MIT and DOE Work on Lower Cost Nuclear Reactors Using AI

The US Department of Energy is working on advanced nuclear reactors that have lower manufacturing and operating costs. The Generating Electricity Managed by Intelligent Nuclear Assets (GEMINA) program as $27 million in funding. GEMINA is accelerating research, discovery, and development of new digital technologies that would produce effective and sustainable reductions in O&M costs.

MIT-led teams will collaborate with leading industry partners with practical O&M experience and automation to support the development of digital twins. Digital twins are virtual replicas of physical systems that are programmed to have the same properties, specifications, and behavioral characteristics as actual systems. The goal is to apply artificial intelligence, advanced control systems, predictive maintenance, and model-based fault detection within the digital twins to inform the design of O&M frameworks for advanced nuclear power plants.

There are three MIT teams and six non-MIT teams in the GEMINA program.

1. NSE professors Emilio Baglietto and Koroush Shirvan will collaborate with researchers from GE Research and GE Hitachi. The GE Hitachi BWRX-300, a small modular reactor designed to provide flexible energy generation, will serve as a reference design. BWRX-300 is a promising small modular reactor concept that aims to be competitive with natural gas to realize market penetration in the United States. The team will assemble, validate, and exercise high-fidelity digital twins of the BWRX-300 systems. Digital twins address mechanical and thermal fatigue failure modes that drive O&M activities well beyond selected BWRX-300 components and extend to all advanced reactors where a flowing fluid is present.

2. MIT Principal Research Engineer and Interim Director of the Nuclear Reactor Lab Gordon Kohse will lead a collaboration with MPR Associates to generate critical irradiation data to be used in digital twinning of molten-salt reactors (MSRs). MSRs produce radioactive materials when nuclear fuel is dissolved in a molten salt at high temperature and undergoes fission as it flows through the reactor core. Understanding the behavior of these radioactive materials is important for MSR design and for predicting and reducing O&M costs — a vital step in bringing safe, clean, next-generation nuclear power to market.

3. A third MIT team will work closely with the Electric Power Research Institute (EPRI) on a new paradigm for reducing advanced reactor O&M. This is a proof-of-concept study that will explore how to move away from the traditional maintenance and repair approach. The EPRI-led project will examine a “replace and refurbish” model in which components are intentionally designed and tested for shorter and more predictable lifetimes with the potential for game-changing O&M cost savings. This approach is similar to that adopted by the commercial airline industry, in which multiple refurbishments — including engine replacement — can keep a jet aircraft flying economically over many decades.

Written By Brian Wang

42 thoughts on “MIT and DOE Work on Lower Cost Nuclear Reactors Using AI”

  1. If you are talking about the Indian Point nuclear power plants I think they are still running. But they are at least 20 years pass their designed lifespan. And the company that runs them wants to run them another twenty years. They are also locate upstream of NYC and in a relatively populated area. Shutting them down is a sensible idea.

    As for keeping COVID patients in the nursing homes, I am not sure Cuomo was aware of the magnitude of the problem. Something about a lack of testing supplies. He was also busy dealing with the overflowing hospital ICUs.

  2. The mirrors could have to be high enough to illuminate the backside of the earth but don’t have to be GEO. And each mirror would only serve a few earth station so they didn’t have to skew that much. I think the biggest problem was the glimmer. But you could shut and open the mirror segments so that only the earth station would receive the light.
    The area lit my each mirror would be large but the accumulated light should only cover a few hundred sq miles.

  3. I hope not, I don’t want to see SpaceX having a monopoly over orbital launches.

    Yes, they are cheap now, but they won’t remain cheap once they have a monopoly. Look at Tesla and their full self-drive module, Elon Musk says that its price is going to be above $100,000, or look at how he wants Starlink to finance a fleet of a thousand Starships to go to Mars, these are not reasonable margins on a healthy market, these are milking their costumers for all they are worth under a monopoly.

    We need old-space (NASA, Boeing, Ariane, …) and new-space (Blue Origin, …) to keep Elon Musk honest.

  4. As a general rule we only criticize grammar and spelling if you do it in the same sentence as calling someone else ignorant and illiterate.

  5. Don’t worry. There won’t be any more ‘Old Space’ mega-projects. SLS represents the last hurrah of the Space Caucus. Now that there is a better and much cheaper way to send astronauts to the Moon, most of the heavy lifting will be done on-the-cheap. But we did not have that option 20 years ago, when we began considering the US space program post-Shuttle (a bargain?). All we need to do now is hope that Musk will get his system flying and is willing to build us a few that are NASA-rated. Either that or we (the United States taxpayers, through our Congressional representatives) could hire non-government contractors instead of astronauts for future projects. Or we can give up on a national program altogether.

    I still want SLS to succeed, Money-pit or no.
    We still need a “national” space program.
    Yes- they are pigs at the trough. That’s politics.

  6. Using Musk’s best cost estimates and halving the needed square footage of panels down to 10,000 square miles to power the U.S. (well over 20k for ground based), I calculated a cost of well over 7 trillion. Where is this cheap/free energy of which you speak?

  7. It’s a pity because, generally speaking, I find Nextbigfuture a very good scientific and technological site. (Sorry for eventual grammar mistakes, I’m Italian).

  8. Good question. I think this a personal beef. Perhaps they scorned Brian at some point in the past. I’ve asked this very question several times in the comments – always crickets.

  9. Moon has advantage in that telebots can be operated from Earth, as does all cis-lunar, where much can be built, and is far easier than Mars in many ways.To avoid common question, by O’Neill I consider Al Globus ELEO to be well down the path to O’Neill Space, depending mostly on how ISRU is advanced. Limiting O’Neill to Island 3 plans, where the “3” refers to generation, not SN, leaves an impossible first task, obviously, so is not the plan! That starts with Space Solar and ISRU in a huge project, far dwarfing all other current or past efforts, but paying the rent too!

  10. Unless someone stumbles across the secret sauce for something approaching a human level AI, even if it only reaches the cognitive capacity of a 5 year old, i would be shocked if construction on an O’Neill cylinder commences prior to the end of this century. I expect it to take that long to get any meaningful infrastructure built up outside of near earth orbit.

  11. European Space Agency(ESA) budget is < 7B€, and that’s 22 European states. I have a hard time seeing them take on such an effort by themselves. I can see China doing it. Japan would definitely have an interest in this, but i think they lack the resources. 

    Perhaps an ITER 2.0 effort?

  12. “Its benefits have been known for some time; in 2008, the Pentagon’s SBSP Study Group concluded that “space-based solar power does present a strategic opportunity that could significantly advance U.S. and partner security, capability, and freedom of action and merits significant further attention on the part of both the US Government and the private sector.” More recently, in 2016, when the Defense Department, the U.S. Agency for International Development, and the State Department held a contest for the biggest ideas that could simultaneously advance diplomacy, development and defense, space solar power swept the awards.”
    -Why the new X-37 space plane mission is a big deal Peter Garretson Politico May 15, 2020, 5:45 AM CDT

  13. The government will not be permitted to fund this, it will convert the existing winners into losers. If this happens, coal, nat gas & oil largely dies, don’t expect anyone to giving up that cash flow.

    There is more than one government on this planet. The USA, Russian and Australian governments might like coal and gas but Germany? Japan? China?
    (Oil isn’t affected by solar power, even when challenged by EVs it’s battery costs that matter to oil).

  14. I have realized that we are using 2 defs of investment, totally my fault! An older version (China) stated that the break even would be at around $5ooB, but that was only 500GWe! The current doc is stating totals, with .8T each Moon and Earth rectennae, say $2T cost, but financed mostly by profit, not *investment*. Even so, this needs re-work in light of all the new electronics, rockets promised, other Lunar activities, strong gov support for Space Solar(see the uncrewed AF orbiter story). Oh, and global heating.

  15. I’m glad you weren’t in charge of Edison’s research budget. We’d be reading these comments by candlelight.

  16. “If this happens, coal, nat gas & oil largely die” as does global heating! As long as the tech looks good to you, I will let the money people fight over who gets the rewards. However long it takes to get O’Neill Space going. There are all sorts of other ways to get started too, so this is just the one that has looked best to me since ~’90. And that is starting from favoring GEO solar sats as O’Neill proposed, a correction on that. As long as LSP is considered, I am happy with the result. I am not happy if it is merely ignored for reasons covered in the proposal, such as “the Moon has a day/nite cycle”.

  17. 800B$ (2000$) = 1224B$(2020$)

    The government will not be permitted to fund this, it will convert the existing winners into losers. If this happens, coal, nat gas & oil largely dies, don’t expect anyone to giving up that cash flow.

    You just can’t say to investors 1.2T$ in cost gets you +80T$ in revenue, it’s not that simple. Anyone looking at the CapEx and ROI will toss you to the curb and invest their money in PV + storage or combined cycle power plants with carbon sequestration.Nuclear(fission) has the same problem, theirs is orders of magnitude less than space solar and they still can’t get funding without a public backstop.If it takes your project many years to get the cash flowing, you’ll need to be able to service your debts over that time. Nuclear projects are 5-10B$ and take ~5 years to build, financing that is too much of a gamble for any private entity without public funds; that is why Nuclear is nearly dead in the US. If 5-1B$ is such a problem financing, 1.2T$ is a practical impossibility given unknowable time lines of developing a new space based industry from scratch. Perhaps such a plan can be implemented if there was pre-existing space industry and you didn’t have to build it yourself.

  18. I don’t know if it is wasted money on research. I’m not qualified to comment on most of this, my math ends at spreadsheet columns and binary 1’s and 0’s. I will say that when you put researchers together working on a project that doesn’t work, the magic of serendipity and smart people banging their heads together can lead to unexpected and successful outcomes. As a taxpayer, I have no problems funding scientific research, more than I support it for most of the other waste projects the government gets involved in and shouldn’t. I agree with you on smart spending of result driven research dollars, but any knowledge we didn’t have before is good knowledge. As long as everything is well documented it adds to our scientific knowledge.

  19. Those who wish to make many many $Ts selling many many TWes to the World, solving global heating and opening O’Neill Space to all’s future. Some smarty pantserati types, Bezos comes to mind. A no-brainer for energy companies. The $1.2 T is unfamiliar to me, unless you are *real* izing the $.5 T Criswell states. Needs new estimate, now with Musk rockets planned and much cooperation possible with other Lunar activities now seen as useful. Much cheaper! Perhaps the money that would go to nukes? (edit: investment will be much more than initial start-up investment, as most of the investment will be internal, from profits, after the “break even” mentioned. Throw all these numbers out, they do not know about Musk rockets or other Lunar needs that can contribute. LSP may just grow from cells on the Moon used to produce fuel for Mars!)

  20. The barrier to doing useful things is almost never physics, it’s almost always the vicissitudes of commerce. 

    Who do you expect will be willing to invest 1.2 T$ to create the infrastructure to manufacture and support Criswell’s vision?

  21. oh. oh. oh. its happening. Will the 50-ton ITER tortoise, slowly lumbering its way to the ‘commercial fusion power’ finish line beat the dozens of fast but under-proven, under-funded but lean/mean, erratic but passionate, and private but multi-university/ grant/ angel investor-funded little Fusion Techno Bunnies??? No-one knows: but the months, years, decades, and Billions are Passing us into the 2020s….
    “…ITER, the world’s largest international scientific collaboration, is beginning assembly of the fusion reactor tokamak that will include 12 different essential hardware systems provided by US ITER, which is managed by Oak Ridge National Laboratory. … The soup-bowl-shaped base of the cryostat was gradually lifted from its frame, carried across the Assembly Hall to the Tokamak Building and eventually lowered into the assembly pit. The operation marked the culmination of a ten-year effort to design, manufacture, deliver, assemble and weld one of the most crucial components. … The first shipment of central solenoid modules will begin later this year…” whew…. agonizing.

  22. How about you critizise what is wrong about his comment instead of him as a person? Just a thought….

  23. “cost of panels hardened for space and launch”. You are in for a surprise! See
    Note that nobody talks of product launch anymore, except to start ISRU. Launch small starter factories only. Note Criswell’s cost estimate does what I describe BEFORE Musk cheap rockets were dreamt of, should be much cheaper now. Enjoy! (I’m assuming you have read “The High Frontier”, or you are in for a REAL surprise!). Also, consider any probs you see with 20-200 TWe nuke on Earth, including the problem, that that solution does not open O’Neill Space for other things, and needs transmission lines.

  24. Brian, not directly connected, but… why you never mention Kairos power. and almost never Oklo Inc. (that has just launched its Aurora power plant)?

  25. The SLS and Orion spacecraft, that’s pork, billions and billions of tax dollars wasted. But a measly $27M on a research that could produce huge benefits if succesful ?. I, for one, welcome this kind of “pork”.

  26. …they can never get as low as Space Solar…

    Can you provide some support? 

    I doubt SpaceX could get the numbers to work even if they’re paying true cost for their own launch service. A company paying retail is looking at $50 per watt in launch costs, too high to be competitive at the utility scale even if the panels are free. It could work, lots of things could, but not the easiest much less cheapest solution. 

    Musk is obsessed with the poor efficiency of space solar, that’s why he hates the idea. I don’t think the efficiency matters if you can get the cost of panels hardened for space and launch down to a manageable level.

    btw, ITER started assembly of their fusion reactor last week. The end of the beginning for the future of fusion is in sight, good times ahead.

  27. Let’s see… well over 20,000 square miles to power the U.S. alone. Let’s simplify and cut down do to greater efficiency in space to 10k square miles. Average solar panel is 15 square feet and weighs 40 lbs. So dividing into 10k square miles that gives us a total of 18,585,600,000‬ panels needed (278784000000 square feet/15 square feet). Take this number and multiply by 40 and you get 743,424,000,000 lbs. Musk’s best predictions for Starship estimate a cost of $20,000 per ton so divide by 2000 then multiply by 20,000 = $7,434,240,000,000‬.

    So how again is this free fusion energy so cheap? I left out maintenance, infrastructure, and construction cost. Your welcome.

  28. Yes, we had a detailed discussion of the concept here; The basic problem being that, either the mirrors were in LEO, and thus couldn’t be useful much beyond daylight hours, and would only have a short window of time to illuminate any particular target while having to slew madly. Or they’d be in GEO, and while they’d be available 24/7 from a given site, they would illuminate a spot the size of an entire state.

    In between you had both problems.

  29. I remember many decades ago reading about a solar energy system composed of many solar mirrors in orbit that would focus sunlight onto earth based power stations. In a way reminds me of StarLink.

    The solar mirrors could be used for many purposes: illuminating fields of crops, illuminating cities at night, provide heating, and providing electrical power.

    Starship low cost lifting capability to orbit could make it economically viable.

  30. No singularity here, The US Department of Energy is advancing all forms of lower emissions energy generation.

  31. Go with Molten Salt using Uranium. Since inherently safe won’t need any complex triple redundant cooling. Won’t need complex fuel rod assemblies. Won’t need special high pressure containment vessel. Focus on cheap and safe and you will get cheap and safe.

  32. If the goal is base energy, they can never get as low as Space Solar. Just collecting existing free fusion energy is cheap! Musk rockets make Criswell Lunar Solar Power so cheap they have already destroyed large scale artificial fusion.

Comments are closed.