Well, so that’s the Fort St.
Rain design. Right. So in Colorado, where I’m sitting, and we toured this a couple weeks ago, right? Like it, it’s an incredible plant and it’s still there. They have all the shielding there, but it’s 10 stories tall, right? It’s like a skyscraper, like, like a moderate size apartment building. And it is this huge concrete monolithic like structure that how is
[00:21:14] Robbie Stewart: the reactor?
Yeah. And you have to ask yourself, is it possible? To build a 10 story tall nuclear grade structure given the regulatory environment that we’re in and the complexity of material traceability on nuclear grade concrete and all of these elements doesn’t make sense for us to be building nuclear grade skyscrapers.
And, is that what a future scalable industry looks like? And so my co-founder and I re looked at that and said, no, that’s not the future. And I should give him all of the credit here. Because we were in a research project, we were talking about H DGRs a lot. And we had a number of faculty advisors who were like,
[00:21:48] Mark Hinaman: it’s high, high temperature gas reactors,
[00:21:50] Robbie Stewart: right?
Yeah. Hgr. Yeah. And we had a number of faculty advisors sort of recommending H C G R for this project we were looking on, and my co-founder consistently was like, with a mechanical civil background, mechanical, structural background. I, it’s like these are incredibly complicated structures, right? You’re talking about an 800 ton vessel.
That is 25 meters tall and its response under seismic loads is incredibly complicated. And the, you know, the constructability question here is highly uncertain and we need to think about a different paradigm. So he pushed back on our recommendation to, explore this technology further. And we would spend a lot of time, cuz in grad school you’re afforded a lot of time to sit around and sort of, kick the can and figure out what’s going on.
So every day for probably weeks, we would just sit in my office and argue about, You know what’s best. Does this make sense? And at some point in time, I think I just like under the pressure of like, I’m not a structural person, right? I’m a thermal hydraulics, sort of thermal fluids.
He transfer background person. I was like, all right, so tell me what the, like the three hardest things are here that you’re the most concerned about. And he’s like, well, I’m worried about the cross vessel. There’s this like vessel between a steam generator and a reactive pressure vessel, and I’m worried about how tall it’s, and I was like, well, Let’s eliminate the cross vessel and integrate the two and into like one long vessel.
And then let’s just lay the whole thing horizontally. So instead of a vertical structure now that’s, you know, 60 meters tall, let’s do something that’s 10 meters tall and we just lay it up flat so it could fit inside that Walmart. Um, and so we take this back to the professors, the faculty on the project we’re working on, and they ask all the questions that we had first thought of, which are operational questions.
So the reason these things are oriented vertically is because it’s. You’re operating the same direction as gravity, which makes a lot of your, operational actions. Easy. Yeah. You start doing things per particular gravity and you end up with a lot of cantilevers complex robotics that you need to do things.
Um, and so that has been the focus, of our work for the last few years, which is can you operate fuel, decommission, install, you know, all those questions, can you do all of the nuclear things that you need to do? With a horizontally integrated high temperature gas reactor that would exist in a much simpler structure.
So, that’s what we’ve set out to do. And we’ve got we’re part of a, the Advanced Reactor Demonstration Program, so we’re one of the tiers of funding there at funds. M I t a number of other institutions in ourselves to sort of look at these research questions and figure out can we put an HGR in an easy to construct and highly scalable building.
[00:24:22] Mark Hinaman: Got it. What’s your guys’ target size?
[00:24:25] Robbie Stewart: So the. Long-term target size is 200 megawatts thermal. Because our focus today is answering this question of feasibility, like, can this be done? We’ve limited that to say, okay, let’s try to get, instead of the 60 year, 200 megawatts thermal plant, let’s get a 10 year feasibility line to 150 megawatts thermal.
And if we can do that, then we think you can sort of unlock the future. But just getting to feasibility is complicated enough in a highly integrated system like nuclear, that you wanna really narrow down what you’re trying to achieve in the first pass. Or you’ll never actually get there.
[00:25:03] Mark Hinaman: Got it. So 150 megawatts thermal electric when Yeah.
Convert to electricity. I mean, that’d be 50 to 75 megawatts electric, which, yeah. In homes per megawatt. So 50 to 75,000 homes. Yeah. I received colleague recently that said, we should always add in how many homes we can power or some other metrics, right? So yeah,
[00:25:25] Robbie Stewart: people love to think about it in terms of homes. Yeah, which I is a helpful metric, I think cause yeah, most people have no idea what a megawatt is. I think about
[00:25:34] Mark Hinaman: in gallons of diesel, but yeah. Yeah. Okay. Um, that’s awesome. So the horizontal piece is kinda your guys’, would you say your primary thesis from where you’ve differed from? The norm.
[00:25:48] Robbie Stewart: Yeah, that’s exactly right.
And as well as integrated. So the inside the hgr, you have two main primary system components. You’ve got the pressure vessel and then the steam generator, which is just another word for heat exchanger. And they’re traditionally offset from each other and connected by a cross vessel. Um, and that cross vessel has, you know, we can get very technical here, but it has some challenges with the installation of it.
How do you get it into place? How does it handle loads? Thermal gradients, how does it thermally expand as it’s constrained by these other two pieces? And so we basically thought, looked at all that and we were like, well, let’s just get rid of it. Basically our thesis is every time something looks complicated, can we eliminate the system through some other more simple system.
And just repeatedly doing that until we achieve something that’s feasible.
[00:26:33] Mark Hinaman: Got it. Cool. Okay, so there’s, lot to dive in there and a bunch of different directions that we could take it. So, I guess from a practical standpoint, when I hear you say, let’s just eliminate things, I mean, a lot of components are in systems.
For a reason. And sometimes when you the classic example is right, a kid takes apart a machine and then puts it all back together and ends up with extra parts. It’s like, well, the machine kind of still works, but then there’s an error at some point down the road. Right? Like, is there any concern about that?
[00:27:01] Robbie Stewart: No, absolutely. So this is again, where I give so much credit to, our C t O and co-founder Enrique. He has done a phenomenal job of digging into the actual why’s behind so many of these systems. He’s, read through so many design and licensing research papers on the H T R P M, which is the Chinese H G G R, which is the only H G G R built in the last several decades.
Around the world. And so these are the folks that know how to do it. Unfortunately we don’t know how to do it in the us. They actually know what, the thermal expansions are on this cross vessel. They actually know these, how difficult it is to manufacture the steam generator, the way that it works and the materials required.
Cuz they’ve built and tested and now are operating one, or at least they did hot functional testing. But, that’s a great question. How do you know what you’re doing? And what you’re eliminating is okay. And I think the classic example here is if you show a civil engineer, a nuclear reactor building, they’ll say, we can get rid of 90% of this concrete.
And people correlate concrete volumes to cost, right? So that it’s pretty common to, for folks to look at the amount of concrete and say, this is directly correlated with how expensive this plant is gonna be. And so they’re like, oh, we don’t need all this concrete and. But there’s integrated reasons for why you need everything, right?
In many places, you don’t need that concrete for structural purpose. It’s actually serving a radiation shielding purpose, right? And so it’s allowing your workers to move in certain parts of the plant. And so, you really have to start to build out an understanding of these systems, their integrated effect, why they operate the way they do, why their performance specs are what they are.
So not even that they do a certain action, but that they do that action in a certain time with a certain reliability. And why that is. And you have to then back out like, okay, how can I meet that same set of specifications with the different system? And we push boundaries on that on in a lot of ways because like for example, we have one system, it’s called the reactor cavity cooling system.
So this is in an accident. Say we lose offsite power, like in the Fukushima style accident, how do you remove the decay heat from your plant? What’s the system to do this? And in a traditional H G G R, this is looks like a series of radiator panels. So this is like tubes with sheet metal welded to them that you’re running water through.
And the heat is hitting these panels through convection and radiation, and that’s how you’re getting rid of all your heat. And the water tank for these things is really high above the reactor so that you can get a lot of pressure to drive your natural circulation loop. You want to get a lot of like heated length to drive a natural circulation loop.
So in a, because traditional related yards are vertical, you can get, 20 meters plus of head to drive your Now after circulation, Enrique was like, well actually I think we only need like three and a half meters of head. That’s why that’s a lot lower. And if you show that to someone who’s only seen the 25 meter head version, 20 meter head version, and they’re like, I don’t know that you’re gonna get the performance you need.
And you’re like, okay, well, let’s analyze it. Let’s get all the way into the details and let’s show that it can perform exactly how we, need it to perform. It works, right? So we’ve had Argonne looking at this as part of our a RDP our 20 grant work, and they’ve shown that it’s gonna work perfectly fine.
It’s gonna remove exactly as much heat as we need it to, and there’s not nothing to be concerned about. And so, it’s sort of shifting. It’s like, that’s not that you need 20 meters of heated length or, and pressure head. It’s that you need a certain performance of heat removal and let’s focus on only providing the necessary spec.
Which it takes a lot of work to drill down into those details.
[00:30:28] Mark Hinaman: Yeah, that’s fascinating. I think about it as know what you’re solving for.
[00:30:33] Robbie Stewart: Yeah.
[00:30:34] Mark Hinaman: So the digital twin you mentioned earlier and how that, you said that was GE was kind of at the forefront of that in 2015.
Are you guys utilizing any of that in your current development? Can you describe what a digital twin is and how they’re useful?
[00:30:50] Robbie Stewart: Yeah. Yeah. So I a bit more about a digital twin. So basically a digital twin is trying to use operating data, in the reduced order model, really simple, whether it’s. A regression, right?
You’ve eliminated all physics and it’s just a simple linear regression or statistical regression of some kind, or maybe you have some really simple physics model. So instead of doing a full FDA plan element analysis of a component, you’re doing a really basic linearized stress analysis, and you’re running that stress analysis or that heat transfer model or whatever it is for every flight.
This is what we’re doing at ge. You look at. This component has flown 1,800 times between, Atlanta and London. And we’re gonna look at how hot the air was that day, what the gross weight of, the aircraft was. All of the data that you can get, how fast the takeoff was. The RP at the engine is what the, the, you know, combustor exit temperature was all stuff.
You’re gonna use all that data. To slowly model the cumulative damage done to that component. So how is that component accumulating damage over time? And then you’re gonna use that simplified model to inform maintenance decisions. So you’re gonna say, based on this, we’re actually gonna inspect this engine earlier, or we’re gonna postpone inspection, or the FAA requires this inspection, but our model tells us the component’s probably fine.
So we’re gonna do the inspection, but we’re not gonna plan to have to take it off wing for any maintenance because we know that we’re gonna be, it’s gonna be okay. And that lets you do some sort of global operations optimization and maintenance scheduling that has huge benefits when you have tens of thousands of engines in the fleet like GE does.
Yeah. Um, so for us, it, we are so early in our design phase, so now shipping to Boston dogs, we’re so early in our design phase. That thinking about lifeing in that standpoint, and modeling in a reduced order way or estimating cumulative damage based on operations is, not at our forefront. And there’s a few systems that folks look at for condition-based maintenance and nuclear.
The huge difference between what GE did and how they were able to capture so much value versus the nuclear industry is the variability in operations. Okay. If you’re operating a flight between Atlanta and London and New York and Dubai, those two flights are gonna generate significantly different amounts of damage or deterioration to components.
But a nuclear plant, if all things going well, you’re operating it in the same place constantly without changing the power level for 18 months to two years at a time. You’re shutting it down for two to three weeks, you’re starting it up again and you’re doing the thing again. And so there’s not changes in environmental parameters, and there’s not huge swings and power levels, so the accumulation of damage looks a bit different, and the way to extract value from something like a digital twin then looks a bit different.
And I’ll be honest with you, I don’t have my mind perfectly wrapped around how exactly you capture value from something like that, but, I know folks around the industry are looking at it.
[00:33:52] Mark Hinaman: Well, it sounds like you at least know what you don’t know. So let’s shift back to the size question.
Why did you guys target the size of reactor that you did, and, why not something smaller Or maybe are you thinking about starting with something smaller, for testing or prototyping?
[00:34:04] Robbie Stewart: Yeah, yeah. This is the usual question. So, why do we not go smaller or why do we not go bigger? So we started bigger.
The very first. Hand drawn we had of this thing was 350 megawatts thermal. We basically took the General Atomics Modular H G G R, which was first designed by GA in the 1980s, and we were like, let’s turn this one horizontal. This is the most developed patents
[00:34:30] Mark Hinaman: expired. Yes, please. Here we go.
[00:34:34] Robbie Stewart: And the challenge then is that you start to get into vessels that are huge.
HGR are notorious for a low power density, so the amount of power per volume inside the core is low, which has great safety advantages unless you do a lot of things passive with passive safety. But it means the size of your components are really big. And even with that GA 350 megawatt vessel, you’re still talking about a vessel that is, Six meters in diameter, which is enormous.
You think about six meters like this is, basically you’re only transporting this thing by barge. You’re not really gonna get this on a road most places. And if it is on a road, it’s a short trip from a coast or a river to Yeah.
[00:35:14] Mark Hinaman: Something that size typically would be manufactured onsite?
Yeah, exactly. Like you said, you can’t transport the modular components necessary. So I guess that fits in well. Is that the modular in your guys’ name or acronym, right? Or you got the Mir reactor?
[00:35:29] Robbie Stewart: Yeah, the Mire reactor. So, modular fits right in there, which is we want to be as we wanna be able to shift as much of the onsite work offsite.
And so if you’re talking about a large vessel, you’re not doing that right? So you want to be sort of in the realm of things that are highly transportable and easy to modularize. In fact, if you look at, Civil construction modularity. Most buildings that are constructed with modular technology are of low spans and low heights.
So these are buildings that are less than 10 or 12 meters tall, and then less than maybe 15 meters in span. Because that is what the construction industry, and so you think about that like, well, what is that? What are the buildings that sort of fit that paradigm? And it’s mostly like office buildings, multi-family housing, that kind of stuff.
This is where people talking about modular construction and the impact it’s had on accelerating the construction industry. It’s in that side of things. It’s not in enormous structures that are 45 meter spans and 60 meter heights. Um, yeah. And, so we wanted to fit everything sort of within the modularity.
But your question on power level, One of the other conclusions of my thesis was that the economy of scale is still a primary driver of cost. Smaller reactors do reduce risk, and a lot of the ways they reduce risk is that the cost of an overrun for a 1 billion reactor if it were overruns, a hundred percent is only a billion dollars.
And the cost of an overrun for a 10 billion reactor if it overruns a hundred percent is 10 billion. Yeah. But the sequential learning you can do, where if you cost over on the first one in a billion and then you cost over the second one at, 300 million, and then you cost over on the third one at, 50 million, the cumulative effect of that is how you reduce the total risk to a sort of a sequentially deployed set of reactors.
But the risk of the first one is still pretty high. The risk of construction delay and cost overrun, even for a smaller reactor at the first one is still pretty high. And so the economy of scale is still a primary driver, for total cost, you want to do things bigger because you wanna sort of reduce the, total cost.
And so when we looked at going smaller, like to the micro reactor size, it’s really hard to do a bottom up cost estimate of a micro reactor, and say this is gonna be. Competitive with the SMR class of reactors. And so, our thesis is to aim for the site work and construction profile of a micro reactor, but with the power level of an smr.
So small modular reactor, 200 megawatts thermal, like that range. And we think that’s like the right balance between, constructability and construction risk as well as the economy of scale. Now I think that the, huge caveat to that, which I have to give cuz I spent a lot of time in the costest estimating world in nuclear.
Is that road transportable like roll off the factory style micro reactors as a completely different operating paradigm for the nuclear industry. And I have no sense of what they will cost or, what levels of scale they’ll reach and regulatory questions, but that they’re sort of completely outside of the traditional economy.
Something they have a
[00:38:21] Mark Hinaman: different customer too. Right. That it sounds like your guys’ thesis, and the problem that you’re solving is still truly the decarbonization utility grid. Scale size. Like, you’re selling electricity, you’re selling electrons and grid. Many of the micro reactors aren’t, they’re selling dispatchable, portable power.
[00:38:41] Robbie Stewart: Yep, exactly. Yeah. And, I’ll add one piece to, our customer market, which I is why I really love what y’all are doing at fire division, which is the industrial heat market. Because the beautiful thing about nuclear power is we produce heat and we use that heat to make electricity. And unfortunately that’s not a perfectly efficient process.
I mean, I don’t know how many folks know this, but even for an HT G r, which is a high temperature process, maybe 40% of that heat gets converted to electricity for a light water reactor, which is what is operating all around the world today. It’s like 30, 33% of that heat gets convert to electricity.
Well, if we could sell that heat directly, and there’s a lot of folks in the oil and gas industry too that need heat, well then you’ve lowered your cost per energy by, a factor between two and three. And that’s a really competitive product then in a decarbonized economy where maybe the other option for a petrochemical plant to decarbonize is electrification.
So they’re gonna fully electrify whether it’s a fire box or, some other system on site to make this heat. And if that’s coming from a solar wind battery, hydro grid. A non-nuclear supported grid that electricity is, it’s electricity to heat is one to one, but nuclear tahi is, one half to one third.
So, the cost advantage there is huge. Um, in fact, last week RPE hosted a great workshop between nuclear folks and pet chemical, um, oil and gas industry. And there were a number of representatives from Exxon Shell bp, B A S F, Eastman Chemical. Dow Chemical there because they recognize like if they’re gonna decarbonize, their plants and they need steam, they need 1600 pound steam, then a great source for that is nuclear because of, the cost advantage there.
[00:40:28] Mark Hinaman: Yeah. You know what comes to mind immediately for me, and, this is just cuz it’s close to home, but I mean I grew up in Northwest Colorado and there’s a ton of um, Oil shale there, that’s not as mature. So the Kein type is less mature. So, if you add heat, it will mobilize and be able to, be moved.
And I mean, people have been working on this since like the eighties. And my dad growing up would be like, cuz he is a geologist. Oh, there’s, we’ve got more oil south of town, 50 miles, a hundred miles south of town here than Saudi Arabia has. It’s just, if we just added heat then we could do it.
And it’s like, Well, there you go. Here’s a way to do it, right? You’ve got an industrial plant, you, pump some heat off nuclear reactor on the ground, and now that’s not necessarily productive for decarbonization, but for mining the raw material, which is getting the hydrocarbons, which are super valuable, use ’em in everyday life.
Um, yeah. Accessing them, it’s still, it’s probably cheaper than direct air capture and recreating them out of air and water, right? Like
[00:41:26] Robbie Stewart: yeah. We’re gonna continue to need oil and gas products, right? We’re gonna, plastics doesn’t go anywhere. Fertilizer’s not going anywhere. Chemicals pretty cool.
[00:41:37] Mark Hinaman: They’re really useful. Yeah. Open ’em in the ocean, but like, man, they’re so useful.
[00:41:43] Robbie Stewart: Yes, we should be more careful about our, where we dispose of and how we end life, these products. But I mean, fertilizer has been, an insane benefit to the global. I got civilization, right? This is enabled life longevity and food security that we’ve never seen before, and it’s requires hydrogen from methane.
[00:42:07] Mark Hinaman: Couldn’t agree more. So, let’s scale back a little bit. The RDP Grant advanced reactor demonstration program. Let’s chat a little bit about this. What is it, for those people that are ignorant about it? How’d you guys apply? What, what was the process? Funding size?
Yeah, let’s just talk about it bit.
[00:42:22] Robbie Stewart: Yeah. So when we first had the idea, this was October, 2019, and we sort of were kicking the idea around. Enrique and I were trying to figure out what to do, and then a r DP was announced, I believe it was that next summer. Or maybe it was later, that spring, I don’t recall exactly, and so.
We, as we were talking about it, Carus Sherman, professor of N Y T and the PI of our a R DP grant. Got excited about it as well and he was like, Hey, this is a great idea. I’d love to do some more research on this question and look at this. And he went out and put a team together cuz he had the connections.
Again, at this point in time, I’m like two years into my nuclear space. I don’t have the networking connections to do this. And so Carus goes out to University of Michigan and finds a few faculty there. University of Buffalo, argon National Lab, NPR Associates. And build a team and set and a proposal to say, Hey, constructability is gonna be key to scalability for nuclear.
This horizontal integrated design has a lot of potential. It has these questions, and we’re gonna spend three years looking at these questions and we’re gonna do it with researchers from these universities in this national lab. And so we supported him in the writing and creating of that proposal, but he gets all the credit as PI, for putting it together and putting the team together.
And so submitted the proposal or awarded the proposal. I think we were the first A R D P awardees to start. Universities are sometimes incredibly slow at contracting, but in this case, they were faster than everybody else that they won awards. So first to start and I think we’ll be first to finish, we finished summer of 2024 at three year program.
And it’s been a really useful scope of work for us. I mean, we talk about everything from seismic analysis at the University of Buffalo to how our fuel blocks are gonna move during an earthquake to radiation shielding requirements At m i t to the, you know, reactor cavity cooling system I described at Argon.
[00:44:15] Mark Hinaman: Generate your design and answer a lot of the unknown questions that you had about your design and leverage the capabilities at the national labs and universities because I mean, many people may think, oh, well we’ll just go to the national lab and it’s a national lab. They should be able to work with us and work for us for free.
Right? There’s sometimes a cost share and you gotta pay them, to do stuff, cuz I mean, in reality there’s still human. There that are getting paid and they’re, they’ve got a paycheck and,
[00:44:40] Robbie Stewart: yep. Exactly. And that also, I mean, to get to your earlier question about like, how do you know when you’re eliminating or changing a system that you’re capturing all the effects, building a team like that, a consortium of experienced, nuclear folks has huge advantages because they have, everyone has their own experience base to say, oh yeah, in the last 15 years, here’s how this particular system ended up looking the way it does for the traditional h gtr.
And we’re able to go to each one, each one of these individual experts who has, their niche knowledge base and say, how do we then shift, that set of requirements.
[00:45:13] Mark Hinaman: Got it. Yeah. That’s awesome. So what’s, what’s your out guys’ outlook for the next 1, 2, 5 years?
[00:45:20] Robbie Stewart: Yeah. So, like I said, our A R D P grant finishes next summer.
And our goal by, that point is to achieve sort of like basic feasibility. This is kinda what I described earlier. Can we do all the nuclear operations that we need to do, achieve. All the safety levels at sort of, the pre preconceptual level, right? This isn’t a preconceptual design stage. This is pre preconceptual, just sort of getting all those pieces together.
So that’s where we want to end up next year that we could, at that point, it’s like confidently, like this is worth looking at further. We can achieve all the things that, we think we need to achieve, but there’s more work to be done. At that point in time, it’s probably another three to four years to get to a preconceptual conceptual design point where, okay, now we’re adding a lot more detail.
We’re adding a lot more integrated. Right now because the pace of development’s changing so much, a lot of teams are looking at different versions and time of what the whole system looks like. And at some point we’ve gotta like bring those together. We’ve gotta like integrate all those pieces and we sort of have how that’s gonna work on paper.
But doing the actual integration piece there, in the horizontal orientation, the biggest questions like technical questions are refueling. So how do we get fuel blocks in and out of our core? And reactivity control. So how do we do, our power control levels and shut down reactor? And that’s because refueling is really easy from above.
It’s like a fishing hook, right? You come down and grab these blocks. So we’ve spent a lot of time designing a robotic refueling machine. So the next few years, what we wanna do is prototype that at the small scale. Prove that it can perform the functions that it needs to do. That’ll be a big technical milestone.
And then this activity control. It’s really hard, rods go vertical. Gravity helps you get them in fast. If there’s an accident to shut down your reactor, if you gotta shove those things in horizontally, the whole system of how that looks, how it slides into place, do you do controlled drums, which instead of inserting a rod to shut down your reactor, you’re rotating a large barrel.
To move neutron absorbing material closer to your core, which shuts down the core. And so you gotta prove some feasibility questions that, so we wanna do some prototyping here and there, and that’s how work over the next three to four years, long term, our deployment timeline is in the 2030s. So, we think, like spend the next, six or seven years doing as much as we can with grants, sort of non-dilutive funding sources, building out the proof of feasibility here.
And then we get to a point where, We de-risk the technology sufficiently that it makes sense to bring in private sources of capital from a risk standpoint to come help us build a few demonstrations non-nuclear at first, because it’ll be lower cost and you can still understand a lot of the constructability advantage that we’re claiming and then getting to a first nuclear plant in the 2030s.
[00:47:53] Mark Hinaman: Cool. You answered my next question. That’s good. Um, okay. What do you see as the, some of the biggest, well, let’s just pick one cuz we’re coming up on our time, but, what’s one of the biggest risks that you guys face in front of you? And I know that there’s a bunch you mentioned earlier on like, Well, this is not an easy problem to solve.
In fact, when you look at all the venture capital funding recommendations they say no hardware fast time to deployment, lots of customers, right? And if you think about nuclear, it’s like it’s tons of hardware, huge designs, lots of capital. Slow deployment, heavy regulatory burden, right? Like picking the hardest problem to solve.
[00:48:31] Robbie Stewart: That’s exactly right. That’s exactly right. And there are venture funds and we should give them credit, right? There are a lot of venture funds moving into hardware. There’s a lot of climate focused venture, but
[00:48:41] Mark Hinaman: the software thing felt like a bubble to us. Mechanical guys, very nice guys. But how does this actually add value?
I don’t really understand.
[00:48:47] Robbie Stewart: I mean, one of things I wish is that we had a climate focused 10 years ago when interest rates were gonna be so low for 10 years, and we would’ve been able to do a lot of different stuff, unfortunately. Right now there’s this big fire in Canada, a number of fires in Canada that are pushing smoke all across the us And a few years ago there was the big California fires and there was, the, sort of the lore in climate venture capital is that it was these California fires that were dropping particular matter on cars in the Bay Area in San Francisco and, the orange and red sunsets and the don’t go outside orders.
That we’re all of a sudden venture capitalists, so we’re focused on software. We’re like, actually climate seems to be a problem now, now that we’re like experiencing this directly from this, these buyers. Yeah. So there is a lot of climate focused investment. The timelines are still short.
They’re typically focused on one system. Right. You’re like, oh, we’re gonna do carbon capture and it’s gonna be one mechanical or chemical system. Right. A nuclear plant is, 30 to 150 different systems of that level of complexity. Yeah. And so putting all those pieces together in a venture fundable or ventable way is very difficult.
And so we are actually really grateful to be supported by the Activate Fellowship. So this is, a decarbonization. Or it’s actually just a, general supporting technologies that will be for the public good. And one of their public goods is climate change, and it’s a two year fellowship that provides us with money for stipends and research money and as well as access to, their network.
And so this will fund us for the next few years, and we’re really grateful for opportunities like this that. Like activate that support companies trying to solve big hard problems that take a long time and need sources of capital that are very patient. Got it.
[00:50:29] Mark Hinaman: So I guess, is that your answer? The funding piece is one of the hardest Yeah,
[00:50:33] Robbie Stewart: I would, yeah.
Yeah, exactly. The funding piece. Yeah, about
[00:50:39] Mark Hinaman: incentives drive the world World, right? Like when am I gonna get my dollar back? And if you’re like, well, it could be 16 years, right? I mean, if it’s the mid 2030s, it’s 2023 today, then yeah, that’s a long time.
[00:50:49] Robbie Stewart: It’s a long time. I’ll say that.
You know the one piece on customers that you mentioned at this ARPA-e workshop last week, all of the, not all, but a number of the petrochemical companies and our, oil and gas companies were like, we would be excited to deploy nuclear reactor at one of our facilities, but we want to go and touch it. We need to go and see it built.
We need to see it operating. I wanna send my engineers to go walk around it. And the question I think everybody had in the room, and this was asked by multiple folks. Is, how do you get the funding? How do you get the capital to go build those demonstrations so that you can then walk a customer around it?
And this is where a R DP is really a phenomenal program, and you have to give all the credit to the think tanks that architected this from the New Innovation Alliance to Third Way to Clear Path. All the folks that sort of put this together to say, Hey, we’re gonna demonstrate these technologies. This is something that is really valuable.
They deserve a lot of credit, and I think we should keep doing demonstrations. I think that the value of demonstrations is Funded by the government or maybe through loan guarantees from the government that sort of offset the capital risk, for other folks they want to bring money in. But, that is how we unlock the future of this industry is we keep demonstrating technology physically.
[00:51:59] Mark Hinaman: I couldn’t agree more. There’s nothing better than going out and building it. I mean, you know, from working on jet engines, right? Like, yeah, you can have a lot of ideas about how something’s gonna work, but so you actually screw things together and make sure that they fit. Like,
[00:52:11] Robbie Stewart: yeah. I get you so far.
[00:52:15] Mark Hinaman: So, where do you see all this going? Robbie? Leave us on an optimistic note. What’s the world look like in 5, 10, 20 years?
[00:52:22] Robbie Stewart: Yeah. Five years. Fortunate, I think a nuclear pipe looks very similar. Hopefully we have a number of reactors under construction, right? I think that would be the most exciting piece.
If we have concrete getting poured at multiple sites in five years, I think that’s a huge win and something we wanna celebrate. In 10 years, hopefully those first concrete pores have multiplied, right? As we’ve had successes at some of these first demonstrations, they’re gonna be challenges. We should own up to the reality that there’re gonna be challenges.
We shouldn’t overpromise how these first plants are gonna perform. This is new tech that we’re trying to build, but once it’s up and running, we should be excited about the potential, and scaling it. So I think in the, 10 to 20 year timeline, I’m hoping that we’re building reactors multiple sites, both for power generation, electricity generation, and heat and steam generation, all across the us.
And then I hope that the export market starts to take off. I hope that the US. Through, whether it’s the XM bank or other, sort of strategic imperatives, we’re able to export US technology, nuclear technology around the world to decarbonize the global economy.
[00:53:20] Mark Hinaman: I love that vision. It’s gonna be people like us to go and do it, so let’s go do it, man.
[00:53:25] Robbie Stewart: Yeah. Hey, we’re trying.
[00:53:28] Mark Hinaman: Awesome. Robbie Stewart, this has been great. Thanks so much.
[00:53:30] Robbie Stewart: Thanks for having me.